research article conflict set and waveform modelling...
TRANSCRIPT
Research ArticleConflict Set and Waveform Modelling forPower Amplifier Design
Anamarija Juhas and Ladislav A Novak
Department of Power Electronics and Communication Engineering Faculty of Technical Sciences University of Novi Sad21000 Novi Sad Serbia
Correspondence should be addressed to Ladislav A Novak ladislavunsacrs
Received 19 July 2014 Accepted 7 February 2015
Academic Editor Ben T Nohara
Copyright copy 2015 A Juhas and L A Novak This is an open access article distributed under the Creative Commons AttributionLicense which permits unrestricted use distribution and reproduction in any medium provided the original work is properlycited
Various classes of nonnegative waveforms containing dc component fundamental and 119896th harmonic (119896 ge 2) which proved to beof interest in waveform modelling for power amplifier (PA) design are considered in this paper In optimization of PA efficiencynonnegative waveforms with maximal amplitude of fundamental harmonic and those with maximal coefficient of cosine term offundamental harmonic (optimal waveforms) play an important role Optimal waveforms have multiple global minima and this factclosely relates the problem of optimization of PA efficiency to the concept of conflict set There is also keen interest in findingdescriptions for various classes of suboptimal waveforms such as nonnegative waveforms with at least one zero nonnegativewaveforms with maximal amplitude of fundamental harmonic for prescribed amplitude of 119896th harmonic nonnegative waveformswith maximal coefficient of cosine part of fundamental harmonic for prescribed coefficients of 119896th harmonic and nonnegativecosine waveforms with at least one zero Closed form descriptions for all these suboptimal types of waveforms are provided in thispaper Suboptimal waveforms may also have multiple global minima and therefore be related to the concept of conflict set Fourcase studies of usage of closed form descriptions of nonnegative waveforms in PA modelling are also provided
1 Introduction
Theorigin of the concept of conflict set goes back to J CMax-well (Maxwell 1831ndash1879) who informally introduced mostof features of what today is called conflict set [1] Fromthis reason Maxwell set or Maxwell stratum is also used assynonyms for conflict set Roughly speaking conflict setassociated with a smooth function 119891 with 119898 parameters isthe set of 119898-tuples in parameter space for which 119891 has mul-tiple global minima Conflict set is also intimately related tosingularity theory and catastrophe theory [1]
Although without explicit reference many max-minmin-max engineering design problems related to nonsmoothoptimizations in parameter spaces (eg see [2]) includingproblems related to the optimization of efficiency of poweramplifiers (PAs) (eg see [3ndash12]) are connected to the con-cept of conflict set The concept of conflict set has been alsoused in mathematics (eg see [13ndash16]) and physics (eg see[17 18]) including subjects like black holes [19]
Nonnegative waveforms with maximal amplitude of fun-damental harmonic and those with maximal coefficient ofcosine term of fundamental harmonic (optimal waveforms)have multiple global minima and therefore are closely relatedto the concept of conflict setThe suboptimal waveforms suchas
(i) nonnegative waveforms with at least one zero(ii) nonnegative waveforms with maximal amplitude of
fundamental harmonic for prescribed amplitude of119896th harmonic
(iii) nonnegative waveforms with maximal coefficient ofcosine part of fundamental harmonic for prescribedcoefficients of 119896th harmonic
(iv) nonnegative cosine waveforms with at least one zeromay also havemultiple global minima [9 11 12] and thereforebe related to the concept of conflict set as well These subop-timal waveforms are clearly of interest in shapingmodelling
Hindawi Publishing CorporationMathematical Problems in EngineeringVolume 2015 Article ID 585962 29 pageshttpdxdoiorg1011552015585962
2 Mathematical Problems in Engineering
drain (collectorplate) waveforms in PA design (eg see [3ndash12 20 21])
Fejer in his seminal paper [22] provided general descrip-tion of all nonnegative trigonometric polynomials with 119899
consecutive harmonics in terms of 2119899 + 2 parameters satis-fying one nonlinear constraint He also derived closed formsolution to the problem of finding maximum possible ampli-tude of the first harmonic of nonnegative cosine polynomialswith consecutive harmonics
Fuzik [3] (see also [10]) considered cosine polynomialswith dc fundamental and 119896th harmonic for arbitrary 119896 ge 2
and provided closed form solution for coefficients of optimalwaveform Rhodes in [7] provided closed form expressionfor maximum possible amplitude of fundamental harmonicof nonnegative waveforms containing consecutive odd har-monics A subclass of nonnegative cosine waveforms withdc fundamental and third harmonic having factorized formdescription has been considered in [23]
High efficiency PA with arbitrary output harmonic ter-minations has been analysed in [9] along with maximal effi-ciency fundamental output power and load impedance
Factorized form of nonnegative waveforms up to secondharmonic with at least one zero has been suggested in [11] inthe context of continuous class BJ mode of PA operation
General description of all nonnegative waveforms up tosecond harmonic in terms of four independent parametershas been provided in [12] This includes nonnegative wave-forms with at least one zero as a special case
End point of conflict set normally corresponds to so-called maximally flat waveform which also belongs to classof suboptimal waveforms First comprehensive usage of max-imally flat waveforms in the context of analysis of PA goesto Raab [20] General description of maximally flat wave-forms with arbitrary number of harmonics has been pre-sented in [21] along with closed form expressions for effi-ciency of class-F and inverse class-F PA with maximally flatwaveforms Description of maximally flat cosine waveformswith consecutive harmonics has been presented in [8] in thecontext of finite harmonic class-C PA
In this paper we provide general descriptions of a num-ber of optimal and suboptimal nonnegative waveforms con-taining dc component fundamental and an arbitrary 119896thharmonic 119896 ge 2 and show how they are related to theconcept of conflict set According to our best knowledge thispaper provides the very first usage of conflict set in the courseof solving problems related to optimization of PA efficiencyMain results are stated in six propositions (Propositions 1 69 18 22 and 26) four corollaries (Corollaries 2ndash5) twentyremarks and three algorithms Four case studies of usage ofclosed form descriptions of nonnegative waveforms in PAefficiency analysis are considered in detail in Section 7
This paper is organized in the following way In Section 2we introduce concepts of minimum function and gain func-tion (Section 21) conflict set (Section 22) and parameterspace (Section 23) In Sections 3ndash6 we provide generaldescriptions of various classes of nonnegative waveformscontaining dc component fundamental and 119896th harmonicwith at least one zero General case of nonnegative waveformswith at least one zero is presented in Section 31The case with
exactly two zeros is considered in Section 32 An algorithmfor calculation of coefficients of fundamental harmonic ofnonnegative waveforms with two zeros for prescribed coeffi-cients of 119896th harmonic is presented in Section 33 Descrip-tion of nonnegative waveforms with maximal amplitude offundamental harmonic for prescribed amplitude of 119896th har-monic is provided in Section 4 Nonnegative waveforms withmaximal coefficient of cosine part of fundamental harmonicfor prescribed coefficients of 119896th harmonic are consideredin Section 51 An illustration of results of Section 51 forparticular case 119896 = 3 is given in Section 52 Section 61 isdevoted to nonnegative cosine waveforms with at least onezero and arbitrary 119896 ge 2 whereas Section 62 considerscosine waveforms with at least one zero for 119896 = 3 InSection 7 four case studies of application of descriptions ofnonnegative waveforms with fundamental and 119896th harmonicin PA modelling are presented In the Appendices list ofsome finite sums of trigonometric functions widely usedthroughout the paper and brief account of the Chebyshevpolynomials are provided
2 Minimum Function Gain Function andConflict Set
In this sectionwe considerminimum function and gain func-tion (Section 21) conflict set (Section 22) and parameterspace (Section 23) in the context of nonnegative waveformswith fundamental and 119896th harmonic
We start with provision of a brief account of the factsrelated to the concepts of minimum function and conflict setFor this purpose let us denote by 119891(119909 119906) a family of smoothfunctions of 119899 variables depending on 119898 parameters where119909 isin 119877
119899 is 119899-tuple of variables and 119906 isin 119877119898 is 119898-tuple of
parameters The minimum function 119865 119877119898
rarr 119877 associatedwith the function 119891 is defined as 119865(119906) = min
119909119891(119909 119906)
Therefore the domain of theminimum function is parameterspace of the function 119891 The minimum function 119865(119906) is con-tinuous but not necessarily smooth function of parameters[13 24] It is a smooth function if 119891(119909 119906) possesses uniqueglobal minimum at nondegenerate critical point [13] (criticalpoint is degenerate if at least first two consecutive derivativesare equal to zero) In this context the conflict set can bedefined as the set of the parameters for which function 119891
has global minimum at a degenerate critical point orandmultiple global minima [13]
For a wide class ofminimum functions when the numberof parameters is not greater than four the behaviour ofminimum function in a neighbourhood of any point can bedescribed by one of ldquonormal formsrdquo from a finite list as statedin [24] For example for smooth function 119891 119877 times 119877
2
rarr 119877the minimum function 119865(119906
1 1199062) = min
119909119891(119909 119906
1 1199062) near the
origin can be locally reduced to one of the following threenormal forms [25] minus|119906
1|min(119906
1 1199062 1199061+ 1199062) or min
119909(1199094
+
11990611199092
+ 1199062119909) In this example the conflict set is the set of all
points (1199061 1199062) for which minimum function 119865(119906
1 1199062) is not
differentiable because function 119891(119909 1199061 1199062) possesses at least
two global minima [25]
Mathematical Problems in Engineering 3
21 Minimum Function and Gain Function In what followswe consider family of waveforms of type
119908 (120591 120574 119860 120572) = 1 minus 120574 (cos 120591 + 119860 cos (119896120591 + 120572)) (1)
where 120591 stands for 120596119905 120574 gt 0 119896 ge 2 119860 gt 0 and 120572 isin [0 2120587)Waveforms of type (1) include all possible shapes which canoccur but not all possiblewaveforms containing fundamentaland 119896th harmonic However shifting of waveforms of type(1) along 120591-axis could recover all possible waveforms withfundamental and 119896th harmonic
The problem of finding nonnegative waveform of type(1) having maximum amplitude of fundamental harmonicplays an important role in optimization of PA efficiency Thisextremal problem can be reformulated as problem of findingnonnegative waveform from family (1) having maximumpossible value of coefficient 120574 Nonnegative waveform offamily (1) with maximum possible value of coefficient 120574 iscalled ldquooptimalrdquo or ldquoextremalrdquo waveform
Furthermore let us introduce an auxiliary waveform
119891 (120591 119860 120572) = minus cos 120591 minus 119860 cos (119896120591 + 120572) (2)
which is smooth function of one variable 120591 and two parame-ters119860 and 120572 In terms of 119891(120591 119860 120572) the above extremal prob-lem reduces to the problem of finding maximum possiblevalue of coefficient 120574 that satisfies
1 + 120574119891 (120591 119860 120572) ge 0 (3)
Clearly for any prescribed pair (119860 120572) there is a uniquemaximal value of coefficient 120574 for which inequality (3) holdsfor all 120591 This maximal value of 120574 associated with the pair(119860 120572) we denote it by 119866(119860 120572) and call it ldquogain functionrdquo
Let
119865min (119860 120572) = min120591
119891 (120591 119860 120572) (4)
be theminimum function associatedwith119891(120591 119860 120572) Accord-ing to (3) 119866(119860 120572) and 119865min(119860 120572) satisfy the following rela-tion 1 + 119866(119860 120572)119865min(119860 120572) = 0 Since 119865min(119860 120572) is obviouslynonzero it follows immediately that
119866 (119860 120572) = minus1
119865min (119860 120572) (5)
A relation analogue to (5) for 119896 = 2 (fundamental andsecond harmonic) has been derived in [4] According toour best knowledge it was the first appearance of gainfunction expressed via associated minimum function Theconsideration presented in [4] has been restricted to theparticular case when 120572 = 120587 The same problem for 120572 = 120587 andarbitrary 119896 ge 2 has been investigated in [3] (see also [10])
According to above consideration the problem of finding3-tuple (120574 119860 120572) with maximum possible value of 120574 for which(3) holds is equivalent to the problem of finding maximumvalue of gain function
120574max = max119860120572
119866 (119860 120572) = minus1
max119860120572
119865min (119860 120572)(6)
14
12
1
08
06
04
G1 G2
001
02
0304
0506 0
1205872
12058731205872
2120587
Parameter 120572
Parameter A
Gai
n fu
nctio
n
k = 2
Figure 1 Graph of 119866(119860 120572) for 119896 = 2 Points 1198661and 119866
2denote
beginning of the ridge and maximum of gain function respectively
and corresponding pair (119860lowast 120572lowast) that satisfies
120574max = max119860120572
119866 (119860 120572) = 119866 (119860lowast
120572lowast
)
max119860120572
119865min (119860 120572) = 119865min (119860lowast
120572lowast
)
(7)
Thus the optimal waveform 119908lowast
(120591) is determined by parame-ters 120574max 119860
lowast and 120572lowast that is
119908lowast
(120591) = 119908 (120591 120574max 119860lowast
120572lowast
) (8)
Optimal waveform has two global minima (this claim willbe justified in Section 4 Remark 21) Consequently the pair(119860lowast
120572lowast
) which corresponds to maximum of gain function119866(119860 120572) belongs to conflict set in (119860 120572) parameter space
Figure 1 shows graph of gain function 119866(119860 120572) for 119896 = 2Notice that it has sharp ridge and that maximum of gainfunction (point 119866
2) lies on the ridge This maximum corre-
sponds to the optimal waveform (solution of the consideredextremal problem) The beginning of the ridge (point 119866
1)
corresponds to the waveform which possesses global mini-mum at degenerate critical point that is corresponds tomax-imally flat waveform (eg see [21]) Gain function 119866(119860 120572)
is not differentiable on the ridge and consequently is notdifferentiable at the point where it has global maximumThisexplains why the approach based on critical points does notwork and why conflict set is so important in the consideredproblem
Positions of global minima of 119891(120591 119860 120572) for 119896 = 2 arepresented in Figure 2 According to Proposition 1 conflict setis the ray defined by119860 gt 14 and120572 = 120587Waveforms119891(120591 119860 120572)with parameters that belong to the conflict set have two globalminimaThewaveform corresponding to the end point of theray (119860 = 14 and 120572 = 120587) has global minimum at degeneratecritical point (so-called maximally flat waveform [21])
Nonnegative waveforms of type (1) with 120574 = 119866(119860 120572) =
minus1119865min(119860 120572) have at least one zero To show that it issufficient to see that 119908(120591 120574 119860 120572) = 0 for 120591 satisfying 119891(120591119860 120572) = 119865min(119860 120572)
4 Mathematical Problems in Engineering
001
02
0304
05
06 01205872
12058731205872
2120587
Parameter 120572
Parameter A
1205872
1205874
0
minus1205874
minus1205872Posit
ions
of g
loba
l min
k = 2
Figure 2 Positions of global minima of 119891(120591 119860 120572) for 119896 = 2
The problem of finding maximum value of fundamentalharmonic cosine part of nonnegative waveform of the form
119908119886(120591 120574119886 119887 119860 120572) = 1 minus 120574
119886(cos 120591 + 119887 sin 120591 + 119860 cos (119896120591 + 120572))
(9)
where 120574119886gt 0 is also related to the problem of finding max-
imumof theminimum functionOptimalwaveformof family(9) has two global minima (this claim will be justified inSection 5 Remark 25) and therefore corresponding 3-tupleof parameters belongs to the conflict set in parameter spaceof family (9)
Let us introduce an auxiliary waveform
119891119886(120591 119887 119860 120572) = minus cos 120591 minus 119887 sin 120591 minus 119860 cos (119896120591 + 120572) (10)
and corresponding minimum function 119865119886min(119887 119860 120572) =
min120591119891119886(120591 119887 119860 120572) Inequality 119908
119886(120591 120574119886 119887 119860 120572) ge 0 can be
rewritten as 1 + 120574119886119865119886min(119887 119860 120572) ge 0 and therefore the
highest value of 120574119886is attained for 120574
119886= minus1119865
119886min(119887 119860 120572) Itimmediately follows that nonnegative waveform of type (9)with 120574
119886= minus1119865
119886min(119887 119860 120572) has zero for 120591 satisfying 119891119886(120591
119887 119860 120572) = 119865119886min(119887 119860 120572)
22 Conflict Set Historically conflict set came into beingfrom the problems in which families of smooth functions(such as potentials distances and waveforms) with two com-peting minima occurThe situation when competing minimabecome equal refers to the presence of conflict set (Maxwellset Maxwell strata) in the associated parameter space
There are many facets of conflict set For example in theproblem involving distances between two sets of points theconflict set is the intersections between iso-distance lines[14] Conflict set also arises in the situation when two wavefronts coming from different objects meet [15 25] In thestudy of black holes conflict set is the line of crossover of thehorizon formed by the merger of two black holes [19] In theclassical Euler problem conflict set is a set of points wheredistinct extremal trajectories with the same value of the costfunctional meet one another [18]
Conflict set is very difficult to calculate both analyticallyand numerically (eg see [15]) because of apparent nondif-ferentiability in some directions In optimization of PA effi-ciency some authors already reported difficulties in findingoptimum via standard analytical tools [4 5]
In this section we consider conflict set in the context offamily of waveforms of type (2) for arbitrary 119896 ge 2 In thiscontext for prescribed integer 119896 ge 2 conflict set is said to bea set of all pairs (119860 120572) for which 119891(120591 119860 120572) possesses multipleglobal minima
Suppose that 1205911015840 and 12059110158401015840 are the positions of global min-
ima of 119891(120591 119860 120572) Then the conflict set is specified by the fol-lowing set of relations
119891 (1205911015840
119860 120572) = 119891 (12059110158401015840
119860 120572) (11)
1198911015840
(1205911015840
119860 120572) = 0 1198911015840
(12059110158401015840
119860 120572) = 0 (12)
11989110158401015840
(1205911015840
119860 120572) gt 0 11989110158401015840
(12059110158401015840
119860 120572) gt 0 (13)
(forall120591) 119891 (120591 119860 120572) ge 119891 (1205911015840
119860 120572) (14)
Relations (12) and (13) say that 119891(120591 119860 120572) has minima at 1205911015840and 12059110158401015840 while relations (11) and (14) imply that these minimaare equal and global
The following proposition describes the conflict set offamily of waveforms of type (2)
Proposition 1 Conflict set of family of waveforms of type (2)is the set of all pairs (119860 120572) such that 119860 gt 1119896
2 and 120572 = 120587
The proof of Proposition 1 which is provided at the endof this section also implies that the following four corollarieshold
Corollary 2 The conflict set has end point at (119860 120572) =
(11198962
120587) This end point corresponds to the maximally flatwaveform [21]
Corollary 3 Waveforms of type (2) with parameters thatbelong to conflict set have two global minima at plusmn120591
Δ where
0 lt 120591Δle 120587119896
Corollary 4 Every waveform with fundamental and 119896th har-monic has either one or two global minima
Corollary 5 Conflict set can be parameterised in terms of 120591Δ
as follows
120572 = 120587 119860 (120591Δ) =
sin 120591Δ
119896 sin 119896120591Δ
0 lt 120591Δle120587
119896 (15)
Notice that119860(120591Δ) ismonotonically increasing function on inter-
val 0 lt 120591Δle 120587119896
Proof of Proposition 1 Without loss of generality we canrestrict our consideration to the interval minus120587 lt 120591 le 120587 This isan immediate consequence of the fact that 119891(120591 119860 120572) is aperiodic function
Suppose that 1205911015840 and 12059110158401015840 where 1205911015840 lt 12059110158401015840 are points at which
119891(120591 119860 120572) has two equal global minima Then conflict set is
Mathematical Problems in Engineering 5
specified by relations (11)ndash(14) From (11)ndash(13) it follows thatrelations
119891 (1205911015840
119860 120572) minus 119891 (12059110158401015840
119860 120572) = 0
1198911015840
(1205911015840
119860 120572) + 1198911015840
(12059110158401015840
119860 120572) = 0
1198911015840
(1205911015840
119860 120572) minus 1198911015840
(12059110158401015840
119860 120572) = 0
11989110158401015840
(1205911015840
119860 120572) + 11989110158401015840
(12059110158401015840
119860 120572) gt 0
(16)
also hold Let
120591119904119903=(1205911015840
+ 12059110158401015840
)
2 120591
Δ=(12059110158401015840
minus 1205911015840
)
2
(17)
be a pair of points associated with (1205911015840 12059110158401015840) Clearly
minus120587 lt 120591119904119903lt 120587 (18)
0 lt 120591Δlt 120587 (19)
12059110158401015840
= 120591119904119903+ 120591Δ 120591
1015840
= 120591119904119903minus 120591Δ (20)
The first and second derivatives of 119891(120591 119860 120572) are equal to
1198911015840
(120591 119860 120572) = sin 120591 + 119896119860 sin (119896120591 + 120572)
11989110158401015840
(120591 119860 120572) = cos 120591 + 1198962119860 cos (119896120591 + 120572) (21)
By using (20)-(21) system (16) can be rewritten as
sin 120591119904119903sin 120591Δ+ 119860 sin (119896120591
119904119903+ 120572) sin 119896120591
Δ= 0 (22)
sin 120591119904119903cos 120591Δ+ 119896119860 sin (119896120591
119904119903+ 120572) cos 119896120591
Δ= 0 (23)
cos 120591119904119903sin 120591Δ+ 119896119860 cos (119896120591
119904119903+ 120572) sin 119896120591
Δ= 0 (24)
cos 120591119904119903cos 120591Δ+ 1198962
119860 cos (119896120591119904119903+ 120572) cos 119896120591
Δgt 0 (25)
From (19) it follows that sin 120591Δgt 0 Multiplying (24) and
(25) with minus cos 120591Δand sin 120591
Δgt 0 respectively and sum-
ming the resulting relations we obtain 119896119860 cos(119896120591119904119903
+
120572)[119896 sin 120591Δcos 119896120591
Δminus sin 119896120591
Δcos 120591Δ] gt 0 The latest relation
immediately implies that
119896 sin 120591Δcos 119896120591
Δminus sin 119896120591
Δcos 120591Δ
= 0 (26)
Equations (22) and (23) can be considered as a system oftwo linear equations in terms of sin 120591
119904119903and 119860 sin(119896120591
119904119903+ 120572)
According to (26) the determinant of this system is nonzeroand therefore it has only trivial solution
sin 120591119904119903= 0 sin (119896120591
119904119903+ 120572) = 0 (27)
According to (18) sin 120591119904119903= 0 implies
120591119904119903= 0 (28)
According to (20) 120591119904119903= 0 implies
12059110158401015840
= 120591Δ 120591
1015840
= minus120591Δ (29)
Furthermore 120591119904119903= 0 and sin(119896120591
119904119903+ 120572) = 0 imply that sin120572 =
0 From (29) it follows that 120591Δis position of global minimum
of 119891(120591 119860 120572) Clearly 119891(120591Δ 119860 120572) le 119891(0 119860 120572) which together
with sin120572 = 0 leads to
1 minus cos 120591Δ+ 119860 cos120572 (1 minus cos 119896120591
Δ) le 0 (30)
From 120591Δ
= 0 (see (19)) 119860 gt 0 and (30) it follows that cos 120572 lt
0 which together with sin120572 = 0 yields
120572 = 120587 (31)
Since 120591Δis position of global minimum it follows that
119891(120591Δ 119860 120587) le 119891(120587119896 119860 120587) Accordingly 119860(1 + cos 119896120591
Δ) le
cos 120591Δminus cos120587119896 which together with 119860 gt 0 implies that
cos 120591Δminus cos120587119896 ge 0 This relation along with (19) yields
0 lt 120591Δle120587
119896 (32)
Substitution of (31) and (28) in (24) leads to
119860 =sin 120591Δ
119896 sin 119896120591Δ
(33)
Notice that sin 119896120591Δ sin 120591
Δis monotonically decreasing func-
tion on interval (32)Therefore parameter119860 is monotonicallyincreasing function on the same interval with lim
120591Δrarr0+119860 =
11198962 Consequently 119860 gt 1119896
2 which completes theproof
23 Parameter Space In parameter space of family of wave-forms (2) there are two subsets playing important role in theclassification of the family instancesThese are conflict set andcatastrophe set
Catastrophe set is subset of parameter space of waveform119891(120591 119860 120572) It consists of those pairs (119860 120572) for which thecorresponding waveforms 119891(120591 119860 120572) have degenerate criticalpoints at which first and second derivatives are equal to zeroThus for finding catastrophe set we have to consider thefollowing system of equations
1198911015840
(120591119889 119860 120572) = 0
11989110158401015840
(120591119889 119860 120572) = 0
(34)
where 120591119889is a degenerate critical point of waveform119891(120591 119860 120572)
Conflict set in parameter space of waveform119891(120591 119860 120572) asshown in Proposition 1 is the ray described by 119860 gt 1119896
2 and120572 = 120587 It is intimately connected to catastrophe set
In what follows in this subsection we use polar coordinatesystem (119860 cos120572 119860 sin120572) instead of Cartesian coordinatesystem (119860 120572) Examples of catastrophe set and conflict setfor 119896 le 5 plotted in parameter space (119860 cos120572 119860 sin120572) arepresented in Figure 3 Solid line represents the catastropheset while dotted line describes conflict set The isolated pickpoints (usually called cusp) which appear in catastrophecurves correspond to maximally flat waveforms with max-imally flat minimum andor maximally flat maximumThereare two such picks in the catastrophe curves for 119896 = 2 and
6 Mathematical Problems in Engineering
00
00
k = 2 k = 3
k = 4 k = 5
Acos 120572
Acos 120572
Acos 120572
Acos 120572
A sin 120572
A sin 120572A sin 120572
A sin 120572
Figure 3 Catastrophe set (solid line) and corresponding conflict set(dotted line) for 119896 le 5 In each plot white triangle dot correspondsto optimal waveform and white circle dot corresponds to maximallyflat waveform
119896 = 4 and one in the catastrophe curves for 119896 = 3 and 119896 = 5Notice that the end point of conflict set is the cusp point
Catastrophe set divides the parameter space (119860 cos120572119860 sin120572) into disjoint subsets In the cases 119896 = 2 and 119896 =
3 catastrophe curve defines inner and outer part For 119896 gt
3 catastrophe curve makes partition of parameter space inseveral inner subsets and one outer subset (see Figure 3)
Notice also that multiplying 119891(120591 119860 120572) with a positiveconstant and adding in turn another constant which leads towaveform of type 119908(120591 120574 119860 120572) (see (1) and (2)) do not makeimpact on the character of catastrophe and conflict sets Thisis because in the course of finding catastrophe set first andsecond derivatives of 119891(120591 119860 120572) are set to zero Clearly (34) interms of 119891(120591 119860 120572) are equivalent to the analogous equationsin terms of119908(120591 120574 119860 120572) Analogously in the course of findingconflict set we consider only the positions of global minima(these positions forwaveforms119891(120591 119860 120572) and119908(120591 120574 119860 120572) arethe same)
3 Nonnegative Waveforms with atLeast One Zero
In what follows let us consider a waveform containing dccomponent fundamental and 119896th (119896 ge 2) harmonic of theform
119879119896(120591) = 1 + 119886
1cos 120591 + 119887
1sin 120591 + 119886
119896cos 119896120591 + 119887
119896sin 119896120591 (35)
The amplitudes of fundamental and 119896th harmonic of wave-form of type (35) respectively are
1205821= radic11988621+ 11988721 (36)
120582119896= radic1198862119896+ 1198872119896 (37)
As it is shown in Section 21 nonnegative waveforms withmaximal amplitude of fundamental harmonic or maximalcoefficient of fundamental harmonic cosine part have atleast one zero It is also shown in Section 22 (Corollary 4)that waveforms of type (35) with nonzero amplitude offundamental harmonic have either one or two globalminimaConsequently if nonnegative waveform of type (35) withnonzero amplitude of fundamental harmonic has at least onezero then it has at most two zeros
In Section 31 we provide general description of nonnega-tive waveforms of type (35) with at least one zero In Sections32 and 33 we consider nonnegative waveforms of type (35)with two zeros
31 General Description of Nonnegative Waveforms with atLeast One Zero The main result of this section is presentedin the following proposition
Proposition 6 Every nonnegative waveform of type (35) withat least one zero can be expressed in the following form
119879119896(120591) = [1 minus cos (120591 minus 120591
0)] [1 minus 120582
119896119903119896(120591)] (38)
where119903119896(120591) = (119896 minus 1) cos 120585
+ 2
119896minus1
sum
119899=1
(119896 minus 119899) cos (119899 (120591 minus 1205910) + 120585)
(39)
providing that
120582119896le [(119896 minus 1) cos 120585 + 119896 sin(120585 minus 120585119896)
sin(120585119896)]
minus1
(40)
10038161003816100381610038161205851003816100381610038161003816 le 120587 (41)
Remark 7 Function on the right hand side of (40) is mono-tonically increasing function of |120585| on interval |120585| le 120587 (formore details about this function see Remark 15) From (57)and (65) it follows that relation
0 le 120582119896le 1 (42)
holds for every nonnegative waveform of type (35) Noticethat according to (40) 120582
119896= 1 implies |120585| = 120587 Substitution
of 120582119896
= 1 and |120585| = 120587 into (55) yields 119879119896(120591) = 1 minus
cos 119896(120591 minus 1205910) Consequently 120582
119896= 1 implies that amplitude
1205821of fundamental harmonic is equal to zero
Remark 8 Conversion of (38) into additive form leads to thefollowing expressions for coefficients of nonnegative wave-forms of type (35) with at least one zero
1198861= minus (1 + 120582
119896cos 120585) cos 120591
0minus 119896120582119896sin 120585 sin 120591
0 (43)
1198871= minus (1 + 120582
119896cos 120585) sin 120591
0+ 119896120582119896sin 120585 cos 120591
0 (44)
119886119896= 120582119896cos (119896120591
0minus 120585) (45)
119887119896= 120582119896sin (119896120591
0minus 120585) (46)
providing that 120582119896satisfy (40) and |120585| le 120587
Mathematical Problems in Engineering 7
2
1
0
0 1 2 3 4
Angle 120591120587
Wav
efor
ms
1205823 = radic2201205823 = radic281205823 = radic24
Figure 4 Nonnegative waveforms with at least one zero for 119896 = 31205910= 1205876 and 120585 = 31205874
Three examples of nonnegative waveforms with at leastone zero for 119896 = 3 are presented in Figure 4 (examples ofnonnegative waveformswith at least one zero for 119896 = 2 can befound in [12]) For all three waveforms presented in Figure 4we assume that 120591
0= 1205876 and 120585 = 31205874 From (40) it follows
that 1205823le radic24 Coefficients of waveform with 120582
3= radic220
(dotted line) are 1198861= minus08977 119887
1= minus03451 119886
3= 005
and 1198873= minus005 Coefficients of waveform with 120582
3= radic28
(dashed line) are 1198861= minus09453 119887
1= minus01127 119886
3= 0125
and 1198873= minus0125 Coefficients of waveform with 120582
3= radic24
(solid line) are 1198861= minus10245 119887
1= 02745 119886
3= 025 and
1198873= minus025 First two waveforms have one zero while third
waveform (presented with solid line) has two zeros
Proof of Proposition 6 Waveform of type (35) containingdc component fundamental and 119896th harmonic can be alsoexpressed in the form
119879119896(120591) = 1 + 120582
1cos (120591 + 120593
1) + 120582119896cos (119896120591 + 120593
119896) (47)
where1205821ge 0120582
119896ge 0120593
1isin (minus120587 120587] and120593
119896isin (minus120587 120587] It is easy
to see that relations between coefficient of (35) and param-eters of (47) read as follows
1198861= 1205821cos1205931 119887
1= minus1205821sin1205931 (48)
119886119896= 120582119896cos120593119896 119887
119896= minus120582119896sin120593119896 (49)
Let us introduce 120585 such that10038161003816100381610038161205851003816100381610038161003816 le 120587 120585 = (119896120591
0+ 120593119896) mod2120587 (50)
Using (50) coefficients (49) can be expressed as (45)-(46)Let us assume that 119879
119896(120591) is nonnegative waveform of type
(35) with at least one zero that is 119879119896(120591) ge 0 and 119879
119896(1205910) = 0
for some 1205910 Notice that conditions 119879
119896(120591) ge 0 and 119879
119896(1205910) = 0
imply that 1198791015840119896(1205910) = 0 From 119879
119896(1205910) = 0 and 1198791015840
119896(1205910) = 0 by
using (50) it follows that
1205821cos (120591
0+ 1205931) = minus (1 + 120582
119896cos 120585)
1205821sin (1205910+ 1205931) = minus119896120582
119896sin 120585
(51)
respectively On the other hand 1205821cos(120591 + 120593
1) can be rewrit-
ten as
1205821cos (120591 + 120593
1) = 1205821cos (120591
0+ 1205931) cos (120591 minus 120591
0)
minus 1205821sin (1205910+ 1205931) sin (120591 minus 120591
0)
(52)
Substitution of (51) into (52) yields
1205821cos (120591 + 120593
1) = minus (1 + 120582
119896cos 120585) cos (120591 minus 120591
0)
+ 119896120582119896sin 120585 sin (120591 minus 120591
0)
(53)
According to (50) it follows that cos(119896120591 + 120593119896) = cos(119896(120591 minus
1205910) + 120585) that is
cos (119896120591 + 120593119896) = cos 120585 cos 119896 (120591 minus 120591
0) minus sin 120585 sin 119896 (120591 minus 120591
0)
(54)
Furthermore substitution of (54) and (53) into (47) leads to
119879119896(120591) = [1 minus cos (120591 minus 120591
0)] [1 + 120582
119896cos 120585]
minus 120582119896[1 minus cos 119896 (120591 minus 120591
0)] cos 120585
+ 120582119896[119896 sin (120591 minus 120591
0) minus sin 119896 (120591 minus 120591
0)] sin 120585
(55)
According to (A2) and (A4) (see Appendices) there is com-mon factor [1minuscos(120591minus120591
0)] for all terms in (55) Consequently
(55) can be written in the form (38) where
119903119896(120591) = minus cos 120585 + [
1 minus cos 119896 (120591 minus 1205910)
1 minus cos (120591 minus 1205910)] cos 120585
minus [119896 sin (120591 minus 120591
0) minus sin 119896 (120591 minus 120591
0)
1 minus cos (120591 minus 1205910)
] sin 120585
(56)
From (56) by using (A2) (A4) and cos 120585 cos 119899(120591 minus 1205910) minus
sin 120585 sin 119899(120591 minus 1205910) = cos(119899(120591 minus 120591
0) + 120585) we obtain (39)
In what follows we are going to prove that (40) also holdsAccording to (38) 119879
119896(120591) is nonnegative if and only if
120582119896max120591
119903119896(120591) le 1 (57)
Let us first show that position of global maximumof 119903119896(120591)
belongs to the interval |120591 minus 1205910| le 2120587119896 Relation (56) can be
rewritten as
119903119896(120591) = 119903
119896(1205910minus2120585
119896) + 119902119896(120591) (58)
where
119903119896(1205910minus2120585
119896) = (119896 minus 1) cos 120585 + 119896 sin (120585 minus 120585119896)
sin (120585119896) (59)
119902119896(120591) =
1
1 minus cos (120591 minus 1205910)
sdot [cos 120585 minus cos(119896(120591 minus 1205910+120585
119896))
+119896 sin 120585sin (120585119896)
(cos(120591 minus 1205910+120585
119896) minus cos(120585
119896))]
(60)
8 Mathematical Problems in Engineering
For |120585| lt 120587 relation sin 120585 sin(120585119896) gt 0 obviously holds Fromcos 119905 gt cos 1199051015840 for |119905| le 120587119896 lt |119905
1015840
| le 120587 it follows that positionof global maximum of the function of type [119888 cos 119905 minus cos(119896119905)]for 119888 gt 0 belongs to interval |119905| le 120587119896 Therefore position ofglobal maximum of the expression in the square brackets in(60) for |120585| lt 120587 belongs to interval |120591 minus 120591
0+ 120585119896| le 120587119896 This
inequality together with |120585| lt 120587 leads to |120591minus1205910| lt 2120587119896 Since
[1 minus cos(120591 minus 1205910)]minus1 decreases with increasing |120591 minus 120591
0| le 120587 it
follows that 119902119896(120591) for |120585| lt 120587 has global maximum on interval
|120591minus1205910| lt 2120587119896 For |120585| = 120587 it is easy to show thatmax
120591119902119896(120591) =
119902119896(1205910plusmn 2120587119896) = 0 Since 119903
119896(120591) minus 119902
119896(120591) is constant (see (58))
it follows from previous considerations that 119903119896(120591) has global
maximum on interval |120591 minus 1205910| le 2120587119896
To find max120591119903119896(120591) let us consider first derivative of 119903
119896(120591)
with respect to 120591 Starting from (56) first derivative of 119903119896(120591)
can be expressed in the following form
119889119903119896(120591)
119889120591= minus119904 (120591) sdot sin(
119896 (120591 minus 1205910)
2+ 120585) (61)
where
119904 (120591) = [sin(119896 (120591 minus 120591
0)
2) cos(
120591 minus 1205910
2)
minus 119896 cos(119896 (120591 minus 120591
0)
2) sin(
120591 minus 1205910
2)]
sdot sinminus3 (120591 minus 1205910
2)
(62)
Using (A6) (see Appendices) (62) can be rewritten as
119904 (120591) = 2
119896minus1
sum
119899=1
119899 (119896 minus 119899) cos((119896 minus 2119899) (120591 minus 120591
0)
2) (63)
From 119899(119896 minus 119899) gt 0 and |119896 minus 2119899| lt 119896 119899 = 1 (119896 minus 1) itfollows that all summands in (63) decrease with increasing|120591 minus 1205910| providing that |120591 minus 120591
0| le 2120587119896 Therefore 119904(120591) ge 119904(120591
0plusmn
2120587119896) = 119896sin2(120587119896) gt 0 for |120591 minus 1205910| le 2120587119896 Consequently
119889119903119896(120591)119889120591 = 0 and |120591minus120591
0| le 2120587119896 imply that sin(119896(120591minus120591
0)2+
120585) = 0From |120585| le 120587 |120591minus120591
0| le 2120587119896 and sin(119896(120591minus120591
0)2+120585) = 0
it follows that 120591minus1205910+120585119896 = minus120585119896 or |120591minus120591
0+120585119896| = (2120587minus|120585|)119896
and therefore cos(119896(120591 minus 1205910+ 120585119896)) = cos 120585 Since cos(120585119896) ge
cos(2120587 minus |120585|)119896 it follows that max120591119902119896(120591) is attained for 120591 =
1205910minus2120585119896 Furthermore from (60) it follows that max
120591119902119896(120591) =
119902119896(1205910minus 2120585119896) = 0 which together with (58)-(59) leads to
max120591
119903119896(120591) = 119903
119896(1205910minus2120585
119896)
= (119896 minus 1) cos 120585 + 119896 sin (120585 minus 120585119896)sin (120585119896)
(64)
Both terms on the right hand side of (64) are even functionsof 120585 and decrease with increase of |120585| |120585| le 120587 Thereforemax120591119903119896(120591) attains its lowest value for |120585| = 120587 It is easy to
show that right hand side of (64) for |120585| = 120587 is equal to 1which further implies that
max120591
119903119896(120591) ge 1 (65)
From (65) it follows that (57) can be rewritten as 120582119896
le
[max120591119903119896(120591)]minus1 Finally substitution of (64) into 120582
119896le
[max120591119903119896(120591)]minus1 leads to (40) which completes the proof
32 Nonnegative Waveforms with Two Zeros Nonnegativewaveforms of type (35) with two zeros always possess twoglobal minima Such nonnegative waveforms are thereforerelated to the conflict set
In this subsection we provide general description of non-negative waveforms of type (35) for 119896 ge 2 and exactly twozeros According to Remark 7 120582
119896= 1 implies |120585| = 120587 and
119879119896(120591) = 1 minus cos 119896(120591 minus 120591
0) Number of zeros of 119879
119896(120591) = 1 minus
cos 119896(120591minus1205910) on fundamental period equals 119896 which is greater
than two for 119896 gt 2 and equal to two for 119896 = 2 In the followingproposition we exclude all waveforms with 120582
119896= 1 (the case
when 119896 = 2 and 1205822= 1 is going to be discussed in Remark 10)
Proposition 9 Every nonnegative waveform of type (35) withexactly two zeros can be expressed in the following form
119879119896(120591) = 120582
119896[1 minus cos (120591 minus 120591
0)] [1 minus cos(120591 minus 120591
0+2120585
119896)]
sdot [1198880+ 2
119896minus2
sum
119899=1
119888119899cos 119899(120591 minus 120591
0+120585
119896)]
(66)
where
119888119899= [sin(120585 minus 119899120585
119896) cos(120585
119896)
minus (119896 minus 119899) cos(120585 minus 119899120585
119896) sin(120585
119896)]
sdot sinminus3 (120585119896)
(67)
120582119896= [(119896 minus 1) cos 120585 + 119896 sin(120585 minus 120585119896)
sin(120585119896)]
minus1
(68)
0 lt10038161003816100381610038161205851003816100381610038161003816 lt 120587 (69)
Remark 10 For 119896 = 2 waveforms with 1205822= 1 also have
exactly two zeros These waveforms can be included in aboveproposition by substituting (69) with 0 lt |120585| le 120587
Remark 11 Apart from nonnegative waveforms of type (35)with two zeros there are another two types of nonnegativewaveforms which can be obtained from (66)ndash(68) These are
(i) nonnegative waveforms with 119896 zeros (correspondingto |120585| = 120587) and
(ii) maximally flat nonnegative waveforms (correspond-ing to 120585 = 0)
Notice that nonnegative waveforms of type (35) with120582119896= 1 can be obtained from (66)ndash(68) by setting |120585| =
120587 Substitution of 120582119896
= 1 and |120585| = 120587 into (66) alongwith execution of all multiplications and usage of (A2) (seeAppendices) leads to 119879
119896(120591) = 1 minus cos 119896(120591 minus 120591
0)
Mathematical Problems in Engineering 9
Also maximally flat nonnegative waveforms (they haveonly one zero [21]) can be obtained from (66)ndash(68) by setting120585 = 0 Thus substitution of 120585 = 0 into (66)ndash(68) leads tothe following form of maximally flat nonnegative waveformof type (35)
119879119896(120591) =
[1 minus cos (120591 minus 1205910)]2
3 (1198962 minus 1)
sdot [119896 (1198962
minus 1)
+ 2
119896minus2
sum
119899=1
(119896 minus 119899) ((119896 minus 119899)2
minus 1) cos 119899 (120591 minus 1205910)]
(70)
Maximally flat nonnegative waveforms of type (35) for 119896 le 4
can be expressed as
1198792(120591) =
2
3[1 minus cos(120591 minus 120591
0)]2
1198793(120591) =
1
2[1 minus cos(120591 minus 120591
0)]2
[2 + cos (120591 minus 1205910)]
1198794(120591) =
4
15[1 minus cos (120591 minus 120591
0)]2
sdot [5 + 4 cos (120591 minus 1205910) + cos 2 (120591 minus 120591
0)]
(71)
Remark 12 Every nonnegative waveform of type (35) withexactly one zero at nondegenerate critical point can bedescribed as in Proposition 6 providing that symbol ldquolerdquoin relation (40) is replaced with ldquoltrdquo This is an immediateconsequence of Propositions 6 and 9 and Remark 11
Remark 13 Identity [1minus cos(120591minus1205910)][1minus cos(120591minus120591
0+2120585119896)] =
[cos 120585119896 minus cos(120591 minus 1205910+ 120585119896)]
2 implies that (66) can be alsorewritten as
119879119896(120591) = 120582
119896[cos 120585119896
minus cos(120591 minus 1205910+120585
119896)]
2
sdot [1198880+ 2
119896minus2
sum
119899=1
119888119899cos 119899(120591 minus 120591
0+120585
119896)]
(72)
Furthermore substitution of (67) into (72) leads to
119879119896(120591)
= 120582119896[cos 120585119896
minus cos(120591 minus 1205910+120585
119896)]
sdot [(119896 minus 1) sin 120585sin (120585119896)
minus 2
119896minus1
sum
119899=1
sin (120585 minus 119899120585119896)sin (120585119896)
cos 119899(120591 minus 1205910+120585
119896)]
(73)
Remark 14 According to (A6) (see Appendices) it followsthat coefficients (67) can be expressed as
119888119899= 2
119896minus119899minus1
sum
119898=1
119898(119896 minus 119899 minus 119898) cos((119896 minus 119899 minus 2119898) 120585119896
) (74)
Furthermore from (74) it follows that coefficients 119888119896minus2
119888119896minus3
119888119896minus4
and 119888119896minus5
are equal to119888119896minus2
= 2 (75)
119888119896minus3
= 8 cos(120585119896) (76)
119888119896minus4
= 8 + 12 cos(2120585119896) (77)
119888119896minus5
= 24 cos(120585119896) + 16 cos(3120585
119896) (78)
For example for 119896 = 2 (75) and (68) lead to 1198880= 2 and
1205822= 1(2+cos 120585) respectively which from (72) further imply
that
1198792(120591) =
2 [cos(1205852) minus cos(120591 minus 1205910+ 1205852)]
2
[2 + cos 120585] (79)
Also for 119896 = 3 (75) (76) and (68) lead to 1198881= 2 119888
0=
8 cos(1205853) and 1205823= [2(3 cos(1205853) + cos 120585)]minus1 respectively
which from (72) further imply that
1198793(120591) =
2 [cos (1205853) minus cos (120591 minus 1205910+ 1205853)]
2
[3 cos (1205853) + cos 120585]
sdot [2 cos(1205853) + cos(120591 minus 120591
0+120585
3)]
(80)
Remark 15 According to (A5) (see Appendices) relation(68) can be rewritten as
120582119896= [(119896 minus 1) cos 120585 + 119896
119896minus1
sum
119899=1
cos((119896 minus 2119899)120585119896
)]
minus1
(81)
Clearly amplitude 120582119896of 119896th harmonic of nonnegative wave-
form of type (35) with exactly two zeros is even functionof 120585 Since cos((119896 minus 2119899)120585119896) 119899 = 0 (119896 minus 1) decreaseswith increase of |120585| on interval 0 le |120585| le 120587 it follows that120582119896monotonically increases with increase of |120585| Right hand
side of (68) is equal to 1(1198962
minus 1) for 120585 = 0 and to one for|120585| = 120587 Therefore for nonnegative waveforms of type (35)with exactly two zeros the following relation holds
1
1198962 minus 1lt 120582119896lt 1 (82)
The left boundary in (82) corresponds to maximally flatnonnegative waveforms (see Remark 11) The right boundaryin (82) corresponds to nonnegative waveforms with 119896 zeros(also see Remark 11)
Amplitude of 119896th harmonic of nonnegative waveform oftype (35) with two zeros as a function of parameter 120585 for 119896 le5 is presented in Figure 5
Remark 16 Nonnegative waveform of type (35) with twozeros can be also expressed in the following form
119879119896(120591) = 1 minus 120582
119896
119896 sin 120585sin (120585119896)
cos(120591 minus 1205910+120585
119896)
+ 120582119896cos (119896 (120591 minus 120591
0) + 120585)
(83)
10 Mathematical Problems in Engineering
1
08
06
04
02
0minus1 minus05 0 05
Am
plitu
de120582k
1
Parameter 120585120587
k = 2k = 3
k = 4k = 5
Figure 5 Amplitude of 119896th harmonic of nonnegative waveformwith two zeros as a function of parameter 120585
where 120582119896is given by (68) and 0 lt |120585| lt 120587 From (83) it follows
that coefficients of fundamental harmonic of nonnegativewaveform of type (35) with two zeros are
1198861= minus1205821cos(120591
0minus120585
119896) 119887
1= minus1205821sin(120591
0minus120585
119896) (84)
where 1205821is amplitude of fundamental harmonic
1205821=
119896 sin 120585sin (120585119896)
120582119896 (85)
Coefficients of 119896th harmonic are given by (45)-(46)Notice that (68) can be rewritten as
120582119896= [cos(120585
119896)
119896 sin 120585sin(120585119896)
minus cos 120585]minus1
(86)
By introducing new variable
119909 = cos(120585119896) (87)
and using the Chebyshev polynomials (eg see Appendices)relations (85) and (86) can be rewritten as
1205821= 119896120582119896119880119896minus1
(119909) (88)
120582119896=
1
119896119909119880119896minus1
(119909) minus 119881119896(119909)
(89)
where119881119896(119909) and119880
119896(119909) denote the Chebyshev polynomials of
the first and second kind respectively From (89) it followsthat
120582119896[119896119909119880119896minus1
(119909) minus 119881119896(119909)] minus 1 = 0 (90)
which is polynomial equation of 119896th degree in terms of var-iable 119909 From 0 lt |120585| lt 120587 and (87) it follows that
cos(120587119896) lt 119909 lt 1 (91)
Since 120582119896is monotonically increasing function of |120585| 0 lt |120585| lt
120587 it follows that 120582119896is monotonically decreasing function of
119909 This further implies that (90) has only one solution thatsatisfies (91) (For 119896 = 2 expression (91) reads cos(1205872) le
119909 lt 1) This solution for 119909 (which can be obtained at leastnumerically) according to (88) leads to amplitude 120582
1of
fundamental harmonicFor 119896 le 4 solutions of (90) and (91) are
119909 = radic1 minus 1205822
21205822
1
3lt 1205822le 1
119909 =1
23radic1205823
1
8lt 1205823lt 1
119909 = radic1
6(1 + radic
51205824+ 3
21205824
)1
15lt 1205824lt 1
(92)
Insertion of (92) into (88) leads to the following relationsbetween amplitude 120582
1of fundamental and amplitude 120582
119896of
119896th harmonic 119896 le 4
1205821= radic8120582
2(1 minus 120582
2)
1
3lt 1205822le 1 (93)
1205821= 3 (
3radic1205823minus 1205823)
1
8lt 1205823lt 1 (94)
1205821= radic
32
27(radic2120582
4(3 + 5120582
4)3
minus 21205824(9 + 7120582
4))
1
15lt 1205824lt 1
(95)
Proof of Proposition 9 As it has been shown earlier (seeProposition 6) nonnegative waveform of type (35) with atleast one zero can be represented in form (38) Since weexclude nonnegative waveforms with 120582
119896= 1 according to
Remark 7 it follows that we exclude case |120585| = 120587Therefore inthe quest for nonnegative waveforms of type (35) having twozeros we will start with waveforms of type (38) for |120585| lt 120587It is clear that nonnegative waveforms of type (38) have twozeros if and only if
120582119896= [max120591
119903119896(120591)]minus1
(96)
and max120591119903119896(120591) = 119903
119896(1205910) According to (64) max
120591119903119896(120591) =
119903119896(1205910) implies |120585| = 0 Therefore it is sufficient to consider
only the interval (69)Substituting (96) into (38) we obtain
119879119896(120591) =
[1 minus cos (120591 minus 1205910)] [max
120591119903119896(120591) minus 119903
119896(120591)]
max120591119903119896(120591)
(97)
Mathematical Problems in Engineering 11
Expression max120591119903119896(120591) minus 119903
119896(120591) according to (64) and (39)
equals
max120591
119903119896(120591) minus 119903
119896(120591) = 119896
sin ((119896 minus 1) 120585119896)sin (120585119896)
minus 2
119896minus1
sum
119899=1
(119896 minus 119899) cos (119899 (120591 minus 1205910) + 120585)
(98)
Comparison of (97) with (66) yields
max120591
119903119896(120591) minus 119903
119896(120591) = [1 minus cos(120591 minus 120591
0+2120585
119896)]
sdot [1198880+ 2
119896minus2
sum
119899=1
119888119899cos 119899(120591 minus 120591
0+120585
119896)]
(99)
where coefficients 119888119899 119899 = 0 119896 minus 2 are given by (67) In
what follows we are going to show that right hand sides of(98) and (99) are equal
From (67) it follows that
1198880minus 1198881cos(120585
119896) = 119896
sin (120585 minus 120585119896)sin (120585119896)
(100)
Also from (67) for 119899 = 1 119896minus3 it follows that the followingrelations hold
(119888119899minus1
+ 119888119899+1
) cos(120585119896) minus 2119888
119899= 2 (119896 minus 119899) cos(120585 minus 119899120585
119896)
(119888119899minus1
minus 119888119899+1
) sin(120585119896) = 2 (119896 minus 119899) sin(120585 minus 119899120585
119896)
(101)
From (99) by using (75) (76) (100)-(101) and trigonometricidentities
cos(120591 minus 1205910+2120585
119896) = cos(120585
119896) cos(120591 minus 120591
0+120585
119896)
minus sin(120585119896) sin(120591 minus 120591
0+120585
119896)
cos(120585 minus 119899120585
119896) cos(119899(120591 minus 120591
0+120585
119896))
minus sin(120585 minus 119899120585
119896) sin(119899(120591 minus 120591
0+120585
119896))
= cos (119899 (120591 minus 1205910) + 120585)
(102)
we obtain (98) Consequently (98) and (99) are equal whichcompletes the proof
33 Nonnegative Waveforms with Two Zeros and PrescribedCoefficients of 119896thHarmonic In this subsectionwe show thatfor prescribed coefficients 119886
119896and 119887119896 there are 119896 nonnegative
waveforms of type (35) with exactly two zeros According to
(37) and (82) coefficients 119886119896and 119887119896of nonnegative waveforms
of type (35) with exactly two zeros satisfy the followingrelation
1
1198962 minus 1lt radic1198862119896+ 1198872119896lt 1 (103)
According to Remark 16 the value of 119909 (see (87)) that cor-responds to 120582
119896= radic1198862
119896+ 1198872119896can be determined from (90)-
(91) As we mentioned earlier (90) has only one solutionthat satisfies (91) This value of 119909 according to (88) leadsto the amplitude 120582
1of fundamental harmonic (closed form
expressions for 1205821in terms of 120582
119896and 119896 le 4 are given by (93)ndash
(95))On the other hand from (45)-(46) it follows that
1198961205910minus 120585 = atan 2 (119887
119896 119886119896) + 2119902120587 119902 = 1 (119896 minus 1) (104)
where function atan 2(119910 119909) is defined as
atan 2 (119910 119909) =
arctan(119910
119909) if 119909 ge 0
arctan(119910
119909) + 120587 if 119909 lt 0 119910 ge 0
arctan(119910
119909) minus 120587 if 119909 lt 0 119910 lt 0
(105)
with the codomain (minus120587 120587] Furthermore according to (84)and (104) the coefficients of fundamental harmonic of non-negative waveforms with two zeros and prescribed coeffi-cients of 119896th harmonic are equal to
1198861= minus1205821cos[
atan 2 (119887119896 119886119896) + 2119902120587
119896]
1198871= minus1205821sin[
atan 2 (119887119896 119886119896) + 2119902120587
119896]
(106)
where 119902 = 0 (119896minus 1) For chosen 119902 according to (104) and(66) positions of zeros are
1205910=1
119896[120585 + atan 2 (119887
119896 119886119896) + 2119902120587]
1205910minus2120585
119896=1
119896[minus120585 + atan 2 (119887
119896 119886119896) + 2119902120587]
(107)
From (106) and 119902 = 0 (119896minus1) it follows that for prescribedcoefficients 119886
119896and 119887119896 there are 119896 nonnegative waveforms of
type (35) with exactly two zerosWe provide here an algorithm to facilitate calculation
of coefficients 1198861and 1198871of nonnegative waveforms of type
(35) with two zeros and prescribed coefficients 119886119896and 119887
119896
providing that 119886119896and 119887119896satisfy (103)
12 Mathematical Problems in Engineering
2
1
0
Wav
efor
ms
0 1 2 3 4
Angle 120591120587
q = 0
q = 1
q = 2
Figure 6 Nonnegative waveforms with two zeros for 119896 = 3 1198863=
minus015 and 1198873= minus02
Algorithm 17 (i) Calculate 120582119896= radic1198862119896+ 1198872119896
(ii) identify 119909 that satisfies both relations (90) and (91)(iii) calculate 120582
1according to (88)
(iv) choose integer 119902 such that 0 le 119902 le 119896 minus 1(v) calculate 119886
1and 1198871according to (106)
For 119896 le 4 by using (93) for 119896 = 2 (94) for 119896 = 3 and (95)for 119896 = 4 it is possible to calculate directly 120582
1from 120582
119896and
proceed to step (iv)For 119896 = 2 and prescribed coefficients 119886
2and 1198872 there are
two waveforms with two zeros one corresponding to 1198861lt 0
and the other corresponding to 1198861gt 0 (see also [12])
Let us take as an input 119896 = 3 1198863= minus015 and 119887
3= minus02
Execution of Algorithm 17 on this input yields 1205823= 025 and
1205821= 11399 (according to (94)) For 119902 = 0 we calculate
1198861= minus08432 and 119887
1= 07670 (corresponding waveform is
presented by solid line in Figure 6) for 119902 = 1 we calculate1198861= minus02426 and 119887
1= minus11138 (corresponding waveform is
presented by dashed line) for 119902 = 2 we calculate 1198861= 10859
and 1198871= 03468 (corresponding waveform is presented by
dotted line)As another example of the usage of Algorithm 17 let us
consider case 119896 = 4 and assume that1198864= minus015 and 119887
4= minus02
Consequently 1205824= 025 and 120582
1= 09861 (according to (95))
For 119902 = 0 3we calculate the following four pairs (1198861 1198871) of
coefficients of fundamental harmonic (minus08388 05184) for119902 = 0 (minus05184 minus08388) for 119902 = 1 (08388 minus05184) for 119902 =2 and (05184 08388) for 119902 = 3 Corresponding waveformsare presented in Figure 7
4 Nonnegative Waveforms with MaximalAmplitude of Fundamental Harmonic
In this section we provide general description of nonnegativewaveforms containing fundamental and 119896th harmonic withmaximal amplitude of fundamental harmonic for prescribedamplitude of 119896th harmonic
The main result of this section is presented in the fol-lowing proposition
3
2
1
0
Wav
efor
ms
0 1 2 3 4
Angle 120591120587
q = 0q = 1
q = 2q = 3
Figure 7 Nonnegative waveforms with two zeros for 119896 = 4 1198864=
minus015 and 1198874= minus02
Proposition 18 Every nonnegativewaveformof type (35)withmaximal amplitude 120582
1of fundamental harmonic and pre-
scribed amplitude 120582119896of 119896th harmonic can be expressed in the
following form
119879119896(120591) = [1 minus cos (120591 minus 120591
0)]
sdot [1 minus (119896 minus 1) 120582119896minus 2120582119896
119896minus1
sum
119899=1
(119896 minus 119899) cos (119899 (120591 minus 1205910))]
(108)
if 0 le 120582119896le 1(119896
2
minus 1) or
119879119896(120591) = 120582
119896[1 minus cos (120591 minus 120591
0)] [1 minus cos(120591 minus 120591
0+2120585
119896)]
sdot [1198880+ 2
119896minus2
sum
119899=1
119888119899cos 119899(120591 minus 120591
0+120585
119896)]
(109)
if 1(1198962 minus 1) le 120582119896le 1 providing that 119888
119899 119899 = 0 119896 minus 2 and
120582119896are related to 120585 via relations (67) and (68) respectively and
|120585| le 120587
Remark 19 Expression (108) can be obtained from (38) bysetting 120585 = 0 Furthermore insertion of 120585 = 0 into (43)ndash(46)leads to the following expressions for coefficients ofwaveformof type (108)
1198861= minus (1 + 120582
119896) cos 120591
0 119887
1= minus (1 + 120582
119896) sin 120591
0
119886119896= 120582119896cos (119896120591
0) 119887
119896= 120582119896sin (119896120591
0)
(110)
On the other hand (109) coincides with (66) Thereforethe expressions for coefficients of (109) and (66) also coincideThus expressions for coefficients of fundamental harmonic ofwaveform (109) are given by (84) where 120582
1is given by (85)
while expressions for coefficients of 119896th harmonic are givenby (45)-(46)
Waveforms described by (108) have exactly one zerowhile waveforms described by (109) for 1(1198962 minus 1) lt 120582
119896lt 1
Mathematical Problems in Engineering 13
14
12
1
08
06
04
02
00 05 1
Amplitude 120582k
Am
plitu
de1205821
k = 2
k = 3
k = 4
Figure 8 Maximal amplitude of fundamental harmonic as a func-tion of amplitude of 119896th harmonic
have exactly two zeros As we mentioned earlier waveforms(109) for 120582
119896= 1 have 119896 zeros
Remark 20 Maximal amplitude of fundamental harmonic ofnonnegative waveforms of type (35) for prescribed amplitudeof 119896th harmonic can be expressed as
1205821= 1 + 120582
119896 (111)
if 0 le 120582119896le 1(119896
2
minus 1) or
1205821=
119896 sin 120585119896 sin 120585 cos (120585119896) minus cos 120585 sin (120585119896)
(112)
if 1(1198962 minus 1) le 120582119896le 1 where 120585 is related to 120582
119896via (68) (or
(86)) and |120585| le 120587From (110) it follows that (111) holds Substitution of (86)
into (85) leads to (112)Notice that 120582
119896= 1(119896
2
minus 1) is the only common point ofthe intervals 0 le 120582
119896le 1(119896
2
minus 1) and 1(1198962
minus 1) le 120582119896le
1 According to (111) 120582119896= 1(119896
2
minus 1) corresponds to 1205821=
1198962
(1198962
minus1) It can be also obtained from (112) by setting 120585 = 0The waveforms corresponding to this pair of amplitudes aremaximally flat nonnegative waveforms
Maximal amplitude of fundamental harmonic of non-negative waveform of type (35) for 119896 le 4 as a function ofamplitude of 119896th harmonic is presented in Figure 8
Remark 21 Maximum value of amplitude of fundamentalharmonic of nonnegative waveform of type (35) is
1205821max =
1
cos (120587 (2119896)) (113)
This maximum value is attained for |120585| = 1205872 (see (112)) Thecorresponding value of amplitude of 119896th harmonic is 120582
119896=
(1119896) tan(120587(2119896)) Nonnegative waveforms of type (35) with1205821= 1205821max have two zeros at 1205910 and 1205910 minus 120587119896 for 120585 = 1205872 or
at 1205910and 1205910+ 120587119896 for 120585 = minus1205872
14
12
1
08
06
04
02
0minus1 minus05 0 05 1
Am
plitu
de1205821
Parameter 120585120587
k = 2k = 3k = 4
Figure 9 Maximal amplitude of fundamental harmonic as a func-tion of parameter 120585
To prove that (113) holds let us first show that the fol-lowing relation holds for 119896 ge 2
cos( 120587
2119896) lt 1 minus
1
1198962 (114)
From 119896 ge 2 it follows that sinc(120587(4119896)) gt sinc(1205874) wheresinc 119909 = (sin119909)119909 and therefore sin(120587(4119896)) gt 1(radic2119896)By using trigonometric identity cos 2119909 = 1 minus 2sin2119909 weimmediately obtain (114)
According to (111) and (112) it is clear that 1205821attains its
maximum value on the interval 1(1198962 minus 1) le 120582119896le 1 Since
120582119896is monotonic function of |120585| on interval |120585| le 120587 (see
Remark 15) it follows that 119889120582119896119889120585 = 0 for 0 lt |120585| lt 120587
Therefore to find critical points of 1205821as a function of 120582
119896
it is sufficient to find critical points of 1205821as a function of
|120585| 0 lt |120585| lt 120587 and consider its values at the end points120585 = 0 and |120585| = 120587 Plot of 120582
1as a function of parameter 120585
for 119896 le 4 is presented in Figure 9 According to (112) firstderivative of 120582
1with respect to 120585 is equal to zero if and only
if (119896 cos 120585 sin(120585119896) minus sin 120585 cos(120585119896)) cos 120585 = 0 On interval0 lt |120585| lt 120587 this is true if and only if |120585| = 1205872 Accordingto (112) 120582
1is equal to 119896
2
(1198962
minus 1) for 120585 = 0 equal to zerofor |120585| = 120587 and equal to 1 cos(120587(2119896)) for |120585| = 1205872 From(114) it follows that 1198962(1198962minus1) lt 1 cos(120587(2119896)) and thereforemaximum value of 120582
1is given by (113) Moreover maximum
value of 1205821is attained for |120585| = 1205872
According to above consideration all nonnegative wave-forms of type (35) having maximum value of amplitude offundamental harmonic can be obtained from (109) by setting|120585| = 1205872 Three of them corresponding to 119896 = 3 120585 = 1205872and three different values of 120591
0(01205876 and1205873) are presented
in Figure 10 Dotted line corresponds to 1205910= 0 (coefficients
of corresponding waveform are 1198861= minus1 119887
1= 1radic3 119886
3= 0
and 1198873= minusradic39) solid line to 120591
0= 1205876 (119886
1= minus2radic3 119887
1= 0
1198863= radic39 and 119887
3= 0) and dashed line to 120591
0= 1205873 (119886
1= minus1
1198871= minus1radic3 119886
3= 0 and 119887
3= radic39)
Proof of Proposition 18 As it has been shown earlier (Propo-sition 6) nonnegative waveform of type (35) with at least
14 Mathematical Problems in Engineering
2
1
0
0 1 2 3 4
Angle 120591120587
Wav
efor
ms
1205910 = 01205910 = 12058761205910 = 1205873
Figure 10 Nonnegative waveforms with maximum amplitude offundamental harmonic for 119896 = 3 and 120585 = 1205872
one zero can be represented in form (38) According to (43)(44) and (36) for amplitude 120582
1of fundamental harmonic of
waveforms of type (38) the following relation holds
1205821= radic(1 + 120582
119896cos 120585)2 + 11989621205822
119896sin2120585 (115)
where 120582119896satisfy (40) and |120585| le 120587
Because of (40) in the quest of finding maximal 1205821for
prescribed 120582119896 we have to consider the following two cases
(Case i)120582119896lt [(119896minus1) cos 120585 + 119896 sin(120585minus120585119896) sin(120585119896)]minus1
(Case ii)120582119896= [(119896minus1) cos 120585+119896 sin(120585minus120585119896) sin(120585119896)]minus1
Case i Since 120582119896lt [(119896 minus 1) cos 120585 + 119896 sin(120585 minus 120585119896) sin(120585119896)]minus1
implies 120582119896
= 1 according to (115) it follows that 1205821
= 0Hence 119889120582
1119889120585 = 0 implies
2120582119896sin 120585 [1 minus (1198962 minus 1) 120582
119896cos 120585] = 0 (116)
Therefore 1198891205821119889120585 = 0 if 120582
119896= 0 (Option 1) or sin 120585 = 0
(Option 2) or (1198962 minus 1)120582119896cos 120585 = 1 (Option 3)
Option 1 According to (115) 120582119896= 0 implies 120582
1= 1 (notice
that this implication shows that 1205821does not depend on 120585 and
therefore we can set 120585 to zero value)
Option 2 According to (115) sin 120585 = 0 implies 1205821= 1 +
120582119896cos 120585 which further leads to the conclusion that 120582
1is
maximal for 120585 = 0 For 120585 = 0 120582119896lt [(119896 minus 1) cos 120585 + 119896 sin(120585 minus
120585119896) sin(120585119896)]minus1 becomes 120582119896lt 1(119896
2
minus 1)
Option 3 This option leads to contradiction To show thatnotice that (119896
2
minus 1)120582119896cos 120585 = 1 and 120582
119896lt [(119896 minus
1) cos 120585 + 119896 sin(120585 minus 120585119896) sin(120585119896)]minus1 imply that (119896 minus 1) cos 120585 gtsin(120585minus120585119896) sin(120585119896) Using (A5) (see Appendices) the latestinequality can be rewritten assum119896minus1
119899=1[cos 120585minuscos((119896minus2119899)120585119896)] gt
0 But from |119896 minus 2119899| lt 119896 119899 = 1 (119896 minus 1) and |120585| le 120587
it follows that all summands are not positive and therefore(119896minus1) cos 120585 gt sin(120585minus120585119896) sin(120585119896) does not hold for |120585| le 120587
Consequently Case i implies 120585 = 0 and 120582119896lt 1(119896
2
minus 1)Finally substitution of 120585 = 0 into (38) leads to (108) whichproves that (108) holds for 120582
119896lt 1(119896
2
minus 1)
Case ii Relation120582119896= [(119896minus1) cos 120585+119896 sin(120585minus120585119896) sin(120585119896)]minus1
according to Proposition 9 and Remark 11 implies that cor-responding waveforms can be expressed via (66)ndash(68) for|120585| le 120587 Furthermore 120582
119896= [(119896 minus 1) cos 120585 + 119896 sin(120585 minus
120585119896) sin(120585119896)]minus1 and |120585| le 120587 imply 1(1198962 minus 1) le 120582119896le 1
This proves that (109) holds for 1(1198962 minus 1) le 120582119896le 1
Finally let us prove that (108) holds for 120582119896= 1(119896
2
minus
1) According to (68) (see also Remark 11) this value of 120582119896
corresponds to 120585 = 0 Furthermore substitution of 120582119896=
1(1198962
minus 1) and 120585 = 0 into (109) leads to (70) which can berewritten as
119879119896(120591) =
[1 minus cos (120591 minus 1205910)]
(1 minus 1198962)
sdot [119896 (119896 minus 1) minus 2
119896minus1
sum
119899=1
(119896 minus 119899) cos (119899 (120591 minus 1205910))]
(117)
Waveform (117) coincides with waveform (108) for 120582119896
=
1(1 minus 1198962
) Consequently (108) holds for 120582119896= 1(1 minus 119896
2
)which completes the proof
5 Nonnegative Waveforms with MaximalAbsolute Value of the Coefficient of CosineTerm of Fundamental Harmonic
In this sectionwe consider general description of nonnegativewaveforms of type (35) with maximal absolute value ofcoefficient 119886
1for prescribed coefficients of 119896th harmonicThis
type of waveform is of particular interest in PA efficiencyanalysis In a number of cases of practical interest eithercurrent or voltage waveform is prescribed In such casesthe problem of finding maximal efficiency of PA can bereduced to the problem of finding nonnegative waveformwith maximal coefficient 119886
1for prescribed coefficients of 119896th
harmonic (see also Section 7)In Section 51 we provide general description of nonneg-
ative waveforms of type (35) with maximal absolute value ofcoefficient 119886
1for prescribed coefficients of 119896th harmonic In
Section 52 we illustrate results of Section 51 for particularcase 119896 = 3
51 Nonnegative Waveforms with Maximal Absolute Value ofCoefficient 119886
1for 119896 ge 2 Waveforms 119879
119896(120591) of type (35) with
1198861ge 0 can be derived from those with 119886
1le 0 by shifting
by 120587 and therefore we can assume without loss of generalitythat 119886
1le 0 Notice that if 119896 is even then shifting 119879
119896(120591) by
120587 produces the same result as replacement of 1198861with minus119886
1
(119886119896remains the same) On the other hand if 119896 is odd then
shifting 119879119896(120591) by 120587 produces the same result as replacement
of 1198861with minus119886
1and 119886119896with minus119886
119896
According to (37) coefficients of 119896th harmonic can beexpressed as
119886119896= 120582119896cos 120575 119887
119896= 120582119896sin 120575 (118)
Mathematical Problems in Engineering 15
where
|120575| le 120587 (119)
Conversely for prescribed coefficients 119886119896and 119887
119896 120575 can be
determined as
120575 = atan 2 (119887119896 119886119896) (120)
where definition of function atan 2(119910 119909) is given by (105)The main result of this section is stated in the following
proposition
Proposition 22 Every nonnegative waveform of type (35)withmaximal absolute value of coefficient 119886
1le 0 for prescribed
coefficients 119886119896and 119887119896of 119896th harmonic can be represented as
119879119896(120591)
= [1 minus cos 120591]
sdot [1 minus (119896 minus 1) 119886119896minus 2
119896minus1
sum
119899=1
(119896 minus 119899) (119886119896cos 119899120591 + 119887
119896sin 119899120591)]
(121)
if 119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) le 1+119886
119896 where 120575 = atan 2(bk
119886119896) or
119879119896(120591) = 120582
119896[1 minus cos(120591 minus (120575 + 120585)
119896)]
sdot [1 minus cos(120591 minus (120575 minus 120585)
119896)]
sdot [1198880+ 2
119896minus2
sum
119899=1
119888119899cos 119899(120591 minus 120575
119896)]
(122)
if 119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) ge 1 + 119886
119896 where 119888
119899 119899 = 0
119896minus2 and 120582119896= radic1198862119896+ 1198872119896are related to 120585 via relations (67) and
(68) respectively and |120585| le 120587
Remark 23 Expression (121) can be obtained from (38) bysetting 120591
0= 0 and 120585 = minus120575 and then replacing 120582
119896cos 120575 with
119886119896(see (118)) and 120582
119896cos(119899120591 minus 120575) with 119886
119896cos 119899120591 + 119887
119896sin 119899120591
(see also (118)) Furthermore insertion of 1205910= 0 and 120585 =
minus120575 into (43)ndash(46) leads to the following relations betweenfundamental and 119896th harmonic coefficients of waveform(121)
1198861= minus (1 + 119886
119896) 119887
1= minus119896119887
119896 (123)
On the other hand expression (122) can be obtained from(66) by replacing 120591
0minus120585119896with 120575119896 Therefore substitution of
1205910minus 120585119896 = 120575119896 in (84) leads to
1198861= minus1205821cos(120575
119896) 119887
1= minus1205821sin(120575
119896) (124)
where 1205821is given by (85)
The fundamental harmonic coefficients 1198861and 1198871of wave-
form of type (35) with maximal absolute value of coefficient1198861le 0 satisfy both relations (123) and (124) if 119886
119896and 119887119896satisfy
1 + 119886119896= 119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) For such waveforms
relations 1205910= 0 and 120585 = minus120575 also hold
Remark 24 Amplitude of 119896th harmonic of nonnegativewaveform of type (35) with maximal absolute value of coeffi-cient 119886
1le 0 and coefficients 119886
119896 119887119896satisfying 1 + 119886
119896=
119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) is
120582119896=
sin (120575119896)119896 sin 120575 cos (120575119896) minus cos 120575 sin (120575119896)
(125)
To show that it is sufficient to substitute 119886119896= 120582119896cos 120575 (see
(118)) into 1 + 119886119896= 119896120582119896[sin 120575 sin(120575119896)] cos(120575119896)
Introducing new variable
119910 = cos(120575119896) (126)
and using the Chebyshev polynomials (eg see Appendices)relations 119886
119896= 120582119896cos 120575 and (125) can be rewritten as
119886119896= 120582119896119881119896(119910) (127)
120582119896=
1
119896119910119880119896minus1
(119910) minus 119881119896(119910)
(128)
where119881119896(119910) and119880
119896(119910) denote the Chebyshev polynomials of
the first and second kind respectively Substitution of (128)into (127) leads to
119886119896119896119910119880119896minus1
(119910) minus (1 + 119886119896) 119881119896(119910) = 0 (129)
which is polynomial equation of 119896th degree in terms of var-iable 119910 From |120575| le 120587 and (126) it follows that
cos(120587119896) le 119910 le 1 (130)
In what follows we show that 119886119896is monotonically increas-
ing function of 119910 on the interval (130) From 120585 = minus120575 (seeRemark 23) and (81) it follows that 120582minus1
119896= (119896 minus 1) cos 120575 +
119896sum119896minus1
119899=1cos((119896 minus 2119899)120575119896) ge 1 and therefore 119886
119896= 120582119896cos 120575 can
be rewritten as
119886119896=
cos 120575(119896 minus 1) cos 120575 + 119896sum119896minus1
119899=1cos ((119896 minus 2119899) 120575119896)
(131)
Obviously 119886119896is even function of 120575 and all cosines in (131)
are monotonically decreasing functions of |120575| on the interval|120575| le 120587 It is easy to show that cos((119896 minus 2119899)120575119896) 119899 =
1 (119896 minus 1) decreases slower than cos 120575 when |120575| increasesThis implies that denominator of the right hand side of(131) decreases slower than numerator Since denominator ispositive for |120575| le 120587 it further implies that 119886
119896is decreasing
function of |120575| on interval |120575| le 120587 Consequently 119886119896is
monotonically increasing function of 119910 on the interval (130)Thus we have shown that 119886
119896is monotonically increasing
function of 119910 on the interval (130) and therefore (129) hasonly one solution that satisfies (130) According to (128) thevalue of 119910 obtained from (129) and (130) either analyticallyor numerically leads to amplitude 120582
119896of 119896th harmonic
16 Mathematical Problems in Engineering
1
05
0
minus05
minus1
minus1 minus05 0 05 1
Coefficient ak
Coe
ffici
entb
k
radica2k+ b2
kle 1
k = 2k = 3k = 4
Figure 11 Plot of (119886119896 119887119896) satisfying 1 + 119886
119896= 119896120582
119896[sin 120575 sin(120575
119896)] cos(120575119896) for 119896 le 4
By solving (129) and (130) for 119896 le 4 we obtain
119910 = radic1 + 1198862
2 (1 minus 1198862) minus1 le 119886
2le1
3
119910 = radic3
4 (1 minus 21198863) minus1 le 119886
3le1
8
119910 =radicradic2 minus 4119886
4+ 1011988624minus 2 (1 minus 119886
4)
4 (1 minus 31198864)
minus1 le 1198864le
1
15
(132)
Insertion of (132) into (128) leads to the following explicitexpressions for the amplitude 120582
119896 119896 le 4
1205822=1
2(1 minus 119886
2) minus1 le 119886
2le1
3 (133)
1205822
3= [
1
3(1 minus 2119886
3)]
3
minus1 le 1198863le1
8 (134)
1205824=1
4(minus1 minus 119886
4+ radic2 minus 4119886
4+ 1011988624) minus1 le 119886
4le
1
15
(135)
Relations (133)ndash(135) define closed lines (see Figure 11) whichseparate points representing waveforms of type (121) frompoints representing waveforms of type (122) For given 119896points inside the corresponding curve refer to nonnegativewaveforms of type (121) whereas points outside curve (andradic1198862119896+ 1198872119896le 1) correspond to nonnegative waveforms of type
(122) Points on the respective curve correspond to the wave-forms which can be expressed in both forms (121) and (122)
Remark 25 Themaximum absolute value of coefficient 1198861of
nonnegative waveform of type (35) is
100381610038161003816100381611988611003816100381610038161003816max =
1
cos (120587 (2119896)) (136)
This maximum value is attained for |120585| = 1205872 and 120575 = 0
(see (124)) Notice that |1198861|max is equal to the maximum value
1205821max of amplitude of fundamental harmonic (see (113))
Coefficients of waveform with maximum absolute value ofcoefficient 119886
1 1198861lt 0 are
1198861= minus
1
cos (120587 (2119896)) 119886
119896=1
119896tan( 120587
(2119896))
1198871= 119887119896= 0
(137)
Waveformdescribed by (137) is cosinewaveformhaving zerosat 120587(2119896) and minus120587(2119896)
In the course of proving (136) notice first that |1198861|max le
1205821max holds According to (123) and (124) maximum of |119886
1|
occurs for 119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) ge 1 + 119886
119896 From (124)
it immediately follows that maximum value of |1198861| is attained
if and only if 1205821= 1205821max and 120575 = 0 which because of
120575119896 = 1205910minus120585119896 further implies 120591
0= 120585119896 Sincemaximumvalue
of 1205821is attained for |120585| = 1205872 it follows that corresponding
waveform has zeros at 120587(2119896) and minus120587(2119896)
Proof of Proposition 22 As it was mentioned earlier in thissection we can assume without loss of generality that 119886
1le 0
We consider waveforms119879119896(120591) of type (35) such that119879
119896(120591) ge 0
and119879119896(120591) = 0 for some 120591
0 Fromassumption that nonnegative
waveform 119879119896(120591) of type (35) has at least one zero it follows
that it can be expressed in form (38)Let us also assume that 120591
0is position of nondegenerate
critical point Therefore 119879119896(1205910) = 0 implies 1198791015840
119896(1205910) = 0 and
11987910158401015840
119896(1205910) gt 0 According to (55) second derivative of 119879
119896(120591) at
1205910can be expressed as 11987910158401015840
119896(1205910) = 1 minus 120582
119896(1198962
minus 1) cos 120585 Since11987910158401015840
119896(1205910) gt 0 it follows immediately that
1 minus 120582119896(1198962
minus 1) cos 120585 gt 0 (138)
Let us further assume that 119879119896(120591) has exactly one zeroThe
problem of finding maximum absolute value of 1198861is con-
nected to the problem of finding maximum of the minimumfunction (see Section 21) If waveforms possess unique globalminimum at nondegenerate critical point then correspond-ing minimum function is a smooth function of parameters[13] Consequently assumption that 119879
119896(120591) has exactly one
zero at nondegenerate critical point leads to the conclusionthat coefficient 119886
1is differentiable function of 120591
0 First
derivative of 1198861(see (43)) with respect to 120591
0 taking into
account that 1205971205851205971205910= 119896 (see (50)) can be expressed in the
following factorized form
1205971198861
1205971205910
= sin 1205910[1 minus 120582
119896(1198962
minus 1) cos 120585] (139)
Mathematical Problems in Engineering 17
From (138) and (139) it is clear that 12059711988611205971205910= 0 if and only if
sin 1205910= 0 According toRemark 12 assumption that119879
119896(120591)has
exactly one zero implies 120582119896lt 1 From (51) (48) and 120582
119896lt 1
it follows that 1198861cos 1205910+ 1198871sin 1205910lt 0 which together with
sin 1205910= 0 implies that 119886
1cos 1205910lt 0 Assumption 119886
1le 0
together with relations 1198861cos 1205910lt 0 and sin 120591
0= 0 further
implies 1198861
= 0 and
1205910= 0 (140)
Insertion of 1205910= 0 into (38) leads to
119879119896(120591)
= [1 minus cos 120591]
sdot [1 minus (119896 minus 1) 120582119896cos 120585 minus 2120582
119896
119896minus1
sum
119899=1
(119896 minus 119899) cos (119899120591 + 120585)]
(141)
Substitution of 1205910= 0 into (45) and (46) yields 119886
119896= 120582119896cos 120585
and 119887119896
= minus120582119896sin 120585 respectively Replacing 120582
119896cos 120585 with
119886119896and 120582
119896cos(119899120591 + 120585) with (119886
119896cos 119899120591 + 119887
119896sin 119899120591) in (141)
immediately leads to (121)Furthermore 119886
119896= 120582119896cos 120585 119887
119896= minus120582
119896sin 120585 and (118)
imply that
120575 = minus120585 (142)
According to (38)ndash(40) and (142) it follows that (141) is non-negative if and only if
120582119896[(119896 minus 1) cos 120575 + 119896 sin (120575 minus 120575119896)
sin (120575119896)] lt 1 (143)
Notice that 119886119896= 120582119896cos 120575 implies that the following relation
holds
120582119896[(119896 minus 1) cos 120575 + 119896 sin (120575 minus 120575119896)
sin (120575119896)]
= minus119886119896+ 119896120582119896
sin 120575sin (120575119896)
cos(120575119896)
(144)
Finally substitution of (144) into (143) leads to 119896120582119896[sin 120575
sin(120575119896)] cos(120575119896) lt 1 + 119886119896 which proves that (121) holds
when 119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) lt 1 + 119886
119896
Apart from nonnegative waveforms with exactly one zeroat nondegenerate critical point in what follows we will alsoconsider other types of nonnegative waveforms with at leastone zero According to Proposition 9 and Remark 11 thesewaveforms can be described by (66)ndash(68) providing that 0 le|120585| le 120587
According to (35) 119879119896(0) ge 0 implies 1 + 119886
1+ 119886119896ge 0
Consequently 1198861le 0 implies that |119886
1| le 1 + 119886
119896 On the other
hand according to (123) |1198861| = 1 + 119886
119896holds for waveforms
of type (121) The converse is also true 1198861le 0 and |119886
1| =
1 + 119886119896imply 119886
1= minus1 minus 119886
119896 which further from (35) implies
119879119896(0) = 0 Therefore in what follows it is enough to consider
only nonnegativewaveformswhich can be described by (66)ndash(68) and 0 le |120585| le 120587 with coefficients 119886
119896and 119887119896satisfying
119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) ge 1 + 119886
119896
For prescribed coefficients 119886119896and 119887119896 the amplitude 120582
119896=
radic1198862119896+ 1198872119896of 119896th harmonic is also prescribed According to
Remark 15 (see also Remark 16) 120582119896is monotonically
decreasing function of 119909 = cos(120585119896) The value of 119909 can beobtained by solving (90) subject to the constraint cos(120587119896) le119909 le 1 Then 120582
1can be determined from (88) From (106) it
immediately follows that maximal absolute value of 1198861le 0
corresponds to 119902 = 0 which from (104) and (120) furtherimplies that
120575 = 1198961205910minus 120585 (145)
Furthermore 119902 = 0 according to (107) implies that waveformzeros are
1205910=(120575 + 120585)
119896 120591
1015840
0= 1205910minus2120585
119896=(120575 minus 120585)
119896 (146)
Substitution of 1205910= (120575 + 120585)119896 into (66) yields (122) which
proves that (122) holds when 119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) ge
1 + 119886119896
In what follows we prove that (121) also holds when119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) = 1 + 119886
119896 Substitution of 119886
119896=
120582119896cos 120575 into 119896120582
119896[sin 120575 sin(120575119896)] cos(120575119896) = 1 + 119886
119896leads to
120582119896[(119896 minus 1) cos 120575 + 119896 sin (120575 minus 120575119896)
sin (120575119896)] = 1 (147)
As we mentioned earlier relation (142) holds for all wave-forms of type (121) Substituting (142) into (147) we obtain
120582119896[(119896 minus 1) cos 120585 + 119896 sin (120585 minus 120585119896)
sin (120585119896)] = 1 (148)
This expression can be rearranged as
120582119896
119896 sin ((119896 minus 1) 120585119896)sin 120585119896
= 1 minus (119896 minus 1) 120582119896cos 120585 (149)
On the other hand for waveforms of type (122) according to(68) relations (148) and (149) also hold Substitution of 120591
0=
(120575 + 120585)119896 (see (145)) and (67) into (122) leads to
119879119896(120591)
= 120582119896[1 minus cos (120591 minus 120591
0)]
sdot [119896 sin ((119896 minus 1) 120585119896)
sin 120585119896minus 2
119896minus1
sum
119899=1
(119896 minus 119899) cos (119899 (120591 minus 1205910) + 120585)]
(150)
Furthermore substitution of (142) into (145) implies that1205910
= 0 Finally substitution of 1205910
= 0 and (149) into(150) leads to (141) Therefore (141) holds when 119896120582
119896[sin 120575
sin(120575119896)] cos(120575119896) = 1 + 119886119896 which in turn shows that (121)
holds when 119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) = 1 + 119886
119896 This
completes the proof
18 Mathematical Problems in Engineering
52 Nonnegative Waveforms with Maximal Absolute Valueof Coefficient 119886
1for 119896 = 3 Nonnegative waveform of type
(35) for 119896 = 3 is widely used in PA design (eg see [10])In this subsection we illustrate results of Section 51 for thisparticular case The case 119896 = 2 is presented in detail in [12]
Coefficients of fundamental harmonic of nonnegativewaveform of type (35) with 119896 = 3 and maximal absolutevalue of coefficient 119886
1le 0 for prescribed coefficients 119886
3and
1198873(1205823= radic11988623+ 11988723) according to (123) (124) (134) (94) and
(120) are equal to
1198861= minus1 minus 119886
3 119887
1= minus3119887
3 (151)
if 12058223le [(1 minus 2119886
3)3]3
1198861= minus1205821cos(120575
3) 119887
1= minus1205821sin(120575
3) (152)
where 1205821= 3(
3radic1205823minus 1205823) and 120575 = atan 2(119887
3 1198863) if [(1 minus
21198863)3]3
le 1205822
3le 1The line 1205822
3= [(1minus2119886
3)3]3 (see case 119896 = 3
in Figure 11) separates points representing waveforms withcoefficients satisfying (151) from points representing wave-forms with coefficients satisfying (152) Waveforms describedby (151) for 1205822
3lt [(1 minus 2119886
3)3]3 have exactly one zero at
1205910= 0 Waveforms described by (151) and (152) for 1205822
3= [(1 minus
21198863)3]3 also have zero at 120591
0= 0 These waveforms as a rule
have exactly two zeros However there are two exceptionsone related to the maximally flat nonnegative waveform withcoefficients 119886
1= minus98 119886
3= 18 and 119887
1= 1198873= 0 which
has only one zero and the other related to the waveform withcoefficients 119886
1= 0 119886
3= minus1 and 119887
1= 1198873= 0 which has three
zerosWaveforms described by (152) for [(1minus21198863)3]3
lt 1205822
3lt
1 have two zeros Waveforms with 1205823= 1 have only third
harmonic (fundamental harmonic is zero)Plot of contours of maximal absolute value of coefficient
1198861 1198861le 0 for prescribed coefficients 119886
3and 1198873is presented
in Figure 12 According to Remark 25 the waveform withmaximum absolute value of 119886
1le 0 is fully described with
the following coefficients 1198861
= minus2radic3 1198863
= radic39 and1198871= 1198873= 0 This waveform has two zeros at plusmn1205876
Two examples of nonnegative waveforms for 119896 = 3
and maximal absolute value of coefficient 1198861 1198861le 0 with
prescribed coefficients 1198863and 1198873are presented in Figure 13
One waveform corresponds to the case 12058223lt [(1 minus 2119886
3)3]3
(solid line) and the other to the case 12058223gt [(1 minus 2119886
3)3]3
(dashed line)Thewaveform represented by solid line has onezero and its coefficients are 119886
3= minus01 119887
3= 01 119886
1= minus09
and 1198871= minus03 Dashed line corresponds to the waveform
having two zeros with coefficients 1198863= minus01 119887
3= 03 119886
1=
minus08844 and 1198871= minus06460 (case 1205822
3gt [(1 minus 2119886
3)3]3)
6 Nonnegative Cosine Waveforms withat Least One Zero
Nonnegative cosine waveforms have proved to be of impor-tance for waveform modelling in PA design (eg see [10])In this section we consider nonnegative cosine waveforms
1
05
0
minus05
minus1
minus1 minus05 0 05 1
Coefficient a3
Coe
ffici
entb
3
02
04
06
08
10
11
Figure 12 Contours ofmaximal absolute value of coefficient 1198861 1198861le
0 as a function of 1198863and 1198873
2
1
0
0 1 2 3 4
Angle 120591120587
Wav
efor
ms
a3 = minus01 b3 = 01
a3 = minus01 b3 = 03
Figure 13 Nonnegative waveforms for 119896 = 3 and maximal absolutevalue of 119886
1 1198861le 0 with prescribed coefficients 119886
3and 1198873
containing fundamental and 119896th harmonic with at least onezero
Cosine waveform with dc component fundamental and119896th harmonic can be obtained from (35) by setting 119887
1= 119887119896=
0 that is
119879119896(120591) = 1 + 119886
1cos 120591 + 119886
119896cos 119896120591 (153)
In Section 61 we provide general description of non-negative cosine waveforms of type (153) with at least onezero We show that nonnegative cosine waveforms with atleast one zero coincide with nonnegative cosine waveformswith maximal absolute value of coefficient 119886
1for prescribed
coefficient 119886119896 In Section 62 we illustrate results of Section 61
for particular case 119896 = 3
61 Nonnegative Cosine Waveforms with at Least One Zerofor 119896 ge 2 Amplitudes of fundamental and 119896th harmonic
Mathematical Problems in Engineering 19
of cosine waveform of type (153) are 1205821= |1198861| and 120582
119896=
|119886119896| respectively According to (42) for nonnegative cosine
waveforms of type (153) the following relation holds
minus1 le 119886119896le 1 (154)
This explains why 119896th harmonic coefficient 119886119896in Proposi-
tion 26 goes through interval [minus1 1]Waveforms (153) with 119886
1ge 0 can be obtained from
waveforms with 1198861le 0 by shifting by 120587 and therefore with-
out loss of generality we can assume that 1198861le 0
Proposition 26 Each nonnegative cosine waveform of type(153) with 119886
1le 0 and at least one zero can be represented as
119879119896(120591) = [1 minus cos 120591] [1 minus (119896 minus 1) 119886
119896minus 2119886119896
119896minus1
sum
119899=1
(119896 minus 119899) cos 119899120591]
(155)
if minus1 le 119886119896le 1(119896
2
minus 1) or
119879119896(120591) = 119886
119896[1 minus cos (120591 minus 120591
0)] [1 minus cos (120591 + 120591
0)]
sdot [1198880+ 2
119896minus2
sum
119899=1
119888119899cos 119899120591]
(156)
where
119888119899=sin ((119896 minus 119899) 120591
0) cos 120591
0minus (119896 minus 119899) cos ((119896 minus 119899) 120591
0) sin 120591
0
sin31205910
(157)
119886119896=
sin 1205910
119896 sin (1198961205910) cos 120591
0minus cos (119896120591
0) sin 120591
0
(158)
100381610038161003816100381612059101003816100381610038161003816 le
120587
119896 (159)
if 1(1198962 minus 1) le 119886119896le 1
Remark 27 Identity [1minuscos(120591minus1205910)][1minuscos(120591+120591
0)] = [cos 120591
0minus
cos 120591]2 implies that (156) can be rewritten as
119879119896(120591) = 119886
119896[cos 120591
0minus cos 120591]2 [119888
0+ 2
119896minus2
sum
119899=1
119888119899cos 119899120591] (160)
Furthermore substitution of (157) into (160) leads to
119879119896(120591) = 119886
119896[cos 120591
0minus cos 120591]
sdot [(119896 minus 1) sin 119896120591
0
sin 1205910
minus 2
119896minus1
sum
119899=1
sin ((119896 minus 119899) 1205910)
sin 1205910
cos 119899120591]
(161)
Remark 28 All nonnegative cosine waveforms of type (153)with at least one zero and 119886
1le 0 except one of them can be
represented either in form (155) or form (156)This exceptionis maximally flat cosine waveform with 119886
1lt 0 which can be
obtained from (155) for 119886119896= 1(119896
2
minus 1) or from (156) for 1205910=
0 Maximally flat cosine waveform with 1198861lt 0 can also be
obtained from (70) by setting 1205910= 0 Furthermore setting
1205910= 0 in (71) leads to maximally flat cosine waveforms for
119896 le 4 and 1198861lt 0
Remark 29 Nonnegative cosine waveform of type (155) with1198861lt 0 and minus1 lt 119886
119896le 1(119896
2
minus 1) has exactly one zero at120591 = 0 Nonnegative cosine waveform described by (156) with1198861lt 0 and 1(1198962 minus 1) lt 119886
119896lt 1 has two zeros at plusmn120591
0 where
0 lt |1205910| lt 120587119896 For 119886
119896= ∓1 nonnegative cosine waveform
of type (153) reduces to 119879119896(120591) = 1 ∓ cos 119896120591 (clearly these two
waveforms both have 119896 zeros)
Remark 30 Transformation of (155) into an additive formleads to the following relation
1198861= minus1 minus 119886
119896 (162)
where minus1 le 119886119896le 1(119896
2
minus1) Similarly transformation of (156)leads to the following relation
1198861= minus119886119896
119896 sin 1198961205910
sin 1205910
(163)
where 119886119896is given by (158) 1(1198962minus1) le 119886
119896le 1 and |120591
0| le 120587119896
Notice that coefficients of maximally flat cosine waveformnamely 119886
119896= 1(119896
2
minus1) and 1198861= minus1198962
(1198962
minus1) satisfy relation(162) They also satisfy relation (163) for 120591
0= 0
Remark 31 Nonnegative cosine waveforms of type (153) withat least one zero coincide with nonnegative cosine waveformswith maximal absolute value of coefficient 119886
1for prescribed
coefficient 119886119896
In proving that Remark 31 holds notice that expression(155) can be obtained from (121) by setting 119887
119896= 0 Fur-
thermore if 119886119896
ge 0 then 120582119896
= 119886119896 which together
with 119887119896
= 0 and (118) implies 120575 = 0 In this case119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) le 1 + 119886
119896becomes 1198962119886
119896le 1 +
119886119896 On the other hand if 119886
119896lt 0 then 120582
119896= minus119886
119896 which
together with 119887119896= 0 and (118) implies |120575| = 120587 In this case
119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) le 1 + 119886
119896becomes 0 le 1 + 119886
119896
Therefore every nonnegative cosine waveform of type (155)has maximal absolute value of coefficient 119886
1for prescribed
coefficient 119886119896 when minus1 le 119886
119896le 1(119896
2
minus 1)Let us now show that expression (156) can be obtained
from (122) by setting 119887119896= 0 and 119886
119896gt 0 For waveforms
of type (122) according to (118) 119887119896= 0 and 119886
119896gt 0 imply
120575 = 0 and 120582119896= 119886119896 Substitution of 120582
119896= 119886119896and 120575 = 0 into
119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) ge 1+119886
119896leads to 119886
119896ge 1(119896
2
minus1)Furthermore substitution of 120575 = 0 into (145) yields 120591
0= 120585119896
Insertion of 120582119896= 119886119896 120575 = 0 and 120591
0= 120585119896 into (122) leads
to (156) Therefore every nonnegative cosine waveform oftype (156) has maximal absolute value of coefficient 119886
1for
prescribed coefficient 119886119896 when 1(1198962 minus 1) le 119886
119896le 1
Proof of Proposition 26 Let us start with nonnegative cosinewaveform of type (153) with 120582
119896= |119886119896| = 1 According to
Remark 7 120582119896= |119886119896| = 1 implies that 120582
1= |1198861| = 0
Substitution of 119886119896
= minus1 into (155) and using (A2) (seeAppendices) lead to 119879
119896(120591) = 1 minus cos 119896120591 Consequently (155)
holds for 119886119896= minus1 On the other hand substitution of 119886
119896= 1
into (158) yields |1205910| = 120587119896 Furthermore substitution of
119886119896= 1 and 120591
0= 120587119896 (or 120591
0= minus120587119896) into (156) along
with performing all multiplications and using (A2) leads to
20 Mathematical Problems in Engineering
119879119896(120591) = 1 + cos 119896120591 Consequently (156)ndash(158) hold for 119886
119896= 1
and |1205910| = 120587119896
It is easy to see that 120582119896= |119886119896| lt 1 and 119879
119896(1205910) = 0 for some
1205910imply 120582
1= |1198861| = 0 Therefore in what follows we assume
that |119886119896| = 1 and 119886
1lt 0
Cosine waveforms are even functions of 120591 Therefore ifnonnegative cosine waveform has exactly one zero it has to beeither at 0 or at 120587 On the other hand if nonnegative cosinewaveform with 119886
1= 0 has exactly two zeros then these zeros
are placed at plusmn1205910 such that 120591
0is neither 0 nor 120587
In order to prove that (155) holds for minus1 lt 119886119896le 1(1minus119896
2
)let us start by referring to the description (38) of nonnegativewaveformswith at least one zero As wementioned earlier fornonnegative cosine waveformwith exactly one zero (denotedby 1205910) it is either 120591
0= 0 or 120591
0= 120587 Therefore in both cases
sin 1205910= 0 Substitution of sin 120591
0= 0 into (43) together with
1198861
= 0 and 120582119896= |119886119896| lt 1 leads to
1205910= 0 (164)
Clearly 1205910= 0 119887
1= 0 and 119887
119896= 0 according to (44) and (46)
imply 120582119896sin 120585 = 0 Since 120582
119896= |119886119896| it follows that |119886
119896| sin 120585 = 0
also holds which further implies 120582119896= 119886119896= 0 or sin 120585 =
0 In the case when 120582119896= 119886119896= 0 from (164) and (43) we
obtain 1198861= minus1 which further implies that 119879
119896(120591) = 1 minus cos 120591
Consequently (155) holds for 119886119896= 0 In the case when sin 120585 =
0 from (164) and (45) we obtain 119886119896= 120582119896if 120585 = 0 or 119886
119896= minus120582119896
if 120585 = 120587 Relations 119886119896= 120582119896and 120585 = 0 according to (40) imply
that 0 le 119886119896le 1(1 minus 119896
2
) Substitution of 120585 = 0 120582119896= 119886119896 and
(164) into (38) leads to (155) which proves that (155) holdsfor 0 le 119886
119896le 1(1 minus 119896
2
) On the other hand relations 119886119896=
minus120582119896and 120585 = 120587 according to (40) imply that minus1 lt 119886
119896le 0
Substitution of 120585 = 120587 120582119896= minus119886119896 and (164) into (38) also leads
to (155) which proves that (155) also holds for minus1 lt 119886119896le 0
Consequently (155) holds for minus1 lt 119886119896le 1(1 minus 119896
2
)In what follows we first prove that (156)-(157) hold for
1(1 minus 1198962
) lt 119886119896lt 1 For this purpose let us start with non-
negative waveforms with two zeros described by (66) As wementioned before nonnegative cosine waveforms with twozeros have zeros at 120591
0and minus120591
0 such that 120591
0= 0 and 120591
0= 120587
Relations 1198861
lt 0 and 1198871
= 0 according to (84) implycos(1205910minus 120585119896) = 1 and therefore
120585
119896= 1205910 (165)
From 120585119896 = 1205910and 0 lt |120585| lt 120587 it follows that 0 lt |120591
0| lt
120587119896 Insertion of 120585119896 = 1205910into (45) yields 119886
119896= 120582119896 Relations
119886119896= 120582119896and (82) imply that 1(1 minus 1198962) lt 119886
119896lt 1 Substitution
of 120582119896= 119886119896and 120585119896 = 120591
0into (66)ndash(68) leads to (156)ndash(158)
which proves that (156)ndash(158) hold for 1(1 minus 1198962
) lt 119886119896lt 1
and 0 lt |1205910| lt 120587119896
Finally substitution of 119886119896= 1(1 minus 119896
2
) and 1205910= 0 into
(161) leads to
119879119896(120591) =
[1 minus cos 120591](1 minus 1198962)
[119896 (119896 minus 1) minus 2
119896minus1
sum
119899=1
(119896 minus 119899) cos 119899120591]
(166)
Waveform (166) coincides with waveform (155) for 119886119896
=
1(1 minus 1198962
) which in turn proves that (156) holds for 119886119896=
1(1 minus 1198962
) and 1205910= 0 This completes the proof
62 Nonnegative CosineWaveforms with at Least One Zero for119896 = 3 In this subsection we consider nonnegative cosinewaveforms with at least one zero for 119896 = 3 (for case 119896 = 2
see [12])Cosine waveform with fundamental and third harmonic
reads
1198793(120591) = 1 + 119886
1cos 120591 + 119886
3cos 3120591 (167)
For 1198861le 0 and minus1 le 119886
3le 18 according to (155) non-
negative cosine waveform of type (167) with at least one zerocan be expressed as
1198793(120591) = (1 minus cos 120591) [1 minus 2119886
3(1 + 2 cos 120591 + cos 2120591)] (168)
From 1198793(120591 + 120587) = 2 minus 119879
3(120591) it immediately follows that for
1198861ge 0 and minus18 le 119886
3le 1 119879
3(120591) can be expressed as
1198793(120591) = (1 + cos 120591) [1 + 2119886
3(1 minus 2 cos 120591 + cos 2120591)] (169)
For 1198861le 0 and 18 le 119886
3le 1 from (158) it follows that 119886
3=
[8cos31205910]minus1 This relation along with (160) and (157) further
implies that 1198793(120591) can be expressed as
1198793(120591) =
[cos 1205910minus cos 120591]2 [2 cos 120591
0+ cos 120591]
2cos31205910
(170)
providing that |1205910| le 1205873 From 119879
3(120591 + 120587) = 2 minus 119879
3(120591) it
follows that (170) also holds for 1198861ge 0 and minus1 le 119886
3le minus18
providing that 1205910isin [21205873 41205873]
Maximally flat nonnegative cosinewaveformof type (167)with 119886
1lt 0 (minimum at 120591
0= 0) reads 119879
3(120591) = [1 minus
cos 120591]2[1 + (12) cos 120591] Dually maximally flat nonnegativecosine waveform with 119886
1gt 0 (minimum at 120591
0= 120587) reads
1198793(120591) = [1 + cos 120591]2[1 minus (12) cos 120591]In what follows we provide relations between coefficients
1198861and 1198863of nonnegative cosine waveforms of type (167) with
at least one zeroFor 1198861le 0 conversion of (168) into an additive form
immediately leads to the following relation
1198861= minus1 minus 119886
3for minus 1 le 119886
3le1
8 (171)
Conversion of (170) into an additive form leads to 1198861
=
minus31198863(2 cos 2120591
0+ 1) which can be also expressed as 119886
1=
minus31198863(4cos2120591
0minus 1) For 119886
1le 0 relations |120591
0| le 1205873 119886
1=
minus31198863(4cos2120591
0minus 1) and 119886
3= [8cos3120591
0]minus1 lead to
1198861= minus3 [ 3radic119886
3minus 1198863] for 1
8le 1198863le 1 (172)
Similarly for 1198861ge 0 conversion of (169) into an additive form
leads to the following relation
1198861= 1 minus 119886
3for minus
1
8le 1198863le 1 (173)
Mathematical Problems in Engineering 21
1
15
15
05
0
minus05
minus15minus15
minus1
minus1 minus05 0 05 1
Coefficient a1
Coe
ffici
enta
3 a1= minus
3(3radica3minus a3)
a1 =
minus1 minus a3
a1 =
1 minusa3
a1 =3(3radic|a3
| +a3)
Figure 14 Parameter space of cosine waveforms for 119896 = 3
For waveform of type (170) with 1198861ge 0 relations 120591
0isin [21205873
41205873] 1198861= minus3119886
3(4cos2120591
0minus 1) and 119886
3= [8cos3120591
0]minus1 lead to
1198861= 3 [
3radic10038161003816100381610038161198863
1003816100381610038161003816 + 1198863] for minus 1 le 1198863le minus
1
8 (174)
Every cosine waveform of type (167) corresponds to apair of real numbers (119886
1 1198863) and vice versa Points (119886
1 1198863)
in grey area in Figure 14 correspond to nonnegative cosinewaveforms for 119896 = 3 The points at the boundary of grey areacorrespond to nonnegative cosinewaveformswith at least onezero A number of shapes of nonnegative cosine waveformswith 119896 = 3 and at least one zero plotted on interval [minus120587 120587]are also presented in Figure 14 The boundary of grey area inFigure 14 consists of four line segments described by relations(171)ndash(174) The common point of line segments (172) and(173) is cusp point with coordinates 119886
1= 0 and 119886
3= 1
Another cusp point with coordinates 1198861= 0 and 119886
3= minus1
is the common point of line segments (171) and (174) Thecommon point of line segments (171)-(172) has coordinates(minus98 18) and common point of line segments (173)-(174)has coordinates (98 minus18) These points are representedby white circle dots and they correspond to maximallyflat cosine waveforms (eg see [21]) White triangle dotswith coordinates (2radic3 minusradic39) and (minus2radic3radic39) refer tothe nonnegative cosine waveforms with maximum value ofamplitude of fundamental harmonic
7 Four Case Studies of Usage of NonnegativeWaveforms in PA Efficiency Analysis
In this section we provide four case studies of usage ofdescription of nonnegative waveforms with fundamental and119896th harmonic in PA efficiency analysis In first two casestudies to be presented in Section 71 voltage is nonnegativewaveform with fundamental and second harmonic with atleast one zero In remaining two case studies to be consideredin Section 72 voltage waveform contains fundamental andthird harmonic
i
v Load
Vdc
Idc
Lch
vL
iL
Cb
+ +in
Figure 15 Generic PA circuit diagram
Let us consider generic PA circuit diagram as shown inFigure 15 We assume here that voltage and current wave-forms at the transistor output are
V (120579) = 1 + 1198861V cos 120579 + 1198871V sin 120579 + 119886119896V cos 119896120579 + 119887119896V sin 119896120579
119894 (120579) = 1 + 1198861119894cos 120579 +
infin
sum
119899=2
119886119899119894cos 119899120579
(175)
where 120579 stands for 120596119905 Both waveforms are normalized in thesense that dc components of voltage and current are 119881dc =
1 and 119868dc = 1 respectively Under assumption that blockingcapacitor 119862
119887behaves as short-circuit at the fundamental and
higher harmonics current and voltage waveforms at the loadare
V119871(120579) = 119886
1V cos 120579 + 1198871V sin 120579 + 119886119896V cos 119896120579 + 119887119896V sin 119896120579
119894119871(120579) = minus119886
1119894cos 120579 minus
infin
sum
119899=2
119886119899119894cos 119899120579
(176)
In terms of coefficients of voltage and current waveforms theload impedance at fundamental harmonic is 119911
1= minus(119886
1V minus
1198951198871V)1198861119894 whereas load impedance at 119896th harmonic is 119911
119896=
minus(119886119896Vminus119895119887119896V)119886119896119894 All other harmonics are short-circuited (119911
119899=
0 for 119899 = 1 and 119899 = 119896) Time average output power of PA (egsee [10]) with waveform pair (175) at fundamental frequencycan be expressed as
1198751= minus
11988611198941198861V
2 (177)
For normalized waveforms (175) with 119881dc = 1 and 119868dc = 1dc power is 119875dc = 1 Consequently PA efficiency 120578 = 119875
1119875dc
(eg see [10 26]) is equal to
120578 = minus11988611198941198861V
2 (178)
Thus time average output power 1198751of PA with pair of nor-
malized waveform (175) is equal to efficiency (178)Power utilization factor (PUF) is defined [26] as ldquothe
ratio of power delivered in a given situation to the power
22 Mathematical Problems in Engineering
delivered by the same device with the same supply voltagein Class A moderdquo Since the output power in class-A modeis 1198751class-A = max[V(120579)] sdotmax[119894(120579)]8 (eg see [9]) it follows
that power utilization factor PUF = 11987511198751class-A for PA with
pair of normalized waveforms (175) can be expressed as
PUF =8120578
max [V (120579)] sdotmax [119894 (120579)] (179)
71 NonnegativeWaveforms for 119896 = 2 in PAEfficiencyAnalysisIn this subsection we provide two case studies of usage ofdescription of nonnegative waveforms with fundamental andsecond harmonic (119896 = 2) in PA efficiency analysis For moreexamples of usage of descriptions of nonnegative waveformswith fundamental and second harmonic in PA efficiencyanalysis see [12]
Case Study 71 In this case study we consider efficiency of PAfor given second harmonic impedance providing that voltageis nonnegative waveform with fundamental and second har-monic and current is ldquohalf-sinerdquo waveform frequently used inefficiency analysis of classical PA operation (eg see [10])
Standard model of current waveform for classical PAoperation has the form (eg see [10 26])
119894119863(120579) =
119868119863[cos 120579 minus cos(120572
2)] |120579| le
120572
2
0120572
2le |120579| le 120587
(180)
where 120572 is conduction angle and 119868119863
gt 0 Since 119894119863(120579) is
even function it immediately follows that its Fourier seriescontains only dc component and cosine terms
119894119863(120579) = 119868dc +
infin
sum
119899=1
119868119899cos 119899120579 (181)
The dc component of the waveform (180) is
119868dc =119868119863120572
2120587[sinc(120572
2) minus cos(120572
2)] (182)
where sinc119909 = (sin119909)119909 The coefficient of the fundamentalharmonic component reads
1198681=119868119863120572
2120587(1 minus sinc120572) (183)
and the coefficient of 119899th harmonic component can bewrittenin the form
119868119899=119868119863
119899120587[sin ((119899 minus 1) 1205722)
(119899 minus 1)minussin ((119899 + 1) 1205722)
(119899 + 1)] 119899 ge 2
(184)
For ldquohalf-sinerdquo current waveform conduction angle is equalto 120587 (class-B conduction angle) According to (182) thisfurther implies that 119868dc = 119868
119863120587 To obtain normalized form of
waveform (180) we set 119868dc = 1which implies that 119868119863= 120587 Fur-
thermore substitution of 120572 = 120587 and 119868119863= 120587 in (180) leads to
119894 (120579) =
120587 cos 120579 |120579| lt120587
2
0120587
2lt |120579| le 120587
(185)
Similarly substitution of 119868119863= 120587 and 120572 = 120587 into (183) and
(184) leads to the coefficients of waveform (185) Coefficientsof fundamental and second harmonic respectively are
1198861119894=120587
2 119886
2119894=2
3 (186)
On the other hand voltage waveform of type (35) for 119896 =2 reads
V (120579) = 1 + 1198861V cos 120579 + 1198871V sin 120579 + 1198862V cos 2120579 + 1198872V sin 2120579
(187)
This waveform contains only fundamental and second har-monic and therefore all harmonics of order higher thantwo are short-circuited (119911
119899= 0 for 119899 gt 2) For current
voltage pair (185) and (187) load impedance at fundamentalharmonic is 119911
1= minus(119886
1V minus 1198951198871V)1198861119894 whereas load impedance
at second harmonic is 1199112= minus(119886
2Vminus1198951198872V)1198862119894 According to ourassumption the load is passive and therefore Re119911
1 gt 0 and
Re1199112 ge 0 which further imply 119886
11198941198861V lt 0 and 119886
21198941198862V le 0
respectivelyIt is easy to see that problem of findingmaximal efficiency
of PA with current-voltage pair (185) and (187) for prescribedsecond harmonic impedance can be reduced to the problemof finding voltage waveform of type (187) with maximal coef-ficient |119886
1V| for prescribed coefficients of second harmonic(see Section 5)
The following algorithm (analogous to Algorithm 22presented in [12]) provides the procedure for calculation ofmaximal efficiency with current-voltage pair (185) and (187)for prescribed second harmonic impedance The definitionof function atan 2(119910 119909) which appears in the step (iii) of thefollowing algorithm is given by (105)
Algorithm 32 (i) Choose 1199112= 1199032+1198951199092such that |119911
2| le 1|119886
2119894|
(ii) calculate 1198862V minus 1198951198872V = minus119911
21198862119894and 120582
2V = radic11988622V + 1198872
2V(iii) if 2120582
2V le 1 minus 1198862V then calculate 119886
1V = minus1 minus 1198862V and
1198871V = minus2119887
2V else calculate 1205821V = radic81205822V(1 minus 1205822V) 1205790V minus 120585V2 =
(12)atan2(1198872V 1198862V) 1198861V = minus120582
1V cos(1205790V minus 120585V2) and 1198871V =
minus1205821V sin(1205790V minus 120585V2)(iv) calculate efficiency 120578 = minus119886
11198941198861V2
(v) calculate 1199111= minus(119886
1V minus 1198951198871V)1198861119894 and 1199112119899 = 1199112Re119911
1
In this case study coefficients of fundamental and secondharmonic of current waveform are given by (186) Maximalefficiency of PA associated with the waveform pair (185)and (187) as a function of normalized second harmonicimpedance 119911
2119899= 1199112Re119911
1 is presented in Figure 16(a)
As can be seen from Figure 16(a) efficiency of 078 isachieved at the edge of Smith chart where second harmonicimpedance has small resistive part Corresponding PUFcalculated according to (179) is presented in Figure 16(b)Peak efficiency 120578 = 1205874 = 07854 and peak value of PUF = 1
are attained when second harmonic is short-circuited (whichcorresponds to ideal class-B operation [10 26])
For example for second harmonic impedance 1199112= 01 minus
11989505 and currentwaveform (185) fromAlgorithm32 it followsthat 2120582
2V le 1 minus 1198862V Furthermore according to step (iii)
of above algorithm maximal efficiency of PA is attained
Mathematical Problems in Engineering 23
+j2
+j5
+j1
+j2
+j5
minusj2
minusj5
minusj1
minusj2
minusj5
infin0
078
02
05
1 2 5
075 07 065
06 05
075
07
065
06
05
120578 lt 05
(a)
099
095
091
083 075
067
0 59 051
051
099
095
091
083
075 0
67 059
+j5
+j1
+j2
+j5
minusj5
minusj1
minusj2
minusj5
infin
+j2
minusj2
0 02
05
1 2 5
120578 lt 05
(b)
Figure 16 (a) Contours of maximal efficiency of PA and (b) contours of corresponding PUF associated with the waveform pair (185) and(187) as functions of normalized second harmonic impedance 119911
2119899= 1199112Re119911
1
2
3
4
1
0
2 3 410
Wav
efor
ms
Angle 120579120587
i(120579)
(120579)
Figure 17 Waveform pair (185) and (187) that provides maximalefficiency for 119911
2= 01 minus 11989505
with voltage waveform of type (187) with coefficients 1198862V =
minus00667 1198872V = minus03333 119886
1V = minus09333 and 1198871V = 06667 (see
Figure 17) Corresponding efficiency PUF and normalizedsecond harmonic impedance are 120578 = 07330 PUF = 07572and 1199112119899= 01683 minus 11989508415 respectively
On the other hand for second harmonic impedance 1199112=
01 minus 11989508 and current waveform (185) from Algorithm 32 itfollows that 2120582
2V gt 1 minus 1198862V Then according to step (iii) of
above algorithm maximal efficiency is attained with voltagewaveform of type (187) with coefficients 119886
2V = minus00667 1198872V =
minus05333 1198861V = minus09333 and 119887
1V = 10572 (see Figure 18)Efficiency PUF andnormalized secondharmonic impedanceare 120578 = 07330 PUF = 06332 and 119911
2119899= 01683 minus 11989513465
respectively
2
3
4
1
0
2 3 410
Wav
efor
ms
Angle 120579120587
i(120579)
(120579)
Figure 18 Waveform pair (185) and (187) that provides maximalefficiency for 119911
2= 01 minus 11989508
Case Study 72 As another case study let us consider the effi-ciency of PA providing that current waveform is nonnegativecosine waveform up to third harmonic with maximum valueof amplitude of fundamental harmonic [22] (see also [8])
119894 (120579) = 1 +1 + radic5
2cos 120579 + 2radic5
5cos 2120579 + 5 minus radic5
10cos 3120579
(188)
and voltage waveform is nonnegative waveform of type(187) Load impedances at fundamental second and thirdharmonic are 119911
1= minus(119886
1Vminus1198951198871V)1198861119894 1199112 = minus(1198862Vminus1198951198872V)1198862119894 and
1199113= 0 respectively According to our assumption the load
is passive and therefore Re1199111 gt 0 and Re119911
2 ge 0 which
further imply 11988611198941198861V lt 0 and 119886
21198941198862V le 0 respectively
Because current waveform (188) contains only cosineterms and voltage waveform is the same as in previous case
24 Mathematical Problems in Engineering
08
075
07
06
06
065
065
05
05
075 0
7
120578 lt 05
+j2
+j5
+j1
+j2
+j5
minusj2
minusj5
minusj1
minusj2
minusj5
infin0
02
05
1 2 5
(a)
085
08 0
75 07 0
6
05 04
08
075
0405
060
7
120578 lt 05
+j2
+j5
+j1
+j2
+j5
minusj2
minusj5
minusj1
minusj2
minusj5
infin0 02
05
1 2 5
(b)
Figure 19 (a) Contours of maximal efficiency of PA and (b) contours of corresponding PUF associated with the waveform pair (187)-(188)as functions of normalized second harmonic impedance 119911
2119899= 1199112Re119911
1
study the procedure for calculation of maximal efficiency ofPA with waveform pair (187)-(188) is the same as presentedin Algorithm 32 In this case study the coefficients of funda-mental and second harmonic of current waveform are 119886
1119894=
(1 + radic5)2 and 1198862119894= 2radic55 respectively
Maximal efficiency of PA associated with the waveformpair (187)-(188) as a function of normalized secondharmonicimpedance 119911
2119899= 1199112Re119911
1 is presented in Figure 19(a)
Efficiency of 08 is achieved at the edge of Smith chart wheresecond harmonic impedance has small resistive part Thetheoretical upper bound 120578 = (1 + radic5)4 asymp 08090 isattained when second harmonic is short-circuitedWhen thisupper bound is reached both second and third harmonic areshort-circuited which implies that we are dealing with finiteharmonic class-C [6 8] or dually when current and voltageinterchange their roles with finite harmonic inverse class-C[6 9] Corresponding PUF calculated according to (179) ispresented in Figure 19(b) Peak value of PUF asymp 08541 isattained when second harmonic is short-circuited
For example for second harmonic impedance 1199112
=
007 minus 11989504 and current waveform (188) from Algorithm 32 itfollows that 2120582
2V le 1 minus 1198862V Furthermore according to step
(iii) of Algorithm 32 maximal efficiency of PA is attainedwith voltage waveform of type (187) with coefficients 119886
2V =
minus00626 1198872V = minus03578 119886
1V = minus09374 and 1198871V = 07155 (see
Figure 20) Corresponding efficiency PUF and normalizedsecond harmonic impedance are 120578 = 07584 PUF = 06337and 1199112119899= 01208 minus 11989506904 respectively
On the other hand for 1199112= 005 minus 11989507 and current
waveform (187) it follows that 21205822V gt 1minus119886
2VThen accordingto step (iii) of Algorithm 32 themaximal efficiency is attainedwith voltage waveform of type (187) with coefficients 119886
2V =
minus00447 1198872V = minus06261 119886
1V = minus09318 and 1198871V = 10007 (see
Figure 21) Efficiency PUF and normalized second harmonic
2
3
4
1
0
2 3 410
Wav
efor
ms
Angle 120579120587
i(120579)
(120579)
Figure 20 Waveform pair (187)-(188) that provides maximal effi-ciency for 119911
2= 007 minus 11989504
impedance are 120578 = 07538 PUF = 05314 and 1199112119899= 00868minus
11989512156 respectively
72 Nonnegative Waveforms for 119896 = 3 in PA EfficiencyAnalysis In this subsection we provide another two casestudies of usage of description of nonnegative waveforms inPA efficiency analysis this time with fundamental and thirdharmonic (119896 = 3)
Case Study 73 Let us consider current-voltage pair such thatvoltage is nonnegative waveform with fundamental and thirdharmonic
V (120579) = 1 + 1198861V cos 120579 + 1198871V sin 120579 + 1198863V cos 3120579 + 1198873V sin 3120579
(189)
Mathematical Problems in Engineering 25
2
3
4
1
0
2 3 410
Wav
efor
ms
Angle 120579120587
i(120579)
(120579)
Figure 21 Waveform pair (187)-(188) that provides maximal effi-ciency for 119911
2= 005 minus 11989507
and current is nonnegative cosine waveform given by (188)Load impedances at fundamental second and third har-monic are 119911
1= minus(119886
1V minus 1198951198871V)1198861119894 1199112 = 0 and 119911
3= minus(119886
3V minus
1198951198873V)1198863119894 respectively According to our assumption the load
is passive and therefore Re1199111 gt 0 and Re119911
3 ge 0 which
further imply 11988611198941198861V lt 0 and 119886
31198941198863V le 0
In this subsection we consider the problem of findingmaximal efficiency of PA with waveform pair (188)-(189)for given third harmonic impedance As we mentionedearlier problem of finding maximal efficiency of PA withcurrent-voltage pair (188)-(189) for prescribed third har-monic impedance can be reduced to the problem of findingvoltage waveform of type (189) withmaximal coefficient |119886
1V|for prescribed coefficients of third harmonic (see Section 52)
The following algorithm provides the procedure forcalculation of maximal efficiency with current-voltage pair(188)-(189) The definition of function atan 2(119910 119909) whichappears in step (iii) of the following algorithm is given by(105)
Algorithm 33 (i) Choose 1199113= 1199033+1198951199093such that |119911
3| le 1|119886
3119894|
(ii) calculate 1198863V minus 1198951198873V = minus119911
31198863119894and 120582
3V = radic11988623V + 1198872
3V(iii) if 271205822
3V le (1 minus 21198863V)3 then calculate 119886
1V = minus1 minus 1198863V
and 1198871V = minus3119887
3V else calculate 1205821V = 3(3radic1205823V minus 120582
3V) 1205790V minus120585V3 = (13)atan 2(119887
3V 1198863V) 1198861V = minus1205821V cos(1205790V minus 120585V3) and
1198871V = minus120582
1V sin(1205790V minus 120585V3)(iv) calculate efficiency 120578 = minus119886
11198941198861V2
(v) calculate 1199111= minus(119886
1V minus 1198951198871V)1198861119894 and 1199113119899 = 1199113Re119911
1
In this case study coefficients of fundamental and thirdharmonic of current waveform are 119886
1119894= (1 + radic5)2 and
1198863119894= (5 minus radic5)10 respectively For the waveform pair (188)-
(189) maximal efficiency of PA as a function of normalizedthird harmonic impedance 119911
3119899= 1199113Re119911
1 is presented in
Figure 22 Efficiency of 08 is reached when third harmonicimpedance has small resistive part Peak efficiency 120578 = (1 +
radic5)4 asymp 08090 is achieved when third harmonic is short-circuited
+j2
+j5
+j1
+j2
+j5
minusj2
minusj5
minusj1
minusj2
minusj5
infin0
08
08
075
075
07
07
06
05
02
05
1 2
Figure 22 Contours of maximal efficiency of PA associated withthe waveform pair (188)-(189) as a function of normalized thirdharmonic impedance 119911
3119899= 1199113Re119911
1
For the present case study in what follows we show thatpower utilization factor is proportional to efficiency For volt-age waveform of type (189) it is easy to see that V(120579 + 120587) =
2minusV(120579) holdsThis relation along with the fact that waveformV(120579) that provides maximal efficiency has at least one zeroimplies that max[V(120579)] = 2 On the other hand currentwaveform (188) is cosine waveform with positive coefficientsand therefore max[119894(120579)] = 119894(0) = 2 + 4radic5 Consequentlyaccording to (179) the following relation holds
PUFCase study 73 = 2 (5 minus 2radic5) 120578 = 10557120578 (190)
Clearly the ratio PUF120578 is constant and therefore in this casestudy PUF can be easily calculated from the correspondingefficiency Accordingly peak efficiency and peak value ofPUFCase study 73 = 3radic52 minus 52 = 08541 are attained forthe same voltage waveform (when third harmonic is short-circuited)
In the first example current waveform (188) and 1199113=
02 minus 11989505 imply that 2712058223V le (1 minus 2119886
3V)3 Then according
to Algorithm 33 the voltage waveform of type (189) thatprovides maximal efficiency has the following coefficients1198863V = minus00553 119887
3V = minus01382 1198861V = minus09447 and 119887
1V =
04146 (see Figure 23) Efficiency PUF and normalized thirdharmonic impedance are 120578 = 07643 PUF = 08069 and1199113119899= 03425 minus 11989508564 respectivelyIn the second example current waveform (188) and 119911
3=
01 minus 11989511 imply that 2712058223V gt (1 minus 2119886
3V)3 Then according
to Algorithm 33 the voltage waveform of type (189) thatprovides maximal efficiency has the following coefficients1198863V = minus00276 119887
3V = minus03040 1198861V = minus09391 and 119887
1V =
05807 (see Figure 24) Efficiency PUF and normalized thirdharmonic impedance are 120578 = 07598 PUF = 08021 and1199113119899= 01723 minus 11989518952 respectively
26 Mathematical Problems in Engineering
2
3
4
1
0
2 3 410
Wav
efor
ms
Angle 120579120587
i(120579)
(120579)
Figure 23 Waveform pair (188)-(189) that provides maximal effi-ciency for 119911
3= 02 minus 11989505
2
3
4
1
0
2 3 410
Wav
efor
ms
Angle 120579120587
i(120579)
(120579)
Figure 24 Waveform pair (188)-(189) that provides maximal effi-ciency for 119911
3= 01 minus 11989511
Case Study 74 In this case study let us consider current-voltage pair where current is normalized waveform of type(180) with conduction angle 120572 = 115120587 (207∘) and voltageis nonnegative waveform of type (189) Substitution of 120572 =
115120587 and 119868dc = 1 into (182) leads to 119868119863
= 22535Furthermore substitution of 120572 = 115120587 and 119868
119863= 22535 into
(180) leads to
119894 (120579)
=
22535 [cos 120579 minus cos(1151205872
)] |120579| le115120587
2
0115120587
2le |120579| le 120587
(191)
Similarly substitution of 120572 = 115120587 and 119868119863
= 22535 into(183) and (184) for 119899 = 3 yields coefficients of fundamentaland third harmonic of waveform (191)
1198861119894= 14586 119886
3119894= minus01026 (192)
Because current waveform (191) contains only cosineterms and voltage waveform is the same as in previous case
+j2
+j5
+j1
+j2
+j5
minusj2
minusj5
minusj1
minusj2
minusj5
infin0 02
05
1 2
074
076
08
084
082
078
076
074
Figure 25 Contours of maximal efficiency of PA associated withthe waveform pair (189) and (191) as a function of normalized thirdharmonic impedance 119911
3119899= 1199113Re119911
1
study the procedure for calculation of maximal efficiencyof PA with waveform pair (189)ndash(191) is the same as thatpresented in Algorithm 33 In this case study the coefficientsof fundamental and third harmonic of current waveform aregiven by (192)
For the waveform pair (189) and (191) maximal efficiencyof PA as a function of normalized third harmonic impedance1199113119899= 1199113Re119911
1 is presented in Figure 25 Efficiency of 084
is obtained in vicinity of 1199113119899= 23685 (corresponding to 119911
3=
18750) Peak efficiency 120578 asymp 08421 is achieved for voltagewaveform of type (189) with coefficients 119886
1V = minus2radic3 1198863V =
radic39 and 1198871V = 1198873V = 0
In the course of finding power utilization factor noticethat currentwaveformof type (191) attains itsmaximumvaluefor 120579 = 0 Insertion of max[119894(120579)] = 119894(0) = 278 andmax[V(120579)] = 2 for voltage waveform of type (189) into (179)leads to
PUFCase study 74 = 1439120578 (193)
Again the ratio PUF120578 is constant and PUF can be easilycalculated from the corresponding efficiency Accordinglypeak value of PUFCase study 74 asymp 12118 and peak efficiencyare attained for the same voltage waveform
In the first example current waveform (191) and 1199113=
1 minus 11989502 imply that 2712058223V le (1 minus 2119886
3V)3 Then according to
Algorithm 33 voltage waveform of type (189) which providesmaximal efficiency has coefficients 119886
3V = 01026 1198873V =
00205 1198861V = minus11026 and 119887
1V = minus00616 (see Figure 26)Efficiency PUF and normalized third harmonic impedanceare 120578 = 08042 PUF = 11572 and 119911
3119899= 13228 minus 11989502646
respectivelyIn second example current waveform (191) and 119911
3=
15 minus 11989512 imply that 2712058223V gt (1 minus 2119886
3V)3 Then according to
Algorithm 33 voltage waveform of type (189) which providesmaximal efficiency has coefficients 119886
3V = 01540 1198873V =
Mathematical Problems in Engineering 27
2
3
4
1
0
2 3 410
Wav
efor
ms
Angle 120579120587
i(120579)
(120579)
Figure 26 Waveform pair (189) and (191) that provides maximalefficiency for 119911
3= 1 minus 11989502
2
3
4
1
0
2 3 410
Wav
efor
ms
Angle 120579120587
i(120579)
(120579)
Figure 27 Waveform pair (189) and (191) that provides maximalefficiency for 119911
3= 15 minus 11989512
01232 1198861V = minus11255 and 119887
1V = minus02575 (see Figure 27)Efficiency PUF and normalized third harmonic impedanceare 120578 = 08208 PUF = 11812 and 119911
3119899= 19439 minus 11989515552
respectively
8 Conclusion
In this paper we consider a problem of finding general de-scriptions of various classes of nonnegative waveforms withfundamental and 119896th harmonic These classes include non-negative waveforms with at least one zero nonnegative wave-forms with maximal amplitude of fundamental harmonic forprescribed amplitude of 119896th harmonic nonnegative wave-forms withmaximal coefficient of cosine part of fundamentalharmonic for prescribed coefficients of 119896th harmonic andnonnegative cosine waveforms with at least one zero Mainresults are stated in six propositions (Propositions 1 6 918 22 and 26) four corollaries (Corollaries 2ndash5) twentyremarks and three algorithms Four case studies of usageof closed form descriptions of nonnegative waveforms in PAefficiency analysis are considered in detail in Section 7
Appendices
Here we provide a list of finite sums of trigonometric func-tions used in this paper (Appendix A) and brief account ofthe Chebyshev polynomials (Appendix B)
A List of Some Finite Sums ofTrigonometric Functions
Dirichlet kernel (eg see [27]) is as follows
119863119896minus1
(120591) = 1 + 2
119896minus1
sum
119899=1
cos 119899120591 = sin ((2119896 minus 1) 1205912)sin (1205912)
(A1)
Fejer kernel (eg see [27]) can be expressed in the fol-lowing equivalent forms
119865119896minus1
(120591) =1
119896
119896minus1
sum
119899=0
119863119899(120591) = 1 +
2
119896
119896minus1
sum
119899=1
(119896 minus 119899) cos 119899120591
=(1 minus cos 119896120591)119896 (1 minus cos 120591)
(A2)
Lagrangersquos trigonometric identity (eg see [28]) is as fol-lows
1198781(120591) =
119896minus1
sum
119899=1
sin 119899120591 = sin (1198961205912) sin ((119896 minus 1) 1205912)sin (1205912)
(A3)
In what follows we show that the following three trigono-metric identities also hold
2
119896minus1
sum
119899=1
(119896 minus 119899) sin 119899120591 = 119896 sin 120591 minus sin 1198961205911 minus cos 120591
(A4)
119896minus1
sum
119899=1
cos (119896 minus 2119899) 120591 = sin (119896 minus 1) 120591sin 120591
(A5)
119896minus1
sum
119899=1
119899 (119896 minus 119899) cos (119896 minus 2119899) 120591
=sin (119896120591) cos 120591 minus 119896 cos (119896120591) sin 120591
2sin3120591
(A6)
Denote 1198782(120591) = 2sum
119896minus1
119899=1(119896 minus 119899) sin 119899120591 119878
3(120591) = sum
119896minus1
119899=1cos(119896 minus
2119899)120591 and 1198784(120591) = sum
119896minus1
119899=1119899(119896 minus 119899) cos(119896 minus 2119899)120591
Notice that 1198782(120591) = 2119896119878
1(120591) + 119889119863
119896minus1(120591)119889120591 which
immediately leads to (A4)
28 Mathematical Problems in Engineering
Identity (A5) can be obtained as follows
sin (119896 minus 1) 120591sin 120591
=119890119895(119896minus1)120591
minus 119890minus119895(119896minus1)120591
119890119895120591 minus 119890minus119895120591
= 119890119895119896120591
119890minus2119895120591
minus 119890minus2119895(119896minus1)120591
1 minus 119890minus2119895120591
= 119890119895119896120591
119896minus1
sum
119899=1
119890minus2119895119899120591
=
119896minus1
sum
119899=1
119890119895(119896minus2119899)120591
=
119896minus1
sum
119899=1
cos (119896 minus 2119899) 120591
(A7)
From 4119899(119896 minus 119899) = 1198962
minus (119896 minus 2119899)2 it follows that 4119878
4(120591) =
1198962
1198783(120591) + 119889
2
1198783(120591)119889120591
2 which leads to (A6)
B The Chebyshev Polynomials
The Chebyshev polynomials of the first kind 119881119899(119909) can be
defined by the following relation (eg see [29])
119881119899(119909) = cos 119899120591 when 119909 = cos 120591 (B1)
The Chebyshev polynomials of the second kind 119880119899(119909) can be
defined by the following relation (eg see [29])
119880119899(119909) =
sin (119899 + 1) 120591sin 120591
when 119909 = cos 120591 (B2)
The Chebyshev polynomials satisfy the following recur-rence relations (eg see [29])
1198810(119909) = 1 119881
1(119909) = 119909
119881119899+1
(119909) = 2119909119881119899(119909) minus 119881
119899minus1(119909)
1198800(119909) = 1 119880
1(119909) = 2119909
119880119899+1
(119909) = 2119909119880119899(119909) minus 119880
119899minus1(119909)
(B3)
The first few Chebyshev polynomials of the first and secondkind are 119881
2(119909) = 2119909
2
minus 1 1198813(119909) = 4119909
3
minus 3119909 1198814(119909) = 8119909
4
minus
81199092
+ 1 1198802(119909) = 4119909
2
minus 1 1198803(119909) = 8119909
3
minus 4119909 and 1198804(119909) =
161199094
minus 121199092
+ 1
Conflict of Interests
The authors declare that there is no conflict of interestsregarding the publication of this paper
Acknowledgment
This work is supported by the SerbianMinistry of EducationScience and Technology Development as a part of ProjectTP32016
References
[1] V I Arnolrsquod V S Afrajmovich Y S Ilrsquoyashenko and L PShilrsquonikov Dynamical Systems V Bifurcation Theory and Catas-trophe Theory Springer Berlin Germany 1994
[2] E Polak ldquoOn the mathematical foundations of nondifferen-tiable optimization in engineering designrdquo SIAM Review vol29 no 1 pp 21ndash89 1987
[3] N S Fuzik ldquoBiharmonic modes of a tuned RF power amplifierrdquoRadiotehnika vol 25 no 7 pp 62ndash71 1970 (Russian)
[4] P Colantonio F Giannini G Leuzzi and E Limiti ldquoClass Gapproach for low-voltage high-efficiency PA designrdquo Interna-tional Journal of RF and Microwave Computer-Aided Engineer-ing vol 10 no 6 pp 366ndash378 2000
[5] F H Raab ldquoMaximum efficiency and output of class-F poweramplifiersrdquo IEEE Transactions on Microwave Theory and Tech-niques vol 49 no 6 pp 1162ndash1166 2001
[6] FH Raab ldquoClass-E class-C and class-F power amplifiers basedupon a finite number of harmonicsrdquo IEEE Transactions onMicrowaveTheory and Techniques vol 49 no 8 pp 1462ndash14682001
[7] J D Rhodes ldquoOutput universality inmaximum efficiency linearpower amplifiersrdquo International Journal of Circuit Theory andApplications vol 31 no 4 pp 385ndash405 2003
[8] A Juhas and L A Novak ldquoComments on lsquoClass-E class-Cand classF power amplifier based upon a finite number ofharmonicsrsquordquo IEEE Transactions on Microwave Theory and Tech-niques vol 57 no 6 pp 1623ndash1625 2009
[9] M Roberg and Z Popovic ldquoAnalysis of high-efficiency poweramplifiers with arbitrary output harmonic terminationsrdquo IEEETransactions on Microwave Theory and Techniques vol 59 no8 pp 2037ndash2048 2011
[10] A Grebennikov N O Sokal and M J Franco Switchmode RFPower Amplifiers ElsevierAcademic Press San Diego CalifUSA 2nd edition 2012
[11] T Canning P J Tasker and S C Cripps ldquoContinuous modepower amplifier design using harmonic clipping contourstheory and practicerdquo IEEE Transactions on Microwave Theoryand Techniques vol 62 no 1 pp 100ndash110 2014
[12] A Juhas and L A Novak ldquoGeneral description of nonnegativewaveforms up to second harmonic for power amplifier mod-ellingrdquoMathematical Problems in Engineering vol 2014 ArticleID 709762 18 pages 2014
[13] V I Arnolrsquod V V Goryunov O V Lyashko and V A VasilrsquoevDynamical Systems VIIImdashSingularity Theory II ApplicationsSpringer Berlin Germany 1993
[14] D Siersma ldquoProperties of conflict sets in the planerdquo BanachCenter Publications Polish Academy of Sciences vol 50 no 1 pp267ndash276 1999 Proceedings of the Banach Center Symposiumon Geometry and Topology of Caustics (Caustics rsquo98) WarsawPoland
[15] M van Manen The geometry of conflict sets [Dissertation]Universiteit Utrecht UtrechtThe Netherlands 2003 httpigi-tur-archivelibraryuunldissertations2003-0912-123058c4pdf
[16] Y L Sachkov ldquoMaxwell strata and symmetries in the problemofoptimal rolling of a sphere over a planerdquo Sbornik Mathematicsvol 201 no 7-8 pp 1029ndash1051 2010
[17] I A Bogaevsky ldquoPerestroikas of shocks and singularities ofminimum functionsrdquoPhysicaDNonlinear Phenomena vol 173no 1-2 pp 1ndash28 2002
[18] Y L Sachkov ldquoMaxwell strata in the Euler elastic problemrdquoJournal of Dynamical andControl Systems vol 14 no 2 pp 169ndash234 2008
[19] M Siino and T Koike ldquoTopological classification of black holesgeneric Maxwell set and crease set of a horizonrdquo InternationalJournal ofModern Physics D Gravitation Astrophysics Cosmol-ogy vol 20 no 6 pp 1095ndash1122 2011
Mathematical Problems in Engineering 29
[20] F H Raab ldquoClass-F power amplifiers with maximally flat wave-formsrdquo IEEETransactions onMicrowaveTheory and Techniquesvol 45 no 11 pp 2007ndash2012 1997
[21] A Juhas and L A Novak ldquoMaximally flat waveforms with finitenumber of harmonics in class-F power amplifiersrdquo Mathemat-ical Problems in Engineering vol 2013 Article ID 169590 9pages 2013
[22] L Fejer ldquoUber trigonometrische polynomerdquo Journal fur dieReine und Angewandte Mathematik vol 1916 no 146 pp 53ndash82 1916 (German)
[23] S C Cripps ldquoBessel Waives [microwave bytes]rdquo IEEE Micro-wave Magazine vol 10 no 7 pp 30ndash36 117 2009
[24] L N Bryzgalova ldquoSingularities of the maximum of para-metrically dependent functionrdquo Functional Analysis and ItsApplications vol 11 no 1 pp 49ndash51 1977
[25] V I Arnold A A Davydov V A Vassiliev and V MZakalyukin Mathematical Models of Catastrophes Control ofCatastrophic Process Encyclopedia of Life Support Systems(EOLSS) EOLSS Publishers Oxford UK 2006
[26] S C Cripps RF Power Amplifiers for Wireless CommunicationsArtech House Norwood Mass USA 2nd edition 2006
[27] A Zygmund Trigonometric Series vol 1 Cambridge UniversityPress Cambridge UK 2nd edition 1959
[28] A Jeffrey and H DaiHandbook of Mathematical Formulas andIntegrals ElsevierAcademic Press San Diego Calif USA 4thedition 2008
[29] J C Mason and D C Handscomb Chebyshev PolynomialsChapman amp Hall CRC Press Boca Raton Fla USA 2003
Submit your manuscripts athttpwwwhindawicom
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical Problems in Engineering
Hindawi Publishing Corporationhttpwwwhindawicom
Differential EquationsInternational Journal of
Volume 2014
Applied MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Probability and StatisticsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical PhysicsAdvances in
Complex AnalysisJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
OptimizationJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
CombinatoricsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Operations ResearchAdvances in
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Function Spaces
Abstract and Applied AnalysisHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of Mathematics and Mathematical Sciences
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Algebra
Discrete Dynamics in Nature and Society
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Decision SciencesAdvances in
Discrete MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom
Volume 2014 Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Stochastic AnalysisInternational Journal of
2 Mathematical Problems in Engineering
drain (collectorplate) waveforms in PA design (eg see [3ndash12 20 21])
Fejer in his seminal paper [22] provided general descrip-tion of all nonnegative trigonometric polynomials with 119899
consecutive harmonics in terms of 2119899 + 2 parameters satis-fying one nonlinear constraint He also derived closed formsolution to the problem of finding maximum possible ampli-tude of the first harmonic of nonnegative cosine polynomialswith consecutive harmonics
Fuzik [3] (see also [10]) considered cosine polynomialswith dc fundamental and 119896th harmonic for arbitrary 119896 ge 2
and provided closed form solution for coefficients of optimalwaveform Rhodes in [7] provided closed form expressionfor maximum possible amplitude of fundamental harmonicof nonnegative waveforms containing consecutive odd har-monics A subclass of nonnegative cosine waveforms withdc fundamental and third harmonic having factorized formdescription has been considered in [23]
High efficiency PA with arbitrary output harmonic ter-minations has been analysed in [9] along with maximal effi-ciency fundamental output power and load impedance
Factorized form of nonnegative waveforms up to secondharmonic with at least one zero has been suggested in [11] inthe context of continuous class BJ mode of PA operation
General description of all nonnegative waveforms up tosecond harmonic in terms of four independent parametershas been provided in [12] This includes nonnegative wave-forms with at least one zero as a special case
End point of conflict set normally corresponds to so-called maximally flat waveform which also belongs to classof suboptimal waveforms First comprehensive usage of max-imally flat waveforms in the context of analysis of PA goesto Raab [20] General description of maximally flat wave-forms with arbitrary number of harmonics has been pre-sented in [21] along with closed form expressions for effi-ciency of class-F and inverse class-F PA with maximally flatwaveforms Description of maximally flat cosine waveformswith consecutive harmonics has been presented in [8] in thecontext of finite harmonic class-C PA
In this paper we provide general descriptions of a num-ber of optimal and suboptimal nonnegative waveforms con-taining dc component fundamental and an arbitrary 119896thharmonic 119896 ge 2 and show how they are related to theconcept of conflict set According to our best knowledge thispaper provides the very first usage of conflict set in the courseof solving problems related to optimization of PA efficiencyMain results are stated in six propositions (Propositions 1 69 18 22 and 26) four corollaries (Corollaries 2ndash5) twentyremarks and three algorithms Four case studies of usage ofclosed form descriptions of nonnegative waveforms in PAefficiency analysis are considered in detail in Section 7
This paper is organized in the following way In Section 2we introduce concepts of minimum function and gain func-tion (Section 21) conflict set (Section 22) and parameterspace (Section 23) In Sections 3ndash6 we provide generaldescriptions of various classes of nonnegative waveformscontaining dc component fundamental and 119896th harmonicwith at least one zero General case of nonnegative waveformswith at least one zero is presented in Section 31The case with
exactly two zeros is considered in Section 32 An algorithmfor calculation of coefficients of fundamental harmonic ofnonnegative waveforms with two zeros for prescribed coeffi-cients of 119896th harmonic is presented in Section 33 Descrip-tion of nonnegative waveforms with maximal amplitude offundamental harmonic for prescribed amplitude of 119896th har-monic is provided in Section 4 Nonnegative waveforms withmaximal coefficient of cosine part of fundamental harmonicfor prescribed coefficients of 119896th harmonic are consideredin Section 51 An illustration of results of Section 51 forparticular case 119896 = 3 is given in Section 52 Section 61 isdevoted to nonnegative cosine waveforms with at least onezero and arbitrary 119896 ge 2 whereas Section 62 considerscosine waveforms with at least one zero for 119896 = 3 InSection 7 four case studies of application of descriptions ofnonnegative waveforms with fundamental and 119896th harmonicin PA modelling are presented In the Appendices list ofsome finite sums of trigonometric functions widely usedthroughout the paper and brief account of the Chebyshevpolynomials are provided
2 Minimum Function Gain Function andConflict Set
In this sectionwe considerminimum function and gain func-tion (Section 21) conflict set (Section 22) and parameterspace (Section 23) in the context of nonnegative waveformswith fundamental and 119896th harmonic
We start with provision of a brief account of the factsrelated to the concepts of minimum function and conflict setFor this purpose let us denote by 119891(119909 119906) a family of smoothfunctions of 119899 variables depending on 119898 parameters where119909 isin 119877
119899 is 119899-tuple of variables and 119906 isin 119877119898 is 119898-tuple of
parameters The minimum function 119865 119877119898
rarr 119877 associatedwith the function 119891 is defined as 119865(119906) = min
119909119891(119909 119906)
Therefore the domain of theminimum function is parameterspace of the function 119891 The minimum function 119865(119906) is con-tinuous but not necessarily smooth function of parameters[13 24] It is a smooth function if 119891(119909 119906) possesses uniqueglobal minimum at nondegenerate critical point [13] (criticalpoint is degenerate if at least first two consecutive derivativesare equal to zero) In this context the conflict set can bedefined as the set of the parameters for which function 119891
has global minimum at a degenerate critical point orandmultiple global minima [13]
For a wide class ofminimum functions when the numberof parameters is not greater than four the behaviour ofminimum function in a neighbourhood of any point can bedescribed by one of ldquonormal formsrdquo from a finite list as statedin [24] For example for smooth function 119891 119877 times 119877
2
rarr 119877the minimum function 119865(119906
1 1199062) = min
119909119891(119909 119906
1 1199062) near the
origin can be locally reduced to one of the following threenormal forms [25] minus|119906
1|min(119906
1 1199062 1199061+ 1199062) or min
119909(1199094
+
11990611199092
+ 1199062119909) In this example the conflict set is the set of all
points (1199061 1199062) for which minimum function 119865(119906
1 1199062) is not
differentiable because function 119891(119909 1199061 1199062) possesses at least
two global minima [25]
Mathematical Problems in Engineering 3
21 Minimum Function and Gain Function In what followswe consider family of waveforms of type
119908 (120591 120574 119860 120572) = 1 minus 120574 (cos 120591 + 119860 cos (119896120591 + 120572)) (1)
where 120591 stands for 120596119905 120574 gt 0 119896 ge 2 119860 gt 0 and 120572 isin [0 2120587)Waveforms of type (1) include all possible shapes which canoccur but not all possiblewaveforms containing fundamentaland 119896th harmonic However shifting of waveforms of type(1) along 120591-axis could recover all possible waveforms withfundamental and 119896th harmonic
The problem of finding nonnegative waveform of type(1) having maximum amplitude of fundamental harmonicplays an important role in optimization of PA efficiency Thisextremal problem can be reformulated as problem of findingnonnegative waveform from family (1) having maximumpossible value of coefficient 120574 Nonnegative waveform offamily (1) with maximum possible value of coefficient 120574 iscalled ldquooptimalrdquo or ldquoextremalrdquo waveform
Furthermore let us introduce an auxiliary waveform
119891 (120591 119860 120572) = minus cos 120591 minus 119860 cos (119896120591 + 120572) (2)
which is smooth function of one variable 120591 and two parame-ters119860 and 120572 In terms of 119891(120591 119860 120572) the above extremal prob-lem reduces to the problem of finding maximum possiblevalue of coefficient 120574 that satisfies
1 + 120574119891 (120591 119860 120572) ge 0 (3)
Clearly for any prescribed pair (119860 120572) there is a uniquemaximal value of coefficient 120574 for which inequality (3) holdsfor all 120591 This maximal value of 120574 associated with the pair(119860 120572) we denote it by 119866(119860 120572) and call it ldquogain functionrdquo
Let
119865min (119860 120572) = min120591
119891 (120591 119860 120572) (4)
be theminimum function associatedwith119891(120591 119860 120572) Accord-ing to (3) 119866(119860 120572) and 119865min(119860 120572) satisfy the following rela-tion 1 + 119866(119860 120572)119865min(119860 120572) = 0 Since 119865min(119860 120572) is obviouslynonzero it follows immediately that
119866 (119860 120572) = minus1
119865min (119860 120572) (5)
A relation analogue to (5) for 119896 = 2 (fundamental andsecond harmonic) has been derived in [4] According toour best knowledge it was the first appearance of gainfunction expressed via associated minimum function Theconsideration presented in [4] has been restricted to theparticular case when 120572 = 120587 The same problem for 120572 = 120587 andarbitrary 119896 ge 2 has been investigated in [3] (see also [10])
According to above consideration the problem of finding3-tuple (120574 119860 120572) with maximum possible value of 120574 for which(3) holds is equivalent to the problem of finding maximumvalue of gain function
120574max = max119860120572
119866 (119860 120572) = minus1
max119860120572
119865min (119860 120572)(6)
14
12
1
08
06
04
G1 G2
001
02
0304
0506 0
1205872
12058731205872
2120587
Parameter 120572
Parameter A
Gai
n fu
nctio
n
k = 2
Figure 1 Graph of 119866(119860 120572) for 119896 = 2 Points 1198661and 119866
2denote
beginning of the ridge and maximum of gain function respectively
and corresponding pair (119860lowast 120572lowast) that satisfies
120574max = max119860120572
119866 (119860 120572) = 119866 (119860lowast
120572lowast
)
max119860120572
119865min (119860 120572) = 119865min (119860lowast
120572lowast
)
(7)
Thus the optimal waveform 119908lowast
(120591) is determined by parame-ters 120574max 119860
lowast and 120572lowast that is
119908lowast
(120591) = 119908 (120591 120574max 119860lowast
120572lowast
) (8)
Optimal waveform has two global minima (this claim willbe justified in Section 4 Remark 21) Consequently the pair(119860lowast
120572lowast
) which corresponds to maximum of gain function119866(119860 120572) belongs to conflict set in (119860 120572) parameter space
Figure 1 shows graph of gain function 119866(119860 120572) for 119896 = 2Notice that it has sharp ridge and that maximum of gainfunction (point 119866
2) lies on the ridge This maximum corre-
sponds to the optimal waveform (solution of the consideredextremal problem) The beginning of the ridge (point 119866
1)
corresponds to the waveform which possesses global mini-mum at degenerate critical point that is corresponds tomax-imally flat waveform (eg see [21]) Gain function 119866(119860 120572)
is not differentiable on the ridge and consequently is notdifferentiable at the point where it has global maximumThisexplains why the approach based on critical points does notwork and why conflict set is so important in the consideredproblem
Positions of global minima of 119891(120591 119860 120572) for 119896 = 2 arepresented in Figure 2 According to Proposition 1 conflict setis the ray defined by119860 gt 14 and120572 = 120587Waveforms119891(120591 119860 120572)with parameters that belong to the conflict set have two globalminimaThewaveform corresponding to the end point of theray (119860 = 14 and 120572 = 120587) has global minimum at degeneratecritical point (so-called maximally flat waveform [21])
Nonnegative waveforms of type (1) with 120574 = 119866(119860 120572) =
minus1119865min(119860 120572) have at least one zero To show that it issufficient to see that 119908(120591 120574 119860 120572) = 0 for 120591 satisfying 119891(120591119860 120572) = 119865min(119860 120572)
4 Mathematical Problems in Engineering
001
02
0304
05
06 01205872
12058731205872
2120587
Parameter 120572
Parameter A
1205872
1205874
0
minus1205874
minus1205872Posit
ions
of g
loba
l min
k = 2
Figure 2 Positions of global minima of 119891(120591 119860 120572) for 119896 = 2
The problem of finding maximum value of fundamentalharmonic cosine part of nonnegative waveform of the form
119908119886(120591 120574119886 119887 119860 120572) = 1 minus 120574
119886(cos 120591 + 119887 sin 120591 + 119860 cos (119896120591 + 120572))
(9)
where 120574119886gt 0 is also related to the problem of finding max-
imumof theminimum functionOptimalwaveformof family(9) has two global minima (this claim will be justified inSection 5 Remark 25) and therefore corresponding 3-tupleof parameters belongs to the conflict set in parameter spaceof family (9)
Let us introduce an auxiliary waveform
119891119886(120591 119887 119860 120572) = minus cos 120591 minus 119887 sin 120591 minus 119860 cos (119896120591 + 120572) (10)
and corresponding minimum function 119865119886min(119887 119860 120572) =
min120591119891119886(120591 119887 119860 120572) Inequality 119908
119886(120591 120574119886 119887 119860 120572) ge 0 can be
rewritten as 1 + 120574119886119865119886min(119887 119860 120572) ge 0 and therefore the
highest value of 120574119886is attained for 120574
119886= minus1119865
119886min(119887 119860 120572) Itimmediately follows that nonnegative waveform of type (9)with 120574
119886= minus1119865
119886min(119887 119860 120572) has zero for 120591 satisfying 119891119886(120591
119887 119860 120572) = 119865119886min(119887 119860 120572)
22 Conflict Set Historically conflict set came into beingfrom the problems in which families of smooth functions(such as potentials distances and waveforms) with two com-peting minima occurThe situation when competing minimabecome equal refers to the presence of conflict set (Maxwellset Maxwell strata) in the associated parameter space
There are many facets of conflict set For example in theproblem involving distances between two sets of points theconflict set is the intersections between iso-distance lines[14] Conflict set also arises in the situation when two wavefronts coming from different objects meet [15 25] In thestudy of black holes conflict set is the line of crossover of thehorizon formed by the merger of two black holes [19] In theclassical Euler problem conflict set is a set of points wheredistinct extremal trajectories with the same value of the costfunctional meet one another [18]
Conflict set is very difficult to calculate both analyticallyand numerically (eg see [15]) because of apparent nondif-ferentiability in some directions In optimization of PA effi-ciency some authors already reported difficulties in findingoptimum via standard analytical tools [4 5]
In this section we consider conflict set in the context offamily of waveforms of type (2) for arbitrary 119896 ge 2 In thiscontext for prescribed integer 119896 ge 2 conflict set is said to bea set of all pairs (119860 120572) for which 119891(120591 119860 120572) possesses multipleglobal minima
Suppose that 1205911015840 and 12059110158401015840 are the positions of global min-
ima of 119891(120591 119860 120572) Then the conflict set is specified by the fol-lowing set of relations
119891 (1205911015840
119860 120572) = 119891 (12059110158401015840
119860 120572) (11)
1198911015840
(1205911015840
119860 120572) = 0 1198911015840
(12059110158401015840
119860 120572) = 0 (12)
11989110158401015840
(1205911015840
119860 120572) gt 0 11989110158401015840
(12059110158401015840
119860 120572) gt 0 (13)
(forall120591) 119891 (120591 119860 120572) ge 119891 (1205911015840
119860 120572) (14)
Relations (12) and (13) say that 119891(120591 119860 120572) has minima at 1205911015840and 12059110158401015840 while relations (11) and (14) imply that these minimaare equal and global
The following proposition describes the conflict set offamily of waveforms of type (2)
Proposition 1 Conflict set of family of waveforms of type (2)is the set of all pairs (119860 120572) such that 119860 gt 1119896
2 and 120572 = 120587
The proof of Proposition 1 which is provided at the endof this section also implies that the following four corollarieshold
Corollary 2 The conflict set has end point at (119860 120572) =
(11198962
120587) This end point corresponds to the maximally flatwaveform [21]
Corollary 3 Waveforms of type (2) with parameters thatbelong to conflict set have two global minima at plusmn120591
Δ where
0 lt 120591Δle 120587119896
Corollary 4 Every waveform with fundamental and 119896th har-monic has either one or two global minima
Corollary 5 Conflict set can be parameterised in terms of 120591Δ
as follows
120572 = 120587 119860 (120591Δ) =
sin 120591Δ
119896 sin 119896120591Δ
0 lt 120591Δle120587
119896 (15)
Notice that119860(120591Δ) ismonotonically increasing function on inter-
val 0 lt 120591Δle 120587119896
Proof of Proposition 1 Without loss of generality we canrestrict our consideration to the interval minus120587 lt 120591 le 120587 This isan immediate consequence of the fact that 119891(120591 119860 120572) is aperiodic function
Suppose that 1205911015840 and 12059110158401015840 where 1205911015840 lt 12059110158401015840 are points at which
119891(120591 119860 120572) has two equal global minima Then conflict set is
Mathematical Problems in Engineering 5
specified by relations (11)ndash(14) From (11)ndash(13) it follows thatrelations
119891 (1205911015840
119860 120572) minus 119891 (12059110158401015840
119860 120572) = 0
1198911015840
(1205911015840
119860 120572) + 1198911015840
(12059110158401015840
119860 120572) = 0
1198911015840
(1205911015840
119860 120572) minus 1198911015840
(12059110158401015840
119860 120572) = 0
11989110158401015840
(1205911015840
119860 120572) + 11989110158401015840
(12059110158401015840
119860 120572) gt 0
(16)
also hold Let
120591119904119903=(1205911015840
+ 12059110158401015840
)
2 120591
Δ=(12059110158401015840
minus 1205911015840
)
2
(17)
be a pair of points associated with (1205911015840 12059110158401015840) Clearly
minus120587 lt 120591119904119903lt 120587 (18)
0 lt 120591Δlt 120587 (19)
12059110158401015840
= 120591119904119903+ 120591Δ 120591
1015840
= 120591119904119903minus 120591Δ (20)
The first and second derivatives of 119891(120591 119860 120572) are equal to
1198911015840
(120591 119860 120572) = sin 120591 + 119896119860 sin (119896120591 + 120572)
11989110158401015840
(120591 119860 120572) = cos 120591 + 1198962119860 cos (119896120591 + 120572) (21)
By using (20)-(21) system (16) can be rewritten as
sin 120591119904119903sin 120591Δ+ 119860 sin (119896120591
119904119903+ 120572) sin 119896120591
Δ= 0 (22)
sin 120591119904119903cos 120591Δ+ 119896119860 sin (119896120591
119904119903+ 120572) cos 119896120591
Δ= 0 (23)
cos 120591119904119903sin 120591Δ+ 119896119860 cos (119896120591
119904119903+ 120572) sin 119896120591
Δ= 0 (24)
cos 120591119904119903cos 120591Δ+ 1198962
119860 cos (119896120591119904119903+ 120572) cos 119896120591
Δgt 0 (25)
From (19) it follows that sin 120591Δgt 0 Multiplying (24) and
(25) with minus cos 120591Δand sin 120591
Δgt 0 respectively and sum-
ming the resulting relations we obtain 119896119860 cos(119896120591119904119903
+
120572)[119896 sin 120591Δcos 119896120591
Δminus sin 119896120591
Δcos 120591Δ] gt 0 The latest relation
immediately implies that
119896 sin 120591Δcos 119896120591
Δminus sin 119896120591
Δcos 120591Δ
= 0 (26)
Equations (22) and (23) can be considered as a system oftwo linear equations in terms of sin 120591
119904119903and 119860 sin(119896120591
119904119903+ 120572)
According to (26) the determinant of this system is nonzeroand therefore it has only trivial solution
sin 120591119904119903= 0 sin (119896120591
119904119903+ 120572) = 0 (27)
According to (18) sin 120591119904119903= 0 implies
120591119904119903= 0 (28)
According to (20) 120591119904119903= 0 implies
12059110158401015840
= 120591Δ 120591
1015840
= minus120591Δ (29)
Furthermore 120591119904119903= 0 and sin(119896120591
119904119903+ 120572) = 0 imply that sin120572 =
0 From (29) it follows that 120591Δis position of global minimum
of 119891(120591 119860 120572) Clearly 119891(120591Δ 119860 120572) le 119891(0 119860 120572) which together
with sin120572 = 0 leads to
1 minus cos 120591Δ+ 119860 cos120572 (1 minus cos 119896120591
Δ) le 0 (30)
From 120591Δ
= 0 (see (19)) 119860 gt 0 and (30) it follows that cos 120572 lt
0 which together with sin120572 = 0 yields
120572 = 120587 (31)
Since 120591Δis position of global minimum it follows that
119891(120591Δ 119860 120587) le 119891(120587119896 119860 120587) Accordingly 119860(1 + cos 119896120591
Δ) le
cos 120591Δminus cos120587119896 which together with 119860 gt 0 implies that
cos 120591Δminus cos120587119896 ge 0 This relation along with (19) yields
0 lt 120591Δle120587
119896 (32)
Substitution of (31) and (28) in (24) leads to
119860 =sin 120591Δ
119896 sin 119896120591Δ
(33)
Notice that sin 119896120591Δ sin 120591
Δis monotonically decreasing func-
tion on interval (32)Therefore parameter119860 is monotonicallyincreasing function on the same interval with lim
120591Δrarr0+119860 =
11198962 Consequently 119860 gt 1119896
2 which completes theproof
23 Parameter Space In parameter space of family of wave-forms (2) there are two subsets playing important role in theclassification of the family instancesThese are conflict set andcatastrophe set
Catastrophe set is subset of parameter space of waveform119891(120591 119860 120572) It consists of those pairs (119860 120572) for which thecorresponding waveforms 119891(120591 119860 120572) have degenerate criticalpoints at which first and second derivatives are equal to zeroThus for finding catastrophe set we have to consider thefollowing system of equations
1198911015840
(120591119889 119860 120572) = 0
11989110158401015840
(120591119889 119860 120572) = 0
(34)
where 120591119889is a degenerate critical point of waveform119891(120591 119860 120572)
Conflict set in parameter space of waveform119891(120591 119860 120572) asshown in Proposition 1 is the ray described by 119860 gt 1119896
2 and120572 = 120587 It is intimately connected to catastrophe set
In what follows in this subsection we use polar coordinatesystem (119860 cos120572 119860 sin120572) instead of Cartesian coordinatesystem (119860 120572) Examples of catastrophe set and conflict setfor 119896 le 5 plotted in parameter space (119860 cos120572 119860 sin120572) arepresented in Figure 3 Solid line represents the catastropheset while dotted line describes conflict set The isolated pickpoints (usually called cusp) which appear in catastrophecurves correspond to maximally flat waveforms with max-imally flat minimum andor maximally flat maximumThereare two such picks in the catastrophe curves for 119896 = 2 and
6 Mathematical Problems in Engineering
00
00
k = 2 k = 3
k = 4 k = 5
Acos 120572
Acos 120572
Acos 120572
Acos 120572
A sin 120572
A sin 120572A sin 120572
A sin 120572
Figure 3 Catastrophe set (solid line) and corresponding conflict set(dotted line) for 119896 le 5 In each plot white triangle dot correspondsto optimal waveform and white circle dot corresponds to maximallyflat waveform
119896 = 4 and one in the catastrophe curves for 119896 = 3 and 119896 = 5Notice that the end point of conflict set is the cusp point
Catastrophe set divides the parameter space (119860 cos120572119860 sin120572) into disjoint subsets In the cases 119896 = 2 and 119896 =
3 catastrophe curve defines inner and outer part For 119896 gt
3 catastrophe curve makes partition of parameter space inseveral inner subsets and one outer subset (see Figure 3)
Notice also that multiplying 119891(120591 119860 120572) with a positiveconstant and adding in turn another constant which leads towaveform of type 119908(120591 120574 119860 120572) (see (1) and (2)) do not makeimpact on the character of catastrophe and conflict sets Thisis because in the course of finding catastrophe set first andsecond derivatives of 119891(120591 119860 120572) are set to zero Clearly (34) interms of 119891(120591 119860 120572) are equivalent to the analogous equationsin terms of119908(120591 120574 119860 120572) Analogously in the course of findingconflict set we consider only the positions of global minima(these positions forwaveforms119891(120591 119860 120572) and119908(120591 120574 119860 120572) arethe same)
3 Nonnegative Waveforms with atLeast One Zero
In what follows let us consider a waveform containing dccomponent fundamental and 119896th (119896 ge 2) harmonic of theform
119879119896(120591) = 1 + 119886
1cos 120591 + 119887
1sin 120591 + 119886
119896cos 119896120591 + 119887
119896sin 119896120591 (35)
The amplitudes of fundamental and 119896th harmonic of wave-form of type (35) respectively are
1205821= radic11988621+ 11988721 (36)
120582119896= radic1198862119896+ 1198872119896 (37)
As it is shown in Section 21 nonnegative waveforms withmaximal amplitude of fundamental harmonic or maximalcoefficient of fundamental harmonic cosine part have atleast one zero It is also shown in Section 22 (Corollary 4)that waveforms of type (35) with nonzero amplitude offundamental harmonic have either one or two globalminimaConsequently if nonnegative waveform of type (35) withnonzero amplitude of fundamental harmonic has at least onezero then it has at most two zeros
In Section 31 we provide general description of nonnega-tive waveforms of type (35) with at least one zero In Sections32 and 33 we consider nonnegative waveforms of type (35)with two zeros
31 General Description of Nonnegative Waveforms with atLeast One Zero The main result of this section is presentedin the following proposition
Proposition 6 Every nonnegative waveform of type (35) withat least one zero can be expressed in the following form
119879119896(120591) = [1 minus cos (120591 minus 120591
0)] [1 minus 120582
119896119903119896(120591)] (38)
where119903119896(120591) = (119896 minus 1) cos 120585
+ 2
119896minus1
sum
119899=1
(119896 minus 119899) cos (119899 (120591 minus 1205910) + 120585)
(39)
providing that
120582119896le [(119896 minus 1) cos 120585 + 119896 sin(120585 minus 120585119896)
sin(120585119896)]
minus1
(40)
10038161003816100381610038161205851003816100381610038161003816 le 120587 (41)
Remark 7 Function on the right hand side of (40) is mono-tonically increasing function of |120585| on interval |120585| le 120587 (formore details about this function see Remark 15) From (57)and (65) it follows that relation
0 le 120582119896le 1 (42)
holds for every nonnegative waveform of type (35) Noticethat according to (40) 120582
119896= 1 implies |120585| = 120587 Substitution
of 120582119896
= 1 and |120585| = 120587 into (55) yields 119879119896(120591) = 1 minus
cos 119896(120591 minus 1205910) Consequently 120582
119896= 1 implies that amplitude
1205821of fundamental harmonic is equal to zero
Remark 8 Conversion of (38) into additive form leads to thefollowing expressions for coefficients of nonnegative wave-forms of type (35) with at least one zero
1198861= minus (1 + 120582
119896cos 120585) cos 120591
0minus 119896120582119896sin 120585 sin 120591
0 (43)
1198871= minus (1 + 120582
119896cos 120585) sin 120591
0+ 119896120582119896sin 120585 cos 120591
0 (44)
119886119896= 120582119896cos (119896120591
0minus 120585) (45)
119887119896= 120582119896sin (119896120591
0minus 120585) (46)
providing that 120582119896satisfy (40) and |120585| le 120587
Mathematical Problems in Engineering 7
2
1
0
0 1 2 3 4
Angle 120591120587
Wav
efor
ms
1205823 = radic2201205823 = radic281205823 = radic24
Figure 4 Nonnegative waveforms with at least one zero for 119896 = 31205910= 1205876 and 120585 = 31205874
Three examples of nonnegative waveforms with at leastone zero for 119896 = 3 are presented in Figure 4 (examples ofnonnegative waveformswith at least one zero for 119896 = 2 can befound in [12]) For all three waveforms presented in Figure 4we assume that 120591
0= 1205876 and 120585 = 31205874 From (40) it follows
that 1205823le radic24 Coefficients of waveform with 120582
3= radic220
(dotted line) are 1198861= minus08977 119887
1= minus03451 119886
3= 005
and 1198873= minus005 Coefficients of waveform with 120582
3= radic28
(dashed line) are 1198861= minus09453 119887
1= minus01127 119886
3= 0125
and 1198873= minus0125 Coefficients of waveform with 120582
3= radic24
(solid line) are 1198861= minus10245 119887
1= 02745 119886
3= 025 and
1198873= minus025 First two waveforms have one zero while third
waveform (presented with solid line) has two zeros
Proof of Proposition 6 Waveform of type (35) containingdc component fundamental and 119896th harmonic can be alsoexpressed in the form
119879119896(120591) = 1 + 120582
1cos (120591 + 120593
1) + 120582119896cos (119896120591 + 120593
119896) (47)
where1205821ge 0120582
119896ge 0120593
1isin (minus120587 120587] and120593
119896isin (minus120587 120587] It is easy
to see that relations between coefficient of (35) and param-eters of (47) read as follows
1198861= 1205821cos1205931 119887
1= minus1205821sin1205931 (48)
119886119896= 120582119896cos120593119896 119887
119896= minus120582119896sin120593119896 (49)
Let us introduce 120585 such that10038161003816100381610038161205851003816100381610038161003816 le 120587 120585 = (119896120591
0+ 120593119896) mod2120587 (50)
Using (50) coefficients (49) can be expressed as (45)-(46)Let us assume that 119879
119896(120591) is nonnegative waveform of type
(35) with at least one zero that is 119879119896(120591) ge 0 and 119879
119896(1205910) = 0
for some 1205910 Notice that conditions 119879
119896(120591) ge 0 and 119879
119896(1205910) = 0
imply that 1198791015840119896(1205910) = 0 From 119879
119896(1205910) = 0 and 1198791015840
119896(1205910) = 0 by
using (50) it follows that
1205821cos (120591
0+ 1205931) = minus (1 + 120582
119896cos 120585)
1205821sin (1205910+ 1205931) = minus119896120582
119896sin 120585
(51)
respectively On the other hand 1205821cos(120591 + 120593
1) can be rewrit-
ten as
1205821cos (120591 + 120593
1) = 1205821cos (120591
0+ 1205931) cos (120591 minus 120591
0)
minus 1205821sin (1205910+ 1205931) sin (120591 minus 120591
0)
(52)
Substitution of (51) into (52) yields
1205821cos (120591 + 120593
1) = minus (1 + 120582
119896cos 120585) cos (120591 minus 120591
0)
+ 119896120582119896sin 120585 sin (120591 minus 120591
0)
(53)
According to (50) it follows that cos(119896120591 + 120593119896) = cos(119896(120591 minus
1205910) + 120585) that is
cos (119896120591 + 120593119896) = cos 120585 cos 119896 (120591 minus 120591
0) minus sin 120585 sin 119896 (120591 minus 120591
0)
(54)
Furthermore substitution of (54) and (53) into (47) leads to
119879119896(120591) = [1 minus cos (120591 minus 120591
0)] [1 + 120582
119896cos 120585]
minus 120582119896[1 minus cos 119896 (120591 minus 120591
0)] cos 120585
+ 120582119896[119896 sin (120591 minus 120591
0) minus sin 119896 (120591 minus 120591
0)] sin 120585
(55)
According to (A2) and (A4) (see Appendices) there is com-mon factor [1minuscos(120591minus120591
0)] for all terms in (55) Consequently
(55) can be written in the form (38) where
119903119896(120591) = minus cos 120585 + [
1 minus cos 119896 (120591 minus 1205910)
1 minus cos (120591 minus 1205910)] cos 120585
minus [119896 sin (120591 minus 120591
0) minus sin 119896 (120591 minus 120591
0)
1 minus cos (120591 minus 1205910)
] sin 120585
(56)
From (56) by using (A2) (A4) and cos 120585 cos 119899(120591 minus 1205910) minus
sin 120585 sin 119899(120591 minus 1205910) = cos(119899(120591 minus 120591
0) + 120585) we obtain (39)
In what follows we are going to prove that (40) also holdsAccording to (38) 119879
119896(120591) is nonnegative if and only if
120582119896max120591
119903119896(120591) le 1 (57)
Let us first show that position of global maximumof 119903119896(120591)
belongs to the interval |120591 minus 1205910| le 2120587119896 Relation (56) can be
rewritten as
119903119896(120591) = 119903
119896(1205910minus2120585
119896) + 119902119896(120591) (58)
where
119903119896(1205910minus2120585
119896) = (119896 minus 1) cos 120585 + 119896 sin (120585 minus 120585119896)
sin (120585119896) (59)
119902119896(120591) =
1
1 minus cos (120591 minus 1205910)
sdot [cos 120585 minus cos(119896(120591 minus 1205910+120585
119896))
+119896 sin 120585sin (120585119896)
(cos(120591 minus 1205910+120585
119896) minus cos(120585
119896))]
(60)
8 Mathematical Problems in Engineering
For |120585| lt 120587 relation sin 120585 sin(120585119896) gt 0 obviously holds Fromcos 119905 gt cos 1199051015840 for |119905| le 120587119896 lt |119905
1015840
| le 120587 it follows that positionof global maximum of the function of type [119888 cos 119905 minus cos(119896119905)]for 119888 gt 0 belongs to interval |119905| le 120587119896 Therefore position ofglobal maximum of the expression in the square brackets in(60) for |120585| lt 120587 belongs to interval |120591 minus 120591
0+ 120585119896| le 120587119896 This
inequality together with |120585| lt 120587 leads to |120591minus1205910| lt 2120587119896 Since
[1 minus cos(120591 minus 1205910)]minus1 decreases with increasing |120591 minus 120591
0| le 120587 it
follows that 119902119896(120591) for |120585| lt 120587 has global maximum on interval
|120591minus1205910| lt 2120587119896 For |120585| = 120587 it is easy to show thatmax
120591119902119896(120591) =
119902119896(1205910plusmn 2120587119896) = 0 Since 119903
119896(120591) minus 119902
119896(120591) is constant (see (58))
it follows from previous considerations that 119903119896(120591) has global
maximum on interval |120591 minus 1205910| le 2120587119896
To find max120591119903119896(120591) let us consider first derivative of 119903
119896(120591)
with respect to 120591 Starting from (56) first derivative of 119903119896(120591)
can be expressed in the following form
119889119903119896(120591)
119889120591= minus119904 (120591) sdot sin(
119896 (120591 minus 1205910)
2+ 120585) (61)
where
119904 (120591) = [sin(119896 (120591 minus 120591
0)
2) cos(
120591 minus 1205910
2)
minus 119896 cos(119896 (120591 minus 120591
0)
2) sin(
120591 minus 1205910
2)]
sdot sinminus3 (120591 minus 1205910
2)
(62)
Using (A6) (see Appendices) (62) can be rewritten as
119904 (120591) = 2
119896minus1
sum
119899=1
119899 (119896 minus 119899) cos((119896 minus 2119899) (120591 minus 120591
0)
2) (63)
From 119899(119896 minus 119899) gt 0 and |119896 minus 2119899| lt 119896 119899 = 1 (119896 minus 1) itfollows that all summands in (63) decrease with increasing|120591 minus 1205910| providing that |120591 minus 120591
0| le 2120587119896 Therefore 119904(120591) ge 119904(120591
0plusmn
2120587119896) = 119896sin2(120587119896) gt 0 for |120591 minus 1205910| le 2120587119896 Consequently
119889119903119896(120591)119889120591 = 0 and |120591minus120591
0| le 2120587119896 imply that sin(119896(120591minus120591
0)2+
120585) = 0From |120585| le 120587 |120591minus120591
0| le 2120587119896 and sin(119896(120591minus120591
0)2+120585) = 0
it follows that 120591minus1205910+120585119896 = minus120585119896 or |120591minus120591
0+120585119896| = (2120587minus|120585|)119896
and therefore cos(119896(120591 minus 1205910+ 120585119896)) = cos 120585 Since cos(120585119896) ge
cos(2120587 minus |120585|)119896 it follows that max120591119902119896(120591) is attained for 120591 =
1205910minus2120585119896 Furthermore from (60) it follows that max
120591119902119896(120591) =
119902119896(1205910minus 2120585119896) = 0 which together with (58)-(59) leads to
max120591
119903119896(120591) = 119903
119896(1205910minus2120585
119896)
= (119896 minus 1) cos 120585 + 119896 sin (120585 minus 120585119896)sin (120585119896)
(64)
Both terms on the right hand side of (64) are even functionsof 120585 and decrease with increase of |120585| |120585| le 120587 Thereforemax120591119903119896(120591) attains its lowest value for |120585| = 120587 It is easy to
show that right hand side of (64) for |120585| = 120587 is equal to 1which further implies that
max120591
119903119896(120591) ge 1 (65)
From (65) it follows that (57) can be rewritten as 120582119896
le
[max120591119903119896(120591)]minus1 Finally substitution of (64) into 120582
119896le
[max120591119903119896(120591)]minus1 leads to (40) which completes the proof
32 Nonnegative Waveforms with Two Zeros Nonnegativewaveforms of type (35) with two zeros always possess twoglobal minima Such nonnegative waveforms are thereforerelated to the conflict set
In this subsection we provide general description of non-negative waveforms of type (35) for 119896 ge 2 and exactly twozeros According to Remark 7 120582
119896= 1 implies |120585| = 120587 and
119879119896(120591) = 1 minus cos 119896(120591 minus 120591
0) Number of zeros of 119879
119896(120591) = 1 minus
cos 119896(120591minus1205910) on fundamental period equals 119896 which is greater
than two for 119896 gt 2 and equal to two for 119896 = 2 In the followingproposition we exclude all waveforms with 120582
119896= 1 (the case
when 119896 = 2 and 1205822= 1 is going to be discussed in Remark 10)
Proposition 9 Every nonnegative waveform of type (35) withexactly two zeros can be expressed in the following form
119879119896(120591) = 120582
119896[1 minus cos (120591 minus 120591
0)] [1 minus cos(120591 minus 120591
0+2120585
119896)]
sdot [1198880+ 2
119896minus2
sum
119899=1
119888119899cos 119899(120591 minus 120591
0+120585
119896)]
(66)
where
119888119899= [sin(120585 minus 119899120585
119896) cos(120585
119896)
minus (119896 minus 119899) cos(120585 minus 119899120585
119896) sin(120585
119896)]
sdot sinminus3 (120585119896)
(67)
120582119896= [(119896 minus 1) cos 120585 + 119896 sin(120585 minus 120585119896)
sin(120585119896)]
minus1
(68)
0 lt10038161003816100381610038161205851003816100381610038161003816 lt 120587 (69)
Remark 10 For 119896 = 2 waveforms with 1205822= 1 also have
exactly two zeros These waveforms can be included in aboveproposition by substituting (69) with 0 lt |120585| le 120587
Remark 11 Apart from nonnegative waveforms of type (35)with two zeros there are another two types of nonnegativewaveforms which can be obtained from (66)ndash(68) These are
(i) nonnegative waveforms with 119896 zeros (correspondingto |120585| = 120587) and
(ii) maximally flat nonnegative waveforms (correspond-ing to 120585 = 0)
Notice that nonnegative waveforms of type (35) with120582119896= 1 can be obtained from (66)ndash(68) by setting |120585| =
120587 Substitution of 120582119896
= 1 and |120585| = 120587 into (66) alongwith execution of all multiplications and usage of (A2) (seeAppendices) leads to 119879
119896(120591) = 1 minus cos 119896(120591 minus 120591
0)
Mathematical Problems in Engineering 9
Also maximally flat nonnegative waveforms (they haveonly one zero [21]) can be obtained from (66)ndash(68) by setting120585 = 0 Thus substitution of 120585 = 0 into (66)ndash(68) leads tothe following form of maximally flat nonnegative waveformof type (35)
119879119896(120591) =
[1 minus cos (120591 minus 1205910)]2
3 (1198962 minus 1)
sdot [119896 (1198962
minus 1)
+ 2
119896minus2
sum
119899=1
(119896 minus 119899) ((119896 minus 119899)2
minus 1) cos 119899 (120591 minus 1205910)]
(70)
Maximally flat nonnegative waveforms of type (35) for 119896 le 4
can be expressed as
1198792(120591) =
2
3[1 minus cos(120591 minus 120591
0)]2
1198793(120591) =
1
2[1 minus cos(120591 minus 120591
0)]2
[2 + cos (120591 minus 1205910)]
1198794(120591) =
4
15[1 minus cos (120591 minus 120591
0)]2
sdot [5 + 4 cos (120591 minus 1205910) + cos 2 (120591 minus 120591
0)]
(71)
Remark 12 Every nonnegative waveform of type (35) withexactly one zero at nondegenerate critical point can bedescribed as in Proposition 6 providing that symbol ldquolerdquoin relation (40) is replaced with ldquoltrdquo This is an immediateconsequence of Propositions 6 and 9 and Remark 11
Remark 13 Identity [1minus cos(120591minus1205910)][1minus cos(120591minus120591
0+2120585119896)] =
[cos 120585119896 minus cos(120591 minus 1205910+ 120585119896)]
2 implies that (66) can be alsorewritten as
119879119896(120591) = 120582
119896[cos 120585119896
minus cos(120591 minus 1205910+120585
119896)]
2
sdot [1198880+ 2
119896minus2
sum
119899=1
119888119899cos 119899(120591 minus 120591
0+120585
119896)]
(72)
Furthermore substitution of (67) into (72) leads to
119879119896(120591)
= 120582119896[cos 120585119896
minus cos(120591 minus 1205910+120585
119896)]
sdot [(119896 minus 1) sin 120585sin (120585119896)
minus 2
119896minus1
sum
119899=1
sin (120585 minus 119899120585119896)sin (120585119896)
cos 119899(120591 minus 1205910+120585
119896)]
(73)
Remark 14 According to (A6) (see Appendices) it followsthat coefficients (67) can be expressed as
119888119899= 2
119896minus119899minus1
sum
119898=1
119898(119896 minus 119899 minus 119898) cos((119896 minus 119899 minus 2119898) 120585119896
) (74)
Furthermore from (74) it follows that coefficients 119888119896minus2
119888119896minus3
119888119896minus4
and 119888119896minus5
are equal to119888119896minus2
= 2 (75)
119888119896minus3
= 8 cos(120585119896) (76)
119888119896minus4
= 8 + 12 cos(2120585119896) (77)
119888119896minus5
= 24 cos(120585119896) + 16 cos(3120585
119896) (78)
For example for 119896 = 2 (75) and (68) lead to 1198880= 2 and
1205822= 1(2+cos 120585) respectively which from (72) further imply
that
1198792(120591) =
2 [cos(1205852) minus cos(120591 minus 1205910+ 1205852)]
2
[2 + cos 120585] (79)
Also for 119896 = 3 (75) (76) and (68) lead to 1198881= 2 119888
0=
8 cos(1205853) and 1205823= [2(3 cos(1205853) + cos 120585)]minus1 respectively
which from (72) further imply that
1198793(120591) =
2 [cos (1205853) minus cos (120591 minus 1205910+ 1205853)]
2
[3 cos (1205853) + cos 120585]
sdot [2 cos(1205853) + cos(120591 minus 120591
0+120585
3)]
(80)
Remark 15 According to (A5) (see Appendices) relation(68) can be rewritten as
120582119896= [(119896 minus 1) cos 120585 + 119896
119896minus1
sum
119899=1
cos((119896 minus 2119899)120585119896
)]
minus1
(81)
Clearly amplitude 120582119896of 119896th harmonic of nonnegative wave-
form of type (35) with exactly two zeros is even functionof 120585 Since cos((119896 minus 2119899)120585119896) 119899 = 0 (119896 minus 1) decreaseswith increase of |120585| on interval 0 le |120585| le 120587 it follows that120582119896monotonically increases with increase of |120585| Right hand
side of (68) is equal to 1(1198962
minus 1) for 120585 = 0 and to one for|120585| = 120587 Therefore for nonnegative waveforms of type (35)with exactly two zeros the following relation holds
1
1198962 minus 1lt 120582119896lt 1 (82)
The left boundary in (82) corresponds to maximally flatnonnegative waveforms (see Remark 11) The right boundaryin (82) corresponds to nonnegative waveforms with 119896 zeros(also see Remark 11)
Amplitude of 119896th harmonic of nonnegative waveform oftype (35) with two zeros as a function of parameter 120585 for 119896 le5 is presented in Figure 5
Remark 16 Nonnegative waveform of type (35) with twozeros can be also expressed in the following form
119879119896(120591) = 1 minus 120582
119896
119896 sin 120585sin (120585119896)
cos(120591 minus 1205910+120585
119896)
+ 120582119896cos (119896 (120591 minus 120591
0) + 120585)
(83)
10 Mathematical Problems in Engineering
1
08
06
04
02
0minus1 minus05 0 05
Am
plitu
de120582k
1
Parameter 120585120587
k = 2k = 3
k = 4k = 5
Figure 5 Amplitude of 119896th harmonic of nonnegative waveformwith two zeros as a function of parameter 120585
where 120582119896is given by (68) and 0 lt |120585| lt 120587 From (83) it follows
that coefficients of fundamental harmonic of nonnegativewaveform of type (35) with two zeros are
1198861= minus1205821cos(120591
0minus120585
119896) 119887
1= minus1205821sin(120591
0minus120585
119896) (84)
where 1205821is amplitude of fundamental harmonic
1205821=
119896 sin 120585sin (120585119896)
120582119896 (85)
Coefficients of 119896th harmonic are given by (45)-(46)Notice that (68) can be rewritten as
120582119896= [cos(120585
119896)
119896 sin 120585sin(120585119896)
minus cos 120585]minus1
(86)
By introducing new variable
119909 = cos(120585119896) (87)
and using the Chebyshev polynomials (eg see Appendices)relations (85) and (86) can be rewritten as
1205821= 119896120582119896119880119896minus1
(119909) (88)
120582119896=
1
119896119909119880119896minus1
(119909) minus 119881119896(119909)
(89)
where119881119896(119909) and119880
119896(119909) denote the Chebyshev polynomials of
the first and second kind respectively From (89) it followsthat
120582119896[119896119909119880119896minus1
(119909) minus 119881119896(119909)] minus 1 = 0 (90)
which is polynomial equation of 119896th degree in terms of var-iable 119909 From 0 lt |120585| lt 120587 and (87) it follows that
cos(120587119896) lt 119909 lt 1 (91)
Since 120582119896is monotonically increasing function of |120585| 0 lt |120585| lt
120587 it follows that 120582119896is monotonically decreasing function of
119909 This further implies that (90) has only one solution thatsatisfies (91) (For 119896 = 2 expression (91) reads cos(1205872) le
119909 lt 1) This solution for 119909 (which can be obtained at leastnumerically) according to (88) leads to amplitude 120582
1of
fundamental harmonicFor 119896 le 4 solutions of (90) and (91) are
119909 = radic1 minus 1205822
21205822
1
3lt 1205822le 1
119909 =1
23radic1205823
1
8lt 1205823lt 1
119909 = radic1
6(1 + radic
51205824+ 3
21205824
)1
15lt 1205824lt 1
(92)
Insertion of (92) into (88) leads to the following relationsbetween amplitude 120582
1of fundamental and amplitude 120582
119896of
119896th harmonic 119896 le 4
1205821= radic8120582
2(1 minus 120582
2)
1
3lt 1205822le 1 (93)
1205821= 3 (
3radic1205823minus 1205823)
1
8lt 1205823lt 1 (94)
1205821= radic
32
27(radic2120582
4(3 + 5120582
4)3
minus 21205824(9 + 7120582
4))
1
15lt 1205824lt 1
(95)
Proof of Proposition 9 As it has been shown earlier (seeProposition 6) nonnegative waveform of type (35) with atleast one zero can be represented in form (38) Since weexclude nonnegative waveforms with 120582
119896= 1 according to
Remark 7 it follows that we exclude case |120585| = 120587Therefore inthe quest for nonnegative waveforms of type (35) having twozeros we will start with waveforms of type (38) for |120585| lt 120587It is clear that nonnegative waveforms of type (38) have twozeros if and only if
120582119896= [max120591
119903119896(120591)]minus1
(96)
and max120591119903119896(120591) = 119903
119896(1205910) According to (64) max
120591119903119896(120591) =
119903119896(1205910) implies |120585| = 0 Therefore it is sufficient to consider
only the interval (69)Substituting (96) into (38) we obtain
119879119896(120591) =
[1 minus cos (120591 minus 1205910)] [max
120591119903119896(120591) minus 119903
119896(120591)]
max120591119903119896(120591)
(97)
Mathematical Problems in Engineering 11
Expression max120591119903119896(120591) minus 119903
119896(120591) according to (64) and (39)
equals
max120591
119903119896(120591) minus 119903
119896(120591) = 119896
sin ((119896 minus 1) 120585119896)sin (120585119896)
minus 2
119896minus1
sum
119899=1
(119896 minus 119899) cos (119899 (120591 minus 1205910) + 120585)
(98)
Comparison of (97) with (66) yields
max120591
119903119896(120591) minus 119903
119896(120591) = [1 minus cos(120591 minus 120591
0+2120585
119896)]
sdot [1198880+ 2
119896minus2
sum
119899=1
119888119899cos 119899(120591 minus 120591
0+120585
119896)]
(99)
where coefficients 119888119899 119899 = 0 119896 minus 2 are given by (67) In
what follows we are going to show that right hand sides of(98) and (99) are equal
From (67) it follows that
1198880minus 1198881cos(120585
119896) = 119896
sin (120585 minus 120585119896)sin (120585119896)
(100)
Also from (67) for 119899 = 1 119896minus3 it follows that the followingrelations hold
(119888119899minus1
+ 119888119899+1
) cos(120585119896) minus 2119888
119899= 2 (119896 minus 119899) cos(120585 minus 119899120585
119896)
(119888119899minus1
minus 119888119899+1
) sin(120585119896) = 2 (119896 minus 119899) sin(120585 minus 119899120585
119896)
(101)
From (99) by using (75) (76) (100)-(101) and trigonometricidentities
cos(120591 minus 1205910+2120585
119896) = cos(120585
119896) cos(120591 minus 120591
0+120585
119896)
minus sin(120585119896) sin(120591 minus 120591
0+120585
119896)
cos(120585 minus 119899120585
119896) cos(119899(120591 minus 120591
0+120585
119896))
minus sin(120585 minus 119899120585
119896) sin(119899(120591 minus 120591
0+120585
119896))
= cos (119899 (120591 minus 1205910) + 120585)
(102)
we obtain (98) Consequently (98) and (99) are equal whichcompletes the proof
33 Nonnegative Waveforms with Two Zeros and PrescribedCoefficients of 119896thHarmonic In this subsectionwe show thatfor prescribed coefficients 119886
119896and 119887119896 there are 119896 nonnegative
waveforms of type (35) with exactly two zeros According to
(37) and (82) coefficients 119886119896and 119887119896of nonnegative waveforms
of type (35) with exactly two zeros satisfy the followingrelation
1
1198962 minus 1lt radic1198862119896+ 1198872119896lt 1 (103)
According to Remark 16 the value of 119909 (see (87)) that cor-responds to 120582
119896= radic1198862
119896+ 1198872119896can be determined from (90)-
(91) As we mentioned earlier (90) has only one solutionthat satisfies (91) This value of 119909 according to (88) leadsto the amplitude 120582
1of fundamental harmonic (closed form
expressions for 1205821in terms of 120582
119896and 119896 le 4 are given by (93)ndash
(95))On the other hand from (45)-(46) it follows that
1198961205910minus 120585 = atan 2 (119887
119896 119886119896) + 2119902120587 119902 = 1 (119896 minus 1) (104)
where function atan 2(119910 119909) is defined as
atan 2 (119910 119909) =
arctan(119910
119909) if 119909 ge 0
arctan(119910
119909) + 120587 if 119909 lt 0 119910 ge 0
arctan(119910
119909) minus 120587 if 119909 lt 0 119910 lt 0
(105)
with the codomain (minus120587 120587] Furthermore according to (84)and (104) the coefficients of fundamental harmonic of non-negative waveforms with two zeros and prescribed coeffi-cients of 119896th harmonic are equal to
1198861= minus1205821cos[
atan 2 (119887119896 119886119896) + 2119902120587
119896]
1198871= minus1205821sin[
atan 2 (119887119896 119886119896) + 2119902120587
119896]
(106)
where 119902 = 0 (119896minus 1) For chosen 119902 according to (104) and(66) positions of zeros are
1205910=1
119896[120585 + atan 2 (119887
119896 119886119896) + 2119902120587]
1205910minus2120585
119896=1
119896[minus120585 + atan 2 (119887
119896 119886119896) + 2119902120587]
(107)
From (106) and 119902 = 0 (119896minus1) it follows that for prescribedcoefficients 119886
119896and 119887119896 there are 119896 nonnegative waveforms of
type (35) with exactly two zerosWe provide here an algorithm to facilitate calculation
of coefficients 1198861and 1198871of nonnegative waveforms of type
(35) with two zeros and prescribed coefficients 119886119896and 119887
119896
providing that 119886119896and 119887119896satisfy (103)
12 Mathematical Problems in Engineering
2
1
0
Wav
efor
ms
0 1 2 3 4
Angle 120591120587
q = 0
q = 1
q = 2
Figure 6 Nonnegative waveforms with two zeros for 119896 = 3 1198863=
minus015 and 1198873= minus02
Algorithm 17 (i) Calculate 120582119896= radic1198862119896+ 1198872119896
(ii) identify 119909 that satisfies both relations (90) and (91)(iii) calculate 120582
1according to (88)
(iv) choose integer 119902 such that 0 le 119902 le 119896 minus 1(v) calculate 119886
1and 1198871according to (106)
For 119896 le 4 by using (93) for 119896 = 2 (94) for 119896 = 3 and (95)for 119896 = 4 it is possible to calculate directly 120582
1from 120582
119896and
proceed to step (iv)For 119896 = 2 and prescribed coefficients 119886
2and 1198872 there are
two waveforms with two zeros one corresponding to 1198861lt 0
and the other corresponding to 1198861gt 0 (see also [12])
Let us take as an input 119896 = 3 1198863= minus015 and 119887
3= minus02
Execution of Algorithm 17 on this input yields 1205823= 025 and
1205821= 11399 (according to (94)) For 119902 = 0 we calculate
1198861= minus08432 and 119887
1= 07670 (corresponding waveform is
presented by solid line in Figure 6) for 119902 = 1 we calculate1198861= minus02426 and 119887
1= minus11138 (corresponding waveform is
presented by dashed line) for 119902 = 2 we calculate 1198861= 10859
and 1198871= 03468 (corresponding waveform is presented by
dotted line)As another example of the usage of Algorithm 17 let us
consider case 119896 = 4 and assume that1198864= minus015 and 119887
4= minus02
Consequently 1205824= 025 and 120582
1= 09861 (according to (95))
For 119902 = 0 3we calculate the following four pairs (1198861 1198871) of
coefficients of fundamental harmonic (minus08388 05184) for119902 = 0 (minus05184 minus08388) for 119902 = 1 (08388 minus05184) for 119902 =2 and (05184 08388) for 119902 = 3 Corresponding waveformsare presented in Figure 7
4 Nonnegative Waveforms with MaximalAmplitude of Fundamental Harmonic
In this section we provide general description of nonnegativewaveforms containing fundamental and 119896th harmonic withmaximal amplitude of fundamental harmonic for prescribedamplitude of 119896th harmonic
The main result of this section is presented in the fol-lowing proposition
3
2
1
0
Wav
efor
ms
0 1 2 3 4
Angle 120591120587
q = 0q = 1
q = 2q = 3
Figure 7 Nonnegative waveforms with two zeros for 119896 = 4 1198864=
minus015 and 1198874= minus02
Proposition 18 Every nonnegativewaveformof type (35)withmaximal amplitude 120582
1of fundamental harmonic and pre-
scribed amplitude 120582119896of 119896th harmonic can be expressed in the
following form
119879119896(120591) = [1 minus cos (120591 minus 120591
0)]
sdot [1 minus (119896 minus 1) 120582119896minus 2120582119896
119896minus1
sum
119899=1
(119896 minus 119899) cos (119899 (120591 minus 1205910))]
(108)
if 0 le 120582119896le 1(119896
2
minus 1) or
119879119896(120591) = 120582
119896[1 minus cos (120591 minus 120591
0)] [1 minus cos(120591 minus 120591
0+2120585
119896)]
sdot [1198880+ 2
119896minus2
sum
119899=1
119888119899cos 119899(120591 minus 120591
0+120585
119896)]
(109)
if 1(1198962 minus 1) le 120582119896le 1 providing that 119888
119899 119899 = 0 119896 minus 2 and
120582119896are related to 120585 via relations (67) and (68) respectively and
|120585| le 120587
Remark 19 Expression (108) can be obtained from (38) bysetting 120585 = 0 Furthermore insertion of 120585 = 0 into (43)ndash(46)leads to the following expressions for coefficients ofwaveformof type (108)
1198861= minus (1 + 120582
119896) cos 120591
0 119887
1= minus (1 + 120582
119896) sin 120591
0
119886119896= 120582119896cos (119896120591
0) 119887
119896= 120582119896sin (119896120591
0)
(110)
On the other hand (109) coincides with (66) Thereforethe expressions for coefficients of (109) and (66) also coincideThus expressions for coefficients of fundamental harmonic ofwaveform (109) are given by (84) where 120582
1is given by (85)
while expressions for coefficients of 119896th harmonic are givenby (45)-(46)
Waveforms described by (108) have exactly one zerowhile waveforms described by (109) for 1(1198962 minus 1) lt 120582
119896lt 1
Mathematical Problems in Engineering 13
14
12
1
08
06
04
02
00 05 1
Amplitude 120582k
Am
plitu
de1205821
k = 2
k = 3
k = 4
Figure 8 Maximal amplitude of fundamental harmonic as a func-tion of amplitude of 119896th harmonic
have exactly two zeros As we mentioned earlier waveforms(109) for 120582
119896= 1 have 119896 zeros
Remark 20 Maximal amplitude of fundamental harmonic ofnonnegative waveforms of type (35) for prescribed amplitudeof 119896th harmonic can be expressed as
1205821= 1 + 120582
119896 (111)
if 0 le 120582119896le 1(119896
2
minus 1) or
1205821=
119896 sin 120585119896 sin 120585 cos (120585119896) minus cos 120585 sin (120585119896)
(112)
if 1(1198962 minus 1) le 120582119896le 1 where 120585 is related to 120582
119896via (68) (or
(86)) and |120585| le 120587From (110) it follows that (111) holds Substitution of (86)
into (85) leads to (112)Notice that 120582
119896= 1(119896
2
minus 1) is the only common point ofthe intervals 0 le 120582
119896le 1(119896
2
minus 1) and 1(1198962
minus 1) le 120582119896le
1 According to (111) 120582119896= 1(119896
2
minus 1) corresponds to 1205821=
1198962
(1198962
minus1) It can be also obtained from (112) by setting 120585 = 0The waveforms corresponding to this pair of amplitudes aremaximally flat nonnegative waveforms
Maximal amplitude of fundamental harmonic of non-negative waveform of type (35) for 119896 le 4 as a function ofamplitude of 119896th harmonic is presented in Figure 8
Remark 21 Maximum value of amplitude of fundamentalharmonic of nonnegative waveform of type (35) is
1205821max =
1
cos (120587 (2119896)) (113)
This maximum value is attained for |120585| = 1205872 (see (112)) Thecorresponding value of amplitude of 119896th harmonic is 120582
119896=
(1119896) tan(120587(2119896)) Nonnegative waveforms of type (35) with1205821= 1205821max have two zeros at 1205910 and 1205910 minus 120587119896 for 120585 = 1205872 or
at 1205910and 1205910+ 120587119896 for 120585 = minus1205872
14
12
1
08
06
04
02
0minus1 minus05 0 05 1
Am
plitu
de1205821
Parameter 120585120587
k = 2k = 3k = 4
Figure 9 Maximal amplitude of fundamental harmonic as a func-tion of parameter 120585
To prove that (113) holds let us first show that the fol-lowing relation holds for 119896 ge 2
cos( 120587
2119896) lt 1 minus
1
1198962 (114)
From 119896 ge 2 it follows that sinc(120587(4119896)) gt sinc(1205874) wheresinc 119909 = (sin119909)119909 and therefore sin(120587(4119896)) gt 1(radic2119896)By using trigonometric identity cos 2119909 = 1 minus 2sin2119909 weimmediately obtain (114)
According to (111) and (112) it is clear that 1205821attains its
maximum value on the interval 1(1198962 minus 1) le 120582119896le 1 Since
120582119896is monotonic function of |120585| on interval |120585| le 120587 (see
Remark 15) it follows that 119889120582119896119889120585 = 0 for 0 lt |120585| lt 120587
Therefore to find critical points of 1205821as a function of 120582
119896
it is sufficient to find critical points of 1205821as a function of
|120585| 0 lt |120585| lt 120587 and consider its values at the end points120585 = 0 and |120585| = 120587 Plot of 120582
1as a function of parameter 120585
for 119896 le 4 is presented in Figure 9 According to (112) firstderivative of 120582
1with respect to 120585 is equal to zero if and only
if (119896 cos 120585 sin(120585119896) minus sin 120585 cos(120585119896)) cos 120585 = 0 On interval0 lt |120585| lt 120587 this is true if and only if |120585| = 1205872 Accordingto (112) 120582
1is equal to 119896
2
(1198962
minus 1) for 120585 = 0 equal to zerofor |120585| = 120587 and equal to 1 cos(120587(2119896)) for |120585| = 1205872 From(114) it follows that 1198962(1198962minus1) lt 1 cos(120587(2119896)) and thereforemaximum value of 120582
1is given by (113) Moreover maximum
value of 1205821is attained for |120585| = 1205872
According to above consideration all nonnegative wave-forms of type (35) having maximum value of amplitude offundamental harmonic can be obtained from (109) by setting|120585| = 1205872 Three of them corresponding to 119896 = 3 120585 = 1205872and three different values of 120591
0(01205876 and1205873) are presented
in Figure 10 Dotted line corresponds to 1205910= 0 (coefficients
of corresponding waveform are 1198861= minus1 119887
1= 1radic3 119886
3= 0
and 1198873= minusradic39) solid line to 120591
0= 1205876 (119886
1= minus2radic3 119887
1= 0
1198863= radic39 and 119887
3= 0) and dashed line to 120591
0= 1205873 (119886
1= minus1
1198871= minus1radic3 119886
3= 0 and 119887
3= radic39)
Proof of Proposition 18 As it has been shown earlier (Propo-sition 6) nonnegative waveform of type (35) with at least
14 Mathematical Problems in Engineering
2
1
0
0 1 2 3 4
Angle 120591120587
Wav
efor
ms
1205910 = 01205910 = 12058761205910 = 1205873
Figure 10 Nonnegative waveforms with maximum amplitude offundamental harmonic for 119896 = 3 and 120585 = 1205872
one zero can be represented in form (38) According to (43)(44) and (36) for amplitude 120582
1of fundamental harmonic of
waveforms of type (38) the following relation holds
1205821= radic(1 + 120582
119896cos 120585)2 + 11989621205822
119896sin2120585 (115)
where 120582119896satisfy (40) and |120585| le 120587
Because of (40) in the quest of finding maximal 1205821for
prescribed 120582119896 we have to consider the following two cases
(Case i)120582119896lt [(119896minus1) cos 120585 + 119896 sin(120585minus120585119896) sin(120585119896)]minus1
(Case ii)120582119896= [(119896minus1) cos 120585+119896 sin(120585minus120585119896) sin(120585119896)]minus1
Case i Since 120582119896lt [(119896 minus 1) cos 120585 + 119896 sin(120585 minus 120585119896) sin(120585119896)]minus1
implies 120582119896
= 1 according to (115) it follows that 1205821
= 0Hence 119889120582
1119889120585 = 0 implies
2120582119896sin 120585 [1 minus (1198962 minus 1) 120582
119896cos 120585] = 0 (116)
Therefore 1198891205821119889120585 = 0 if 120582
119896= 0 (Option 1) or sin 120585 = 0
(Option 2) or (1198962 minus 1)120582119896cos 120585 = 1 (Option 3)
Option 1 According to (115) 120582119896= 0 implies 120582
1= 1 (notice
that this implication shows that 1205821does not depend on 120585 and
therefore we can set 120585 to zero value)
Option 2 According to (115) sin 120585 = 0 implies 1205821= 1 +
120582119896cos 120585 which further leads to the conclusion that 120582
1is
maximal for 120585 = 0 For 120585 = 0 120582119896lt [(119896 minus 1) cos 120585 + 119896 sin(120585 minus
120585119896) sin(120585119896)]minus1 becomes 120582119896lt 1(119896
2
minus 1)
Option 3 This option leads to contradiction To show thatnotice that (119896
2
minus 1)120582119896cos 120585 = 1 and 120582
119896lt [(119896 minus
1) cos 120585 + 119896 sin(120585 minus 120585119896) sin(120585119896)]minus1 imply that (119896 minus 1) cos 120585 gtsin(120585minus120585119896) sin(120585119896) Using (A5) (see Appendices) the latestinequality can be rewritten assum119896minus1
119899=1[cos 120585minuscos((119896minus2119899)120585119896)] gt
0 But from |119896 minus 2119899| lt 119896 119899 = 1 (119896 minus 1) and |120585| le 120587
it follows that all summands are not positive and therefore(119896minus1) cos 120585 gt sin(120585minus120585119896) sin(120585119896) does not hold for |120585| le 120587
Consequently Case i implies 120585 = 0 and 120582119896lt 1(119896
2
minus 1)Finally substitution of 120585 = 0 into (38) leads to (108) whichproves that (108) holds for 120582
119896lt 1(119896
2
minus 1)
Case ii Relation120582119896= [(119896minus1) cos 120585+119896 sin(120585minus120585119896) sin(120585119896)]minus1
according to Proposition 9 and Remark 11 implies that cor-responding waveforms can be expressed via (66)ndash(68) for|120585| le 120587 Furthermore 120582
119896= [(119896 minus 1) cos 120585 + 119896 sin(120585 minus
120585119896) sin(120585119896)]minus1 and |120585| le 120587 imply 1(1198962 minus 1) le 120582119896le 1
This proves that (109) holds for 1(1198962 minus 1) le 120582119896le 1
Finally let us prove that (108) holds for 120582119896= 1(119896
2
minus
1) According to (68) (see also Remark 11) this value of 120582119896
corresponds to 120585 = 0 Furthermore substitution of 120582119896=
1(1198962
minus 1) and 120585 = 0 into (109) leads to (70) which can berewritten as
119879119896(120591) =
[1 minus cos (120591 minus 1205910)]
(1 minus 1198962)
sdot [119896 (119896 minus 1) minus 2
119896minus1
sum
119899=1
(119896 minus 119899) cos (119899 (120591 minus 1205910))]
(117)
Waveform (117) coincides with waveform (108) for 120582119896
=
1(1 minus 1198962
) Consequently (108) holds for 120582119896= 1(1 minus 119896
2
)which completes the proof
5 Nonnegative Waveforms with MaximalAbsolute Value of the Coefficient of CosineTerm of Fundamental Harmonic
In this sectionwe consider general description of nonnegativewaveforms of type (35) with maximal absolute value ofcoefficient 119886
1for prescribed coefficients of 119896th harmonicThis
type of waveform is of particular interest in PA efficiencyanalysis In a number of cases of practical interest eithercurrent or voltage waveform is prescribed In such casesthe problem of finding maximal efficiency of PA can bereduced to the problem of finding nonnegative waveformwith maximal coefficient 119886
1for prescribed coefficients of 119896th
harmonic (see also Section 7)In Section 51 we provide general description of nonneg-
ative waveforms of type (35) with maximal absolute value ofcoefficient 119886
1for prescribed coefficients of 119896th harmonic In
Section 52 we illustrate results of Section 51 for particularcase 119896 = 3
51 Nonnegative Waveforms with Maximal Absolute Value ofCoefficient 119886
1for 119896 ge 2 Waveforms 119879
119896(120591) of type (35) with
1198861ge 0 can be derived from those with 119886
1le 0 by shifting
by 120587 and therefore we can assume without loss of generalitythat 119886
1le 0 Notice that if 119896 is even then shifting 119879
119896(120591) by
120587 produces the same result as replacement of 1198861with minus119886
1
(119886119896remains the same) On the other hand if 119896 is odd then
shifting 119879119896(120591) by 120587 produces the same result as replacement
of 1198861with minus119886
1and 119886119896with minus119886
119896
According to (37) coefficients of 119896th harmonic can beexpressed as
119886119896= 120582119896cos 120575 119887
119896= 120582119896sin 120575 (118)
Mathematical Problems in Engineering 15
where
|120575| le 120587 (119)
Conversely for prescribed coefficients 119886119896and 119887
119896 120575 can be
determined as
120575 = atan 2 (119887119896 119886119896) (120)
where definition of function atan 2(119910 119909) is given by (105)The main result of this section is stated in the following
proposition
Proposition 22 Every nonnegative waveform of type (35)withmaximal absolute value of coefficient 119886
1le 0 for prescribed
coefficients 119886119896and 119887119896of 119896th harmonic can be represented as
119879119896(120591)
= [1 minus cos 120591]
sdot [1 minus (119896 minus 1) 119886119896minus 2
119896minus1
sum
119899=1
(119896 minus 119899) (119886119896cos 119899120591 + 119887
119896sin 119899120591)]
(121)
if 119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) le 1+119886
119896 where 120575 = atan 2(bk
119886119896) or
119879119896(120591) = 120582
119896[1 minus cos(120591 minus (120575 + 120585)
119896)]
sdot [1 minus cos(120591 minus (120575 minus 120585)
119896)]
sdot [1198880+ 2
119896minus2
sum
119899=1
119888119899cos 119899(120591 minus 120575
119896)]
(122)
if 119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) ge 1 + 119886
119896 where 119888
119899 119899 = 0
119896minus2 and 120582119896= radic1198862119896+ 1198872119896are related to 120585 via relations (67) and
(68) respectively and |120585| le 120587
Remark 23 Expression (121) can be obtained from (38) bysetting 120591
0= 0 and 120585 = minus120575 and then replacing 120582
119896cos 120575 with
119886119896(see (118)) and 120582
119896cos(119899120591 minus 120575) with 119886
119896cos 119899120591 + 119887
119896sin 119899120591
(see also (118)) Furthermore insertion of 1205910= 0 and 120585 =
minus120575 into (43)ndash(46) leads to the following relations betweenfundamental and 119896th harmonic coefficients of waveform(121)
1198861= minus (1 + 119886
119896) 119887
1= minus119896119887
119896 (123)
On the other hand expression (122) can be obtained from(66) by replacing 120591
0minus120585119896with 120575119896 Therefore substitution of
1205910minus 120585119896 = 120575119896 in (84) leads to
1198861= minus1205821cos(120575
119896) 119887
1= minus1205821sin(120575
119896) (124)
where 1205821is given by (85)
The fundamental harmonic coefficients 1198861and 1198871of wave-
form of type (35) with maximal absolute value of coefficient1198861le 0 satisfy both relations (123) and (124) if 119886
119896and 119887119896satisfy
1 + 119886119896= 119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) For such waveforms
relations 1205910= 0 and 120585 = minus120575 also hold
Remark 24 Amplitude of 119896th harmonic of nonnegativewaveform of type (35) with maximal absolute value of coeffi-cient 119886
1le 0 and coefficients 119886
119896 119887119896satisfying 1 + 119886
119896=
119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) is
120582119896=
sin (120575119896)119896 sin 120575 cos (120575119896) minus cos 120575 sin (120575119896)
(125)
To show that it is sufficient to substitute 119886119896= 120582119896cos 120575 (see
(118)) into 1 + 119886119896= 119896120582119896[sin 120575 sin(120575119896)] cos(120575119896)
Introducing new variable
119910 = cos(120575119896) (126)
and using the Chebyshev polynomials (eg see Appendices)relations 119886
119896= 120582119896cos 120575 and (125) can be rewritten as
119886119896= 120582119896119881119896(119910) (127)
120582119896=
1
119896119910119880119896minus1
(119910) minus 119881119896(119910)
(128)
where119881119896(119910) and119880
119896(119910) denote the Chebyshev polynomials of
the first and second kind respectively Substitution of (128)into (127) leads to
119886119896119896119910119880119896minus1
(119910) minus (1 + 119886119896) 119881119896(119910) = 0 (129)
which is polynomial equation of 119896th degree in terms of var-iable 119910 From |120575| le 120587 and (126) it follows that
cos(120587119896) le 119910 le 1 (130)
In what follows we show that 119886119896is monotonically increas-
ing function of 119910 on the interval (130) From 120585 = minus120575 (seeRemark 23) and (81) it follows that 120582minus1
119896= (119896 minus 1) cos 120575 +
119896sum119896minus1
119899=1cos((119896 minus 2119899)120575119896) ge 1 and therefore 119886
119896= 120582119896cos 120575 can
be rewritten as
119886119896=
cos 120575(119896 minus 1) cos 120575 + 119896sum119896minus1
119899=1cos ((119896 minus 2119899) 120575119896)
(131)
Obviously 119886119896is even function of 120575 and all cosines in (131)
are monotonically decreasing functions of |120575| on the interval|120575| le 120587 It is easy to show that cos((119896 minus 2119899)120575119896) 119899 =
1 (119896 minus 1) decreases slower than cos 120575 when |120575| increasesThis implies that denominator of the right hand side of(131) decreases slower than numerator Since denominator ispositive for |120575| le 120587 it further implies that 119886
119896is decreasing
function of |120575| on interval |120575| le 120587 Consequently 119886119896is
monotonically increasing function of 119910 on the interval (130)Thus we have shown that 119886
119896is monotonically increasing
function of 119910 on the interval (130) and therefore (129) hasonly one solution that satisfies (130) According to (128) thevalue of 119910 obtained from (129) and (130) either analyticallyor numerically leads to amplitude 120582
119896of 119896th harmonic
16 Mathematical Problems in Engineering
1
05
0
minus05
minus1
minus1 minus05 0 05 1
Coefficient ak
Coe
ffici
entb
k
radica2k+ b2
kle 1
k = 2k = 3k = 4
Figure 11 Plot of (119886119896 119887119896) satisfying 1 + 119886
119896= 119896120582
119896[sin 120575 sin(120575
119896)] cos(120575119896) for 119896 le 4
By solving (129) and (130) for 119896 le 4 we obtain
119910 = radic1 + 1198862
2 (1 minus 1198862) minus1 le 119886
2le1
3
119910 = radic3
4 (1 minus 21198863) minus1 le 119886
3le1
8
119910 =radicradic2 minus 4119886
4+ 1011988624minus 2 (1 minus 119886
4)
4 (1 minus 31198864)
minus1 le 1198864le
1
15
(132)
Insertion of (132) into (128) leads to the following explicitexpressions for the amplitude 120582
119896 119896 le 4
1205822=1
2(1 minus 119886
2) minus1 le 119886
2le1
3 (133)
1205822
3= [
1
3(1 minus 2119886
3)]
3
minus1 le 1198863le1
8 (134)
1205824=1
4(minus1 minus 119886
4+ radic2 minus 4119886
4+ 1011988624) minus1 le 119886
4le
1
15
(135)
Relations (133)ndash(135) define closed lines (see Figure 11) whichseparate points representing waveforms of type (121) frompoints representing waveforms of type (122) For given 119896points inside the corresponding curve refer to nonnegativewaveforms of type (121) whereas points outside curve (andradic1198862119896+ 1198872119896le 1) correspond to nonnegative waveforms of type
(122) Points on the respective curve correspond to the wave-forms which can be expressed in both forms (121) and (122)
Remark 25 Themaximum absolute value of coefficient 1198861of
nonnegative waveform of type (35) is
100381610038161003816100381611988611003816100381610038161003816max =
1
cos (120587 (2119896)) (136)
This maximum value is attained for |120585| = 1205872 and 120575 = 0
(see (124)) Notice that |1198861|max is equal to the maximum value
1205821max of amplitude of fundamental harmonic (see (113))
Coefficients of waveform with maximum absolute value ofcoefficient 119886
1 1198861lt 0 are
1198861= minus
1
cos (120587 (2119896)) 119886
119896=1
119896tan( 120587
(2119896))
1198871= 119887119896= 0
(137)
Waveformdescribed by (137) is cosinewaveformhaving zerosat 120587(2119896) and minus120587(2119896)
In the course of proving (136) notice first that |1198861|max le
1205821max holds According to (123) and (124) maximum of |119886
1|
occurs for 119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) ge 1 + 119886
119896 From (124)
it immediately follows that maximum value of |1198861| is attained
if and only if 1205821= 1205821max and 120575 = 0 which because of
120575119896 = 1205910minus120585119896 further implies 120591
0= 120585119896 Sincemaximumvalue
of 1205821is attained for |120585| = 1205872 it follows that corresponding
waveform has zeros at 120587(2119896) and minus120587(2119896)
Proof of Proposition 22 As it was mentioned earlier in thissection we can assume without loss of generality that 119886
1le 0
We consider waveforms119879119896(120591) of type (35) such that119879
119896(120591) ge 0
and119879119896(120591) = 0 for some 120591
0 Fromassumption that nonnegative
waveform 119879119896(120591) of type (35) has at least one zero it follows
that it can be expressed in form (38)Let us also assume that 120591
0is position of nondegenerate
critical point Therefore 119879119896(1205910) = 0 implies 1198791015840
119896(1205910) = 0 and
11987910158401015840
119896(1205910) gt 0 According to (55) second derivative of 119879
119896(120591) at
1205910can be expressed as 11987910158401015840
119896(1205910) = 1 minus 120582
119896(1198962
minus 1) cos 120585 Since11987910158401015840
119896(1205910) gt 0 it follows immediately that
1 minus 120582119896(1198962
minus 1) cos 120585 gt 0 (138)
Let us further assume that 119879119896(120591) has exactly one zeroThe
problem of finding maximum absolute value of 1198861is con-
nected to the problem of finding maximum of the minimumfunction (see Section 21) If waveforms possess unique globalminimum at nondegenerate critical point then correspond-ing minimum function is a smooth function of parameters[13] Consequently assumption that 119879
119896(120591) has exactly one
zero at nondegenerate critical point leads to the conclusionthat coefficient 119886
1is differentiable function of 120591
0 First
derivative of 1198861(see (43)) with respect to 120591
0 taking into
account that 1205971205851205971205910= 119896 (see (50)) can be expressed in the
following factorized form
1205971198861
1205971205910
= sin 1205910[1 minus 120582
119896(1198962
minus 1) cos 120585] (139)
Mathematical Problems in Engineering 17
From (138) and (139) it is clear that 12059711988611205971205910= 0 if and only if
sin 1205910= 0 According toRemark 12 assumption that119879
119896(120591)has
exactly one zero implies 120582119896lt 1 From (51) (48) and 120582
119896lt 1
it follows that 1198861cos 1205910+ 1198871sin 1205910lt 0 which together with
sin 1205910= 0 implies that 119886
1cos 1205910lt 0 Assumption 119886
1le 0
together with relations 1198861cos 1205910lt 0 and sin 120591
0= 0 further
implies 1198861
= 0 and
1205910= 0 (140)
Insertion of 1205910= 0 into (38) leads to
119879119896(120591)
= [1 minus cos 120591]
sdot [1 minus (119896 minus 1) 120582119896cos 120585 minus 2120582
119896
119896minus1
sum
119899=1
(119896 minus 119899) cos (119899120591 + 120585)]
(141)
Substitution of 1205910= 0 into (45) and (46) yields 119886
119896= 120582119896cos 120585
and 119887119896
= minus120582119896sin 120585 respectively Replacing 120582
119896cos 120585 with
119886119896and 120582
119896cos(119899120591 + 120585) with (119886
119896cos 119899120591 + 119887
119896sin 119899120591) in (141)
immediately leads to (121)Furthermore 119886
119896= 120582119896cos 120585 119887
119896= minus120582
119896sin 120585 and (118)
imply that
120575 = minus120585 (142)
According to (38)ndash(40) and (142) it follows that (141) is non-negative if and only if
120582119896[(119896 minus 1) cos 120575 + 119896 sin (120575 minus 120575119896)
sin (120575119896)] lt 1 (143)
Notice that 119886119896= 120582119896cos 120575 implies that the following relation
holds
120582119896[(119896 minus 1) cos 120575 + 119896 sin (120575 minus 120575119896)
sin (120575119896)]
= minus119886119896+ 119896120582119896
sin 120575sin (120575119896)
cos(120575119896)
(144)
Finally substitution of (144) into (143) leads to 119896120582119896[sin 120575
sin(120575119896)] cos(120575119896) lt 1 + 119886119896 which proves that (121) holds
when 119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) lt 1 + 119886
119896
Apart from nonnegative waveforms with exactly one zeroat nondegenerate critical point in what follows we will alsoconsider other types of nonnegative waveforms with at leastone zero According to Proposition 9 and Remark 11 thesewaveforms can be described by (66)ndash(68) providing that 0 le|120585| le 120587
According to (35) 119879119896(0) ge 0 implies 1 + 119886
1+ 119886119896ge 0
Consequently 1198861le 0 implies that |119886
1| le 1 + 119886
119896 On the other
hand according to (123) |1198861| = 1 + 119886
119896holds for waveforms
of type (121) The converse is also true 1198861le 0 and |119886
1| =
1 + 119886119896imply 119886
1= minus1 minus 119886
119896 which further from (35) implies
119879119896(0) = 0 Therefore in what follows it is enough to consider
only nonnegativewaveformswhich can be described by (66)ndash(68) and 0 le |120585| le 120587 with coefficients 119886
119896and 119887119896satisfying
119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) ge 1 + 119886
119896
For prescribed coefficients 119886119896and 119887119896 the amplitude 120582
119896=
radic1198862119896+ 1198872119896of 119896th harmonic is also prescribed According to
Remark 15 (see also Remark 16) 120582119896is monotonically
decreasing function of 119909 = cos(120585119896) The value of 119909 can beobtained by solving (90) subject to the constraint cos(120587119896) le119909 le 1 Then 120582
1can be determined from (88) From (106) it
immediately follows that maximal absolute value of 1198861le 0
corresponds to 119902 = 0 which from (104) and (120) furtherimplies that
120575 = 1198961205910minus 120585 (145)
Furthermore 119902 = 0 according to (107) implies that waveformzeros are
1205910=(120575 + 120585)
119896 120591
1015840
0= 1205910minus2120585
119896=(120575 minus 120585)
119896 (146)
Substitution of 1205910= (120575 + 120585)119896 into (66) yields (122) which
proves that (122) holds when 119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) ge
1 + 119886119896
In what follows we prove that (121) also holds when119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) = 1 + 119886
119896 Substitution of 119886
119896=
120582119896cos 120575 into 119896120582
119896[sin 120575 sin(120575119896)] cos(120575119896) = 1 + 119886
119896leads to
120582119896[(119896 minus 1) cos 120575 + 119896 sin (120575 minus 120575119896)
sin (120575119896)] = 1 (147)
As we mentioned earlier relation (142) holds for all wave-forms of type (121) Substituting (142) into (147) we obtain
120582119896[(119896 minus 1) cos 120585 + 119896 sin (120585 minus 120585119896)
sin (120585119896)] = 1 (148)
This expression can be rearranged as
120582119896
119896 sin ((119896 minus 1) 120585119896)sin 120585119896
= 1 minus (119896 minus 1) 120582119896cos 120585 (149)
On the other hand for waveforms of type (122) according to(68) relations (148) and (149) also hold Substitution of 120591
0=
(120575 + 120585)119896 (see (145)) and (67) into (122) leads to
119879119896(120591)
= 120582119896[1 minus cos (120591 minus 120591
0)]
sdot [119896 sin ((119896 minus 1) 120585119896)
sin 120585119896minus 2
119896minus1
sum
119899=1
(119896 minus 119899) cos (119899 (120591 minus 1205910) + 120585)]
(150)
Furthermore substitution of (142) into (145) implies that1205910
= 0 Finally substitution of 1205910
= 0 and (149) into(150) leads to (141) Therefore (141) holds when 119896120582
119896[sin 120575
sin(120575119896)] cos(120575119896) = 1 + 119886119896 which in turn shows that (121)
holds when 119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) = 1 + 119886
119896 This
completes the proof
18 Mathematical Problems in Engineering
52 Nonnegative Waveforms with Maximal Absolute Valueof Coefficient 119886
1for 119896 = 3 Nonnegative waveform of type
(35) for 119896 = 3 is widely used in PA design (eg see [10])In this subsection we illustrate results of Section 51 for thisparticular case The case 119896 = 2 is presented in detail in [12]
Coefficients of fundamental harmonic of nonnegativewaveform of type (35) with 119896 = 3 and maximal absolutevalue of coefficient 119886
1le 0 for prescribed coefficients 119886
3and
1198873(1205823= radic11988623+ 11988723) according to (123) (124) (134) (94) and
(120) are equal to
1198861= minus1 minus 119886
3 119887
1= minus3119887
3 (151)
if 12058223le [(1 minus 2119886
3)3]3
1198861= minus1205821cos(120575
3) 119887
1= minus1205821sin(120575
3) (152)
where 1205821= 3(
3radic1205823minus 1205823) and 120575 = atan 2(119887
3 1198863) if [(1 minus
21198863)3]3
le 1205822
3le 1The line 1205822
3= [(1minus2119886
3)3]3 (see case 119896 = 3
in Figure 11) separates points representing waveforms withcoefficients satisfying (151) from points representing wave-forms with coefficients satisfying (152) Waveforms describedby (151) for 1205822
3lt [(1 minus 2119886
3)3]3 have exactly one zero at
1205910= 0 Waveforms described by (151) and (152) for 1205822
3= [(1 minus
21198863)3]3 also have zero at 120591
0= 0 These waveforms as a rule
have exactly two zeros However there are two exceptionsone related to the maximally flat nonnegative waveform withcoefficients 119886
1= minus98 119886
3= 18 and 119887
1= 1198873= 0 which
has only one zero and the other related to the waveform withcoefficients 119886
1= 0 119886
3= minus1 and 119887
1= 1198873= 0 which has three
zerosWaveforms described by (152) for [(1minus21198863)3]3
lt 1205822
3lt
1 have two zeros Waveforms with 1205823= 1 have only third
harmonic (fundamental harmonic is zero)Plot of contours of maximal absolute value of coefficient
1198861 1198861le 0 for prescribed coefficients 119886
3and 1198873is presented
in Figure 12 According to Remark 25 the waveform withmaximum absolute value of 119886
1le 0 is fully described with
the following coefficients 1198861
= minus2radic3 1198863
= radic39 and1198871= 1198873= 0 This waveform has two zeros at plusmn1205876
Two examples of nonnegative waveforms for 119896 = 3
and maximal absolute value of coefficient 1198861 1198861le 0 with
prescribed coefficients 1198863and 1198873are presented in Figure 13
One waveform corresponds to the case 12058223lt [(1 minus 2119886
3)3]3
(solid line) and the other to the case 12058223gt [(1 minus 2119886
3)3]3
(dashed line)Thewaveform represented by solid line has onezero and its coefficients are 119886
3= minus01 119887
3= 01 119886
1= minus09
and 1198871= minus03 Dashed line corresponds to the waveform
having two zeros with coefficients 1198863= minus01 119887
3= 03 119886
1=
minus08844 and 1198871= minus06460 (case 1205822
3gt [(1 minus 2119886
3)3]3)
6 Nonnegative Cosine Waveforms withat Least One Zero
Nonnegative cosine waveforms have proved to be of impor-tance for waveform modelling in PA design (eg see [10])In this section we consider nonnegative cosine waveforms
1
05
0
minus05
minus1
minus1 minus05 0 05 1
Coefficient a3
Coe
ffici
entb
3
02
04
06
08
10
11
Figure 12 Contours ofmaximal absolute value of coefficient 1198861 1198861le
0 as a function of 1198863and 1198873
2
1
0
0 1 2 3 4
Angle 120591120587
Wav
efor
ms
a3 = minus01 b3 = 01
a3 = minus01 b3 = 03
Figure 13 Nonnegative waveforms for 119896 = 3 and maximal absolutevalue of 119886
1 1198861le 0 with prescribed coefficients 119886
3and 1198873
containing fundamental and 119896th harmonic with at least onezero
Cosine waveform with dc component fundamental and119896th harmonic can be obtained from (35) by setting 119887
1= 119887119896=
0 that is
119879119896(120591) = 1 + 119886
1cos 120591 + 119886
119896cos 119896120591 (153)
In Section 61 we provide general description of non-negative cosine waveforms of type (153) with at least onezero We show that nonnegative cosine waveforms with atleast one zero coincide with nonnegative cosine waveformswith maximal absolute value of coefficient 119886
1for prescribed
coefficient 119886119896 In Section 62 we illustrate results of Section 61
for particular case 119896 = 3
61 Nonnegative Cosine Waveforms with at Least One Zerofor 119896 ge 2 Amplitudes of fundamental and 119896th harmonic
Mathematical Problems in Engineering 19
of cosine waveform of type (153) are 1205821= |1198861| and 120582
119896=
|119886119896| respectively According to (42) for nonnegative cosine
waveforms of type (153) the following relation holds
minus1 le 119886119896le 1 (154)
This explains why 119896th harmonic coefficient 119886119896in Proposi-
tion 26 goes through interval [minus1 1]Waveforms (153) with 119886
1ge 0 can be obtained from
waveforms with 1198861le 0 by shifting by 120587 and therefore with-
out loss of generality we can assume that 1198861le 0
Proposition 26 Each nonnegative cosine waveform of type(153) with 119886
1le 0 and at least one zero can be represented as
119879119896(120591) = [1 minus cos 120591] [1 minus (119896 minus 1) 119886
119896minus 2119886119896
119896minus1
sum
119899=1
(119896 minus 119899) cos 119899120591]
(155)
if minus1 le 119886119896le 1(119896
2
minus 1) or
119879119896(120591) = 119886
119896[1 minus cos (120591 minus 120591
0)] [1 minus cos (120591 + 120591
0)]
sdot [1198880+ 2
119896minus2
sum
119899=1
119888119899cos 119899120591]
(156)
where
119888119899=sin ((119896 minus 119899) 120591
0) cos 120591
0minus (119896 minus 119899) cos ((119896 minus 119899) 120591
0) sin 120591
0
sin31205910
(157)
119886119896=
sin 1205910
119896 sin (1198961205910) cos 120591
0minus cos (119896120591
0) sin 120591
0
(158)
100381610038161003816100381612059101003816100381610038161003816 le
120587
119896 (159)
if 1(1198962 minus 1) le 119886119896le 1
Remark 27 Identity [1minuscos(120591minus1205910)][1minuscos(120591+120591
0)] = [cos 120591
0minus
cos 120591]2 implies that (156) can be rewritten as
119879119896(120591) = 119886
119896[cos 120591
0minus cos 120591]2 [119888
0+ 2
119896minus2
sum
119899=1
119888119899cos 119899120591] (160)
Furthermore substitution of (157) into (160) leads to
119879119896(120591) = 119886
119896[cos 120591
0minus cos 120591]
sdot [(119896 minus 1) sin 119896120591
0
sin 1205910
minus 2
119896minus1
sum
119899=1
sin ((119896 minus 119899) 1205910)
sin 1205910
cos 119899120591]
(161)
Remark 28 All nonnegative cosine waveforms of type (153)with at least one zero and 119886
1le 0 except one of them can be
represented either in form (155) or form (156)This exceptionis maximally flat cosine waveform with 119886
1lt 0 which can be
obtained from (155) for 119886119896= 1(119896
2
minus 1) or from (156) for 1205910=
0 Maximally flat cosine waveform with 1198861lt 0 can also be
obtained from (70) by setting 1205910= 0 Furthermore setting
1205910= 0 in (71) leads to maximally flat cosine waveforms for
119896 le 4 and 1198861lt 0
Remark 29 Nonnegative cosine waveform of type (155) with1198861lt 0 and minus1 lt 119886
119896le 1(119896
2
minus 1) has exactly one zero at120591 = 0 Nonnegative cosine waveform described by (156) with1198861lt 0 and 1(1198962 minus 1) lt 119886
119896lt 1 has two zeros at plusmn120591
0 where
0 lt |1205910| lt 120587119896 For 119886
119896= ∓1 nonnegative cosine waveform
of type (153) reduces to 119879119896(120591) = 1 ∓ cos 119896120591 (clearly these two
waveforms both have 119896 zeros)
Remark 30 Transformation of (155) into an additive formleads to the following relation
1198861= minus1 minus 119886
119896 (162)
where minus1 le 119886119896le 1(119896
2
minus1) Similarly transformation of (156)leads to the following relation
1198861= minus119886119896
119896 sin 1198961205910
sin 1205910
(163)
where 119886119896is given by (158) 1(1198962minus1) le 119886
119896le 1 and |120591
0| le 120587119896
Notice that coefficients of maximally flat cosine waveformnamely 119886
119896= 1(119896
2
minus1) and 1198861= minus1198962
(1198962
minus1) satisfy relation(162) They also satisfy relation (163) for 120591
0= 0
Remark 31 Nonnegative cosine waveforms of type (153) withat least one zero coincide with nonnegative cosine waveformswith maximal absolute value of coefficient 119886
1for prescribed
coefficient 119886119896
In proving that Remark 31 holds notice that expression(155) can be obtained from (121) by setting 119887
119896= 0 Fur-
thermore if 119886119896
ge 0 then 120582119896
= 119886119896 which together
with 119887119896
= 0 and (118) implies 120575 = 0 In this case119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) le 1 + 119886
119896becomes 1198962119886
119896le 1 +
119886119896 On the other hand if 119886
119896lt 0 then 120582
119896= minus119886
119896 which
together with 119887119896= 0 and (118) implies |120575| = 120587 In this case
119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) le 1 + 119886
119896becomes 0 le 1 + 119886
119896
Therefore every nonnegative cosine waveform of type (155)has maximal absolute value of coefficient 119886
1for prescribed
coefficient 119886119896 when minus1 le 119886
119896le 1(119896
2
minus 1)Let us now show that expression (156) can be obtained
from (122) by setting 119887119896= 0 and 119886
119896gt 0 For waveforms
of type (122) according to (118) 119887119896= 0 and 119886
119896gt 0 imply
120575 = 0 and 120582119896= 119886119896 Substitution of 120582
119896= 119886119896and 120575 = 0 into
119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) ge 1+119886
119896leads to 119886
119896ge 1(119896
2
minus1)Furthermore substitution of 120575 = 0 into (145) yields 120591
0= 120585119896
Insertion of 120582119896= 119886119896 120575 = 0 and 120591
0= 120585119896 into (122) leads
to (156) Therefore every nonnegative cosine waveform oftype (156) has maximal absolute value of coefficient 119886
1for
prescribed coefficient 119886119896 when 1(1198962 minus 1) le 119886
119896le 1
Proof of Proposition 26 Let us start with nonnegative cosinewaveform of type (153) with 120582
119896= |119886119896| = 1 According to
Remark 7 120582119896= |119886119896| = 1 implies that 120582
1= |1198861| = 0
Substitution of 119886119896
= minus1 into (155) and using (A2) (seeAppendices) lead to 119879
119896(120591) = 1 minus cos 119896120591 Consequently (155)
holds for 119886119896= minus1 On the other hand substitution of 119886
119896= 1
into (158) yields |1205910| = 120587119896 Furthermore substitution of
119886119896= 1 and 120591
0= 120587119896 (or 120591
0= minus120587119896) into (156) along
with performing all multiplications and using (A2) leads to
20 Mathematical Problems in Engineering
119879119896(120591) = 1 + cos 119896120591 Consequently (156)ndash(158) hold for 119886
119896= 1
and |1205910| = 120587119896
It is easy to see that 120582119896= |119886119896| lt 1 and 119879
119896(1205910) = 0 for some
1205910imply 120582
1= |1198861| = 0 Therefore in what follows we assume
that |119886119896| = 1 and 119886
1lt 0
Cosine waveforms are even functions of 120591 Therefore ifnonnegative cosine waveform has exactly one zero it has to beeither at 0 or at 120587 On the other hand if nonnegative cosinewaveform with 119886
1= 0 has exactly two zeros then these zeros
are placed at plusmn1205910 such that 120591
0is neither 0 nor 120587
In order to prove that (155) holds for minus1 lt 119886119896le 1(1minus119896
2
)let us start by referring to the description (38) of nonnegativewaveformswith at least one zero As wementioned earlier fornonnegative cosine waveformwith exactly one zero (denotedby 1205910) it is either 120591
0= 0 or 120591
0= 120587 Therefore in both cases
sin 1205910= 0 Substitution of sin 120591
0= 0 into (43) together with
1198861
= 0 and 120582119896= |119886119896| lt 1 leads to
1205910= 0 (164)
Clearly 1205910= 0 119887
1= 0 and 119887
119896= 0 according to (44) and (46)
imply 120582119896sin 120585 = 0 Since 120582
119896= |119886119896| it follows that |119886
119896| sin 120585 = 0
also holds which further implies 120582119896= 119886119896= 0 or sin 120585 =
0 In the case when 120582119896= 119886119896= 0 from (164) and (43) we
obtain 1198861= minus1 which further implies that 119879
119896(120591) = 1 minus cos 120591
Consequently (155) holds for 119886119896= 0 In the case when sin 120585 =
0 from (164) and (45) we obtain 119886119896= 120582119896if 120585 = 0 or 119886
119896= minus120582119896
if 120585 = 120587 Relations 119886119896= 120582119896and 120585 = 0 according to (40) imply
that 0 le 119886119896le 1(1 minus 119896
2
) Substitution of 120585 = 0 120582119896= 119886119896 and
(164) into (38) leads to (155) which proves that (155) holdsfor 0 le 119886
119896le 1(1 minus 119896
2
) On the other hand relations 119886119896=
minus120582119896and 120585 = 120587 according to (40) imply that minus1 lt 119886
119896le 0
Substitution of 120585 = 120587 120582119896= minus119886119896 and (164) into (38) also leads
to (155) which proves that (155) also holds for minus1 lt 119886119896le 0
Consequently (155) holds for minus1 lt 119886119896le 1(1 minus 119896
2
)In what follows we first prove that (156)-(157) hold for
1(1 minus 1198962
) lt 119886119896lt 1 For this purpose let us start with non-
negative waveforms with two zeros described by (66) As wementioned before nonnegative cosine waveforms with twozeros have zeros at 120591
0and minus120591
0 such that 120591
0= 0 and 120591
0= 120587
Relations 1198861
lt 0 and 1198871
= 0 according to (84) implycos(1205910minus 120585119896) = 1 and therefore
120585
119896= 1205910 (165)
From 120585119896 = 1205910and 0 lt |120585| lt 120587 it follows that 0 lt |120591
0| lt
120587119896 Insertion of 120585119896 = 1205910into (45) yields 119886
119896= 120582119896 Relations
119886119896= 120582119896and (82) imply that 1(1 minus 1198962) lt 119886
119896lt 1 Substitution
of 120582119896= 119886119896and 120585119896 = 120591
0into (66)ndash(68) leads to (156)ndash(158)
which proves that (156)ndash(158) hold for 1(1 minus 1198962
) lt 119886119896lt 1
and 0 lt |1205910| lt 120587119896
Finally substitution of 119886119896= 1(1 minus 119896
2
) and 1205910= 0 into
(161) leads to
119879119896(120591) =
[1 minus cos 120591](1 minus 1198962)
[119896 (119896 minus 1) minus 2
119896minus1
sum
119899=1
(119896 minus 119899) cos 119899120591]
(166)
Waveform (166) coincides with waveform (155) for 119886119896
=
1(1 minus 1198962
) which in turn proves that (156) holds for 119886119896=
1(1 minus 1198962
) and 1205910= 0 This completes the proof
62 Nonnegative CosineWaveforms with at Least One Zero for119896 = 3 In this subsection we consider nonnegative cosinewaveforms with at least one zero for 119896 = 3 (for case 119896 = 2
see [12])Cosine waveform with fundamental and third harmonic
reads
1198793(120591) = 1 + 119886
1cos 120591 + 119886
3cos 3120591 (167)
For 1198861le 0 and minus1 le 119886
3le 18 according to (155) non-
negative cosine waveform of type (167) with at least one zerocan be expressed as
1198793(120591) = (1 minus cos 120591) [1 minus 2119886
3(1 + 2 cos 120591 + cos 2120591)] (168)
From 1198793(120591 + 120587) = 2 minus 119879
3(120591) it immediately follows that for
1198861ge 0 and minus18 le 119886
3le 1 119879
3(120591) can be expressed as
1198793(120591) = (1 + cos 120591) [1 + 2119886
3(1 minus 2 cos 120591 + cos 2120591)] (169)
For 1198861le 0 and 18 le 119886
3le 1 from (158) it follows that 119886
3=
[8cos31205910]minus1 This relation along with (160) and (157) further
implies that 1198793(120591) can be expressed as
1198793(120591) =
[cos 1205910minus cos 120591]2 [2 cos 120591
0+ cos 120591]
2cos31205910
(170)
providing that |1205910| le 1205873 From 119879
3(120591 + 120587) = 2 minus 119879
3(120591) it
follows that (170) also holds for 1198861ge 0 and minus1 le 119886
3le minus18
providing that 1205910isin [21205873 41205873]
Maximally flat nonnegative cosinewaveformof type (167)with 119886
1lt 0 (minimum at 120591
0= 0) reads 119879
3(120591) = [1 minus
cos 120591]2[1 + (12) cos 120591] Dually maximally flat nonnegativecosine waveform with 119886
1gt 0 (minimum at 120591
0= 120587) reads
1198793(120591) = [1 + cos 120591]2[1 minus (12) cos 120591]In what follows we provide relations between coefficients
1198861and 1198863of nonnegative cosine waveforms of type (167) with
at least one zeroFor 1198861le 0 conversion of (168) into an additive form
immediately leads to the following relation
1198861= minus1 minus 119886
3for minus 1 le 119886
3le1
8 (171)
Conversion of (170) into an additive form leads to 1198861
=
minus31198863(2 cos 2120591
0+ 1) which can be also expressed as 119886
1=
minus31198863(4cos2120591
0minus 1) For 119886
1le 0 relations |120591
0| le 1205873 119886
1=
minus31198863(4cos2120591
0minus 1) and 119886
3= [8cos3120591
0]minus1 lead to
1198861= minus3 [ 3radic119886
3minus 1198863] for 1
8le 1198863le 1 (172)
Similarly for 1198861ge 0 conversion of (169) into an additive form
leads to the following relation
1198861= 1 minus 119886
3for minus
1
8le 1198863le 1 (173)
Mathematical Problems in Engineering 21
1
15
15
05
0
minus05
minus15minus15
minus1
minus1 minus05 0 05 1
Coefficient a1
Coe
ffici
enta
3 a1= minus
3(3radica3minus a3)
a1 =
minus1 minus a3
a1 =
1 minusa3
a1 =3(3radic|a3
| +a3)
Figure 14 Parameter space of cosine waveforms for 119896 = 3
For waveform of type (170) with 1198861ge 0 relations 120591
0isin [21205873
41205873] 1198861= minus3119886
3(4cos2120591
0minus 1) and 119886
3= [8cos3120591
0]minus1 lead to
1198861= 3 [
3radic10038161003816100381610038161198863
1003816100381610038161003816 + 1198863] for minus 1 le 1198863le minus
1
8 (174)
Every cosine waveform of type (167) corresponds to apair of real numbers (119886
1 1198863) and vice versa Points (119886
1 1198863)
in grey area in Figure 14 correspond to nonnegative cosinewaveforms for 119896 = 3 The points at the boundary of grey areacorrespond to nonnegative cosinewaveformswith at least onezero A number of shapes of nonnegative cosine waveformswith 119896 = 3 and at least one zero plotted on interval [minus120587 120587]are also presented in Figure 14 The boundary of grey area inFigure 14 consists of four line segments described by relations(171)ndash(174) The common point of line segments (172) and(173) is cusp point with coordinates 119886
1= 0 and 119886
3= 1
Another cusp point with coordinates 1198861= 0 and 119886
3= minus1
is the common point of line segments (171) and (174) Thecommon point of line segments (171)-(172) has coordinates(minus98 18) and common point of line segments (173)-(174)has coordinates (98 minus18) These points are representedby white circle dots and they correspond to maximallyflat cosine waveforms (eg see [21]) White triangle dotswith coordinates (2radic3 minusradic39) and (minus2radic3radic39) refer tothe nonnegative cosine waveforms with maximum value ofamplitude of fundamental harmonic
7 Four Case Studies of Usage of NonnegativeWaveforms in PA Efficiency Analysis
In this section we provide four case studies of usage ofdescription of nonnegative waveforms with fundamental and119896th harmonic in PA efficiency analysis In first two casestudies to be presented in Section 71 voltage is nonnegativewaveform with fundamental and second harmonic with atleast one zero In remaining two case studies to be consideredin Section 72 voltage waveform contains fundamental andthird harmonic
i
v Load
Vdc
Idc
Lch
vL
iL
Cb
+ +in
Figure 15 Generic PA circuit diagram
Let us consider generic PA circuit diagram as shown inFigure 15 We assume here that voltage and current wave-forms at the transistor output are
V (120579) = 1 + 1198861V cos 120579 + 1198871V sin 120579 + 119886119896V cos 119896120579 + 119887119896V sin 119896120579
119894 (120579) = 1 + 1198861119894cos 120579 +
infin
sum
119899=2
119886119899119894cos 119899120579
(175)
where 120579 stands for 120596119905 Both waveforms are normalized in thesense that dc components of voltage and current are 119881dc =
1 and 119868dc = 1 respectively Under assumption that blockingcapacitor 119862
119887behaves as short-circuit at the fundamental and
higher harmonics current and voltage waveforms at the loadare
V119871(120579) = 119886
1V cos 120579 + 1198871V sin 120579 + 119886119896V cos 119896120579 + 119887119896V sin 119896120579
119894119871(120579) = minus119886
1119894cos 120579 minus
infin
sum
119899=2
119886119899119894cos 119899120579
(176)
In terms of coefficients of voltage and current waveforms theload impedance at fundamental harmonic is 119911
1= minus(119886
1V minus
1198951198871V)1198861119894 whereas load impedance at 119896th harmonic is 119911
119896=
minus(119886119896Vminus119895119887119896V)119886119896119894 All other harmonics are short-circuited (119911
119899=
0 for 119899 = 1 and 119899 = 119896) Time average output power of PA (egsee [10]) with waveform pair (175) at fundamental frequencycan be expressed as
1198751= minus
11988611198941198861V
2 (177)
For normalized waveforms (175) with 119881dc = 1 and 119868dc = 1dc power is 119875dc = 1 Consequently PA efficiency 120578 = 119875
1119875dc
(eg see [10 26]) is equal to
120578 = minus11988611198941198861V
2 (178)
Thus time average output power 1198751of PA with pair of nor-
malized waveform (175) is equal to efficiency (178)Power utilization factor (PUF) is defined [26] as ldquothe
ratio of power delivered in a given situation to the power
22 Mathematical Problems in Engineering
delivered by the same device with the same supply voltagein Class A moderdquo Since the output power in class-A modeis 1198751class-A = max[V(120579)] sdotmax[119894(120579)]8 (eg see [9]) it follows
that power utilization factor PUF = 11987511198751class-A for PA with
pair of normalized waveforms (175) can be expressed as
PUF =8120578
max [V (120579)] sdotmax [119894 (120579)] (179)
71 NonnegativeWaveforms for 119896 = 2 in PAEfficiencyAnalysisIn this subsection we provide two case studies of usage ofdescription of nonnegative waveforms with fundamental andsecond harmonic (119896 = 2) in PA efficiency analysis For moreexamples of usage of descriptions of nonnegative waveformswith fundamental and second harmonic in PA efficiencyanalysis see [12]
Case Study 71 In this case study we consider efficiency of PAfor given second harmonic impedance providing that voltageis nonnegative waveform with fundamental and second har-monic and current is ldquohalf-sinerdquo waveform frequently used inefficiency analysis of classical PA operation (eg see [10])
Standard model of current waveform for classical PAoperation has the form (eg see [10 26])
119894119863(120579) =
119868119863[cos 120579 minus cos(120572
2)] |120579| le
120572
2
0120572
2le |120579| le 120587
(180)
where 120572 is conduction angle and 119868119863
gt 0 Since 119894119863(120579) is
even function it immediately follows that its Fourier seriescontains only dc component and cosine terms
119894119863(120579) = 119868dc +
infin
sum
119899=1
119868119899cos 119899120579 (181)
The dc component of the waveform (180) is
119868dc =119868119863120572
2120587[sinc(120572
2) minus cos(120572
2)] (182)
where sinc119909 = (sin119909)119909 The coefficient of the fundamentalharmonic component reads
1198681=119868119863120572
2120587(1 minus sinc120572) (183)
and the coefficient of 119899th harmonic component can bewrittenin the form
119868119899=119868119863
119899120587[sin ((119899 minus 1) 1205722)
(119899 minus 1)minussin ((119899 + 1) 1205722)
(119899 + 1)] 119899 ge 2
(184)
For ldquohalf-sinerdquo current waveform conduction angle is equalto 120587 (class-B conduction angle) According to (182) thisfurther implies that 119868dc = 119868
119863120587 To obtain normalized form of
waveform (180) we set 119868dc = 1which implies that 119868119863= 120587 Fur-
thermore substitution of 120572 = 120587 and 119868119863= 120587 in (180) leads to
119894 (120579) =
120587 cos 120579 |120579| lt120587
2
0120587
2lt |120579| le 120587
(185)
Similarly substitution of 119868119863= 120587 and 120572 = 120587 into (183) and
(184) leads to the coefficients of waveform (185) Coefficientsof fundamental and second harmonic respectively are
1198861119894=120587
2 119886
2119894=2
3 (186)
On the other hand voltage waveform of type (35) for 119896 =2 reads
V (120579) = 1 + 1198861V cos 120579 + 1198871V sin 120579 + 1198862V cos 2120579 + 1198872V sin 2120579
(187)
This waveform contains only fundamental and second har-monic and therefore all harmonics of order higher thantwo are short-circuited (119911
119899= 0 for 119899 gt 2) For current
voltage pair (185) and (187) load impedance at fundamentalharmonic is 119911
1= minus(119886
1V minus 1198951198871V)1198861119894 whereas load impedance
at second harmonic is 1199112= minus(119886
2Vminus1198951198872V)1198862119894 According to ourassumption the load is passive and therefore Re119911
1 gt 0 and
Re1199112 ge 0 which further imply 119886
11198941198861V lt 0 and 119886
21198941198862V le 0
respectivelyIt is easy to see that problem of findingmaximal efficiency
of PA with current-voltage pair (185) and (187) for prescribedsecond harmonic impedance can be reduced to the problemof finding voltage waveform of type (187) with maximal coef-ficient |119886
1V| for prescribed coefficients of second harmonic(see Section 5)
The following algorithm (analogous to Algorithm 22presented in [12]) provides the procedure for calculation ofmaximal efficiency with current-voltage pair (185) and (187)for prescribed second harmonic impedance The definitionof function atan 2(119910 119909) which appears in the step (iii) of thefollowing algorithm is given by (105)
Algorithm 32 (i) Choose 1199112= 1199032+1198951199092such that |119911
2| le 1|119886
2119894|
(ii) calculate 1198862V minus 1198951198872V = minus119911
21198862119894and 120582
2V = radic11988622V + 1198872
2V(iii) if 2120582
2V le 1 minus 1198862V then calculate 119886
1V = minus1 minus 1198862V and
1198871V = minus2119887
2V else calculate 1205821V = radic81205822V(1 minus 1205822V) 1205790V minus 120585V2 =
(12)atan2(1198872V 1198862V) 1198861V = minus120582
1V cos(1205790V minus 120585V2) and 1198871V =
minus1205821V sin(1205790V minus 120585V2)(iv) calculate efficiency 120578 = minus119886
11198941198861V2
(v) calculate 1199111= minus(119886
1V minus 1198951198871V)1198861119894 and 1199112119899 = 1199112Re119911
1
In this case study coefficients of fundamental and secondharmonic of current waveform are given by (186) Maximalefficiency of PA associated with the waveform pair (185)and (187) as a function of normalized second harmonicimpedance 119911
2119899= 1199112Re119911
1 is presented in Figure 16(a)
As can be seen from Figure 16(a) efficiency of 078 isachieved at the edge of Smith chart where second harmonicimpedance has small resistive part Corresponding PUFcalculated according to (179) is presented in Figure 16(b)Peak efficiency 120578 = 1205874 = 07854 and peak value of PUF = 1
are attained when second harmonic is short-circuited (whichcorresponds to ideal class-B operation [10 26])
For example for second harmonic impedance 1199112= 01 minus
11989505 and currentwaveform (185) fromAlgorithm32 it followsthat 2120582
2V le 1 minus 1198862V Furthermore according to step (iii)
of above algorithm maximal efficiency of PA is attained
Mathematical Problems in Engineering 23
+j2
+j5
+j1
+j2
+j5
minusj2
minusj5
minusj1
minusj2
minusj5
infin0
078
02
05
1 2 5
075 07 065
06 05
075
07
065
06
05
120578 lt 05
(a)
099
095
091
083 075
067
0 59 051
051
099
095
091
083
075 0
67 059
+j5
+j1
+j2
+j5
minusj5
minusj1
minusj2
minusj5
infin
+j2
minusj2
0 02
05
1 2 5
120578 lt 05
(b)
Figure 16 (a) Contours of maximal efficiency of PA and (b) contours of corresponding PUF associated with the waveform pair (185) and(187) as functions of normalized second harmonic impedance 119911
2119899= 1199112Re119911
1
2
3
4
1
0
2 3 410
Wav
efor
ms
Angle 120579120587
i(120579)
(120579)
Figure 17 Waveform pair (185) and (187) that provides maximalefficiency for 119911
2= 01 minus 11989505
with voltage waveform of type (187) with coefficients 1198862V =
minus00667 1198872V = minus03333 119886
1V = minus09333 and 1198871V = 06667 (see
Figure 17) Corresponding efficiency PUF and normalizedsecond harmonic impedance are 120578 = 07330 PUF = 07572and 1199112119899= 01683 minus 11989508415 respectively
On the other hand for second harmonic impedance 1199112=
01 minus 11989508 and current waveform (185) from Algorithm 32 itfollows that 2120582
2V gt 1 minus 1198862V Then according to step (iii) of
above algorithm maximal efficiency is attained with voltagewaveform of type (187) with coefficients 119886
2V = minus00667 1198872V =
minus05333 1198861V = minus09333 and 119887
1V = 10572 (see Figure 18)Efficiency PUF andnormalized secondharmonic impedanceare 120578 = 07330 PUF = 06332 and 119911
2119899= 01683 minus 11989513465
respectively
2
3
4
1
0
2 3 410
Wav
efor
ms
Angle 120579120587
i(120579)
(120579)
Figure 18 Waveform pair (185) and (187) that provides maximalefficiency for 119911
2= 01 minus 11989508
Case Study 72 As another case study let us consider the effi-ciency of PA providing that current waveform is nonnegativecosine waveform up to third harmonic with maximum valueof amplitude of fundamental harmonic [22] (see also [8])
119894 (120579) = 1 +1 + radic5
2cos 120579 + 2radic5
5cos 2120579 + 5 minus radic5
10cos 3120579
(188)
and voltage waveform is nonnegative waveform of type(187) Load impedances at fundamental second and thirdharmonic are 119911
1= minus(119886
1Vminus1198951198871V)1198861119894 1199112 = minus(1198862Vminus1198951198872V)1198862119894 and
1199113= 0 respectively According to our assumption the load
is passive and therefore Re1199111 gt 0 and Re119911
2 ge 0 which
further imply 11988611198941198861V lt 0 and 119886
21198941198862V le 0 respectively
Because current waveform (188) contains only cosineterms and voltage waveform is the same as in previous case
24 Mathematical Problems in Engineering
08
075
07
06
06
065
065
05
05
075 0
7
120578 lt 05
+j2
+j5
+j1
+j2
+j5
minusj2
minusj5
minusj1
minusj2
minusj5
infin0
02
05
1 2 5
(a)
085
08 0
75 07 0
6
05 04
08
075
0405
060
7
120578 lt 05
+j2
+j5
+j1
+j2
+j5
minusj2
minusj5
minusj1
minusj2
minusj5
infin0 02
05
1 2 5
(b)
Figure 19 (a) Contours of maximal efficiency of PA and (b) contours of corresponding PUF associated with the waveform pair (187)-(188)as functions of normalized second harmonic impedance 119911
2119899= 1199112Re119911
1
study the procedure for calculation of maximal efficiency ofPA with waveform pair (187)-(188) is the same as presentedin Algorithm 32 In this case study the coefficients of funda-mental and second harmonic of current waveform are 119886
1119894=
(1 + radic5)2 and 1198862119894= 2radic55 respectively
Maximal efficiency of PA associated with the waveformpair (187)-(188) as a function of normalized secondharmonicimpedance 119911
2119899= 1199112Re119911
1 is presented in Figure 19(a)
Efficiency of 08 is achieved at the edge of Smith chart wheresecond harmonic impedance has small resistive part Thetheoretical upper bound 120578 = (1 + radic5)4 asymp 08090 isattained when second harmonic is short-circuitedWhen thisupper bound is reached both second and third harmonic areshort-circuited which implies that we are dealing with finiteharmonic class-C [6 8] or dually when current and voltageinterchange their roles with finite harmonic inverse class-C[6 9] Corresponding PUF calculated according to (179) ispresented in Figure 19(b) Peak value of PUF asymp 08541 isattained when second harmonic is short-circuited
For example for second harmonic impedance 1199112
=
007 minus 11989504 and current waveform (188) from Algorithm 32 itfollows that 2120582
2V le 1 minus 1198862V Furthermore according to step
(iii) of Algorithm 32 maximal efficiency of PA is attainedwith voltage waveform of type (187) with coefficients 119886
2V =
minus00626 1198872V = minus03578 119886
1V = minus09374 and 1198871V = 07155 (see
Figure 20) Corresponding efficiency PUF and normalizedsecond harmonic impedance are 120578 = 07584 PUF = 06337and 1199112119899= 01208 minus 11989506904 respectively
On the other hand for 1199112= 005 minus 11989507 and current
waveform (187) it follows that 21205822V gt 1minus119886
2VThen accordingto step (iii) of Algorithm 32 themaximal efficiency is attainedwith voltage waveform of type (187) with coefficients 119886
2V =
minus00447 1198872V = minus06261 119886
1V = minus09318 and 1198871V = 10007 (see
Figure 21) Efficiency PUF and normalized second harmonic
2
3
4
1
0
2 3 410
Wav
efor
ms
Angle 120579120587
i(120579)
(120579)
Figure 20 Waveform pair (187)-(188) that provides maximal effi-ciency for 119911
2= 007 minus 11989504
impedance are 120578 = 07538 PUF = 05314 and 1199112119899= 00868minus
11989512156 respectively
72 Nonnegative Waveforms for 119896 = 3 in PA EfficiencyAnalysis In this subsection we provide another two casestudies of usage of description of nonnegative waveforms inPA efficiency analysis this time with fundamental and thirdharmonic (119896 = 3)
Case Study 73 Let us consider current-voltage pair such thatvoltage is nonnegative waveform with fundamental and thirdharmonic
V (120579) = 1 + 1198861V cos 120579 + 1198871V sin 120579 + 1198863V cos 3120579 + 1198873V sin 3120579
(189)
Mathematical Problems in Engineering 25
2
3
4
1
0
2 3 410
Wav
efor
ms
Angle 120579120587
i(120579)
(120579)
Figure 21 Waveform pair (187)-(188) that provides maximal effi-ciency for 119911
2= 005 minus 11989507
and current is nonnegative cosine waveform given by (188)Load impedances at fundamental second and third har-monic are 119911
1= minus(119886
1V minus 1198951198871V)1198861119894 1199112 = 0 and 119911
3= minus(119886
3V minus
1198951198873V)1198863119894 respectively According to our assumption the load
is passive and therefore Re1199111 gt 0 and Re119911
3 ge 0 which
further imply 11988611198941198861V lt 0 and 119886
31198941198863V le 0
In this subsection we consider the problem of findingmaximal efficiency of PA with waveform pair (188)-(189)for given third harmonic impedance As we mentionedearlier problem of finding maximal efficiency of PA withcurrent-voltage pair (188)-(189) for prescribed third har-monic impedance can be reduced to the problem of findingvoltage waveform of type (189) withmaximal coefficient |119886
1V|for prescribed coefficients of third harmonic (see Section 52)
The following algorithm provides the procedure forcalculation of maximal efficiency with current-voltage pair(188)-(189) The definition of function atan 2(119910 119909) whichappears in step (iii) of the following algorithm is given by(105)
Algorithm 33 (i) Choose 1199113= 1199033+1198951199093such that |119911
3| le 1|119886
3119894|
(ii) calculate 1198863V minus 1198951198873V = minus119911
31198863119894and 120582
3V = radic11988623V + 1198872
3V(iii) if 271205822
3V le (1 minus 21198863V)3 then calculate 119886
1V = minus1 minus 1198863V
and 1198871V = minus3119887
3V else calculate 1205821V = 3(3radic1205823V minus 120582
3V) 1205790V minus120585V3 = (13)atan 2(119887
3V 1198863V) 1198861V = minus1205821V cos(1205790V minus 120585V3) and
1198871V = minus120582
1V sin(1205790V minus 120585V3)(iv) calculate efficiency 120578 = minus119886
11198941198861V2
(v) calculate 1199111= minus(119886
1V minus 1198951198871V)1198861119894 and 1199113119899 = 1199113Re119911
1
In this case study coefficients of fundamental and thirdharmonic of current waveform are 119886
1119894= (1 + radic5)2 and
1198863119894= (5 minus radic5)10 respectively For the waveform pair (188)-
(189) maximal efficiency of PA as a function of normalizedthird harmonic impedance 119911
3119899= 1199113Re119911
1 is presented in
Figure 22 Efficiency of 08 is reached when third harmonicimpedance has small resistive part Peak efficiency 120578 = (1 +
radic5)4 asymp 08090 is achieved when third harmonic is short-circuited
+j2
+j5
+j1
+j2
+j5
minusj2
minusj5
minusj1
minusj2
minusj5
infin0
08
08
075
075
07
07
06
05
02
05
1 2
Figure 22 Contours of maximal efficiency of PA associated withthe waveform pair (188)-(189) as a function of normalized thirdharmonic impedance 119911
3119899= 1199113Re119911
1
For the present case study in what follows we show thatpower utilization factor is proportional to efficiency For volt-age waveform of type (189) it is easy to see that V(120579 + 120587) =
2minusV(120579) holdsThis relation along with the fact that waveformV(120579) that provides maximal efficiency has at least one zeroimplies that max[V(120579)] = 2 On the other hand currentwaveform (188) is cosine waveform with positive coefficientsand therefore max[119894(120579)] = 119894(0) = 2 + 4radic5 Consequentlyaccording to (179) the following relation holds
PUFCase study 73 = 2 (5 minus 2radic5) 120578 = 10557120578 (190)
Clearly the ratio PUF120578 is constant and therefore in this casestudy PUF can be easily calculated from the correspondingefficiency Accordingly peak efficiency and peak value ofPUFCase study 73 = 3radic52 minus 52 = 08541 are attained forthe same voltage waveform (when third harmonic is short-circuited)
In the first example current waveform (188) and 1199113=
02 minus 11989505 imply that 2712058223V le (1 minus 2119886
3V)3 Then according
to Algorithm 33 the voltage waveform of type (189) thatprovides maximal efficiency has the following coefficients1198863V = minus00553 119887
3V = minus01382 1198861V = minus09447 and 119887
1V =
04146 (see Figure 23) Efficiency PUF and normalized thirdharmonic impedance are 120578 = 07643 PUF = 08069 and1199113119899= 03425 minus 11989508564 respectivelyIn the second example current waveform (188) and 119911
3=
01 minus 11989511 imply that 2712058223V gt (1 minus 2119886
3V)3 Then according
to Algorithm 33 the voltage waveform of type (189) thatprovides maximal efficiency has the following coefficients1198863V = minus00276 119887
3V = minus03040 1198861V = minus09391 and 119887
1V =
05807 (see Figure 24) Efficiency PUF and normalized thirdharmonic impedance are 120578 = 07598 PUF = 08021 and1199113119899= 01723 minus 11989518952 respectively
26 Mathematical Problems in Engineering
2
3
4
1
0
2 3 410
Wav
efor
ms
Angle 120579120587
i(120579)
(120579)
Figure 23 Waveform pair (188)-(189) that provides maximal effi-ciency for 119911
3= 02 minus 11989505
2
3
4
1
0
2 3 410
Wav
efor
ms
Angle 120579120587
i(120579)
(120579)
Figure 24 Waveform pair (188)-(189) that provides maximal effi-ciency for 119911
3= 01 minus 11989511
Case Study 74 In this case study let us consider current-voltage pair where current is normalized waveform of type(180) with conduction angle 120572 = 115120587 (207∘) and voltageis nonnegative waveform of type (189) Substitution of 120572 =
115120587 and 119868dc = 1 into (182) leads to 119868119863
= 22535Furthermore substitution of 120572 = 115120587 and 119868
119863= 22535 into
(180) leads to
119894 (120579)
=
22535 [cos 120579 minus cos(1151205872
)] |120579| le115120587
2
0115120587
2le |120579| le 120587
(191)
Similarly substitution of 120572 = 115120587 and 119868119863
= 22535 into(183) and (184) for 119899 = 3 yields coefficients of fundamentaland third harmonic of waveform (191)
1198861119894= 14586 119886
3119894= minus01026 (192)
Because current waveform (191) contains only cosineterms and voltage waveform is the same as in previous case
+j2
+j5
+j1
+j2
+j5
minusj2
minusj5
minusj1
minusj2
minusj5
infin0 02
05
1 2
074
076
08
084
082
078
076
074
Figure 25 Contours of maximal efficiency of PA associated withthe waveform pair (189) and (191) as a function of normalized thirdharmonic impedance 119911
3119899= 1199113Re119911
1
study the procedure for calculation of maximal efficiencyof PA with waveform pair (189)ndash(191) is the same as thatpresented in Algorithm 33 In this case study the coefficientsof fundamental and third harmonic of current waveform aregiven by (192)
For the waveform pair (189) and (191) maximal efficiencyof PA as a function of normalized third harmonic impedance1199113119899= 1199113Re119911
1 is presented in Figure 25 Efficiency of 084
is obtained in vicinity of 1199113119899= 23685 (corresponding to 119911
3=
18750) Peak efficiency 120578 asymp 08421 is achieved for voltagewaveform of type (189) with coefficients 119886
1V = minus2radic3 1198863V =
radic39 and 1198871V = 1198873V = 0
In the course of finding power utilization factor noticethat currentwaveformof type (191) attains itsmaximumvaluefor 120579 = 0 Insertion of max[119894(120579)] = 119894(0) = 278 andmax[V(120579)] = 2 for voltage waveform of type (189) into (179)leads to
PUFCase study 74 = 1439120578 (193)
Again the ratio PUF120578 is constant and PUF can be easilycalculated from the corresponding efficiency Accordinglypeak value of PUFCase study 74 asymp 12118 and peak efficiencyare attained for the same voltage waveform
In the first example current waveform (191) and 1199113=
1 minus 11989502 imply that 2712058223V le (1 minus 2119886
3V)3 Then according to
Algorithm 33 voltage waveform of type (189) which providesmaximal efficiency has coefficients 119886
3V = 01026 1198873V =
00205 1198861V = minus11026 and 119887
1V = minus00616 (see Figure 26)Efficiency PUF and normalized third harmonic impedanceare 120578 = 08042 PUF = 11572 and 119911
3119899= 13228 minus 11989502646
respectivelyIn second example current waveform (191) and 119911
3=
15 minus 11989512 imply that 2712058223V gt (1 minus 2119886
3V)3 Then according to
Algorithm 33 voltage waveform of type (189) which providesmaximal efficiency has coefficients 119886
3V = 01540 1198873V =
Mathematical Problems in Engineering 27
2
3
4
1
0
2 3 410
Wav
efor
ms
Angle 120579120587
i(120579)
(120579)
Figure 26 Waveform pair (189) and (191) that provides maximalefficiency for 119911
3= 1 minus 11989502
2
3
4
1
0
2 3 410
Wav
efor
ms
Angle 120579120587
i(120579)
(120579)
Figure 27 Waveform pair (189) and (191) that provides maximalefficiency for 119911
3= 15 minus 11989512
01232 1198861V = minus11255 and 119887
1V = minus02575 (see Figure 27)Efficiency PUF and normalized third harmonic impedanceare 120578 = 08208 PUF = 11812 and 119911
3119899= 19439 minus 11989515552
respectively
8 Conclusion
In this paper we consider a problem of finding general de-scriptions of various classes of nonnegative waveforms withfundamental and 119896th harmonic These classes include non-negative waveforms with at least one zero nonnegative wave-forms with maximal amplitude of fundamental harmonic forprescribed amplitude of 119896th harmonic nonnegative wave-forms withmaximal coefficient of cosine part of fundamentalharmonic for prescribed coefficients of 119896th harmonic andnonnegative cosine waveforms with at least one zero Mainresults are stated in six propositions (Propositions 1 6 918 22 and 26) four corollaries (Corollaries 2ndash5) twentyremarks and three algorithms Four case studies of usageof closed form descriptions of nonnegative waveforms in PAefficiency analysis are considered in detail in Section 7
Appendices
Here we provide a list of finite sums of trigonometric func-tions used in this paper (Appendix A) and brief account ofthe Chebyshev polynomials (Appendix B)
A List of Some Finite Sums ofTrigonometric Functions
Dirichlet kernel (eg see [27]) is as follows
119863119896minus1
(120591) = 1 + 2
119896minus1
sum
119899=1
cos 119899120591 = sin ((2119896 minus 1) 1205912)sin (1205912)
(A1)
Fejer kernel (eg see [27]) can be expressed in the fol-lowing equivalent forms
119865119896minus1
(120591) =1
119896
119896minus1
sum
119899=0
119863119899(120591) = 1 +
2
119896
119896minus1
sum
119899=1
(119896 minus 119899) cos 119899120591
=(1 minus cos 119896120591)119896 (1 minus cos 120591)
(A2)
Lagrangersquos trigonometric identity (eg see [28]) is as fol-lows
1198781(120591) =
119896minus1
sum
119899=1
sin 119899120591 = sin (1198961205912) sin ((119896 minus 1) 1205912)sin (1205912)
(A3)
In what follows we show that the following three trigono-metric identities also hold
2
119896minus1
sum
119899=1
(119896 minus 119899) sin 119899120591 = 119896 sin 120591 minus sin 1198961205911 minus cos 120591
(A4)
119896minus1
sum
119899=1
cos (119896 minus 2119899) 120591 = sin (119896 minus 1) 120591sin 120591
(A5)
119896minus1
sum
119899=1
119899 (119896 minus 119899) cos (119896 minus 2119899) 120591
=sin (119896120591) cos 120591 minus 119896 cos (119896120591) sin 120591
2sin3120591
(A6)
Denote 1198782(120591) = 2sum
119896minus1
119899=1(119896 minus 119899) sin 119899120591 119878
3(120591) = sum
119896minus1
119899=1cos(119896 minus
2119899)120591 and 1198784(120591) = sum
119896minus1
119899=1119899(119896 minus 119899) cos(119896 minus 2119899)120591
Notice that 1198782(120591) = 2119896119878
1(120591) + 119889119863
119896minus1(120591)119889120591 which
immediately leads to (A4)
28 Mathematical Problems in Engineering
Identity (A5) can be obtained as follows
sin (119896 minus 1) 120591sin 120591
=119890119895(119896minus1)120591
minus 119890minus119895(119896minus1)120591
119890119895120591 minus 119890minus119895120591
= 119890119895119896120591
119890minus2119895120591
minus 119890minus2119895(119896minus1)120591
1 minus 119890minus2119895120591
= 119890119895119896120591
119896minus1
sum
119899=1
119890minus2119895119899120591
=
119896minus1
sum
119899=1
119890119895(119896minus2119899)120591
=
119896minus1
sum
119899=1
cos (119896 minus 2119899) 120591
(A7)
From 4119899(119896 minus 119899) = 1198962
minus (119896 minus 2119899)2 it follows that 4119878
4(120591) =
1198962
1198783(120591) + 119889
2
1198783(120591)119889120591
2 which leads to (A6)
B The Chebyshev Polynomials
The Chebyshev polynomials of the first kind 119881119899(119909) can be
defined by the following relation (eg see [29])
119881119899(119909) = cos 119899120591 when 119909 = cos 120591 (B1)
The Chebyshev polynomials of the second kind 119880119899(119909) can be
defined by the following relation (eg see [29])
119880119899(119909) =
sin (119899 + 1) 120591sin 120591
when 119909 = cos 120591 (B2)
The Chebyshev polynomials satisfy the following recur-rence relations (eg see [29])
1198810(119909) = 1 119881
1(119909) = 119909
119881119899+1
(119909) = 2119909119881119899(119909) minus 119881
119899minus1(119909)
1198800(119909) = 1 119880
1(119909) = 2119909
119880119899+1
(119909) = 2119909119880119899(119909) minus 119880
119899minus1(119909)
(B3)
The first few Chebyshev polynomials of the first and secondkind are 119881
2(119909) = 2119909
2
minus 1 1198813(119909) = 4119909
3
minus 3119909 1198814(119909) = 8119909
4
minus
81199092
+ 1 1198802(119909) = 4119909
2
minus 1 1198803(119909) = 8119909
3
minus 4119909 and 1198804(119909) =
161199094
minus 121199092
+ 1
Conflict of Interests
The authors declare that there is no conflict of interestsregarding the publication of this paper
Acknowledgment
This work is supported by the SerbianMinistry of EducationScience and Technology Development as a part of ProjectTP32016
References
[1] V I Arnolrsquod V S Afrajmovich Y S Ilrsquoyashenko and L PShilrsquonikov Dynamical Systems V Bifurcation Theory and Catas-trophe Theory Springer Berlin Germany 1994
[2] E Polak ldquoOn the mathematical foundations of nondifferen-tiable optimization in engineering designrdquo SIAM Review vol29 no 1 pp 21ndash89 1987
[3] N S Fuzik ldquoBiharmonic modes of a tuned RF power amplifierrdquoRadiotehnika vol 25 no 7 pp 62ndash71 1970 (Russian)
[4] P Colantonio F Giannini G Leuzzi and E Limiti ldquoClass Gapproach for low-voltage high-efficiency PA designrdquo Interna-tional Journal of RF and Microwave Computer-Aided Engineer-ing vol 10 no 6 pp 366ndash378 2000
[5] F H Raab ldquoMaximum efficiency and output of class-F poweramplifiersrdquo IEEE Transactions on Microwave Theory and Tech-niques vol 49 no 6 pp 1162ndash1166 2001
[6] FH Raab ldquoClass-E class-C and class-F power amplifiers basedupon a finite number of harmonicsrdquo IEEE Transactions onMicrowaveTheory and Techniques vol 49 no 8 pp 1462ndash14682001
[7] J D Rhodes ldquoOutput universality inmaximum efficiency linearpower amplifiersrdquo International Journal of Circuit Theory andApplications vol 31 no 4 pp 385ndash405 2003
[8] A Juhas and L A Novak ldquoComments on lsquoClass-E class-Cand classF power amplifier based upon a finite number ofharmonicsrsquordquo IEEE Transactions on Microwave Theory and Tech-niques vol 57 no 6 pp 1623ndash1625 2009
[9] M Roberg and Z Popovic ldquoAnalysis of high-efficiency poweramplifiers with arbitrary output harmonic terminationsrdquo IEEETransactions on Microwave Theory and Techniques vol 59 no8 pp 2037ndash2048 2011
[10] A Grebennikov N O Sokal and M J Franco Switchmode RFPower Amplifiers ElsevierAcademic Press San Diego CalifUSA 2nd edition 2012
[11] T Canning P J Tasker and S C Cripps ldquoContinuous modepower amplifier design using harmonic clipping contourstheory and practicerdquo IEEE Transactions on Microwave Theoryand Techniques vol 62 no 1 pp 100ndash110 2014
[12] A Juhas and L A Novak ldquoGeneral description of nonnegativewaveforms up to second harmonic for power amplifier mod-ellingrdquoMathematical Problems in Engineering vol 2014 ArticleID 709762 18 pages 2014
[13] V I Arnolrsquod V V Goryunov O V Lyashko and V A VasilrsquoevDynamical Systems VIIImdashSingularity Theory II ApplicationsSpringer Berlin Germany 1993
[14] D Siersma ldquoProperties of conflict sets in the planerdquo BanachCenter Publications Polish Academy of Sciences vol 50 no 1 pp267ndash276 1999 Proceedings of the Banach Center Symposiumon Geometry and Topology of Caustics (Caustics rsquo98) WarsawPoland
[15] M van Manen The geometry of conflict sets [Dissertation]Universiteit Utrecht UtrechtThe Netherlands 2003 httpigi-tur-archivelibraryuunldissertations2003-0912-123058c4pdf
[16] Y L Sachkov ldquoMaxwell strata and symmetries in the problemofoptimal rolling of a sphere over a planerdquo Sbornik Mathematicsvol 201 no 7-8 pp 1029ndash1051 2010
[17] I A Bogaevsky ldquoPerestroikas of shocks and singularities ofminimum functionsrdquoPhysicaDNonlinear Phenomena vol 173no 1-2 pp 1ndash28 2002
[18] Y L Sachkov ldquoMaxwell strata in the Euler elastic problemrdquoJournal of Dynamical andControl Systems vol 14 no 2 pp 169ndash234 2008
[19] M Siino and T Koike ldquoTopological classification of black holesgeneric Maxwell set and crease set of a horizonrdquo InternationalJournal ofModern Physics D Gravitation Astrophysics Cosmol-ogy vol 20 no 6 pp 1095ndash1122 2011
Mathematical Problems in Engineering 29
[20] F H Raab ldquoClass-F power amplifiers with maximally flat wave-formsrdquo IEEETransactions onMicrowaveTheory and Techniquesvol 45 no 11 pp 2007ndash2012 1997
[21] A Juhas and L A Novak ldquoMaximally flat waveforms with finitenumber of harmonics in class-F power amplifiersrdquo Mathemat-ical Problems in Engineering vol 2013 Article ID 169590 9pages 2013
[22] L Fejer ldquoUber trigonometrische polynomerdquo Journal fur dieReine und Angewandte Mathematik vol 1916 no 146 pp 53ndash82 1916 (German)
[23] S C Cripps ldquoBessel Waives [microwave bytes]rdquo IEEE Micro-wave Magazine vol 10 no 7 pp 30ndash36 117 2009
[24] L N Bryzgalova ldquoSingularities of the maximum of para-metrically dependent functionrdquo Functional Analysis and ItsApplications vol 11 no 1 pp 49ndash51 1977
[25] V I Arnold A A Davydov V A Vassiliev and V MZakalyukin Mathematical Models of Catastrophes Control ofCatastrophic Process Encyclopedia of Life Support Systems(EOLSS) EOLSS Publishers Oxford UK 2006
[26] S C Cripps RF Power Amplifiers for Wireless CommunicationsArtech House Norwood Mass USA 2nd edition 2006
[27] A Zygmund Trigonometric Series vol 1 Cambridge UniversityPress Cambridge UK 2nd edition 1959
[28] A Jeffrey and H DaiHandbook of Mathematical Formulas andIntegrals ElsevierAcademic Press San Diego Calif USA 4thedition 2008
[29] J C Mason and D C Handscomb Chebyshev PolynomialsChapman amp Hall CRC Press Boca Raton Fla USA 2003
Submit your manuscripts athttpwwwhindawicom
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical Problems in Engineering
Hindawi Publishing Corporationhttpwwwhindawicom
Differential EquationsInternational Journal of
Volume 2014
Applied MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Probability and StatisticsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical PhysicsAdvances in
Complex AnalysisJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
OptimizationJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
CombinatoricsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Operations ResearchAdvances in
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Function Spaces
Abstract and Applied AnalysisHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of Mathematics and Mathematical Sciences
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Algebra
Discrete Dynamics in Nature and Society
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Decision SciencesAdvances in
Discrete MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom
Volume 2014 Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Stochastic AnalysisInternational Journal of
Mathematical Problems in Engineering 3
21 Minimum Function and Gain Function In what followswe consider family of waveforms of type
119908 (120591 120574 119860 120572) = 1 minus 120574 (cos 120591 + 119860 cos (119896120591 + 120572)) (1)
where 120591 stands for 120596119905 120574 gt 0 119896 ge 2 119860 gt 0 and 120572 isin [0 2120587)Waveforms of type (1) include all possible shapes which canoccur but not all possiblewaveforms containing fundamentaland 119896th harmonic However shifting of waveforms of type(1) along 120591-axis could recover all possible waveforms withfundamental and 119896th harmonic
The problem of finding nonnegative waveform of type(1) having maximum amplitude of fundamental harmonicplays an important role in optimization of PA efficiency Thisextremal problem can be reformulated as problem of findingnonnegative waveform from family (1) having maximumpossible value of coefficient 120574 Nonnegative waveform offamily (1) with maximum possible value of coefficient 120574 iscalled ldquooptimalrdquo or ldquoextremalrdquo waveform
Furthermore let us introduce an auxiliary waveform
119891 (120591 119860 120572) = minus cos 120591 minus 119860 cos (119896120591 + 120572) (2)
which is smooth function of one variable 120591 and two parame-ters119860 and 120572 In terms of 119891(120591 119860 120572) the above extremal prob-lem reduces to the problem of finding maximum possiblevalue of coefficient 120574 that satisfies
1 + 120574119891 (120591 119860 120572) ge 0 (3)
Clearly for any prescribed pair (119860 120572) there is a uniquemaximal value of coefficient 120574 for which inequality (3) holdsfor all 120591 This maximal value of 120574 associated with the pair(119860 120572) we denote it by 119866(119860 120572) and call it ldquogain functionrdquo
Let
119865min (119860 120572) = min120591
119891 (120591 119860 120572) (4)
be theminimum function associatedwith119891(120591 119860 120572) Accord-ing to (3) 119866(119860 120572) and 119865min(119860 120572) satisfy the following rela-tion 1 + 119866(119860 120572)119865min(119860 120572) = 0 Since 119865min(119860 120572) is obviouslynonzero it follows immediately that
119866 (119860 120572) = minus1
119865min (119860 120572) (5)
A relation analogue to (5) for 119896 = 2 (fundamental andsecond harmonic) has been derived in [4] According toour best knowledge it was the first appearance of gainfunction expressed via associated minimum function Theconsideration presented in [4] has been restricted to theparticular case when 120572 = 120587 The same problem for 120572 = 120587 andarbitrary 119896 ge 2 has been investigated in [3] (see also [10])
According to above consideration the problem of finding3-tuple (120574 119860 120572) with maximum possible value of 120574 for which(3) holds is equivalent to the problem of finding maximumvalue of gain function
120574max = max119860120572
119866 (119860 120572) = minus1
max119860120572
119865min (119860 120572)(6)
14
12
1
08
06
04
G1 G2
001
02
0304
0506 0
1205872
12058731205872
2120587
Parameter 120572
Parameter A
Gai
n fu
nctio
n
k = 2
Figure 1 Graph of 119866(119860 120572) for 119896 = 2 Points 1198661and 119866
2denote
beginning of the ridge and maximum of gain function respectively
and corresponding pair (119860lowast 120572lowast) that satisfies
120574max = max119860120572
119866 (119860 120572) = 119866 (119860lowast
120572lowast
)
max119860120572
119865min (119860 120572) = 119865min (119860lowast
120572lowast
)
(7)
Thus the optimal waveform 119908lowast
(120591) is determined by parame-ters 120574max 119860
lowast and 120572lowast that is
119908lowast
(120591) = 119908 (120591 120574max 119860lowast
120572lowast
) (8)
Optimal waveform has two global minima (this claim willbe justified in Section 4 Remark 21) Consequently the pair(119860lowast
120572lowast
) which corresponds to maximum of gain function119866(119860 120572) belongs to conflict set in (119860 120572) parameter space
Figure 1 shows graph of gain function 119866(119860 120572) for 119896 = 2Notice that it has sharp ridge and that maximum of gainfunction (point 119866
2) lies on the ridge This maximum corre-
sponds to the optimal waveform (solution of the consideredextremal problem) The beginning of the ridge (point 119866
1)
corresponds to the waveform which possesses global mini-mum at degenerate critical point that is corresponds tomax-imally flat waveform (eg see [21]) Gain function 119866(119860 120572)
is not differentiable on the ridge and consequently is notdifferentiable at the point where it has global maximumThisexplains why the approach based on critical points does notwork and why conflict set is so important in the consideredproblem
Positions of global minima of 119891(120591 119860 120572) for 119896 = 2 arepresented in Figure 2 According to Proposition 1 conflict setis the ray defined by119860 gt 14 and120572 = 120587Waveforms119891(120591 119860 120572)with parameters that belong to the conflict set have two globalminimaThewaveform corresponding to the end point of theray (119860 = 14 and 120572 = 120587) has global minimum at degeneratecritical point (so-called maximally flat waveform [21])
Nonnegative waveforms of type (1) with 120574 = 119866(119860 120572) =
minus1119865min(119860 120572) have at least one zero To show that it issufficient to see that 119908(120591 120574 119860 120572) = 0 for 120591 satisfying 119891(120591119860 120572) = 119865min(119860 120572)
4 Mathematical Problems in Engineering
001
02
0304
05
06 01205872
12058731205872
2120587
Parameter 120572
Parameter A
1205872
1205874
0
minus1205874
minus1205872Posit
ions
of g
loba
l min
k = 2
Figure 2 Positions of global minima of 119891(120591 119860 120572) for 119896 = 2
The problem of finding maximum value of fundamentalharmonic cosine part of nonnegative waveform of the form
119908119886(120591 120574119886 119887 119860 120572) = 1 minus 120574
119886(cos 120591 + 119887 sin 120591 + 119860 cos (119896120591 + 120572))
(9)
where 120574119886gt 0 is also related to the problem of finding max-
imumof theminimum functionOptimalwaveformof family(9) has two global minima (this claim will be justified inSection 5 Remark 25) and therefore corresponding 3-tupleof parameters belongs to the conflict set in parameter spaceof family (9)
Let us introduce an auxiliary waveform
119891119886(120591 119887 119860 120572) = minus cos 120591 minus 119887 sin 120591 minus 119860 cos (119896120591 + 120572) (10)
and corresponding minimum function 119865119886min(119887 119860 120572) =
min120591119891119886(120591 119887 119860 120572) Inequality 119908
119886(120591 120574119886 119887 119860 120572) ge 0 can be
rewritten as 1 + 120574119886119865119886min(119887 119860 120572) ge 0 and therefore the
highest value of 120574119886is attained for 120574
119886= minus1119865
119886min(119887 119860 120572) Itimmediately follows that nonnegative waveform of type (9)with 120574
119886= minus1119865
119886min(119887 119860 120572) has zero for 120591 satisfying 119891119886(120591
119887 119860 120572) = 119865119886min(119887 119860 120572)
22 Conflict Set Historically conflict set came into beingfrom the problems in which families of smooth functions(such as potentials distances and waveforms) with two com-peting minima occurThe situation when competing minimabecome equal refers to the presence of conflict set (Maxwellset Maxwell strata) in the associated parameter space
There are many facets of conflict set For example in theproblem involving distances between two sets of points theconflict set is the intersections between iso-distance lines[14] Conflict set also arises in the situation when two wavefronts coming from different objects meet [15 25] In thestudy of black holes conflict set is the line of crossover of thehorizon formed by the merger of two black holes [19] In theclassical Euler problem conflict set is a set of points wheredistinct extremal trajectories with the same value of the costfunctional meet one another [18]
Conflict set is very difficult to calculate both analyticallyand numerically (eg see [15]) because of apparent nondif-ferentiability in some directions In optimization of PA effi-ciency some authors already reported difficulties in findingoptimum via standard analytical tools [4 5]
In this section we consider conflict set in the context offamily of waveforms of type (2) for arbitrary 119896 ge 2 In thiscontext for prescribed integer 119896 ge 2 conflict set is said to bea set of all pairs (119860 120572) for which 119891(120591 119860 120572) possesses multipleglobal minima
Suppose that 1205911015840 and 12059110158401015840 are the positions of global min-
ima of 119891(120591 119860 120572) Then the conflict set is specified by the fol-lowing set of relations
119891 (1205911015840
119860 120572) = 119891 (12059110158401015840
119860 120572) (11)
1198911015840
(1205911015840
119860 120572) = 0 1198911015840
(12059110158401015840
119860 120572) = 0 (12)
11989110158401015840
(1205911015840
119860 120572) gt 0 11989110158401015840
(12059110158401015840
119860 120572) gt 0 (13)
(forall120591) 119891 (120591 119860 120572) ge 119891 (1205911015840
119860 120572) (14)
Relations (12) and (13) say that 119891(120591 119860 120572) has minima at 1205911015840and 12059110158401015840 while relations (11) and (14) imply that these minimaare equal and global
The following proposition describes the conflict set offamily of waveforms of type (2)
Proposition 1 Conflict set of family of waveforms of type (2)is the set of all pairs (119860 120572) such that 119860 gt 1119896
2 and 120572 = 120587
The proof of Proposition 1 which is provided at the endof this section also implies that the following four corollarieshold
Corollary 2 The conflict set has end point at (119860 120572) =
(11198962
120587) This end point corresponds to the maximally flatwaveform [21]
Corollary 3 Waveforms of type (2) with parameters thatbelong to conflict set have two global minima at plusmn120591
Δ where
0 lt 120591Δle 120587119896
Corollary 4 Every waveform with fundamental and 119896th har-monic has either one or two global minima
Corollary 5 Conflict set can be parameterised in terms of 120591Δ
as follows
120572 = 120587 119860 (120591Δ) =
sin 120591Δ
119896 sin 119896120591Δ
0 lt 120591Δle120587
119896 (15)
Notice that119860(120591Δ) ismonotonically increasing function on inter-
val 0 lt 120591Δle 120587119896
Proof of Proposition 1 Without loss of generality we canrestrict our consideration to the interval minus120587 lt 120591 le 120587 This isan immediate consequence of the fact that 119891(120591 119860 120572) is aperiodic function
Suppose that 1205911015840 and 12059110158401015840 where 1205911015840 lt 12059110158401015840 are points at which
119891(120591 119860 120572) has two equal global minima Then conflict set is
Mathematical Problems in Engineering 5
specified by relations (11)ndash(14) From (11)ndash(13) it follows thatrelations
119891 (1205911015840
119860 120572) minus 119891 (12059110158401015840
119860 120572) = 0
1198911015840
(1205911015840
119860 120572) + 1198911015840
(12059110158401015840
119860 120572) = 0
1198911015840
(1205911015840
119860 120572) minus 1198911015840
(12059110158401015840
119860 120572) = 0
11989110158401015840
(1205911015840
119860 120572) + 11989110158401015840
(12059110158401015840
119860 120572) gt 0
(16)
also hold Let
120591119904119903=(1205911015840
+ 12059110158401015840
)
2 120591
Δ=(12059110158401015840
minus 1205911015840
)
2
(17)
be a pair of points associated with (1205911015840 12059110158401015840) Clearly
minus120587 lt 120591119904119903lt 120587 (18)
0 lt 120591Δlt 120587 (19)
12059110158401015840
= 120591119904119903+ 120591Δ 120591
1015840
= 120591119904119903minus 120591Δ (20)
The first and second derivatives of 119891(120591 119860 120572) are equal to
1198911015840
(120591 119860 120572) = sin 120591 + 119896119860 sin (119896120591 + 120572)
11989110158401015840
(120591 119860 120572) = cos 120591 + 1198962119860 cos (119896120591 + 120572) (21)
By using (20)-(21) system (16) can be rewritten as
sin 120591119904119903sin 120591Δ+ 119860 sin (119896120591
119904119903+ 120572) sin 119896120591
Δ= 0 (22)
sin 120591119904119903cos 120591Δ+ 119896119860 sin (119896120591
119904119903+ 120572) cos 119896120591
Δ= 0 (23)
cos 120591119904119903sin 120591Δ+ 119896119860 cos (119896120591
119904119903+ 120572) sin 119896120591
Δ= 0 (24)
cos 120591119904119903cos 120591Δ+ 1198962
119860 cos (119896120591119904119903+ 120572) cos 119896120591
Δgt 0 (25)
From (19) it follows that sin 120591Δgt 0 Multiplying (24) and
(25) with minus cos 120591Δand sin 120591
Δgt 0 respectively and sum-
ming the resulting relations we obtain 119896119860 cos(119896120591119904119903
+
120572)[119896 sin 120591Δcos 119896120591
Δminus sin 119896120591
Δcos 120591Δ] gt 0 The latest relation
immediately implies that
119896 sin 120591Δcos 119896120591
Δminus sin 119896120591
Δcos 120591Δ
= 0 (26)
Equations (22) and (23) can be considered as a system oftwo linear equations in terms of sin 120591
119904119903and 119860 sin(119896120591
119904119903+ 120572)
According to (26) the determinant of this system is nonzeroand therefore it has only trivial solution
sin 120591119904119903= 0 sin (119896120591
119904119903+ 120572) = 0 (27)
According to (18) sin 120591119904119903= 0 implies
120591119904119903= 0 (28)
According to (20) 120591119904119903= 0 implies
12059110158401015840
= 120591Δ 120591
1015840
= minus120591Δ (29)
Furthermore 120591119904119903= 0 and sin(119896120591
119904119903+ 120572) = 0 imply that sin120572 =
0 From (29) it follows that 120591Δis position of global minimum
of 119891(120591 119860 120572) Clearly 119891(120591Δ 119860 120572) le 119891(0 119860 120572) which together
with sin120572 = 0 leads to
1 minus cos 120591Δ+ 119860 cos120572 (1 minus cos 119896120591
Δ) le 0 (30)
From 120591Δ
= 0 (see (19)) 119860 gt 0 and (30) it follows that cos 120572 lt
0 which together with sin120572 = 0 yields
120572 = 120587 (31)
Since 120591Δis position of global minimum it follows that
119891(120591Δ 119860 120587) le 119891(120587119896 119860 120587) Accordingly 119860(1 + cos 119896120591
Δ) le
cos 120591Δminus cos120587119896 which together with 119860 gt 0 implies that
cos 120591Δminus cos120587119896 ge 0 This relation along with (19) yields
0 lt 120591Δle120587
119896 (32)
Substitution of (31) and (28) in (24) leads to
119860 =sin 120591Δ
119896 sin 119896120591Δ
(33)
Notice that sin 119896120591Δ sin 120591
Δis monotonically decreasing func-
tion on interval (32)Therefore parameter119860 is monotonicallyincreasing function on the same interval with lim
120591Δrarr0+119860 =
11198962 Consequently 119860 gt 1119896
2 which completes theproof
23 Parameter Space In parameter space of family of wave-forms (2) there are two subsets playing important role in theclassification of the family instancesThese are conflict set andcatastrophe set
Catastrophe set is subset of parameter space of waveform119891(120591 119860 120572) It consists of those pairs (119860 120572) for which thecorresponding waveforms 119891(120591 119860 120572) have degenerate criticalpoints at which first and second derivatives are equal to zeroThus for finding catastrophe set we have to consider thefollowing system of equations
1198911015840
(120591119889 119860 120572) = 0
11989110158401015840
(120591119889 119860 120572) = 0
(34)
where 120591119889is a degenerate critical point of waveform119891(120591 119860 120572)
Conflict set in parameter space of waveform119891(120591 119860 120572) asshown in Proposition 1 is the ray described by 119860 gt 1119896
2 and120572 = 120587 It is intimately connected to catastrophe set
In what follows in this subsection we use polar coordinatesystem (119860 cos120572 119860 sin120572) instead of Cartesian coordinatesystem (119860 120572) Examples of catastrophe set and conflict setfor 119896 le 5 plotted in parameter space (119860 cos120572 119860 sin120572) arepresented in Figure 3 Solid line represents the catastropheset while dotted line describes conflict set The isolated pickpoints (usually called cusp) which appear in catastrophecurves correspond to maximally flat waveforms with max-imally flat minimum andor maximally flat maximumThereare two such picks in the catastrophe curves for 119896 = 2 and
6 Mathematical Problems in Engineering
00
00
k = 2 k = 3
k = 4 k = 5
Acos 120572
Acos 120572
Acos 120572
Acos 120572
A sin 120572
A sin 120572A sin 120572
A sin 120572
Figure 3 Catastrophe set (solid line) and corresponding conflict set(dotted line) for 119896 le 5 In each plot white triangle dot correspondsto optimal waveform and white circle dot corresponds to maximallyflat waveform
119896 = 4 and one in the catastrophe curves for 119896 = 3 and 119896 = 5Notice that the end point of conflict set is the cusp point
Catastrophe set divides the parameter space (119860 cos120572119860 sin120572) into disjoint subsets In the cases 119896 = 2 and 119896 =
3 catastrophe curve defines inner and outer part For 119896 gt
3 catastrophe curve makes partition of parameter space inseveral inner subsets and one outer subset (see Figure 3)
Notice also that multiplying 119891(120591 119860 120572) with a positiveconstant and adding in turn another constant which leads towaveform of type 119908(120591 120574 119860 120572) (see (1) and (2)) do not makeimpact on the character of catastrophe and conflict sets Thisis because in the course of finding catastrophe set first andsecond derivatives of 119891(120591 119860 120572) are set to zero Clearly (34) interms of 119891(120591 119860 120572) are equivalent to the analogous equationsin terms of119908(120591 120574 119860 120572) Analogously in the course of findingconflict set we consider only the positions of global minima(these positions forwaveforms119891(120591 119860 120572) and119908(120591 120574 119860 120572) arethe same)
3 Nonnegative Waveforms with atLeast One Zero
In what follows let us consider a waveform containing dccomponent fundamental and 119896th (119896 ge 2) harmonic of theform
119879119896(120591) = 1 + 119886
1cos 120591 + 119887
1sin 120591 + 119886
119896cos 119896120591 + 119887
119896sin 119896120591 (35)
The amplitudes of fundamental and 119896th harmonic of wave-form of type (35) respectively are
1205821= radic11988621+ 11988721 (36)
120582119896= radic1198862119896+ 1198872119896 (37)
As it is shown in Section 21 nonnegative waveforms withmaximal amplitude of fundamental harmonic or maximalcoefficient of fundamental harmonic cosine part have atleast one zero It is also shown in Section 22 (Corollary 4)that waveforms of type (35) with nonzero amplitude offundamental harmonic have either one or two globalminimaConsequently if nonnegative waveform of type (35) withnonzero amplitude of fundamental harmonic has at least onezero then it has at most two zeros
In Section 31 we provide general description of nonnega-tive waveforms of type (35) with at least one zero In Sections32 and 33 we consider nonnegative waveforms of type (35)with two zeros
31 General Description of Nonnegative Waveforms with atLeast One Zero The main result of this section is presentedin the following proposition
Proposition 6 Every nonnegative waveform of type (35) withat least one zero can be expressed in the following form
119879119896(120591) = [1 minus cos (120591 minus 120591
0)] [1 minus 120582
119896119903119896(120591)] (38)
where119903119896(120591) = (119896 minus 1) cos 120585
+ 2
119896minus1
sum
119899=1
(119896 minus 119899) cos (119899 (120591 minus 1205910) + 120585)
(39)
providing that
120582119896le [(119896 minus 1) cos 120585 + 119896 sin(120585 minus 120585119896)
sin(120585119896)]
minus1
(40)
10038161003816100381610038161205851003816100381610038161003816 le 120587 (41)
Remark 7 Function on the right hand side of (40) is mono-tonically increasing function of |120585| on interval |120585| le 120587 (formore details about this function see Remark 15) From (57)and (65) it follows that relation
0 le 120582119896le 1 (42)
holds for every nonnegative waveform of type (35) Noticethat according to (40) 120582
119896= 1 implies |120585| = 120587 Substitution
of 120582119896
= 1 and |120585| = 120587 into (55) yields 119879119896(120591) = 1 minus
cos 119896(120591 minus 1205910) Consequently 120582
119896= 1 implies that amplitude
1205821of fundamental harmonic is equal to zero
Remark 8 Conversion of (38) into additive form leads to thefollowing expressions for coefficients of nonnegative wave-forms of type (35) with at least one zero
1198861= minus (1 + 120582
119896cos 120585) cos 120591
0minus 119896120582119896sin 120585 sin 120591
0 (43)
1198871= minus (1 + 120582
119896cos 120585) sin 120591
0+ 119896120582119896sin 120585 cos 120591
0 (44)
119886119896= 120582119896cos (119896120591
0minus 120585) (45)
119887119896= 120582119896sin (119896120591
0minus 120585) (46)
providing that 120582119896satisfy (40) and |120585| le 120587
Mathematical Problems in Engineering 7
2
1
0
0 1 2 3 4
Angle 120591120587
Wav
efor
ms
1205823 = radic2201205823 = radic281205823 = radic24
Figure 4 Nonnegative waveforms with at least one zero for 119896 = 31205910= 1205876 and 120585 = 31205874
Three examples of nonnegative waveforms with at leastone zero for 119896 = 3 are presented in Figure 4 (examples ofnonnegative waveformswith at least one zero for 119896 = 2 can befound in [12]) For all three waveforms presented in Figure 4we assume that 120591
0= 1205876 and 120585 = 31205874 From (40) it follows
that 1205823le radic24 Coefficients of waveform with 120582
3= radic220
(dotted line) are 1198861= minus08977 119887
1= minus03451 119886
3= 005
and 1198873= minus005 Coefficients of waveform with 120582
3= radic28
(dashed line) are 1198861= minus09453 119887
1= minus01127 119886
3= 0125
and 1198873= minus0125 Coefficients of waveform with 120582
3= radic24
(solid line) are 1198861= minus10245 119887
1= 02745 119886
3= 025 and
1198873= minus025 First two waveforms have one zero while third
waveform (presented with solid line) has two zeros
Proof of Proposition 6 Waveform of type (35) containingdc component fundamental and 119896th harmonic can be alsoexpressed in the form
119879119896(120591) = 1 + 120582
1cos (120591 + 120593
1) + 120582119896cos (119896120591 + 120593
119896) (47)
where1205821ge 0120582
119896ge 0120593
1isin (minus120587 120587] and120593
119896isin (minus120587 120587] It is easy
to see that relations between coefficient of (35) and param-eters of (47) read as follows
1198861= 1205821cos1205931 119887
1= minus1205821sin1205931 (48)
119886119896= 120582119896cos120593119896 119887
119896= minus120582119896sin120593119896 (49)
Let us introduce 120585 such that10038161003816100381610038161205851003816100381610038161003816 le 120587 120585 = (119896120591
0+ 120593119896) mod2120587 (50)
Using (50) coefficients (49) can be expressed as (45)-(46)Let us assume that 119879
119896(120591) is nonnegative waveform of type
(35) with at least one zero that is 119879119896(120591) ge 0 and 119879
119896(1205910) = 0
for some 1205910 Notice that conditions 119879
119896(120591) ge 0 and 119879
119896(1205910) = 0
imply that 1198791015840119896(1205910) = 0 From 119879
119896(1205910) = 0 and 1198791015840
119896(1205910) = 0 by
using (50) it follows that
1205821cos (120591
0+ 1205931) = minus (1 + 120582
119896cos 120585)
1205821sin (1205910+ 1205931) = minus119896120582
119896sin 120585
(51)
respectively On the other hand 1205821cos(120591 + 120593
1) can be rewrit-
ten as
1205821cos (120591 + 120593
1) = 1205821cos (120591
0+ 1205931) cos (120591 minus 120591
0)
minus 1205821sin (1205910+ 1205931) sin (120591 minus 120591
0)
(52)
Substitution of (51) into (52) yields
1205821cos (120591 + 120593
1) = minus (1 + 120582
119896cos 120585) cos (120591 minus 120591
0)
+ 119896120582119896sin 120585 sin (120591 minus 120591
0)
(53)
According to (50) it follows that cos(119896120591 + 120593119896) = cos(119896(120591 minus
1205910) + 120585) that is
cos (119896120591 + 120593119896) = cos 120585 cos 119896 (120591 minus 120591
0) minus sin 120585 sin 119896 (120591 minus 120591
0)
(54)
Furthermore substitution of (54) and (53) into (47) leads to
119879119896(120591) = [1 minus cos (120591 minus 120591
0)] [1 + 120582
119896cos 120585]
minus 120582119896[1 minus cos 119896 (120591 minus 120591
0)] cos 120585
+ 120582119896[119896 sin (120591 minus 120591
0) minus sin 119896 (120591 minus 120591
0)] sin 120585
(55)
According to (A2) and (A4) (see Appendices) there is com-mon factor [1minuscos(120591minus120591
0)] for all terms in (55) Consequently
(55) can be written in the form (38) where
119903119896(120591) = minus cos 120585 + [
1 minus cos 119896 (120591 minus 1205910)
1 minus cos (120591 minus 1205910)] cos 120585
minus [119896 sin (120591 minus 120591
0) minus sin 119896 (120591 minus 120591
0)
1 minus cos (120591 minus 1205910)
] sin 120585
(56)
From (56) by using (A2) (A4) and cos 120585 cos 119899(120591 minus 1205910) minus
sin 120585 sin 119899(120591 minus 1205910) = cos(119899(120591 minus 120591
0) + 120585) we obtain (39)
In what follows we are going to prove that (40) also holdsAccording to (38) 119879
119896(120591) is nonnegative if and only if
120582119896max120591
119903119896(120591) le 1 (57)
Let us first show that position of global maximumof 119903119896(120591)
belongs to the interval |120591 minus 1205910| le 2120587119896 Relation (56) can be
rewritten as
119903119896(120591) = 119903
119896(1205910minus2120585
119896) + 119902119896(120591) (58)
where
119903119896(1205910minus2120585
119896) = (119896 minus 1) cos 120585 + 119896 sin (120585 minus 120585119896)
sin (120585119896) (59)
119902119896(120591) =
1
1 minus cos (120591 minus 1205910)
sdot [cos 120585 minus cos(119896(120591 minus 1205910+120585
119896))
+119896 sin 120585sin (120585119896)
(cos(120591 minus 1205910+120585
119896) minus cos(120585
119896))]
(60)
8 Mathematical Problems in Engineering
For |120585| lt 120587 relation sin 120585 sin(120585119896) gt 0 obviously holds Fromcos 119905 gt cos 1199051015840 for |119905| le 120587119896 lt |119905
1015840
| le 120587 it follows that positionof global maximum of the function of type [119888 cos 119905 minus cos(119896119905)]for 119888 gt 0 belongs to interval |119905| le 120587119896 Therefore position ofglobal maximum of the expression in the square brackets in(60) for |120585| lt 120587 belongs to interval |120591 minus 120591
0+ 120585119896| le 120587119896 This
inequality together with |120585| lt 120587 leads to |120591minus1205910| lt 2120587119896 Since
[1 minus cos(120591 minus 1205910)]minus1 decreases with increasing |120591 minus 120591
0| le 120587 it
follows that 119902119896(120591) for |120585| lt 120587 has global maximum on interval
|120591minus1205910| lt 2120587119896 For |120585| = 120587 it is easy to show thatmax
120591119902119896(120591) =
119902119896(1205910plusmn 2120587119896) = 0 Since 119903
119896(120591) minus 119902
119896(120591) is constant (see (58))
it follows from previous considerations that 119903119896(120591) has global
maximum on interval |120591 minus 1205910| le 2120587119896
To find max120591119903119896(120591) let us consider first derivative of 119903
119896(120591)
with respect to 120591 Starting from (56) first derivative of 119903119896(120591)
can be expressed in the following form
119889119903119896(120591)
119889120591= minus119904 (120591) sdot sin(
119896 (120591 minus 1205910)
2+ 120585) (61)
where
119904 (120591) = [sin(119896 (120591 minus 120591
0)
2) cos(
120591 minus 1205910
2)
minus 119896 cos(119896 (120591 minus 120591
0)
2) sin(
120591 minus 1205910
2)]
sdot sinminus3 (120591 minus 1205910
2)
(62)
Using (A6) (see Appendices) (62) can be rewritten as
119904 (120591) = 2
119896minus1
sum
119899=1
119899 (119896 minus 119899) cos((119896 minus 2119899) (120591 minus 120591
0)
2) (63)
From 119899(119896 minus 119899) gt 0 and |119896 minus 2119899| lt 119896 119899 = 1 (119896 minus 1) itfollows that all summands in (63) decrease with increasing|120591 minus 1205910| providing that |120591 minus 120591
0| le 2120587119896 Therefore 119904(120591) ge 119904(120591
0plusmn
2120587119896) = 119896sin2(120587119896) gt 0 for |120591 minus 1205910| le 2120587119896 Consequently
119889119903119896(120591)119889120591 = 0 and |120591minus120591
0| le 2120587119896 imply that sin(119896(120591minus120591
0)2+
120585) = 0From |120585| le 120587 |120591minus120591
0| le 2120587119896 and sin(119896(120591minus120591
0)2+120585) = 0
it follows that 120591minus1205910+120585119896 = minus120585119896 or |120591minus120591
0+120585119896| = (2120587minus|120585|)119896
and therefore cos(119896(120591 minus 1205910+ 120585119896)) = cos 120585 Since cos(120585119896) ge
cos(2120587 minus |120585|)119896 it follows that max120591119902119896(120591) is attained for 120591 =
1205910minus2120585119896 Furthermore from (60) it follows that max
120591119902119896(120591) =
119902119896(1205910minus 2120585119896) = 0 which together with (58)-(59) leads to
max120591
119903119896(120591) = 119903
119896(1205910minus2120585
119896)
= (119896 minus 1) cos 120585 + 119896 sin (120585 minus 120585119896)sin (120585119896)
(64)
Both terms on the right hand side of (64) are even functionsof 120585 and decrease with increase of |120585| |120585| le 120587 Thereforemax120591119903119896(120591) attains its lowest value for |120585| = 120587 It is easy to
show that right hand side of (64) for |120585| = 120587 is equal to 1which further implies that
max120591
119903119896(120591) ge 1 (65)
From (65) it follows that (57) can be rewritten as 120582119896
le
[max120591119903119896(120591)]minus1 Finally substitution of (64) into 120582
119896le
[max120591119903119896(120591)]minus1 leads to (40) which completes the proof
32 Nonnegative Waveforms with Two Zeros Nonnegativewaveforms of type (35) with two zeros always possess twoglobal minima Such nonnegative waveforms are thereforerelated to the conflict set
In this subsection we provide general description of non-negative waveforms of type (35) for 119896 ge 2 and exactly twozeros According to Remark 7 120582
119896= 1 implies |120585| = 120587 and
119879119896(120591) = 1 minus cos 119896(120591 minus 120591
0) Number of zeros of 119879
119896(120591) = 1 minus
cos 119896(120591minus1205910) on fundamental period equals 119896 which is greater
than two for 119896 gt 2 and equal to two for 119896 = 2 In the followingproposition we exclude all waveforms with 120582
119896= 1 (the case
when 119896 = 2 and 1205822= 1 is going to be discussed in Remark 10)
Proposition 9 Every nonnegative waveform of type (35) withexactly two zeros can be expressed in the following form
119879119896(120591) = 120582
119896[1 minus cos (120591 minus 120591
0)] [1 minus cos(120591 minus 120591
0+2120585
119896)]
sdot [1198880+ 2
119896minus2
sum
119899=1
119888119899cos 119899(120591 minus 120591
0+120585
119896)]
(66)
where
119888119899= [sin(120585 minus 119899120585
119896) cos(120585
119896)
minus (119896 minus 119899) cos(120585 minus 119899120585
119896) sin(120585
119896)]
sdot sinminus3 (120585119896)
(67)
120582119896= [(119896 minus 1) cos 120585 + 119896 sin(120585 minus 120585119896)
sin(120585119896)]
minus1
(68)
0 lt10038161003816100381610038161205851003816100381610038161003816 lt 120587 (69)
Remark 10 For 119896 = 2 waveforms with 1205822= 1 also have
exactly two zeros These waveforms can be included in aboveproposition by substituting (69) with 0 lt |120585| le 120587
Remark 11 Apart from nonnegative waveforms of type (35)with two zeros there are another two types of nonnegativewaveforms which can be obtained from (66)ndash(68) These are
(i) nonnegative waveforms with 119896 zeros (correspondingto |120585| = 120587) and
(ii) maximally flat nonnegative waveforms (correspond-ing to 120585 = 0)
Notice that nonnegative waveforms of type (35) with120582119896= 1 can be obtained from (66)ndash(68) by setting |120585| =
120587 Substitution of 120582119896
= 1 and |120585| = 120587 into (66) alongwith execution of all multiplications and usage of (A2) (seeAppendices) leads to 119879
119896(120591) = 1 minus cos 119896(120591 minus 120591
0)
Mathematical Problems in Engineering 9
Also maximally flat nonnegative waveforms (they haveonly one zero [21]) can be obtained from (66)ndash(68) by setting120585 = 0 Thus substitution of 120585 = 0 into (66)ndash(68) leads tothe following form of maximally flat nonnegative waveformof type (35)
119879119896(120591) =
[1 minus cos (120591 minus 1205910)]2
3 (1198962 minus 1)
sdot [119896 (1198962
minus 1)
+ 2
119896minus2
sum
119899=1
(119896 minus 119899) ((119896 minus 119899)2
minus 1) cos 119899 (120591 minus 1205910)]
(70)
Maximally flat nonnegative waveforms of type (35) for 119896 le 4
can be expressed as
1198792(120591) =
2
3[1 minus cos(120591 minus 120591
0)]2
1198793(120591) =
1
2[1 minus cos(120591 minus 120591
0)]2
[2 + cos (120591 minus 1205910)]
1198794(120591) =
4
15[1 minus cos (120591 minus 120591
0)]2
sdot [5 + 4 cos (120591 minus 1205910) + cos 2 (120591 minus 120591
0)]
(71)
Remark 12 Every nonnegative waveform of type (35) withexactly one zero at nondegenerate critical point can bedescribed as in Proposition 6 providing that symbol ldquolerdquoin relation (40) is replaced with ldquoltrdquo This is an immediateconsequence of Propositions 6 and 9 and Remark 11
Remark 13 Identity [1minus cos(120591minus1205910)][1minus cos(120591minus120591
0+2120585119896)] =
[cos 120585119896 minus cos(120591 minus 1205910+ 120585119896)]
2 implies that (66) can be alsorewritten as
119879119896(120591) = 120582
119896[cos 120585119896
minus cos(120591 minus 1205910+120585
119896)]
2
sdot [1198880+ 2
119896minus2
sum
119899=1
119888119899cos 119899(120591 minus 120591
0+120585
119896)]
(72)
Furthermore substitution of (67) into (72) leads to
119879119896(120591)
= 120582119896[cos 120585119896
minus cos(120591 minus 1205910+120585
119896)]
sdot [(119896 minus 1) sin 120585sin (120585119896)
minus 2
119896minus1
sum
119899=1
sin (120585 minus 119899120585119896)sin (120585119896)
cos 119899(120591 minus 1205910+120585
119896)]
(73)
Remark 14 According to (A6) (see Appendices) it followsthat coefficients (67) can be expressed as
119888119899= 2
119896minus119899minus1
sum
119898=1
119898(119896 minus 119899 minus 119898) cos((119896 minus 119899 minus 2119898) 120585119896
) (74)
Furthermore from (74) it follows that coefficients 119888119896minus2
119888119896minus3
119888119896minus4
and 119888119896minus5
are equal to119888119896minus2
= 2 (75)
119888119896minus3
= 8 cos(120585119896) (76)
119888119896minus4
= 8 + 12 cos(2120585119896) (77)
119888119896minus5
= 24 cos(120585119896) + 16 cos(3120585
119896) (78)
For example for 119896 = 2 (75) and (68) lead to 1198880= 2 and
1205822= 1(2+cos 120585) respectively which from (72) further imply
that
1198792(120591) =
2 [cos(1205852) minus cos(120591 minus 1205910+ 1205852)]
2
[2 + cos 120585] (79)
Also for 119896 = 3 (75) (76) and (68) lead to 1198881= 2 119888
0=
8 cos(1205853) and 1205823= [2(3 cos(1205853) + cos 120585)]minus1 respectively
which from (72) further imply that
1198793(120591) =
2 [cos (1205853) minus cos (120591 minus 1205910+ 1205853)]
2
[3 cos (1205853) + cos 120585]
sdot [2 cos(1205853) + cos(120591 minus 120591
0+120585
3)]
(80)
Remark 15 According to (A5) (see Appendices) relation(68) can be rewritten as
120582119896= [(119896 minus 1) cos 120585 + 119896
119896minus1
sum
119899=1
cos((119896 minus 2119899)120585119896
)]
minus1
(81)
Clearly amplitude 120582119896of 119896th harmonic of nonnegative wave-
form of type (35) with exactly two zeros is even functionof 120585 Since cos((119896 minus 2119899)120585119896) 119899 = 0 (119896 minus 1) decreaseswith increase of |120585| on interval 0 le |120585| le 120587 it follows that120582119896monotonically increases with increase of |120585| Right hand
side of (68) is equal to 1(1198962
minus 1) for 120585 = 0 and to one for|120585| = 120587 Therefore for nonnegative waveforms of type (35)with exactly two zeros the following relation holds
1
1198962 minus 1lt 120582119896lt 1 (82)
The left boundary in (82) corresponds to maximally flatnonnegative waveforms (see Remark 11) The right boundaryin (82) corresponds to nonnegative waveforms with 119896 zeros(also see Remark 11)
Amplitude of 119896th harmonic of nonnegative waveform oftype (35) with two zeros as a function of parameter 120585 for 119896 le5 is presented in Figure 5
Remark 16 Nonnegative waveform of type (35) with twozeros can be also expressed in the following form
119879119896(120591) = 1 minus 120582
119896
119896 sin 120585sin (120585119896)
cos(120591 minus 1205910+120585
119896)
+ 120582119896cos (119896 (120591 minus 120591
0) + 120585)
(83)
10 Mathematical Problems in Engineering
1
08
06
04
02
0minus1 minus05 0 05
Am
plitu
de120582k
1
Parameter 120585120587
k = 2k = 3
k = 4k = 5
Figure 5 Amplitude of 119896th harmonic of nonnegative waveformwith two zeros as a function of parameter 120585
where 120582119896is given by (68) and 0 lt |120585| lt 120587 From (83) it follows
that coefficients of fundamental harmonic of nonnegativewaveform of type (35) with two zeros are
1198861= minus1205821cos(120591
0minus120585
119896) 119887
1= minus1205821sin(120591
0minus120585
119896) (84)
where 1205821is amplitude of fundamental harmonic
1205821=
119896 sin 120585sin (120585119896)
120582119896 (85)
Coefficients of 119896th harmonic are given by (45)-(46)Notice that (68) can be rewritten as
120582119896= [cos(120585
119896)
119896 sin 120585sin(120585119896)
minus cos 120585]minus1
(86)
By introducing new variable
119909 = cos(120585119896) (87)
and using the Chebyshev polynomials (eg see Appendices)relations (85) and (86) can be rewritten as
1205821= 119896120582119896119880119896minus1
(119909) (88)
120582119896=
1
119896119909119880119896minus1
(119909) minus 119881119896(119909)
(89)
where119881119896(119909) and119880
119896(119909) denote the Chebyshev polynomials of
the first and second kind respectively From (89) it followsthat
120582119896[119896119909119880119896minus1
(119909) minus 119881119896(119909)] minus 1 = 0 (90)
which is polynomial equation of 119896th degree in terms of var-iable 119909 From 0 lt |120585| lt 120587 and (87) it follows that
cos(120587119896) lt 119909 lt 1 (91)
Since 120582119896is monotonically increasing function of |120585| 0 lt |120585| lt
120587 it follows that 120582119896is monotonically decreasing function of
119909 This further implies that (90) has only one solution thatsatisfies (91) (For 119896 = 2 expression (91) reads cos(1205872) le
119909 lt 1) This solution for 119909 (which can be obtained at leastnumerically) according to (88) leads to amplitude 120582
1of
fundamental harmonicFor 119896 le 4 solutions of (90) and (91) are
119909 = radic1 minus 1205822
21205822
1
3lt 1205822le 1
119909 =1
23radic1205823
1
8lt 1205823lt 1
119909 = radic1
6(1 + radic
51205824+ 3
21205824
)1
15lt 1205824lt 1
(92)
Insertion of (92) into (88) leads to the following relationsbetween amplitude 120582
1of fundamental and amplitude 120582
119896of
119896th harmonic 119896 le 4
1205821= radic8120582
2(1 minus 120582
2)
1
3lt 1205822le 1 (93)
1205821= 3 (
3radic1205823minus 1205823)
1
8lt 1205823lt 1 (94)
1205821= radic
32
27(radic2120582
4(3 + 5120582
4)3
minus 21205824(9 + 7120582
4))
1
15lt 1205824lt 1
(95)
Proof of Proposition 9 As it has been shown earlier (seeProposition 6) nonnegative waveform of type (35) with atleast one zero can be represented in form (38) Since weexclude nonnegative waveforms with 120582
119896= 1 according to
Remark 7 it follows that we exclude case |120585| = 120587Therefore inthe quest for nonnegative waveforms of type (35) having twozeros we will start with waveforms of type (38) for |120585| lt 120587It is clear that nonnegative waveforms of type (38) have twozeros if and only if
120582119896= [max120591
119903119896(120591)]minus1
(96)
and max120591119903119896(120591) = 119903
119896(1205910) According to (64) max
120591119903119896(120591) =
119903119896(1205910) implies |120585| = 0 Therefore it is sufficient to consider
only the interval (69)Substituting (96) into (38) we obtain
119879119896(120591) =
[1 minus cos (120591 minus 1205910)] [max
120591119903119896(120591) minus 119903
119896(120591)]
max120591119903119896(120591)
(97)
Mathematical Problems in Engineering 11
Expression max120591119903119896(120591) minus 119903
119896(120591) according to (64) and (39)
equals
max120591
119903119896(120591) minus 119903
119896(120591) = 119896
sin ((119896 minus 1) 120585119896)sin (120585119896)
minus 2
119896minus1
sum
119899=1
(119896 minus 119899) cos (119899 (120591 minus 1205910) + 120585)
(98)
Comparison of (97) with (66) yields
max120591
119903119896(120591) minus 119903
119896(120591) = [1 minus cos(120591 minus 120591
0+2120585
119896)]
sdot [1198880+ 2
119896minus2
sum
119899=1
119888119899cos 119899(120591 minus 120591
0+120585
119896)]
(99)
where coefficients 119888119899 119899 = 0 119896 minus 2 are given by (67) In
what follows we are going to show that right hand sides of(98) and (99) are equal
From (67) it follows that
1198880minus 1198881cos(120585
119896) = 119896
sin (120585 minus 120585119896)sin (120585119896)
(100)
Also from (67) for 119899 = 1 119896minus3 it follows that the followingrelations hold
(119888119899minus1
+ 119888119899+1
) cos(120585119896) minus 2119888
119899= 2 (119896 minus 119899) cos(120585 minus 119899120585
119896)
(119888119899minus1
minus 119888119899+1
) sin(120585119896) = 2 (119896 minus 119899) sin(120585 minus 119899120585
119896)
(101)
From (99) by using (75) (76) (100)-(101) and trigonometricidentities
cos(120591 minus 1205910+2120585
119896) = cos(120585
119896) cos(120591 minus 120591
0+120585
119896)
minus sin(120585119896) sin(120591 minus 120591
0+120585
119896)
cos(120585 minus 119899120585
119896) cos(119899(120591 minus 120591
0+120585
119896))
minus sin(120585 minus 119899120585
119896) sin(119899(120591 minus 120591
0+120585
119896))
= cos (119899 (120591 minus 1205910) + 120585)
(102)
we obtain (98) Consequently (98) and (99) are equal whichcompletes the proof
33 Nonnegative Waveforms with Two Zeros and PrescribedCoefficients of 119896thHarmonic In this subsectionwe show thatfor prescribed coefficients 119886
119896and 119887119896 there are 119896 nonnegative
waveforms of type (35) with exactly two zeros According to
(37) and (82) coefficients 119886119896and 119887119896of nonnegative waveforms
of type (35) with exactly two zeros satisfy the followingrelation
1
1198962 minus 1lt radic1198862119896+ 1198872119896lt 1 (103)
According to Remark 16 the value of 119909 (see (87)) that cor-responds to 120582
119896= radic1198862
119896+ 1198872119896can be determined from (90)-
(91) As we mentioned earlier (90) has only one solutionthat satisfies (91) This value of 119909 according to (88) leadsto the amplitude 120582
1of fundamental harmonic (closed form
expressions for 1205821in terms of 120582
119896and 119896 le 4 are given by (93)ndash
(95))On the other hand from (45)-(46) it follows that
1198961205910minus 120585 = atan 2 (119887
119896 119886119896) + 2119902120587 119902 = 1 (119896 minus 1) (104)
where function atan 2(119910 119909) is defined as
atan 2 (119910 119909) =
arctan(119910
119909) if 119909 ge 0
arctan(119910
119909) + 120587 if 119909 lt 0 119910 ge 0
arctan(119910
119909) minus 120587 if 119909 lt 0 119910 lt 0
(105)
with the codomain (minus120587 120587] Furthermore according to (84)and (104) the coefficients of fundamental harmonic of non-negative waveforms with two zeros and prescribed coeffi-cients of 119896th harmonic are equal to
1198861= minus1205821cos[
atan 2 (119887119896 119886119896) + 2119902120587
119896]
1198871= minus1205821sin[
atan 2 (119887119896 119886119896) + 2119902120587
119896]
(106)
where 119902 = 0 (119896minus 1) For chosen 119902 according to (104) and(66) positions of zeros are
1205910=1
119896[120585 + atan 2 (119887
119896 119886119896) + 2119902120587]
1205910minus2120585
119896=1
119896[minus120585 + atan 2 (119887
119896 119886119896) + 2119902120587]
(107)
From (106) and 119902 = 0 (119896minus1) it follows that for prescribedcoefficients 119886
119896and 119887119896 there are 119896 nonnegative waveforms of
type (35) with exactly two zerosWe provide here an algorithm to facilitate calculation
of coefficients 1198861and 1198871of nonnegative waveforms of type
(35) with two zeros and prescribed coefficients 119886119896and 119887
119896
providing that 119886119896and 119887119896satisfy (103)
12 Mathematical Problems in Engineering
2
1
0
Wav
efor
ms
0 1 2 3 4
Angle 120591120587
q = 0
q = 1
q = 2
Figure 6 Nonnegative waveforms with two zeros for 119896 = 3 1198863=
minus015 and 1198873= minus02
Algorithm 17 (i) Calculate 120582119896= radic1198862119896+ 1198872119896
(ii) identify 119909 that satisfies both relations (90) and (91)(iii) calculate 120582
1according to (88)
(iv) choose integer 119902 such that 0 le 119902 le 119896 minus 1(v) calculate 119886
1and 1198871according to (106)
For 119896 le 4 by using (93) for 119896 = 2 (94) for 119896 = 3 and (95)for 119896 = 4 it is possible to calculate directly 120582
1from 120582
119896and
proceed to step (iv)For 119896 = 2 and prescribed coefficients 119886
2and 1198872 there are
two waveforms with two zeros one corresponding to 1198861lt 0
and the other corresponding to 1198861gt 0 (see also [12])
Let us take as an input 119896 = 3 1198863= minus015 and 119887
3= minus02
Execution of Algorithm 17 on this input yields 1205823= 025 and
1205821= 11399 (according to (94)) For 119902 = 0 we calculate
1198861= minus08432 and 119887
1= 07670 (corresponding waveform is
presented by solid line in Figure 6) for 119902 = 1 we calculate1198861= minus02426 and 119887
1= minus11138 (corresponding waveform is
presented by dashed line) for 119902 = 2 we calculate 1198861= 10859
and 1198871= 03468 (corresponding waveform is presented by
dotted line)As another example of the usage of Algorithm 17 let us
consider case 119896 = 4 and assume that1198864= minus015 and 119887
4= minus02
Consequently 1205824= 025 and 120582
1= 09861 (according to (95))
For 119902 = 0 3we calculate the following four pairs (1198861 1198871) of
coefficients of fundamental harmonic (minus08388 05184) for119902 = 0 (minus05184 minus08388) for 119902 = 1 (08388 minus05184) for 119902 =2 and (05184 08388) for 119902 = 3 Corresponding waveformsare presented in Figure 7
4 Nonnegative Waveforms with MaximalAmplitude of Fundamental Harmonic
In this section we provide general description of nonnegativewaveforms containing fundamental and 119896th harmonic withmaximal amplitude of fundamental harmonic for prescribedamplitude of 119896th harmonic
The main result of this section is presented in the fol-lowing proposition
3
2
1
0
Wav
efor
ms
0 1 2 3 4
Angle 120591120587
q = 0q = 1
q = 2q = 3
Figure 7 Nonnegative waveforms with two zeros for 119896 = 4 1198864=
minus015 and 1198874= minus02
Proposition 18 Every nonnegativewaveformof type (35)withmaximal amplitude 120582
1of fundamental harmonic and pre-
scribed amplitude 120582119896of 119896th harmonic can be expressed in the
following form
119879119896(120591) = [1 minus cos (120591 minus 120591
0)]
sdot [1 minus (119896 minus 1) 120582119896minus 2120582119896
119896minus1
sum
119899=1
(119896 minus 119899) cos (119899 (120591 minus 1205910))]
(108)
if 0 le 120582119896le 1(119896
2
minus 1) or
119879119896(120591) = 120582
119896[1 minus cos (120591 minus 120591
0)] [1 minus cos(120591 minus 120591
0+2120585
119896)]
sdot [1198880+ 2
119896minus2
sum
119899=1
119888119899cos 119899(120591 minus 120591
0+120585
119896)]
(109)
if 1(1198962 minus 1) le 120582119896le 1 providing that 119888
119899 119899 = 0 119896 minus 2 and
120582119896are related to 120585 via relations (67) and (68) respectively and
|120585| le 120587
Remark 19 Expression (108) can be obtained from (38) bysetting 120585 = 0 Furthermore insertion of 120585 = 0 into (43)ndash(46)leads to the following expressions for coefficients ofwaveformof type (108)
1198861= minus (1 + 120582
119896) cos 120591
0 119887
1= minus (1 + 120582
119896) sin 120591
0
119886119896= 120582119896cos (119896120591
0) 119887
119896= 120582119896sin (119896120591
0)
(110)
On the other hand (109) coincides with (66) Thereforethe expressions for coefficients of (109) and (66) also coincideThus expressions for coefficients of fundamental harmonic ofwaveform (109) are given by (84) where 120582
1is given by (85)
while expressions for coefficients of 119896th harmonic are givenby (45)-(46)
Waveforms described by (108) have exactly one zerowhile waveforms described by (109) for 1(1198962 minus 1) lt 120582
119896lt 1
Mathematical Problems in Engineering 13
14
12
1
08
06
04
02
00 05 1
Amplitude 120582k
Am
plitu
de1205821
k = 2
k = 3
k = 4
Figure 8 Maximal amplitude of fundamental harmonic as a func-tion of amplitude of 119896th harmonic
have exactly two zeros As we mentioned earlier waveforms(109) for 120582
119896= 1 have 119896 zeros
Remark 20 Maximal amplitude of fundamental harmonic ofnonnegative waveforms of type (35) for prescribed amplitudeof 119896th harmonic can be expressed as
1205821= 1 + 120582
119896 (111)
if 0 le 120582119896le 1(119896
2
minus 1) or
1205821=
119896 sin 120585119896 sin 120585 cos (120585119896) minus cos 120585 sin (120585119896)
(112)
if 1(1198962 minus 1) le 120582119896le 1 where 120585 is related to 120582
119896via (68) (or
(86)) and |120585| le 120587From (110) it follows that (111) holds Substitution of (86)
into (85) leads to (112)Notice that 120582
119896= 1(119896
2
minus 1) is the only common point ofthe intervals 0 le 120582
119896le 1(119896
2
minus 1) and 1(1198962
minus 1) le 120582119896le
1 According to (111) 120582119896= 1(119896
2
minus 1) corresponds to 1205821=
1198962
(1198962
minus1) It can be also obtained from (112) by setting 120585 = 0The waveforms corresponding to this pair of amplitudes aremaximally flat nonnegative waveforms
Maximal amplitude of fundamental harmonic of non-negative waveform of type (35) for 119896 le 4 as a function ofamplitude of 119896th harmonic is presented in Figure 8
Remark 21 Maximum value of amplitude of fundamentalharmonic of nonnegative waveform of type (35) is
1205821max =
1
cos (120587 (2119896)) (113)
This maximum value is attained for |120585| = 1205872 (see (112)) Thecorresponding value of amplitude of 119896th harmonic is 120582
119896=
(1119896) tan(120587(2119896)) Nonnegative waveforms of type (35) with1205821= 1205821max have two zeros at 1205910 and 1205910 minus 120587119896 for 120585 = 1205872 or
at 1205910and 1205910+ 120587119896 for 120585 = minus1205872
14
12
1
08
06
04
02
0minus1 minus05 0 05 1
Am
plitu
de1205821
Parameter 120585120587
k = 2k = 3k = 4
Figure 9 Maximal amplitude of fundamental harmonic as a func-tion of parameter 120585
To prove that (113) holds let us first show that the fol-lowing relation holds for 119896 ge 2
cos( 120587
2119896) lt 1 minus
1
1198962 (114)
From 119896 ge 2 it follows that sinc(120587(4119896)) gt sinc(1205874) wheresinc 119909 = (sin119909)119909 and therefore sin(120587(4119896)) gt 1(radic2119896)By using trigonometric identity cos 2119909 = 1 minus 2sin2119909 weimmediately obtain (114)
According to (111) and (112) it is clear that 1205821attains its
maximum value on the interval 1(1198962 minus 1) le 120582119896le 1 Since
120582119896is monotonic function of |120585| on interval |120585| le 120587 (see
Remark 15) it follows that 119889120582119896119889120585 = 0 for 0 lt |120585| lt 120587
Therefore to find critical points of 1205821as a function of 120582
119896
it is sufficient to find critical points of 1205821as a function of
|120585| 0 lt |120585| lt 120587 and consider its values at the end points120585 = 0 and |120585| = 120587 Plot of 120582
1as a function of parameter 120585
for 119896 le 4 is presented in Figure 9 According to (112) firstderivative of 120582
1with respect to 120585 is equal to zero if and only
if (119896 cos 120585 sin(120585119896) minus sin 120585 cos(120585119896)) cos 120585 = 0 On interval0 lt |120585| lt 120587 this is true if and only if |120585| = 1205872 Accordingto (112) 120582
1is equal to 119896
2
(1198962
minus 1) for 120585 = 0 equal to zerofor |120585| = 120587 and equal to 1 cos(120587(2119896)) for |120585| = 1205872 From(114) it follows that 1198962(1198962minus1) lt 1 cos(120587(2119896)) and thereforemaximum value of 120582
1is given by (113) Moreover maximum
value of 1205821is attained for |120585| = 1205872
According to above consideration all nonnegative wave-forms of type (35) having maximum value of amplitude offundamental harmonic can be obtained from (109) by setting|120585| = 1205872 Three of them corresponding to 119896 = 3 120585 = 1205872and three different values of 120591
0(01205876 and1205873) are presented
in Figure 10 Dotted line corresponds to 1205910= 0 (coefficients
of corresponding waveform are 1198861= minus1 119887
1= 1radic3 119886
3= 0
and 1198873= minusradic39) solid line to 120591
0= 1205876 (119886
1= minus2radic3 119887
1= 0
1198863= radic39 and 119887
3= 0) and dashed line to 120591
0= 1205873 (119886
1= minus1
1198871= minus1radic3 119886
3= 0 and 119887
3= radic39)
Proof of Proposition 18 As it has been shown earlier (Propo-sition 6) nonnegative waveform of type (35) with at least
14 Mathematical Problems in Engineering
2
1
0
0 1 2 3 4
Angle 120591120587
Wav
efor
ms
1205910 = 01205910 = 12058761205910 = 1205873
Figure 10 Nonnegative waveforms with maximum amplitude offundamental harmonic for 119896 = 3 and 120585 = 1205872
one zero can be represented in form (38) According to (43)(44) and (36) for amplitude 120582
1of fundamental harmonic of
waveforms of type (38) the following relation holds
1205821= radic(1 + 120582
119896cos 120585)2 + 11989621205822
119896sin2120585 (115)
where 120582119896satisfy (40) and |120585| le 120587
Because of (40) in the quest of finding maximal 1205821for
prescribed 120582119896 we have to consider the following two cases
(Case i)120582119896lt [(119896minus1) cos 120585 + 119896 sin(120585minus120585119896) sin(120585119896)]minus1
(Case ii)120582119896= [(119896minus1) cos 120585+119896 sin(120585minus120585119896) sin(120585119896)]minus1
Case i Since 120582119896lt [(119896 minus 1) cos 120585 + 119896 sin(120585 minus 120585119896) sin(120585119896)]minus1
implies 120582119896
= 1 according to (115) it follows that 1205821
= 0Hence 119889120582
1119889120585 = 0 implies
2120582119896sin 120585 [1 minus (1198962 minus 1) 120582
119896cos 120585] = 0 (116)
Therefore 1198891205821119889120585 = 0 if 120582
119896= 0 (Option 1) or sin 120585 = 0
(Option 2) or (1198962 minus 1)120582119896cos 120585 = 1 (Option 3)
Option 1 According to (115) 120582119896= 0 implies 120582
1= 1 (notice
that this implication shows that 1205821does not depend on 120585 and
therefore we can set 120585 to zero value)
Option 2 According to (115) sin 120585 = 0 implies 1205821= 1 +
120582119896cos 120585 which further leads to the conclusion that 120582
1is
maximal for 120585 = 0 For 120585 = 0 120582119896lt [(119896 minus 1) cos 120585 + 119896 sin(120585 minus
120585119896) sin(120585119896)]minus1 becomes 120582119896lt 1(119896
2
minus 1)
Option 3 This option leads to contradiction To show thatnotice that (119896
2
minus 1)120582119896cos 120585 = 1 and 120582
119896lt [(119896 minus
1) cos 120585 + 119896 sin(120585 minus 120585119896) sin(120585119896)]minus1 imply that (119896 minus 1) cos 120585 gtsin(120585minus120585119896) sin(120585119896) Using (A5) (see Appendices) the latestinequality can be rewritten assum119896minus1
119899=1[cos 120585minuscos((119896minus2119899)120585119896)] gt
0 But from |119896 minus 2119899| lt 119896 119899 = 1 (119896 minus 1) and |120585| le 120587
it follows that all summands are not positive and therefore(119896minus1) cos 120585 gt sin(120585minus120585119896) sin(120585119896) does not hold for |120585| le 120587
Consequently Case i implies 120585 = 0 and 120582119896lt 1(119896
2
minus 1)Finally substitution of 120585 = 0 into (38) leads to (108) whichproves that (108) holds for 120582
119896lt 1(119896
2
minus 1)
Case ii Relation120582119896= [(119896minus1) cos 120585+119896 sin(120585minus120585119896) sin(120585119896)]minus1
according to Proposition 9 and Remark 11 implies that cor-responding waveforms can be expressed via (66)ndash(68) for|120585| le 120587 Furthermore 120582
119896= [(119896 minus 1) cos 120585 + 119896 sin(120585 minus
120585119896) sin(120585119896)]minus1 and |120585| le 120587 imply 1(1198962 minus 1) le 120582119896le 1
This proves that (109) holds for 1(1198962 minus 1) le 120582119896le 1
Finally let us prove that (108) holds for 120582119896= 1(119896
2
minus
1) According to (68) (see also Remark 11) this value of 120582119896
corresponds to 120585 = 0 Furthermore substitution of 120582119896=
1(1198962
minus 1) and 120585 = 0 into (109) leads to (70) which can berewritten as
119879119896(120591) =
[1 minus cos (120591 minus 1205910)]
(1 minus 1198962)
sdot [119896 (119896 minus 1) minus 2
119896minus1
sum
119899=1
(119896 minus 119899) cos (119899 (120591 minus 1205910))]
(117)
Waveform (117) coincides with waveform (108) for 120582119896
=
1(1 minus 1198962
) Consequently (108) holds for 120582119896= 1(1 minus 119896
2
)which completes the proof
5 Nonnegative Waveforms with MaximalAbsolute Value of the Coefficient of CosineTerm of Fundamental Harmonic
In this sectionwe consider general description of nonnegativewaveforms of type (35) with maximal absolute value ofcoefficient 119886
1for prescribed coefficients of 119896th harmonicThis
type of waveform is of particular interest in PA efficiencyanalysis In a number of cases of practical interest eithercurrent or voltage waveform is prescribed In such casesthe problem of finding maximal efficiency of PA can bereduced to the problem of finding nonnegative waveformwith maximal coefficient 119886
1for prescribed coefficients of 119896th
harmonic (see also Section 7)In Section 51 we provide general description of nonneg-
ative waveforms of type (35) with maximal absolute value ofcoefficient 119886
1for prescribed coefficients of 119896th harmonic In
Section 52 we illustrate results of Section 51 for particularcase 119896 = 3
51 Nonnegative Waveforms with Maximal Absolute Value ofCoefficient 119886
1for 119896 ge 2 Waveforms 119879
119896(120591) of type (35) with
1198861ge 0 can be derived from those with 119886
1le 0 by shifting
by 120587 and therefore we can assume without loss of generalitythat 119886
1le 0 Notice that if 119896 is even then shifting 119879
119896(120591) by
120587 produces the same result as replacement of 1198861with minus119886
1
(119886119896remains the same) On the other hand if 119896 is odd then
shifting 119879119896(120591) by 120587 produces the same result as replacement
of 1198861with minus119886
1and 119886119896with minus119886
119896
According to (37) coefficients of 119896th harmonic can beexpressed as
119886119896= 120582119896cos 120575 119887
119896= 120582119896sin 120575 (118)
Mathematical Problems in Engineering 15
where
|120575| le 120587 (119)
Conversely for prescribed coefficients 119886119896and 119887
119896 120575 can be
determined as
120575 = atan 2 (119887119896 119886119896) (120)
where definition of function atan 2(119910 119909) is given by (105)The main result of this section is stated in the following
proposition
Proposition 22 Every nonnegative waveform of type (35)withmaximal absolute value of coefficient 119886
1le 0 for prescribed
coefficients 119886119896and 119887119896of 119896th harmonic can be represented as
119879119896(120591)
= [1 minus cos 120591]
sdot [1 minus (119896 minus 1) 119886119896minus 2
119896minus1
sum
119899=1
(119896 minus 119899) (119886119896cos 119899120591 + 119887
119896sin 119899120591)]
(121)
if 119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) le 1+119886
119896 where 120575 = atan 2(bk
119886119896) or
119879119896(120591) = 120582
119896[1 minus cos(120591 minus (120575 + 120585)
119896)]
sdot [1 minus cos(120591 minus (120575 minus 120585)
119896)]
sdot [1198880+ 2
119896minus2
sum
119899=1
119888119899cos 119899(120591 minus 120575
119896)]
(122)
if 119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) ge 1 + 119886
119896 where 119888
119899 119899 = 0
119896minus2 and 120582119896= radic1198862119896+ 1198872119896are related to 120585 via relations (67) and
(68) respectively and |120585| le 120587
Remark 23 Expression (121) can be obtained from (38) bysetting 120591
0= 0 and 120585 = minus120575 and then replacing 120582
119896cos 120575 with
119886119896(see (118)) and 120582
119896cos(119899120591 minus 120575) with 119886
119896cos 119899120591 + 119887
119896sin 119899120591
(see also (118)) Furthermore insertion of 1205910= 0 and 120585 =
minus120575 into (43)ndash(46) leads to the following relations betweenfundamental and 119896th harmonic coefficients of waveform(121)
1198861= minus (1 + 119886
119896) 119887
1= minus119896119887
119896 (123)
On the other hand expression (122) can be obtained from(66) by replacing 120591
0minus120585119896with 120575119896 Therefore substitution of
1205910minus 120585119896 = 120575119896 in (84) leads to
1198861= minus1205821cos(120575
119896) 119887
1= minus1205821sin(120575
119896) (124)
where 1205821is given by (85)
The fundamental harmonic coefficients 1198861and 1198871of wave-
form of type (35) with maximal absolute value of coefficient1198861le 0 satisfy both relations (123) and (124) if 119886
119896and 119887119896satisfy
1 + 119886119896= 119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) For such waveforms
relations 1205910= 0 and 120585 = minus120575 also hold
Remark 24 Amplitude of 119896th harmonic of nonnegativewaveform of type (35) with maximal absolute value of coeffi-cient 119886
1le 0 and coefficients 119886
119896 119887119896satisfying 1 + 119886
119896=
119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) is
120582119896=
sin (120575119896)119896 sin 120575 cos (120575119896) minus cos 120575 sin (120575119896)
(125)
To show that it is sufficient to substitute 119886119896= 120582119896cos 120575 (see
(118)) into 1 + 119886119896= 119896120582119896[sin 120575 sin(120575119896)] cos(120575119896)
Introducing new variable
119910 = cos(120575119896) (126)
and using the Chebyshev polynomials (eg see Appendices)relations 119886
119896= 120582119896cos 120575 and (125) can be rewritten as
119886119896= 120582119896119881119896(119910) (127)
120582119896=
1
119896119910119880119896minus1
(119910) minus 119881119896(119910)
(128)
where119881119896(119910) and119880
119896(119910) denote the Chebyshev polynomials of
the first and second kind respectively Substitution of (128)into (127) leads to
119886119896119896119910119880119896minus1
(119910) minus (1 + 119886119896) 119881119896(119910) = 0 (129)
which is polynomial equation of 119896th degree in terms of var-iable 119910 From |120575| le 120587 and (126) it follows that
cos(120587119896) le 119910 le 1 (130)
In what follows we show that 119886119896is monotonically increas-
ing function of 119910 on the interval (130) From 120585 = minus120575 (seeRemark 23) and (81) it follows that 120582minus1
119896= (119896 minus 1) cos 120575 +
119896sum119896minus1
119899=1cos((119896 minus 2119899)120575119896) ge 1 and therefore 119886
119896= 120582119896cos 120575 can
be rewritten as
119886119896=
cos 120575(119896 minus 1) cos 120575 + 119896sum119896minus1
119899=1cos ((119896 minus 2119899) 120575119896)
(131)
Obviously 119886119896is even function of 120575 and all cosines in (131)
are monotonically decreasing functions of |120575| on the interval|120575| le 120587 It is easy to show that cos((119896 minus 2119899)120575119896) 119899 =
1 (119896 minus 1) decreases slower than cos 120575 when |120575| increasesThis implies that denominator of the right hand side of(131) decreases slower than numerator Since denominator ispositive for |120575| le 120587 it further implies that 119886
119896is decreasing
function of |120575| on interval |120575| le 120587 Consequently 119886119896is
monotonically increasing function of 119910 on the interval (130)Thus we have shown that 119886
119896is monotonically increasing
function of 119910 on the interval (130) and therefore (129) hasonly one solution that satisfies (130) According to (128) thevalue of 119910 obtained from (129) and (130) either analyticallyor numerically leads to amplitude 120582
119896of 119896th harmonic
16 Mathematical Problems in Engineering
1
05
0
minus05
minus1
minus1 minus05 0 05 1
Coefficient ak
Coe
ffici
entb
k
radica2k+ b2
kle 1
k = 2k = 3k = 4
Figure 11 Plot of (119886119896 119887119896) satisfying 1 + 119886
119896= 119896120582
119896[sin 120575 sin(120575
119896)] cos(120575119896) for 119896 le 4
By solving (129) and (130) for 119896 le 4 we obtain
119910 = radic1 + 1198862
2 (1 minus 1198862) minus1 le 119886
2le1
3
119910 = radic3
4 (1 minus 21198863) minus1 le 119886
3le1
8
119910 =radicradic2 minus 4119886
4+ 1011988624minus 2 (1 minus 119886
4)
4 (1 minus 31198864)
minus1 le 1198864le
1
15
(132)
Insertion of (132) into (128) leads to the following explicitexpressions for the amplitude 120582
119896 119896 le 4
1205822=1
2(1 minus 119886
2) minus1 le 119886
2le1
3 (133)
1205822
3= [
1
3(1 minus 2119886
3)]
3
minus1 le 1198863le1
8 (134)
1205824=1
4(minus1 minus 119886
4+ radic2 minus 4119886
4+ 1011988624) minus1 le 119886
4le
1
15
(135)
Relations (133)ndash(135) define closed lines (see Figure 11) whichseparate points representing waveforms of type (121) frompoints representing waveforms of type (122) For given 119896points inside the corresponding curve refer to nonnegativewaveforms of type (121) whereas points outside curve (andradic1198862119896+ 1198872119896le 1) correspond to nonnegative waveforms of type
(122) Points on the respective curve correspond to the wave-forms which can be expressed in both forms (121) and (122)
Remark 25 Themaximum absolute value of coefficient 1198861of
nonnegative waveform of type (35) is
100381610038161003816100381611988611003816100381610038161003816max =
1
cos (120587 (2119896)) (136)
This maximum value is attained for |120585| = 1205872 and 120575 = 0
(see (124)) Notice that |1198861|max is equal to the maximum value
1205821max of amplitude of fundamental harmonic (see (113))
Coefficients of waveform with maximum absolute value ofcoefficient 119886
1 1198861lt 0 are
1198861= minus
1
cos (120587 (2119896)) 119886
119896=1
119896tan( 120587
(2119896))
1198871= 119887119896= 0
(137)
Waveformdescribed by (137) is cosinewaveformhaving zerosat 120587(2119896) and minus120587(2119896)
In the course of proving (136) notice first that |1198861|max le
1205821max holds According to (123) and (124) maximum of |119886
1|
occurs for 119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) ge 1 + 119886
119896 From (124)
it immediately follows that maximum value of |1198861| is attained
if and only if 1205821= 1205821max and 120575 = 0 which because of
120575119896 = 1205910minus120585119896 further implies 120591
0= 120585119896 Sincemaximumvalue
of 1205821is attained for |120585| = 1205872 it follows that corresponding
waveform has zeros at 120587(2119896) and minus120587(2119896)
Proof of Proposition 22 As it was mentioned earlier in thissection we can assume without loss of generality that 119886
1le 0
We consider waveforms119879119896(120591) of type (35) such that119879
119896(120591) ge 0
and119879119896(120591) = 0 for some 120591
0 Fromassumption that nonnegative
waveform 119879119896(120591) of type (35) has at least one zero it follows
that it can be expressed in form (38)Let us also assume that 120591
0is position of nondegenerate
critical point Therefore 119879119896(1205910) = 0 implies 1198791015840
119896(1205910) = 0 and
11987910158401015840
119896(1205910) gt 0 According to (55) second derivative of 119879
119896(120591) at
1205910can be expressed as 11987910158401015840
119896(1205910) = 1 minus 120582
119896(1198962
minus 1) cos 120585 Since11987910158401015840
119896(1205910) gt 0 it follows immediately that
1 minus 120582119896(1198962
minus 1) cos 120585 gt 0 (138)
Let us further assume that 119879119896(120591) has exactly one zeroThe
problem of finding maximum absolute value of 1198861is con-
nected to the problem of finding maximum of the minimumfunction (see Section 21) If waveforms possess unique globalminimum at nondegenerate critical point then correspond-ing minimum function is a smooth function of parameters[13] Consequently assumption that 119879
119896(120591) has exactly one
zero at nondegenerate critical point leads to the conclusionthat coefficient 119886
1is differentiable function of 120591
0 First
derivative of 1198861(see (43)) with respect to 120591
0 taking into
account that 1205971205851205971205910= 119896 (see (50)) can be expressed in the
following factorized form
1205971198861
1205971205910
= sin 1205910[1 minus 120582
119896(1198962
minus 1) cos 120585] (139)
Mathematical Problems in Engineering 17
From (138) and (139) it is clear that 12059711988611205971205910= 0 if and only if
sin 1205910= 0 According toRemark 12 assumption that119879
119896(120591)has
exactly one zero implies 120582119896lt 1 From (51) (48) and 120582
119896lt 1
it follows that 1198861cos 1205910+ 1198871sin 1205910lt 0 which together with
sin 1205910= 0 implies that 119886
1cos 1205910lt 0 Assumption 119886
1le 0
together with relations 1198861cos 1205910lt 0 and sin 120591
0= 0 further
implies 1198861
= 0 and
1205910= 0 (140)
Insertion of 1205910= 0 into (38) leads to
119879119896(120591)
= [1 minus cos 120591]
sdot [1 minus (119896 minus 1) 120582119896cos 120585 minus 2120582
119896
119896minus1
sum
119899=1
(119896 minus 119899) cos (119899120591 + 120585)]
(141)
Substitution of 1205910= 0 into (45) and (46) yields 119886
119896= 120582119896cos 120585
and 119887119896
= minus120582119896sin 120585 respectively Replacing 120582
119896cos 120585 with
119886119896and 120582
119896cos(119899120591 + 120585) with (119886
119896cos 119899120591 + 119887
119896sin 119899120591) in (141)
immediately leads to (121)Furthermore 119886
119896= 120582119896cos 120585 119887
119896= minus120582
119896sin 120585 and (118)
imply that
120575 = minus120585 (142)
According to (38)ndash(40) and (142) it follows that (141) is non-negative if and only if
120582119896[(119896 minus 1) cos 120575 + 119896 sin (120575 minus 120575119896)
sin (120575119896)] lt 1 (143)
Notice that 119886119896= 120582119896cos 120575 implies that the following relation
holds
120582119896[(119896 minus 1) cos 120575 + 119896 sin (120575 minus 120575119896)
sin (120575119896)]
= minus119886119896+ 119896120582119896
sin 120575sin (120575119896)
cos(120575119896)
(144)
Finally substitution of (144) into (143) leads to 119896120582119896[sin 120575
sin(120575119896)] cos(120575119896) lt 1 + 119886119896 which proves that (121) holds
when 119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) lt 1 + 119886
119896
Apart from nonnegative waveforms with exactly one zeroat nondegenerate critical point in what follows we will alsoconsider other types of nonnegative waveforms with at leastone zero According to Proposition 9 and Remark 11 thesewaveforms can be described by (66)ndash(68) providing that 0 le|120585| le 120587
According to (35) 119879119896(0) ge 0 implies 1 + 119886
1+ 119886119896ge 0
Consequently 1198861le 0 implies that |119886
1| le 1 + 119886
119896 On the other
hand according to (123) |1198861| = 1 + 119886
119896holds for waveforms
of type (121) The converse is also true 1198861le 0 and |119886
1| =
1 + 119886119896imply 119886
1= minus1 minus 119886
119896 which further from (35) implies
119879119896(0) = 0 Therefore in what follows it is enough to consider
only nonnegativewaveformswhich can be described by (66)ndash(68) and 0 le |120585| le 120587 with coefficients 119886
119896and 119887119896satisfying
119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) ge 1 + 119886
119896
For prescribed coefficients 119886119896and 119887119896 the amplitude 120582
119896=
radic1198862119896+ 1198872119896of 119896th harmonic is also prescribed According to
Remark 15 (see also Remark 16) 120582119896is monotonically
decreasing function of 119909 = cos(120585119896) The value of 119909 can beobtained by solving (90) subject to the constraint cos(120587119896) le119909 le 1 Then 120582
1can be determined from (88) From (106) it
immediately follows that maximal absolute value of 1198861le 0
corresponds to 119902 = 0 which from (104) and (120) furtherimplies that
120575 = 1198961205910minus 120585 (145)
Furthermore 119902 = 0 according to (107) implies that waveformzeros are
1205910=(120575 + 120585)
119896 120591
1015840
0= 1205910minus2120585
119896=(120575 minus 120585)
119896 (146)
Substitution of 1205910= (120575 + 120585)119896 into (66) yields (122) which
proves that (122) holds when 119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) ge
1 + 119886119896
In what follows we prove that (121) also holds when119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) = 1 + 119886
119896 Substitution of 119886
119896=
120582119896cos 120575 into 119896120582
119896[sin 120575 sin(120575119896)] cos(120575119896) = 1 + 119886
119896leads to
120582119896[(119896 minus 1) cos 120575 + 119896 sin (120575 minus 120575119896)
sin (120575119896)] = 1 (147)
As we mentioned earlier relation (142) holds for all wave-forms of type (121) Substituting (142) into (147) we obtain
120582119896[(119896 minus 1) cos 120585 + 119896 sin (120585 minus 120585119896)
sin (120585119896)] = 1 (148)
This expression can be rearranged as
120582119896
119896 sin ((119896 minus 1) 120585119896)sin 120585119896
= 1 minus (119896 minus 1) 120582119896cos 120585 (149)
On the other hand for waveforms of type (122) according to(68) relations (148) and (149) also hold Substitution of 120591
0=
(120575 + 120585)119896 (see (145)) and (67) into (122) leads to
119879119896(120591)
= 120582119896[1 minus cos (120591 minus 120591
0)]
sdot [119896 sin ((119896 minus 1) 120585119896)
sin 120585119896minus 2
119896minus1
sum
119899=1
(119896 minus 119899) cos (119899 (120591 minus 1205910) + 120585)]
(150)
Furthermore substitution of (142) into (145) implies that1205910
= 0 Finally substitution of 1205910
= 0 and (149) into(150) leads to (141) Therefore (141) holds when 119896120582
119896[sin 120575
sin(120575119896)] cos(120575119896) = 1 + 119886119896 which in turn shows that (121)
holds when 119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) = 1 + 119886
119896 This
completes the proof
18 Mathematical Problems in Engineering
52 Nonnegative Waveforms with Maximal Absolute Valueof Coefficient 119886
1for 119896 = 3 Nonnegative waveform of type
(35) for 119896 = 3 is widely used in PA design (eg see [10])In this subsection we illustrate results of Section 51 for thisparticular case The case 119896 = 2 is presented in detail in [12]
Coefficients of fundamental harmonic of nonnegativewaveform of type (35) with 119896 = 3 and maximal absolutevalue of coefficient 119886
1le 0 for prescribed coefficients 119886
3and
1198873(1205823= radic11988623+ 11988723) according to (123) (124) (134) (94) and
(120) are equal to
1198861= minus1 minus 119886
3 119887
1= minus3119887
3 (151)
if 12058223le [(1 minus 2119886
3)3]3
1198861= minus1205821cos(120575
3) 119887
1= minus1205821sin(120575
3) (152)
where 1205821= 3(
3radic1205823minus 1205823) and 120575 = atan 2(119887
3 1198863) if [(1 minus
21198863)3]3
le 1205822
3le 1The line 1205822
3= [(1minus2119886
3)3]3 (see case 119896 = 3
in Figure 11) separates points representing waveforms withcoefficients satisfying (151) from points representing wave-forms with coefficients satisfying (152) Waveforms describedby (151) for 1205822
3lt [(1 minus 2119886
3)3]3 have exactly one zero at
1205910= 0 Waveforms described by (151) and (152) for 1205822
3= [(1 minus
21198863)3]3 also have zero at 120591
0= 0 These waveforms as a rule
have exactly two zeros However there are two exceptionsone related to the maximally flat nonnegative waveform withcoefficients 119886
1= minus98 119886
3= 18 and 119887
1= 1198873= 0 which
has only one zero and the other related to the waveform withcoefficients 119886
1= 0 119886
3= minus1 and 119887
1= 1198873= 0 which has three
zerosWaveforms described by (152) for [(1minus21198863)3]3
lt 1205822
3lt
1 have two zeros Waveforms with 1205823= 1 have only third
harmonic (fundamental harmonic is zero)Plot of contours of maximal absolute value of coefficient
1198861 1198861le 0 for prescribed coefficients 119886
3and 1198873is presented
in Figure 12 According to Remark 25 the waveform withmaximum absolute value of 119886
1le 0 is fully described with
the following coefficients 1198861
= minus2radic3 1198863
= radic39 and1198871= 1198873= 0 This waveform has two zeros at plusmn1205876
Two examples of nonnegative waveforms for 119896 = 3
and maximal absolute value of coefficient 1198861 1198861le 0 with
prescribed coefficients 1198863and 1198873are presented in Figure 13
One waveform corresponds to the case 12058223lt [(1 minus 2119886
3)3]3
(solid line) and the other to the case 12058223gt [(1 minus 2119886
3)3]3
(dashed line)Thewaveform represented by solid line has onezero and its coefficients are 119886
3= minus01 119887
3= 01 119886
1= minus09
and 1198871= minus03 Dashed line corresponds to the waveform
having two zeros with coefficients 1198863= minus01 119887
3= 03 119886
1=
minus08844 and 1198871= minus06460 (case 1205822
3gt [(1 minus 2119886
3)3]3)
6 Nonnegative Cosine Waveforms withat Least One Zero
Nonnegative cosine waveforms have proved to be of impor-tance for waveform modelling in PA design (eg see [10])In this section we consider nonnegative cosine waveforms
1
05
0
minus05
minus1
minus1 minus05 0 05 1
Coefficient a3
Coe
ffici
entb
3
02
04
06
08
10
11
Figure 12 Contours ofmaximal absolute value of coefficient 1198861 1198861le
0 as a function of 1198863and 1198873
2
1
0
0 1 2 3 4
Angle 120591120587
Wav
efor
ms
a3 = minus01 b3 = 01
a3 = minus01 b3 = 03
Figure 13 Nonnegative waveforms for 119896 = 3 and maximal absolutevalue of 119886
1 1198861le 0 with prescribed coefficients 119886
3and 1198873
containing fundamental and 119896th harmonic with at least onezero
Cosine waveform with dc component fundamental and119896th harmonic can be obtained from (35) by setting 119887
1= 119887119896=
0 that is
119879119896(120591) = 1 + 119886
1cos 120591 + 119886
119896cos 119896120591 (153)
In Section 61 we provide general description of non-negative cosine waveforms of type (153) with at least onezero We show that nonnegative cosine waveforms with atleast one zero coincide with nonnegative cosine waveformswith maximal absolute value of coefficient 119886
1for prescribed
coefficient 119886119896 In Section 62 we illustrate results of Section 61
for particular case 119896 = 3
61 Nonnegative Cosine Waveforms with at Least One Zerofor 119896 ge 2 Amplitudes of fundamental and 119896th harmonic
Mathematical Problems in Engineering 19
of cosine waveform of type (153) are 1205821= |1198861| and 120582
119896=
|119886119896| respectively According to (42) for nonnegative cosine
waveforms of type (153) the following relation holds
minus1 le 119886119896le 1 (154)
This explains why 119896th harmonic coefficient 119886119896in Proposi-
tion 26 goes through interval [minus1 1]Waveforms (153) with 119886
1ge 0 can be obtained from
waveforms with 1198861le 0 by shifting by 120587 and therefore with-
out loss of generality we can assume that 1198861le 0
Proposition 26 Each nonnegative cosine waveform of type(153) with 119886
1le 0 and at least one zero can be represented as
119879119896(120591) = [1 minus cos 120591] [1 minus (119896 minus 1) 119886
119896minus 2119886119896
119896minus1
sum
119899=1
(119896 minus 119899) cos 119899120591]
(155)
if minus1 le 119886119896le 1(119896
2
minus 1) or
119879119896(120591) = 119886
119896[1 minus cos (120591 minus 120591
0)] [1 minus cos (120591 + 120591
0)]
sdot [1198880+ 2
119896minus2
sum
119899=1
119888119899cos 119899120591]
(156)
where
119888119899=sin ((119896 minus 119899) 120591
0) cos 120591
0minus (119896 minus 119899) cos ((119896 minus 119899) 120591
0) sin 120591
0
sin31205910
(157)
119886119896=
sin 1205910
119896 sin (1198961205910) cos 120591
0minus cos (119896120591
0) sin 120591
0
(158)
100381610038161003816100381612059101003816100381610038161003816 le
120587
119896 (159)
if 1(1198962 minus 1) le 119886119896le 1
Remark 27 Identity [1minuscos(120591minus1205910)][1minuscos(120591+120591
0)] = [cos 120591
0minus
cos 120591]2 implies that (156) can be rewritten as
119879119896(120591) = 119886
119896[cos 120591
0minus cos 120591]2 [119888
0+ 2
119896minus2
sum
119899=1
119888119899cos 119899120591] (160)
Furthermore substitution of (157) into (160) leads to
119879119896(120591) = 119886
119896[cos 120591
0minus cos 120591]
sdot [(119896 minus 1) sin 119896120591
0
sin 1205910
minus 2
119896minus1
sum
119899=1
sin ((119896 minus 119899) 1205910)
sin 1205910
cos 119899120591]
(161)
Remark 28 All nonnegative cosine waveforms of type (153)with at least one zero and 119886
1le 0 except one of them can be
represented either in form (155) or form (156)This exceptionis maximally flat cosine waveform with 119886
1lt 0 which can be
obtained from (155) for 119886119896= 1(119896
2
minus 1) or from (156) for 1205910=
0 Maximally flat cosine waveform with 1198861lt 0 can also be
obtained from (70) by setting 1205910= 0 Furthermore setting
1205910= 0 in (71) leads to maximally flat cosine waveforms for
119896 le 4 and 1198861lt 0
Remark 29 Nonnegative cosine waveform of type (155) with1198861lt 0 and minus1 lt 119886
119896le 1(119896
2
minus 1) has exactly one zero at120591 = 0 Nonnegative cosine waveform described by (156) with1198861lt 0 and 1(1198962 minus 1) lt 119886
119896lt 1 has two zeros at plusmn120591
0 where
0 lt |1205910| lt 120587119896 For 119886
119896= ∓1 nonnegative cosine waveform
of type (153) reduces to 119879119896(120591) = 1 ∓ cos 119896120591 (clearly these two
waveforms both have 119896 zeros)
Remark 30 Transformation of (155) into an additive formleads to the following relation
1198861= minus1 minus 119886
119896 (162)
where minus1 le 119886119896le 1(119896
2
minus1) Similarly transformation of (156)leads to the following relation
1198861= minus119886119896
119896 sin 1198961205910
sin 1205910
(163)
where 119886119896is given by (158) 1(1198962minus1) le 119886
119896le 1 and |120591
0| le 120587119896
Notice that coefficients of maximally flat cosine waveformnamely 119886
119896= 1(119896
2
minus1) and 1198861= minus1198962
(1198962
minus1) satisfy relation(162) They also satisfy relation (163) for 120591
0= 0
Remark 31 Nonnegative cosine waveforms of type (153) withat least one zero coincide with nonnegative cosine waveformswith maximal absolute value of coefficient 119886
1for prescribed
coefficient 119886119896
In proving that Remark 31 holds notice that expression(155) can be obtained from (121) by setting 119887
119896= 0 Fur-
thermore if 119886119896
ge 0 then 120582119896
= 119886119896 which together
with 119887119896
= 0 and (118) implies 120575 = 0 In this case119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) le 1 + 119886
119896becomes 1198962119886
119896le 1 +
119886119896 On the other hand if 119886
119896lt 0 then 120582
119896= minus119886
119896 which
together with 119887119896= 0 and (118) implies |120575| = 120587 In this case
119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) le 1 + 119886
119896becomes 0 le 1 + 119886
119896
Therefore every nonnegative cosine waveform of type (155)has maximal absolute value of coefficient 119886
1for prescribed
coefficient 119886119896 when minus1 le 119886
119896le 1(119896
2
minus 1)Let us now show that expression (156) can be obtained
from (122) by setting 119887119896= 0 and 119886
119896gt 0 For waveforms
of type (122) according to (118) 119887119896= 0 and 119886
119896gt 0 imply
120575 = 0 and 120582119896= 119886119896 Substitution of 120582
119896= 119886119896and 120575 = 0 into
119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) ge 1+119886
119896leads to 119886
119896ge 1(119896
2
minus1)Furthermore substitution of 120575 = 0 into (145) yields 120591
0= 120585119896
Insertion of 120582119896= 119886119896 120575 = 0 and 120591
0= 120585119896 into (122) leads
to (156) Therefore every nonnegative cosine waveform oftype (156) has maximal absolute value of coefficient 119886
1for
prescribed coefficient 119886119896 when 1(1198962 minus 1) le 119886
119896le 1
Proof of Proposition 26 Let us start with nonnegative cosinewaveform of type (153) with 120582
119896= |119886119896| = 1 According to
Remark 7 120582119896= |119886119896| = 1 implies that 120582
1= |1198861| = 0
Substitution of 119886119896
= minus1 into (155) and using (A2) (seeAppendices) lead to 119879
119896(120591) = 1 minus cos 119896120591 Consequently (155)
holds for 119886119896= minus1 On the other hand substitution of 119886
119896= 1
into (158) yields |1205910| = 120587119896 Furthermore substitution of
119886119896= 1 and 120591
0= 120587119896 (or 120591
0= minus120587119896) into (156) along
with performing all multiplications and using (A2) leads to
20 Mathematical Problems in Engineering
119879119896(120591) = 1 + cos 119896120591 Consequently (156)ndash(158) hold for 119886
119896= 1
and |1205910| = 120587119896
It is easy to see that 120582119896= |119886119896| lt 1 and 119879
119896(1205910) = 0 for some
1205910imply 120582
1= |1198861| = 0 Therefore in what follows we assume
that |119886119896| = 1 and 119886
1lt 0
Cosine waveforms are even functions of 120591 Therefore ifnonnegative cosine waveform has exactly one zero it has to beeither at 0 or at 120587 On the other hand if nonnegative cosinewaveform with 119886
1= 0 has exactly two zeros then these zeros
are placed at plusmn1205910 such that 120591
0is neither 0 nor 120587
In order to prove that (155) holds for minus1 lt 119886119896le 1(1minus119896
2
)let us start by referring to the description (38) of nonnegativewaveformswith at least one zero As wementioned earlier fornonnegative cosine waveformwith exactly one zero (denotedby 1205910) it is either 120591
0= 0 or 120591
0= 120587 Therefore in both cases
sin 1205910= 0 Substitution of sin 120591
0= 0 into (43) together with
1198861
= 0 and 120582119896= |119886119896| lt 1 leads to
1205910= 0 (164)
Clearly 1205910= 0 119887
1= 0 and 119887
119896= 0 according to (44) and (46)
imply 120582119896sin 120585 = 0 Since 120582
119896= |119886119896| it follows that |119886
119896| sin 120585 = 0
also holds which further implies 120582119896= 119886119896= 0 or sin 120585 =
0 In the case when 120582119896= 119886119896= 0 from (164) and (43) we
obtain 1198861= minus1 which further implies that 119879
119896(120591) = 1 minus cos 120591
Consequently (155) holds for 119886119896= 0 In the case when sin 120585 =
0 from (164) and (45) we obtain 119886119896= 120582119896if 120585 = 0 or 119886
119896= minus120582119896
if 120585 = 120587 Relations 119886119896= 120582119896and 120585 = 0 according to (40) imply
that 0 le 119886119896le 1(1 minus 119896
2
) Substitution of 120585 = 0 120582119896= 119886119896 and
(164) into (38) leads to (155) which proves that (155) holdsfor 0 le 119886
119896le 1(1 minus 119896
2
) On the other hand relations 119886119896=
minus120582119896and 120585 = 120587 according to (40) imply that minus1 lt 119886
119896le 0
Substitution of 120585 = 120587 120582119896= minus119886119896 and (164) into (38) also leads
to (155) which proves that (155) also holds for minus1 lt 119886119896le 0
Consequently (155) holds for minus1 lt 119886119896le 1(1 minus 119896
2
)In what follows we first prove that (156)-(157) hold for
1(1 minus 1198962
) lt 119886119896lt 1 For this purpose let us start with non-
negative waveforms with two zeros described by (66) As wementioned before nonnegative cosine waveforms with twozeros have zeros at 120591
0and minus120591
0 such that 120591
0= 0 and 120591
0= 120587
Relations 1198861
lt 0 and 1198871
= 0 according to (84) implycos(1205910minus 120585119896) = 1 and therefore
120585
119896= 1205910 (165)
From 120585119896 = 1205910and 0 lt |120585| lt 120587 it follows that 0 lt |120591
0| lt
120587119896 Insertion of 120585119896 = 1205910into (45) yields 119886
119896= 120582119896 Relations
119886119896= 120582119896and (82) imply that 1(1 minus 1198962) lt 119886
119896lt 1 Substitution
of 120582119896= 119886119896and 120585119896 = 120591
0into (66)ndash(68) leads to (156)ndash(158)
which proves that (156)ndash(158) hold for 1(1 minus 1198962
) lt 119886119896lt 1
and 0 lt |1205910| lt 120587119896
Finally substitution of 119886119896= 1(1 minus 119896
2
) and 1205910= 0 into
(161) leads to
119879119896(120591) =
[1 minus cos 120591](1 minus 1198962)
[119896 (119896 minus 1) minus 2
119896minus1
sum
119899=1
(119896 minus 119899) cos 119899120591]
(166)
Waveform (166) coincides with waveform (155) for 119886119896
=
1(1 minus 1198962
) which in turn proves that (156) holds for 119886119896=
1(1 minus 1198962
) and 1205910= 0 This completes the proof
62 Nonnegative CosineWaveforms with at Least One Zero for119896 = 3 In this subsection we consider nonnegative cosinewaveforms with at least one zero for 119896 = 3 (for case 119896 = 2
see [12])Cosine waveform with fundamental and third harmonic
reads
1198793(120591) = 1 + 119886
1cos 120591 + 119886
3cos 3120591 (167)
For 1198861le 0 and minus1 le 119886
3le 18 according to (155) non-
negative cosine waveform of type (167) with at least one zerocan be expressed as
1198793(120591) = (1 minus cos 120591) [1 minus 2119886
3(1 + 2 cos 120591 + cos 2120591)] (168)
From 1198793(120591 + 120587) = 2 minus 119879
3(120591) it immediately follows that for
1198861ge 0 and minus18 le 119886
3le 1 119879
3(120591) can be expressed as
1198793(120591) = (1 + cos 120591) [1 + 2119886
3(1 minus 2 cos 120591 + cos 2120591)] (169)
For 1198861le 0 and 18 le 119886
3le 1 from (158) it follows that 119886
3=
[8cos31205910]minus1 This relation along with (160) and (157) further
implies that 1198793(120591) can be expressed as
1198793(120591) =
[cos 1205910minus cos 120591]2 [2 cos 120591
0+ cos 120591]
2cos31205910
(170)
providing that |1205910| le 1205873 From 119879
3(120591 + 120587) = 2 minus 119879
3(120591) it
follows that (170) also holds for 1198861ge 0 and minus1 le 119886
3le minus18
providing that 1205910isin [21205873 41205873]
Maximally flat nonnegative cosinewaveformof type (167)with 119886
1lt 0 (minimum at 120591
0= 0) reads 119879
3(120591) = [1 minus
cos 120591]2[1 + (12) cos 120591] Dually maximally flat nonnegativecosine waveform with 119886
1gt 0 (minimum at 120591
0= 120587) reads
1198793(120591) = [1 + cos 120591]2[1 minus (12) cos 120591]In what follows we provide relations between coefficients
1198861and 1198863of nonnegative cosine waveforms of type (167) with
at least one zeroFor 1198861le 0 conversion of (168) into an additive form
immediately leads to the following relation
1198861= minus1 minus 119886
3for minus 1 le 119886
3le1
8 (171)
Conversion of (170) into an additive form leads to 1198861
=
minus31198863(2 cos 2120591
0+ 1) which can be also expressed as 119886
1=
minus31198863(4cos2120591
0minus 1) For 119886
1le 0 relations |120591
0| le 1205873 119886
1=
minus31198863(4cos2120591
0minus 1) and 119886
3= [8cos3120591
0]minus1 lead to
1198861= minus3 [ 3radic119886
3minus 1198863] for 1
8le 1198863le 1 (172)
Similarly for 1198861ge 0 conversion of (169) into an additive form
leads to the following relation
1198861= 1 minus 119886
3for minus
1
8le 1198863le 1 (173)
Mathematical Problems in Engineering 21
1
15
15
05
0
minus05
minus15minus15
minus1
minus1 minus05 0 05 1
Coefficient a1
Coe
ffici
enta
3 a1= minus
3(3radica3minus a3)
a1 =
minus1 minus a3
a1 =
1 minusa3
a1 =3(3radic|a3
| +a3)
Figure 14 Parameter space of cosine waveforms for 119896 = 3
For waveform of type (170) with 1198861ge 0 relations 120591
0isin [21205873
41205873] 1198861= minus3119886
3(4cos2120591
0minus 1) and 119886
3= [8cos3120591
0]minus1 lead to
1198861= 3 [
3radic10038161003816100381610038161198863
1003816100381610038161003816 + 1198863] for minus 1 le 1198863le minus
1
8 (174)
Every cosine waveform of type (167) corresponds to apair of real numbers (119886
1 1198863) and vice versa Points (119886
1 1198863)
in grey area in Figure 14 correspond to nonnegative cosinewaveforms for 119896 = 3 The points at the boundary of grey areacorrespond to nonnegative cosinewaveformswith at least onezero A number of shapes of nonnegative cosine waveformswith 119896 = 3 and at least one zero plotted on interval [minus120587 120587]are also presented in Figure 14 The boundary of grey area inFigure 14 consists of four line segments described by relations(171)ndash(174) The common point of line segments (172) and(173) is cusp point with coordinates 119886
1= 0 and 119886
3= 1
Another cusp point with coordinates 1198861= 0 and 119886
3= minus1
is the common point of line segments (171) and (174) Thecommon point of line segments (171)-(172) has coordinates(minus98 18) and common point of line segments (173)-(174)has coordinates (98 minus18) These points are representedby white circle dots and they correspond to maximallyflat cosine waveforms (eg see [21]) White triangle dotswith coordinates (2radic3 minusradic39) and (minus2radic3radic39) refer tothe nonnegative cosine waveforms with maximum value ofamplitude of fundamental harmonic
7 Four Case Studies of Usage of NonnegativeWaveforms in PA Efficiency Analysis
In this section we provide four case studies of usage ofdescription of nonnegative waveforms with fundamental and119896th harmonic in PA efficiency analysis In first two casestudies to be presented in Section 71 voltage is nonnegativewaveform with fundamental and second harmonic with atleast one zero In remaining two case studies to be consideredin Section 72 voltage waveform contains fundamental andthird harmonic
i
v Load
Vdc
Idc
Lch
vL
iL
Cb
+ +in
Figure 15 Generic PA circuit diagram
Let us consider generic PA circuit diagram as shown inFigure 15 We assume here that voltage and current wave-forms at the transistor output are
V (120579) = 1 + 1198861V cos 120579 + 1198871V sin 120579 + 119886119896V cos 119896120579 + 119887119896V sin 119896120579
119894 (120579) = 1 + 1198861119894cos 120579 +
infin
sum
119899=2
119886119899119894cos 119899120579
(175)
where 120579 stands for 120596119905 Both waveforms are normalized in thesense that dc components of voltage and current are 119881dc =
1 and 119868dc = 1 respectively Under assumption that blockingcapacitor 119862
119887behaves as short-circuit at the fundamental and
higher harmonics current and voltage waveforms at the loadare
V119871(120579) = 119886
1V cos 120579 + 1198871V sin 120579 + 119886119896V cos 119896120579 + 119887119896V sin 119896120579
119894119871(120579) = minus119886
1119894cos 120579 minus
infin
sum
119899=2
119886119899119894cos 119899120579
(176)
In terms of coefficients of voltage and current waveforms theload impedance at fundamental harmonic is 119911
1= minus(119886
1V minus
1198951198871V)1198861119894 whereas load impedance at 119896th harmonic is 119911
119896=
minus(119886119896Vminus119895119887119896V)119886119896119894 All other harmonics are short-circuited (119911
119899=
0 for 119899 = 1 and 119899 = 119896) Time average output power of PA (egsee [10]) with waveform pair (175) at fundamental frequencycan be expressed as
1198751= minus
11988611198941198861V
2 (177)
For normalized waveforms (175) with 119881dc = 1 and 119868dc = 1dc power is 119875dc = 1 Consequently PA efficiency 120578 = 119875
1119875dc
(eg see [10 26]) is equal to
120578 = minus11988611198941198861V
2 (178)
Thus time average output power 1198751of PA with pair of nor-
malized waveform (175) is equal to efficiency (178)Power utilization factor (PUF) is defined [26] as ldquothe
ratio of power delivered in a given situation to the power
22 Mathematical Problems in Engineering
delivered by the same device with the same supply voltagein Class A moderdquo Since the output power in class-A modeis 1198751class-A = max[V(120579)] sdotmax[119894(120579)]8 (eg see [9]) it follows
that power utilization factor PUF = 11987511198751class-A for PA with
pair of normalized waveforms (175) can be expressed as
PUF =8120578
max [V (120579)] sdotmax [119894 (120579)] (179)
71 NonnegativeWaveforms for 119896 = 2 in PAEfficiencyAnalysisIn this subsection we provide two case studies of usage ofdescription of nonnegative waveforms with fundamental andsecond harmonic (119896 = 2) in PA efficiency analysis For moreexamples of usage of descriptions of nonnegative waveformswith fundamental and second harmonic in PA efficiencyanalysis see [12]
Case Study 71 In this case study we consider efficiency of PAfor given second harmonic impedance providing that voltageis nonnegative waveform with fundamental and second har-monic and current is ldquohalf-sinerdquo waveform frequently used inefficiency analysis of classical PA operation (eg see [10])
Standard model of current waveform for classical PAoperation has the form (eg see [10 26])
119894119863(120579) =
119868119863[cos 120579 minus cos(120572
2)] |120579| le
120572
2
0120572
2le |120579| le 120587
(180)
where 120572 is conduction angle and 119868119863
gt 0 Since 119894119863(120579) is
even function it immediately follows that its Fourier seriescontains only dc component and cosine terms
119894119863(120579) = 119868dc +
infin
sum
119899=1
119868119899cos 119899120579 (181)
The dc component of the waveform (180) is
119868dc =119868119863120572
2120587[sinc(120572
2) minus cos(120572
2)] (182)
where sinc119909 = (sin119909)119909 The coefficient of the fundamentalharmonic component reads
1198681=119868119863120572
2120587(1 minus sinc120572) (183)
and the coefficient of 119899th harmonic component can bewrittenin the form
119868119899=119868119863
119899120587[sin ((119899 minus 1) 1205722)
(119899 minus 1)minussin ((119899 + 1) 1205722)
(119899 + 1)] 119899 ge 2
(184)
For ldquohalf-sinerdquo current waveform conduction angle is equalto 120587 (class-B conduction angle) According to (182) thisfurther implies that 119868dc = 119868
119863120587 To obtain normalized form of
waveform (180) we set 119868dc = 1which implies that 119868119863= 120587 Fur-
thermore substitution of 120572 = 120587 and 119868119863= 120587 in (180) leads to
119894 (120579) =
120587 cos 120579 |120579| lt120587
2
0120587
2lt |120579| le 120587
(185)
Similarly substitution of 119868119863= 120587 and 120572 = 120587 into (183) and
(184) leads to the coefficients of waveform (185) Coefficientsof fundamental and second harmonic respectively are
1198861119894=120587
2 119886
2119894=2
3 (186)
On the other hand voltage waveform of type (35) for 119896 =2 reads
V (120579) = 1 + 1198861V cos 120579 + 1198871V sin 120579 + 1198862V cos 2120579 + 1198872V sin 2120579
(187)
This waveform contains only fundamental and second har-monic and therefore all harmonics of order higher thantwo are short-circuited (119911
119899= 0 for 119899 gt 2) For current
voltage pair (185) and (187) load impedance at fundamentalharmonic is 119911
1= minus(119886
1V minus 1198951198871V)1198861119894 whereas load impedance
at second harmonic is 1199112= minus(119886
2Vminus1198951198872V)1198862119894 According to ourassumption the load is passive and therefore Re119911
1 gt 0 and
Re1199112 ge 0 which further imply 119886
11198941198861V lt 0 and 119886
21198941198862V le 0
respectivelyIt is easy to see that problem of findingmaximal efficiency
of PA with current-voltage pair (185) and (187) for prescribedsecond harmonic impedance can be reduced to the problemof finding voltage waveform of type (187) with maximal coef-ficient |119886
1V| for prescribed coefficients of second harmonic(see Section 5)
The following algorithm (analogous to Algorithm 22presented in [12]) provides the procedure for calculation ofmaximal efficiency with current-voltage pair (185) and (187)for prescribed second harmonic impedance The definitionof function atan 2(119910 119909) which appears in the step (iii) of thefollowing algorithm is given by (105)
Algorithm 32 (i) Choose 1199112= 1199032+1198951199092such that |119911
2| le 1|119886
2119894|
(ii) calculate 1198862V minus 1198951198872V = minus119911
21198862119894and 120582
2V = radic11988622V + 1198872
2V(iii) if 2120582
2V le 1 minus 1198862V then calculate 119886
1V = minus1 minus 1198862V and
1198871V = minus2119887
2V else calculate 1205821V = radic81205822V(1 minus 1205822V) 1205790V minus 120585V2 =
(12)atan2(1198872V 1198862V) 1198861V = minus120582
1V cos(1205790V minus 120585V2) and 1198871V =
minus1205821V sin(1205790V minus 120585V2)(iv) calculate efficiency 120578 = minus119886
11198941198861V2
(v) calculate 1199111= minus(119886
1V minus 1198951198871V)1198861119894 and 1199112119899 = 1199112Re119911
1
In this case study coefficients of fundamental and secondharmonic of current waveform are given by (186) Maximalefficiency of PA associated with the waveform pair (185)and (187) as a function of normalized second harmonicimpedance 119911
2119899= 1199112Re119911
1 is presented in Figure 16(a)
As can be seen from Figure 16(a) efficiency of 078 isachieved at the edge of Smith chart where second harmonicimpedance has small resistive part Corresponding PUFcalculated according to (179) is presented in Figure 16(b)Peak efficiency 120578 = 1205874 = 07854 and peak value of PUF = 1
are attained when second harmonic is short-circuited (whichcorresponds to ideal class-B operation [10 26])
For example for second harmonic impedance 1199112= 01 minus
11989505 and currentwaveform (185) fromAlgorithm32 it followsthat 2120582
2V le 1 minus 1198862V Furthermore according to step (iii)
of above algorithm maximal efficiency of PA is attained
Mathematical Problems in Engineering 23
+j2
+j5
+j1
+j2
+j5
minusj2
minusj5
minusj1
minusj2
minusj5
infin0
078
02
05
1 2 5
075 07 065
06 05
075
07
065
06
05
120578 lt 05
(a)
099
095
091
083 075
067
0 59 051
051
099
095
091
083
075 0
67 059
+j5
+j1
+j2
+j5
minusj5
minusj1
minusj2
minusj5
infin
+j2
minusj2
0 02
05
1 2 5
120578 lt 05
(b)
Figure 16 (a) Contours of maximal efficiency of PA and (b) contours of corresponding PUF associated with the waveform pair (185) and(187) as functions of normalized second harmonic impedance 119911
2119899= 1199112Re119911
1
2
3
4
1
0
2 3 410
Wav
efor
ms
Angle 120579120587
i(120579)
(120579)
Figure 17 Waveform pair (185) and (187) that provides maximalefficiency for 119911
2= 01 minus 11989505
with voltage waveform of type (187) with coefficients 1198862V =
minus00667 1198872V = minus03333 119886
1V = minus09333 and 1198871V = 06667 (see
Figure 17) Corresponding efficiency PUF and normalizedsecond harmonic impedance are 120578 = 07330 PUF = 07572and 1199112119899= 01683 minus 11989508415 respectively
On the other hand for second harmonic impedance 1199112=
01 minus 11989508 and current waveform (185) from Algorithm 32 itfollows that 2120582
2V gt 1 minus 1198862V Then according to step (iii) of
above algorithm maximal efficiency is attained with voltagewaveform of type (187) with coefficients 119886
2V = minus00667 1198872V =
minus05333 1198861V = minus09333 and 119887
1V = 10572 (see Figure 18)Efficiency PUF andnormalized secondharmonic impedanceare 120578 = 07330 PUF = 06332 and 119911
2119899= 01683 minus 11989513465
respectively
2
3
4
1
0
2 3 410
Wav
efor
ms
Angle 120579120587
i(120579)
(120579)
Figure 18 Waveform pair (185) and (187) that provides maximalefficiency for 119911
2= 01 minus 11989508
Case Study 72 As another case study let us consider the effi-ciency of PA providing that current waveform is nonnegativecosine waveform up to third harmonic with maximum valueof amplitude of fundamental harmonic [22] (see also [8])
119894 (120579) = 1 +1 + radic5
2cos 120579 + 2radic5
5cos 2120579 + 5 minus radic5
10cos 3120579
(188)
and voltage waveform is nonnegative waveform of type(187) Load impedances at fundamental second and thirdharmonic are 119911
1= minus(119886
1Vminus1198951198871V)1198861119894 1199112 = minus(1198862Vminus1198951198872V)1198862119894 and
1199113= 0 respectively According to our assumption the load
is passive and therefore Re1199111 gt 0 and Re119911
2 ge 0 which
further imply 11988611198941198861V lt 0 and 119886
21198941198862V le 0 respectively
Because current waveform (188) contains only cosineterms and voltage waveform is the same as in previous case
24 Mathematical Problems in Engineering
08
075
07
06
06
065
065
05
05
075 0
7
120578 lt 05
+j2
+j5
+j1
+j2
+j5
minusj2
minusj5
minusj1
minusj2
minusj5
infin0
02
05
1 2 5
(a)
085
08 0
75 07 0
6
05 04
08
075
0405
060
7
120578 lt 05
+j2
+j5
+j1
+j2
+j5
minusj2
minusj5
minusj1
minusj2
minusj5
infin0 02
05
1 2 5
(b)
Figure 19 (a) Contours of maximal efficiency of PA and (b) contours of corresponding PUF associated with the waveform pair (187)-(188)as functions of normalized second harmonic impedance 119911
2119899= 1199112Re119911
1
study the procedure for calculation of maximal efficiency ofPA with waveform pair (187)-(188) is the same as presentedin Algorithm 32 In this case study the coefficients of funda-mental and second harmonic of current waveform are 119886
1119894=
(1 + radic5)2 and 1198862119894= 2radic55 respectively
Maximal efficiency of PA associated with the waveformpair (187)-(188) as a function of normalized secondharmonicimpedance 119911
2119899= 1199112Re119911
1 is presented in Figure 19(a)
Efficiency of 08 is achieved at the edge of Smith chart wheresecond harmonic impedance has small resistive part Thetheoretical upper bound 120578 = (1 + radic5)4 asymp 08090 isattained when second harmonic is short-circuitedWhen thisupper bound is reached both second and third harmonic areshort-circuited which implies that we are dealing with finiteharmonic class-C [6 8] or dually when current and voltageinterchange their roles with finite harmonic inverse class-C[6 9] Corresponding PUF calculated according to (179) ispresented in Figure 19(b) Peak value of PUF asymp 08541 isattained when second harmonic is short-circuited
For example for second harmonic impedance 1199112
=
007 minus 11989504 and current waveform (188) from Algorithm 32 itfollows that 2120582
2V le 1 minus 1198862V Furthermore according to step
(iii) of Algorithm 32 maximal efficiency of PA is attainedwith voltage waveform of type (187) with coefficients 119886
2V =
minus00626 1198872V = minus03578 119886
1V = minus09374 and 1198871V = 07155 (see
Figure 20) Corresponding efficiency PUF and normalizedsecond harmonic impedance are 120578 = 07584 PUF = 06337and 1199112119899= 01208 minus 11989506904 respectively
On the other hand for 1199112= 005 minus 11989507 and current
waveform (187) it follows that 21205822V gt 1minus119886
2VThen accordingto step (iii) of Algorithm 32 themaximal efficiency is attainedwith voltage waveform of type (187) with coefficients 119886
2V =
minus00447 1198872V = minus06261 119886
1V = minus09318 and 1198871V = 10007 (see
Figure 21) Efficiency PUF and normalized second harmonic
2
3
4
1
0
2 3 410
Wav
efor
ms
Angle 120579120587
i(120579)
(120579)
Figure 20 Waveform pair (187)-(188) that provides maximal effi-ciency for 119911
2= 007 minus 11989504
impedance are 120578 = 07538 PUF = 05314 and 1199112119899= 00868minus
11989512156 respectively
72 Nonnegative Waveforms for 119896 = 3 in PA EfficiencyAnalysis In this subsection we provide another two casestudies of usage of description of nonnegative waveforms inPA efficiency analysis this time with fundamental and thirdharmonic (119896 = 3)
Case Study 73 Let us consider current-voltage pair such thatvoltage is nonnegative waveform with fundamental and thirdharmonic
V (120579) = 1 + 1198861V cos 120579 + 1198871V sin 120579 + 1198863V cos 3120579 + 1198873V sin 3120579
(189)
Mathematical Problems in Engineering 25
2
3
4
1
0
2 3 410
Wav
efor
ms
Angle 120579120587
i(120579)
(120579)
Figure 21 Waveform pair (187)-(188) that provides maximal effi-ciency for 119911
2= 005 minus 11989507
and current is nonnegative cosine waveform given by (188)Load impedances at fundamental second and third har-monic are 119911
1= minus(119886
1V minus 1198951198871V)1198861119894 1199112 = 0 and 119911
3= minus(119886
3V minus
1198951198873V)1198863119894 respectively According to our assumption the load
is passive and therefore Re1199111 gt 0 and Re119911
3 ge 0 which
further imply 11988611198941198861V lt 0 and 119886
31198941198863V le 0
In this subsection we consider the problem of findingmaximal efficiency of PA with waveform pair (188)-(189)for given third harmonic impedance As we mentionedearlier problem of finding maximal efficiency of PA withcurrent-voltage pair (188)-(189) for prescribed third har-monic impedance can be reduced to the problem of findingvoltage waveform of type (189) withmaximal coefficient |119886
1V|for prescribed coefficients of third harmonic (see Section 52)
The following algorithm provides the procedure forcalculation of maximal efficiency with current-voltage pair(188)-(189) The definition of function atan 2(119910 119909) whichappears in step (iii) of the following algorithm is given by(105)
Algorithm 33 (i) Choose 1199113= 1199033+1198951199093such that |119911
3| le 1|119886
3119894|
(ii) calculate 1198863V minus 1198951198873V = minus119911
31198863119894and 120582
3V = radic11988623V + 1198872
3V(iii) if 271205822
3V le (1 minus 21198863V)3 then calculate 119886
1V = minus1 minus 1198863V
and 1198871V = minus3119887
3V else calculate 1205821V = 3(3radic1205823V minus 120582
3V) 1205790V minus120585V3 = (13)atan 2(119887
3V 1198863V) 1198861V = minus1205821V cos(1205790V minus 120585V3) and
1198871V = minus120582
1V sin(1205790V minus 120585V3)(iv) calculate efficiency 120578 = minus119886
11198941198861V2
(v) calculate 1199111= minus(119886
1V minus 1198951198871V)1198861119894 and 1199113119899 = 1199113Re119911
1
In this case study coefficients of fundamental and thirdharmonic of current waveform are 119886
1119894= (1 + radic5)2 and
1198863119894= (5 minus radic5)10 respectively For the waveform pair (188)-
(189) maximal efficiency of PA as a function of normalizedthird harmonic impedance 119911
3119899= 1199113Re119911
1 is presented in
Figure 22 Efficiency of 08 is reached when third harmonicimpedance has small resistive part Peak efficiency 120578 = (1 +
radic5)4 asymp 08090 is achieved when third harmonic is short-circuited
+j2
+j5
+j1
+j2
+j5
minusj2
minusj5
minusj1
minusj2
minusj5
infin0
08
08
075
075
07
07
06
05
02
05
1 2
Figure 22 Contours of maximal efficiency of PA associated withthe waveform pair (188)-(189) as a function of normalized thirdharmonic impedance 119911
3119899= 1199113Re119911
1
For the present case study in what follows we show thatpower utilization factor is proportional to efficiency For volt-age waveform of type (189) it is easy to see that V(120579 + 120587) =
2minusV(120579) holdsThis relation along with the fact that waveformV(120579) that provides maximal efficiency has at least one zeroimplies that max[V(120579)] = 2 On the other hand currentwaveform (188) is cosine waveform with positive coefficientsand therefore max[119894(120579)] = 119894(0) = 2 + 4radic5 Consequentlyaccording to (179) the following relation holds
PUFCase study 73 = 2 (5 minus 2radic5) 120578 = 10557120578 (190)
Clearly the ratio PUF120578 is constant and therefore in this casestudy PUF can be easily calculated from the correspondingefficiency Accordingly peak efficiency and peak value ofPUFCase study 73 = 3radic52 minus 52 = 08541 are attained forthe same voltage waveform (when third harmonic is short-circuited)
In the first example current waveform (188) and 1199113=
02 minus 11989505 imply that 2712058223V le (1 minus 2119886
3V)3 Then according
to Algorithm 33 the voltage waveform of type (189) thatprovides maximal efficiency has the following coefficients1198863V = minus00553 119887
3V = minus01382 1198861V = minus09447 and 119887
1V =
04146 (see Figure 23) Efficiency PUF and normalized thirdharmonic impedance are 120578 = 07643 PUF = 08069 and1199113119899= 03425 minus 11989508564 respectivelyIn the second example current waveform (188) and 119911
3=
01 minus 11989511 imply that 2712058223V gt (1 minus 2119886
3V)3 Then according
to Algorithm 33 the voltage waveform of type (189) thatprovides maximal efficiency has the following coefficients1198863V = minus00276 119887
3V = minus03040 1198861V = minus09391 and 119887
1V =
05807 (see Figure 24) Efficiency PUF and normalized thirdharmonic impedance are 120578 = 07598 PUF = 08021 and1199113119899= 01723 minus 11989518952 respectively
26 Mathematical Problems in Engineering
2
3
4
1
0
2 3 410
Wav
efor
ms
Angle 120579120587
i(120579)
(120579)
Figure 23 Waveform pair (188)-(189) that provides maximal effi-ciency for 119911
3= 02 minus 11989505
2
3
4
1
0
2 3 410
Wav
efor
ms
Angle 120579120587
i(120579)
(120579)
Figure 24 Waveform pair (188)-(189) that provides maximal effi-ciency for 119911
3= 01 minus 11989511
Case Study 74 In this case study let us consider current-voltage pair where current is normalized waveform of type(180) with conduction angle 120572 = 115120587 (207∘) and voltageis nonnegative waveform of type (189) Substitution of 120572 =
115120587 and 119868dc = 1 into (182) leads to 119868119863
= 22535Furthermore substitution of 120572 = 115120587 and 119868
119863= 22535 into
(180) leads to
119894 (120579)
=
22535 [cos 120579 minus cos(1151205872
)] |120579| le115120587
2
0115120587
2le |120579| le 120587
(191)
Similarly substitution of 120572 = 115120587 and 119868119863
= 22535 into(183) and (184) for 119899 = 3 yields coefficients of fundamentaland third harmonic of waveform (191)
1198861119894= 14586 119886
3119894= minus01026 (192)
Because current waveform (191) contains only cosineterms and voltage waveform is the same as in previous case
+j2
+j5
+j1
+j2
+j5
minusj2
minusj5
minusj1
minusj2
minusj5
infin0 02
05
1 2
074
076
08
084
082
078
076
074
Figure 25 Contours of maximal efficiency of PA associated withthe waveform pair (189) and (191) as a function of normalized thirdharmonic impedance 119911
3119899= 1199113Re119911
1
study the procedure for calculation of maximal efficiencyof PA with waveform pair (189)ndash(191) is the same as thatpresented in Algorithm 33 In this case study the coefficientsof fundamental and third harmonic of current waveform aregiven by (192)
For the waveform pair (189) and (191) maximal efficiencyof PA as a function of normalized third harmonic impedance1199113119899= 1199113Re119911
1 is presented in Figure 25 Efficiency of 084
is obtained in vicinity of 1199113119899= 23685 (corresponding to 119911
3=
18750) Peak efficiency 120578 asymp 08421 is achieved for voltagewaveform of type (189) with coefficients 119886
1V = minus2radic3 1198863V =
radic39 and 1198871V = 1198873V = 0
In the course of finding power utilization factor noticethat currentwaveformof type (191) attains itsmaximumvaluefor 120579 = 0 Insertion of max[119894(120579)] = 119894(0) = 278 andmax[V(120579)] = 2 for voltage waveform of type (189) into (179)leads to
PUFCase study 74 = 1439120578 (193)
Again the ratio PUF120578 is constant and PUF can be easilycalculated from the corresponding efficiency Accordinglypeak value of PUFCase study 74 asymp 12118 and peak efficiencyare attained for the same voltage waveform
In the first example current waveform (191) and 1199113=
1 minus 11989502 imply that 2712058223V le (1 minus 2119886
3V)3 Then according to
Algorithm 33 voltage waveform of type (189) which providesmaximal efficiency has coefficients 119886
3V = 01026 1198873V =
00205 1198861V = minus11026 and 119887
1V = minus00616 (see Figure 26)Efficiency PUF and normalized third harmonic impedanceare 120578 = 08042 PUF = 11572 and 119911
3119899= 13228 minus 11989502646
respectivelyIn second example current waveform (191) and 119911
3=
15 minus 11989512 imply that 2712058223V gt (1 minus 2119886
3V)3 Then according to
Algorithm 33 voltage waveform of type (189) which providesmaximal efficiency has coefficients 119886
3V = 01540 1198873V =
Mathematical Problems in Engineering 27
2
3
4
1
0
2 3 410
Wav
efor
ms
Angle 120579120587
i(120579)
(120579)
Figure 26 Waveform pair (189) and (191) that provides maximalefficiency for 119911
3= 1 minus 11989502
2
3
4
1
0
2 3 410
Wav
efor
ms
Angle 120579120587
i(120579)
(120579)
Figure 27 Waveform pair (189) and (191) that provides maximalefficiency for 119911
3= 15 minus 11989512
01232 1198861V = minus11255 and 119887
1V = minus02575 (see Figure 27)Efficiency PUF and normalized third harmonic impedanceare 120578 = 08208 PUF = 11812 and 119911
3119899= 19439 minus 11989515552
respectively
8 Conclusion
In this paper we consider a problem of finding general de-scriptions of various classes of nonnegative waveforms withfundamental and 119896th harmonic These classes include non-negative waveforms with at least one zero nonnegative wave-forms with maximal amplitude of fundamental harmonic forprescribed amplitude of 119896th harmonic nonnegative wave-forms withmaximal coefficient of cosine part of fundamentalharmonic for prescribed coefficients of 119896th harmonic andnonnegative cosine waveforms with at least one zero Mainresults are stated in six propositions (Propositions 1 6 918 22 and 26) four corollaries (Corollaries 2ndash5) twentyremarks and three algorithms Four case studies of usageof closed form descriptions of nonnegative waveforms in PAefficiency analysis are considered in detail in Section 7
Appendices
Here we provide a list of finite sums of trigonometric func-tions used in this paper (Appendix A) and brief account ofthe Chebyshev polynomials (Appendix B)
A List of Some Finite Sums ofTrigonometric Functions
Dirichlet kernel (eg see [27]) is as follows
119863119896minus1
(120591) = 1 + 2
119896minus1
sum
119899=1
cos 119899120591 = sin ((2119896 minus 1) 1205912)sin (1205912)
(A1)
Fejer kernel (eg see [27]) can be expressed in the fol-lowing equivalent forms
119865119896minus1
(120591) =1
119896
119896minus1
sum
119899=0
119863119899(120591) = 1 +
2
119896
119896minus1
sum
119899=1
(119896 minus 119899) cos 119899120591
=(1 minus cos 119896120591)119896 (1 minus cos 120591)
(A2)
Lagrangersquos trigonometric identity (eg see [28]) is as fol-lows
1198781(120591) =
119896minus1
sum
119899=1
sin 119899120591 = sin (1198961205912) sin ((119896 minus 1) 1205912)sin (1205912)
(A3)
In what follows we show that the following three trigono-metric identities also hold
2
119896minus1
sum
119899=1
(119896 minus 119899) sin 119899120591 = 119896 sin 120591 minus sin 1198961205911 minus cos 120591
(A4)
119896minus1
sum
119899=1
cos (119896 minus 2119899) 120591 = sin (119896 minus 1) 120591sin 120591
(A5)
119896minus1
sum
119899=1
119899 (119896 minus 119899) cos (119896 minus 2119899) 120591
=sin (119896120591) cos 120591 minus 119896 cos (119896120591) sin 120591
2sin3120591
(A6)
Denote 1198782(120591) = 2sum
119896minus1
119899=1(119896 minus 119899) sin 119899120591 119878
3(120591) = sum
119896minus1
119899=1cos(119896 minus
2119899)120591 and 1198784(120591) = sum
119896minus1
119899=1119899(119896 minus 119899) cos(119896 minus 2119899)120591
Notice that 1198782(120591) = 2119896119878
1(120591) + 119889119863
119896minus1(120591)119889120591 which
immediately leads to (A4)
28 Mathematical Problems in Engineering
Identity (A5) can be obtained as follows
sin (119896 minus 1) 120591sin 120591
=119890119895(119896minus1)120591
minus 119890minus119895(119896minus1)120591
119890119895120591 minus 119890minus119895120591
= 119890119895119896120591
119890minus2119895120591
minus 119890minus2119895(119896minus1)120591
1 minus 119890minus2119895120591
= 119890119895119896120591
119896minus1
sum
119899=1
119890minus2119895119899120591
=
119896minus1
sum
119899=1
119890119895(119896minus2119899)120591
=
119896minus1
sum
119899=1
cos (119896 minus 2119899) 120591
(A7)
From 4119899(119896 minus 119899) = 1198962
minus (119896 minus 2119899)2 it follows that 4119878
4(120591) =
1198962
1198783(120591) + 119889
2
1198783(120591)119889120591
2 which leads to (A6)
B The Chebyshev Polynomials
The Chebyshev polynomials of the first kind 119881119899(119909) can be
defined by the following relation (eg see [29])
119881119899(119909) = cos 119899120591 when 119909 = cos 120591 (B1)
The Chebyshev polynomials of the second kind 119880119899(119909) can be
defined by the following relation (eg see [29])
119880119899(119909) =
sin (119899 + 1) 120591sin 120591
when 119909 = cos 120591 (B2)
The Chebyshev polynomials satisfy the following recur-rence relations (eg see [29])
1198810(119909) = 1 119881
1(119909) = 119909
119881119899+1
(119909) = 2119909119881119899(119909) minus 119881
119899minus1(119909)
1198800(119909) = 1 119880
1(119909) = 2119909
119880119899+1
(119909) = 2119909119880119899(119909) minus 119880
119899minus1(119909)
(B3)
The first few Chebyshev polynomials of the first and secondkind are 119881
2(119909) = 2119909
2
minus 1 1198813(119909) = 4119909
3
minus 3119909 1198814(119909) = 8119909
4
minus
81199092
+ 1 1198802(119909) = 4119909
2
minus 1 1198803(119909) = 8119909
3
minus 4119909 and 1198804(119909) =
161199094
minus 121199092
+ 1
Conflict of Interests
The authors declare that there is no conflict of interestsregarding the publication of this paper
Acknowledgment
This work is supported by the SerbianMinistry of EducationScience and Technology Development as a part of ProjectTP32016
References
[1] V I Arnolrsquod V S Afrajmovich Y S Ilrsquoyashenko and L PShilrsquonikov Dynamical Systems V Bifurcation Theory and Catas-trophe Theory Springer Berlin Germany 1994
[2] E Polak ldquoOn the mathematical foundations of nondifferen-tiable optimization in engineering designrdquo SIAM Review vol29 no 1 pp 21ndash89 1987
[3] N S Fuzik ldquoBiharmonic modes of a tuned RF power amplifierrdquoRadiotehnika vol 25 no 7 pp 62ndash71 1970 (Russian)
[4] P Colantonio F Giannini G Leuzzi and E Limiti ldquoClass Gapproach for low-voltage high-efficiency PA designrdquo Interna-tional Journal of RF and Microwave Computer-Aided Engineer-ing vol 10 no 6 pp 366ndash378 2000
[5] F H Raab ldquoMaximum efficiency and output of class-F poweramplifiersrdquo IEEE Transactions on Microwave Theory and Tech-niques vol 49 no 6 pp 1162ndash1166 2001
[6] FH Raab ldquoClass-E class-C and class-F power amplifiers basedupon a finite number of harmonicsrdquo IEEE Transactions onMicrowaveTheory and Techniques vol 49 no 8 pp 1462ndash14682001
[7] J D Rhodes ldquoOutput universality inmaximum efficiency linearpower amplifiersrdquo International Journal of Circuit Theory andApplications vol 31 no 4 pp 385ndash405 2003
[8] A Juhas and L A Novak ldquoComments on lsquoClass-E class-Cand classF power amplifier based upon a finite number ofharmonicsrsquordquo IEEE Transactions on Microwave Theory and Tech-niques vol 57 no 6 pp 1623ndash1625 2009
[9] M Roberg and Z Popovic ldquoAnalysis of high-efficiency poweramplifiers with arbitrary output harmonic terminationsrdquo IEEETransactions on Microwave Theory and Techniques vol 59 no8 pp 2037ndash2048 2011
[10] A Grebennikov N O Sokal and M J Franco Switchmode RFPower Amplifiers ElsevierAcademic Press San Diego CalifUSA 2nd edition 2012
[11] T Canning P J Tasker and S C Cripps ldquoContinuous modepower amplifier design using harmonic clipping contourstheory and practicerdquo IEEE Transactions on Microwave Theoryand Techniques vol 62 no 1 pp 100ndash110 2014
[12] A Juhas and L A Novak ldquoGeneral description of nonnegativewaveforms up to second harmonic for power amplifier mod-ellingrdquoMathematical Problems in Engineering vol 2014 ArticleID 709762 18 pages 2014
[13] V I Arnolrsquod V V Goryunov O V Lyashko and V A VasilrsquoevDynamical Systems VIIImdashSingularity Theory II ApplicationsSpringer Berlin Germany 1993
[14] D Siersma ldquoProperties of conflict sets in the planerdquo BanachCenter Publications Polish Academy of Sciences vol 50 no 1 pp267ndash276 1999 Proceedings of the Banach Center Symposiumon Geometry and Topology of Caustics (Caustics rsquo98) WarsawPoland
[15] M van Manen The geometry of conflict sets [Dissertation]Universiteit Utrecht UtrechtThe Netherlands 2003 httpigi-tur-archivelibraryuunldissertations2003-0912-123058c4pdf
[16] Y L Sachkov ldquoMaxwell strata and symmetries in the problemofoptimal rolling of a sphere over a planerdquo Sbornik Mathematicsvol 201 no 7-8 pp 1029ndash1051 2010
[17] I A Bogaevsky ldquoPerestroikas of shocks and singularities ofminimum functionsrdquoPhysicaDNonlinear Phenomena vol 173no 1-2 pp 1ndash28 2002
[18] Y L Sachkov ldquoMaxwell strata in the Euler elastic problemrdquoJournal of Dynamical andControl Systems vol 14 no 2 pp 169ndash234 2008
[19] M Siino and T Koike ldquoTopological classification of black holesgeneric Maxwell set and crease set of a horizonrdquo InternationalJournal ofModern Physics D Gravitation Astrophysics Cosmol-ogy vol 20 no 6 pp 1095ndash1122 2011
Mathematical Problems in Engineering 29
[20] F H Raab ldquoClass-F power amplifiers with maximally flat wave-formsrdquo IEEETransactions onMicrowaveTheory and Techniquesvol 45 no 11 pp 2007ndash2012 1997
[21] A Juhas and L A Novak ldquoMaximally flat waveforms with finitenumber of harmonics in class-F power amplifiersrdquo Mathemat-ical Problems in Engineering vol 2013 Article ID 169590 9pages 2013
[22] L Fejer ldquoUber trigonometrische polynomerdquo Journal fur dieReine und Angewandte Mathematik vol 1916 no 146 pp 53ndash82 1916 (German)
[23] S C Cripps ldquoBessel Waives [microwave bytes]rdquo IEEE Micro-wave Magazine vol 10 no 7 pp 30ndash36 117 2009
[24] L N Bryzgalova ldquoSingularities of the maximum of para-metrically dependent functionrdquo Functional Analysis and ItsApplications vol 11 no 1 pp 49ndash51 1977
[25] V I Arnold A A Davydov V A Vassiliev and V MZakalyukin Mathematical Models of Catastrophes Control ofCatastrophic Process Encyclopedia of Life Support Systems(EOLSS) EOLSS Publishers Oxford UK 2006
[26] S C Cripps RF Power Amplifiers for Wireless CommunicationsArtech House Norwood Mass USA 2nd edition 2006
[27] A Zygmund Trigonometric Series vol 1 Cambridge UniversityPress Cambridge UK 2nd edition 1959
[28] A Jeffrey and H DaiHandbook of Mathematical Formulas andIntegrals ElsevierAcademic Press San Diego Calif USA 4thedition 2008
[29] J C Mason and D C Handscomb Chebyshev PolynomialsChapman amp Hall CRC Press Boca Raton Fla USA 2003
Submit your manuscripts athttpwwwhindawicom
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical Problems in Engineering
Hindawi Publishing Corporationhttpwwwhindawicom
Differential EquationsInternational Journal of
Volume 2014
Applied MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Probability and StatisticsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical PhysicsAdvances in
Complex AnalysisJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
OptimizationJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
CombinatoricsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Operations ResearchAdvances in
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Function Spaces
Abstract and Applied AnalysisHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of Mathematics and Mathematical Sciences
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Algebra
Discrete Dynamics in Nature and Society
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Decision SciencesAdvances in
Discrete MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom
Volume 2014 Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Stochastic AnalysisInternational Journal of
4 Mathematical Problems in Engineering
001
02
0304
05
06 01205872
12058731205872
2120587
Parameter 120572
Parameter A
1205872
1205874
0
minus1205874
minus1205872Posit
ions
of g
loba
l min
k = 2
Figure 2 Positions of global minima of 119891(120591 119860 120572) for 119896 = 2
The problem of finding maximum value of fundamentalharmonic cosine part of nonnegative waveform of the form
119908119886(120591 120574119886 119887 119860 120572) = 1 minus 120574
119886(cos 120591 + 119887 sin 120591 + 119860 cos (119896120591 + 120572))
(9)
where 120574119886gt 0 is also related to the problem of finding max-
imumof theminimum functionOptimalwaveformof family(9) has two global minima (this claim will be justified inSection 5 Remark 25) and therefore corresponding 3-tupleof parameters belongs to the conflict set in parameter spaceof family (9)
Let us introduce an auxiliary waveform
119891119886(120591 119887 119860 120572) = minus cos 120591 minus 119887 sin 120591 minus 119860 cos (119896120591 + 120572) (10)
and corresponding minimum function 119865119886min(119887 119860 120572) =
min120591119891119886(120591 119887 119860 120572) Inequality 119908
119886(120591 120574119886 119887 119860 120572) ge 0 can be
rewritten as 1 + 120574119886119865119886min(119887 119860 120572) ge 0 and therefore the
highest value of 120574119886is attained for 120574
119886= minus1119865
119886min(119887 119860 120572) Itimmediately follows that nonnegative waveform of type (9)with 120574
119886= minus1119865
119886min(119887 119860 120572) has zero for 120591 satisfying 119891119886(120591
119887 119860 120572) = 119865119886min(119887 119860 120572)
22 Conflict Set Historically conflict set came into beingfrom the problems in which families of smooth functions(such as potentials distances and waveforms) with two com-peting minima occurThe situation when competing minimabecome equal refers to the presence of conflict set (Maxwellset Maxwell strata) in the associated parameter space
There are many facets of conflict set For example in theproblem involving distances between two sets of points theconflict set is the intersections between iso-distance lines[14] Conflict set also arises in the situation when two wavefronts coming from different objects meet [15 25] In thestudy of black holes conflict set is the line of crossover of thehorizon formed by the merger of two black holes [19] In theclassical Euler problem conflict set is a set of points wheredistinct extremal trajectories with the same value of the costfunctional meet one another [18]
Conflict set is very difficult to calculate both analyticallyand numerically (eg see [15]) because of apparent nondif-ferentiability in some directions In optimization of PA effi-ciency some authors already reported difficulties in findingoptimum via standard analytical tools [4 5]
In this section we consider conflict set in the context offamily of waveforms of type (2) for arbitrary 119896 ge 2 In thiscontext for prescribed integer 119896 ge 2 conflict set is said to bea set of all pairs (119860 120572) for which 119891(120591 119860 120572) possesses multipleglobal minima
Suppose that 1205911015840 and 12059110158401015840 are the positions of global min-
ima of 119891(120591 119860 120572) Then the conflict set is specified by the fol-lowing set of relations
119891 (1205911015840
119860 120572) = 119891 (12059110158401015840
119860 120572) (11)
1198911015840
(1205911015840
119860 120572) = 0 1198911015840
(12059110158401015840
119860 120572) = 0 (12)
11989110158401015840
(1205911015840
119860 120572) gt 0 11989110158401015840
(12059110158401015840
119860 120572) gt 0 (13)
(forall120591) 119891 (120591 119860 120572) ge 119891 (1205911015840
119860 120572) (14)
Relations (12) and (13) say that 119891(120591 119860 120572) has minima at 1205911015840and 12059110158401015840 while relations (11) and (14) imply that these minimaare equal and global
The following proposition describes the conflict set offamily of waveforms of type (2)
Proposition 1 Conflict set of family of waveforms of type (2)is the set of all pairs (119860 120572) such that 119860 gt 1119896
2 and 120572 = 120587
The proof of Proposition 1 which is provided at the endof this section also implies that the following four corollarieshold
Corollary 2 The conflict set has end point at (119860 120572) =
(11198962
120587) This end point corresponds to the maximally flatwaveform [21]
Corollary 3 Waveforms of type (2) with parameters thatbelong to conflict set have two global minima at plusmn120591
Δ where
0 lt 120591Δle 120587119896
Corollary 4 Every waveform with fundamental and 119896th har-monic has either one or two global minima
Corollary 5 Conflict set can be parameterised in terms of 120591Δ
as follows
120572 = 120587 119860 (120591Δ) =
sin 120591Δ
119896 sin 119896120591Δ
0 lt 120591Δle120587
119896 (15)
Notice that119860(120591Δ) ismonotonically increasing function on inter-
val 0 lt 120591Δle 120587119896
Proof of Proposition 1 Without loss of generality we canrestrict our consideration to the interval minus120587 lt 120591 le 120587 This isan immediate consequence of the fact that 119891(120591 119860 120572) is aperiodic function
Suppose that 1205911015840 and 12059110158401015840 where 1205911015840 lt 12059110158401015840 are points at which
119891(120591 119860 120572) has two equal global minima Then conflict set is
Mathematical Problems in Engineering 5
specified by relations (11)ndash(14) From (11)ndash(13) it follows thatrelations
119891 (1205911015840
119860 120572) minus 119891 (12059110158401015840
119860 120572) = 0
1198911015840
(1205911015840
119860 120572) + 1198911015840
(12059110158401015840
119860 120572) = 0
1198911015840
(1205911015840
119860 120572) minus 1198911015840
(12059110158401015840
119860 120572) = 0
11989110158401015840
(1205911015840
119860 120572) + 11989110158401015840
(12059110158401015840
119860 120572) gt 0
(16)
also hold Let
120591119904119903=(1205911015840
+ 12059110158401015840
)
2 120591
Δ=(12059110158401015840
minus 1205911015840
)
2
(17)
be a pair of points associated with (1205911015840 12059110158401015840) Clearly
minus120587 lt 120591119904119903lt 120587 (18)
0 lt 120591Δlt 120587 (19)
12059110158401015840
= 120591119904119903+ 120591Δ 120591
1015840
= 120591119904119903minus 120591Δ (20)
The first and second derivatives of 119891(120591 119860 120572) are equal to
1198911015840
(120591 119860 120572) = sin 120591 + 119896119860 sin (119896120591 + 120572)
11989110158401015840
(120591 119860 120572) = cos 120591 + 1198962119860 cos (119896120591 + 120572) (21)
By using (20)-(21) system (16) can be rewritten as
sin 120591119904119903sin 120591Δ+ 119860 sin (119896120591
119904119903+ 120572) sin 119896120591
Δ= 0 (22)
sin 120591119904119903cos 120591Δ+ 119896119860 sin (119896120591
119904119903+ 120572) cos 119896120591
Δ= 0 (23)
cos 120591119904119903sin 120591Δ+ 119896119860 cos (119896120591
119904119903+ 120572) sin 119896120591
Δ= 0 (24)
cos 120591119904119903cos 120591Δ+ 1198962
119860 cos (119896120591119904119903+ 120572) cos 119896120591
Δgt 0 (25)
From (19) it follows that sin 120591Δgt 0 Multiplying (24) and
(25) with minus cos 120591Δand sin 120591
Δgt 0 respectively and sum-
ming the resulting relations we obtain 119896119860 cos(119896120591119904119903
+
120572)[119896 sin 120591Δcos 119896120591
Δminus sin 119896120591
Δcos 120591Δ] gt 0 The latest relation
immediately implies that
119896 sin 120591Δcos 119896120591
Δminus sin 119896120591
Δcos 120591Δ
= 0 (26)
Equations (22) and (23) can be considered as a system oftwo linear equations in terms of sin 120591
119904119903and 119860 sin(119896120591
119904119903+ 120572)
According to (26) the determinant of this system is nonzeroand therefore it has only trivial solution
sin 120591119904119903= 0 sin (119896120591
119904119903+ 120572) = 0 (27)
According to (18) sin 120591119904119903= 0 implies
120591119904119903= 0 (28)
According to (20) 120591119904119903= 0 implies
12059110158401015840
= 120591Δ 120591
1015840
= minus120591Δ (29)
Furthermore 120591119904119903= 0 and sin(119896120591
119904119903+ 120572) = 0 imply that sin120572 =
0 From (29) it follows that 120591Δis position of global minimum
of 119891(120591 119860 120572) Clearly 119891(120591Δ 119860 120572) le 119891(0 119860 120572) which together
with sin120572 = 0 leads to
1 minus cos 120591Δ+ 119860 cos120572 (1 minus cos 119896120591
Δ) le 0 (30)
From 120591Δ
= 0 (see (19)) 119860 gt 0 and (30) it follows that cos 120572 lt
0 which together with sin120572 = 0 yields
120572 = 120587 (31)
Since 120591Δis position of global minimum it follows that
119891(120591Δ 119860 120587) le 119891(120587119896 119860 120587) Accordingly 119860(1 + cos 119896120591
Δ) le
cos 120591Δminus cos120587119896 which together with 119860 gt 0 implies that
cos 120591Δminus cos120587119896 ge 0 This relation along with (19) yields
0 lt 120591Δle120587
119896 (32)
Substitution of (31) and (28) in (24) leads to
119860 =sin 120591Δ
119896 sin 119896120591Δ
(33)
Notice that sin 119896120591Δ sin 120591
Δis monotonically decreasing func-
tion on interval (32)Therefore parameter119860 is monotonicallyincreasing function on the same interval with lim
120591Δrarr0+119860 =
11198962 Consequently 119860 gt 1119896
2 which completes theproof
23 Parameter Space In parameter space of family of wave-forms (2) there are two subsets playing important role in theclassification of the family instancesThese are conflict set andcatastrophe set
Catastrophe set is subset of parameter space of waveform119891(120591 119860 120572) It consists of those pairs (119860 120572) for which thecorresponding waveforms 119891(120591 119860 120572) have degenerate criticalpoints at which first and second derivatives are equal to zeroThus for finding catastrophe set we have to consider thefollowing system of equations
1198911015840
(120591119889 119860 120572) = 0
11989110158401015840
(120591119889 119860 120572) = 0
(34)
where 120591119889is a degenerate critical point of waveform119891(120591 119860 120572)
Conflict set in parameter space of waveform119891(120591 119860 120572) asshown in Proposition 1 is the ray described by 119860 gt 1119896
2 and120572 = 120587 It is intimately connected to catastrophe set
In what follows in this subsection we use polar coordinatesystem (119860 cos120572 119860 sin120572) instead of Cartesian coordinatesystem (119860 120572) Examples of catastrophe set and conflict setfor 119896 le 5 plotted in parameter space (119860 cos120572 119860 sin120572) arepresented in Figure 3 Solid line represents the catastropheset while dotted line describes conflict set The isolated pickpoints (usually called cusp) which appear in catastrophecurves correspond to maximally flat waveforms with max-imally flat minimum andor maximally flat maximumThereare two such picks in the catastrophe curves for 119896 = 2 and
6 Mathematical Problems in Engineering
00
00
k = 2 k = 3
k = 4 k = 5
Acos 120572
Acos 120572
Acos 120572
Acos 120572
A sin 120572
A sin 120572A sin 120572
A sin 120572
Figure 3 Catastrophe set (solid line) and corresponding conflict set(dotted line) for 119896 le 5 In each plot white triangle dot correspondsto optimal waveform and white circle dot corresponds to maximallyflat waveform
119896 = 4 and one in the catastrophe curves for 119896 = 3 and 119896 = 5Notice that the end point of conflict set is the cusp point
Catastrophe set divides the parameter space (119860 cos120572119860 sin120572) into disjoint subsets In the cases 119896 = 2 and 119896 =
3 catastrophe curve defines inner and outer part For 119896 gt
3 catastrophe curve makes partition of parameter space inseveral inner subsets and one outer subset (see Figure 3)
Notice also that multiplying 119891(120591 119860 120572) with a positiveconstant and adding in turn another constant which leads towaveform of type 119908(120591 120574 119860 120572) (see (1) and (2)) do not makeimpact on the character of catastrophe and conflict sets Thisis because in the course of finding catastrophe set first andsecond derivatives of 119891(120591 119860 120572) are set to zero Clearly (34) interms of 119891(120591 119860 120572) are equivalent to the analogous equationsin terms of119908(120591 120574 119860 120572) Analogously in the course of findingconflict set we consider only the positions of global minima(these positions forwaveforms119891(120591 119860 120572) and119908(120591 120574 119860 120572) arethe same)
3 Nonnegative Waveforms with atLeast One Zero
In what follows let us consider a waveform containing dccomponent fundamental and 119896th (119896 ge 2) harmonic of theform
119879119896(120591) = 1 + 119886
1cos 120591 + 119887
1sin 120591 + 119886
119896cos 119896120591 + 119887
119896sin 119896120591 (35)
The amplitudes of fundamental and 119896th harmonic of wave-form of type (35) respectively are
1205821= radic11988621+ 11988721 (36)
120582119896= radic1198862119896+ 1198872119896 (37)
As it is shown in Section 21 nonnegative waveforms withmaximal amplitude of fundamental harmonic or maximalcoefficient of fundamental harmonic cosine part have atleast one zero It is also shown in Section 22 (Corollary 4)that waveforms of type (35) with nonzero amplitude offundamental harmonic have either one or two globalminimaConsequently if nonnegative waveform of type (35) withnonzero amplitude of fundamental harmonic has at least onezero then it has at most two zeros
In Section 31 we provide general description of nonnega-tive waveforms of type (35) with at least one zero In Sections32 and 33 we consider nonnegative waveforms of type (35)with two zeros
31 General Description of Nonnegative Waveforms with atLeast One Zero The main result of this section is presentedin the following proposition
Proposition 6 Every nonnegative waveform of type (35) withat least one zero can be expressed in the following form
119879119896(120591) = [1 minus cos (120591 minus 120591
0)] [1 minus 120582
119896119903119896(120591)] (38)
where119903119896(120591) = (119896 minus 1) cos 120585
+ 2
119896minus1
sum
119899=1
(119896 minus 119899) cos (119899 (120591 minus 1205910) + 120585)
(39)
providing that
120582119896le [(119896 minus 1) cos 120585 + 119896 sin(120585 minus 120585119896)
sin(120585119896)]
minus1
(40)
10038161003816100381610038161205851003816100381610038161003816 le 120587 (41)
Remark 7 Function on the right hand side of (40) is mono-tonically increasing function of |120585| on interval |120585| le 120587 (formore details about this function see Remark 15) From (57)and (65) it follows that relation
0 le 120582119896le 1 (42)
holds for every nonnegative waveform of type (35) Noticethat according to (40) 120582
119896= 1 implies |120585| = 120587 Substitution
of 120582119896
= 1 and |120585| = 120587 into (55) yields 119879119896(120591) = 1 minus
cos 119896(120591 minus 1205910) Consequently 120582
119896= 1 implies that amplitude
1205821of fundamental harmonic is equal to zero
Remark 8 Conversion of (38) into additive form leads to thefollowing expressions for coefficients of nonnegative wave-forms of type (35) with at least one zero
1198861= minus (1 + 120582
119896cos 120585) cos 120591
0minus 119896120582119896sin 120585 sin 120591
0 (43)
1198871= minus (1 + 120582
119896cos 120585) sin 120591
0+ 119896120582119896sin 120585 cos 120591
0 (44)
119886119896= 120582119896cos (119896120591
0minus 120585) (45)
119887119896= 120582119896sin (119896120591
0minus 120585) (46)
providing that 120582119896satisfy (40) and |120585| le 120587
Mathematical Problems in Engineering 7
2
1
0
0 1 2 3 4
Angle 120591120587
Wav
efor
ms
1205823 = radic2201205823 = radic281205823 = radic24
Figure 4 Nonnegative waveforms with at least one zero for 119896 = 31205910= 1205876 and 120585 = 31205874
Three examples of nonnegative waveforms with at leastone zero for 119896 = 3 are presented in Figure 4 (examples ofnonnegative waveformswith at least one zero for 119896 = 2 can befound in [12]) For all three waveforms presented in Figure 4we assume that 120591
0= 1205876 and 120585 = 31205874 From (40) it follows
that 1205823le radic24 Coefficients of waveform with 120582
3= radic220
(dotted line) are 1198861= minus08977 119887
1= minus03451 119886
3= 005
and 1198873= minus005 Coefficients of waveform with 120582
3= radic28
(dashed line) are 1198861= minus09453 119887
1= minus01127 119886
3= 0125
and 1198873= minus0125 Coefficients of waveform with 120582
3= radic24
(solid line) are 1198861= minus10245 119887
1= 02745 119886
3= 025 and
1198873= minus025 First two waveforms have one zero while third
waveform (presented with solid line) has two zeros
Proof of Proposition 6 Waveform of type (35) containingdc component fundamental and 119896th harmonic can be alsoexpressed in the form
119879119896(120591) = 1 + 120582
1cos (120591 + 120593
1) + 120582119896cos (119896120591 + 120593
119896) (47)
where1205821ge 0120582
119896ge 0120593
1isin (minus120587 120587] and120593
119896isin (minus120587 120587] It is easy
to see that relations between coefficient of (35) and param-eters of (47) read as follows
1198861= 1205821cos1205931 119887
1= minus1205821sin1205931 (48)
119886119896= 120582119896cos120593119896 119887
119896= minus120582119896sin120593119896 (49)
Let us introduce 120585 such that10038161003816100381610038161205851003816100381610038161003816 le 120587 120585 = (119896120591
0+ 120593119896) mod2120587 (50)
Using (50) coefficients (49) can be expressed as (45)-(46)Let us assume that 119879
119896(120591) is nonnegative waveform of type
(35) with at least one zero that is 119879119896(120591) ge 0 and 119879
119896(1205910) = 0
for some 1205910 Notice that conditions 119879
119896(120591) ge 0 and 119879
119896(1205910) = 0
imply that 1198791015840119896(1205910) = 0 From 119879
119896(1205910) = 0 and 1198791015840
119896(1205910) = 0 by
using (50) it follows that
1205821cos (120591
0+ 1205931) = minus (1 + 120582
119896cos 120585)
1205821sin (1205910+ 1205931) = minus119896120582
119896sin 120585
(51)
respectively On the other hand 1205821cos(120591 + 120593
1) can be rewrit-
ten as
1205821cos (120591 + 120593
1) = 1205821cos (120591
0+ 1205931) cos (120591 minus 120591
0)
minus 1205821sin (1205910+ 1205931) sin (120591 minus 120591
0)
(52)
Substitution of (51) into (52) yields
1205821cos (120591 + 120593
1) = minus (1 + 120582
119896cos 120585) cos (120591 minus 120591
0)
+ 119896120582119896sin 120585 sin (120591 minus 120591
0)
(53)
According to (50) it follows that cos(119896120591 + 120593119896) = cos(119896(120591 minus
1205910) + 120585) that is
cos (119896120591 + 120593119896) = cos 120585 cos 119896 (120591 minus 120591
0) minus sin 120585 sin 119896 (120591 minus 120591
0)
(54)
Furthermore substitution of (54) and (53) into (47) leads to
119879119896(120591) = [1 minus cos (120591 minus 120591
0)] [1 + 120582
119896cos 120585]
minus 120582119896[1 minus cos 119896 (120591 minus 120591
0)] cos 120585
+ 120582119896[119896 sin (120591 minus 120591
0) minus sin 119896 (120591 minus 120591
0)] sin 120585
(55)
According to (A2) and (A4) (see Appendices) there is com-mon factor [1minuscos(120591minus120591
0)] for all terms in (55) Consequently
(55) can be written in the form (38) where
119903119896(120591) = minus cos 120585 + [
1 minus cos 119896 (120591 minus 1205910)
1 minus cos (120591 minus 1205910)] cos 120585
minus [119896 sin (120591 minus 120591
0) minus sin 119896 (120591 minus 120591
0)
1 minus cos (120591 minus 1205910)
] sin 120585
(56)
From (56) by using (A2) (A4) and cos 120585 cos 119899(120591 minus 1205910) minus
sin 120585 sin 119899(120591 minus 1205910) = cos(119899(120591 minus 120591
0) + 120585) we obtain (39)
In what follows we are going to prove that (40) also holdsAccording to (38) 119879
119896(120591) is nonnegative if and only if
120582119896max120591
119903119896(120591) le 1 (57)
Let us first show that position of global maximumof 119903119896(120591)
belongs to the interval |120591 minus 1205910| le 2120587119896 Relation (56) can be
rewritten as
119903119896(120591) = 119903
119896(1205910minus2120585
119896) + 119902119896(120591) (58)
where
119903119896(1205910minus2120585
119896) = (119896 minus 1) cos 120585 + 119896 sin (120585 minus 120585119896)
sin (120585119896) (59)
119902119896(120591) =
1
1 minus cos (120591 minus 1205910)
sdot [cos 120585 minus cos(119896(120591 minus 1205910+120585
119896))
+119896 sin 120585sin (120585119896)
(cos(120591 minus 1205910+120585
119896) minus cos(120585
119896))]
(60)
8 Mathematical Problems in Engineering
For |120585| lt 120587 relation sin 120585 sin(120585119896) gt 0 obviously holds Fromcos 119905 gt cos 1199051015840 for |119905| le 120587119896 lt |119905
1015840
| le 120587 it follows that positionof global maximum of the function of type [119888 cos 119905 minus cos(119896119905)]for 119888 gt 0 belongs to interval |119905| le 120587119896 Therefore position ofglobal maximum of the expression in the square brackets in(60) for |120585| lt 120587 belongs to interval |120591 minus 120591
0+ 120585119896| le 120587119896 This
inequality together with |120585| lt 120587 leads to |120591minus1205910| lt 2120587119896 Since
[1 minus cos(120591 minus 1205910)]minus1 decreases with increasing |120591 minus 120591
0| le 120587 it
follows that 119902119896(120591) for |120585| lt 120587 has global maximum on interval
|120591minus1205910| lt 2120587119896 For |120585| = 120587 it is easy to show thatmax
120591119902119896(120591) =
119902119896(1205910plusmn 2120587119896) = 0 Since 119903
119896(120591) minus 119902
119896(120591) is constant (see (58))
it follows from previous considerations that 119903119896(120591) has global
maximum on interval |120591 minus 1205910| le 2120587119896
To find max120591119903119896(120591) let us consider first derivative of 119903
119896(120591)
with respect to 120591 Starting from (56) first derivative of 119903119896(120591)
can be expressed in the following form
119889119903119896(120591)
119889120591= minus119904 (120591) sdot sin(
119896 (120591 minus 1205910)
2+ 120585) (61)
where
119904 (120591) = [sin(119896 (120591 minus 120591
0)
2) cos(
120591 minus 1205910
2)
minus 119896 cos(119896 (120591 minus 120591
0)
2) sin(
120591 minus 1205910
2)]
sdot sinminus3 (120591 minus 1205910
2)
(62)
Using (A6) (see Appendices) (62) can be rewritten as
119904 (120591) = 2
119896minus1
sum
119899=1
119899 (119896 minus 119899) cos((119896 minus 2119899) (120591 minus 120591
0)
2) (63)
From 119899(119896 minus 119899) gt 0 and |119896 minus 2119899| lt 119896 119899 = 1 (119896 minus 1) itfollows that all summands in (63) decrease with increasing|120591 minus 1205910| providing that |120591 minus 120591
0| le 2120587119896 Therefore 119904(120591) ge 119904(120591
0plusmn
2120587119896) = 119896sin2(120587119896) gt 0 for |120591 minus 1205910| le 2120587119896 Consequently
119889119903119896(120591)119889120591 = 0 and |120591minus120591
0| le 2120587119896 imply that sin(119896(120591minus120591
0)2+
120585) = 0From |120585| le 120587 |120591minus120591
0| le 2120587119896 and sin(119896(120591minus120591
0)2+120585) = 0
it follows that 120591minus1205910+120585119896 = minus120585119896 or |120591minus120591
0+120585119896| = (2120587minus|120585|)119896
and therefore cos(119896(120591 minus 1205910+ 120585119896)) = cos 120585 Since cos(120585119896) ge
cos(2120587 minus |120585|)119896 it follows that max120591119902119896(120591) is attained for 120591 =
1205910minus2120585119896 Furthermore from (60) it follows that max
120591119902119896(120591) =
119902119896(1205910minus 2120585119896) = 0 which together with (58)-(59) leads to
max120591
119903119896(120591) = 119903
119896(1205910minus2120585
119896)
= (119896 minus 1) cos 120585 + 119896 sin (120585 minus 120585119896)sin (120585119896)
(64)
Both terms on the right hand side of (64) are even functionsof 120585 and decrease with increase of |120585| |120585| le 120587 Thereforemax120591119903119896(120591) attains its lowest value for |120585| = 120587 It is easy to
show that right hand side of (64) for |120585| = 120587 is equal to 1which further implies that
max120591
119903119896(120591) ge 1 (65)
From (65) it follows that (57) can be rewritten as 120582119896
le
[max120591119903119896(120591)]minus1 Finally substitution of (64) into 120582
119896le
[max120591119903119896(120591)]minus1 leads to (40) which completes the proof
32 Nonnegative Waveforms with Two Zeros Nonnegativewaveforms of type (35) with two zeros always possess twoglobal minima Such nonnegative waveforms are thereforerelated to the conflict set
In this subsection we provide general description of non-negative waveforms of type (35) for 119896 ge 2 and exactly twozeros According to Remark 7 120582
119896= 1 implies |120585| = 120587 and
119879119896(120591) = 1 minus cos 119896(120591 minus 120591
0) Number of zeros of 119879
119896(120591) = 1 minus
cos 119896(120591minus1205910) on fundamental period equals 119896 which is greater
than two for 119896 gt 2 and equal to two for 119896 = 2 In the followingproposition we exclude all waveforms with 120582
119896= 1 (the case
when 119896 = 2 and 1205822= 1 is going to be discussed in Remark 10)
Proposition 9 Every nonnegative waveform of type (35) withexactly two zeros can be expressed in the following form
119879119896(120591) = 120582
119896[1 minus cos (120591 minus 120591
0)] [1 minus cos(120591 minus 120591
0+2120585
119896)]
sdot [1198880+ 2
119896minus2
sum
119899=1
119888119899cos 119899(120591 minus 120591
0+120585
119896)]
(66)
where
119888119899= [sin(120585 minus 119899120585
119896) cos(120585
119896)
minus (119896 minus 119899) cos(120585 minus 119899120585
119896) sin(120585
119896)]
sdot sinminus3 (120585119896)
(67)
120582119896= [(119896 minus 1) cos 120585 + 119896 sin(120585 minus 120585119896)
sin(120585119896)]
minus1
(68)
0 lt10038161003816100381610038161205851003816100381610038161003816 lt 120587 (69)
Remark 10 For 119896 = 2 waveforms with 1205822= 1 also have
exactly two zeros These waveforms can be included in aboveproposition by substituting (69) with 0 lt |120585| le 120587
Remark 11 Apart from nonnegative waveforms of type (35)with two zeros there are another two types of nonnegativewaveforms which can be obtained from (66)ndash(68) These are
(i) nonnegative waveforms with 119896 zeros (correspondingto |120585| = 120587) and
(ii) maximally flat nonnegative waveforms (correspond-ing to 120585 = 0)
Notice that nonnegative waveforms of type (35) with120582119896= 1 can be obtained from (66)ndash(68) by setting |120585| =
120587 Substitution of 120582119896
= 1 and |120585| = 120587 into (66) alongwith execution of all multiplications and usage of (A2) (seeAppendices) leads to 119879
119896(120591) = 1 minus cos 119896(120591 minus 120591
0)
Mathematical Problems in Engineering 9
Also maximally flat nonnegative waveforms (they haveonly one zero [21]) can be obtained from (66)ndash(68) by setting120585 = 0 Thus substitution of 120585 = 0 into (66)ndash(68) leads tothe following form of maximally flat nonnegative waveformof type (35)
119879119896(120591) =
[1 minus cos (120591 minus 1205910)]2
3 (1198962 minus 1)
sdot [119896 (1198962
minus 1)
+ 2
119896minus2
sum
119899=1
(119896 minus 119899) ((119896 minus 119899)2
minus 1) cos 119899 (120591 minus 1205910)]
(70)
Maximally flat nonnegative waveforms of type (35) for 119896 le 4
can be expressed as
1198792(120591) =
2
3[1 minus cos(120591 minus 120591
0)]2
1198793(120591) =
1
2[1 minus cos(120591 minus 120591
0)]2
[2 + cos (120591 minus 1205910)]
1198794(120591) =
4
15[1 minus cos (120591 minus 120591
0)]2
sdot [5 + 4 cos (120591 minus 1205910) + cos 2 (120591 minus 120591
0)]
(71)
Remark 12 Every nonnegative waveform of type (35) withexactly one zero at nondegenerate critical point can bedescribed as in Proposition 6 providing that symbol ldquolerdquoin relation (40) is replaced with ldquoltrdquo This is an immediateconsequence of Propositions 6 and 9 and Remark 11
Remark 13 Identity [1minus cos(120591minus1205910)][1minus cos(120591minus120591
0+2120585119896)] =
[cos 120585119896 minus cos(120591 minus 1205910+ 120585119896)]
2 implies that (66) can be alsorewritten as
119879119896(120591) = 120582
119896[cos 120585119896
minus cos(120591 minus 1205910+120585
119896)]
2
sdot [1198880+ 2
119896minus2
sum
119899=1
119888119899cos 119899(120591 minus 120591
0+120585
119896)]
(72)
Furthermore substitution of (67) into (72) leads to
119879119896(120591)
= 120582119896[cos 120585119896
minus cos(120591 minus 1205910+120585
119896)]
sdot [(119896 minus 1) sin 120585sin (120585119896)
minus 2
119896minus1
sum
119899=1
sin (120585 minus 119899120585119896)sin (120585119896)
cos 119899(120591 minus 1205910+120585
119896)]
(73)
Remark 14 According to (A6) (see Appendices) it followsthat coefficients (67) can be expressed as
119888119899= 2
119896minus119899minus1
sum
119898=1
119898(119896 minus 119899 minus 119898) cos((119896 minus 119899 minus 2119898) 120585119896
) (74)
Furthermore from (74) it follows that coefficients 119888119896minus2
119888119896minus3
119888119896minus4
and 119888119896minus5
are equal to119888119896minus2
= 2 (75)
119888119896minus3
= 8 cos(120585119896) (76)
119888119896minus4
= 8 + 12 cos(2120585119896) (77)
119888119896minus5
= 24 cos(120585119896) + 16 cos(3120585
119896) (78)
For example for 119896 = 2 (75) and (68) lead to 1198880= 2 and
1205822= 1(2+cos 120585) respectively which from (72) further imply
that
1198792(120591) =
2 [cos(1205852) minus cos(120591 minus 1205910+ 1205852)]
2
[2 + cos 120585] (79)
Also for 119896 = 3 (75) (76) and (68) lead to 1198881= 2 119888
0=
8 cos(1205853) and 1205823= [2(3 cos(1205853) + cos 120585)]minus1 respectively
which from (72) further imply that
1198793(120591) =
2 [cos (1205853) minus cos (120591 minus 1205910+ 1205853)]
2
[3 cos (1205853) + cos 120585]
sdot [2 cos(1205853) + cos(120591 minus 120591
0+120585
3)]
(80)
Remark 15 According to (A5) (see Appendices) relation(68) can be rewritten as
120582119896= [(119896 minus 1) cos 120585 + 119896
119896minus1
sum
119899=1
cos((119896 minus 2119899)120585119896
)]
minus1
(81)
Clearly amplitude 120582119896of 119896th harmonic of nonnegative wave-
form of type (35) with exactly two zeros is even functionof 120585 Since cos((119896 minus 2119899)120585119896) 119899 = 0 (119896 minus 1) decreaseswith increase of |120585| on interval 0 le |120585| le 120587 it follows that120582119896monotonically increases with increase of |120585| Right hand
side of (68) is equal to 1(1198962
minus 1) for 120585 = 0 and to one for|120585| = 120587 Therefore for nonnegative waveforms of type (35)with exactly two zeros the following relation holds
1
1198962 minus 1lt 120582119896lt 1 (82)
The left boundary in (82) corresponds to maximally flatnonnegative waveforms (see Remark 11) The right boundaryin (82) corresponds to nonnegative waveforms with 119896 zeros(also see Remark 11)
Amplitude of 119896th harmonic of nonnegative waveform oftype (35) with two zeros as a function of parameter 120585 for 119896 le5 is presented in Figure 5
Remark 16 Nonnegative waveform of type (35) with twozeros can be also expressed in the following form
119879119896(120591) = 1 minus 120582
119896
119896 sin 120585sin (120585119896)
cos(120591 minus 1205910+120585
119896)
+ 120582119896cos (119896 (120591 minus 120591
0) + 120585)
(83)
10 Mathematical Problems in Engineering
1
08
06
04
02
0minus1 minus05 0 05
Am
plitu
de120582k
1
Parameter 120585120587
k = 2k = 3
k = 4k = 5
Figure 5 Amplitude of 119896th harmonic of nonnegative waveformwith two zeros as a function of parameter 120585
where 120582119896is given by (68) and 0 lt |120585| lt 120587 From (83) it follows
that coefficients of fundamental harmonic of nonnegativewaveform of type (35) with two zeros are
1198861= minus1205821cos(120591
0minus120585
119896) 119887
1= minus1205821sin(120591
0minus120585
119896) (84)
where 1205821is amplitude of fundamental harmonic
1205821=
119896 sin 120585sin (120585119896)
120582119896 (85)
Coefficients of 119896th harmonic are given by (45)-(46)Notice that (68) can be rewritten as
120582119896= [cos(120585
119896)
119896 sin 120585sin(120585119896)
minus cos 120585]minus1
(86)
By introducing new variable
119909 = cos(120585119896) (87)
and using the Chebyshev polynomials (eg see Appendices)relations (85) and (86) can be rewritten as
1205821= 119896120582119896119880119896minus1
(119909) (88)
120582119896=
1
119896119909119880119896minus1
(119909) minus 119881119896(119909)
(89)
where119881119896(119909) and119880
119896(119909) denote the Chebyshev polynomials of
the first and second kind respectively From (89) it followsthat
120582119896[119896119909119880119896minus1
(119909) minus 119881119896(119909)] minus 1 = 0 (90)
which is polynomial equation of 119896th degree in terms of var-iable 119909 From 0 lt |120585| lt 120587 and (87) it follows that
cos(120587119896) lt 119909 lt 1 (91)
Since 120582119896is monotonically increasing function of |120585| 0 lt |120585| lt
120587 it follows that 120582119896is monotonically decreasing function of
119909 This further implies that (90) has only one solution thatsatisfies (91) (For 119896 = 2 expression (91) reads cos(1205872) le
119909 lt 1) This solution for 119909 (which can be obtained at leastnumerically) according to (88) leads to amplitude 120582
1of
fundamental harmonicFor 119896 le 4 solutions of (90) and (91) are
119909 = radic1 minus 1205822
21205822
1
3lt 1205822le 1
119909 =1
23radic1205823
1
8lt 1205823lt 1
119909 = radic1
6(1 + radic
51205824+ 3
21205824
)1
15lt 1205824lt 1
(92)
Insertion of (92) into (88) leads to the following relationsbetween amplitude 120582
1of fundamental and amplitude 120582
119896of
119896th harmonic 119896 le 4
1205821= radic8120582
2(1 minus 120582
2)
1
3lt 1205822le 1 (93)
1205821= 3 (
3radic1205823minus 1205823)
1
8lt 1205823lt 1 (94)
1205821= radic
32
27(radic2120582
4(3 + 5120582
4)3
minus 21205824(9 + 7120582
4))
1
15lt 1205824lt 1
(95)
Proof of Proposition 9 As it has been shown earlier (seeProposition 6) nonnegative waveform of type (35) with atleast one zero can be represented in form (38) Since weexclude nonnegative waveforms with 120582
119896= 1 according to
Remark 7 it follows that we exclude case |120585| = 120587Therefore inthe quest for nonnegative waveforms of type (35) having twozeros we will start with waveforms of type (38) for |120585| lt 120587It is clear that nonnegative waveforms of type (38) have twozeros if and only if
120582119896= [max120591
119903119896(120591)]minus1
(96)
and max120591119903119896(120591) = 119903
119896(1205910) According to (64) max
120591119903119896(120591) =
119903119896(1205910) implies |120585| = 0 Therefore it is sufficient to consider
only the interval (69)Substituting (96) into (38) we obtain
119879119896(120591) =
[1 minus cos (120591 minus 1205910)] [max
120591119903119896(120591) minus 119903
119896(120591)]
max120591119903119896(120591)
(97)
Mathematical Problems in Engineering 11
Expression max120591119903119896(120591) minus 119903
119896(120591) according to (64) and (39)
equals
max120591
119903119896(120591) minus 119903
119896(120591) = 119896
sin ((119896 minus 1) 120585119896)sin (120585119896)
minus 2
119896minus1
sum
119899=1
(119896 minus 119899) cos (119899 (120591 minus 1205910) + 120585)
(98)
Comparison of (97) with (66) yields
max120591
119903119896(120591) minus 119903
119896(120591) = [1 minus cos(120591 minus 120591
0+2120585
119896)]
sdot [1198880+ 2
119896minus2
sum
119899=1
119888119899cos 119899(120591 minus 120591
0+120585
119896)]
(99)
where coefficients 119888119899 119899 = 0 119896 minus 2 are given by (67) In
what follows we are going to show that right hand sides of(98) and (99) are equal
From (67) it follows that
1198880minus 1198881cos(120585
119896) = 119896
sin (120585 minus 120585119896)sin (120585119896)
(100)
Also from (67) for 119899 = 1 119896minus3 it follows that the followingrelations hold
(119888119899minus1
+ 119888119899+1
) cos(120585119896) minus 2119888
119899= 2 (119896 minus 119899) cos(120585 minus 119899120585
119896)
(119888119899minus1
minus 119888119899+1
) sin(120585119896) = 2 (119896 minus 119899) sin(120585 minus 119899120585
119896)
(101)
From (99) by using (75) (76) (100)-(101) and trigonometricidentities
cos(120591 minus 1205910+2120585
119896) = cos(120585
119896) cos(120591 minus 120591
0+120585
119896)
minus sin(120585119896) sin(120591 minus 120591
0+120585
119896)
cos(120585 minus 119899120585
119896) cos(119899(120591 minus 120591
0+120585
119896))
minus sin(120585 minus 119899120585
119896) sin(119899(120591 minus 120591
0+120585
119896))
= cos (119899 (120591 minus 1205910) + 120585)
(102)
we obtain (98) Consequently (98) and (99) are equal whichcompletes the proof
33 Nonnegative Waveforms with Two Zeros and PrescribedCoefficients of 119896thHarmonic In this subsectionwe show thatfor prescribed coefficients 119886
119896and 119887119896 there are 119896 nonnegative
waveforms of type (35) with exactly two zeros According to
(37) and (82) coefficients 119886119896and 119887119896of nonnegative waveforms
of type (35) with exactly two zeros satisfy the followingrelation
1
1198962 minus 1lt radic1198862119896+ 1198872119896lt 1 (103)
According to Remark 16 the value of 119909 (see (87)) that cor-responds to 120582
119896= radic1198862
119896+ 1198872119896can be determined from (90)-
(91) As we mentioned earlier (90) has only one solutionthat satisfies (91) This value of 119909 according to (88) leadsto the amplitude 120582
1of fundamental harmonic (closed form
expressions for 1205821in terms of 120582
119896and 119896 le 4 are given by (93)ndash
(95))On the other hand from (45)-(46) it follows that
1198961205910minus 120585 = atan 2 (119887
119896 119886119896) + 2119902120587 119902 = 1 (119896 minus 1) (104)
where function atan 2(119910 119909) is defined as
atan 2 (119910 119909) =
arctan(119910
119909) if 119909 ge 0
arctan(119910
119909) + 120587 if 119909 lt 0 119910 ge 0
arctan(119910
119909) minus 120587 if 119909 lt 0 119910 lt 0
(105)
with the codomain (minus120587 120587] Furthermore according to (84)and (104) the coefficients of fundamental harmonic of non-negative waveforms with two zeros and prescribed coeffi-cients of 119896th harmonic are equal to
1198861= minus1205821cos[
atan 2 (119887119896 119886119896) + 2119902120587
119896]
1198871= minus1205821sin[
atan 2 (119887119896 119886119896) + 2119902120587
119896]
(106)
where 119902 = 0 (119896minus 1) For chosen 119902 according to (104) and(66) positions of zeros are
1205910=1
119896[120585 + atan 2 (119887
119896 119886119896) + 2119902120587]
1205910minus2120585
119896=1
119896[minus120585 + atan 2 (119887
119896 119886119896) + 2119902120587]
(107)
From (106) and 119902 = 0 (119896minus1) it follows that for prescribedcoefficients 119886
119896and 119887119896 there are 119896 nonnegative waveforms of
type (35) with exactly two zerosWe provide here an algorithm to facilitate calculation
of coefficients 1198861and 1198871of nonnegative waveforms of type
(35) with two zeros and prescribed coefficients 119886119896and 119887
119896
providing that 119886119896and 119887119896satisfy (103)
12 Mathematical Problems in Engineering
2
1
0
Wav
efor
ms
0 1 2 3 4
Angle 120591120587
q = 0
q = 1
q = 2
Figure 6 Nonnegative waveforms with two zeros for 119896 = 3 1198863=
minus015 and 1198873= minus02
Algorithm 17 (i) Calculate 120582119896= radic1198862119896+ 1198872119896
(ii) identify 119909 that satisfies both relations (90) and (91)(iii) calculate 120582
1according to (88)
(iv) choose integer 119902 such that 0 le 119902 le 119896 minus 1(v) calculate 119886
1and 1198871according to (106)
For 119896 le 4 by using (93) for 119896 = 2 (94) for 119896 = 3 and (95)for 119896 = 4 it is possible to calculate directly 120582
1from 120582
119896and
proceed to step (iv)For 119896 = 2 and prescribed coefficients 119886
2and 1198872 there are
two waveforms with two zeros one corresponding to 1198861lt 0
and the other corresponding to 1198861gt 0 (see also [12])
Let us take as an input 119896 = 3 1198863= minus015 and 119887
3= minus02
Execution of Algorithm 17 on this input yields 1205823= 025 and
1205821= 11399 (according to (94)) For 119902 = 0 we calculate
1198861= minus08432 and 119887
1= 07670 (corresponding waveform is
presented by solid line in Figure 6) for 119902 = 1 we calculate1198861= minus02426 and 119887
1= minus11138 (corresponding waveform is
presented by dashed line) for 119902 = 2 we calculate 1198861= 10859
and 1198871= 03468 (corresponding waveform is presented by
dotted line)As another example of the usage of Algorithm 17 let us
consider case 119896 = 4 and assume that1198864= minus015 and 119887
4= minus02
Consequently 1205824= 025 and 120582
1= 09861 (according to (95))
For 119902 = 0 3we calculate the following four pairs (1198861 1198871) of
coefficients of fundamental harmonic (minus08388 05184) for119902 = 0 (minus05184 minus08388) for 119902 = 1 (08388 minus05184) for 119902 =2 and (05184 08388) for 119902 = 3 Corresponding waveformsare presented in Figure 7
4 Nonnegative Waveforms with MaximalAmplitude of Fundamental Harmonic
In this section we provide general description of nonnegativewaveforms containing fundamental and 119896th harmonic withmaximal amplitude of fundamental harmonic for prescribedamplitude of 119896th harmonic
The main result of this section is presented in the fol-lowing proposition
3
2
1
0
Wav
efor
ms
0 1 2 3 4
Angle 120591120587
q = 0q = 1
q = 2q = 3
Figure 7 Nonnegative waveforms with two zeros for 119896 = 4 1198864=
minus015 and 1198874= minus02
Proposition 18 Every nonnegativewaveformof type (35)withmaximal amplitude 120582
1of fundamental harmonic and pre-
scribed amplitude 120582119896of 119896th harmonic can be expressed in the
following form
119879119896(120591) = [1 minus cos (120591 minus 120591
0)]
sdot [1 minus (119896 minus 1) 120582119896minus 2120582119896
119896minus1
sum
119899=1
(119896 minus 119899) cos (119899 (120591 minus 1205910))]
(108)
if 0 le 120582119896le 1(119896
2
minus 1) or
119879119896(120591) = 120582
119896[1 minus cos (120591 minus 120591
0)] [1 minus cos(120591 minus 120591
0+2120585
119896)]
sdot [1198880+ 2
119896minus2
sum
119899=1
119888119899cos 119899(120591 minus 120591
0+120585
119896)]
(109)
if 1(1198962 minus 1) le 120582119896le 1 providing that 119888
119899 119899 = 0 119896 minus 2 and
120582119896are related to 120585 via relations (67) and (68) respectively and
|120585| le 120587
Remark 19 Expression (108) can be obtained from (38) bysetting 120585 = 0 Furthermore insertion of 120585 = 0 into (43)ndash(46)leads to the following expressions for coefficients ofwaveformof type (108)
1198861= minus (1 + 120582
119896) cos 120591
0 119887
1= minus (1 + 120582
119896) sin 120591
0
119886119896= 120582119896cos (119896120591
0) 119887
119896= 120582119896sin (119896120591
0)
(110)
On the other hand (109) coincides with (66) Thereforethe expressions for coefficients of (109) and (66) also coincideThus expressions for coefficients of fundamental harmonic ofwaveform (109) are given by (84) where 120582
1is given by (85)
while expressions for coefficients of 119896th harmonic are givenby (45)-(46)
Waveforms described by (108) have exactly one zerowhile waveforms described by (109) for 1(1198962 minus 1) lt 120582
119896lt 1
Mathematical Problems in Engineering 13
14
12
1
08
06
04
02
00 05 1
Amplitude 120582k
Am
plitu
de1205821
k = 2
k = 3
k = 4
Figure 8 Maximal amplitude of fundamental harmonic as a func-tion of amplitude of 119896th harmonic
have exactly two zeros As we mentioned earlier waveforms(109) for 120582
119896= 1 have 119896 zeros
Remark 20 Maximal amplitude of fundamental harmonic ofnonnegative waveforms of type (35) for prescribed amplitudeof 119896th harmonic can be expressed as
1205821= 1 + 120582
119896 (111)
if 0 le 120582119896le 1(119896
2
minus 1) or
1205821=
119896 sin 120585119896 sin 120585 cos (120585119896) minus cos 120585 sin (120585119896)
(112)
if 1(1198962 minus 1) le 120582119896le 1 where 120585 is related to 120582
119896via (68) (or
(86)) and |120585| le 120587From (110) it follows that (111) holds Substitution of (86)
into (85) leads to (112)Notice that 120582
119896= 1(119896
2
minus 1) is the only common point ofthe intervals 0 le 120582
119896le 1(119896
2
minus 1) and 1(1198962
minus 1) le 120582119896le
1 According to (111) 120582119896= 1(119896
2
minus 1) corresponds to 1205821=
1198962
(1198962
minus1) It can be also obtained from (112) by setting 120585 = 0The waveforms corresponding to this pair of amplitudes aremaximally flat nonnegative waveforms
Maximal amplitude of fundamental harmonic of non-negative waveform of type (35) for 119896 le 4 as a function ofamplitude of 119896th harmonic is presented in Figure 8
Remark 21 Maximum value of amplitude of fundamentalharmonic of nonnegative waveform of type (35) is
1205821max =
1
cos (120587 (2119896)) (113)
This maximum value is attained for |120585| = 1205872 (see (112)) Thecorresponding value of amplitude of 119896th harmonic is 120582
119896=
(1119896) tan(120587(2119896)) Nonnegative waveforms of type (35) with1205821= 1205821max have two zeros at 1205910 and 1205910 minus 120587119896 for 120585 = 1205872 or
at 1205910and 1205910+ 120587119896 for 120585 = minus1205872
14
12
1
08
06
04
02
0minus1 minus05 0 05 1
Am
plitu
de1205821
Parameter 120585120587
k = 2k = 3k = 4
Figure 9 Maximal amplitude of fundamental harmonic as a func-tion of parameter 120585
To prove that (113) holds let us first show that the fol-lowing relation holds for 119896 ge 2
cos( 120587
2119896) lt 1 minus
1
1198962 (114)
From 119896 ge 2 it follows that sinc(120587(4119896)) gt sinc(1205874) wheresinc 119909 = (sin119909)119909 and therefore sin(120587(4119896)) gt 1(radic2119896)By using trigonometric identity cos 2119909 = 1 minus 2sin2119909 weimmediately obtain (114)
According to (111) and (112) it is clear that 1205821attains its
maximum value on the interval 1(1198962 minus 1) le 120582119896le 1 Since
120582119896is monotonic function of |120585| on interval |120585| le 120587 (see
Remark 15) it follows that 119889120582119896119889120585 = 0 for 0 lt |120585| lt 120587
Therefore to find critical points of 1205821as a function of 120582
119896
it is sufficient to find critical points of 1205821as a function of
|120585| 0 lt |120585| lt 120587 and consider its values at the end points120585 = 0 and |120585| = 120587 Plot of 120582
1as a function of parameter 120585
for 119896 le 4 is presented in Figure 9 According to (112) firstderivative of 120582
1with respect to 120585 is equal to zero if and only
if (119896 cos 120585 sin(120585119896) minus sin 120585 cos(120585119896)) cos 120585 = 0 On interval0 lt |120585| lt 120587 this is true if and only if |120585| = 1205872 Accordingto (112) 120582
1is equal to 119896
2
(1198962
minus 1) for 120585 = 0 equal to zerofor |120585| = 120587 and equal to 1 cos(120587(2119896)) for |120585| = 1205872 From(114) it follows that 1198962(1198962minus1) lt 1 cos(120587(2119896)) and thereforemaximum value of 120582
1is given by (113) Moreover maximum
value of 1205821is attained for |120585| = 1205872
According to above consideration all nonnegative wave-forms of type (35) having maximum value of amplitude offundamental harmonic can be obtained from (109) by setting|120585| = 1205872 Three of them corresponding to 119896 = 3 120585 = 1205872and three different values of 120591
0(01205876 and1205873) are presented
in Figure 10 Dotted line corresponds to 1205910= 0 (coefficients
of corresponding waveform are 1198861= minus1 119887
1= 1radic3 119886
3= 0
and 1198873= minusradic39) solid line to 120591
0= 1205876 (119886
1= minus2radic3 119887
1= 0
1198863= radic39 and 119887
3= 0) and dashed line to 120591
0= 1205873 (119886
1= minus1
1198871= minus1radic3 119886
3= 0 and 119887
3= radic39)
Proof of Proposition 18 As it has been shown earlier (Propo-sition 6) nonnegative waveform of type (35) with at least
14 Mathematical Problems in Engineering
2
1
0
0 1 2 3 4
Angle 120591120587
Wav
efor
ms
1205910 = 01205910 = 12058761205910 = 1205873
Figure 10 Nonnegative waveforms with maximum amplitude offundamental harmonic for 119896 = 3 and 120585 = 1205872
one zero can be represented in form (38) According to (43)(44) and (36) for amplitude 120582
1of fundamental harmonic of
waveforms of type (38) the following relation holds
1205821= radic(1 + 120582
119896cos 120585)2 + 11989621205822
119896sin2120585 (115)
where 120582119896satisfy (40) and |120585| le 120587
Because of (40) in the quest of finding maximal 1205821for
prescribed 120582119896 we have to consider the following two cases
(Case i)120582119896lt [(119896minus1) cos 120585 + 119896 sin(120585minus120585119896) sin(120585119896)]minus1
(Case ii)120582119896= [(119896minus1) cos 120585+119896 sin(120585minus120585119896) sin(120585119896)]minus1
Case i Since 120582119896lt [(119896 minus 1) cos 120585 + 119896 sin(120585 minus 120585119896) sin(120585119896)]minus1
implies 120582119896
= 1 according to (115) it follows that 1205821
= 0Hence 119889120582
1119889120585 = 0 implies
2120582119896sin 120585 [1 minus (1198962 minus 1) 120582
119896cos 120585] = 0 (116)
Therefore 1198891205821119889120585 = 0 if 120582
119896= 0 (Option 1) or sin 120585 = 0
(Option 2) or (1198962 minus 1)120582119896cos 120585 = 1 (Option 3)
Option 1 According to (115) 120582119896= 0 implies 120582
1= 1 (notice
that this implication shows that 1205821does not depend on 120585 and
therefore we can set 120585 to zero value)
Option 2 According to (115) sin 120585 = 0 implies 1205821= 1 +
120582119896cos 120585 which further leads to the conclusion that 120582
1is
maximal for 120585 = 0 For 120585 = 0 120582119896lt [(119896 minus 1) cos 120585 + 119896 sin(120585 minus
120585119896) sin(120585119896)]minus1 becomes 120582119896lt 1(119896
2
minus 1)
Option 3 This option leads to contradiction To show thatnotice that (119896
2
minus 1)120582119896cos 120585 = 1 and 120582
119896lt [(119896 minus
1) cos 120585 + 119896 sin(120585 minus 120585119896) sin(120585119896)]minus1 imply that (119896 minus 1) cos 120585 gtsin(120585minus120585119896) sin(120585119896) Using (A5) (see Appendices) the latestinequality can be rewritten assum119896minus1
119899=1[cos 120585minuscos((119896minus2119899)120585119896)] gt
0 But from |119896 minus 2119899| lt 119896 119899 = 1 (119896 minus 1) and |120585| le 120587
it follows that all summands are not positive and therefore(119896minus1) cos 120585 gt sin(120585minus120585119896) sin(120585119896) does not hold for |120585| le 120587
Consequently Case i implies 120585 = 0 and 120582119896lt 1(119896
2
minus 1)Finally substitution of 120585 = 0 into (38) leads to (108) whichproves that (108) holds for 120582
119896lt 1(119896
2
minus 1)
Case ii Relation120582119896= [(119896minus1) cos 120585+119896 sin(120585minus120585119896) sin(120585119896)]minus1
according to Proposition 9 and Remark 11 implies that cor-responding waveforms can be expressed via (66)ndash(68) for|120585| le 120587 Furthermore 120582
119896= [(119896 minus 1) cos 120585 + 119896 sin(120585 minus
120585119896) sin(120585119896)]minus1 and |120585| le 120587 imply 1(1198962 minus 1) le 120582119896le 1
This proves that (109) holds for 1(1198962 minus 1) le 120582119896le 1
Finally let us prove that (108) holds for 120582119896= 1(119896
2
minus
1) According to (68) (see also Remark 11) this value of 120582119896
corresponds to 120585 = 0 Furthermore substitution of 120582119896=
1(1198962
minus 1) and 120585 = 0 into (109) leads to (70) which can berewritten as
119879119896(120591) =
[1 minus cos (120591 minus 1205910)]
(1 minus 1198962)
sdot [119896 (119896 minus 1) minus 2
119896minus1
sum
119899=1
(119896 minus 119899) cos (119899 (120591 minus 1205910))]
(117)
Waveform (117) coincides with waveform (108) for 120582119896
=
1(1 minus 1198962
) Consequently (108) holds for 120582119896= 1(1 minus 119896
2
)which completes the proof
5 Nonnegative Waveforms with MaximalAbsolute Value of the Coefficient of CosineTerm of Fundamental Harmonic
In this sectionwe consider general description of nonnegativewaveforms of type (35) with maximal absolute value ofcoefficient 119886
1for prescribed coefficients of 119896th harmonicThis
type of waveform is of particular interest in PA efficiencyanalysis In a number of cases of practical interest eithercurrent or voltage waveform is prescribed In such casesthe problem of finding maximal efficiency of PA can bereduced to the problem of finding nonnegative waveformwith maximal coefficient 119886
1for prescribed coefficients of 119896th
harmonic (see also Section 7)In Section 51 we provide general description of nonneg-
ative waveforms of type (35) with maximal absolute value ofcoefficient 119886
1for prescribed coefficients of 119896th harmonic In
Section 52 we illustrate results of Section 51 for particularcase 119896 = 3
51 Nonnegative Waveforms with Maximal Absolute Value ofCoefficient 119886
1for 119896 ge 2 Waveforms 119879
119896(120591) of type (35) with
1198861ge 0 can be derived from those with 119886
1le 0 by shifting
by 120587 and therefore we can assume without loss of generalitythat 119886
1le 0 Notice that if 119896 is even then shifting 119879
119896(120591) by
120587 produces the same result as replacement of 1198861with minus119886
1
(119886119896remains the same) On the other hand if 119896 is odd then
shifting 119879119896(120591) by 120587 produces the same result as replacement
of 1198861with minus119886
1and 119886119896with minus119886
119896
According to (37) coefficients of 119896th harmonic can beexpressed as
119886119896= 120582119896cos 120575 119887
119896= 120582119896sin 120575 (118)
Mathematical Problems in Engineering 15
where
|120575| le 120587 (119)
Conversely for prescribed coefficients 119886119896and 119887
119896 120575 can be
determined as
120575 = atan 2 (119887119896 119886119896) (120)
where definition of function atan 2(119910 119909) is given by (105)The main result of this section is stated in the following
proposition
Proposition 22 Every nonnegative waveform of type (35)withmaximal absolute value of coefficient 119886
1le 0 for prescribed
coefficients 119886119896and 119887119896of 119896th harmonic can be represented as
119879119896(120591)
= [1 minus cos 120591]
sdot [1 minus (119896 minus 1) 119886119896minus 2
119896minus1
sum
119899=1
(119896 minus 119899) (119886119896cos 119899120591 + 119887
119896sin 119899120591)]
(121)
if 119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) le 1+119886
119896 where 120575 = atan 2(bk
119886119896) or
119879119896(120591) = 120582
119896[1 minus cos(120591 minus (120575 + 120585)
119896)]
sdot [1 minus cos(120591 minus (120575 minus 120585)
119896)]
sdot [1198880+ 2
119896minus2
sum
119899=1
119888119899cos 119899(120591 minus 120575
119896)]
(122)
if 119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) ge 1 + 119886
119896 where 119888
119899 119899 = 0
119896minus2 and 120582119896= radic1198862119896+ 1198872119896are related to 120585 via relations (67) and
(68) respectively and |120585| le 120587
Remark 23 Expression (121) can be obtained from (38) bysetting 120591
0= 0 and 120585 = minus120575 and then replacing 120582
119896cos 120575 with
119886119896(see (118)) and 120582
119896cos(119899120591 minus 120575) with 119886
119896cos 119899120591 + 119887
119896sin 119899120591
(see also (118)) Furthermore insertion of 1205910= 0 and 120585 =
minus120575 into (43)ndash(46) leads to the following relations betweenfundamental and 119896th harmonic coefficients of waveform(121)
1198861= minus (1 + 119886
119896) 119887
1= minus119896119887
119896 (123)
On the other hand expression (122) can be obtained from(66) by replacing 120591
0minus120585119896with 120575119896 Therefore substitution of
1205910minus 120585119896 = 120575119896 in (84) leads to
1198861= minus1205821cos(120575
119896) 119887
1= minus1205821sin(120575
119896) (124)
where 1205821is given by (85)
The fundamental harmonic coefficients 1198861and 1198871of wave-
form of type (35) with maximal absolute value of coefficient1198861le 0 satisfy both relations (123) and (124) if 119886
119896and 119887119896satisfy
1 + 119886119896= 119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) For such waveforms
relations 1205910= 0 and 120585 = minus120575 also hold
Remark 24 Amplitude of 119896th harmonic of nonnegativewaveform of type (35) with maximal absolute value of coeffi-cient 119886
1le 0 and coefficients 119886
119896 119887119896satisfying 1 + 119886
119896=
119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) is
120582119896=
sin (120575119896)119896 sin 120575 cos (120575119896) minus cos 120575 sin (120575119896)
(125)
To show that it is sufficient to substitute 119886119896= 120582119896cos 120575 (see
(118)) into 1 + 119886119896= 119896120582119896[sin 120575 sin(120575119896)] cos(120575119896)
Introducing new variable
119910 = cos(120575119896) (126)
and using the Chebyshev polynomials (eg see Appendices)relations 119886
119896= 120582119896cos 120575 and (125) can be rewritten as
119886119896= 120582119896119881119896(119910) (127)
120582119896=
1
119896119910119880119896minus1
(119910) minus 119881119896(119910)
(128)
where119881119896(119910) and119880
119896(119910) denote the Chebyshev polynomials of
the first and second kind respectively Substitution of (128)into (127) leads to
119886119896119896119910119880119896minus1
(119910) minus (1 + 119886119896) 119881119896(119910) = 0 (129)
which is polynomial equation of 119896th degree in terms of var-iable 119910 From |120575| le 120587 and (126) it follows that
cos(120587119896) le 119910 le 1 (130)
In what follows we show that 119886119896is monotonically increas-
ing function of 119910 on the interval (130) From 120585 = minus120575 (seeRemark 23) and (81) it follows that 120582minus1
119896= (119896 minus 1) cos 120575 +
119896sum119896minus1
119899=1cos((119896 minus 2119899)120575119896) ge 1 and therefore 119886
119896= 120582119896cos 120575 can
be rewritten as
119886119896=
cos 120575(119896 minus 1) cos 120575 + 119896sum119896minus1
119899=1cos ((119896 minus 2119899) 120575119896)
(131)
Obviously 119886119896is even function of 120575 and all cosines in (131)
are monotonically decreasing functions of |120575| on the interval|120575| le 120587 It is easy to show that cos((119896 minus 2119899)120575119896) 119899 =
1 (119896 minus 1) decreases slower than cos 120575 when |120575| increasesThis implies that denominator of the right hand side of(131) decreases slower than numerator Since denominator ispositive for |120575| le 120587 it further implies that 119886
119896is decreasing
function of |120575| on interval |120575| le 120587 Consequently 119886119896is
monotonically increasing function of 119910 on the interval (130)Thus we have shown that 119886
119896is monotonically increasing
function of 119910 on the interval (130) and therefore (129) hasonly one solution that satisfies (130) According to (128) thevalue of 119910 obtained from (129) and (130) either analyticallyor numerically leads to amplitude 120582
119896of 119896th harmonic
16 Mathematical Problems in Engineering
1
05
0
minus05
minus1
minus1 minus05 0 05 1
Coefficient ak
Coe
ffici
entb
k
radica2k+ b2
kle 1
k = 2k = 3k = 4
Figure 11 Plot of (119886119896 119887119896) satisfying 1 + 119886
119896= 119896120582
119896[sin 120575 sin(120575
119896)] cos(120575119896) for 119896 le 4
By solving (129) and (130) for 119896 le 4 we obtain
119910 = radic1 + 1198862
2 (1 minus 1198862) minus1 le 119886
2le1
3
119910 = radic3
4 (1 minus 21198863) minus1 le 119886
3le1
8
119910 =radicradic2 minus 4119886
4+ 1011988624minus 2 (1 minus 119886
4)
4 (1 minus 31198864)
minus1 le 1198864le
1
15
(132)
Insertion of (132) into (128) leads to the following explicitexpressions for the amplitude 120582
119896 119896 le 4
1205822=1
2(1 minus 119886
2) minus1 le 119886
2le1
3 (133)
1205822
3= [
1
3(1 minus 2119886
3)]
3
minus1 le 1198863le1
8 (134)
1205824=1
4(minus1 minus 119886
4+ radic2 minus 4119886
4+ 1011988624) minus1 le 119886
4le
1
15
(135)
Relations (133)ndash(135) define closed lines (see Figure 11) whichseparate points representing waveforms of type (121) frompoints representing waveforms of type (122) For given 119896points inside the corresponding curve refer to nonnegativewaveforms of type (121) whereas points outside curve (andradic1198862119896+ 1198872119896le 1) correspond to nonnegative waveforms of type
(122) Points on the respective curve correspond to the wave-forms which can be expressed in both forms (121) and (122)
Remark 25 Themaximum absolute value of coefficient 1198861of
nonnegative waveform of type (35) is
100381610038161003816100381611988611003816100381610038161003816max =
1
cos (120587 (2119896)) (136)
This maximum value is attained for |120585| = 1205872 and 120575 = 0
(see (124)) Notice that |1198861|max is equal to the maximum value
1205821max of amplitude of fundamental harmonic (see (113))
Coefficients of waveform with maximum absolute value ofcoefficient 119886
1 1198861lt 0 are
1198861= minus
1
cos (120587 (2119896)) 119886
119896=1
119896tan( 120587
(2119896))
1198871= 119887119896= 0
(137)
Waveformdescribed by (137) is cosinewaveformhaving zerosat 120587(2119896) and minus120587(2119896)
In the course of proving (136) notice first that |1198861|max le
1205821max holds According to (123) and (124) maximum of |119886
1|
occurs for 119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) ge 1 + 119886
119896 From (124)
it immediately follows that maximum value of |1198861| is attained
if and only if 1205821= 1205821max and 120575 = 0 which because of
120575119896 = 1205910minus120585119896 further implies 120591
0= 120585119896 Sincemaximumvalue
of 1205821is attained for |120585| = 1205872 it follows that corresponding
waveform has zeros at 120587(2119896) and minus120587(2119896)
Proof of Proposition 22 As it was mentioned earlier in thissection we can assume without loss of generality that 119886
1le 0
We consider waveforms119879119896(120591) of type (35) such that119879
119896(120591) ge 0
and119879119896(120591) = 0 for some 120591
0 Fromassumption that nonnegative
waveform 119879119896(120591) of type (35) has at least one zero it follows
that it can be expressed in form (38)Let us also assume that 120591
0is position of nondegenerate
critical point Therefore 119879119896(1205910) = 0 implies 1198791015840
119896(1205910) = 0 and
11987910158401015840
119896(1205910) gt 0 According to (55) second derivative of 119879
119896(120591) at
1205910can be expressed as 11987910158401015840
119896(1205910) = 1 minus 120582
119896(1198962
minus 1) cos 120585 Since11987910158401015840
119896(1205910) gt 0 it follows immediately that
1 minus 120582119896(1198962
minus 1) cos 120585 gt 0 (138)
Let us further assume that 119879119896(120591) has exactly one zeroThe
problem of finding maximum absolute value of 1198861is con-
nected to the problem of finding maximum of the minimumfunction (see Section 21) If waveforms possess unique globalminimum at nondegenerate critical point then correspond-ing minimum function is a smooth function of parameters[13] Consequently assumption that 119879
119896(120591) has exactly one
zero at nondegenerate critical point leads to the conclusionthat coefficient 119886
1is differentiable function of 120591
0 First
derivative of 1198861(see (43)) with respect to 120591
0 taking into
account that 1205971205851205971205910= 119896 (see (50)) can be expressed in the
following factorized form
1205971198861
1205971205910
= sin 1205910[1 minus 120582
119896(1198962
minus 1) cos 120585] (139)
Mathematical Problems in Engineering 17
From (138) and (139) it is clear that 12059711988611205971205910= 0 if and only if
sin 1205910= 0 According toRemark 12 assumption that119879
119896(120591)has
exactly one zero implies 120582119896lt 1 From (51) (48) and 120582
119896lt 1
it follows that 1198861cos 1205910+ 1198871sin 1205910lt 0 which together with
sin 1205910= 0 implies that 119886
1cos 1205910lt 0 Assumption 119886
1le 0
together with relations 1198861cos 1205910lt 0 and sin 120591
0= 0 further
implies 1198861
= 0 and
1205910= 0 (140)
Insertion of 1205910= 0 into (38) leads to
119879119896(120591)
= [1 minus cos 120591]
sdot [1 minus (119896 minus 1) 120582119896cos 120585 minus 2120582
119896
119896minus1
sum
119899=1
(119896 minus 119899) cos (119899120591 + 120585)]
(141)
Substitution of 1205910= 0 into (45) and (46) yields 119886
119896= 120582119896cos 120585
and 119887119896
= minus120582119896sin 120585 respectively Replacing 120582
119896cos 120585 with
119886119896and 120582
119896cos(119899120591 + 120585) with (119886
119896cos 119899120591 + 119887
119896sin 119899120591) in (141)
immediately leads to (121)Furthermore 119886
119896= 120582119896cos 120585 119887
119896= minus120582
119896sin 120585 and (118)
imply that
120575 = minus120585 (142)
According to (38)ndash(40) and (142) it follows that (141) is non-negative if and only if
120582119896[(119896 minus 1) cos 120575 + 119896 sin (120575 minus 120575119896)
sin (120575119896)] lt 1 (143)
Notice that 119886119896= 120582119896cos 120575 implies that the following relation
holds
120582119896[(119896 minus 1) cos 120575 + 119896 sin (120575 minus 120575119896)
sin (120575119896)]
= minus119886119896+ 119896120582119896
sin 120575sin (120575119896)
cos(120575119896)
(144)
Finally substitution of (144) into (143) leads to 119896120582119896[sin 120575
sin(120575119896)] cos(120575119896) lt 1 + 119886119896 which proves that (121) holds
when 119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) lt 1 + 119886
119896
Apart from nonnegative waveforms with exactly one zeroat nondegenerate critical point in what follows we will alsoconsider other types of nonnegative waveforms with at leastone zero According to Proposition 9 and Remark 11 thesewaveforms can be described by (66)ndash(68) providing that 0 le|120585| le 120587
According to (35) 119879119896(0) ge 0 implies 1 + 119886
1+ 119886119896ge 0
Consequently 1198861le 0 implies that |119886
1| le 1 + 119886
119896 On the other
hand according to (123) |1198861| = 1 + 119886
119896holds for waveforms
of type (121) The converse is also true 1198861le 0 and |119886
1| =
1 + 119886119896imply 119886
1= minus1 minus 119886
119896 which further from (35) implies
119879119896(0) = 0 Therefore in what follows it is enough to consider
only nonnegativewaveformswhich can be described by (66)ndash(68) and 0 le |120585| le 120587 with coefficients 119886
119896and 119887119896satisfying
119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) ge 1 + 119886
119896
For prescribed coefficients 119886119896and 119887119896 the amplitude 120582
119896=
radic1198862119896+ 1198872119896of 119896th harmonic is also prescribed According to
Remark 15 (see also Remark 16) 120582119896is monotonically
decreasing function of 119909 = cos(120585119896) The value of 119909 can beobtained by solving (90) subject to the constraint cos(120587119896) le119909 le 1 Then 120582
1can be determined from (88) From (106) it
immediately follows that maximal absolute value of 1198861le 0
corresponds to 119902 = 0 which from (104) and (120) furtherimplies that
120575 = 1198961205910minus 120585 (145)
Furthermore 119902 = 0 according to (107) implies that waveformzeros are
1205910=(120575 + 120585)
119896 120591
1015840
0= 1205910minus2120585
119896=(120575 minus 120585)
119896 (146)
Substitution of 1205910= (120575 + 120585)119896 into (66) yields (122) which
proves that (122) holds when 119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) ge
1 + 119886119896
In what follows we prove that (121) also holds when119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) = 1 + 119886
119896 Substitution of 119886
119896=
120582119896cos 120575 into 119896120582
119896[sin 120575 sin(120575119896)] cos(120575119896) = 1 + 119886
119896leads to
120582119896[(119896 minus 1) cos 120575 + 119896 sin (120575 minus 120575119896)
sin (120575119896)] = 1 (147)
As we mentioned earlier relation (142) holds for all wave-forms of type (121) Substituting (142) into (147) we obtain
120582119896[(119896 minus 1) cos 120585 + 119896 sin (120585 minus 120585119896)
sin (120585119896)] = 1 (148)
This expression can be rearranged as
120582119896
119896 sin ((119896 minus 1) 120585119896)sin 120585119896
= 1 minus (119896 minus 1) 120582119896cos 120585 (149)
On the other hand for waveforms of type (122) according to(68) relations (148) and (149) also hold Substitution of 120591
0=
(120575 + 120585)119896 (see (145)) and (67) into (122) leads to
119879119896(120591)
= 120582119896[1 minus cos (120591 minus 120591
0)]
sdot [119896 sin ((119896 minus 1) 120585119896)
sin 120585119896minus 2
119896minus1
sum
119899=1
(119896 minus 119899) cos (119899 (120591 minus 1205910) + 120585)]
(150)
Furthermore substitution of (142) into (145) implies that1205910
= 0 Finally substitution of 1205910
= 0 and (149) into(150) leads to (141) Therefore (141) holds when 119896120582
119896[sin 120575
sin(120575119896)] cos(120575119896) = 1 + 119886119896 which in turn shows that (121)
holds when 119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) = 1 + 119886
119896 This
completes the proof
18 Mathematical Problems in Engineering
52 Nonnegative Waveforms with Maximal Absolute Valueof Coefficient 119886
1for 119896 = 3 Nonnegative waveform of type
(35) for 119896 = 3 is widely used in PA design (eg see [10])In this subsection we illustrate results of Section 51 for thisparticular case The case 119896 = 2 is presented in detail in [12]
Coefficients of fundamental harmonic of nonnegativewaveform of type (35) with 119896 = 3 and maximal absolutevalue of coefficient 119886
1le 0 for prescribed coefficients 119886
3and
1198873(1205823= radic11988623+ 11988723) according to (123) (124) (134) (94) and
(120) are equal to
1198861= minus1 minus 119886
3 119887
1= minus3119887
3 (151)
if 12058223le [(1 minus 2119886
3)3]3
1198861= minus1205821cos(120575
3) 119887
1= minus1205821sin(120575
3) (152)
where 1205821= 3(
3radic1205823minus 1205823) and 120575 = atan 2(119887
3 1198863) if [(1 minus
21198863)3]3
le 1205822
3le 1The line 1205822
3= [(1minus2119886
3)3]3 (see case 119896 = 3
in Figure 11) separates points representing waveforms withcoefficients satisfying (151) from points representing wave-forms with coefficients satisfying (152) Waveforms describedby (151) for 1205822
3lt [(1 minus 2119886
3)3]3 have exactly one zero at
1205910= 0 Waveforms described by (151) and (152) for 1205822
3= [(1 minus
21198863)3]3 also have zero at 120591
0= 0 These waveforms as a rule
have exactly two zeros However there are two exceptionsone related to the maximally flat nonnegative waveform withcoefficients 119886
1= minus98 119886
3= 18 and 119887
1= 1198873= 0 which
has only one zero and the other related to the waveform withcoefficients 119886
1= 0 119886
3= minus1 and 119887
1= 1198873= 0 which has three
zerosWaveforms described by (152) for [(1minus21198863)3]3
lt 1205822
3lt
1 have two zeros Waveforms with 1205823= 1 have only third
harmonic (fundamental harmonic is zero)Plot of contours of maximal absolute value of coefficient
1198861 1198861le 0 for prescribed coefficients 119886
3and 1198873is presented
in Figure 12 According to Remark 25 the waveform withmaximum absolute value of 119886
1le 0 is fully described with
the following coefficients 1198861
= minus2radic3 1198863
= radic39 and1198871= 1198873= 0 This waveform has two zeros at plusmn1205876
Two examples of nonnegative waveforms for 119896 = 3
and maximal absolute value of coefficient 1198861 1198861le 0 with
prescribed coefficients 1198863and 1198873are presented in Figure 13
One waveform corresponds to the case 12058223lt [(1 minus 2119886
3)3]3
(solid line) and the other to the case 12058223gt [(1 minus 2119886
3)3]3
(dashed line)Thewaveform represented by solid line has onezero and its coefficients are 119886
3= minus01 119887
3= 01 119886
1= minus09
and 1198871= minus03 Dashed line corresponds to the waveform
having two zeros with coefficients 1198863= minus01 119887
3= 03 119886
1=
minus08844 and 1198871= minus06460 (case 1205822
3gt [(1 minus 2119886
3)3]3)
6 Nonnegative Cosine Waveforms withat Least One Zero
Nonnegative cosine waveforms have proved to be of impor-tance for waveform modelling in PA design (eg see [10])In this section we consider nonnegative cosine waveforms
1
05
0
minus05
minus1
minus1 minus05 0 05 1
Coefficient a3
Coe
ffici
entb
3
02
04
06
08
10
11
Figure 12 Contours ofmaximal absolute value of coefficient 1198861 1198861le
0 as a function of 1198863and 1198873
2
1
0
0 1 2 3 4
Angle 120591120587
Wav
efor
ms
a3 = minus01 b3 = 01
a3 = minus01 b3 = 03
Figure 13 Nonnegative waveforms for 119896 = 3 and maximal absolutevalue of 119886
1 1198861le 0 with prescribed coefficients 119886
3and 1198873
containing fundamental and 119896th harmonic with at least onezero
Cosine waveform with dc component fundamental and119896th harmonic can be obtained from (35) by setting 119887
1= 119887119896=
0 that is
119879119896(120591) = 1 + 119886
1cos 120591 + 119886
119896cos 119896120591 (153)
In Section 61 we provide general description of non-negative cosine waveforms of type (153) with at least onezero We show that nonnegative cosine waveforms with atleast one zero coincide with nonnegative cosine waveformswith maximal absolute value of coefficient 119886
1for prescribed
coefficient 119886119896 In Section 62 we illustrate results of Section 61
for particular case 119896 = 3
61 Nonnegative Cosine Waveforms with at Least One Zerofor 119896 ge 2 Amplitudes of fundamental and 119896th harmonic
Mathematical Problems in Engineering 19
of cosine waveform of type (153) are 1205821= |1198861| and 120582
119896=
|119886119896| respectively According to (42) for nonnegative cosine
waveforms of type (153) the following relation holds
minus1 le 119886119896le 1 (154)
This explains why 119896th harmonic coefficient 119886119896in Proposi-
tion 26 goes through interval [minus1 1]Waveforms (153) with 119886
1ge 0 can be obtained from
waveforms with 1198861le 0 by shifting by 120587 and therefore with-
out loss of generality we can assume that 1198861le 0
Proposition 26 Each nonnegative cosine waveform of type(153) with 119886
1le 0 and at least one zero can be represented as
119879119896(120591) = [1 minus cos 120591] [1 minus (119896 minus 1) 119886
119896minus 2119886119896
119896minus1
sum
119899=1
(119896 minus 119899) cos 119899120591]
(155)
if minus1 le 119886119896le 1(119896
2
minus 1) or
119879119896(120591) = 119886
119896[1 minus cos (120591 minus 120591
0)] [1 minus cos (120591 + 120591
0)]
sdot [1198880+ 2
119896minus2
sum
119899=1
119888119899cos 119899120591]
(156)
where
119888119899=sin ((119896 minus 119899) 120591
0) cos 120591
0minus (119896 minus 119899) cos ((119896 minus 119899) 120591
0) sin 120591
0
sin31205910
(157)
119886119896=
sin 1205910
119896 sin (1198961205910) cos 120591
0minus cos (119896120591
0) sin 120591
0
(158)
100381610038161003816100381612059101003816100381610038161003816 le
120587
119896 (159)
if 1(1198962 minus 1) le 119886119896le 1
Remark 27 Identity [1minuscos(120591minus1205910)][1minuscos(120591+120591
0)] = [cos 120591
0minus
cos 120591]2 implies that (156) can be rewritten as
119879119896(120591) = 119886
119896[cos 120591
0minus cos 120591]2 [119888
0+ 2
119896minus2
sum
119899=1
119888119899cos 119899120591] (160)
Furthermore substitution of (157) into (160) leads to
119879119896(120591) = 119886
119896[cos 120591
0minus cos 120591]
sdot [(119896 minus 1) sin 119896120591
0
sin 1205910
minus 2
119896minus1
sum
119899=1
sin ((119896 minus 119899) 1205910)
sin 1205910
cos 119899120591]
(161)
Remark 28 All nonnegative cosine waveforms of type (153)with at least one zero and 119886
1le 0 except one of them can be
represented either in form (155) or form (156)This exceptionis maximally flat cosine waveform with 119886
1lt 0 which can be
obtained from (155) for 119886119896= 1(119896
2
minus 1) or from (156) for 1205910=
0 Maximally flat cosine waveform with 1198861lt 0 can also be
obtained from (70) by setting 1205910= 0 Furthermore setting
1205910= 0 in (71) leads to maximally flat cosine waveforms for
119896 le 4 and 1198861lt 0
Remark 29 Nonnegative cosine waveform of type (155) with1198861lt 0 and minus1 lt 119886
119896le 1(119896
2
minus 1) has exactly one zero at120591 = 0 Nonnegative cosine waveform described by (156) with1198861lt 0 and 1(1198962 minus 1) lt 119886
119896lt 1 has two zeros at plusmn120591
0 where
0 lt |1205910| lt 120587119896 For 119886
119896= ∓1 nonnegative cosine waveform
of type (153) reduces to 119879119896(120591) = 1 ∓ cos 119896120591 (clearly these two
waveforms both have 119896 zeros)
Remark 30 Transformation of (155) into an additive formleads to the following relation
1198861= minus1 minus 119886
119896 (162)
where minus1 le 119886119896le 1(119896
2
minus1) Similarly transformation of (156)leads to the following relation
1198861= minus119886119896
119896 sin 1198961205910
sin 1205910
(163)
where 119886119896is given by (158) 1(1198962minus1) le 119886
119896le 1 and |120591
0| le 120587119896
Notice that coefficients of maximally flat cosine waveformnamely 119886
119896= 1(119896
2
minus1) and 1198861= minus1198962
(1198962
minus1) satisfy relation(162) They also satisfy relation (163) for 120591
0= 0
Remark 31 Nonnegative cosine waveforms of type (153) withat least one zero coincide with nonnegative cosine waveformswith maximal absolute value of coefficient 119886
1for prescribed
coefficient 119886119896
In proving that Remark 31 holds notice that expression(155) can be obtained from (121) by setting 119887
119896= 0 Fur-
thermore if 119886119896
ge 0 then 120582119896
= 119886119896 which together
with 119887119896
= 0 and (118) implies 120575 = 0 In this case119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) le 1 + 119886
119896becomes 1198962119886
119896le 1 +
119886119896 On the other hand if 119886
119896lt 0 then 120582
119896= minus119886
119896 which
together with 119887119896= 0 and (118) implies |120575| = 120587 In this case
119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) le 1 + 119886
119896becomes 0 le 1 + 119886
119896
Therefore every nonnegative cosine waveform of type (155)has maximal absolute value of coefficient 119886
1for prescribed
coefficient 119886119896 when minus1 le 119886
119896le 1(119896
2
minus 1)Let us now show that expression (156) can be obtained
from (122) by setting 119887119896= 0 and 119886
119896gt 0 For waveforms
of type (122) according to (118) 119887119896= 0 and 119886
119896gt 0 imply
120575 = 0 and 120582119896= 119886119896 Substitution of 120582
119896= 119886119896and 120575 = 0 into
119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) ge 1+119886
119896leads to 119886
119896ge 1(119896
2
minus1)Furthermore substitution of 120575 = 0 into (145) yields 120591
0= 120585119896
Insertion of 120582119896= 119886119896 120575 = 0 and 120591
0= 120585119896 into (122) leads
to (156) Therefore every nonnegative cosine waveform oftype (156) has maximal absolute value of coefficient 119886
1for
prescribed coefficient 119886119896 when 1(1198962 minus 1) le 119886
119896le 1
Proof of Proposition 26 Let us start with nonnegative cosinewaveform of type (153) with 120582
119896= |119886119896| = 1 According to
Remark 7 120582119896= |119886119896| = 1 implies that 120582
1= |1198861| = 0
Substitution of 119886119896
= minus1 into (155) and using (A2) (seeAppendices) lead to 119879
119896(120591) = 1 minus cos 119896120591 Consequently (155)
holds for 119886119896= minus1 On the other hand substitution of 119886
119896= 1
into (158) yields |1205910| = 120587119896 Furthermore substitution of
119886119896= 1 and 120591
0= 120587119896 (or 120591
0= minus120587119896) into (156) along
with performing all multiplications and using (A2) leads to
20 Mathematical Problems in Engineering
119879119896(120591) = 1 + cos 119896120591 Consequently (156)ndash(158) hold for 119886
119896= 1
and |1205910| = 120587119896
It is easy to see that 120582119896= |119886119896| lt 1 and 119879
119896(1205910) = 0 for some
1205910imply 120582
1= |1198861| = 0 Therefore in what follows we assume
that |119886119896| = 1 and 119886
1lt 0
Cosine waveforms are even functions of 120591 Therefore ifnonnegative cosine waveform has exactly one zero it has to beeither at 0 or at 120587 On the other hand if nonnegative cosinewaveform with 119886
1= 0 has exactly two zeros then these zeros
are placed at plusmn1205910 such that 120591
0is neither 0 nor 120587
In order to prove that (155) holds for minus1 lt 119886119896le 1(1minus119896
2
)let us start by referring to the description (38) of nonnegativewaveformswith at least one zero As wementioned earlier fornonnegative cosine waveformwith exactly one zero (denotedby 1205910) it is either 120591
0= 0 or 120591
0= 120587 Therefore in both cases
sin 1205910= 0 Substitution of sin 120591
0= 0 into (43) together with
1198861
= 0 and 120582119896= |119886119896| lt 1 leads to
1205910= 0 (164)
Clearly 1205910= 0 119887
1= 0 and 119887
119896= 0 according to (44) and (46)
imply 120582119896sin 120585 = 0 Since 120582
119896= |119886119896| it follows that |119886
119896| sin 120585 = 0
also holds which further implies 120582119896= 119886119896= 0 or sin 120585 =
0 In the case when 120582119896= 119886119896= 0 from (164) and (43) we
obtain 1198861= minus1 which further implies that 119879
119896(120591) = 1 minus cos 120591
Consequently (155) holds for 119886119896= 0 In the case when sin 120585 =
0 from (164) and (45) we obtain 119886119896= 120582119896if 120585 = 0 or 119886
119896= minus120582119896
if 120585 = 120587 Relations 119886119896= 120582119896and 120585 = 0 according to (40) imply
that 0 le 119886119896le 1(1 minus 119896
2
) Substitution of 120585 = 0 120582119896= 119886119896 and
(164) into (38) leads to (155) which proves that (155) holdsfor 0 le 119886
119896le 1(1 minus 119896
2
) On the other hand relations 119886119896=
minus120582119896and 120585 = 120587 according to (40) imply that minus1 lt 119886
119896le 0
Substitution of 120585 = 120587 120582119896= minus119886119896 and (164) into (38) also leads
to (155) which proves that (155) also holds for minus1 lt 119886119896le 0
Consequently (155) holds for minus1 lt 119886119896le 1(1 minus 119896
2
)In what follows we first prove that (156)-(157) hold for
1(1 minus 1198962
) lt 119886119896lt 1 For this purpose let us start with non-
negative waveforms with two zeros described by (66) As wementioned before nonnegative cosine waveforms with twozeros have zeros at 120591
0and minus120591
0 such that 120591
0= 0 and 120591
0= 120587
Relations 1198861
lt 0 and 1198871
= 0 according to (84) implycos(1205910minus 120585119896) = 1 and therefore
120585
119896= 1205910 (165)
From 120585119896 = 1205910and 0 lt |120585| lt 120587 it follows that 0 lt |120591
0| lt
120587119896 Insertion of 120585119896 = 1205910into (45) yields 119886
119896= 120582119896 Relations
119886119896= 120582119896and (82) imply that 1(1 minus 1198962) lt 119886
119896lt 1 Substitution
of 120582119896= 119886119896and 120585119896 = 120591
0into (66)ndash(68) leads to (156)ndash(158)
which proves that (156)ndash(158) hold for 1(1 minus 1198962
) lt 119886119896lt 1
and 0 lt |1205910| lt 120587119896
Finally substitution of 119886119896= 1(1 minus 119896
2
) and 1205910= 0 into
(161) leads to
119879119896(120591) =
[1 minus cos 120591](1 minus 1198962)
[119896 (119896 minus 1) minus 2
119896minus1
sum
119899=1
(119896 minus 119899) cos 119899120591]
(166)
Waveform (166) coincides with waveform (155) for 119886119896
=
1(1 minus 1198962
) which in turn proves that (156) holds for 119886119896=
1(1 minus 1198962
) and 1205910= 0 This completes the proof
62 Nonnegative CosineWaveforms with at Least One Zero for119896 = 3 In this subsection we consider nonnegative cosinewaveforms with at least one zero for 119896 = 3 (for case 119896 = 2
see [12])Cosine waveform with fundamental and third harmonic
reads
1198793(120591) = 1 + 119886
1cos 120591 + 119886
3cos 3120591 (167)
For 1198861le 0 and minus1 le 119886
3le 18 according to (155) non-
negative cosine waveform of type (167) with at least one zerocan be expressed as
1198793(120591) = (1 minus cos 120591) [1 minus 2119886
3(1 + 2 cos 120591 + cos 2120591)] (168)
From 1198793(120591 + 120587) = 2 minus 119879
3(120591) it immediately follows that for
1198861ge 0 and minus18 le 119886
3le 1 119879
3(120591) can be expressed as
1198793(120591) = (1 + cos 120591) [1 + 2119886
3(1 minus 2 cos 120591 + cos 2120591)] (169)
For 1198861le 0 and 18 le 119886
3le 1 from (158) it follows that 119886
3=
[8cos31205910]minus1 This relation along with (160) and (157) further
implies that 1198793(120591) can be expressed as
1198793(120591) =
[cos 1205910minus cos 120591]2 [2 cos 120591
0+ cos 120591]
2cos31205910
(170)
providing that |1205910| le 1205873 From 119879
3(120591 + 120587) = 2 minus 119879
3(120591) it
follows that (170) also holds for 1198861ge 0 and minus1 le 119886
3le minus18
providing that 1205910isin [21205873 41205873]
Maximally flat nonnegative cosinewaveformof type (167)with 119886
1lt 0 (minimum at 120591
0= 0) reads 119879
3(120591) = [1 minus
cos 120591]2[1 + (12) cos 120591] Dually maximally flat nonnegativecosine waveform with 119886
1gt 0 (minimum at 120591
0= 120587) reads
1198793(120591) = [1 + cos 120591]2[1 minus (12) cos 120591]In what follows we provide relations between coefficients
1198861and 1198863of nonnegative cosine waveforms of type (167) with
at least one zeroFor 1198861le 0 conversion of (168) into an additive form
immediately leads to the following relation
1198861= minus1 minus 119886
3for minus 1 le 119886
3le1
8 (171)
Conversion of (170) into an additive form leads to 1198861
=
minus31198863(2 cos 2120591
0+ 1) which can be also expressed as 119886
1=
minus31198863(4cos2120591
0minus 1) For 119886
1le 0 relations |120591
0| le 1205873 119886
1=
minus31198863(4cos2120591
0minus 1) and 119886
3= [8cos3120591
0]minus1 lead to
1198861= minus3 [ 3radic119886
3minus 1198863] for 1
8le 1198863le 1 (172)
Similarly for 1198861ge 0 conversion of (169) into an additive form
leads to the following relation
1198861= 1 minus 119886
3for minus
1
8le 1198863le 1 (173)
Mathematical Problems in Engineering 21
1
15
15
05
0
minus05
minus15minus15
minus1
minus1 minus05 0 05 1
Coefficient a1
Coe
ffici
enta
3 a1= minus
3(3radica3minus a3)
a1 =
minus1 minus a3
a1 =
1 minusa3
a1 =3(3radic|a3
| +a3)
Figure 14 Parameter space of cosine waveforms for 119896 = 3
For waveform of type (170) with 1198861ge 0 relations 120591
0isin [21205873
41205873] 1198861= minus3119886
3(4cos2120591
0minus 1) and 119886
3= [8cos3120591
0]minus1 lead to
1198861= 3 [
3radic10038161003816100381610038161198863
1003816100381610038161003816 + 1198863] for minus 1 le 1198863le minus
1
8 (174)
Every cosine waveform of type (167) corresponds to apair of real numbers (119886
1 1198863) and vice versa Points (119886
1 1198863)
in grey area in Figure 14 correspond to nonnegative cosinewaveforms for 119896 = 3 The points at the boundary of grey areacorrespond to nonnegative cosinewaveformswith at least onezero A number of shapes of nonnegative cosine waveformswith 119896 = 3 and at least one zero plotted on interval [minus120587 120587]are also presented in Figure 14 The boundary of grey area inFigure 14 consists of four line segments described by relations(171)ndash(174) The common point of line segments (172) and(173) is cusp point with coordinates 119886
1= 0 and 119886
3= 1
Another cusp point with coordinates 1198861= 0 and 119886
3= minus1
is the common point of line segments (171) and (174) Thecommon point of line segments (171)-(172) has coordinates(minus98 18) and common point of line segments (173)-(174)has coordinates (98 minus18) These points are representedby white circle dots and they correspond to maximallyflat cosine waveforms (eg see [21]) White triangle dotswith coordinates (2radic3 minusradic39) and (minus2radic3radic39) refer tothe nonnegative cosine waveforms with maximum value ofamplitude of fundamental harmonic
7 Four Case Studies of Usage of NonnegativeWaveforms in PA Efficiency Analysis
In this section we provide four case studies of usage ofdescription of nonnegative waveforms with fundamental and119896th harmonic in PA efficiency analysis In first two casestudies to be presented in Section 71 voltage is nonnegativewaveform with fundamental and second harmonic with atleast one zero In remaining two case studies to be consideredin Section 72 voltage waveform contains fundamental andthird harmonic
i
v Load
Vdc
Idc
Lch
vL
iL
Cb
+ +in
Figure 15 Generic PA circuit diagram
Let us consider generic PA circuit diagram as shown inFigure 15 We assume here that voltage and current wave-forms at the transistor output are
V (120579) = 1 + 1198861V cos 120579 + 1198871V sin 120579 + 119886119896V cos 119896120579 + 119887119896V sin 119896120579
119894 (120579) = 1 + 1198861119894cos 120579 +
infin
sum
119899=2
119886119899119894cos 119899120579
(175)
where 120579 stands for 120596119905 Both waveforms are normalized in thesense that dc components of voltage and current are 119881dc =
1 and 119868dc = 1 respectively Under assumption that blockingcapacitor 119862
119887behaves as short-circuit at the fundamental and
higher harmonics current and voltage waveforms at the loadare
V119871(120579) = 119886
1V cos 120579 + 1198871V sin 120579 + 119886119896V cos 119896120579 + 119887119896V sin 119896120579
119894119871(120579) = minus119886
1119894cos 120579 minus
infin
sum
119899=2
119886119899119894cos 119899120579
(176)
In terms of coefficients of voltage and current waveforms theload impedance at fundamental harmonic is 119911
1= minus(119886
1V minus
1198951198871V)1198861119894 whereas load impedance at 119896th harmonic is 119911
119896=
minus(119886119896Vminus119895119887119896V)119886119896119894 All other harmonics are short-circuited (119911
119899=
0 for 119899 = 1 and 119899 = 119896) Time average output power of PA (egsee [10]) with waveform pair (175) at fundamental frequencycan be expressed as
1198751= minus
11988611198941198861V
2 (177)
For normalized waveforms (175) with 119881dc = 1 and 119868dc = 1dc power is 119875dc = 1 Consequently PA efficiency 120578 = 119875
1119875dc
(eg see [10 26]) is equal to
120578 = minus11988611198941198861V
2 (178)
Thus time average output power 1198751of PA with pair of nor-
malized waveform (175) is equal to efficiency (178)Power utilization factor (PUF) is defined [26] as ldquothe
ratio of power delivered in a given situation to the power
22 Mathematical Problems in Engineering
delivered by the same device with the same supply voltagein Class A moderdquo Since the output power in class-A modeis 1198751class-A = max[V(120579)] sdotmax[119894(120579)]8 (eg see [9]) it follows
that power utilization factor PUF = 11987511198751class-A for PA with
pair of normalized waveforms (175) can be expressed as
PUF =8120578
max [V (120579)] sdotmax [119894 (120579)] (179)
71 NonnegativeWaveforms for 119896 = 2 in PAEfficiencyAnalysisIn this subsection we provide two case studies of usage ofdescription of nonnegative waveforms with fundamental andsecond harmonic (119896 = 2) in PA efficiency analysis For moreexamples of usage of descriptions of nonnegative waveformswith fundamental and second harmonic in PA efficiencyanalysis see [12]
Case Study 71 In this case study we consider efficiency of PAfor given second harmonic impedance providing that voltageis nonnegative waveform with fundamental and second har-monic and current is ldquohalf-sinerdquo waveform frequently used inefficiency analysis of classical PA operation (eg see [10])
Standard model of current waveform for classical PAoperation has the form (eg see [10 26])
119894119863(120579) =
119868119863[cos 120579 minus cos(120572
2)] |120579| le
120572
2
0120572
2le |120579| le 120587
(180)
where 120572 is conduction angle and 119868119863
gt 0 Since 119894119863(120579) is
even function it immediately follows that its Fourier seriescontains only dc component and cosine terms
119894119863(120579) = 119868dc +
infin
sum
119899=1
119868119899cos 119899120579 (181)
The dc component of the waveform (180) is
119868dc =119868119863120572
2120587[sinc(120572
2) minus cos(120572
2)] (182)
where sinc119909 = (sin119909)119909 The coefficient of the fundamentalharmonic component reads
1198681=119868119863120572
2120587(1 minus sinc120572) (183)
and the coefficient of 119899th harmonic component can bewrittenin the form
119868119899=119868119863
119899120587[sin ((119899 minus 1) 1205722)
(119899 minus 1)minussin ((119899 + 1) 1205722)
(119899 + 1)] 119899 ge 2
(184)
For ldquohalf-sinerdquo current waveform conduction angle is equalto 120587 (class-B conduction angle) According to (182) thisfurther implies that 119868dc = 119868
119863120587 To obtain normalized form of
waveform (180) we set 119868dc = 1which implies that 119868119863= 120587 Fur-
thermore substitution of 120572 = 120587 and 119868119863= 120587 in (180) leads to
119894 (120579) =
120587 cos 120579 |120579| lt120587
2
0120587
2lt |120579| le 120587
(185)
Similarly substitution of 119868119863= 120587 and 120572 = 120587 into (183) and
(184) leads to the coefficients of waveform (185) Coefficientsof fundamental and second harmonic respectively are
1198861119894=120587
2 119886
2119894=2
3 (186)
On the other hand voltage waveform of type (35) for 119896 =2 reads
V (120579) = 1 + 1198861V cos 120579 + 1198871V sin 120579 + 1198862V cos 2120579 + 1198872V sin 2120579
(187)
This waveform contains only fundamental and second har-monic and therefore all harmonics of order higher thantwo are short-circuited (119911
119899= 0 for 119899 gt 2) For current
voltage pair (185) and (187) load impedance at fundamentalharmonic is 119911
1= minus(119886
1V minus 1198951198871V)1198861119894 whereas load impedance
at second harmonic is 1199112= minus(119886
2Vminus1198951198872V)1198862119894 According to ourassumption the load is passive and therefore Re119911
1 gt 0 and
Re1199112 ge 0 which further imply 119886
11198941198861V lt 0 and 119886
21198941198862V le 0
respectivelyIt is easy to see that problem of findingmaximal efficiency
of PA with current-voltage pair (185) and (187) for prescribedsecond harmonic impedance can be reduced to the problemof finding voltage waveform of type (187) with maximal coef-ficient |119886
1V| for prescribed coefficients of second harmonic(see Section 5)
The following algorithm (analogous to Algorithm 22presented in [12]) provides the procedure for calculation ofmaximal efficiency with current-voltage pair (185) and (187)for prescribed second harmonic impedance The definitionof function atan 2(119910 119909) which appears in the step (iii) of thefollowing algorithm is given by (105)
Algorithm 32 (i) Choose 1199112= 1199032+1198951199092such that |119911
2| le 1|119886
2119894|
(ii) calculate 1198862V minus 1198951198872V = minus119911
21198862119894and 120582
2V = radic11988622V + 1198872
2V(iii) if 2120582
2V le 1 minus 1198862V then calculate 119886
1V = minus1 minus 1198862V and
1198871V = minus2119887
2V else calculate 1205821V = radic81205822V(1 minus 1205822V) 1205790V minus 120585V2 =
(12)atan2(1198872V 1198862V) 1198861V = minus120582
1V cos(1205790V minus 120585V2) and 1198871V =
minus1205821V sin(1205790V minus 120585V2)(iv) calculate efficiency 120578 = minus119886
11198941198861V2
(v) calculate 1199111= minus(119886
1V minus 1198951198871V)1198861119894 and 1199112119899 = 1199112Re119911
1
In this case study coefficients of fundamental and secondharmonic of current waveform are given by (186) Maximalefficiency of PA associated with the waveform pair (185)and (187) as a function of normalized second harmonicimpedance 119911
2119899= 1199112Re119911
1 is presented in Figure 16(a)
As can be seen from Figure 16(a) efficiency of 078 isachieved at the edge of Smith chart where second harmonicimpedance has small resistive part Corresponding PUFcalculated according to (179) is presented in Figure 16(b)Peak efficiency 120578 = 1205874 = 07854 and peak value of PUF = 1
are attained when second harmonic is short-circuited (whichcorresponds to ideal class-B operation [10 26])
For example for second harmonic impedance 1199112= 01 minus
11989505 and currentwaveform (185) fromAlgorithm32 it followsthat 2120582
2V le 1 minus 1198862V Furthermore according to step (iii)
of above algorithm maximal efficiency of PA is attained
Mathematical Problems in Engineering 23
+j2
+j5
+j1
+j2
+j5
minusj2
minusj5
minusj1
minusj2
minusj5
infin0
078
02
05
1 2 5
075 07 065
06 05
075
07
065
06
05
120578 lt 05
(a)
099
095
091
083 075
067
0 59 051
051
099
095
091
083
075 0
67 059
+j5
+j1
+j2
+j5
minusj5
minusj1
minusj2
minusj5
infin
+j2
minusj2
0 02
05
1 2 5
120578 lt 05
(b)
Figure 16 (a) Contours of maximal efficiency of PA and (b) contours of corresponding PUF associated with the waveform pair (185) and(187) as functions of normalized second harmonic impedance 119911
2119899= 1199112Re119911
1
2
3
4
1
0
2 3 410
Wav
efor
ms
Angle 120579120587
i(120579)
(120579)
Figure 17 Waveform pair (185) and (187) that provides maximalefficiency for 119911
2= 01 minus 11989505
with voltage waveform of type (187) with coefficients 1198862V =
minus00667 1198872V = minus03333 119886
1V = minus09333 and 1198871V = 06667 (see
Figure 17) Corresponding efficiency PUF and normalizedsecond harmonic impedance are 120578 = 07330 PUF = 07572and 1199112119899= 01683 minus 11989508415 respectively
On the other hand for second harmonic impedance 1199112=
01 minus 11989508 and current waveform (185) from Algorithm 32 itfollows that 2120582
2V gt 1 minus 1198862V Then according to step (iii) of
above algorithm maximal efficiency is attained with voltagewaveform of type (187) with coefficients 119886
2V = minus00667 1198872V =
minus05333 1198861V = minus09333 and 119887
1V = 10572 (see Figure 18)Efficiency PUF andnormalized secondharmonic impedanceare 120578 = 07330 PUF = 06332 and 119911
2119899= 01683 minus 11989513465
respectively
2
3
4
1
0
2 3 410
Wav
efor
ms
Angle 120579120587
i(120579)
(120579)
Figure 18 Waveform pair (185) and (187) that provides maximalefficiency for 119911
2= 01 minus 11989508
Case Study 72 As another case study let us consider the effi-ciency of PA providing that current waveform is nonnegativecosine waveform up to third harmonic with maximum valueof amplitude of fundamental harmonic [22] (see also [8])
119894 (120579) = 1 +1 + radic5
2cos 120579 + 2radic5
5cos 2120579 + 5 minus radic5
10cos 3120579
(188)
and voltage waveform is nonnegative waveform of type(187) Load impedances at fundamental second and thirdharmonic are 119911
1= minus(119886
1Vminus1198951198871V)1198861119894 1199112 = minus(1198862Vminus1198951198872V)1198862119894 and
1199113= 0 respectively According to our assumption the load
is passive and therefore Re1199111 gt 0 and Re119911
2 ge 0 which
further imply 11988611198941198861V lt 0 and 119886
21198941198862V le 0 respectively
Because current waveform (188) contains only cosineterms and voltage waveform is the same as in previous case
24 Mathematical Problems in Engineering
08
075
07
06
06
065
065
05
05
075 0
7
120578 lt 05
+j2
+j5
+j1
+j2
+j5
minusj2
minusj5
minusj1
minusj2
minusj5
infin0
02
05
1 2 5
(a)
085
08 0
75 07 0
6
05 04
08
075
0405
060
7
120578 lt 05
+j2
+j5
+j1
+j2
+j5
minusj2
minusj5
minusj1
minusj2
minusj5
infin0 02
05
1 2 5
(b)
Figure 19 (a) Contours of maximal efficiency of PA and (b) contours of corresponding PUF associated with the waveform pair (187)-(188)as functions of normalized second harmonic impedance 119911
2119899= 1199112Re119911
1
study the procedure for calculation of maximal efficiency ofPA with waveform pair (187)-(188) is the same as presentedin Algorithm 32 In this case study the coefficients of funda-mental and second harmonic of current waveform are 119886
1119894=
(1 + radic5)2 and 1198862119894= 2radic55 respectively
Maximal efficiency of PA associated with the waveformpair (187)-(188) as a function of normalized secondharmonicimpedance 119911
2119899= 1199112Re119911
1 is presented in Figure 19(a)
Efficiency of 08 is achieved at the edge of Smith chart wheresecond harmonic impedance has small resistive part Thetheoretical upper bound 120578 = (1 + radic5)4 asymp 08090 isattained when second harmonic is short-circuitedWhen thisupper bound is reached both second and third harmonic areshort-circuited which implies that we are dealing with finiteharmonic class-C [6 8] or dually when current and voltageinterchange their roles with finite harmonic inverse class-C[6 9] Corresponding PUF calculated according to (179) ispresented in Figure 19(b) Peak value of PUF asymp 08541 isattained when second harmonic is short-circuited
For example for second harmonic impedance 1199112
=
007 minus 11989504 and current waveform (188) from Algorithm 32 itfollows that 2120582
2V le 1 minus 1198862V Furthermore according to step
(iii) of Algorithm 32 maximal efficiency of PA is attainedwith voltage waveform of type (187) with coefficients 119886
2V =
minus00626 1198872V = minus03578 119886
1V = minus09374 and 1198871V = 07155 (see
Figure 20) Corresponding efficiency PUF and normalizedsecond harmonic impedance are 120578 = 07584 PUF = 06337and 1199112119899= 01208 minus 11989506904 respectively
On the other hand for 1199112= 005 minus 11989507 and current
waveform (187) it follows that 21205822V gt 1minus119886
2VThen accordingto step (iii) of Algorithm 32 themaximal efficiency is attainedwith voltage waveform of type (187) with coefficients 119886
2V =
minus00447 1198872V = minus06261 119886
1V = minus09318 and 1198871V = 10007 (see
Figure 21) Efficiency PUF and normalized second harmonic
2
3
4
1
0
2 3 410
Wav
efor
ms
Angle 120579120587
i(120579)
(120579)
Figure 20 Waveform pair (187)-(188) that provides maximal effi-ciency for 119911
2= 007 minus 11989504
impedance are 120578 = 07538 PUF = 05314 and 1199112119899= 00868minus
11989512156 respectively
72 Nonnegative Waveforms for 119896 = 3 in PA EfficiencyAnalysis In this subsection we provide another two casestudies of usage of description of nonnegative waveforms inPA efficiency analysis this time with fundamental and thirdharmonic (119896 = 3)
Case Study 73 Let us consider current-voltage pair such thatvoltage is nonnegative waveform with fundamental and thirdharmonic
V (120579) = 1 + 1198861V cos 120579 + 1198871V sin 120579 + 1198863V cos 3120579 + 1198873V sin 3120579
(189)
Mathematical Problems in Engineering 25
2
3
4
1
0
2 3 410
Wav
efor
ms
Angle 120579120587
i(120579)
(120579)
Figure 21 Waveform pair (187)-(188) that provides maximal effi-ciency for 119911
2= 005 minus 11989507
and current is nonnegative cosine waveform given by (188)Load impedances at fundamental second and third har-monic are 119911
1= minus(119886
1V minus 1198951198871V)1198861119894 1199112 = 0 and 119911
3= minus(119886
3V minus
1198951198873V)1198863119894 respectively According to our assumption the load
is passive and therefore Re1199111 gt 0 and Re119911
3 ge 0 which
further imply 11988611198941198861V lt 0 and 119886
31198941198863V le 0
In this subsection we consider the problem of findingmaximal efficiency of PA with waveform pair (188)-(189)for given third harmonic impedance As we mentionedearlier problem of finding maximal efficiency of PA withcurrent-voltage pair (188)-(189) for prescribed third har-monic impedance can be reduced to the problem of findingvoltage waveform of type (189) withmaximal coefficient |119886
1V|for prescribed coefficients of third harmonic (see Section 52)
The following algorithm provides the procedure forcalculation of maximal efficiency with current-voltage pair(188)-(189) The definition of function atan 2(119910 119909) whichappears in step (iii) of the following algorithm is given by(105)
Algorithm 33 (i) Choose 1199113= 1199033+1198951199093such that |119911
3| le 1|119886
3119894|
(ii) calculate 1198863V minus 1198951198873V = minus119911
31198863119894and 120582
3V = radic11988623V + 1198872
3V(iii) if 271205822
3V le (1 minus 21198863V)3 then calculate 119886
1V = minus1 minus 1198863V
and 1198871V = minus3119887
3V else calculate 1205821V = 3(3radic1205823V minus 120582
3V) 1205790V minus120585V3 = (13)atan 2(119887
3V 1198863V) 1198861V = minus1205821V cos(1205790V minus 120585V3) and
1198871V = minus120582
1V sin(1205790V minus 120585V3)(iv) calculate efficiency 120578 = minus119886
11198941198861V2
(v) calculate 1199111= minus(119886
1V minus 1198951198871V)1198861119894 and 1199113119899 = 1199113Re119911
1
In this case study coefficients of fundamental and thirdharmonic of current waveform are 119886
1119894= (1 + radic5)2 and
1198863119894= (5 minus radic5)10 respectively For the waveform pair (188)-
(189) maximal efficiency of PA as a function of normalizedthird harmonic impedance 119911
3119899= 1199113Re119911
1 is presented in
Figure 22 Efficiency of 08 is reached when third harmonicimpedance has small resistive part Peak efficiency 120578 = (1 +
radic5)4 asymp 08090 is achieved when third harmonic is short-circuited
+j2
+j5
+j1
+j2
+j5
minusj2
minusj5
minusj1
minusj2
minusj5
infin0
08
08
075
075
07
07
06
05
02
05
1 2
Figure 22 Contours of maximal efficiency of PA associated withthe waveform pair (188)-(189) as a function of normalized thirdharmonic impedance 119911
3119899= 1199113Re119911
1
For the present case study in what follows we show thatpower utilization factor is proportional to efficiency For volt-age waveform of type (189) it is easy to see that V(120579 + 120587) =
2minusV(120579) holdsThis relation along with the fact that waveformV(120579) that provides maximal efficiency has at least one zeroimplies that max[V(120579)] = 2 On the other hand currentwaveform (188) is cosine waveform with positive coefficientsand therefore max[119894(120579)] = 119894(0) = 2 + 4radic5 Consequentlyaccording to (179) the following relation holds
PUFCase study 73 = 2 (5 minus 2radic5) 120578 = 10557120578 (190)
Clearly the ratio PUF120578 is constant and therefore in this casestudy PUF can be easily calculated from the correspondingefficiency Accordingly peak efficiency and peak value ofPUFCase study 73 = 3radic52 minus 52 = 08541 are attained forthe same voltage waveform (when third harmonic is short-circuited)
In the first example current waveform (188) and 1199113=
02 minus 11989505 imply that 2712058223V le (1 minus 2119886
3V)3 Then according
to Algorithm 33 the voltage waveform of type (189) thatprovides maximal efficiency has the following coefficients1198863V = minus00553 119887
3V = minus01382 1198861V = minus09447 and 119887
1V =
04146 (see Figure 23) Efficiency PUF and normalized thirdharmonic impedance are 120578 = 07643 PUF = 08069 and1199113119899= 03425 minus 11989508564 respectivelyIn the second example current waveform (188) and 119911
3=
01 minus 11989511 imply that 2712058223V gt (1 minus 2119886
3V)3 Then according
to Algorithm 33 the voltage waveform of type (189) thatprovides maximal efficiency has the following coefficients1198863V = minus00276 119887
3V = minus03040 1198861V = minus09391 and 119887
1V =
05807 (see Figure 24) Efficiency PUF and normalized thirdharmonic impedance are 120578 = 07598 PUF = 08021 and1199113119899= 01723 minus 11989518952 respectively
26 Mathematical Problems in Engineering
2
3
4
1
0
2 3 410
Wav
efor
ms
Angle 120579120587
i(120579)
(120579)
Figure 23 Waveform pair (188)-(189) that provides maximal effi-ciency for 119911
3= 02 minus 11989505
2
3
4
1
0
2 3 410
Wav
efor
ms
Angle 120579120587
i(120579)
(120579)
Figure 24 Waveform pair (188)-(189) that provides maximal effi-ciency for 119911
3= 01 minus 11989511
Case Study 74 In this case study let us consider current-voltage pair where current is normalized waveform of type(180) with conduction angle 120572 = 115120587 (207∘) and voltageis nonnegative waveform of type (189) Substitution of 120572 =
115120587 and 119868dc = 1 into (182) leads to 119868119863
= 22535Furthermore substitution of 120572 = 115120587 and 119868
119863= 22535 into
(180) leads to
119894 (120579)
=
22535 [cos 120579 minus cos(1151205872
)] |120579| le115120587
2
0115120587
2le |120579| le 120587
(191)
Similarly substitution of 120572 = 115120587 and 119868119863
= 22535 into(183) and (184) for 119899 = 3 yields coefficients of fundamentaland third harmonic of waveform (191)
1198861119894= 14586 119886
3119894= minus01026 (192)
Because current waveform (191) contains only cosineterms and voltage waveform is the same as in previous case
+j2
+j5
+j1
+j2
+j5
minusj2
minusj5
minusj1
minusj2
minusj5
infin0 02
05
1 2
074
076
08
084
082
078
076
074
Figure 25 Contours of maximal efficiency of PA associated withthe waveform pair (189) and (191) as a function of normalized thirdharmonic impedance 119911
3119899= 1199113Re119911
1
study the procedure for calculation of maximal efficiencyof PA with waveform pair (189)ndash(191) is the same as thatpresented in Algorithm 33 In this case study the coefficientsof fundamental and third harmonic of current waveform aregiven by (192)
For the waveform pair (189) and (191) maximal efficiencyof PA as a function of normalized third harmonic impedance1199113119899= 1199113Re119911
1 is presented in Figure 25 Efficiency of 084
is obtained in vicinity of 1199113119899= 23685 (corresponding to 119911
3=
18750) Peak efficiency 120578 asymp 08421 is achieved for voltagewaveform of type (189) with coefficients 119886
1V = minus2radic3 1198863V =
radic39 and 1198871V = 1198873V = 0
In the course of finding power utilization factor noticethat currentwaveformof type (191) attains itsmaximumvaluefor 120579 = 0 Insertion of max[119894(120579)] = 119894(0) = 278 andmax[V(120579)] = 2 for voltage waveform of type (189) into (179)leads to
PUFCase study 74 = 1439120578 (193)
Again the ratio PUF120578 is constant and PUF can be easilycalculated from the corresponding efficiency Accordinglypeak value of PUFCase study 74 asymp 12118 and peak efficiencyare attained for the same voltage waveform
In the first example current waveform (191) and 1199113=
1 minus 11989502 imply that 2712058223V le (1 minus 2119886
3V)3 Then according to
Algorithm 33 voltage waveform of type (189) which providesmaximal efficiency has coefficients 119886
3V = 01026 1198873V =
00205 1198861V = minus11026 and 119887
1V = minus00616 (see Figure 26)Efficiency PUF and normalized third harmonic impedanceare 120578 = 08042 PUF = 11572 and 119911
3119899= 13228 minus 11989502646
respectivelyIn second example current waveform (191) and 119911
3=
15 minus 11989512 imply that 2712058223V gt (1 minus 2119886
3V)3 Then according to
Algorithm 33 voltage waveform of type (189) which providesmaximal efficiency has coefficients 119886
3V = 01540 1198873V =
Mathematical Problems in Engineering 27
2
3
4
1
0
2 3 410
Wav
efor
ms
Angle 120579120587
i(120579)
(120579)
Figure 26 Waveform pair (189) and (191) that provides maximalefficiency for 119911
3= 1 minus 11989502
2
3
4
1
0
2 3 410
Wav
efor
ms
Angle 120579120587
i(120579)
(120579)
Figure 27 Waveform pair (189) and (191) that provides maximalefficiency for 119911
3= 15 minus 11989512
01232 1198861V = minus11255 and 119887
1V = minus02575 (see Figure 27)Efficiency PUF and normalized third harmonic impedanceare 120578 = 08208 PUF = 11812 and 119911
3119899= 19439 minus 11989515552
respectively
8 Conclusion
In this paper we consider a problem of finding general de-scriptions of various classes of nonnegative waveforms withfundamental and 119896th harmonic These classes include non-negative waveforms with at least one zero nonnegative wave-forms with maximal amplitude of fundamental harmonic forprescribed amplitude of 119896th harmonic nonnegative wave-forms withmaximal coefficient of cosine part of fundamentalharmonic for prescribed coefficients of 119896th harmonic andnonnegative cosine waveforms with at least one zero Mainresults are stated in six propositions (Propositions 1 6 918 22 and 26) four corollaries (Corollaries 2ndash5) twentyremarks and three algorithms Four case studies of usageof closed form descriptions of nonnegative waveforms in PAefficiency analysis are considered in detail in Section 7
Appendices
Here we provide a list of finite sums of trigonometric func-tions used in this paper (Appendix A) and brief account ofthe Chebyshev polynomials (Appendix B)
A List of Some Finite Sums ofTrigonometric Functions
Dirichlet kernel (eg see [27]) is as follows
119863119896minus1
(120591) = 1 + 2
119896minus1
sum
119899=1
cos 119899120591 = sin ((2119896 minus 1) 1205912)sin (1205912)
(A1)
Fejer kernel (eg see [27]) can be expressed in the fol-lowing equivalent forms
119865119896minus1
(120591) =1
119896
119896minus1
sum
119899=0
119863119899(120591) = 1 +
2
119896
119896minus1
sum
119899=1
(119896 minus 119899) cos 119899120591
=(1 minus cos 119896120591)119896 (1 minus cos 120591)
(A2)
Lagrangersquos trigonometric identity (eg see [28]) is as fol-lows
1198781(120591) =
119896minus1
sum
119899=1
sin 119899120591 = sin (1198961205912) sin ((119896 minus 1) 1205912)sin (1205912)
(A3)
In what follows we show that the following three trigono-metric identities also hold
2
119896minus1
sum
119899=1
(119896 minus 119899) sin 119899120591 = 119896 sin 120591 minus sin 1198961205911 minus cos 120591
(A4)
119896minus1
sum
119899=1
cos (119896 minus 2119899) 120591 = sin (119896 minus 1) 120591sin 120591
(A5)
119896minus1
sum
119899=1
119899 (119896 minus 119899) cos (119896 minus 2119899) 120591
=sin (119896120591) cos 120591 minus 119896 cos (119896120591) sin 120591
2sin3120591
(A6)
Denote 1198782(120591) = 2sum
119896minus1
119899=1(119896 minus 119899) sin 119899120591 119878
3(120591) = sum
119896minus1
119899=1cos(119896 minus
2119899)120591 and 1198784(120591) = sum
119896minus1
119899=1119899(119896 minus 119899) cos(119896 minus 2119899)120591
Notice that 1198782(120591) = 2119896119878
1(120591) + 119889119863
119896minus1(120591)119889120591 which
immediately leads to (A4)
28 Mathematical Problems in Engineering
Identity (A5) can be obtained as follows
sin (119896 minus 1) 120591sin 120591
=119890119895(119896minus1)120591
minus 119890minus119895(119896minus1)120591
119890119895120591 minus 119890minus119895120591
= 119890119895119896120591
119890minus2119895120591
minus 119890minus2119895(119896minus1)120591
1 minus 119890minus2119895120591
= 119890119895119896120591
119896minus1
sum
119899=1
119890minus2119895119899120591
=
119896minus1
sum
119899=1
119890119895(119896minus2119899)120591
=
119896minus1
sum
119899=1
cos (119896 minus 2119899) 120591
(A7)
From 4119899(119896 minus 119899) = 1198962
minus (119896 minus 2119899)2 it follows that 4119878
4(120591) =
1198962
1198783(120591) + 119889
2
1198783(120591)119889120591
2 which leads to (A6)
B The Chebyshev Polynomials
The Chebyshev polynomials of the first kind 119881119899(119909) can be
defined by the following relation (eg see [29])
119881119899(119909) = cos 119899120591 when 119909 = cos 120591 (B1)
The Chebyshev polynomials of the second kind 119880119899(119909) can be
defined by the following relation (eg see [29])
119880119899(119909) =
sin (119899 + 1) 120591sin 120591
when 119909 = cos 120591 (B2)
The Chebyshev polynomials satisfy the following recur-rence relations (eg see [29])
1198810(119909) = 1 119881
1(119909) = 119909
119881119899+1
(119909) = 2119909119881119899(119909) minus 119881
119899minus1(119909)
1198800(119909) = 1 119880
1(119909) = 2119909
119880119899+1
(119909) = 2119909119880119899(119909) minus 119880
119899minus1(119909)
(B3)
The first few Chebyshev polynomials of the first and secondkind are 119881
2(119909) = 2119909
2
minus 1 1198813(119909) = 4119909
3
minus 3119909 1198814(119909) = 8119909
4
minus
81199092
+ 1 1198802(119909) = 4119909
2
minus 1 1198803(119909) = 8119909
3
minus 4119909 and 1198804(119909) =
161199094
minus 121199092
+ 1
Conflict of Interests
The authors declare that there is no conflict of interestsregarding the publication of this paper
Acknowledgment
This work is supported by the SerbianMinistry of EducationScience and Technology Development as a part of ProjectTP32016
References
[1] V I Arnolrsquod V S Afrajmovich Y S Ilrsquoyashenko and L PShilrsquonikov Dynamical Systems V Bifurcation Theory and Catas-trophe Theory Springer Berlin Germany 1994
[2] E Polak ldquoOn the mathematical foundations of nondifferen-tiable optimization in engineering designrdquo SIAM Review vol29 no 1 pp 21ndash89 1987
[3] N S Fuzik ldquoBiharmonic modes of a tuned RF power amplifierrdquoRadiotehnika vol 25 no 7 pp 62ndash71 1970 (Russian)
[4] P Colantonio F Giannini G Leuzzi and E Limiti ldquoClass Gapproach for low-voltage high-efficiency PA designrdquo Interna-tional Journal of RF and Microwave Computer-Aided Engineer-ing vol 10 no 6 pp 366ndash378 2000
[5] F H Raab ldquoMaximum efficiency and output of class-F poweramplifiersrdquo IEEE Transactions on Microwave Theory and Tech-niques vol 49 no 6 pp 1162ndash1166 2001
[6] FH Raab ldquoClass-E class-C and class-F power amplifiers basedupon a finite number of harmonicsrdquo IEEE Transactions onMicrowaveTheory and Techniques vol 49 no 8 pp 1462ndash14682001
[7] J D Rhodes ldquoOutput universality inmaximum efficiency linearpower amplifiersrdquo International Journal of Circuit Theory andApplications vol 31 no 4 pp 385ndash405 2003
[8] A Juhas and L A Novak ldquoComments on lsquoClass-E class-Cand classF power amplifier based upon a finite number ofharmonicsrsquordquo IEEE Transactions on Microwave Theory and Tech-niques vol 57 no 6 pp 1623ndash1625 2009
[9] M Roberg and Z Popovic ldquoAnalysis of high-efficiency poweramplifiers with arbitrary output harmonic terminationsrdquo IEEETransactions on Microwave Theory and Techniques vol 59 no8 pp 2037ndash2048 2011
[10] A Grebennikov N O Sokal and M J Franco Switchmode RFPower Amplifiers ElsevierAcademic Press San Diego CalifUSA 2nd edition 2012
[11] T Canning P J Tasker and S C Cripps ldquoContinuous modepower amplifier design using harmonic clipping contourstheory and practicerdquo IEEE Transactions on Microwave Theoryand Techniques vol 62 no 1 pp 100ndash110 2014
[12] A Juhas and L A Novak ldquoGeneral description of nonnegativewaveforms up to second harmonic for power amplifier mod-ellingrdquoMathematical Problems in Engineering vol 2014 ArticleID 709762 18 pages 2014
[13] V I Arnolrsquod V V Goryunov O V Lyashko and V A VasilrsquoevDynamical Systems VIIImdashSingularity Theory II ApplicationsSpringer Berlin Germany 1993
[14] D Siersma ldquoProperties of conflict sets in the planerdquo BanachCenter Publications Polish Academy of Sciences vol 50 no 1 pp267ndash276 1999 Proceedings of the Banach Center Symposiumon Geometry and Topology of Caustics (Caustics rsquo98) WarsawPoland
[15] M van Manen The geometry of conflict sets [Dissertation]Universiteit Utrecht UtrechtThe Netherlands 2003 httpigi-tur-archivelibraryuunldissertations2003-0912-123058c4pdf
[16] Y L Sachkov ldquoMaxwell strata and symmetries in the problemofoptimal rolling of a sphere over a planerdquo Sbornik Mathematicsvol 201 no 7-8 pp 1029ndash1051 2010
[17] I A Bogaevsky ldquoPerestroikas of shocks and singularities ofminimum functionsrdquoPhysicaDNonlinear Phenomena vol 173no 1-2 pp 1ndash28 2002
[18] Y L Sachkov ldquoMaxwell strata in the Euler elastic problemrdquoJournal of Dynamical andControl Systems vol 14 no 2 pp 169ndash234 2008
[19] M Siino and T Koike ldquoTopological classification of black holesgeneric Maxwell set and crease set of a horizonrdquo InternationalJournal ofModern Physics D Gravitation Astrophysics Cosmol-ogy vol 20 no 6 pp 1095ndash1122 2011
Mathematical Problems in Engineering 29
[20] F H Raab ldquoClass-F power amplifiers with maximally flat wave-formsrdquo IEEETransactions onMicrowaveTheory and Techniquesvol 45 no 11 pp 2007ndash2012 1997
[21] A Juhas and L A Novak ldquoMaximally flat waveforms with finitenumber of harmonics in class-F power amplifiersrdquo Mathemat-ical Problems in Engineering vol 2013 Article ID 169590 9pages 2013
[22] L Fejer ldquoUber trigonometrische polynomerdquo Journal fur dieReine und Angewandte Mathematik vol 1916 no 146 pp 53ndash82 1916 (German)
[23] S C Cripps ldquoBessel Waives [microwave bytes]rdquo IEEE Micro-wave Magazine vol 10 no 7 pp 30ndash36 117 2009
[24] L N Bryzgalova ldquoSingularities of the maximum of para-metrically dependent functionrdquo Functional Analysis and ItsApplications vol 11 no 1 pp 49ndash51 1977
[25] V I Arnold A A Davydov V A Vassiliev and V MZakalyukin Mathematical Models of Catastrophes Control ofCatastrophic Process Encyclopedia of Life Support Systems(EOLSS) EOLSS Publishers Oxford UK 2006
[26] S C Cripps RF Power Amplifiers for Wireless CommunicationsArtech House Norwood Mass USA 2nd edition 2006
[27] A Zygmund Trigonometric Series vol 1 Cambridge UniversityPress Cambridge UK 2nd edition 1959
[28] A Jeffrey and H DaiHandbook of Mathematical Formulas andIntegrals ElsevierAcademic Press San Diego Calif USA 4thedition 2008
[29] J C Mason and D C Handscomb Chebyshev PolynomialsChapman amp Hall CRC Press Boca Raton Fla USA 2003
Submit your manuscripts athttpwwwhindawicom
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical Problems in Engineering
Hindawi Publishing Corporationhttpwwwhindawicom
Differential EquationsInternational Journal of
Volume 2014
Applied MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Probability and StatisticsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical PhysicsAdvances in
Complex AnalysisJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
OptimizationJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
CombinatoricsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Operations ResearchAdvances in
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Function Spaces
Abstract and Applied AnalysisHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of Mathematics and Mathematical Sciences
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Algebra
Discrete Dynamics in Nature and Society
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Decision SciencesAdvances in
Discrete MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom
Volume 2014 Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Stochastic AnalysisInternational Journal of
Mathematical Problems in Engineering 5
specified by relations (11)ndash(14) From (11)ndash(13) it follows thatrelations
119891 (1205911015840
119860 120572) minus 119891 (12059110158401015840
119860 120572) = 0
1198911015840
(1205911015840
119860 120572) + 1198911015840
(12059110158401015840
119860 120572) = 0
1198911015840
(1205911015840
119860 120572) minus 1198911015840
(12059110158401015840
119860 120572) = 0
11989110158401015840
(1205911015840
119860 120572) + 11989110158401015840
(12059110158401015840
119860 120572) gt 0
(16)
also hold Let
120591119904119903=(1205911015840
+ 12059110158401015840
)
2 120591
Δ=(12059110158401015840
minus 1205911015840
)
2
(17)
be a pair of points associated with (1205911015840 12059110158401015840) Clearly
minus120587 lt 120591119904119903lt 120587 (18)
0 lt 120591Δlt 120587 (19)
12059110158401015840
= 120591119904119903+ 120591Δ 120591
1015840
= 120591119904119903minus 120591Δ (20)
The first and second derivatives of 119891(120591 119860 120572) are equal to
1198911015840
(120591 119860 120572) = sin 120591 + 119896119860 sin (119896120591 + 120572)
11989110158401015840
(120591 119860 120572) = cos 120591 + 1198962119860 cos (119896120591 + 120572) (21)
By using (20)-(21) system (16) can be rewritten as
sin 120591119904119903sin 120591Δ+ 119860 sin (119896120591
119904119903+ 120572) sin 119896120591
Δ= 0 (22)
sin 120591119904119903cos 120591Δ+ 119896119860 sin (119896120591
119904119903+ 120572) cos 119896120591
Δ= 0 (23)
cos 120591119904119903sin 120591Δ+ 119896119860 cos (119896120591
119904119903+ 120572) sin 119896120591
Δ= 0 (24)
cos 120591119904119903cos 120591Δ+ 1198962
119860 cos (119896120591119904119903+ 120572) cos 119896120591
Δgt 0 (25)
From (19) it follows that sin 120591Δgt 0 Multiplying (24) and
(25) with minus cos 120591Δand sin 120591
Δgt 0 respectively and sum-
ming the resulting relations we obtain 119896119860 cos(119896120591119904119903
+
120572)[119896 sin 120591Δcos 119896120591
Δminus sin 119896120591
Δcos 120591Δ] gt 0 The latest relation
immediately implies that
119896 sin 120591Δcos 119896120591
Δminus sin 119896120591
Δcos 120591Δ
= 0 (26)
Equations (22) and (23) can be considered as a system oftwo linear equations in terms of sin 120591
119904119903and 119860 sin(119896120591
119904119903+ 120572)
According to (26) the determinant of this system is nonzeroand therefore it has only trivial solution
sin 120591119904119903= 0 sin (119896120591
119904119903+ 120572) = 0 (27)
According to (18) sin 120591119904119903= 0 implies
120591119904119903= 0 (28)
According to (20) 120591119904119903= 0 implies
12059110158401015840
= 120591Δ 120591
1015840
= minus120591Δ (29)
Furthermore 120591119904119903= 0 and sin(119896120591
119904119903+ 120572) = 0 imply that sin120572 =
0 From (29) it follows that 120591Δis position of global minimum
of 119891(120591 119860 120572) Clearly 119891(120591Δ 119860 120572) le 119891(0 119860 120572) which together
with sin120572 = 0 leads to
1 minus cos 120591Δ+ 119860 cos120572 (1 minus cos 119896120591
Δ) le 0 (30)
From 120591Δ
= 0 (see (19)) 119860 gt 0 and (30) it follows that cos 120572 lt
0 which together with sin120572 = 0 yields
120572 = 120587 (31)
Since 120591Δis position of global minimum it follows that
119891(120591Δ 119860 120587) le 119891(120587119896 119860 120587) Accordingly 119860(1 + cos 119896120591
Δ) le
cos 120591Δminus cos120587119896 which together with 119860 gt 0 implies that
cos 120591Δminus cos120587119896 ge 0 This relation along with (19) yields
0 lt 120591Δle120587
119896 (32)
Substitution of (31) and (28) in (24) leads to
119860 =sin 120591Δ
119896 sin 119896120591Δ
(33)
Notice that sin 119896120591Δ sin 120591
Δis monotonically decreasing func-
tion on interval (32)Therefore parameter119860 is monotonicallyincreasing function on the same interval with lim
120591Δrarr0+119860 =
11198962 Consequently 119860 gt 1119896
2 which completes theproof
23 Parameter Space In parameter space of family of wave-forms (2) there are two subsets playing important role in theclassification of the family instancesThese are conflict set andcatastrophe set
Catastrophe set is subset of parameter space of waveform119891(120591 119860 120572) It consists of those pairs (119860 120572) for which thecorresponding waveforms 119891(120591 119860 120572) have degenerate criticalpoints at which first and second derivatives are equal to zeroThus for finding catastrophe set we have to consider thefollowing system of equations
1198911015840
(120591119889 119860 120572) = 0
11989110158401015840
(120591119889 119860 120572) = 0
(34)
where 120591119889is a degenerate critical point of waveform119891(120591 119860 120572)
Conflict set in parameter space of waveform119891(120591 119860 120572) asshown in Proposition 1 is the ray described by 119860 gt 1119896
2 and120572 = 120587 It is intimately connected to catastrophe set
In what follows in this subsection we use polar coordinatesystem (119860 cos120572 119860 sin120572) instead of Cartesian coordinatesystem (119860 120572) Examples of catastrophe set and conflict setfor 119896 le 5 plotted in parameter space (119860 cos120572 119860 sin120572) arepresented in Figure 3 Solid line represents the catastropheset while dotted line describes conflict set The isolated pickpoints (usually called cusp) which appear in catastrophecurves correspond to maximally flat waveforms with max-imally flat minimum andor maximally flat maximumThereare two such picks in the catastrophe curves for 119896 = 2 and
6 Mathematical Problems in Engineering
00
00
k = 2 k = 3
k = 4 k = 5
Acos 120572
Acos 120572
Acos 120572
Acos 120572
A sin 120572
A sin 120572A sin 120572
A sin 120572
Figure 3 Catastrophe set (solid line) and corresponding conflict set(dotted line) for 119896 le 5 In each plot white triangle dot correspondsto optimal waveform and white circle dot corresponds to maximallyflat waveform
119896 = 4 and one in the catastrophe curves for 119896 = 3 and 119896 = 5Notice that the end point of conflict set is the cusp point
Catastrophe set divides the parameter space (119860 cos120572119860 sin120572) into disjoint subsets In the cases 119896 = 2 and 119896 =
3 catastrophe curve defines inner and outer part For 119896 gt
3 catastrophe curve makes partition of parameter space inseveral inner subsets and one outer subset (see Figure 3)
Notice also that multiplying 119891(120591 119860 120572) with a positiveconstant and adding in turn another constant which leads towaveform of type 119908(120591 120574 119860 120572) (see (1) and (2)) do not makeimpact on the character of catastrophe and conflict sets Thisis because in the course of finding catastrophe set first andsecond derivatives of 119891(120591 119860 120572) are set to zero Clearly (34) interms of 119891(120591 119860 120572) are equivalent to the analogous equationsin terms of119908(120591 120574 119860 120572) Analogously in the course of findingconflict set we consider only the positions of global minima(these positions forwaveforms119891(120591 119860 120572) and119908(120591 120574 119860 120572) arethe same)
3 Nonnegative Waveforms with atLeast One Zero
In what follows let us consider a waveform containing dccomponent fundamental and 119896th (119896 ge 2) harmonic of theform
119879119896(120591) = 1 + 119886
1cos 120591 + 119887
1sin 120591 + 119886
119896cos 119896120591 + 119887
119896sin 119896120591 (35)
The amplitudes of fundamental and 119896th harmonic of wave-form of type (35) respectively are
1205821= radic11988621+ 11988721 (36)
120582119896= radic1198862119896+ 1198872119896 (37)
As it is shown in Section 21 nonnegative waveforms withmaximal amplitude of fundamental harmonic or maximalcoefficient of fundamental harmonic cosine part have atleast one zero It is also shown in Section 22 (Corollary 4)that waveforms of type (35) with nonzero amplitude offundamental harmonic have either one or two globalminimaConsequently if nonnegative waveform of type (35) withnonzero amplitude of fundamental harmonic has at least onezero then it has at most two zeros
In Section 31 we provide general description of nonnega-tive waveforms of type (35) with at least one zero In Sections32 and 33 we consider nonnegative waveforms of type (35)with two zeros
31 General Description of Nonnegative Waveforms with atLeast One Zero The main result of this section is presentedin the following proposition
Proposition 6 Every nonnegative waveform of type (35) withat least one zero can be expressed in the following form
119879119896(120591) = [1 minus cos (120591 minus 120591
0)] [1 minus 120582
119896119903119896(120591)] (38)
where119903119896(120591) = (119896 minus 1) cos 120585
+ 2
119896minus1
sum
119899=1
(119896 minus 119899) cos (119899 (120591 minus 1205910) + 120585)
(39)
providing that
120582119896le [(119896 minus 1) cos 120585 + 119896 sin(120585 minus 120585119896)
sin(120585119896)]
minus1
(40)
10038161003816100381610038161205851003816100381610038161003816 le 120587 (41)
Remark 7 Function on the right hand side of (40) is mono-tonically increasing function of |120585| on interval |120585| le 120587 (formore details about this function see Remark 15) From (57)and (65) it follows that relation
0 le 120582119896le 1 (42)
holds for every nonnegative waveform of type (35) Noticethat according to (40) 120582
119896= 1 implies |120585| = 120587 Substitution
of 120582119896
= 1 and |120585| = 120587 into (55) yields 119879119896(120591) = 1 minus
cos 119896(120591 minus 1205910) Consequently 120582
119896= 1 implies that amplitude
1205821of fundamental harmonic is equal to zero
Remark 8 Conversion of (38) into additive form leads to thefollowing expressions for coefficients of nonnegative wave-forms of type (35) with at least one zero
1198861= minus (1 + 120582
119896cos 120585) cos 120591
0minus 119896120582119896sin 120585 sin 120591
0 (43)
1198871= minus (1 + 120582
119896cos 120585) sin 120591
0+ 119896120582119896sin 120585 cos 120591
0 (44)
119886119896= 120582119896cos (119896120591
0minus 120585) (45)
119887119896= 120582119896sin (119896120591
0minus 120585) (46)
providing that 120582119896satisfy (40) and |120585| le 120587
Mathematical Problems in Engineering 7
2
1
0
0 1 2 3 4
Angle 120591120587
Wav
efor
ms
1205823 = radic2201205823 = radic281205823 = radic24
Figure 4 Nonnegative waveforms with at least one zero for 119896 = 31205910= 1205876 and 120585 = 31205874
Three examples of nonnegative waveforms with at leastone zero for 119896 = 3 are presented in Figure 4 (examples ofnonnegative waveformswith at least one zero for 119896 = 2 can befound in [12]) For all three waveforms presented in Figure 4we assume that 120591
0= 1205876 and 120585 = 31205874 From (40) it follows
that 1205823le radic24 Coefficients of waveform with 120582
3= radic220
(dotted line) are 1198861= minus08977 119887
1= minus03451 119886
3= 005
and 1198873= minus005 Coefficients of waveform with 120582
3= radic28
(dashed line) are 1198861= minus09453 119887
1= minus01127 119886
3= 0125
and 1198873= minus0125 Coefficients of waveform with 120582
3= radic24
(solid line) are 1198861= minus10245 119887
1= 02745 119886
3= 025 and
1198873= minus025 First two waveforms have one zero while third
waveform (presented with solid line) has two zeros
Proof of Proposition 6 Waveform of type (35) containingdc component fundamental and 119896th harmonic can be alsoexpressed in the form
119879119896(120591) = 1 + 120582
1cos (120591 + 120593
1) + 120582119896cos (119896120591 + 120593
119896) (47)
where1205821ge 0120582
119896ge 0120593
1isin (minus120587 120587] and120593
119896isin (minus120587 120587] It is easy
to see that relations between coefficient of (35) and param-eters of (47) read as follows
1198861= 1205821cos1205931 119887
1= minus1205821sin1205931 (48)
119886119896= 120582119896cos120593119896 119887
119896= minus120582119896sin120593119896 (49)
Let us introduce 120585 such that10038161003816100381610038161205851003816100381610038161003816 le 120587 120585 = (119896120591
0+ 120593119896) mod2120587 (50)
Using (50) coefficients (49) can be expressed as (45)-(46)Let us assume that 119879
119896(120591) is nonnegative waveform of type
(35) with at least one zero that is 119879119896(120591) ge 0 and 119879
119896(1205910) = 0
for some 1205910 Notice that conditions 119879
119896(120591) ge 0 and 119879
119896(1205910) = 0
imply that 1198791015840119896(1205910) = 0 From 119879
119896(1205910) = 0 and 1198791015840
119896(1205910) = 0 by
using (50) it follows that
1205821cos (120591
0+ 1205931) = minus (1 + 120582
119896cos 120585)
1205821sin (1205910+ 1205931) = minus119896120582
119896sin 120585
(51)
respectively On the other hand 1205821cos(120591 + 120593
1) can be rewrit-
ten as
1205821cos (120591 + 120593
1) = 1205821cos (120591
0+ 1205931) cos (120591 minus 120591
0)
minus 1205821sin (1205910+ 1205931) sin (120591 minus 120591
0)
(52)
Substitution of (51) into (52) yields
1205821cos (120591 + 120593
1) = minus (1 + 120582
119896cos 120585) cos (120591 minus 120591
0)
+ 119896120582119896sin 120585 sin (120591 minus 120591
0)
(53)
According to (50) it follows that cos(119896120591 + 120593119896) = cos(119896(120591 minus
1205910) + 120585) that is
cos (119896120591 + 120593119896) = cos 120585 cos 119896 (120591 minus 120591
0) minus sin 120585 sin 119896 (120591 minus 120591
0)
(54)
Furthermore substitution of (54) and (53) into (47) leads to
119879119896(120591) = [1 minus cos (120591 minus 120591
0)] [1 + 120582
119896cos 120585]
minus 120582119896[1 minus cos 119896 (120591 minus 120591
0)] cos 120585
+ 120582119896[119896 sin (120591 minus 120591
0) minus sin 119896 (120591 minus 120591
0)] sin 120585
(55)
According to (A2) and (A4) (see Appendices) there is com-mon factor [1minuscos(120591minus120591
0)] for all terms in (55) Consequently
(55) can be written in the form (38) where
119903119896(120591) = minus cos 120585 + [
1 minus cos 119896 (120591 minus 1205910)
1 minus cos (120591 minus 1205910)] cos 120585
minus [119896 sin (120591 minus 120591
0) minus sin 119896 (120591 minus 120591
0)
1 minus cos (120591 minus 1205910)
] sin 120585
(56)
From (56) by using (A2) (A4) and cos 120585 cos 119899(120591 minus 1205910) minus
sin 120585 sin 119899(120591 minus 1205910) = cos(119899(120591 minus 120591
0) + 120585) we obtain (39)
In what follows we are going to prove that (40) also holdsAccording to (38) 119879
119896(120591) is nonnegative if and only if
120582119896max120591
119903119896(120591) le 1 (57)
Let us first show that position of global maximumof 119903119896(120591)
belongs to the interval |120591 minus 1205910| le 2120587119896 Relation (56) can be
rewritten as
119903119896(120591) = 119903
119896(1205910minus2120585
119896) + 119902119896(120591) (58)
where
119903119896(1205910minus2120585
119896) = (119896 minus 1) cos 120585 + 119896 sin (120585 minus 120585119896)
sin (120585119896) (59)
119902119896(120591) =
1
1 minus cos (120591 minus 1205910)
sdot [cos 120585 minus cos(119896(120591 minus 1205910+120585
119896))
+119896 sin 120585sin (120585119896)
(cos(120591 minus 1205910+120585
119896) minus cos(120585
119896))]
(60)
8 Mathematical Problems in Engineering
For |120585| lt 120587 relation sin 120585 sin(120585119896) gt 0 obviously holds Fromcos 119905 gt cos 1199051015840 for |119905| le 120587119896 lt |119905
1015840
| le 120587 it follows that positionof global maximum of the function of type [119888 cos 119905 minus cos(119896119905)]for 119888 gt 0 belongs to interval |119905| le 120587119896 Therefore position ofglobal maximum of the expression in the square brackets in(60) for |120585| lt 120587 belongs to interval |120591 minus 120591
0+ 120585119896| le 120587119896 This
inequality together with |120585| lt 120587 leads to |120591minus1205910| lt 2120587119896 Since
[1 minus cos(120591 minus 1205910)]minus1 decreases with increasing |120591 minus 120591
0| le 120587 it
follows that 119902119896(120591) for |120585| lt 120587 has global maximum on interval
|120591minus1205910| lt 2120587119896 For |120585| = 120587 it is easy to show thatmax
120591119902119896(120591) =
119902119896(1205910plusmn 2120587119896) = 0 Since 119903
119896(120591) minus 119902
119896(120591) is constant (see (58))
it follows from previous considerations that 119903119896(120591) has global
maximum on interval |120591 minus 1205910| le 2120587119896
To find max120591119903119896(120591) let us consider first derivative of 119903
119896(120591)
with respect to 120591 Starting from (56) first derivative of 119903119896(120591)
can be expressed in the following form
119889119903119896(120591)
119889120591= minus119904 (120591) sdot sin(
119896 (120591 minus 1205910)
2+ 120585) (61)
where
119904 (120591) = [sin(119896 (120591 minus 120591
0)
2) cos(
120591 minus 1205910
2)
minus 119896 cos(119896 (120591 minus 120591
0)
2) sin(
120591 minus 1205910
2)]
sdot sinminus3 (120591 minus 1205910
2)
(62)
Using (A6) (see Appendices) (62) can be rewritten as
119904 (120591) = 2
119896minus1
sum
119899=1
119899 (119896 minus 119899) cos((119896 minus 2119899) (120591 minus 120591
0)
2) (63)
From 119899(119896 minus 119899) gt 0 and |119896 minus 2119899| lt 119896 119899 = 1 (119896 minus 1) itfollows that all summands in (63) decrease with increasing|120591 minus 1205910| providing that |120591 minus 120591
0| le 2120587119896 Therefore 119904(120591) ge 119904(120591
0plusmn
2120587119896) = 119896sin2(120587119896) gt 0 for |120591 minus 1205910| le 2120587119896 Consequently
119889119903119896(120591)119889120591 = 0 and |120591minus120591
0| le 2120587119896 imply that sin(119896(120591minus120591
0)2+
120585) = 0From |120585| le 120587 |120591minus120591
0| le 2120587119896 and sin(119896(120591minus120591
0)2+120585) = 0
it follows that 120591minus1205910+120585119896 = minus120585119896 or |120591minus120591
0+120585119896| = (2120587minus|120585|)119896
and therefore cos(119896(120591 minus 1205910+ 120585119896)) = cos 120585 Since cos(120585119896) ge
cos(2120587 minus |120585|)119896 it follows that max120591119902119896(120591) is attained for 120591 =
1205910minus2120585119896 Furthermore from (60) it follows that max
120591119902119896(120591) =
119902119896(1205910minus 2120585119896) = 0 which together with (58)-(59) leads to
max120591
119903119896(120591) = 119903
119896(1205910minus2120585
119896)
= (119896 minus 1) cos 120585 + 119896 sin (120585 minus 120585119896)sin (120585119896)
(64)
Both terms on the right hand side of (64) are even functionsof 120585 and decrease with increase of |120585| |120585| le 120587 Thereforemax120591119903119896(120591) attains its lowest value for |120585| = 120587 It is easy to
show that right hand side of (64) for |120585| = 120587 is equal to 1which further implies that
max120591
119903119896(120591) ge 1 (65)
From (65) it follows that (57) can be rewritten as 120582119896
le
[max120591119903119896(120591)]minus1 Finally substitution of (64) into 120582
119896le
[max120591119903119896(120591)]minus1 leads to (40) which completes the proof
32 Nonnegative Waveforms with Two Zeros Nonnegativewaveforms of type (35) with two zeros always possess twoglobal minima Such nonnegative waveforms are thereforerelated to the conflict set
In this subsection we provide general description of non-negative waveforms of type (35) for 119896 ge 2 and exactly twozeros According to Remark 7 120582
119896= 1 implies |120585| = 120587 and
119879119896(120591) = 1 minus cos 119896(120591 minus 120591
0) Number of zeros of 119879
119896(120591) = 1 minus
cos 119896(120591minus1205910) on fundamental period equals 119896 which is greater
than two for 119896 gt 2 and equal to two for 119896 = 2 In the followingproposition we exclude all waveforms with 120582
119896= 1 (the case
when 119896 = 2 and 1205822= 1 is going to be discussed in Remark 10)
Proposition 9 Every nonnegative waveform of type (35) withexactly two zeros can be expressed in the following form
119879119896(120591) = 120582
119896[1 minus cos (120591 minus 120591
0)] [1 minus cos(120591 minus 120591
0+2120585
119896)]
sdot [1198880+ 2
119896minus2
sum
119899=1
119888119899cos 119899(120591 minus 120591
0+120585
119896)]
(66)
where
119888119899= [sin(120585 minus 119899120585
119896) cos(120585
119896)
minus (119896 minus 119899) cos(120585 minus 119899120585
119896) sin(120585
119896)]
sdot sinminus3 (120585119896)
(67)
120582119896= [(119896 minus 1) cos 120585 + 119896 sin(120585 minus 120585119896)
sin(120585119896)]
minus1
(68)
0 lt10038161003816100381610038161205851003816100381610038161003816 lt 120587 (69)
Remark 10 For 119896 = 2 waveforms with 1205822= 1 also have
exactly two zeros These waveforms can be included in aboveproposition by substituting (69) with 0 lt |120585| le 120587
Remark 11 Apart from nonnegative waveforms of type (35)with two zeros there are another two types of nonnegativewaveforms which can be obtained from (66)ndash(68) These are
(i) nonnegative waveforms with 119896 zeros (correspondingto |120585| = 120587) and
(ii) maximally flat nonnegative waveforms (correspond-ing to 120585 = 0)
Notice that nonnegative waveforms of type (35) with120582119896= 1 can be obtained from (66)ndash(68) by setting |120585| =
120587 Substitution of 120582119896
= 1 and |120585| = 120587 into (66) alongwith execution of all multiplications and usage of (A2) (seeAppendices) leads to 119879
119896(120591) = 1 minus cos 119896(120591 minus 120591
0)
Mathematical Problems in Engineering 9
Also maximally flat nonnegative waveforms (they haveonly one zero [21]) can be obtained from (66)ndash(68) by setting120585 = 0 Thus substitution of 120585 = 0 into (66)ndash(68) leads tothe following form of maximally flat nonnegative waveformof type (35)
119879119896(120591) =
[1 minus cos (120591 minus 1205910)]2
3 (1198962 minus 1)
sdot [119896 (1198962
minus 1)
+ 2
119896minus2
sum
119899=1
(119896 minus 119899) ((119896 minus 119899)2
minus 1) cos 119899 (120591 minus 1205910)]
(70)
Maximally flat nonnegative waveforms of type (35) for 119896 le 4
can be expressed as
1198792(120591) =
2
3[1 minus cos(120591 minus 120591
0)]2
1198793(120591) =
1
2[1 minus cos(120591 minus 120591
0)]2
[2 + cos (120591 minus 1205910)]
1198794(120591) =
4
15[1 minus cos (120591 minus 120591
0)]2
sdot [5 + 4 cos (120591 minus 1205910) + cos 2 (120591 minus 120591
0)]
(71)
Remark 12 Every nonnegative waveform of type (35) withexactly one zero at nondegenerate critical point can bedescribed as in Proposition 6 providing that symbol ldquolerdquoin relation (40) is replaced with ldquoltrdquo This is an immediateconsequence of Propositions 6 and 9 and Remark 11
Remark 13 Identity [1minus cos(120591minus1205910)][1minus cos(120591minus120591
0+2120585119896)] =
[cos 120585119896 minus cos(120591 minus 1205910+ 120585119896)]
2 implies that (66) can be alsorewritten as
119879119896(120591) = 120582
119896[cos 120585119896
minus cos(120591 minus 1205910+120585
119896)]
2
sdot [1198880+ 2
119896minus2
sum
119899=1
119888119899cos 119899(120591 minus 120591
0+120585
119896)]
(72)
Furthermore substitution of (67) into (72) leads to
119879119896(120591)
= 120582119896[cos 120585119896
minus cos(120591 minus 1205910+120585
119896)]
sdot [(119896 minus 1) sin 120585sin (120585119896)
minus 2
119896minus1
sum
119899=1
sin (120585 minus 119899120585119896)sin (120585119896)
cos 119899(120591 minus 1205910+120585
119896)]
(73)
Remark 14 According to (A6) (see Appendices) it followsthat coefficients (67) can be expressed as
119888119899= 2
119896minus119899minus1
sum
119898=1
119898(119896 minus 119899 minus 119898) cos((119896 minus 119899 minus 2119898) 120585119896
) (74)
Furthermore from (74) it follows that coefficients 119888119896minus2
119888119896minus3
119888119896minus4
and 119888119896minus5
are equal to119888119896minus2
= 2 (75)
119888119896minus3
= 8 cos(120585119896) (76)
119888119896minus4
= 8 + 12 cos(2120585119896) (77)
119888119896minus5
= 24 cos(120585119896) + 16 cos(3120585
119896) (78)
For example for 119896 = 2 (75) and (68) lead to 1198880= 2 and
1205822= 1(2+cos 120585) respectively which from (72) further imply
that
1198792(120591) =
2 [cos(1205852) minus cos(120591 minus 1205910+ 1205852)]
2
[2 + cos 120585] (79)
Also for 119896 = 3 (75) (76) and (68) lead to 1198881= 2 119888
0=
8 cos(1205853) and 1205823= [2(3 cos(1205853) + cos 120585)]minus1 respectively
which from (72) further imply that
1198793(120591) =
2 [cos (1205853) minus cos (120591 minus 1205910+ 1205853)]
2
[3 cos (1205853) + cos 120585]
sdot [2 cos(1205853) + cos(120591 minus 120591
0+120585
3)]
(80)
Remark 15 According to (A5) (see Appendices) relation(68) can be rewritten as
120582119896= [(119896 minus 1) cos 120585 + 119896
119896minus1
sum
119899=1
cos((119896 minus 2119899)120585119896
)]
minus1
(81)
Clearly amplitude 120582119896of 119896th harmonic of nonnegative wave-
form of type (35) with exactly two zeros is even functionof 120585 Since cos((119896 minus 2119899)120585119896) 119899 = 0 (119896 minus 1) decreaseswith increase of |120585| on interval 0 le |120585| le 120587 it follows that120582119896monotonically increases with increase of |120585| Right hand
side of (68) is equal to 1(1198962
minus 1) for 120585 = 0 and to one for|120585| = 120587 Therefore for nonnegative waveforms of type (35)with exactly two zeros the following relation holds
1
1198962 minus 1lt 120582119896lt 1 (82)
The left boundary in (82) corresponds to maximally flatnonnegative waveforms (see Remark 11) The right boundaryin (82) corresponds to nonnegative waveforms with 119896 zeros(also see Remark 11)
Amplitude of 119896th harmonic of nonnegative waveform oftype (35) with two zeros as a function of parameter 120585 for 119896 le5 is presented in Figure 5
Remark 16 Nonnegative waveform of type (35) with twozeros can be also expressed in the following form
119879119896(120591) = 1 minus 120582
119896
119896 sin 120585sin (120585119896)
cos(120591 minus 1205910+120585
119896)
+ 120582119896cos (119896 (120591 minus 120591
0) + 120585)
(83)
10 Mathematical Problems in Engineering
1
08
06
04
02
0minus1 minus05 0 05
Am
plitu
de120582k
1
Parameter 120585120587
k = 2k = 3
k = 4k = 5
Figure 5 Amplitude of 119896th harmonic of nonnegative waveformwith two zeros as a function of parameter 120585
where 120582119896is given by (68) and 0 lt |120585| lt 120587 From (83) it follows
that coefficients of fundamental harmonic of nonnegativewaveform of type (35) with two zeros are
1198861= minus1205821cos(120591
0minus120585
119896) 119887
1= minus1205821sin(120591
0minus120585
119896) (84)
where 1205821is amplitude of fundamental harmonic
1205821=
119896 sin 120585sin (120585119896)
120582119896 (85)
Coefficients of 119896th harmonic are given by (45)-(46)Notice that (68) can be rewritten as
120582119896= [cos(120585
119896)
119896 sin 120585sin(120585119896)
minus cos 120585]minus1
(86)
By introducing new variable
119909 = cos(120585119896) (87)
and using the Chebyshev polynomials (eg see Appendices)relations (85) and (86) can be rewritten as
1205821= 119896120582119896119880119896minus1
(119909) (88)
120582119896=
1
119896119909119880119896minus1
(119909) minus 119881119896(119909)
(89)
where119881119896(119909) and119880
119896(119909) denote the Chebyshev polynomials of
the first and second kind respectively From (89) it followsthat
120582119896[119896119909119880119896minus1
(119909) minus 119881119896(119909)] minus 1 = 0 (90)
which is polynomial equation of 119896th degree in terms of var-iable 119909 From 0 lt |120585| lt 120587 and (87) it follows that
cos(120587119896) lt 119909 lt 1 (91)
Since 120582119896is monotonically increasing function of |120585| 0 lt |120585| lt
120587 it follows that 120582119896is monotonically decreasing function of
119909 This further implies that (90) has only one solution thatsatisfies (91) (For 119896 = 2 expression (91) reads cos(1205872) le
119909 lt 1) This solution for 119909 (which can be obtained at leastnumerically) according to (88) leads to amplitude 120582
1of
fundamental harmonicFor 119896 le 4 solutions of (90) and (91) are
119909 = radic1 minus 1205822
21205822
1
3lt 1205822le 1
119909 =1
23radic1205823
1
8lt 1205823lt 1
119909 = radic1
6(1 + radic
51205824+ 3
21205824
)1
15lt 1205824lt 1
(92)
Insertion of (92) into (88) leads to the following relationsbetween amplitude 120582
1of fundamental and amplitude 120582
119896of
119896th harmonic 119896 le 4
1205821= radic8120582
2(1 minus 120582
2)
1
3lt 1205822le 1 (93)
1205821= 3 (
3radic1205823minus 1205823)
1
8lt 1205823lt 1 (94)
1205821= radic
32
27(radic2120582
4(3 + 5120582
4)3
minus 21205824(9 + 7120582
4))
1
15lt 1205824lt 1
(95)
Proof of Proposition 9 As it has been shown earlier (seeProposition 6) nonnegative waveform of type (35) with atleast one zero can be represented in form (38) Since weexclude nonnegative waveforms with 120582
119896= 1 according to
Remark 7 it follows that we exclude case |120585| = 120587Therefore inthe quest for nonnegative waveforms of type (35) having twozeros we will start with waveforms of type (38) for |120585| lt 120587It is clear that nonnegative waveforms of type (38) have twozeros if and only if
120582119896= [max120591
119903119896(120591)]minus1
(96)
and max120591119903119896(120591) = 119903
119896(1205910) According to (64) max
120591119903119896(120591) =
119903119896(1205910) implies |120585| = 0 Therefore it is sufficient to consider
only the interval (69)Substituting (96) into (38) we obtain
119879119896(120591) =
[1 minus cos (120591 minus 1205910)] [max
120591119903119896(120591) minus 119903
119896(120591)]
max120591119903119896(120591)
(97)
Mathematical Problems in Engineering 11
Expression max120591119903119896(120591) minus 119903
119896(120591) according to (64) and (39)
equals
max120591
119903119896(120591) minus 119903
119896(120591) = 119896
sin ((119896 minus 1) 120585119896)sin (120585119896)
minus 2
119896minus1
sum
119899=1
(119896 minus 119899) cos (119899 (120591 minus 1205910) + 120585)
(98)
Comparison of (97) with (66) yields
max120591
119903119896(120591) minus 119903
119896(120591) = [1 minus cos(120591 minus 120591
0+2120585
119896)]
sdot [1198880+ 2
119896minus2
sum
119899=1
119888119899cos 119899(120591 minus 120591
0+120585
119896)]
(99)
where coefficients 119888119899 119899 = 0 119896 minus 2 are given by (67) In
what follows we are going to show that right hand sides of(98) and (99) are equal
From (67) it follows that
1198880minus 1198881cos(120585
119896) = 119896
sin (120585 minus 120585119896)sin (120585119896)
(100)
Also from (67) for 119899 = 1 119896minus3 it follows that the followingrelations hold
(119888119899minus1
+ 119888119899+1
) cos(120585119896) minus 2119888
119899= 2 (119896 minus 119899) cos(120585 minus 119899120585
119896)
(119888119899minus1
minus 119888119899+1
) sin(120585119896) = 2 (119896 minus 119899) sin(120585 minus 119899120585
119896)
(101)
From (99) by using (75) (76) (100)-(101) and trigonometricidentities
cos(120591 minus 1205910+2120585
119896) = cos(120585
119896) cos(120591 minus 120591
0+120585
119896)
minus sin(120585119896) sin(120591 minus 120591
0+120585
119896)
cos(120585 minus 119899120585
119896) cos(119899(120591 minus 120591
0+120585
119896))
minus sin(120585 minus 119899120585
119896) sin(119899(120591 minus 120591
0+120585
119896))
= cos (119899 (120591 minus 1205910) + 120585)
(102)
we obtain (98) Consequently (98) and (99) are equal whichcompletes the proof
33 Nonnegative Waveforms with Two Zeros and PrescribedCoefficients of 119896thHarmonic In this subsectionwe show thatfor prescribed coefficients 119886
119896and 119887119896 there are 119896 nonnegative
waveforms of type (35) with exactly two zeros According to
(37) and (82) coefficients 119886119896and 119887119896of nonnegative waveforms
of type (35) with exactly two zeros satisfy the followingrelation
1
1198962 minus 1lt radic1198862119896+ 1198872119896lt 1 (103)
According to Remark 16 the value of 119909 (see (87)) that cor-responds to 120582
119896= radic1198862
119896+ 1198872119896can be determined from (90)-
(91) As we mentioned earlier (90) has only one solutionthat satisfies (91) This value of 119909 according to (88) leadsto the amplitude 120582
1of fundamental harmonic (closed form
expressions for 1205821in terms of 120582
119896and 119896 le 4 are given by (93)ndash
(95))On the other hand from (45)-(46) it follows that
1198961205910minus 120585 = atan 2 (119887
119896 119886119896) + 2119902120587 119902 = 1 (119896 minus 1) (104)
where function atan 2(119910 119909) is defined as
atan 2 (119910 119909) =
arctan(119910
119909) if 119909 ge 0
arctan(119910
119909) + 120587 if 119909 lt 0 119910 ge 0
arctan(119910
119909) minus 120587 if 119909 lt 0 119910 lt 0
(105)
with the codomain (minus120587 120587] Furthermore according to (84)and (104) the coefficients of fundamental harmonic of non-negative waveforms with two zeros and prescribed coeffi-cients of 119896th harmonic are equal to
1198861= minus1205821cos[
atan 2 (119887119896 119886119896) + 2119902120587
119896]
1198871= minus1205821sin[
atan 2 (119887119896 119886119896) + 2119902120587
119896]
(106)
where 119902 = 0 (119896minus 1) For chosen 119902 according to (104) and(66) positions of zeros are
1205910=1
119896[120585 + atan 2 (119887
119896 119886119896) + 2119902120587]
1205910minus2120585
119896=1
119896[minus120585 + atan 2 (119887
119896 119886119896) + 2119902120587]
(107)
From (106) and 119902 = 0 (119896minus1) it follows that for prescribedcoefficients 119886
119896and 119887119896 there are 119896 nonnegative waveforms of
type (35) with exactly two zerosWe provide here an algorithm to facilitate calculation
of coefficients 1198861and 1198871of nonnegative waveforms of type
(35) with two zeros and prescribed coefficients 119886119896and 119887
119896
providing that 119886119896and 119887119896satisfy (103)
12 Mathematical Problems in Engineering
2
1
0
Wav
efor
ms
0 1 2 3 4
Angle 120591120587
q = 0
q = 1
q = 2
Figure 6 Nonnegative waveforms with two zeros for 119896 = 3 1198863=
minus015 and 1198873= minus02
Algorithm 17 (i) Calculate 120582119896= radic1198862119896+ 1198872119896
(ii) identify 119909 that satisfies both relations (90) and (91)(iii) calculate 120582
1according to (88)
(iv) choose integer 119902 such that 0 le 119902 le 119896 minus 1(v) calculate 119886
1and 1198871according to (106)
For 119896 le 4 by using (93) for 119896 = 2 (94) for 119896 = 3 and (95)for 119896 = 4 it is possible to calculate directly 120582
1from 120582
119896and
proceed to step (iv)For 119896 = 2 and prescribed coefficients 119886
2and 1198872 there are
two waveforms with two zeros one corresponding to 1198861lt 0
and the other corresponding to 1198861gt 0 (see also [12])
Let us take as an input 119896 = 3 1198863= minus015 and 119887
3= minus02
Execution of Algorithm 17 on this input yields 1205823= 025 and
1205821= 11399 (according to (94)) For 119902 = 0 we calculate
1198861= minus08432 and 119887
1= 07670 (corresponding waveform is
presented by solid line in Figure 6) for 119902 = 1 we calculate1198861= minus02426 and 119887
1= minus11138 (corresponding waveform is
presented by dashed line) for 119902 = 2 we calculate 1198861= 10859
and 1198871= 03468 (corresponding waveform is presented by
dotted line)As another example of the usage of Algorithm 17 let us
consider case 119896 = 4 and assume that1198864= minus015 and 119887
4= minus02
Consequently 1205824= 025 and 120582
1= 09861 (according to (95))
For 119902 = 0 3we calculate the following four pairs (1198861 1198871) of
coefficients of fundamental harmonic (minus08388 05184) for119902 = 0 (minus05184 minus08388) for 119902 = 1 (08388 minus05184) for 119902 =2 and (05184 08388) for 119902 = 3 Corresponding waveformsare presented in Figure 7
4 Nonnegative Waveforms with MaximalAmplitude of Fundamental Harmonic
In this section we provide general description of nonnegativewaveforms containing fundamental and 119896th harmonic withmaximal amplitude of fundamental harmonic for prescribedamplitude of 119896th harmonic
The main result of this section is presented in the fol-lowing proposition
3
2
1
0
Wav
efor
ms
0 1 2 3 4
Angle 120591120587
q = 0q = 1
q = 2q = 3
Figure 7 Nonnegative waveforms with two zeros for 119896 = 4 1198864=
minus015 and 1198874= minus02
Proposition 18 Every nonnegativewaveformof type (35)withmaximal amplitude 120582
1of fundamental harmonic and pre-
scribed amplitude 120582119896of 119896th harmonic can be expressed in the
following form
119879119896(120591) = [1 minus cos (120591 minus 120591
0)]
sdot [1 minus (119896 minus 1) 120582119896minus 2120582119896
119896minus1
sum
119899=1
(119896 minus 119899) cos (119899 (120591 minus 1205910))]
(108)
if 0 le 120582119896le 1(119896
2
minus 1) or
119879119896(120591) = 120582
119896[1 minus cos (120591 minus 120591
0)] [1 minus cos(120591 minus 120591
0+2120585
119896)]
sdot [1198880+ 2
119896minus2
sum
119899=1
119888119899cos 119899(120591 minus 120591
0+120585
119896)]
(109)
if 1(1198962 minus 1) le 120582119896le 1 providing that 119888
119899 119899 = 0 119896 minus 2 and
120582119896are related to 120585 via relations (67) and (68) respectively and
|120585| le 120587
Remark 19 Expression (108) can be obtained from (38) bysetting 120585 = 0 Furthermore insertion of 120585 = 0 into (43)ndash(46)leads to the following expressions for coefficients ofwaveformof type (108)
1198861= minus (1 + 120582
119896) cos 120591
0 119887
1= minus (1 + 120582
119896) sin 120591
0
119886119896= 120582119896cos (119896120591
0) 119887
119896= 120582119896sin (119896120591
0)
(110)
On the other hand (109) coincides with (66) Thereforethe expressions for coefficients of (109) and (66) also coincideThus expressions for coefficients of fundamental harmonic ofwaveform (109) are given by (84) where 120582
1is given by (85)
while expressions for coefficients of 119896th harmonic are givenby (45)-(46)
Waveforms described by (108) have exactly one zerowhile waveforms described by (109) for 1(1198962 minus 1) lt 120582
119896lt 1
Mathematical Problems in Engineering 13
14
12
1
08
06
04
02
00 05 1
Amplitude 120582k
Am
plitu
de1205821
k = 2
k = 3
k = 4
Figure 8 Maximal amplitude of fundamental harmonic as a func-tion of amplitude of 119896th harmonic
have exactly two zeros As we mentioned earlier waveforms(109) for 120582
119896= 1 have 119896 zeros
Remark 20 Maximal amplitude of fundamental harmonic ofnonnegative waveforms of type (35) for prescribed amplitudeof 119896th harmonic can be expressed as
1205821= 1 + 120582
119896 (111)
if 0 le 120582119896le 1(119896
2
minus 1) or
1205821=
119896 sin 120585119896 sin 120585 cos (120585119896) minus cos 120585 sin (120585119896)
(112)
if 1(1198962 minus 1) le 120582119896le 1 where 120585 is related to 120582
119896via (68) (or
(86)) and |120585| le 120587From (110) it follows that (111) holds Substitution of (86)
into (85) leads to (112)Notice that 120582
119896= 1(119896
2
minus 1) is the only common point ofthe intervals 0 le 120582
119896le 1(119896
2
minus 1) and 1(1198962
minus 1) le 120582119896le
1 According to (111) 120582119896= 1(119896
2
minus 1) corresponds to 1205821=
1198962
(1198962
minus1) It can be also obtained from (112) by setting 120585 = 0The waveforms corresponding to this pair of amplitudes aremaximally flat nonnegative waveforms
Maximal amplitude of fundamental harmonic of non-negative waveform of type (35) for 119896 le 4 as a function ofamplitude of 119896th harmonic is presented in Figure 8
Remark 21 Maximum value of amplitude of fundamentalharmonic of nonnegative waveform of type (35) is
1205821max =
1
cos (120587 (2119896)) (113)
This maximum value is attained for |120585| = 1205872 (see (112)) Thecorresponding value of amplitude of 119896th harmonic is 120582
119896=
(1119896) tan(120587(2119896)) Nonnegative waveforms of type (35) with1205821= 1205821max have two zeros at 1205910 and 1205910 minus 120587119896 for 120585 = 1205872 or
at 1205910and 1205910+ 120587119896 for 120585 = minus1205872
14
12
1
08
06
04
02
0minus1 minus05 0 05 1
Am
plitu
de1205821
Parameter 120585120587
k = 2k = 3k = 4
Figure 9 Maximal amplitude of fundamental harmonic as a func-tion of parameter 120585
To prove that (113) holds let us first show that the fol-lowing relation holds for 119896 ge 2
cos( 120587
2119896) lt 1 minus
1
1198962 (114)
From 119896 ge 2 it follows that sinc(120587(4119896)) gt sinc(1205874) wheresinc 119909 = (sin119909)119909 and therefore sin(120587(4119896)) gt 1(radic2119896)By using trigonometric identity cos 2119909 = 1 minus 2sin2119909 weimmediately obtain (114)
According to (111) and (112) it is clear that 1205821attains its
maximum value on the interval 1(1198962 minus 1) le 120582119896le 1 Since
120582119896is monotonic function of |120585| on interval |120585| le 120587 (see
Remark 15) it follows that 119889120582119896119889120585 = 0 for 0 lt |120585| lt 120587
Therefore to find critical points of 1205821as a function of 120582
119896
it is sufficient to find critical points of 1205821as a function of
|120585| 0 lt |120585| lt 120587 and consider its values at the end points120585 = 0 and |120585| = 120587 Plot of 120582
1as a function of parameter 120585
for 119896 le 4 is presented in Figure 9 According to (112) firstderivative of 120582
1with respect to 120585 is equal to zero if and only
if (119896 cos 120585 sin(120585119896) minus sin 120585 cos(120585119896)) cos 120585 = 0 On interval0 lt |120585| lt 120587 this is true if and only if |120585| = 1205872 Accordingto (112) 120582
1is equal to 119896
2
(1198962
minus 1) for 120585 = 0 equal to zerofor |120585| = 120587 and equal to 1 cos(120587(2119896)) for |120585| = 1205872 From(114) it follows that 1198962(1198962minus1) lt 1 cos(120587(2119896)) and thereforemaximum value of 120582
1is given by (113) Moreover maximum
value of 1205821is attained for |120585| = 1205872
According to above consideration all nonnegative wave-forms of type (35) having maximum value of amplitude offundamental harmonic can be obtained from (109) by setting|120585| = 1205872 Three of them corresponding to 119896 = 3 120585 = 1205872and three different values of 120591
0(01205876 and1205873) are presented
in Figure 10 Dotted line corresponds to 1205910= 0 (coefficients
of corresponding waveform are 1198861= minus1 119887
1= 1radic3 119886
3= 0
and 1198873= minusradic39) solid line to 120591
0= 1205876 (119886
1= minus2radic3 119887
1= 0
1198863= radic39 and 119887
3= 0) and dashed line to 120591
0= 1205873 (119886
1= minus1
1198871= minus1radic3 119886
3= 0 and 119887
3= radic39)
Proof of Proposition 18 As it has been shown earlier (Propo-sition 6) nonnegative waveform of type (35) with at least
14 Mathematical Problems in Engineering
2
1
0
0 1 2 3 4
Angle 120591120587
Wav
efor
ms
1205910 = 01205910 = 12058761205910 = 1205873
Figure 10 Nonnegative waveforms with maximum amplitude offundamental harmonic for 119896 = 3 and 120585 = 1205872
one zero can be represented in form (38) According to (43)(44) and (36) for amplitude 120582
1of fundamental harmonic of
waveforms of type (38) the following relation holds
1205821= radic(1 + 120582
119896cos 120585)2 + 11989621205822
119896sin2120585 (115)
where 120582119896satisfy (40) and |120585| le 120587
Because of (40) in the quest of finding maximal 1205821for
prescribed 120582119896 we have to consider the following two cases
(Case i)120582119896lt [(119896minus1) cos 120585 + 119896 sin(120585minus120585119896) sin(120585119896)]minus1
(Case ii)120582119896= [(119896minus1) cos 120585+119896 sin(120585minus120585119896) sin(120585119896)]minus1
Case i Since 120582119896lt [(119896 minus 1) cos 120585 + 119896 sin(120585 minus 120585119896) sin(120585119896)]minus1
implies 120582119896
= 1 according to (115) it follows that 1205821
= 0Hence 119889120582
1119889120585 = 0 implies
2120582119896sin 120585 [1 minus (1198962 minus 1) 120582
119896cos 120585] = 0 (116)
Therefore 1198891205821119889120585 = 0 if 120582
119896= 0 (Option 1) or sin 120585 = 0
(Option 2) or (1198962 minus 1)120582119896cos 120585 = 1 (Option 3)
Option 1 According to (115) 120582119896= 0 implies 120582
1= 1 (notice
that this implication shows that 1205821does not depend on 120585 and
therefore we can set 120585 to zero value)
Option 2 According to (115) sin 120585 = 0 implies 1205821= 1 +
120582119896cos 120585 which further leads to the conclusion that 120582
1is
maximal for 120585 = 0 For 120585 = 0 120582119896lt [(119896 minus 1) cos 120585 + 119896 sin(120585 minus
120585119896) sin(120585119896)]minus1 becomes 120582119896lt 1(119896
2
minus 1)
Option 3 This option leads to contradiction To show thatnotice that (119896
2
minus 1)120582119896cos 120585 = 1 and 120582
119896lt [(119896 minus
1) cos 120585 + 119896 sin(120585 minus 120585119896) sin(120585119896)]minus1 imply that (119896 minus 1) cos 120585 gtsin(120585minus120585119896) sin(120585119896) Using (A5) (see Appendices) the latestinequality can be rewritten assum119896minus1
119899=1[cos 120585minuscos((119896minus2119899)120585119896)] gt
0 But from |119896 minus 2119899| lt 119896 119899 = 1 (119896 minus 1) and |120585| le 120587
it follows that all summands are not positive and therefore(119896minus1) cos 120585 gt sin(120585minus120585119896) sin(120585119896) does not hold for |120585| le 120587
Consequently Case i implies 120585 = 0 and 120582119896lt 1(119896
2
minus 1)Finally substitution of 120585 = 0 into (38) leads to (108) whichproves that (108) holds for 120582
119896lt 1(119896
2
minus 1)
Case ii Relation120582119896= [(119896minus1) cos 120585+119896 sin(120585minus120585119896) sin(120585119896)]minus1
according to Proposition 9 and Remark 11 implies that cor-responding waveforms can be expressed via (66)ndash(68) for|120585| le 120587 Furthermore 120582
119896= [(119896 minus 1) cos 120585 + 119896 sin(120585 minus
120585119896) sin(120585119896)]minus1 and |120585| le 120587 imply 1(1198962 minus 1) le 120582119896le 1
This proves that (109) holds for 1(1198962 minus 1) le 120582119896le 1
Finally let us prove that (108) holds for 120582119896= 1(119896
2
minus
1) According to (68) (see also Remark 11) this value of 120582119896
corresponds to 120585 = 0 Furthermore substitution of 120582119896=
1(1198962
minus 1) and 120585 = 0 into (109) leads to (70) which can berewritten as
119879119896(120591) =
[1 minus cos (120591 minus 1205910)]
(1 minus 1198962)
sdot [119896 (119896 minus 1) minus 2
119896minus1
sum
119899=1
(119896 minus 119899) cos (119899 (120591 minus 1205910))]
(117)
Waveform (117) coincides with waveform (108) for 120582119896
=
1(1 minus 1198962
) Consequently (108) holds for 120582119896= 1(1 minus 119896
2
)which completes the proof
5 Nonnegative Waveforms with MaximalAbsolute Value of the Coefficient of CosineTerm of Fundamental Harmonic
In this sectionwe consider general description of nonnegativewaveforms of type (35) with maximal absolute value ofcoefficient 119886
1for prescribed coefficients of 119896th harmonicThis
type of waveform is of particular interest in PA efficiencyanalysis In a number of cases of practical interest eithercurrent or voltage waveform is prescribed In such casesthe problem of finding maximal efficiency of PA can bereduced to the problem of finding nonnegative waveformwith maximal coefficient 119886
1for prescribed coefficients of 119896th
harmonic (see also Section 7)In Section 51 we provide general description of nonneg-
ative waveforms of type (35) with maximal absolute value ofcoefficient 119886
1for prescribed coefficients of 119896th harmonic In
Section 52 we illustrate results of Section 51 for particularcase 119896 = 3
51 Nonnegative Waveforms with Maximal Absolute Value ofCoefficient 119886
1for 119896 ge 2 Waveforms 119879
119896(120591) of type (35) with
1198861ge 0 can be derived from those with 119886
1le 0 by shifting
by 120587 and therefore we can assume without loss of generalitythat 119886
1le 0 Notice that if 119896 is even then shifting 119879
119896(120591) by
120587 produces the same result as replacement of 1198861with minus119886
1
(119886119896remains the same) On the other hand if 119896 is odd then
shifting 119879119896(120591) by 120587 produces the same result as replacement
of 1198861with minus119886
1and 119886119896with minus119886
119896
According to (37) coefficients of 119896th harmonic can beexpressed as
119886119896= 120582119896cos 120575 119887
119896= 120582119896sin 120575 (118)
Mathematical Problems in Engineering 15
where
|120575| le 120587 (119)
Conversely for prescribed coefficients 119886119896and 119887
119896 120575 can be
determined as
120575 = atan 2 (119887119896 119886119896) (120)
where definition of function atan 2(119910 119909) is given by (105)The main result of this section is stated in the following
proposition
Proposition 22 Every nonnegative waveform of type (35)withmaximal absolute value of coefficient 119886
1le 0 for prescribed
coefficients 119886119896and 119887119896of 119896th harmonic can be represented as
119879119896(120591)
= [1 minus cos 120591]
sdot [1 minus (119896 minus 1) 119886119896minus 2
119896minus1
sum
119899=1
(119896 minus 119899) (119886119896cos 119899120591 + 119887
119896sin 119899120591)]
(121)
if 119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) le 1+119886
119896 where 120575 = atan 2(bk
119886119896) or
119879119896(120591) = 120582
119896[1 minus cos(120591 minus (120575 + 120585)
119896)]
sdot [1 minus cos(120591 minus (120575 minus 120585)
119896)]
sdot [1198880+ 2
119896minus2
sum
119899=1
119888119899cos 119899(120591 minus 120575
119896)]
(122)
if 119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) ge 1 + 119886
119896 where 119888
119899 119899 = 0
119896minus2 and 120582119896= radic1198862119896+ 1198872119896are related to 120585 via relations (67) and
(68) respectively and |120585| le 120587
Remark 23 Expression (121) can be obtained from (38) bysetting 120591
0= 0 and 120585 = minus120575 and then replacing 120582
119896cos 120575 with
119886119896(see (118)) and 120582
119896cos(119899120591 minus 120575) with 119886
119896cos 119899120591 + 119887
119896sin 119899120591
(see also (118)) Furthermore insertion of 1205910= 0 and 120585 =
minus120575 into (43)ndash(46) leads to the following relations betweenfundamental and 119896th harmonic coefficients of waveform(121)
1198861= minus (1 + 119886
119896) 119887
1= minus119896119887
119896 (123)
On the other hand expression (122) can be obtained from(66) by replacing 120591
0minus120585119896with 120575119896 Therefore substitution of
1205910minus 120585119896 = 120575119896 in (84) leads to
1198861= minus1205821cos(120575
119896) 119887
1= minus1205821sin(120575
119896) (124)
where 1205821is given by (85)
The fundamental harmonic coefficients 1198861and 1198871of wave-
form of type (35) with maximal absolute value of coefficient1198861le 0 satisfy both relations (123) and (124) if 119886
119896and 119887119896satisfy
1 + 119886119896= 119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) For such waveforms
relations 1205910= 0 and 120585 = minus120575 also hold
Remark 24 Amplitude of 119896th harmonic of nonnegativewaveform of type (35) with maximal absolute value of coeffi-cient 119886
1le 0 and coefficients 119886
119896 119887119896satisfying 1 + 119886
119896=
119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) is
120582119896=
sin (120575119896)119896 sin 120575 cos (120575119896) minus cos 120575 sin (120575119896)
(125)
To show that it is sufficient to substitute 119886119896= 120582119896cos 120575 (see
(118)) into 1 + 119886119896= 119896120582119896[sin 120575 sin(120575119896)] cos(120575119896)
Introducing new variable
119910 = cos(120575119896) (126)
and using the Chebyshev polynomials (eg see Appendices)relations 119886
119896= 120582119896cos 120575 and (125) can be rewritten as
119886119896= 120582119896119881119896(119910) (127)
120582119896=
1
119896119910119880119896minus1
(119910) minus 119881119896(119910)
(128)
where119881119896(119910) and119880
119896(119910) denote the Chebyshev polynomials of
the first and second kind respectively Substitution of (128)into (127) leads to
119886119896119896119910119880119896minus1
(119910) minus (1 + 119886119896) 119881119896(119910) = 0 (129)
which is polynomial equation of 119896th degree in terms of var-iable 119910 From |120575| le 120587 and (126) it follows that
cos(120587119896) le 119910 le 1 (130)
In what follows we show that 119886119896is monotonically increas-
ing function of 119910 on the interval (130) From 120585 = minus120575 (seeRemark 23) and (81) it follows that 120582minus1
119896= (119896 minus 1) cos 120575 +
119896sum119896minus1
119899=1cos((119896 minus 2119899)120575119896) ge 1 and therefore 119886
119896= 120582119896cos 120575 can
be rewritten as
119886119896=
cos 120575(119896 minus 1) cos 120575 + 119896sum119896minus1
119899=1cos ((119896 minus 2119899) 120575119896)
(131)
Obviously 119886119896is even function of 120575 and all cosines in (131)
are monotonically decreasing functions of |120575| on the interval|120575| le 120587 It is easy to show that cos((119896 minus 2119899)120575119896) 119899 =
1 (119896 minus 1) decreases slower than cos 120575 when |120575| increasesThis implies that denominator of the right hand side of(131) decreases slower than numerator Since denominator ispositive for |120575| le 120587 it further implies that 119886
119896is decreasing
function of |120575| on interval |120575| le 120587 Consequently 119886119896is
monotonically increasing function of 119910 on the interval (130)Thus we have shown that 119886
119896is monotonically increasing
function of 119910 on the interval (130) and therefore (129) hasonly one solution that satisfies (130) According to (128) thevalue of 119910 obtained from (129) and (130) either analyticallyor numerically leads to amplitude 120582
119896of 119896th harmonic
16 Mathematical Problems in Engineering
1
05
0
minus05
minus1
minus1 minus05 0 05 1
Coefficient ak
Coe
ffici
entb
k
radica2k+ b2
kle 1
k = 2k = 3k = 4
Figure 11 Plot of (119886119896 119887119896) satisfying 1 + 119886
119896= 119896120582
119896[sin 120575 sin(120575
119896)] cos(120575119896) for 119896 le 4
By solving (129) and (130) for 119896 le 4 we obtain
119910 = radic1 + 1198862
2 (1 minus 1198862) minus1 le 119886
2le1
3
119910 = radic3
4 (1 minus 21198863) minus1 le 119886
3le1
8
119910 =radicradic2 minus 4119886
4+ 1011988624minus 2 (1 minus 119886
4)
4 (1 minus 31198864)
minus1 le 1198864le
1
15
(132)
Insertion of (132) into (128) leads to the following explicitexpressions for the amplitude 120582
119896 119896 le 4
1205822=1
2(1 minus 119886
2) minus1 le 119886
2le1
3 (133)
1205822
3= [
1
3(1 minus 2119886
3)]
3
minus1 le 1198863le1
8 (134)
1205824=1
4(minus1 minus 119886
4+ radic2 minus 4119886
4+ 1011988624) minus1 le 119886
4le
1
15
(135)
Relations (133)ndash(135) define closed lines (see Figure 11) whichseparate points representing waveforms of type (121) frompoints representing waveforms of type (122) For given 119896points inside the corresponding curve refer to nonnegativewaveforms of type (121) whereas points outside curve (andradic1198862119896+ 1198872119896le 1) correspond to nonnegative waveforms of type
(122) Points on the respective curve correspond to the wave-forms which can be expressed in both forms (121) and (122)
Remark 25 Themaximum absolute value of coefficient 1198861of
nonnegative waveform of type (35) is
100381610038161003816100381611988611003816100381610038161003816max =
1
cos (120587 (2119896)) (136)
This maximum value is attained for |120585| = 1205872 and 120575 = 0
(see (124)) Notice that |1198861|max is equal to the maximum value
1205821max of amplitude of fundamental harmonic (see (113))
Coefficients of waveform with maximum absolute value ofcoefficient 119886
1 1198861lt 0 are
1198861= minus
1
cos (120587 (2119896)) 119886
119896=1
119896tan( 120587
(2119896))
1198871= 119887119896= 0
(137)
Waveformdescribed by (137) is cosinewaveformhaving zerosat 120587(2119896) and minus120587(2119896)
In the course of proving (136) notice first that |1198861|max le
1205821max holds According to (123) and (124) maximum of |119886
1|
occurs for 119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) ge 1 + 119886
119896 From (124)
it immediately follows that maximum value of |1198861| is attained
if and only if 1205821= 1205821max and 120575 = 0 which because of
120575119896 = 1205910minus120585119896 further implies 120591
0= 120585119896 Sincemaximumvalue
of 1205821is attained for |120585| = 1205872 it follows that corresponding
waveform has zeros at 120587(2119896) and minus120587(2119896)
Proof of Proposition 22 As it was mentioned earlier in thissection we can assume without loss of generality that 119886
1le 0
We consider waveforms119879119896(120591) of type (35) such that119879
119896(120591) ge 0
and119879119896(120591) = 0 for some 120591
0 Fromassumption that nonnegative
waveform 119879119896(120591) of type (35) has at least one zero it follows
that it can be expressed in form (38)Let us also assume that 120591
0is position of nondegenerate
critical point Therefore 119879119896(1205910) = 0 implies 1198791015840
119896(1205910) = 0 and
11987910158401015840
119896(1205910) gt 0 According to (55) second derivative of 119879
119896(120591) at
1205910can be expressed as 11987910158401015840
119896(1205910) = 1 minus 120582
119896(1198962
minus 1) cos 120585 Since11987910158401015840
119896(1205910) gt 0 it follows immediately that
1 minus 120582119896(1198962
minus 1) cos 120585 gt 0 (138)
Let us further assume that 119879119896(120591) has exactly one zeroThe
problem of finding maximum absolute value of 1198861is con-
nected to the problem of finding maximum of the minimumfunction (see Section 21) If waveforms possess unique globalminimum at nondegenerate critical point then correspond-ing minimum function is a smooth function of parameters[13] Consequently assumption that 119879
119896(120591) has exactly one
zero at nondegenerate critical point leads to the conclusionthat coefficient 119886
1is differentiable function of 120591
0 First
derivative of 1198861(see (43)) with respect to 120591
0 taking into
account that 1205971205851205971205910= 119896 (see (50)) can be expressed in the
following factorized form
1205971198861
1205971205910
= sin 1205910[1 minus 120582
119896(1198962
minus 1) cos 120585] (139)
Mathematical Problems in Engineering 17
From (138) and (139) it is clear that 12059711988611205971205910= 0 if and only if
sin 1205910= 0 According toRemark 12 assumption that119879
119896(120591)has
exactly one zero implies 120582119896lt 1 From (51) (48) and 120582
119896lt 1
it follows that 1198861cos 1205910+ 1198871sin 1205910lt 0 which together with
sin 1205910= 0 implies that 119886
1cos 1205910lt 0 Assumption 119886
1le 0
together with relations 1198861cos 1205910lt 0 and sin 120591
0= 0 further
implies 1198861
= 0 and
1205910= 0 (140)
Insertion of 1205910= 0 into (38) leads to
119879119896(120591)
= [1 minus cos 120591]
sdot [1 minus (119896 minus 1) 120582119896cos 120585 minus 2120582
119896
119896minus1
sum
119899=1
(119896 minus 119899) cos (119899120591 + 120585)]
(141)
Substitution of 1205910= 0 into (45) and (46) yields 119886
119896= 120582119896cos 120585
and 119887119896
= minus120582119896sin 120585 respectively Replacing 120582
119896cos 120585 with
119886119896and 120582
119896cos(119899120591 + 120585) with (119886
119896cos 119899120591 + 119887
119896sin 119899120591) in (141)
immediately leads to (121)Furthermore 119886
119896= 120582119896cos 120585 119887
119896= minus120582
119896sin 120585 and (118)
imply that
120575 = minus120585 (142)
According to (38)ndash(40) and (142) it follows that (141) is non-negative if and only if
120582119896[(119896 minus 1) cos 120575 + 119896 sin (120575 minus 120575119896)
sin (120575119896)] lt 1 (143)
Notice that 119886119896= 120582119896cos 120575 implies that the following relation
holds
120582119896[(119896 minus 1) cos 120575 + 119896 sin (120575 minus 120575119896)
sin (120575119896)]
= minus119886119896+ 119896120582119896
sin 120575sin (120575119896)
cos(120575119896)
(144)
Finally substitution of (144) into (143) leads to 119896120582119896[sin 120575
sin(120575119896)] cos(120575119896) lt 1 + 119886119896 which proves that (121) holds
when 119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) lt 1 + 119886
119896
Apart from nonnegative waveforms with exactly one zeroat nondegenerate critical point in what follows we will alsoconsider other types of nonnegative waveforms with at leastone zero According to Proposition 9 and Remark 11 thesewaveforms can be described by (66)ndash(68) providing that 0 le|120585| le 120587
According to (35) 119879119896(0) ge 0 implies 1 + 119886
1+ 119886119896ge 0
Consequently 1198861le 0 implies that |119886
1| le 1 + 119886
119896 On the other
hand according to (123) |1198861| = 1 + 119886
119896holds for waveforms
of type (121) The converse is also true 1198861le 0 and |119886
1| =
1 + 119886119896imply 119886
1= minus1 minus 119886
119896 which further from (35) implies
119879119896(0) = 0 Therefore in what follows it is enough to consider
only nonnegativewaveformswhich can be described by (66)ndash(68) and 0 le |120585| le 120587 with coefficients 119886
119896and 119887119896satisfying
119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) ge 1 + 119886
119896
For prescribed coefficients 119886119896and 119887119896 the amplitude 120582
119896=
radic1198862119896+ 1198872119896of 119896th harmonic is also prescribed According to
Remark 15 (see also Remark 16) 120582119896is monotonically
decreasing function of 119909 = cos(120585119896) The value of 119909 can beobtained by solving (90) subject to the constraint cos(120587119896) le119909 le 1 Then 120582
1can be determined from (88) From (106) it
immediately follows that maximal absolute value of 1198861le 0
corresponds to 119902 = 0 which from (104) and (120) furtherimplies that
120575 = 1198961205910minus 120585 (145)
Furthermore 119902 = 0 according to (107) implies that waveformzeros are
1205910=(120575 + 120585)
119896 120591
1015840
0= 1205910minus2120585
119896=(120575 minus 120585)
119896 (146)
Substitution of 1205910= (120575 + 120585)119896 into (66) yields (122) which
proves that (122) holds when 119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) ge
1 + 119886119896
In what follows we prove that (121) also holds when119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) = 1 + 119886
119896 Substitution of 119886
119896=
120582119896cos 120575 into 119896120582
119896[sin 120575 sin(120575119896)] cos(120575119896) = 1 + 119886
119896leads to
120582119896[(119896 minus 1) cos 120575 + 119896 sin (120575 minus 120575119896)
sin (120575119896)] = 1 (147)
As we mentioned earlier relation (142) holds for all wave-forms of type (121) Substituting (142) into (147) we obtain
120582119896[(119896 minus 1) cos 120585 + 119896 sin (120585 minus 120585119896)
sin (120585119896)] = 1 (148)
This expression can be rearranged as
120582119896
119896 sin ((119896 minus 1) 120585119896)sin 120585119896
= 1 minus (119896 minus 1) 120582119896cos 120585 (149)
On the other hand for waveforms of type (122) according to(68) relations (148) and (149) also hold Substitution of 120591
0=
(120575 + 120585)119896 (see (145)) and (67) into (122) leads to
119879119896(120591)
= 120582119896[1 minus cos (120591 minus 120591
0)]
sdot [119896 sin ((119896 minus 1) 120585119896)
sin 120585119896minus 2
119896minus1
sum
119899=1
(119896 minus 119899) cos (119899 (120591 minus 1205910) + 120585)]
(150)
Furthermore substitution of (142) into (145) implies that1205910
= 0 Finally substitution of 1205910
= 0 and (149) into(150) leads to (141) Therefore (141) holds when 119896120582
119896[sin 120575
sin(120575119896)] cos(120575119896) = 1 + 119886119896 which in turn shows that (121)
holds when 119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) = 1 + 119886
119896 This
completes the proof
18 Mathematical Problems in Engineering
52 Nonnegative Waveforms with Maximal Absolute Valueof Coefficient 119886
1for 119896 = 3 Nonnegative waveform of type
(35) for 119896 = 3 is widely used in PA design (eg see [10])In this subsection we illustrate results of Section 51 for thisparticular case The case 119896 = 2 is presented in detail in [12]
Coefficients of fundamental harmonic of nonnegativewaveform of type (35) with 119896 = 3 and maximal absolutevalue of coefficient 119886
1le 0 for prescribed coefficients 119886
3and
1198873(1205823= radic11988623+ 11988723) according to (123) (124) (134) (94) and
(120) are equal to
1198861= minus1 minus 119886
3 119887
1= minus3119887
3 (151)
if 12058223le [(1 minus 2119886
3)3]3
1198861= minus1205821cos(120575
3) 119887
1= minus1205821sin(120575
3) (152)
where 1205821= 3(
3radic1205823minus 1205823) and 120575 = atan 2(119887
3 1198863) if [(1 minus
21198863)3]3
le 1205822
3le 1The line 1205822
3= [(1minus2119886
3)3]3 (see case 119896 = 3
in Figure 11) separates points representing waveforms withcoefficients satisfying (151) from points representing wave-forms with coefficients satisfying (152) Waveforms describedby (151) for 1205822
3lt [(1 minus 2119886
3)3]3 have exactly one zero at
1205910= 0 Waveforms described by (151) and (152) for 1205822
3= [(1 minus
21198863)3]3 also have zero at 120591
0= 0 These waveforms as a rule
have exactly two zeros However there are two exceptionsone related to the maximally flat nonnegative waveform withcoefficients 119886
1= minus98 119886
3= 18 and 119887
1= 1198873= 0 which
has only one zero and the other related to the waveform withcoefficients 119886
1= 0 119886
3= minus1 and 119887
1= 1198873= 0 which has three
zerosWaveforms described by (152) for [(1minus21198863)3]3
lt 1205822
3lt
1 have two zeros Waveforms with 1205823= 1 have only third
harmonic (fundamental harmonic is zero)Plot of contours of maximal absolute value of coefficient
1198861 1198861le 0 for prescribed coefficients 119886
3and 1198873is presented
in Figure 12 According to Remark 25 the waveform withmaximum absolute value of 119886
1le 0 is fully described with
the following coefficients 1198861
= minus2radic3 1198863
= radic39 and1198871= 1198873= 0 This waveform has two zeros at plusmn1205876
Two examples of nonnegative waveforms for 119896 = 3
and maximal absolute value of coefficient 1198861 1198861le 0 with
prescribed coefficients 1198863and 1198873are presented in Figure 13
One waveform corresponds to the case 12058223lt [(1 minus 2119886
3)3]3
(solid line) and the other to the case 12058223gt [(1 minus 2119886
3)3]3
(dashed line)Thewaveform represented by solid line has onezero and its coefficients are 119886
3= minus01 119887
3= 01 119886
1= minus09
and 1198871= minus03 Dashed line corresponds to the waveform
having two zeros with coefficients 1198863= minus01 119887
3= 03 119886
1=
minus08844 and 1198871= minus06460 (case 1205822
3gt [(1 minus 2119886
3)3]3)
6 Nonnegative Cosine Waveforms withat Least One Zero
Nonnegative cosine waveforms have proved to be of impor-tance for waveform modelling in PA design (eg see [10])In this section we consider nonnegative cosine waveforms
1
05
0
minus05
minus1
minus1 minus05 0 05 1
Coefficient a3
Coe
ffici
entb
3
02
04
06
08
10
11
Figure 12 Contours ofmaximal absolute value of coefficient 1198861 1198861le
0 as a function of 1198863and 1198873
2
1
0
0 1 2 3 4
Angle 120591120587
Wav
efor
ms
a3 = minus01 b3 = 01
a3 = minus01 b3 = 03
Figure 13 Nonnegative waveforms for 119896 = 3 and maximal absolutevalue of 119886
1 1198861le 0 with prescribed coefficients 119886
3and 1198873
containing fundamental and 119896th harmonic with at least onezero
Cosine waveform with dc component fundamental and119896th harmonic can be obtained from (35) by setting 119887
1= 119887119896=
0 that is
119879119896(120591) = 1 + 119886
1cos 120591 + 119886
119896cos 119896120591 (153)
In Section 61 we provide general description of non-negative cosine waveforms of type (153) with at least onezero We show that nonnegative cosine waveforms with atleast one zero coincide with nonnegative cosine waveformswith maximal absolute value of coefficient 119886
1for prescribed
coefficient 119886119896 In Section 62 we illustrate results of Section 61
for particular case 119896 = 3
61 Nonnegative Cosine Waveforms with at Least One Zerofor 119896 ge 2 Amplitudes of fundamental and 119896th harmonic
Mathematical Problems in Engineering 19
of cosine waveform of type (153) are 1205821= |1198861| and 120582
119896=
|119886119896| respectively According to (42) for nonnegative cosine
waveforms of type (153) the following relation holds
minus1 le 119886119896le 1 (154)
This explains why 119896th harmonic coefficient 119886119896in Proposi-
tion 26 goes through interval [minus1 1]Waveforms (153) with 119886
1ge 0 can be obtained from
waveforms with 1198861le 0 by shifting by 120587 and therefore with-
out loss of generality we can assume that 1198861le 0
Proposition 26 Each nonnegative cosine waveform of type(153) with 119886
1le 0 and at least one zero can be represented as
119879119896(120591) = [1 minus cos 120591] [1 minus (119896 minus 1) 119886
119896minus 2119886119896
119896minus1
sum
119899=1
(119896 minus 119899) cos 119899120591]
(155)
if minus1 le 119886119896le 1(119896
2
minus 1) or
119879119896(120591) = 119886
119896[1 minus cos (120591 minus 120591
0)] [1 minus cos (120591 + 120591
0)]
sdot [1198880+ 2
119896minus2
sum
119899=1
119888119899cos 119899120591]
(156)
where
119888119899=sin ((119896 minus 119899) 120591
0) cos 120591
0minus (119896 minus 119899) cos ((119896 minus 119899) 120591
0) sin 120591
0
sin31205910
(157)
119886119896=
sin 1205910
119896 sin (1198961205910) cos 120591
0minus cos (119896120591
0) sin 120591
0
(158)
100381610038161003816100381612059101003816100381610038161003816 le
120587
119896 (159)
if 1(1198962 minus 1) le 119886119896le 1
Remark 27 Identity [1minuscos(120591minus1205910)][1minuscos(120591+120591
0)] = [cos 120591
0minus
cos 120591]2 implies that (156) can be rewritten as
119879119896(120591) = 119886
119896[cos 120591
0minus cos 120591]2 [119888
0+ 2
119896minus2
sum
119899=1
119888119899cos 119899120591] (160)
Furthermore substitution of (157) into (160) leads to
119879119896(120591) = 119886
119896[cos 120591
0minus cos 120591]
sdot [(119896 minus 1) sin 119896120591
0
sin 1205910
minus 2
119896minus1
sum
119899=1
sin ((119896 minus 119899) 1205910)
sin 1205910
cos 119899120591]
(161)
Remark 28 All nonnegative cosine waveforms of type (153)with at least one zero and 119886
1le 0 except one of them can be
represented either in form (155) or form (156)This exceptionis maximally flat cosine waveform with 119886
1lt 0 which can be
obtained from (155) for 119886119896= 1(119896
2
minus 1) or from (156) for 1205910=
0 Maximally flat cosine waveform with 1198861lt 0 can also be
obtained from (70) by setting 1205910= 0 Furthermore setting
1205910= 0 in (71) leads to maximally flat cosine waveforms for
119896 le 4 and 1198861lt 0
Remark 29 Nonnegative cosine waveform of type (155) with1198861lt 0 and minus1 lt 119886
119896le 1(119896
2
minus 1) has exactly one zero at120591 = 0 Nonnegative cosine waveform described by (156) with1198861lt 0 and 1(1198962 minus 1) lt 119886
119896lt 1 has two zeros at plusmn120591
0 where
0 lt |1205910| lt 120587119896 For 119886
119896= ∓1 nonnegative cosine waveform
of type (153) reduces to 119879119896(120591) = 1 ∓ cos 119896120591 (clearly these two
waveforms both have 119896 zeros)
Remark 30 Transformation of (155) into an additive formleads to the following relation
1198861= minus1 minus 119886
119896 (162)
where minus1 le 119886119896le 1(119896
2
minus1) Similarly transformation of (156)leads to the following relation
1198861= minus119886119896
119896 sin 1198961205910
sin 1205910
(163)
where 119886119896is given by (158) 1(1198962minus1) le 119886
119896le 1 and |120591
0| le 120587119896
Notice that coefficients of maximally flat cosine waveformnamely 119886
119896= 1(119896
2
minus1) and 1198861= minus1198962
(1198962
minus1) satisfy relation(162) They also satisfy relation (163) for 120591
0= 0
Remark 31 Nonnegative cosine waveforms of type (153) withat least one zero coincide with nonnegative cosine waveformswith maximal absolute value of coefficient 119886
1for prescribed
coefficient 119886119896
In proving that Remark 31 holds notice that expression(155) can be obtained from (121) by setting 119887
119896= 0 Fur-
thermore if 119886119896
ge 0 then 120582119896
= 119886119896 which together
with 119887119896
= 0 and (118) implies 120575 = 0 In this case119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) le 1 + 119886
119896becomes 1198962119886
119896le 1 +
119886119896 On the other hand if 119886
119896lt 0 then 120582
119896= minus119886
119896 which
together with 119887119896= 0 and (118) implies |120575| = 120587 In this case
119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) le 1 + 119886
119896becomes 0 le 1 + 119886
119896
Therefore every nonnegative cosine waveform of type (155)has maximal absolute value of coefficient 119886
1for prescribed
coefficient 119886119896 when minus1 le 119886
119896le 1(119896
2
minus 1)Let us now show that expression (156) can be obtained
from (122) by setting 119887119896= 0 and 119886
119896gt 0 For waveforms
of type (122) according to (118) 119887119896= 0 and 119886
119896gt 0 imply
120575 = 0 and 120582119896= 119886119896 Substitution of 120582
119896= 119886119896and 120575 = 0 into
119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) ge 1+119886
119896leads to 119886
119896ge 1(119896
2
minus1)Furthermore substitution of 120575 = 0 into (145) yields 120591
0= 120585119896
Insertion of 120582119896= 119886119896 120575 = 0 and 120591
0= 120585119896 into (122) leads
to (156) Therefore every nonnegative cosine waveform oftype (156) has maximal absolute value of coefficient 119886
1for
prescribed coefficient 119886119896 when 1(1198962 minus 1) le 119886
119896le 1
Proof of Proposition 26 Let us start with nonnegative cosinewaveform of type (153) with 120582
119896= |119886119896| = 1 According to
Remark 7 120582119896= |119886119896| = 1 implies that 120582
1= |1198861| = 0
Substitution of 119886119896
= minus1 into (155) and using (A2) (seeAppendices) lead to 119879
119896(120591) = 1 minus cos 119896120591 Consequently (155)
holds for 119886119896= minus1 On the other hand substitution of 119886
119896= 1
into (158) yields |1205910| = 120587119896 Furthermore substitution of
119886119896= 1 and 120591
0= 120587119896 (or 120591
0= minus120587119896) into (156) along
with performing all multiplications and using (A2) leads to
20 Mathematical Problems in Engineering
119879119896(120591) = 1 + cos 119896120591 Consequently (156)ndash(158) hold for 119886
119896= 1
and |1205910| = 120587119896
It is easy to see that 120582119896= |119886119896| lt 1 and 119879
119896(1205910) = 0 for some
1205910imply 120582
1= |1198861| = 0 Therefore in what follows we assume
that |119886119896| = 1 and 119886
1lt 0
Cosine waveforms are even functions of 120591 Therefore ifnonnegative cosine waveform has exactly one zero it has to beeither at 0 or at 120587 On the other hand if nonnegative cosinewaveform with 119886
1= 0 has exactly two zeros then these zeros
are placed at plusmn1205910 such that 120591
0is neither 0 nor 120587
In order to prove that (155) holds for minus1 lt 119886119896le 1(1minus119896
2
)let us start by referring to the description (38) of nonnegativewaveformswith at least one zero As wementioned earlier fornonnegative cosine waveformwith exactly one zero (denotedby 1205910) it is either 120591
0= 0 or 120591
0= 120587 Therefore in both cases
sin 1205910= 0 Substitution of sin 120591
0= 0 into (43) together with
1198861
= 0 and 120582119896= |119886119896| lt 1 leads to
1205910= 0 (164)
Clearly 1205910= 0 119887
1= 0 and 119887
119896= 0 according to (44) and (46)
imply 120582119896sin 120585 = 0 Since 120582
119896= |119886119896| it follows that |119886
119896| sin 120585 = 0
also holds which further implies 120582119896= 119886119896= 0 or sin 120585 =
0 In the case when 120582119896= 119886119896= 0 from (164) and (43) we
obtain 1198861= minus1 which further implies that 119879
119896(120591) = 1 minus cos 120591
Consequently (155) holds for 119886119896= 0 In the case when sin 120585 =
0 from (164) and (45) we obtain 119886119896= 120582119896if 120585 = 0 or 119886
119896= minus120582119896
if 120585 = 120587 Relations 119886119896= 120582119896and 120585 = 0 according to (40) imply
that 0 le 119886119896le 1(1 minus 119896
2
) Substitution of 120585 = 0 120582119896= 119886119896 and
(164) into (38) leads to (155) which proves that (155) holdsfor 0 le 119886
119896le 1(1 minus 119896
2
) On the other hand relations 119886119896=
minus120582119896and 120585 = 120587 according to (40) imply that minus1 lt 119886
119896le 0
Substitution of 120585 = 120587 120582119896= minus119886119896 and (164) into (38) also leads
to (155) which proves that (155) also holds for minus1 lt 119886119896le 0
Consequently (155) holds for minus1 lt 119886119896le 1(1 minus 119896
2
)In what follows we first prove that (156)-(157) hold for
1(1 minus 1198962
) lt 119886119896lt 1 For this purpose let us start with non-
negative waveforms with two zeros described by (66) As wementioned before nonnegative cosine waveforms with twozeros have zeros at 120591
0and minus120591
0 such that 120591
0= 0 and 120591
0= 120587
Relations 1198861
lt 0 and 1198871
= 0 according to (84) implycos(1205910minus 120585119896) = 1 and therefore
120585
119896= 1205910 (165)
From 120585119896 = 1205910and 0 lt |120585| lt 120587 it follows that 0 lt |120591
0| lt
120587119896 Insertion of 120585119896 = 1205910into (45) yields 119886
119896= 120582119896 Relations
119886119896= 120582119896and (82) imply that 1(1 minus 1198962) lt 119886
119896lt 1 Substitution
of 120582119896= 119886119896and 120585119896 = 120591
0into (66)ndash(68) leads to (156)ndash(158)
which proves that (156)ndash(158) hold for 1(1 minus 1198962
) lt 119886119896lt 1
and 0 lt |1205910| lt 120587119896
Finally substitution of 119886119896= 1(1 minus 119896
2
) and 1205910= 0 into
(161) leads to
119879119896(120591) =
[1 minus cos 120591](1 minus 1198962)
[119896 (119896 minus 1) minus 2
119896minus1
sum
119899=1
(119896 minus 119899) cos 119899120591]
(166)
Waveform (166) coincides with waveform (155) for 119886119896
=
1(1 minus 1198962
) which in turn proves that (156) holds for 119886119896=
1(1 minus 1198962
) and 1205910= 0 This completes the proof
62 Nonnegative CosineWaveforms with at Least One Zero for119896 = 3 In this subsection we consider nonnegative cosinewaveforms with at least one zero for 119896 = 3 (for case 119896 = 2
see [12])Cosine waveform with fundamental and third harmonic
reads
1198793(120591) = 1 + 119886
1cos 120591 + 119886
3cos 3120591 (167)
For 1198861le 0 and minus1 le 119886
3le 18 according to (155) non-
negative cosine waveform of type (167) with at least one zerocan be expressed as
1198793(120591) = (1 minus cos 120591) [1 minus 2119886
3(1 + 2 cos 120591 + cos 2120591)] (168)
From 1198793(120591 + 120587) = 2 minus 119879
3(120591) it immediately follows that for
1198861ge 0 and minus18 le 119886
3le 1 119879
3(120591) can be expressed as
1198793(120591) = (1 + cos 120591) [1 + 2119886
3(1 minus 2 cos 120591 + cos 2120591)] (169)
For 1198861le 0 and 18 le 119886
3le 1 from (158) it follows that 119886
3=
[8cos31205910]minus1 This relation along with (160) and (157) further
implies that 1198793(120591) can be expressed as
1198793(120591) =
[cos 1205910minus cos 120591]2 [2 cos 120591
0+ cos 120591]
2cos31205910
(170)
providing that |1205910| le 1205873 From 119879
3(120591 + 120587) = 2 minus 119879
3(120591) it
follows that (170) also holds for 1198861ge 0 and minus1 le 119886
3le minus18
providing that 1205910isin [21205873 41205873]
Maximally flat nonnegative cosinewaveformof type (167)with 119886
1lt 0 (minimum at 120591
0= 0) reads 119879
3(120591) = [1 minus
cos 120591]2[1 + (12) cos 120591] Dually maximally flat nonnegativecosine waveform with 119886
1gt 0 (minimum at 120591
0= 120587) reads
1198793(120591) = [1 + cos 120591]2[1 minus (12) cos 120591]In what follows we provide relations between coefficients
1198861and 1198863of nonnegative cosine waveforms of type (167) with
at least one zeroFor 1198861le 0 conversion of (168) into an additive form
immediately leads to the following relation
1198861= minus1 minus 119886
3for minus 1 le 119886
3le1
8 (171)
Conversion of (170) into an additive form leads to 1198861
=
minus31198863(2 cos 2120591
0+ 1) which can be also expressed as 119886
1=
minus31198863(4cos2120591
0minus 1) For 119886
1le 0 relations |120591
0| le 1205873 119886
1=
minus31198863(4cos2120591
0minus 1) and 119886
3= [8cos3120591
0]minus1 lead to
1198861= minus3 [ 3radic119886
3minus 1198863] for 1
8le 1198863le 1 (172)
Similarly for 1198861ge 0 conversion of (169) into an additive form
leads to the following relation
1198861= 1 minus 119886
3for minus
1
8le 1198863le 1 (173)
Mathematical Problems in Engineering 21
1
15
15
05
0
minus05
minus15minus15
minus1
minus1 minus05 0 05 1
Coefficient a1
Coe
ffici
enta
3 a1= minus
3(3radica3minus a3)
a1 =
minus1 minus a3
a1 =
1 minusa3
a1 =3(3radic|a3
| +a3)
Figure 14 Parameter space of cosine waveforms for 119896 = 3
For waveform of type (170) with 1198861ge 0 relations 120591
0isin [21205873
41205873] 1198861= minus3119886
3(4cos2120591
0minus 1) and 119886
3= [8cos3120591
0]minus1 lead to
1198861= 3 [
3radic10038161003816100381610038161198863
1003816100381610038161003816 + 1198863] for minus 1 le 1198863le minus
1
8 (174)
Every cosine waveform of type (167) corresponds to apair of real numbers (119886
1 1198863) and vice versa Points (119886
1 1198863)
in grey area in Figure 14 correspond to nonnegative cosinewaveforms for 119896 = 3 The points at the boundary of grey areacorrespond to nonnegative cosinewaveformswith at least onezero A number of shapes of nonnegative cosine waveformswith 119896 = 3 and at least one zero plotted on interval [minus120587 120587]are also presented in Figure 14 The boundary of grey area inFigure 14 consists of four line segments described by relations(171)ndash(174) The common point of line segments (172) and(173) is cusp point with coordinates 119886
1= 0 and 119886
3= 1
Another cusp point with coordinates 1198861= 0 and 119886
3= minus1
is the common point of line segments (171) and (174) Thecommon point of line segments (171)-(172) has coordinates(minus98 18) and common point of line segments (173)-(174)has coordinates (98 minus18) These points are representedby white circle dots and they correspond to maximallyflat cosine waveforms (eg see [21]) White triangle dotswith coordinates (2radic3 minusradic39) and (minus2radic3radic39) refer tothe nonnegative cosine waveforms with maximum value ofamplitude of fundamental harmonic
7 Four Case Studies of Usage of NonnegativeWaveforms in PA Efficiency Analysis
In this section we provide four case studies of usage ofdescription of nonnegative waveforms with fundamental and119896th harmonic in PA efficiency analysis In first two casestudies to be presented in Section 71 voltage is nonnegativewaveform with fundamental and second harmonic with atleast one zero In remaining two case studies to be consideredin Section 72 voltage waveform contains fundamental andthird harmonic
i
v Load
Vdc
Idc
Lch
vL
iL
Cb
+ +in
Figure 15 Generic PA circuit diagram
Let us consider generic PA circuit diagram as shown inFigure 15 We assume here that voltage and current wave-forms at the transistor output are
V (120579) = 1 + 1198861V cos 120579 + 1198871V sin 120579 + 119886119896V cos 119896120579 + 119887119896V sin 119896120579
119894 (120579) = 1 + 1198861119894cos 120579 +
infin
sum
119899=2
119886119899119894cos 119899120579
(175)
where 120579 stands for 120596119905 Both waveforms are normalized in thesense that dc components of voltage and current are 119881dc =
1 and 119868dc = 1 respectively Under assumption that blockingcapacitor 119862
119887behaves as short-circuit at the fundamental and
higher harmonics current and voltage waveforms at the loadare
V119871(120579) = 119886
1V cos 120579 + 1198871V sin 120579 + 119886119896V cos 119896120579 + 119887119896V sin 119896120579
119894119871(120579) = minus119886
1119894cos 120579 minus
infin
sum
119899=2
119886119899119894cos 119899120579
(176)
In terms of coefficients of voltage and current waveforms theload impedance at fundamental harmonic is 119911
1= minus(119886
1V minus
1198951198871V)1198861119894 whereas load impedance at 119896th harmonic is 119911
119896=
minus(119886119896Vminus119895119887119896V)119886119896119894 All other harmonics are short-circuited (119911
119899=
0 for 119899 = 1 and 119899 = 119896) Time average output power of PA (egsee [10]) with waveform pair (175) at fundamental frequencycan be expressed as
1198751= minus
11988611198941198861V
2 (177)
For normalized waveforms (175) with 119881dc = 1 and 119868dc = 1dc power is 119875dc = 1 Consequently PA efficiency 120578 = 119875
1119875dc
(eg see [10 26]) is equal to
120578 = minus11988611198941198861V
2 (178)
Thus time average output power 1198751of PA with pair of nor-
malized waveform (175) is equal to efficiency (178)Power utilization factor (PUF) is defined [26] as ldquothe
ratio of power delivered in a given situation to the power
22 Mathematical Problems in Engineering
delivered by the same device with the same supply voltagein Class A moderdquo Since the output power in class-A modeis 1198751class-A = max[V(120579)] sdotmax[119894(120579)]8 (eg see [9]) it follows
that power utilization factor PUF = 11987511198751class-A for PA with
pair of normalized waveforms (175) can be expressed as
PUF =8120578
max [V (120579)] sdotmax [119894 (120579)] (179)
71 NonnegativeWaveforms for 119896 = 2 in PAEfficiencyAnalysisIn this subsection we provide two case studies of usage ofdescription of nonnegative waveforms with fundamental andsecond harmonic (119896 = 2) in PA efficiency analysis For moreexamples of usage of descriptions of nonnegative waveformswith fundamental and second harmonic in PA efficiencyanalysis see [12]
Case Study 71 In this case study we consider efficiency of PAfor given second harmonic impedance providing that voltageis nonnegative waveform with fundamental and second har-monic and current is ldquohalf-sinerdquo waveform frequently used inefficiency analysis of classical PA operation (eg see [10])
Standard model of current waveform for classical PAoperation has the form (eg see [10 26])
119894119863(120579) =
119868119863[cos 120579 minus cos(120572
2)] |120579| le
120572
2
0120572
2le |120579| le 120587
(180)
where 120572 is conduction angle and 119868119863
gt 0 Since 119894119863(120579) is
even function it immediately follows that its Fourier seriescontains only dc component and cosine terms
119894119863(120579) = 119868dc +
infin
sum
119899=1
119868119899cos 119899120579 (181)
The dc component of the waveform (180) is
119868dc =119868119863120572
2120587[sinc(120572
2) minus cos(120572
2)] (182)
where sinc119909 = (sin119909)119909 The coefficient of the fundamentalharmonic component reads
1198681=119868119863120572
2120587(1 minus sinc120572) (183)
and the coefficient of 119899th harmonic component can bewrittenin the form
119868119899=119868119863
119899120587[sin ((119899 minus 1) 1205722)
(119899 minus 1)minussin ((119899 + 1) 1205722)
(119899 + 1)] 119899 ge 2
(184)
For ldquohalf-sinerdquo current waveform conduction angle is equalto 120587 (class-B conduction angle) According to (182) thisfurther implies that 119868dc = 119868
119863120587 To obtain normalized form of
waveform (180) we set 119868dc = 1which implies that 119868119863= 120587 Fur-
thermore substitution of 120572 = 120587 and 119868119863= 120587 in (180) leads to
119894 (120579) =
120587 cos 120579 |120579| lt120587
2
0120587
2lt |120579| le 120587
(185)
Similarly substitution of 119868119863= 120587 and 120572 = 120587 into (183) and
(184) leads to the coefficients of waveform (185) Coefficientsof fundamental and second harmonic respectively are
1198861119894=120587
2 119886
2119894=2
3 (186)
On the other hand voltage waveform of type (35) for 119896 =2 reads
V (120579) = 1 + 1198861V cos 120579 + 1198871V sin 120579 + 1198862V cos 2120579 + 1198872V sin 2120579
(187)
This waveform contains only fundamental and second har-monic and therefore all harmonics of order higher thantwo are short-circuited (119911
119899= 0 for 119899 gt 2) For current
voltage pair (185) and (187) load impedance at fundamentalharmonic is 119911
1= minus(119886
1V minus 1198951198871V)1198861119894 whereas load impedance
at second harmonic is 1199112= minus(119886
2Vminus1198951198872V)1198862119894 According to ourassumption the load is passive and therefore Re119911
1 gt 0 and
Re1199112 ge 0 which further imply 119886
11198941198861V lt 0 and 119886
21198941198862V le 0
respectivelyIt is easy to see that problem of findingmaximal efficiency
of PA with current-voltage pair (185) and (187) for prescribedsecond harmonic impedance can be reduced to the problemof finding voltage waveform of type (187) with maximal coef-ficient |119886
1V| for prescribed coefficients of second harmonic(see Section 5)
The following algorithm (analogous to Algorithm 22presented in [12]) provides the procedure for calculation ofmaximal efficiency with current-voltage pair (185) and (187)for prescribed second harmonic impedance The definitionof function atan 2(119910 119909) which appears in the step (iii) of thefollowing algorithm is given by (105)
Algorithm 32 (i) Choose 1199112= 1199032+1198951199092such that |119911
2| le 1|119886
2119894|
(ii) calculate 1198862V minus 1198951198872V = minus119911
21198862119894and 120582
2V = radic11988622V + 1198872
2V(iii) if 2120582
2V le 1 minus 1198862V then calculate 119886
1V = minus1 minus 1198862V and
1198871V = minus2119887
2V else calculate 1205821V = radic81205822V(1 minus 1205822V) 1205790V minus 120585V2 =
(12)atan2(1198872V 1198862V) 1198861V = minus120582
1V cos(1205790V minus 120585V2) and 1198871V =
minus1205821V sin(1205790V minus 120585V2)(iv) calculate efficiency 120578 = minus119886
11198941198861V2
(v) calculate 1199111= minus(119886
1V minus 1198951198871V)1198861119894 and 1199112119899 = 1199112Re119911
1
In this case study coefficients of fundamental and secondharmonic of current waveform are given by (186) Maximalefficiency of PA associated with the waveform pair (185)and (187) as a function of normalized second harmonicimpedance 119911
2119899= 1199112Re119911
1 is presented in Figure 16(a)
As can be seen from Figure 16(a) efficiency of 078 isachieved at the edge of Smith chart where second harmonicimpedance has small resistive part Corresponding PUFcalculated according to (179) is presented in Figure 16(b)Peak efficiency 120578 = 1205874 = 07854 and peak value of PUF = 1
are attained when second harmonic is short-circuited (whichcorresponds to ideal class-B operation [10 26])
For example for second harmonic impedance 1199112= 01 minus
11989505 and currentwaveform (185) fromAlgorithm32 it followsthat 2120582
2V le 1 minus 1198862V Furthermore according to step (iii)
of above algorithm maximal efficiency of PA is attained
Mathematical Problems in Engineering 23
+j2
+j5
+j1
+j2
+j5
minusj2
minusj5
minusj1
minusj2
minusj5
infin0
078
02
05
1 2 5
075 07 065
06 05
075
07
065
06
05
120578 lt 05
(a)
099
095
091
083 075
067
0 59 051
051
099
095
091
083
075 0
67 059
+j5
+j1
+j2
+j5
minusj5
minusj1
minusj2
minusj5
infin
+j2
minusj2
0 02
05
1 2 5
120578 lt 05
(b)
Figure 16 (a) Contours of maximal efficiency of PA and (b) contours of corresponding PUF associated with the waveform pair (185) and(187) as functions of normalized second harmonic impedance 119911
2119899= 1199112Re119911
1
2
3
4
1
0
2 3 410
Wav
efor
ms
Angle 120579120587
i(120579)
(120579)
Figure 17 Waveform pair (185) and (187) that provides maximalefficiency for 119911
2= 01 minus 11989505
with voltage waveform of type (187) with coefficients 1198862V =
minus00667 1198872V = minus03333 119886
1V = minus09333 and 1198871V = 06667 (see
Figure 17) Corresponding efficiency PUF and normalizedsecond harmonic impedance are 120578 = 07330 PUF = 07572and 1199112119899= 01683 minus 11989508415 respectively
On the other hand for second harmonic impedance 1199112=
01 minus 11989508 and current waveform (185) from Algorithm 32 itfollows that 2120582
2V gt 1 minus 1198862V Then according to step (iii) of
above algorithm maximal efficiency is attained with voltagewaveform of type (187) with coefficients 119886
2V = minus00667 1198872V =
minus05333 1198861V = minus09333 and 119887
1V = 10572 (see Figure 18)Efficiency PUF andnormalized secondharmonic impedanceare 120578 = 07330 PUF = 06332 and 119911
2119899= 01683 minus 11989513465
respectively
2
3
4
1
0
2 3 410
Wav
efor
ms
Angle 120579120587
i(120579)
(120579)
Figure 18 Waveform pair (185) and (187) that provides maximalefficiency for 119911
2= 01 minus 11989508
Case Study 72 As another case study let us consider the effi-ciency of PA providing that current waveform is nonnegativecosine waveform up to third harmonic with maximum valueof amplitude of fundamental harmonic [22] (see also [8])
119894 (120579) = 1 +1 + radic5
2cos 120579 + 2radic5
5cos 2120579 + 5 minus radic5
10cos 3120579
(188)
and voltage waveform is nonnegative waveform of type(187) Load impedances at fundamental second and thirdharmonic are 119911
1= minus(119886
1Vminus1198951198871V)1198861119894 1199112 = minus(1198862Vminus1198951198872V)1198862119894 and
1199113= 0 respectively According to our assumption the load
is passive and therefore Re1199111 gt 0 and Re119911
2 ge 0 which
further imply 11988611198941198861V lt 0 and 119886
21198941198862V le 0 respectively
Because current waveform (188) contains only cosineterms and voltage waveform is the same as in previous case
24 Mathematical Problems in Engineering
08
075
07
06
06
065
065
05
05
075 0
7
120578 lt 05
+j2
+j5
+j1
+j2
+j5
minusj2
minusj5
minusj1
minusj2
minusj5
infin0
02
05
1 2 5
(a)
085
08 0
75 07 0
6
05 04
08
075
0405
060
7
120578 lt 05
+j2
+j5
+j1
+j2
+j5
minusj2
minusj5
minusj1
minusj2
minusj5
infin0 02
05
1 2 5
(b)
Figure 19 (a) Contours of maximal efficiency of PA and (b) contours of corresponding PUF associated with the waveform pair (187)-(188)as functions of normalized second harmonic impedance 119911
2119899= 1199112Re119911
1
study the procedure for calculation of maximal efficiency ofPA with waveform pair (187)-(188) is the same as presentedin Algorithm 32 In this case study the coefficients of funda-mental and second harmonic of current waveform are 119886
1119894=
(1 + radic5)2 and 1198862119894= 2radic55 respectively
Maximal efficiency of PA associated with the waveformpair (187)-(188) as a function of normalized secondharmonicimpedance 119911
2119899= 1199112Re119911
1 is presented in Figure 19(a)
Efficiency of 08 is achieved at the edge of Smith chart wheresecond harmonic impedance has small resistive part Thetheoretical upper bound 120578 = (1 + radic5)4 asymp 08090 isattained when second harmonic is short-circuitedWhen thisupper bound is reached both second and third harmonic areshort-circuited which implies that we are dealing with finiteharmonic class-C [6 8] or dually when current and voltageinterchange their roles with finite harmonic inverse class-C[6 9] Corresponding PUF calculated according to (179) ispresented in Figure 19(b) Peak value of PUF asymp 08541 isattained when second harmonic is short-circuited
For example for second harmonic impedance 1199112
=
007 minus 11989504 and current waveform (188) from Algorithm 32 itfollows that 2120582
2V le 1 minus 1198862V Furthermore according to step
(iii) of Algorithm 32 maximal efficiency of PA is attainedwith voltage waveform of type (187) with coefficients 119886
2V =
minus00626 1198872V = minus03578 119886
1V = minus09374 and 1198871V = 07155 (see
Figure 20) Corresponding efficiency PUF and normalizedsecond harmonic impedance are 120578 = 07584 PUF = 06337and 1199112119899= 01208 minus 11989506904 respectively
On the other hand for 1199112= 005 minus 11989507 and current
waveform (187) it follows that 21205822V gt 1minus119886
2VThen accordingto step (iii) of Algorithm 32 themaximal efficiency is attainedwith voltage waveform of type (187) with coefficients 119886
2V =
minus00447 1198872V = minus06261 119886
1V = minus09318 and 1198871V = 10007 (see
Figure 21) Efficiency PUF and normalized second harmonic
2
3
4
1
0
2 3 410
Wav
efor
ms
Angle 120579120587
i(120579)
(120579)
Figure 20 Waveform pair (187)-(188) that provides maximal effi-ciency for 119911
2= 007 minus 11989504
impedance are 120578 = 07538 PUF = 05314 and 1199112119899= 00868minus
11989512156 respectively
72 Nonnegative Waveforms for 119896 = 3 in PA EfficiencyAnalysis In this subsection we provide another two casestudies of usage of description of nonnegative waveforms inPA efficiency analysis this time with fundamental and thirdharmonic (119896 = 3)
Case Study 73 Let us consider current-voltage pair such thatvoltage is nonnegative waveform with fundamental and thirdharmonic
V (120579) = 1 + 1198861V cos 120579 + 1198871V sin 120579 + 1198863V cos 3120579 + 1198873V sin 3120579
(189)
Mathematical Problems in Engineering 25
2
3
4
1
0
2 3 410
Wav
efor
ms
Angle 120579120587
i(120579)
(120579)
Figure 21 Waveform pair (187)-(188) that provides maximal effi-ciency for 119911
2= 005 minus 11989507
and current is nonnegative cosine waveform given by (188)Load impedances at fundamental second and third har-monic are 119911
1= minus(119886
1V minus 1198951198871V)1198861119894 1199112 = 0 and 119911
3= minus(119886
3V minus
1198951198873V)1198863119894 respectively According to our assumption the load
is passive and therefore Re1199111 gt 0 and Re119911
3 ge 0 which
further imply 11988611198941198861V lt 0 and 119886
31198941198863V le 0
In this subsection we consider the problem of findingmaximal efficiency of PA with waveform pair (188)-(189)for given third harmonic impedance As we mentionedearlier problem of finding maximal efficiency of PA withcurrent-voltage pair (188)-(189) for prescribed third har-monic impedance can be reduced to the problem of findingvoltage waveform of type (189) withmaximal coefficient |119886
1V|for prescribed coefficients of third harmonic (see Section 52)
The following algorithm provides the procedure forcalculation of maximal efficiency with current-voltage pair(188)-(189) The definition of function atan 2(119910 119909) whichappears in step (iii) of the following algorithm is given by(105)
Algorithm 33 (i) Choose 1199113= 1199033+1198951199093such that |119911
3| le 1|119886
3119894|
(ii) calculate 1198863V minus 1198951198873V = minus119911
31198863119894and 120582
3V = radic11988623V + 1198872
3V(iii) if 271205822
3V le (1 minus 21198863V)3 then calculate 119886
1V = minus1 minus 1198863V
and 1198871V = minus3119887
3V else calculate 1205821V = 3(3radic1205823V minus 120582
3V) 1205790V minus120585V3 = (13)atan 2(119887
3V 1198863V) 1198861V = minus1205821V cos(1205790V minus 120585V3) and
1198871V = minus120582
1V sin(1205790V minus 120585V3)(iv) calculate efficiency 120578 = minus119886
11198941198861V2
(v) calculate 1199111= minus(119886
1V minus 1198951198871V)1198861119894 and 1199113119899 = 1199113Re119911
1
In this case study coefficients of fundamental and thirdharmonic of current waveform are 119886
1119894= (1 + radic5)2 and
1198863119894= (5 minus radic5)10 respectively For the waveform pair (188)-
(189) maximal efficiency of PA as a function of normalizedthird harmonic impedance 119911
3119899= 1199113Re119911
1 is presented in
Figure 22 Efficiency of 08 is reached when third harmonicimpedance has small resistive part Peak efficiency 120578 = (1 +
radic5)4 asymp 08090 is achieved when third harmonic is short-circuited
+j2
+j5
+j1
+j2
+j5
minusj2
minusj5
minusj1
minusj2
minusj5
infin0
08
08
075
075
07
07
06
05
02
05
1 2
Figure 22 Contours of maximal efficiency of PA associated withthe waveform pair (188)-(189) as a function of normalized thirdharmonic impedance 119911
3119899= 1199113Re119911
1
For the present case study in what follows we show thatpower utilization factor is proportional to efficiency For volt-age waveform of type (189) it is easy to see that V(120579 + 120587) =
2minusV(120579) holdsThis relation along with the fact that waveformV(120579) that provides maximal efficiency has at least one zeroimplies that max[V(120579)] = 2 On the other hand currentwaveform (188) is cosine waveform with positive coefficientsand therefore max[119894(120579)] = 119894(0) = 2 + 4radic5 Consequentlyaccording to (179) the following relation holds
PUFCase study 73 = 2 (5 minus 2radic5) 120578 = 10557120578 (190)
Clearly the ratio PUF120578 is constant and therefore in this casestudy PUF can be easily calculated from the correspondingefficiency Accordingly peak efficiency and peak value ofPUFCase study 73 = 3radic52 minus 52 = 08541 are attained forthe same voltage waveform (when third harmonic is short-circuited)
In the first example current waveform (188) and 1199113=
02 minus 11989505 imply that 2712058223V le (1 minus 2119886
3V)3 Then according
to Algorithm 33 the voltage waveform of type (189) thatprovides maximal efficiency has the following coefficients1198863V = minus00553 119887
3V = minus01382 1198861V = minus09447 and 119887
1V =
04146 (see Figure 23) Efficiency PUF and normalized thirdharmonic impedance are 120578 = 07643 PUF = 08069 and1199113119899= 03425 minus 11989508564 respectivelyIn the second example current waveform (188) and 119911
3=
01 minus 11989511 imply that 2712058223V gt (1 minus 2119886
3V)3 Then according
to Algorithm 33 the voltage waveform of type (189) thatprovides maximal efficiency has the following coefficients1198863V = minus00276 119887
3V = minus03040 1198861V = minus09391 and 119887
1V =
05807 (see Figure 24) Efficiency PUF and normalized thirdharmonic impedance are 120578 = 07598 PUF = 08021 and1199113119899= 01723 minus 11989518952 respectively
26 Mathematical Problems in Engineering
2
3
4
1
0
2 3 410
Wav
efor
ms
Angle 120579120587
i(120579)
(120579)
Figure 23 Waveform pair (188)-(189) that provides maximal effi-ciency for 119911
3= 02 minus 11989505
2
3
4
1
0
2 3 410
Wav
efor
ms
Angle 120579120587
i(120579)
(120579)
Figure 24 Waveform pair (188)-(189) that provides maximal effi-ciency for 119911
3= 01 minus 11989511
Case Study 74 In this case study let us consider current-voltage pair where current is normalized waveform of type(180) with conduction angle 120572 = 115120587 (207∘) and voltageis nonnegative waveform of type (189) Substitution of 120572 =
115120587 and 119868dc = 1 into (182) leads to 119868119863
= 22535Furthermore substitution of 120572 = 115120587 and 119868
119863= 22535 into
(180) leads to
119894 (120579)
=
22535 [cos 120579 minus cos(1151205872
)] |120579| le115120587
2
0115120587
2le |120579| le 120587
(191)
Similarly substitution of 120572 = 115120587 and 119868119863
= 22535 into(183) and (184) for 119899 = 3 yields coefficients of fundamentaland third harmonic of waveform (191)
1198861119894= 14586 119886
3119894= minus01026 (192)
Because current waveform (191) contains only cosineterms and voltage waveform is the same as in previous case
+j2
+j5
+j1
+j2
+j5
minusj2
minusj5
minusj1
minusj2
minusj5
infin0 02
05
1 2
074
076
08
084
082
078
076
074
Figure 25 Contours of maximal efficiency of PA associated withthe waveform pair (189) and (191) as a function of normalized thirdharmonic impedance 119911
3119899= 1199113Re119911
1
study the procedure for calculation of maximal efficiencyof PA with waveform pair (189)ndash(191) is the same as thatpresented in Algorithm 33 In this case study the coefficientsof fundamental and third harmonic of current waveform aregiven by (192)
For the waveform pair (189) and (191) maximal efficiencyof PA as a function of normalized third harmonic impedance1199113119899= 1199113Re119911
1 is presented in Figure 25 Efficiency of 084
is obtained in vicinity of 1199113119899= 23685 (corresponding to 119911
3=
18750) Peak efficiency 120578 asymp 08421 is achieved for voltagewaveform of type (189) with coefficients 119886
1V = minus2radic3 1198863V =
radic39 and 1198871V = 1198873V = 0
In the course of finding power utilization factor noticethat currentwaveformof type (191) attains itsmaximumvaluefor 120579 = 0 Insertion of max[119894(120579)] = 119894(0) = 278 andmax[V(120579)] = 2 for voltage waveform of type (189) into (179)leads to
PUFCase study 74 = 1439120578 (193)
Again the ratio PUF120578 is constant and PUF can be easilycalculated from the corresponding efficiency Accordinglypeak value of PUFCase study 74 asymp 12118 and peak efficiencyare attained for the same voltage waveform
In the first example current waveform (191) and 1199113=
1 minus 11989502 imply that 2712058223V le (1 minus 2119886
3V)3 Then according to
Algorithm 33 voltage waveform of type (189) which providesmaximal efficiency has coefficients 119886
3V = 01026 1198873V =
00205 1198861V = minus11026 and 119887
1V = minus00616 (see Figure 26)Efficiency PUF and normalized third harmonic impedanceare 120578 = 08042 PUF = 11572 and 119911
3119899= 13228 minus 11989502646
respectivelyIn second example current waveform (191) and 119911
3=
15 minus 11989512 imply that 2712058223V gt (1 minus 2119886
3V)3 Then according to
Algorithm 33 voltage waveform of type (189) which providesmaximal efficiency has coefficients 119886
3V = 01540 1198873V =
Mathematical Problems in Engineering 27
2
3
4
1
0
2 3 410
Wav
efor
ms
Angle 120579120587
i(120579)
(120579)
Figure 26 Waveform pair (189) and (191) that provides maximalefficiency for 119911
3= 1 minus 11989502
2
3
4
1
0
2 3 410
Wav
efor
ms
Angle 120579120587
i(120579)
(120579)
Figure 27 Waveform pair (189) and (191) that provides maximalefficiency for 119911
3= 15 minus 11989512
01232 1198861V = minus11255 and 119887
1V = minus02575 (see Figure 27)Efficiency PUF and normalized third harmonic impedanceare 120578 = 08208 PUF = 11812 and 119911
3119899= 19439 minus 11989515552
respectively
8 Conclusion
In this paper we consider a problem of finding general de-scriptions of various classes of nonnegative waveforms withfundamental and 119896th harmonic These classes include non-negative waveforms with at least one zero nonnegative wave-forms with maximal amplitude of fundamental harmonic forprescribed amplitude of 119896th harmonic nonnegative wave-forms withmaximal coefficient of cosine part of fundamentalharmonic for prescribed coefficients of 119896th harmonic andnonnegative cosine waveforms with at least one zero Mainresults are stated in six propositions (Propositions 1 6 918 22 and 26) four corollaries (Corollaries 2ndash5) twentyremarks and three algorithms Four case studies of usageof closed form descriptions of nonnegative waveforms in PAefficiency analysis are considered in detail in Section 7
Appendices
Here we provide a list of finite sums of trigonometric func-tions used in this paper (Appendix A) and brief account ofthe Chebyshev polynomials (Appendix B)
A List of Some Finite Sums ofTrigonometric Functions
Dirichlet kernel (eg see [27]) is as follows
119863119896minus1
(120591) = 1 + 2
119896minus1
sum
119899=1
cos 119899120591 = sin ((2119896 minus 1) 1205912)sin (1205912)
(A1)
Fejer kernel (eg see [27]) can be expressed in the fol-lowing equivalent forms
119865119896minus1
(120591) =1
119896
119896minus1
sum
119899=0
119863119899(120591) = 1 +
2
119896
119896minus1
sum
119899=1
(119896 minus 119899) cos 119899120591
=(1 minus cos 119896120591)119896 (1 minus cos 120591)
(A2)
Lagrangersquos trigonometric identity (eg see [28]) is as fol-lows
1198781(120591) =
119896minus1
sum
119899=1
sin 119899120591 = sin (1198961205912) sin ((119896 minus 1) 1205912)sin (1205912)
(A3)
In what follows we show that the following three trigono-metric identities also hold
2
119896minus1
sum
119899=1
(119896 minus 119899) sin 119899120591 = 119896 sin 120591 minus sin 1198961205911 minus cos 120591
(A4)
119896minus1
sum
119899=1
cos (119896 minus 2119899) 120591 = sin (119896 minus 1) 120591sin 120591
(A5)
119896minus1
sum
119899=1
119899 (119896 minus 119899) cos (119896 minus 2119899) 120591
=sin (119896120591) cos 120591 minus 119896 cos (119896120591) sin 120591
2sin3120591
(A6)
Denote 1198782(120591) = 2sum
119896minus1
119899=1(119896 minus 119899) sin 119899120591 119878
3(120591) = sum
119896minus1
119899=1cos(119896 minus
2119899)120591 and 1198784(120591) = sum
119896minus1
119899=1119899(119896 minus 119899) cos(119896 minus 2119899)120591
Notice that 1198782(120591) = 2119896119878
1(120591) + 119889119863
119896minus1(120591)119889120591 which
immediately leads to (A4)
28 Mathematical Problems in Engineering
Identity (A5) can be obtained as follows
sin (119896 minus 1) 120591sin 120591
=119890119895(119896minus1)120591
minus 119890minus119895(119896minus1)120591
119890119895120591 minus 119890minus119895120591
= 119890119895119896120591
119890minus2119895120591
minus 119890minus2119895(119896minus1)120591
1 minus 119890minus2119895120591
= 119890119895119896120591
119896minus1
sum
119899=1
119890minus2119895119899120591
=
119896minus1
sum
119899=1
119890119895(119896minus2119899)120591
=
119896minus1
sum
119899=1
cos (119896 minus 2119899) 120591
(A7)
From 4119899(119896 minus 119899) = 1198962
minus (119896 minus 2119899)2 it follows that 4119878
4(120591) =
1198962
1198783(120591) + 119889
2
1198783(120591)119889120591
2 which leads to (A6)
B The Chebyshev Polynomials
The Chebyshev polynomials of the first kind 119881119899(119909) can be
defined by the following relation (eg see [29])
119881119899(119909) = cos 119899120591 when 119909 = cos 120591 (B1)
The Chebyshev polynomials of the second kind 119880119899(119909) can be
defined by the following relation (eg see [29])
119880119899(119909) =
sin (119899 + 1) 120591sin 120591
when 119909 = cos 120591 (B2)
The Chebyshev polynomials satisfy the following recur-rence relations (eg see [29])
1198810(119909) = 1 119881
1(119909) = 119909
119881119899+1
(119909) = 2119909119881119899(119909) minus 119881
119899minus1(119909)
1198800(119909) = 1 119880
1(119909) = 2119909
119880119899+1
(119909) = 2119909119880119899(119909) minus 119880
119899minus1(119909)
(B3)
The first few Chebyshev polynomials of the first and secondkind are 119881
2(119909) = 2119909
2
minus 1 1198813(119909) = 4119909
3
minus 3119909 1198814(119909) = 8119909
4
minus
81199092
+ 1 1198802(119909) = 4119909
2
minus 1 1198803(119909) = 8119909
3
minus 4119909 and 1198804(119909) =
161199094
minus 121199092
+ 1
Conflict of Interests
The authors declare that there is no conflict of interestsregarding the publication of this paper
Acknowledgment
This work is supported by the SerbianMinistry of EducationScience and Technology Development as a part of ProjectTP32016
References
[1] V I Arnolrsquod V S Afrajmovich Y S Ilrsquoyashenko and L PShilrsquonikov Dynamical Systems V Bifurcation Theory and Catas-trophe Theory Springer Berlin Germany 1994
[2] E Polak ldquoOn the mathematical foundations of nondifferen-tiable optimization in engineering designrdquo SIAM Review vol29 no 1 pp 21ndash89 1987
[3] N S Fuzik ldquoBiharmonic modes of a tuned RF power amplifierrdquoRadiotehnika vol 25 no 7 pp 62ndash71 1970 (Russian)
[4] P Colantonio F Giannini G Leuzzi and E Limiti ldquoClass Gapproach for low-voltage high-efficiency PA designrdquo Interna-tional Journal of RF and Microwave Computer-Aided Engineer-ing vol 10 no 6 pp 366ndash378 2000
[5] F H Raab ldquoMaximum efficiency and output of class-F poweramplifiersrdquo IEEE Transactions on Microwave Theory and Tech-niques vol 49 no 6 pp 1162ndash1166 2001
[6] FH Raab ldquoClass-E class-C and class-F power amplifiers basedupon a finite number of harmonicsrdquo IEEE Transactions onMicrowaveTheory and Techniques vol 49 no 8 pp 1462ndash14682001
[7] J D Rhodes ldquoOutput universality inmaximum efficiency linearpower amplifiersrdquo International Journal of Circuit Theory andApplications vol 31 no 4 pp 385ndash405 2003
[8] A Juhas and L A Novak ldquoComments on lsquoClass-E class-Cand classF power amplifier based upon a finite number ofharmonicsrsquordquo IEEE Transactions on Microwave Theory and Tech-niques vol 57 no 6 pp 1623ndash1625 2009
[9] M Roberg and Z Popovic ldquoAnalysis of high-efficiency poweramplifiers with arbitrary output harmonic terminationsrdquo IEEETransactions on Microwave Theory and Techniques vol 59 no8 pp 2037ndash2048 2011
[10] A Grebennikov N O Sokal and M J Franco Switchmode RFPower Amplifiers ElsevierAcademic Press San Diego CalifUSA 2nd edition 2012
[11] T Canning P J Tasker and S C Cripps ldquoContinuous modepower amplifier design using harmonic clipping contourstheory and practicerdquo IEEE Transactions on Microwave Theoryand Techniques vol 62 no 1 pp 100ndash110 2014
[12] A Juhas and L A Novak ldquoGeneral description of nonnegativewaveforms up to second harmonic for power amplifier mod-ellingrdquoMathematical Problems in Engineering vol 2014 ArticleID 709762 18 pages 2014
[13] V I Arnolrsquod V V Goryunov O V Lyashko and V A VasilrsquoevDynamical Systems VIIImdashSingularity Theory II ApplicationsSpringer Berlin Germany 1993
[14] D Siersma ldquoProperties of conflict sets in the planerdquo BanachCenter Publications Polish Academy of Sciences vol 50 no 1 pp267ndash276 1999 Proceedings of the Banach Center Symposiumon Geometry and Topology of Caustics (Caustics rsquo98) WarsawPoland
[15] M van Manen The geometry of conflict sets [Dissertation]Universiteit Utrecht UtrechtThe Netherlands 2003 httpigi-tur-archivelibraryuunldissertations2003-0912-123058c4pdf
[16] Y L Sachkov ldquoMaxwell strata and symmetries in the problemofoptimal rolling of a sphere over a planerdquo Sbornik Mathematicsvol 201 no 7-8 pp 1029ndash1051 2010
[17] I A Bogaevsky ldquoPerestroikas of shocks and singularities ofminimum functionsrdquoPhysicaDNonlinear Phenomena vol 173no 1-2 pp 1ndash28 2002
[18] Y L Sachkov ldquoMaxwell strata in the Euler elastic problemrdquoJournal of Dynamical andControl Systems vol 14 no 2 pp 169ndash234 2008
[19] M Siino and T Koike ldquoTopological classification of black holesgeneric Maxwell set and crease set of a horizonrdquo InternationalJournal ofModern Physics D Gravitation Astrophysics Cosmol-ogy vol 20 no 6 pp 1095ndash1122 2011
Mathematical Problems in Engineering 29
[20] F H Raab ldquoClass-F power amplifiers with maximally flat wave-formsrdquo IEEETransactions onMicrowaveTheory and Techniquesvol 45 no 11 pp 2007ndash2012 1997
[21] A Juhas and L A Novak ldquoMaximally flat waveforms with finitenumber of harmonics in class-F power amplifiersrdquo Mathemat-ical Problems in Engineering vol 2013 Article ID 169590 9pages 2013
[22] L Fejer ldquoUber trigonometrische polynomerdquo Journal fur dieReine und Angewandte Mathematik vol 1916 no 146 pp 53ndash82 1916 (German)
[23] S C Cripps ldquoBessel Waives [microwave bytes]rdquo IEEE Micro-wave Magazine vol 10 no 7 pp 30ndash36 117 2009
[24] L N Bryzgalova ldquoSingularities of the maximum of para-metrically dependent functionrdquo Functional Analysis and ItsApplications vol 11 no 1 pp 49ndash51 1977
[25] V I Arnold A A Davydov V A Vassiliev and V MZakalyukin Mathematical Models of Catastrophes Control ofCatastrophic Process Encyclopedia of Life Support Systems(EOLSS) EOLSS Publishers Oxford UK 2006
[26] S C Cripps RF Power Amplifiers for Wireless CommunicationsArtech House Norwood Mass USA 2nd edition 2006
[27] A Zygmund Trigonometric Series vol 1 Cambridge UniversityPress Cambridge UK 2nd edition 1959
[28] A Jeffrey and H DaiHandbook of Mathematical Formulas andIntegrals ElsevierAcademic Press San Diego Calif USA 4thedition 2008
[29] J C Mason and D C Handscomb Chebyshev PolynomialsChapman amp Hall CRC Press Boca Raton Fla USA 2003
Submit your manuscripts athttpwwwhindawicom
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical Problems in Engineering
Hindawi Publishing Corporationhttpwwwhindawicom
Differential EquationsInternational Journal of
Volume 2014
Applied MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Probability and StatisticsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical PhysicsAdvances in
Complex AnalysisJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
OptimizationJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
CombinatoricsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Operations ResearchAdvances in
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Function Spaces
Abstract and Applied AnalysisHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of Mathematics and Mathematical Sciences
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Algebra
Discrete Dynamics in Nature and Society
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Decision SciencesAdvances in
Discrete MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom
Volume 2014 Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Stochastic AnalysisInternational Journal of
6 Mathematical Problems in Engineering
00
00
k = 2 k = 3
k = 4 k = 5
Acos 120572
Acos 120572
Acos 120572
Acos 120572
A sin 120572
A sin 120572A sin 120572
A sin 120572
Figure 3 Catastrophe set (solid line) and corresponding conflict set(dotted line) for 119896 le 5 In each plot white triangle dot correspondsto optimal waveform and white circle dot corresponds to maximallyflat waveform
119896 = 4 and one in the catastrophe curves for 119896 = 3 and 119896 = 5Notice that the end point of conflict set is the cusp point
Catastrophe set divides the parameter space (119860 cos120572119860 sin120572) into disjoint subsets In the cases 119896 = 2 and 119896 =
3 catastrophe curve defines inner and outer part For 119896 gt
3 catastrophe curve makes partition of parameter space inseveral inner subsets and one outer subset (see Figure 3)
Notice also that multiplying 119891(120591 119860 120572) with a positiveconstant and adding in turn another constant which leads towaveform of type 119908(120591 120574 119860 120572) (see (1) and (2)) do not makeimpact on the character of catastrophe and conflict sets Thisis because in the course of finding catastrophe set first andsecond derivatives of 119891(120591 119860 120572) are set to zero Clearly (34) interms of 119891(120591 119860 120572) are equivalent to the analogous equationsin terms of119908(120591 120574 119860 120572) Analogously in the course of findingconflict set we consider only the positions of global minima(these positions forwaveforms119891(120591 119860 120572) and119908(120591 120574 119860 120572) arethe same)
3 Nonnegative Waveforms with atLeast One Zero
In what follows let us consider a waveform containing dccomponent fundamental and 119896th (119896 ge 2) harmonic of theform
119879119896(120591) = 1 + 119886
1cos 120591 + 119887
1sin 120591 + 119886
119896cos 119896120591 + 119887
119896sin 119896120591 (35)
The amplitudes of fundamental and 119896th harmonic of wave-form of type (35) respectively are
1205821= radic11988621+ 11988721 (36)
120582119896= radic1198862119896+ 1198872119896 (37)
As it is shown in Section 21 nonnegative waveforms withmaximal amplitude of fundamental harmonic or maximalcoefficient of fundamental harmonic cosine part have atleast one zero It is also shown in Section 22 (Corollary 4)that waveforms of type (35) with nonzero amplitude offundamental harmonic have either one or two globalminimaConsequently if nonnegative waveform of type (35) withnonzero amplitude of fundamental harmonic has at least onezero then it has at most two zeros
In Section 31 we provide general description of nonnega-tive waveforms of type (35) with at least one zero In Sections32 and 33 we consider nonnegative waveforms of type (35)with two zeros
31 General Description of Nonnegative Waveforms with atLeast One Zero The main result of this section is presentedin the following proposition
Proposition 6 Every nonnegative waveform of type (35) withat least one zero can be expressed in the following form
119879119896(120591) = [1 minus cos (120591 minus 120591
0)] [1 minus 120582
119896119903119896(120591)] (38)
where119903119896(120591) = (119896 minus 1) cos 120585
+ 2
119896minus1
sum
119899=1
(119896 minus 119899) cos (119899 (120591 minus 1205910) + 120585)
(39)
providing that
120582119896le [(119896 minus 1) cos 120585 + 119896 sin(120585 minus 120585119896)
sin(120585119896)]
minus1
(40)
10038161003816100381610038161205851003816100381610038161003816 le 120587 (41)
Remark 7 Function on the right hand side of (40) is mono-tonically increasing function of |120585| on interval |120585| le 120587 (formore details about this function see Remark 15) From (57)and (65) it follows that relation
0 le 120582119896le 1 (42)
holds for every nonnegative waveform of type (35) Noticethat according to (40) 120582
119896= 1 implies |120585| = 120587 Substitution
of 120582119896
= 1 and |120585| = 120587 into (55) yields 119879119896(120591) = 1 minus
cos 119896(120591 minus 1205910) Consequently 120582
119896= 1 implies that amplitude
1205821of fundamental harmonic is equal to zero
Remark 8 Conversion of (38) into additive form leads to thefollowing expressions for coefficients of nonnegative wave-forms of type (35) with at least one zero
1198861= minus (1 + 120582
119896cos 120585) cos 120591
0minus 119896120582119896sin 120585 sin 120591
0 (43)
1198871= minus (1 + 120582
119896cos 120585) sin 120591
0+ 119896120582119896sin 120585 cos 120591
0 (44)
119886119896= 120582119896cos (119896120591
0minus 120585) (45)
119887119896= 120582119896sin (119896120591
0minus 120585) (46)
providing that 120582119896satisfy (40) and |120585| le 120587
Mathematical Problems in Engineering 7
2
1
0
0 1 2 3 4
Angle 120591120587
Wav
efor
ms
1205823 = radic2201205823 = radic281205823 = radic24
Figure 4 Nonnegative waveforms with at least one zero for 119896 = 31205910= 1205876 and 120585 = 31205874
Three examples of nonnegative waveforms with at leastone zero for 119896 = 3 are presented in Figure 4 (examples ofnonnegative waveformswith at least one zero for 119896 = 2 can befound in [12]) For all three waveforms presented in Figure 4we assume that 120591
0= 1205876 and 120585 = 31205874 From (40) it follows
that 1205823le radic24 Coefficients of waveform with 120582
3= radic220
(dotted line) are 1198861= minus08977 119887
1= minus03451 119886
3= 005
and 1198873= minus005 Coefficients of waveform with 120582
3= radic28
(dashed line) are 1198861= minus09453 119887
1= minus01127 119886
3= 0125
and 1198873= minus0125 Coefficients of waveform with 120582
3= radic24
(solid line) are 1198861= minus10245 119887
1= 02745 119886
3= 025 and
1198873= minus025 First two waveforms have one zero while third
waveform (presented with solid line) has two zeros
Proof of Proposition 6 Waveform of type (35) containingdc component fundamental and 119896th harmonic can be alsoexpressed in the form
119879119896(120591) = 1 + 120582
1cos (120591 + 120593
1) + 120582119896cos (119896120591 + 120593
119896) (47)
where1205821ge 0120582
119896ge 0120593
1isin (minus120587 120587] and120593
119896isin (minus120587 120587] It is easy
to see that relations between coefficient of (35) and param-eters of (47) read as follows
1198861= 1205821cos1205931 119887
1= minus1205821sin1205931 (48)
119886119896= 120582119896cos120593119896 119887
119896= minus120582119896sin120593119896 (49)
Let us introduce 120585 such that10038161003816100381610038161205851003816100381610038161003816 le 120587 120585 = (119896120591
0+ 120593119896) mod2120587 (50)
Using (50) coefficients (49) can be expressed as (45)-(46)Let us assume that 119879
119896(120591) is nonnegative waveform of type
(35) with at least one zero that is 119879119896(120591) ge 0 and 119879
119896(1205910) = 0
for some 1205910 Notice that conditions 119879
119896(120591) ge 0 and 119879
119896(1205910) = 0
imply that 1198791015840119896(1205910) = 0 From 119879
119896(1205910) = 0 and 1198791015840
119896(1205910) = 0 by
using (50) it follows that
1205821cos (120591
0+ 1205931) = minus (1 + 120582
119896cos 120585)
1205821sin (1205910+ 1205931) = minus119896120582
119896sin 120585
(51)
respectively On the other hand 1205821cos(120591 + 120593
1) can be rewrit-
ten as
1205821cos (120591 + 120593
1) = 1205821cos (120591
0+ 1205931) cos (120591 minus 120591
0)
minus 1205821sin (1205910+ 1205931) sin (120591 minus 120591
0)
(52)
Substitution of (51) into (52) yields
1205821cos (120591 + 120593
1) = minus (1 + 120582
119896cos 120585) cos (120591 minus 120591
0)
+ 119896120582119896sin 120585 sin (120591 minus 120591
0)
(53)
According to (50) it follows that cos(119896120591 + 120593119896) = cos(119896(120591 minus
1205910) + 120585) that is
cos (119896120591 + 120593119896) = cos 120585 cos 119896 (120591 minus 120591
0) minus sin 120585 sin 119896 (120591 minus 120591
0)
(54)
Furthermore substitution of (54) and (53) into (47) leads to
119879119896(120591) = [1 minus cos (120591 minus 120591
0)] [1 + 120582
119896cos 120585]
minus 120582119896[1 minus cos 119896 (120591 minus 120591
0)] cos 120585
+ 120582119896[119896 sin (120591 minus 120591
0) minus sin 119896 (120591 minus 120591
0)] sin 120585
(55)
According to (A2) and (A4) (see Appendices) there is com-mon factor [1minuscos(120591minus120591
0)] for all terms in (55) Consequently
(55) can be written in the form (38) where
119903119896(120591) = minus cos 120585 + [
1 minus cos 119896 (120591 minus 1205910)
1 minus cos (120591 minus 1205910)] cos 120585
minus [119896 sin (120591 minus 120591
0) minus sin 119896 (120591 minus 120591
0)
1 minus cos (120591 minus 1205910)
] sin 120585
(56)
From (56) by using (A2) (A4) and cos 120585 cos 119899(120591 minus 1205910) minus
sin 120585 sin 119899(120591 minus 1205910) = cos(119899(120591 minus 120591
0) + 120585) we obtain (39)
In what follows we are going to prove that (40) also holdsAccording to (38) 119879
119896(120591) is nonnegative if and only if
120582119896max120591
119903119896(120591) le 1 (57)
Let us first show that position of global maximumof 119903119896(120591)
belongs to the interval |120591 minus 1205910| le 2120587119896 Relation (56) can be
rewritten as
119903119896(120591) = 119903
119896(1205910minus2120585
119896) + 119902119896(120591) (58)
where
119903119896(1205910minus2120585
119896) = (119896 minus 1) cos 120585 + 119896 sin (120585 minus 120585119896)
sin (120585119896) (59)
119902119896(120591) =
1
1 minus cos (120591 minus 1205910)
sdot [cos 120585 minus cos(119896(120591 minus 1205910+120585
119896))
+119896 sin 120585sin (120585119896)
(cos(120591 minus 1205910+120585
119896) minus cos(120585
119896))]
(60)
8 Mathematical Problems in Engineering
For |120585| lt 120587 relation sin 120585 sin(120585119896) gt 0 obviously holds Fromcos 119905 gt cos 1199051015840 for |119905| le 120587119896 lt |119905
1015840
| le 120587 it follows that positionof global maximum of the function of type [119888 cos 119905 minus cos(119896119905)]for 119888 gt 0 belongs to interval |119905| le 120587119896 Therefore position ofglobal maximum of the expression in the square brackets in(60) for |120585| lt 120587 belongs to interval |120591 minus 120591
0+ 120585119896| le 120587119896 This
inequality together with |120585| lt 120587 leads to |120591minus1205910| lt 2120587119896 Since
[1 minus cos(120591 minus 1205910)]minus1 decreases with increasing |120591 minus 120591
0| le 120587 it
follows that 119902119896(120591) for |120585| lt 120587 has global maximum on interval
|120591minus1205910| lt 2120587119896 For |120585| = 120587 it is easy to show thatmax
120591119902119896(120591) =
119902119896(1205910plusmn 2120587119896) = 0 Since 119903
119896(120591) minus 119902
119896(120591) is constant (see (58))
it follows from previous considerations that 119903119896(120591) has global
maximum on interval |120591 minus 1205910| le 2120587119896
To find max120591119903119896(120591) let us consider first derivative of 119903
119896(120591)
with respect to 120591 Starting from (56) first derivative of 119903119896(120591)
can be expressed in the following form
119889119903119896(120591)
119889120591= minus119904 (120591) sdot sin(
119896 (120591 minus 1205910)
2+ 120585) (61)
where
119904 (120591) = [sin(119896 (120591 minus 120591
0)
2) cos(
120591 minus 1205910
2)
minus 119896 cos(119896 (120591 minus 120591
0)
2) sin(
120591 minus 1205910
2)]
sdot sinminus3 (120591 minus 1205910
2)
(62)
Using (A6) (see Appendices) (62) can be rewritten as
119904 (120591) = 2
119896minus1
sum
119899=1
119899 (119896 minus 119899) cos((119896 minus 2119899) (120591 minus 120591
0)
2) (63)
From 119899(119896 minus 119899) gt 0 and |119896 minus 2119899| lt 119896 119899 = 1 (119896 minus 1) itfollows that all summands in (63) decrease with increasing|120591 minus 1205910| providing that |120591 minus 120591
0| le 2120587119896 Therefore 119904(120591) ge 119904(120591
0plusmn
2120587119896) = 119896sin2(120587119896) gt 0 for |120591 minus 1205910| le 2120587119896 Consequently
119889119903119896(120591)119889120591 = 0 and |120591minus120591
0| le 2120587119896 imply that sin(119896(120591minus120591
0)2+
120585) = 0From |120585| le 120587 |120591minus120591
0| le 2120587119896 and sin(119896(120591minus120591
0)2+120585) = 0
it follows that 120591minus1205910+120585119896 = minus120585119896 or |120591minus120591
0+120585119896| = (2120587minus|120585|)119896
and therefore cos(119896(120591 minus 1205910+ 120585119896)) = cos 120585 Since cos(120585119896) ge
cos(2120587 minus |120585|)119896 it follows that max120591119902119896(120591) is attained for 120591 =
1205910minus2120585119896 Furthermore from (60) it follows that max
120591119902119896(120591) =
119902119896(1205910minus 2120585119896) = 0 which together with (58)-(59) leads to
max120591
119903119896(120591) = 119903
119896(1205910minus2120585
119896)
= (119896 minus 1) cos 120585 + 119896 sin (120585 minus 120585119896)sin (120585119896)
(64)
Both terms on the right hand side of (64) are even functionsof 120585 and decrease with increase of |120585| |120585| le 120587 Thereforemax120591119903119896(120591) attains its lowest value for |120585| = 120587 It is easy to
show that right hand side of (64) for |120585| = 120587 is equal to 1which further implies that
max120591
119903119896(120591) ge 1 (65)
From (65) it follows that (57) can be rewritten as 120582119896
le
[max120591119903119896(120591)]minus1 Finally substitution of (64) into 120582
119896le
[max120591119903119896(120591)]minus1 leads to (40) which completes the proof
32 Nonnegative Waveforms with Two Zeros Nonnegativewaveforms of type (35) with two zeros always possess twoglobal minima Such nonnegative waveforms are thereforerelated to the conflict set
In this subsection we provide general description of non-negative waveforms of type (35) for 119896 ge 2 and exactly twozeros According to Remark 7 120582
119896= 1 implies |120585| = 120587 and
119879119896(120591) = 1 minus cos 119896(120591 minus 120591
0) Number of zeros of 119879
119896(120591) = 1 minus
cos 119896(120591minus1205910) on fundamental period equals 119896 which is greater
than two for 119896 gt 2 and equal to two for 119896 = 2 In the followingproposition we exclude all waveforms with 120582
119896= 1 (the case
when 119896 = 2 and 1205822= 1 is going to be discussed in Remark 10)
Proposition 9 Every nonnegative waveform of type (35) withexactly two zeros can be expressed in the following form
119879119896(120591) = 120582
119896[1 minus cos (120591 minus 120591
0)] [1 minus cos(120591 minus 120591
0+2120585
119896)]
sdot [1198880+ 2
119896minus2
sum
119899=1
119888119899cos 119899(120591 minus 120591
0+120585
119896)]
(66)
where
119888119899= [sin(120585 minus 119899120585
119896) cos(120585
119896)
minus (119896 minus 119899) cos(120585 minus 119899120585
119896) sin(120585
119896)]
sdot sinminus3 (120585119896)
(67)
120582119896= [(119896 minus 1) cos 120585 + 119896 sin(120585 minus 120585119896)
sin(120585119896)]
minus1
(68)
0 lt10038161003816100381610038161205851003816100381610038161003816 lt 120587 (69)
Remark 10 For 119896 = 2 waveforms with 1205822= 1 also have
exactly two zeros These waveforms can be included in aboveproposition by substituting (69) with 0 lt |120585| le 120587
Remark 11 Apart from nonnegative waveforms of type (35)with two zeros there are another two types of nonnegativewaveforms which can be obtained from (66)ndash(68) These are
(i) nonnegative waveforms with 119896 zeros (correspondingto |120585| = 120587) and
(ii) maximally flat nonnegative waveforms (correspond-ing to 120585 = 0)
Notice that nonnegative waveforms of type (35) with120582119896= 1 can be obtained from (66)ndash(68) by setting |120585| =
120587 Substitution of 120582119896
= 1 and |120585| = 120587 into (66) alongwith execution of all multiplications and usage of (A2) (seeAppendices) leads to 119879
119896(120591) = 1 minus cos 119896(120591 minus 120591
0)
Mathematical Problems in Engineering 9
Also maximally flat nonnegative waveforms (they haveonly one zero [21]) can be obtained from (66)ndash(68) by setting120585 = 0 Thus substitution of 120585 = 0 into (66)ndash(68) leads tothe following form of maximally flat nonnegative waveformof type (35)
119879119896(120591) =
[1 minus cos (120591 minus 1205910)]2
3 (1198962 minus 1)
sdot [119896 (1198962
minus 1)
+ 2
119896minus2
sum
119899=1
(119896 minus 119899) ((119896 minus 119899)2
minus 1) cos 119899 (120591 minus 1205910)]
(70)
Maximally flat nonnegative waveforms of type (35) for 119896 le 4
can be expressed as
1198792(120591) =
2
3[1 minus cos(120591 minus 120591
0)]2
1198793(120591) =
1
2[1 minus cos(120591 minus 120591
0)]2
[2 + cos (120591 minus 1205910)]
1198794(120591) =
4
15[1 minus cos (120591 minus 120591
0)]2
sdot [5 + 4 cos (120591 minus 1205910) + cos 2 (120591 minus 120591
0)]
(71)
Remark 12 Every nonnegative waveform of type (35) withexactly one zero at nondegenerate critical point can bedescribed as in Proposition 6 providing that symbol ldquolerdquoin relation (40) is replaced with ldquoltrdquo This is an immediateconsequence of Propositions 6 and 9 and Remark 11
Remark 13 Identity [1minus cos(120591minus1205910)][1minus cos(120591minus120591
0+2120585119896)] =
[cos 120585119896 minus cos(120591 minus 1205910+ 120585119896)]
2 implies that (66) can be alsorewritten as
119879119896(120591) = 120582
119896[cos 120585119896
minus cos(120591 minus 1205910+120585
119896)]
2
sdot [1198880+ 2
119896minus2
sum
119899=1
119888119899cos 119899(120591 minus 120591
0+120585
119896)]
(72)
Furthermore substitution of (67) into (72) leads to
119879119896(120591)
= 120582119896[cos 120585119896
minus cos(120591 minus 1205910+120585
119896)]
sdot [(119896 minus 1) sin 120585sin (120585119896)
minus 2
119896minus1
sum
119899=1
sin (120585 minus 119899120585119896)sin (120585119896)
cos 119899(120591 minus 1205910+120585
119896)]
(73)
Remark 14 According to (A6) (see Appendices) it followsthat coefficients (67) can be expressed as
119888119899= 2
119896minus119899minus1
sum
119898=1
119898(119896 minus 119899 minus 119898) cos((119896 minus 119899 minus 2119898) 120585119896
) (74)
Furthermore from (74) it follows that coefficients 119888119896minus2
119888119896minus3
119888119896minus4
and 119888119896minus5
are equal to119888119896minus2
= 2 (75)
119888119896minus3
= 8 cos(120585119896) (76)
119888119896minus4
= 8 + 12 cos(2120585119896) (77)
119888119896minus5
= 24 cos(120585119896) + 16 cos(3120585
119896) (78)
For example for 119896 = 2 (75) and (68) lead to 1198880= 2 and
1205822= 1(2+cos 120585) respectively which from (72) further imply
that
1198792(120591) =
2 [cos(1205852) minus cos(120591 minus 1205910+ 1205852)]
2
[2 + cos 120585] (79)
Also for 119896 = 3 (75) (76) and (68) lead to 1198881= 2 119888
0=
8 cos(1205853) and 1205823= [2(3 cos(1205853) + cos 120585)]minus1 respectively
which from (72) further imply that
1198793(120591) =
2 [cos (1205853) minus cos (120591 minus 1205910+ 1205853)]
2
[3 cos (1205853) + cos 120585]
sdot [2 cos(1205853) + cos(120591 minus 120591
0+120585
3)]
(80)
Remark 15 According to (A5) (see Appendices) relation(68) can be rewritten as
120582119896= [(119896 minus 1) cos 120585 + 119896
119896minus1
sum
119899=1
cos((119896 minus 2119899)120585119896
)]
minus1
(81)
Clearly amplitude 120582119896of 119896th harmonic of nonnegative wave-
form of type (35) with exactly two zeros is even functionof 120585 Since cos((119896 minus 2119899)120585119896) 119899 = 0 (119896 minus 1) decreaseswith increase of |120585| on interval 0 le |120585| le 120587 it follows that120582119896monotonically increases with increase of |120585| Right hand
side of (68) is equal to 1(1198962
minus 1) for 120585 = 0 and to one for|120585| = 120587 Therefore for nonnegative waveforms of type (35)with exactly two zeros the following relation holds
1
1198962 minus 1lt 120582119896lt 1 (82)
The left boundary in (82) corresponds to maximally flatnonnegative waveforms (see Remark 11) The right boundaryin (82) corresponds to nonnegative waveforms with 119896 zeros(also see Remark 11)
Amplitude of 119896th harmonic of nonnegative waveform oftype (35) with two zeros as a function of parameter 120585 for 119896 le5 is presented in Figure 5
Remark 16 Nonnegative waveform of type (35) with twozeros can be also expressed in the following form
119879119896(120591) = 1 minus 120582
119896
119896 sin 120585sin (120585119896)
cos(120591 minus 1205910+120585
119896)
+ 120582119896cos (119896 (120591 minus 120591
0) + 120585)
(83)
10 Mathematical Problems in Engineering
1
08
06
04
02
0minus1 minus05 0 05
Am
plitu
de120582k
1
Parameter 120585120587
k = 2k = 3
k = 4k = 5
Figure 5 Amplitude of 119896th harmonic of nonnegative waveformwith two zeros as a function of parameter 120585
where 120582119896is given by (68) and 0 lt |120585| lt 120587 From (83) it follows
that coefficients of fundamental harmonic of nonnegativewaveform of type (35) with two zeros are
1198861= minus1205821cos(120591
0minus120585
119896) 119887
1= minus1205821sin(120591
0minus120585
119896) (84)
where 1205821is amplitude of fundamental harmonic
1205821=
119896 sin 120585sin (120585119896)
120582119896 (85)
Coefficients of 119896th harmonic are given by (45)-(46)Notice that (68) can be rewritten as
120582119896= [cos(120585
119896)
119896 sin 120585sin(120585119896)
minus cos 120585]minus1
(86)
By introducing new variable
119909 = cos(120585119896) (87)
and using the Chebyshev polynomials (eg see Appendices)relations (85) and (86) can be rewritten as
1205821= 119896120582119896119880119896minus1
(119909) (88)
120582119896=
1
119896119909119880119896minus1
(119909) minus 119881119896(119909)
(89)
where119881119896(119909) and119880
119896(119909) denote the Chebyshev polynomials of
the first and second kind respectively From (89) it followsthat
120582119896[119896119909119880119896minus1
(119909) minus 119881119896(119909)] minus 1 = 0 (90)
which is polynomial equation of 119896th degree in terms of var-iable 119909 From 0 lt |120585| lt 120587 and (87) it follows that
cos(120587119896) lt 119909 lt 1 (91)
Since 120582119896is monotonically increasing function of |120585| 0 lt |120585| lt
120587 it follows that 120582119896is monotonically decreasing function of
119909 This further implies that (90) has only one solution thatsatisfies (91) (For 119896 = 2 expression (91) reads cos(1205872) le
119909 lt 1) This solution for 119909 (which can be obtained at leastnumerically) according to (88) leads to amplitude 120582
1of
fundamental harmonicFor 119896 le 4 solutions of (90) and (91) are
119909 = radic1 minus 1205822
21205822
1
3lt 1205822le 1
119909 =1
23radic1205823
1
8lt 1205823lt 1
119909 = radic1
6(1 + radic
51205824+ 3
21205824
)1
15lt 1205824lt 1
(92)
Insertion of (92) into (88) leads to the following relationsbetween amplitude 120582
1of fundamental and amplitude 120582
119896of
119896th harmonic 119896 le 4
1205821= radic8120582
2(1 minus 120582
2)
1
3lt 1205822le 1 (93)
1205821= 3 (
3radic1205823minus 1205823)
1
8lt 1205823lt 1 (94)
1205821= radic
32
27(radic2120582
4(3 + 5120582
4)3
minus 21205824(9 + 7120582
4))
1
15lt 1205824lt 1
(95)
Proof of Proposition 9 As it has been shown earlier (seeProposition 6) nonnegative waveform of type (35) with atleast one zero can be represented in form (38) Since weexclude nonnegative waveforms with 120582
119896= 1 according to
Remark 7 it follows that we exclude case |120585| = 120587Therefore inthe quest for nonnegative waveforms of type (35) having twozeros we will start with waveforms of type (38) for |120585| lt 120587It is clear that nonnegative waveforms of type (38) have twozeros if and only if
120582119896= [max120591
119903119896(120591)]minus1
(96)
and max120591119903119896(120591) = 119903
119896(1205910) According to (64) max
120591119903119896(120591) =
119903119896(1205910) implies |120585| = 0 Therefore it is sufficient to consider
only the interval (69)Substituting (96) into (38) we obtain
119879119896(120591) =
[1 minus cos (120591 minus 1205910)] [max
120591119903119896(120591) minus 119903
119896(120591)]
max120591119903119896(120591)
(97)
Mathematical Problems in Engineering 11
Expression max120591119903119896(120591) minus 119903
119896(120591) according to (64) and (39)
equals
max120591
119903119896(120591) minus 119903
119896(120591) = 119896
sin ((119896 minus 1) 120585119896)sin (120585119896)
minus 2
119896minus1
sum
119899=1
(119896 minus 119899) cos (119899 (120591 minus 1205910) + 120585)
(98)
Comparison of (97) with (66) yields
max120591
119903119896(120591) minus 119903
119896(120591) = [1 minus cos(120591 minus 120591
0+2120585
119896)]
sdot [1198880+ 2
119896minus2
sum
119899=1
119888119899cos 119899(120591 minus 120591
0+120585
119896)]
(99)
where coefficients 119888119899 119899 = 0 119896 minus 2 are given by (67) In
what follows we are going to show that right hand sides of(98) and (99) are equal
From (67) it follows that
1198880minus 1198881cos(120585
119896) = 119896
sin (120585 minus 120585119896)sin (120585119896)
(100)
Also from (67) for 119899 = 1 119896minus3 it follows that the followingrelations hold
(119888119899minus1
+ 119888119899+1
) cos(120585119896) minus 2119888
119899= 2 (119896 minus 119899) cos(120585 minus 119899120585
119896)
(119888119899minus1
minus 119888119899+1
) sin(120585119896) = 2 (119896 minus 119899) sin(120585 minus 119899120585
119896)
(101)
From (99) by using (75) (76) (100)-(101) and trigonometricidentities
cos(120591 minus 1205910+2120585
119896) = cos(120585
119896) cos(120591 minus 120591
0+120585
119896)
minus sin(120585119896) sin(120591 minus 120591
0+120585
119896)
cos(120585 minus 119899120585
119896) cos(119899(120591 minus 120591
0+120585
119896))
minus sin(120585 minus 119899120585
119896) sin(119899(120591 minus 120591
0+120585
119896))
= cos (119899 (120591 minus 1205910) + 120585)
(102)
we obtain (98) Consequently (98) and (99) are equal whichcompletes the proof
33 Nonnegative Waveforms with Two Zeros and PrescribedCoefficients of 119896thHarmonic In this subsectionwe show thatfor prescribed coefficients 119886
119896and 119887119896 there are 119896 nonnegative
waveforms of type (35) with exactly two zeros According to
(37) and (82) coefficients 119886119896and 119887119896of nonnegative waveforms
of type (35) with exactly two zeros satisfy the followingrelation
1
1198962 minus 1lt radic1198862119896+ 1198872119896lt 1 (103)
According to Remark 16 the value of 119909 (see (87)) that cor-responds to 120582
119896= radic1198862
119896+ 1198872119896can be determined from (90)-
(91) As we mentioned earlier (90) has only one solutionthat satisfies (91) This value of 119909 according to (88) leadsto the amplitude 120582
1of fundamental harmonic (closed form
expressions for 1205821in terms of 120582
119896and 119896 le 4 are given by (93)ndash
(95))On the other hand from (45)-(46) it follows that
1198961205910minus 120585 = atan 2 (119887
119896 119886119896) + 2119902120587 119902 = 1 (119896 minus 1) (104)
where function atan 2(119910 119909) is defined as
atan 2 (119910 119909) =
arctan(119910
119909) if 119909 ge 0
arctan(119910
119909) + 120587 if 119909 lt 0 119910 ge 0
arctan(119910
119909) minus 120587 if 119909 lt 0 119910 lt 0
(105)
with the codomain (minus120587 120587] Furthermore according to (84)and (104) the coefficients of fundamental harmonic of non-negative waveforms with two zeros and prescribed coeffi-cients of 119896th harmonic are equal to
1198861= minus1205821cos[
atan 2 (119887119896 119886119896) + 2119902120587
119896]
1198871= minus1205821sin[
atan 2 (119887119896 119886119896) + 2119902120587
119896]
(106)
where 119902 = 0 (119896minus 1) For chosen 119902 according to (104) and(66) positions of zeros are
1205910=1
119896[120585 + atan 2 (119887
119896 119886119896) + 2119902120587]
1205910minus2120585
119896=1
119896[minus120585 + atan 2 (119887
119896 119886119896) + 2119902120587]
(107)
From (106) and 119902 = 0 (119896minus1) it follows that for prescribedcoefficients 119886
119896and 119887119896 there are 119896 nonnegative waveforms of
type (35) with exactly two zerosWe provide here an algorithm to facilitate calculation
of coefficients 1198861and 1198871of nonnegative waveforms of type
(35) with two zeros and prescribed coefficients 119886119896and 119887
119896
providing that 119886119896and 119887119896satisfy (103)
12 Mathematical Problems in Engineering
2
1
0
Wav
efor
ms
0 1 2 3 4
Angle 120591120587
q = 0
q = 1
q = 2
Figure 6 Nonnegative waveforms with two zeros for 119896 = 3 1198863=
minus015 and 1198873= minus02
Algorithm 17 (i) Calculate 120582119896= radic1198862119896+ 1198872119896
(ii) identify 119909 that satisfies both relations (90) and (91)(iii) calculate 120582
1according to (88)
(iv) choose integer 119902 such that 0 le 119902 le 119896 minus 1(v) calculate 119886
1and 1198871according to (106)
For 119896 le 4 by using (93) for 119896 = 2 (94) for 119896 = 3 and (95)for 119896 = 4 it is possible to calculate directly 120582
1from 120582
119896and
proceed to step (iv)For 119896 = 2 and prescribed coefficients 119886
2and 1198872 there are
two waveforms with two zeros one corresponding to 1198861lt 0
and the other corresponding to 1198861gt 0 (see also [12])
Let us take as an input 119896 = 3 1198863= minus015 and 119887
3= minus02
Execution of Algorithm 17 on this input yields 1205823= 025 and
1205821= 11399 (according to (94)) For 119902 = 0 we calculate
1198861= minus08432 and 119887
1= 07670 (corresponding waveform is
presented by solid line in Figure 6) for 119902 = 1 we calculate1198861= minus02426 and 119887
1= minus11138 (corresponding waveform is
presented by dashed line) for 119902 = 2 we calculate 1198861= 10859
and 1198871= 03468 (corresponding waveform is presented by
dotted line)As another example of the usage of Algorithm 17 let us
consider case 119896 = 4 and assume that1198864= minus015 and 119887
4= minus02
Consequently 1205824= 025 and 120582
1= 09861 (according to (95))
For 119902 = 0 3we calculate the following four pairs (1198861 1198871) of
coefficients of fundamental harmonic (minus08388 05184) for119902 = 0 (minus05184 minus08388) for 119902 = 1 (08388 minus05184) for 119902 =2 and (05184 08388) for 119902 = 3 Corresponding waveformsare presented in Figure 7
4 Nonnegative Waveforms with MaximalAmplitude of Fundamental Harmonic
In this section we provide general description of nonnegativewaveforms containing fundamental and 119896th harmonic withmaximal amplitude of fundamental harmonic for prescribedamplitude of 119896th harmonic
The main result of this section is presented in the fol-lowing proposition
3
2
1
0
Wav
efor
ms
0 1 2 3 4
Angle 120591120587
q = 0q = 1
q = 2q = 3
Figure 7 Nonnegative waveforms with two zeros for 119896 = 4 1198864=
minus015 and 1198874= minus02
Proposition 18 Every nonnegativewaveformof type (35)withmaximal amplitude 120582
1of fundamental harmonic and pre-
scribed amplitude 120582119896of 119896th harmonic can be expressed in the
following form
119879119896(120591) = [1 minus cos (120591 minus 120591
0)]
sdot [1 minus (119896 minus 1) 120582119896minus 2120582119896
119896minus1
sum
119899=1
(119896 minus 119899) cos (119899 (120591 minus 1205910))]
(108)
if 0 le 120582119896le 1(119896
2
minus 1) or
119879119896(120591) = 120582
119896[1 minus cos (120591 minus 120591
0)] [1 minus cos(120591 minus 120591
0+2120585
119896)]
sdot [1198880+ 2
119896minus2
sum
119899=1
119888119899cos 119899(120591 minus 120591
0+120585
119896)]
(109)
if 1(1198962 minus 1) le 120582119896le 1 providing that 119888
119899 119899 = 0 119896 minus 2 and
120582119896are related to 120585 via relations (67) and (68) respectively and
|120585| le 120587
Remark 19 Expression (108) can be obtained from (38) bysetting 120585 = 0 Furthermore insertion of 120585 = 0 into (43)ndash(46)leads to the following expressions for coefficients ofwaveformof type (108)
1198861= minus (1 + 120582
119896) cos 120591
0 119887
1= minus (1 + 120582
119896) sin 120591
0
119886119896= 120582119896cos (119896120591
0) 119887
119896= 120582119896sin (119896120591
0)
(110)
On the other hand (109) coincides with (66) Thereforethe expressions for coefficients of (109) and (66) also coincideThus expressions for coefficients of fundamental harmonic ofwaveform (109) are given by (84) where 120582
1is given by (85)
while expressions for coefficients of 119896th harmonic are givenby (45)-(46)
Waveforms described by (108) have exactly one zerowhile waveforms described by (109) for 1(1198962 minus 1) lt 120582
119896lt 1
Mathematical Problems in Engineering 13
14
12
1
08
06
04
02
00 05 1
Amplitude 120582k
Am
plitu
de1205821
k = 2
k = 3
k = 4
Figure 8 Maximal amplitude of fundamental harmonic as a func-tion of amplitude of 119896th harmonic
have exactly two zeros As we mentioned earlier waveforms(109) for 120582
119896= 1 have 119896 zeros
Remark 20 Maximal amplitude of fundamental harmonic ofnonnegative waveforms of type (35) for prescribed amplitudeof 119896th harmonic can be expressed as
1205821= 1 + 120582
119896 (111)
if 0 le 120582119896le 1(119896
2
minus 1) or
1205821=
119896 sin 120585119896 sin 120585 cos (120585119896) minus cos 120585 sin (120585119896)
(112)
if 1(1198962 minus 1) le 120582119896le 1 where 120585 is related to 120582
119896via (68) (or
(86)) and |120585| le 120587From (110) it follows that (111) holds Substitution of (86)
into (85) leads to (112)Notice that 120582
119896= 1(119896
2
minus 1) is the only common point ofthe intervals 0 le 120582
119896le 1(119896
2
minus 1) and 1(1198962
minus 1) le 120582119896le
1 According to (111) 120582119896= 1(119896
2
minus 1) corresponds to 1205821=
1198962
(1198962
minus1) It can be also obtained from (112) by setting 120585 = 0The waveforms corresponding to this pair of amplitudes aremaximally flat nonnegative waveforms
Maximal amplitude of fundamental harmonic of non-negative waveform of type (35) for 119896 le 4 as a function ofamplitude of 119896th harmonic is presented in Figure 8
Remark 21 Maximum value of amplitude of fundamentalharmonic of nonnegative waveform of type (35) is
1205821max =
1
cos (120587 (2119896)) (113)
This maximum value is attained for |120585| = 1205872 (see (112)) Thecorresponding value of amplitude of 119896th harmonic is 120582
119896=
(1119896) tan(120587(2119896)) Nonnegative waveforms of type (35) with1205821= 1205821max have two zeros at 1205910 and 1205910 minus 120587119896 for 120585 = 1205872 or
at 1205910and 1205910+ 120587119896 for 120585 = minus1205872
14
12
1
08
06
04
02
0minus1 minus05 0 05 1
Am
plitu
de1205821
Parameter 120585120587
k = 2k = 3k = 4
Figure 9 Maximal amplitude of fundamental harmonic as a func-tion of parameter 120585
To prove that (113) holds let us first show that the fol-lowing relation holds for 119896 ge 2
cos( 120587
2119896) lt 1 minus
1
1198962 (114)
From 119896 ge 2 it follows that sinc(120587(4119896)) gt sinc(1205874) wheresinc 119909 = (sin119909)119909 and therefore sin(120587(4119896)) gt 1(radic2119896)By using trigonometric identity cos 2119909 = 1 minus 2sin2119909 weimmediately obtain (114)
According to (111) and (112) it is clear that 1205821attains its
maximum value on the interval 1(1198962 minus 1) le 120582119896le 1 Since
120582119896is monotonic function of |120585| on interval |120585| le 120587 (see
Remark 15) it follows that 119889120582119896119889120585 = 0 for 0 lt |120585| lt 120587
Therefore to find critical points of 1205821as a function of 120582
119896
it is sufficient to find critical points of 1205821as a function of
|120585| 0 lt |120585| lt 120587 and consider its values at the end points120585 = 0 and |120585| = 120587 Plot of 120582
1as a function of parameter 120585
for 119896 le 4 is presented in Figure 9 According to (112) firstderivative of 120582
1with respect to 120585 is equal to zero if and only
if (119896 cos 120585 sin(120585119896) minus sin 120585 cos(120585119896)) cos 120585 = 0 On interval0 lt |120585| lt 120587 this is true if and only if |120585| = 1205872 Accordingto (112) 120582
1is equal to 119896
2
(1198962
minus 1) for 120585 = 0 equal to zerofor |120585| = 120587 and equal to 1 cos(120587(2119896)) for |120585| = 1205872 From(114) it follows that 1198962(1198962minus1) lt 1 cos(120587(2119896)) and thereforemaximum value of 120582
1is given by (113) Moreover maximum
value of 1205821is attained for |120585| = 1205872
According to above consideration all nonnegative wave-forms of type (35) having maximum value of amplitude offundamental harmonic can be obtained from (109) by setting|120585| = 1205872 Three of them corresponding to 119896 = 3 120585 = 1205872and three different values of 120591
0(01205876 and1205873) are presented
in Figure 10 Dotted line corresponds to 1205910= 0 (coefficients
of corresponding waveform are 1198861= minus1 119887
1= 1radic3 119886
3= 0
and 1198873= minusradic39) solid line to 120591
0= 1205876 (119886
1= minus2radic3 119887
1= 0
1198863= radic39 and 119887
3= 0) and dashed line to 120591
0= 1205873 (119886
1= minus1
1198871= minus1radic3 119886
3= 0 and 119887
3= radic39)
Proof of Proposition 18 As it has been shown earlier (Propo-sition 6) nonnegative waveform of type (35) with at least
14 Mathematical Problems in Engineering
2
1
0
0 1 2 3 4
Angle 120591120587
Wav
efor
ms
1205910 = 01205910 = 12058761205910 = 1205873
Figure 10 Nonnegative waveforms with maximum amplitude offundamental harmonic for 119896 = 3 and 120585 = 1205872
one zero can be represented in form (38) According to (43)(44) and (36) for amplitude 120582
1of fundamental harmonic of
waveforms of type (38) the following relation holds
1205821= radic(1 + 120582
119896cos 120585)2 + 11989621205822
119896sin2120585 (115)
where 120582119896satisfy (40) and |120585| le 120587
Because of (40) in the quest of finding maximal 1205821for
prescribed 120582119896 we have to consider the following two cases
(Case i)120582119896lt [(119896minus1) cos 120585 + 119896 sin(120585minus120585119896) sin(120585119896)]minus1
(Case ii)120582119896= [(119896minus1) cos 120585+119896 sin(120585minus120585119896) sin(120585119896)]minus1
Case i Since 120582119896lt [(119896 minus 1) cos 120585 + 119896 sin(120585 minus 120585119896) sin(120585119896)]minus1
implies 120582119896
= 1 according to (115) it follows that 1205821
= 0Hence 119889120582
1119889120585 = 0 implies
2120582119896sin 120585 [1 minus (1198962 minus 1) 120582
119896cos 120585] = 0 (116)
Therefore 1198891205821119889120585 = 0 if 120582
119896= 0 (Option 1) or sin 120585 = 0
(Option 2) or (1198962 minus 1)120582119896cos 120585 = 1 (Option 3)
Option 1 According to (115) 120582119896= 0 implies 120582
1= 1 (notice
that this implication shows that 1205821does not depend on 120585 and
therefore we can set 120585 to zero value)
Option 2 According to (115) sin 120585 = 0 implies 1205821= 1 +
120582119896cos 120585 which further leads to the conclusion that 120582
1is
maximal for 120585 = 0 For 120585 = 0 120582119896lt [(119896 minus 1) cos 120585 + 119896 sin(120585 minus
120585119896) sin(120585119896)]minus1 becomes 120582119896lt 1(119896
2
minus 1)
Option 3 This option leads to contradiction To show thatnotice that (119896
2
minus 1)120582119896cos 120585 = 1 and 120582
119896lt [(119896 minus
1) cos 120585 + 119896 sin(120585 minus 120585119896) sin(120585119896)]minus1 imply that (119896 minus 1) cos 120585 gtsin(120585minus120585119896) sin(120585119896) Using (A5) (see Appendices) the latestinequality can be rewritten assum119896minus1
119899=1[cos 120585minuscos((119896minus2119899)120585119896)] gt
0 But from |119896 minus 2119899| lt 119896 119899 = 1 (119896 minus 1) and |120585| le 120587
it follows that all summands are not positive and therefore(119896minus1) cos 120585 gt sin(120585minus120585119896) sin(120585119896) does not hold for |120585| le 120587
Consequently Case i implies 120585 = 0 and 120582119896lt 1(119896
2
minus 1)Finally substitution of 120585 = 0 into (38) leads to (108) whichproves that (108) holds for 120582
119896lt 1(119896
2
minus 1)
Case ii Relation120582119896= [(119896minus1) cos 120585+119896 sin(120585minus120585119896) sin(120585119896)]minus1
according to Proposition 9 and Remark 11 implies that cor-responding waveforms can be expressed via (66)ndash(68) for|120585| le 120587 Furthermore 120582
119896= [(119896 minus 1) cos 120585 + 119896 sin(120585 minus
120585119896) sin(120585119896)]minus1 and |120585| le 120587 imply 1(1198962 minus 1) le 120582119896le 1
This proves that (109) holds for 1(1198962 minus 1) le 120582119896le 1
Finally let us prove that (108) holds for 120582119896= 1(119896
2
minus
1) According to (68) (see also Remark 11) this value of 120582119896
corresponds to 120585 = 0 Furthermore substitution of 120582119896=
1(1198962
minus 1) and 120585 = 0 into (109) leads to (70) which can berewritten as
119879119896(120591) =
[1 minus cos (120591 minus 1205910)]
(1 minus 1198962)
sdot [119896 (119896 minus 1) minus 2
119896minus1
sum
119899=1
(119896 minus 119899) cos (119899 (120591 minus 1205910))]
(117)
Waveform (117) coincides with waveform (108) for 120582119896
=
1(1 minus 1198962
) Consequently (108) holds for 120582119896= 1(1 minus 119896
2
)which completes the proof
5 Nonnegative Waveforms with MaximalAbsolute Value of the Coefficient of CosineTerm of Fundamental Harmonic
In this sectionwe consider general description of nonnegativewaveforms of type (35) with maximal absolute value ofcoefficient 119886
1for prescribed coefficients of 119896th harmonicThis
type of waveform is of particular interest in PA efficiencyanalysis In a number of cases of practical interest eithercurrent or voltage waveform is prescribed In such casesthe problem of finding maximal efficiency of PA can bereduced to the problem of finding nonnegative waveformwith maximal coefficient 119886
1for prescribed coefficients of 119896th
harmonic (see also Section 7)In Section 51 we provide general description of nonneg-
ative waveforms of type (35) with maximal absolute value ofcoefficient 119886
1for prescribed coefficients of 119896th harmonic In
Section 52 we illustrate results of Section 51 for particularcase 119896 = 3
51 Nonnegative Waveforms with Maximal Absolute Value ofCoefficient 119886
1for 119896 ge 2 Waveforms 119879
119896(120591) of type (35) with
1198861ge 0 can be derived from those with 119886
1le 0 by shifting
by 120587 and therefore we can assume without loss of generalitythat 119886
1le 0 Notice that if 119896 is even then shifting 119879
119896(120591) by
120587 produces the same result as replacement of 1198861with minus119886
1
(119886119896remains the same) On the other hand if 119896 is odd then
shifting 119879119896(120591) by 120587 produces the same result as replacement
of 1198861with minus119886
1and 119886119896with minus119886
119896
According to (37) coefficients of 119896th harmonic can beexpressed as
119886119896= 120582119896cos 120575 119887
119896= 120582119896sin 120575 (118)
Mathematical Problems in Engineering 15
where
|120575| le 120587 (119)
Conversely for prescribed coefficients 119886119896and 119887
119896 120575 can be
determined as
120575 = atan 2 (119887119896 119886119896) (120)
where definition of function atan 2(119910 119909) is given by (105)The main result of this section is stated in the following
proposition
Proposition 22 Every nonnegative waveform of type (35)withmaximal absolute value of coefficient 119886
1le 0 for prescribed
coefficients 119886119896and 119887119896of 119896th harmonic can be represented as
119879119896(120591)
= [1 minus cos 120591]
sdot [1 minus (119896 minus 1) 119886119896minus 2
119896minus1
sum
119899=1
(119896 minus 119899) (119886119896cos 119899120591 + 119887
119896sin 119899120591)]
(121)
if 119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) le 1+119886
119896 where 120575 = atan 2(bk
119886119896) or
119879119896(120591) = 120582
119896[1 minus cos(120591 minus (120575 + 120585)
119896)]
sdot [1 minus cos(120591 minus (120575 minus 120585)
119896)]
sdot [1198880+ 2
119896minus2
sum
119899=1
119888119899cos 119899(120591 minus 120575
119896)]
(122)
if 119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) ge 1 + 119886
119896 where 119888
119899 119899 = 0
119896minus2 and 120582119896= radic1198862119896+ 1198872119896are related to 120585 via relations (67) and
(68) respectively and |120585| le 120587
Remark 23 Expression (121) can be obtained from (38) bysetting 120591
0= 0 and 120585 = minus120575 and then replacing 120582
119896cos 120575 with
119886119896(see (118)) and 120582
119896cos(119899120591 minus 120575) with 119886
119896cos 119899120591 + 119887
119896sin 119899120591
(see also (118)) Furthermore insertion of 1205910= 0 and 120585 =
minus120575 into (43)ndash(46) leads to the following relations betweenfundamental and 119896th harmonic coefficients of waveform(121)
1198861= minus (1 + 119886
119896) 119887
1= minus119896119887
119896 (123)
On the other hand expression (122) can be obtained from(66) by replacing 120591
0minus120585119896with 120575119896 Therefore substitution of
1205910minus 120585119896 = 120575119896 in (84) leads to
1198861= minus1205821cos(120575
119896) 119887
1= minus1205821sin(120575
119896) (124)
where 1205821is given by (85)
The fundamental harmonic coefficients 1198861and 1198871of wave-
form of type (35) with maximal absolute value of coefficient1198861le 0 satisfy both relations (123) and (124) if 119886
119896and 119887119896satisfy
1 + 119886119896= 119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) For such waveforms
relations 1205910= 0 and 120585 = minus120575 also hold
Remark 24 Amplitude of 119896th harmonic of nonnegativewaveform of type (35) with maximal absolute value of coeffi-cient 119886
1le 0 and coefficients 119886
119896 119887119896satisfying 1 + 119886
119896=
119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) is
120582119896=
sin (120575119896)119896 sin 120575 cos (120575119896) minus cos 120575 sin (120575119896)
(125)
To show that it is sufficient to substitute 119886119896= 120582119896cos 120575 (see
(118)) into 1 + 119886119896= 119896120582119896[sin 120575 sin(120575119896)] cos(120575119896)
Introducing new variable
119910 = cos(120575119896) (126)
and using the Chebyshev polynomials (eg see Appendices)relations 119886
119896= 120582119896cos 120575 and (125) can be rewritten as
119886119896= 120582119896119881119896(119910) (127)
120582119896=
1
119896119910119880119896minus1
(119910) minus 119881119896(119910)
(128)
where119881119896(119910) and119880
119896(119910) denote the Chebyshev polynomials of
the first and second kind respectively Substitution of (128)into (127) leads to
119886119896119896119910119880119896minus1
(119910) minus (1 + 119886119896) 119881119896(119910) = 0 (129)
which is polynomial equation of 119896th degree in terms of var-iable 119910 From |120575| le 120587 and (126) it follows that
cos(120587119896) le 119910 le 1 (130)
In what follows we show that 119886119896is monotonically increas-
ing function of 119910 on the interval (130) From 120585 = minus120575 (seeRemark 23) and (81) it follows that 120582minus1
119896= (119896 minus 1) cos 120575 +
119896sum119896minus1
119899=1cos((119896 minus 2119899)120575119896) ge 1 and therefore 119886
119896= 120582119896cos 120575 can
be rewritten as
119886119896=
cos 120575(119896 minus 1) cos 120575 + 119896sum119896minus1
119899=1cos ((119896 minus 2119899) 120575119896)
(131)
Obviously 119886119896is even function of 120575 and all cosines in (131)
are monotonically decreasing functions of |120575| on the interval|120575| le 120587 It is easy to show that cos((119896 minus 2119899)120575119896) 119899 =
1 (119896 minus 1) decreases slower than cos 120575 when |120575| increasesThis implies that denominator of the right hand side of(131) decreases slower than numerator Since denominator ispositive for |120575| le 120587 it further implies that 119886
119896is decreasing
function of |120575| on interval |120575| le 120587 Consequently 119886119896is
monotonically increasing function of 119910 on the interval (130)Thus we have shown that 119886
119896is monotonically increasing
function of 119910 on the interval (130) and therefore (129) hasonly one solution that satisfies (130) According to (128) thevalue of 119910 obtained from (129) and (130) either analyticallyor numerically leads to amplitude 120582
119896of 119896th harmonic
16 Mathematical Problems in Engineering
1
05
0
minus05
minus1
minus1 minus05 0 05 1
Coefficient ak
Coe
ffici
entb
k
radica2k+ b2
kle 1
k = 2k = 3k = 4
Figure 11 Plot of (119886119896 119887119896) satisfying 1 + 119886
119896= 119896120582
119896[sin 120575 sin(120575
119896)] cos(120575119896) for 119896 le 4
By solving (129) and (130) for 119896 le 4 we obtain
119910 = radic1 + 1198862
2 (1 minus 1198862) minus1 le 119886
2le1
3
119910 = radic3
4 (1 minus 21198863) minus1 le 119886
3le1
8
119910 =radicradic2 minus 4119886
4+ 1011988624minus 2 (1 minus 119886
4)
4 (1 minus 31198864)
minus1 le 1198864le
1
15
(132)
Insertion of (132) into (128) leads to the following explicitexpressions for the amplitude 120582
119896 119896 le 4
1205822=1
2(1 minus 119886
2) minus1 le 119886
2le1
3 (133)
1205822
3= [
1
3(1 minus 2119886
3)]
3
minus1 le 1198863le1
8 (134)
1205824=1
4(minus1 minus 119886
4+ radic2 minus 4119886
4+ 1011988624) minus1 le 119886
4le
1
15
(135)
Relations (133)ndash(135) define closed lines (see Figure 11) whichseparate points representing waveforms of type (121) frompoints representing waveforms of type (122) For given 119896points inside the corresponding curve refer to nonnegativewaveforms of type (121) whereas points outside curve (andradic1198862119896+ 1198872119896le 1) correspond to nonnegative waveforms of type
(122) Points on the respective curve correspond to the wave-forms which can be expressed in both forms (121) and (122)
Remark 25 Themaximum absolute value of coefficient 1198861of
nonnegative waveform of type (35) is
100381610038161003816100381611988611003816100381610038161003816max =
1
cos (120587 (2119896)) (136)
This maximum value is attained for |120585| = 1205872 and 120575 = 0
(see (124)) Notice that |1198861|max is equal to the maximum value
1205821max of amplitude of fundamental harmonic (see (113))
Coefficients of waveform with maximum absolute value ofcoefficient 119886
1 1198861lt 0 are
1198861= minus
1
cos (120587 (2119896)) 119886
119896=1
119896tan( 120587
(2119896))
1198871= 119887119896= 0
(137)
Waveformdescribed by (137) is cosinewaveformhaving zerosat 120587(2119896) and minus120587(2119896)
In the course of proving (136) notice first that |1198861|max le
1205821max holds According to (123) and (124) maximum of |119886
1|
occurs for 119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) ge 1 + 119886
119896 From (124)
it immediately follows that maximum value of |1198861| is attained
if and only if 1205821= 1205821max and 120575 = 0 which because of
120575119896 = 1205910minus120585119896 further implies 120591
0= 120585119896 Sincemaximumvalue
of 1205821is attained for |120585| = 1205872 it follows that corresponding
waveform has zeros at 120587(2119896) and minus120587(2119896)
Proof of Proposition 22 As it was mentioned earlier in thissection we can assume without loss of generality that 119886
1le 0
We consider waveforms119879119896(120591) of type (35) such that119879
119896(120591) ge 0
and119879119896(120591) = 0 for some 120591
0 Fromassumption that nonnegative
waveform 119879119896(120591) of type (35) has at least one zero it follows
that it can be expressed in form (38)Let us also assume that 120591
0is position of nondegenerate
critical point Therefore 119879119896(1205910) = 0 implies 1198791015840
119896(1205910) = 0 and
11987910158401015840
119896(1205910) gt 0 According to (55) second derivative of 119879
119896(120591) at
1205910can be expressed as 11987910158401015840
119896(1205910) = 1 minus 120582
119896(1198962
minus 1) cos 120585 Since11987910158401015840
119896(1205910) gt 0 it follows immediately that
1 minus 120582119896(1198962
minus 1) cos 120585 gt 0 (138)
Let us further assume that 119879119896(120591) has exactly one zeroThe
problem of finding maximum absolute value of 1198861is con-
nected to the problem of finding maximum of the minimumfunction (see Section 21) If waveforms possess unique globalminimum at nondegenerate critical point then correspond-ing minimum function is a smooth function of parameters[13] Consequently assumption that 119879
119896(120591) has exactly one
zero at nondegenerate critical point leads to the conclusionthat coefficient 119886
1is differentiable function of 120591
0 First
derivative of 1198861(see (43)) with respect to 120591
0 taking into
account that 1205971205851205971205910= 119896 (see (50)) can be expressed in the
following factorized form
1205971198861
1205971205910
= sin 1205910[1 minus 120582
119896(1198962
minus 1) cos 120585] (139)
Mathematical Problems in Engineering 17
From (138) and (139) it is clear that 12059711988611205971205910= 0 if and only if
sin 1205910= 0 According toRemark 12 assumption that119879
119896(120591)has
exactly one zero implies 120582119896lt 1 From (51) (48) and 120582
119896lt 1
it follows that 1198861cos 1205910+ 1198871sin 1205910lt 0 which together with
sin 1205910= 0 implies that 119886
1cos 1205910lt 0 Assumption 119886
1le 0
together with relations 1198861cos 1205910lt 0 and sin 120591
0= 0 further
implies 1198861
= 0 and
1205910= 0 (140)
Insertion of 1205910= 0 into (38) leads to
119879119896(120591)
= [1 minus cos 120591]
sdot [1 minus (119896 minus 1) 120582119896cos 120585 minus 2120582
119896
119896minus1
sum
119899=1
(119896 minus 119899) cos (119899120591 + 120585)]
(141)
Substitution of 1205910= 0 into (45) and (46) yields 119886
119896= 120582119896cos 120585
and 119887119896
= minus120582119896sin 120585 respectively Replacing 120582
119896cos 120585 with
119886119896and 120582
119896cos(119899120591 + 120585) with (119886
119896cos 119899120591 + 119887
119896sin 119899120591) in (141)
immediately leads to (121)Furthermore 119886
119896= 120582119896cos 120585 119887
119896= minus120582
119896sin 120585 and (118)
imply that
120575 = minus120585 (142)
According to (38)ndash(40) and (142) it follows that (141) is non-negative if and only if
120582119896[(119896 minus 1) cos 120575 + 119896 sin (120575 minus 120575119896)
sin (120575119896)] lt 1 (143)
Notice that 119886119896= 120582119896cos 120575 implies that the following relation
holds
120582119896[(119896 minus 1) cos 120575 + 119896 sin (120575 minus 120575119896)
sin (120575119896)]
= minus119886119896+ 119896120582119896
sin 120575sin (120575119896)
cos(120575119896)
(144)
Finally substitution of (144) into (143) leads to 119896120582119896[sin 120575
sin(120575119896)] cos(120575119896) lt 1 + 119886119896 which proves that (121) holds
when 119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) lt 1 + 119886
119896
Apart from nonnegative waveforms with exactly one zeroat nondegenerate critical point in what follows we will alsoconsider other types of nonnegative waveforms with at leastone zero According to Proposition 9 and Remark 11 thesewaveforms can be described by (66)ndash(68) providing that 0 le|120585| le 120587
According to (35) 119879119896(0) ge 0 implies 1 + 119886
1+ 119886119896ge 0
Consequently 1198861le 0 implies that |119886
1| le 1 + 119886
119896 On the other
hand according to (123) |1198861| = 1 + 119886
119896holds for waveforms
of type (121) The converse is also true 1198861le 0 and |119886
1| =
1 + 119886119896imply 119886
1= minus1 minus 119886
119896 which further from (35) implies
119879119896(0) = 0 Therefore in what follows it is enough to consider
only nonnegativewaveformswhich can be described by (66)ndash(68) and 0 le |120585| le 120587 with coefficients 119886
119896and 119887119896satisfying
119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) ge 1 + 119886
119896
For prescribed coefficients 119886119896and 119887119896 the amplitude 120582
119896=
radic1198862119896+ 1198872119896of 119896th harmonic is also prescribed According to
Remark 15 (see also Remark 16) 120582119896is monotonically
decreasing function of 119909 = cos(120585119896) The value of 119909 can beobtained by solving (90) subject to the constraint cos(120587119896) le119909 le 1 Then 120582
1can be determined from (88) From (106) it
immediately follows that maximal absolute value of 1198861le 0
corresponds to 119902 = 0 which from (104) and (120) furtherimplies that
120575 = 1198961205910minus 120585 (145)
Furthermore 119902 = 0 according to (107) implies that waveformzeros are
1205910=(120575 + 120585)
119896 120591
1015840
0= 1205910minus2120585
119896=(120575 minus 120585)
119896 (146)
Substitution of 1205910= (120575 + 120585)119896 into (66) yields (122) which
proves that (122) holds when 119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) ge
1 + 119886119896
In what follows we prove that (121) also holds when119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) = 1 + 119886
119896 Substitution of 119886
119896=
120582119896cos 120575 into 119896120582
119896[sin 120575 sin(120575119896)] cos(120575119896) = 1 + 119886
119896leads to
120582119896[(119896 minus 1) cos 120575 + 119896 sin (120575 minus 120575119896)
sin (120575119896)] = 1 (147)
As we mentioned earlier relation (142) holds for all wave-forms of type (121) Substituting (142) into (147) we obtain
120582119896[(119896 minus 1) cos 120585 + 119896 sin (120585 minus 120585119896)
sin (120585119896)] = 1 (148)
This expression can be rearranged as
120582119896
119896 sin ((119896 minus 1) 120585119896)sin 120585119896
= 1 minus (119896 minus 1) 120582119896cos 120585 (149)
On the other hand for waveforms of type (122) according to(68) relations (148) and (149) also hold Substitution of 120591
0=
(120575 + 120585)119896 (see (145)) and (67) into (122) leads to
119879119896(120591)
= 120582119896[1 minus cos (120591 minus 120591
0)]
sdot [119896 sin ((119896 minus 1) 120585119896)
sin 120585119896minus 2
119896minus1
sum
119899=1
(119896 minus 119899) cos (119899 (120591 minus 1205910) + 120585)]
(150)
Furthermore substitution of (142) into (145) implies that1205910
= 0 Finally substitution of 1205910
= 0 and (149) into(150) leads to (141) Therefore (141) holds when 119896120582
119896[sin 120575
sin(120575119896)] cos(120575119896) = 1 + 119886119896 which in turn shows that (121)
holds when 119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) = 1 + 119886
119896 This
completes the proof
18 Mathematical Problems in Engineering
52 Nonnegative Waveforms with Maximal Absolute Valueof Coefficient 119886
1for 119896 = 3 Nonnegative waveform of type
(35) for 119896 = 3 is widely used in PA design (eg see [10])In this subsection we illustrate results of Section 51 for thisparticular case The case 119896 = 2 is presented in detail in [12]
Coefficients of fundamental harmonic of nonnegativewaveform of type (35) with 119896 = 3 and maximal absolutevalue of coefficient 119886
1le 0 for prescribed coefficients 119886
3and
1198873(1205823= radic11988623+ 11988723) according to (123) (124) (134) (94) and
(120) are equal to
1198861= minus1 minus 119886
3 119887
1= minus3119887
3 (151)
if 12058223le [(1 minus 2119886
3)3]3
1198861= minus1205821cos(120575
3) 119887
1= minus1205821sin(120575
3) (152)
where 1205821= 3(
3radic1205823minus 1205823) and 120575 = atan 2(119887
3 1198863) if [(1 minus
21198863)3]3
le 1205822
3le 1The line 1205822
3= [(1minus2119886
3)3]3 (see case 119896 = 3
in Figure 11) separates points representing waveforms withcoefficients satisfying (151) from points representing wave-forms with coefficients satisfying (152) Waveforms describedby (151) for 1205822
3lt [(1 minus 2119886
3)3]3 have exactly one zero at
1205910= 0 Waveforms described by (151) and (152) for 1205822
3= [(1 minus
21198863)3]3 also have zero at 120591
0= 0 These waveforms as a rule
have exactly two zeros However there are two exceptionsone related to the maximally flat nonnegative waveform withcoefficients 119886
1= minus98 119886
3= 18 and 119887
1= 1198873= 0 which
has only one zero and the other related to the waveform withcoefficients 119886
1= 0 119886
3= minus1 and 119887
1= 1198873= 0 which has three
zerosWaveforms described by (152) for [(1minus21198863)3]3
lt 1205822
3lt
1 have two zeros Waveforms with 1205823= 1 have only third
harmonic (fundamental harmonic is zero)Plot of contours of maximal absolute value of coefficient
1198861 1198861le 0 for prescribed coefficients 119886
3and 1198873is presented
in Figure 12 According to Remark 25 the waveform withmaximum absolute value of 119886
1le 0 is fully described with
the following coefficients 1198861
= minus2radic3 1198863
= radic39 and1198871= 1198873= 0 This waveform has two zeros at plusmn1205876
Two examples of nonnegative waveforms for 119896 = 3
and maximal absolute value of coefficient 1198861 1198861le 0 with
prescribed coefficients 1198863and 1198873are presented in Figure 13
One waveform corresponds to the case 12058223lt [(1 minus 2119886
3)3]3
(solid line) and the other to the case 12058223gt [(1 minus 2119886
3)3]3
(dashed line)Thewaveform represented by solid line has onezero and its coefficients are 119886
3= minus01 119887
3= 01 119886
1= minus09
and 1198871= minus03 Dashed line corresponds to the waveform
having two zeros with coefficients 1198863= minus01 119887
3= 03 119886
1=
minus08844 and 1198871= minus06460 (case 1205822
3gt [(1 minus 2119886
3)3]3)
6 Nonnegative Cosine Waveforms withat Least One Zero
Nonnegative cosine waveforms have proved to be of impor-tance for waveform modelling in PA design (eg see [10])In this section we consider nonnegative cosine waveforms
1
05
0
minus05
minus1
minus1 minus05 0 05 1
Coefficient a3
Coe
ffici
entb
3
02
04
06
08
10
11
Figure 12 Contours ofmaximal absolute value of coefficient 1198861 1198861le
0 as a function of 1198863and 1198873
2
1
0
0 1 2 3 4
Angle 120591120587
Wav
efor
ms
a3 = minus01 b3 = 01
a3 = minus01 b3 = 03
Figure 13 Nonnegative waveforms for 119896 = 3 and maximal absolutevalue of 119886
1 1198861le 0 with prescribed coefficients 119886
3and 1198873
containing fundamental and 119896th harmonic with at least onezero
Cosine waveform with dc component fundamental and119896th harmonic can be obtained from (35) by setting 119887
1= 119887119896=
0 that is
119879119896(120591) = 1 + 119886
1cos 120591 + 119886
119896cos 119896120591 (153)
In Section 61 we provide general description of non-negative cosine waveforms of type (153) with at least onezero We show that nonnegative cosine waveforms with atleast one zero coincide with nonnegative cosine waveformswith maximal absolute value of coefficient 119886
1for prescribed
coefficient 119886119896 In Section 62 we illustrate results of Section 61
for particular case 119896 = 3
61 Nonnegative Cosine Waveforms with at Least One Zerofor 119896 ge 2 Amplitudes of fundamental and 119896th harmonic
Mathematical Problems in Engineering 19
of cosine waveform of type (153) are 1205821= |1198861| and 120582
119896=
|119886119896| respectively According to (42) for nonnegative cosine
waveforms of type (153) the following relation holds
minus1 le 119886119896le 1 (154)
This explains why 119896th harmonic coefficient 119886119896in Proposi-
tion 26 goes through interval [minus1 1]Waveforms (153) with 119886
1ge 0 can be obtained from
waveforms with 1198861le 0 by shifting by 120587 and therefore with-
out loss of generality we can assume that 1198861le 0
Proposition 26 Each nonnegative cosine waveform of type(153) with 119886
1le 0 and at least one zero can be represented as
119879119896(120591) = [1 minus cos 120591] [1 minus (119896 minus 1) 119886
119896minus 2119886119896
119896minus1
sum
119899=1
(119896 minus 119899) cos 119899120591]
(155)
if minus1 le 119886119896le 1(119896
2
minus 1) or
119879119896(120591) = 119886
119896[1 minus cos (120591 minus 120591
0)] [1 minus cos (120591 + 120591
0)]
sdot [1198880+ 2
119896minus2
sum
119899=1
119888119899cos 119899120591]
(156)
where
119888119899=sin ((119896 minus 119899) 120591
0) cos 120591
0minus (119896 minus 119899) cos ((119896 minus 119899) 120591
0) sin 120591
0
sin31205910
(157)
119886119896=
sin 1205910
119896 sin (1198961205910) cos 120591
0minus cos (119896120591
0) sin 120591
0
(158)
100381610038161003816100381612059101003816100381610038161003816 le
120587
119896 (159)
if 1(1198962 minus 1) le 119886119896le 1
Remark 27 Identity [1minuscos(120591minus1205910)][1minuscos(120591+120591
0)] = [cos 120591
0minus
cos 120591]2 implies that (156) can be rewritten as
119879119896(120591) = 119886
119896[cos 120591
0minus cos 120591]2 [119888
0+ 2
119896minus2
sum
119899=1
119888119899cos 119899120591] (160)
Furthermore substitution of (157) into (160) leads to
119879119896(120591) = 119886
119896[cos 120591
0minus cos 120591]
sdot [(119896 minus 1) sin 119896120591
0
sin 1205910
minus 2
119896minus1
sum
119899=1
sin ((119896 minus 119899) 1205910)
sin 1205910
cos 119899120591]
(161)
Remark 28 All nonnegative cosine waveforms of type (153)with at least one zero and 119886
1le 0 except one of them can be
represented either in form (155) or form (156)This exceptionis maximally flat cosine waveform with 119886
1lt 0 which can be
obtained from (155) for 119886119896= 1(119896
2
minus 1) or from (156) for 1205910=
0 Maximally flat cosine waveform with 1198861lt 0 can also be
obtained from (70) by setting 1205910= 0 Furthermore setting
1205910= 0 in (71) leads to maximally flat cosine waveforms for
119896 le 4 and 1198861lt 0
Remark 29 Nonnegative cosine waveform of type (155) with1198861lt 0 and minus1 lt 119886
119896le 1(119896
2
minus 1) has exactly one zero at120591 = 0 Nonnegative cosine waveform described by (156) with1198861lt 0 and 1(1198962 minus 1) lt 119886
119896lt 1 has two zeros at plusmn120591
0 where
0 lt |1205910| lt 120587119896 For 119886
119896= ∓1 nonnegative cosine waveform
of type (153) reduces to 119879119896(120591) = 1 ∓ cos 119896120591 (clearly these two
waveforms both have 119896 zeros)
Remark 30 Transformation of (155) into an additive formleads to the following relation
1198861= minus1 minus 119886
119896 (162)
where minus1 le 119886119896le 1(119896
2
minus1) Similarly transformation of (156)leads to the following relation
1198861= minus119886119896
119896 sin 1198961205910
sin 1205910
(163)
where 119886119896is given by (158) 1(1198962minus1) le 119886
119896le 1 and |120591
0| le 120587119896
Notice that coefficients of maximally flat cosine waveformnamely 119886
119896= 1(119896
2
minus1) and 1198861= minus1198962
(1198962
minus1) satisfy relation(162) They also satisfy relation (163) for 120591
0= 0
Remark 31 Nonnegative cosine waveforms of type (153) withat least one zero coincide with nonnegative cosine waveformswith maximal absolute value of coefficient 119886
1for prescribed
coefficient 119886119896
In proving that Remark 31 holds notice that expression(155) can be obtained from (121) by setting 119887
119896= 0 Fur-
thermore if 119886119896
ge 0 then 120582119896
= 119886119896 which together
with 119887119896
= 0 and (118) implies 120575 = 0 In this case119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) le 1 + 119886
119896becomes 1198962119886
119896le 1 +
119886119896 On the other hand if 119886
119896lt 0 then 120582
119896= minus119886
119896 which
together with 119887119896= 0 and (118) implies |120575| = 120587 In this case
119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) le 1 + 119886
119896becomes 0 le 1 + 119886
119896
Therefore every nonnegative cosine waveform of type (155)has maximal absolute value of coefficient 119886
1for prescribed
coefficient 119886119896 when minus1 le 119886
119896le 1(119896
2
minus 1)Let us now show that expression (156) can be obtained
from (122) by setting 119887119896= 0 and 119886
119896gt 0 For waveforms
of type (122) according to (118) 119887119896= 0 and 119886
119896gt 0 imply
120575 = 0 and 120582119896= 119886119896 Substitution of 120582
119896= 119886119896and 120575 = 0 into
119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) ge 1+119886
119896leads to 119886
119896ge 1(119896
2
minus1)Furthermore substitution of 120575 = 0 into (145) yields 120591
0= 120585119896
Insertion of 120582119896= 119886119896 120575 = 0 and 120591
0= 120585119896 into (122) leads
to (156) Therefore every nonnegative cosine waveform oftype (156) has maximal absolute value of coefficient 119886
1for
prescribed coefficient 119886119896 when 1(1198962 minus 1) le 119886
119896le 1
Proof of Proposition 26 Let us start with nonnegative cosinewaveform of type (153) with 120582
119896= |119886119896| = 1 According to
Remark 7 120582119896= |119886119896| = 1 implies that 120582
1= |1198861| = 0
Substitution of 119886119896
= minus1 into (155) and using (A2) (seeAppendices) lead to 119879
119896(120591) = 1 minus cos 119896120591 Consequently (155)
holds for 119886119896= minus1 On the other hand substitution of 119886
119896= 1
into (158) yields |1205910| = 120587119896 Furthermore substitution of
119886119896= 1 and 120591
0= 120587119896 (or 120591
0= minus120587119896) into (156) along
with performing all multiplications and using (A2) leads to
20 Mathematical Problems in Engineering
119879119896(120591) = 1 + cos 119896120591 Consequently (156)ndash(158) hold for 119886
119896= 1
and |1205910| = 120587119896
It is easy to see that 120582119896= |119886119896| lt 1 and 119879
119896(1205910) = 0 for some
1205910imply 120582
1= |1198861| = 0 Therefore in what follows we assume
that |119886119896| = 1 and 119886
1lt 0
Cosine waveforms are even functions of 120591 Therefore ifnonnegative cosine waveform has exactly one zero it has to beeither at 0 or at 120587 On the other hand if nonnegative cosinewaveform with 119886
1= 0 has exactly two zeros then these zeros
are placed at plusmn1205910 such that 120591
0is neither 0 nor 120587
In order to prove that (155) holds for minus1 lt 119886119896le 1(1minus119896
2
)let us start by referring to the description (38) of nonnegativewaveformswith at least one zero As wementioned earlier fornonnegative cosine waveformwith exactly one zero (denotedby 1205910) it is either 120591
0= 0 or 120591
0= 120587 Therefore in both cases
sin 1205910= 0 Substitution of sin 120591
0= 0 into (43) together with
1198861
= 0 and 120582119896= |119886119896| lt 1 leads to
1205910= 0 (164)
Clearly 1205910= 0 119887
1= 0 and 119887
119896= 0 according to (44) and (46)
imply 120582119896sin 120585 = 0 Since 120582
119896= |119886119896| it follows that |119886
119896| sin 120585 = 0
also holds which further implies 120582119896= 119886119896= 0 or sin 120585 =
0 In the case when 120582119896= 119886119896= 0 from (164) and (43) we
obtain 1198861= minus1 which further implies that 119879
119896(120591) = 1 minus cos 120591
Consequently (155) holds for 119886119896= 0 In the case when sin 120585 =
0 from (164) and (45) we obtain 119886119896= 120582119896if 120585 = 0 or 119886
119896= minus120582119896
if 120585 = 120587 Relations 119886119896= 120582119896and 120585 = 0 according to (40) imply
that 0 le 119886119896le 1(1 minus 119896
2
) Substitution of 120585 = 0 120582119896= 119886119896 and
(164) into (38) leads to (155) which proves that (155) holdsfor 0 le 119886
119896le 1(1 minus 119896
2
) On the other hand relations 119886119896=
minus120582119896and 120585 = 120587 according to (40) imply that minus1 lt 119886
119896le 0
Substitution of 120585 = 120587 120582119896= minus119886119896 and (164) into (38) also leads
to (155) which proves that (155) also holds for minus1 lt 119886119896le 0
Consequently (155) holds for minus1 lt 119886119896le 1(1 minus 119896
2
)In what follows we first prove that (156)-(157) hold for
1(1 minus 1198962
) lt 119886119896lt 1 For this purpose let us start with non-
negative waveforms with two zeros described by (66) As wementioned before nonnegative cosine waveforms with twozeros have zeros at 120591
0and minus120591
0 such that 120591
0= 0 and 120591
0= 120587
Relations 1198861
lt 0 and 1198871
= 0 according to (84) implycos(1205910minus 120585119896) = 1 and therefore
120585
119896= 1205910 (165)
From 120585119896 = 1205910and 0 lt |120585| lt 120587 it follows that 0 lt |120591
0| lt
120587119896 Insertion of 120585119896 = 1205910into (45) yields 119886
119896= 120582119896 Relations
119886119896= 120582119896and (82) imply that 1(1 minus 1198962) lt 119886
119896lt 1 Substitution
of 120582119896= 119886119896and 120585119896 = 120591
0into (66)ndash(68) leads to (156)ndash(158)
which proves that (156)ndash(158) hold for 1(1 minus 1198962
) lt 119886119896lt 1
and 0 lt |1205910| lt 120587119896
Finally substitution of 119886119896= 1(1 minus 119896
2
) and 1205910= 0 into
(161) leads to
119879119896(120591) =
[1 minus cos 120591](1 minus 1198962)
[119896 (119896 minus 1) minus 2
119896minus1
sum
119899=1
(119896 minus 119899) cos 119899120591]
(166)
Waveform (166) coincides with waveform (155) for 119886119896
=
1(1 minus 1198962
) which in turn proves that (156) holds for 119886119896=
1(1 minus 1198962
) and 1205910= 0 This completes the proof
62 Nonnegative CosineWaveforms with at Least One Zero for119896 = 3 In this subsection we consider nonnegative cosinewaveforms with at least one zero for 119896 = 3 (for case 119896 = 2
see [12])Cosine waveform with fundamental and third harmonic
reads
1198793(120591) = 1 + 119886
1cos 120591 + 119886
3cos 3120591 (167)
For 1198861le 0 and minus1 le 119886
3le 18 according to (155) non-
negative cosine waveform of type (167) with at least one zerocan be expressed as
1198793(120591) = (1 minus cos 120591) [1 minus 2119886
3(1 + 2 cos 120591 + cos 2120591)] (168)
From 1198793(120591 + 120587) = 2 minus 119879
3(120591) it immediately follows that for
1198861ge 0 and minus18 le 119886
3le 1 119879
3(120591) can be expressed as
1198793(120591) = (1 + cos 120591) [1 + 2119886
3(1 minus 2 cos 120591 + cos 2120591)] (169)
For 1198861le 0 and 18 le 119886
3le 1 from (158) it follows that 119886
3=
[8cos31205910]minus1 This relation along with (160) and (157) further
implies that 1198793(120591) can be expressed as
1198793(120591) =
[cos 1205910minus cos 120591]2 [2 cos 120591
0+ cos 120591]
2cos31205910
(170)
providing that |1205910| le 1205873 From 119879
3(120591 + 120587) = 2 minus 119879
3(120591) it
follows that (170) also holds for 1198861ge 0 and minus1 le 119886
3le minus18
providing that 1205910isin [21205873 41205873]
Maximally flat nonnegative cosinewaveformof type (167)with 119886
1lt 0 (minimum at 120591
0= 0) reads 119879
3(120591) = [1 minus
cos 120591]2[1 + (12) cos 120591] Dually maximally flat nonnegativecosine waveform with 119886
1gt 0 (minimum at 120591
0= 120587) reads
1198793(120591) = [1 + cos 120591]2[1 minus (12) cos 120591]In what follows we provide relations between coefficients
1198861and 1198863of nonnegative cosine waveforms of type (167) with
at least one zeroFor 1198861le 0 conversion of (168) into an additive form
immediately leads to the following relation
1198861= minus1 minus 119886
3for minus 1 le 119886
3le1
8 (171)
Conversion of (170) into an additive form leads to 1198861
=
minus31198863(2 cos 2120591
0+ 1) which can be also expressed as 119886
1=
minus31198863(4cos2120591
0minus 1) For 119886
1le 0 relations |120591
0| le 1205873 119886
1=
minus31198863(4cos2120591
0minus 1) and 119886
3= [8cos3120591
0]minus1 lead to
1198861= minus3 [ 3radic119886
3minus 1198863] for 1
8le 1198863le 1 (172)
Similarly for 1198861ge 0 conversion of (169) into an additive form
leads to the following relation
1198861= 1 minus 119886
3for minus
1
8le 1198863le 1 (173)
Mathematical Problems in Engineering 21
1
15
15
05
0
minus05
minus15minus15
minus1
minus1 minus05 0 05 1
Coefficient a1
Coe
ffici
enta
3 a1= minus
3(3radica3minus a3)
a1 =
minus1 minus a3
a1 =
1 minusa3
a1 =3(3radic|a3
| +a3)
Figure 14 Parameter space of cosine waveforms for 119896 = 3
For waveform of type (170) with 1198861ge 0 relations 120591
0isin [21205873
41205873] 1198861= minus3119886
3(4cos2120591
0minus 1) and 119886
3= [8cos3120591
0]minus1 lead to
1198861= 3 [
3radic10038161003816100381610038161198863
1003816100381610038161003816 + 1198863] for minus 1 le 1198863le minus
1
8 (174)
Every cosine waveform of type (167) corresponds to apair of real numbers (119886
1 1198863) and vice versa Points (119886
1 1198863)
in grey area in Figure 14 correspond to nonnegative cosinewaveforms for 119896 = 3 The points at the boundary of grey areacorrespond to nonnegative cosinewaveformswith at least onezero A number of shapes of nonnegative cosine waveformswith 119896 = 3 and at least one zero plotted on interval [minus120587 120587]are also presented in Figure 14 The boundary of grey area inFigure 14 consists of four line segments described by relations(171)ndash(174) The common point of line segments (172) and(173) is cusp point with coordinates 119886
1= 0 and 119886
3= 1
Another cusp point with coordinates 1198861= 0 and 119886
3= minus1
is the common point of line segments (171) and (174) Thecommon point of line segments (171)-(172) has coordinates(minus98 18) and common point of line segments (173)-(174)has coordinates (98 minus18) These points are representedby white circle dots and they correspond to maximallyflat cosine waveforms (eg see [21]) White triangle dotswith coordinates (2radic3 minusradic39) and (minus2radic3radic39) refer tothe nonnegative cosine waveforms with maximum value ofamplitude of fundamental harmonic
7 Four Case Studies of Usage of NonnegativeWaveforms in PA Efficiency Analysis
In this section we provide four case studies of usage ofdescription of nonnegative waveforms with fundamental and119896th harmonic in PA efficiency analysis In first two casestudies to be presented in Section 71 voltage is nonnegativewaveform with fundamental and second harmonic with atleast one zero In remaining two case studies to be consideredin Section 72 voltage waveform contains fundamental andthird harmonic
i
v Load
Vdc
Idc
Lch
vL
iL
Cb
+ +in
Figure 15 Generic PA circuit diagram
Let us consider generic PA circuit diagram as shown inFigure 15 We assume here that voltage and current wave-forms at the transistor output are
V (120579) = 1 + 1198861V cos 120579 + 1198871V sin 120579 + 119886119896V cos 119896120579 + 119887119896V sin 119896120579
119894 (120579) = 1 + 1198861119894cos 120579 +
infin
sum
119899=2
119886119899119894cos 119899120579
(175)
where 120579 stands for 120596119905 Both waveforms are normalized in thesense that dc components of voltage and current are 119881dc =
1 and 119868dc = 1 respectively Under assumption that blockingcapacitor 119862
119887behaves as short-circuit at the fundamental and
higher harmonics current and voltage waveforms at the loadare
V119871(120579) = 119886
1V cos 120579 + 1198871V sin 120579 + 119886119896V cos 119896120579 + 119887119896V sin 119896120579
119894119871(120579) = minus119886
1119894cos 120579 minus
infin
sum
119899=2
119886119899119894cos 119899120579
(176)
In terms of coefficients of voltage and current waveforms theload impedance at fundamental harmonic is 119911
1= minus(119886
1V minus
1198951198871V)1198861119894 whereas load impedance at 119896th harmonic is 119911
119896=
minus(119886119896Vminus119895119887119896V)119886119896119894 All other harmonics are short-circuited (119911
119899=
0 for 119899 = 1 and 119899 = 119896) Time average output power of PA (egsee [10]) with waveform pair (175) at fundamental frequencycan be expressed as
1198751= minus
11988611198941198861V
2 (177)
For normalized waveforms (175) with 119881dc = 1 and 119868dc = 1dc power is 119875dc = 1 Consequently PA efficiency 120578 = 119875
1119875dc
(eg see [10 26]) is equal to
120578 = minus11988611198941198861V
2 (178)
Thus time average output power 1198751of PA with pair of nor-
malized waveform (175) is equal to efficiency (178)Power utilization factor (PUF) is defined [26] as ldquothe
ratio of power delivered in a given situation to the power
22 Mathematical Problems in Engineering
delivered by the same device with the same supply voltagein Class A moderdquo Since the output power in class-A modeis 1198751class-A = max[V(120579)] sdotmax[119894(120579)]8 (eg see [9]) it follows
that power utilization factor PUF = 11987511198751class-A for PA with
pair of normalized waveforms (175) can be expressed as
PUF =8120578
max [V (120579)] sdotmax [119894 (120579)] (179)
71 NonnegativeWaveforms for 119896 = 2 in PAEfficiencyAnalysisIn this subsection we provide two case studies of usage ofdescription of nonnegative waveforms with fundamental andsecond harmonic (119896 = 2) in PA efficiency analysis For moreexamples of usage of descriptions of nonnegative waveformswith fundamental and second harmonic in PA efficiencyanalysis see [12]
Case Study 71 In this case study we consider efficiency of PAfor given second harmonic impedance providing that voltageis nonnegative waveform with fundamental and second har-monic and current is ldquohalf-sinerdquo waveform frequently used inefficiency analysis of classical PA operation (eg see [10])
Standard model of current waveform for classical PAoperation has the form (eg see [10 26])
119894119863(120579) =
119868119863[cos 120579 minus cos(120572
2)] |120579| le
120572
2
0120572
2le |120579| le 120587
(180)
where 120572 is conduction angle and 119868119863
gt 0 Since 119894119863(120579) is
even function it immediately follows that its Fourier seriescontains only dc component and cosine terms
119894119863(120579) = 119868dc +
infin
sum
119899=1
119868119899cos 119899120579 (181)
The dc component of the waveform (180) is
119868dc =119868119863120572
2120587[sinc(120572
2) minus cos(120572
2)] (182)
where sinc119909 = (sin119909)119909 The coefficient of the fundamentalharmonic component reads
1198681=119868119863120572
2120587(1 minus sinc120572) (183)
and the coefficient of 119899th harmonic component can bewrittenin the form
119868119899=119868119863
119899120587[sin ((119899 minus 1) 1205722)
(119899 minus 1)minussin ((119899 + 1) 1205722)
(119899 + 1)] 119899 ge 2
(184)
For ldquohalf-sinerdquo current waveform conduction angle is equalto 120587 (class-B conduction angle) According to (182) thisfurther implies that 119868dc = 119868
119863120587 To obtain normalized form of
waveform (180) we set 119868dc = 1which implies that 119868119863= 120587 Fur-
thermore substitution of 120572 = 120587 and 119868119863= 120587 in (180) leads to
119894 (120579) =
120587 cos 120579 |120579| lt120587
2
0120587
2lt |120579| le 120587
(185)
Similarly substitution of 119868119863= 120587 and 120572 = 120587 into (183) and
(184) leads to the coefficients of waveform (185) Coefficientsof fundamental and second harmonic respectively are
1198861119894=120587
2 119886
2119894=2
3 (186)
On the other hand voltage waveform of type (35) for 119896 =2 reads
V (120579) = 1 + 1198861V cos 120579 + 1198871V sin 120579 + 1198862V cos 2120579 + 1198872V sin 2120579
(187)
This waveform contains only fundamental and second har-monic and therefore all harmonics of order higher thantwo are short-circuited (119911
119899= 0 for 119899 gt 2) For current
voltage pair (185) and (187) load impedance at fundamentalharmonic is 119911
1= minus(119886
1V minus 1198951198871V)1198861119894 whereas load impedance
at second harmonic is 1199112= minus(119886
2Vminus1198951198872V)1198862119894 According to ourassumption the load is passive and therefore Re119911
1 gt 0 and
Re1199112 ge 0 which further imply 119886
11198941198861V lt 0 and 119886
21198941198862V le 0
respectivelyIt is easy to see that problem of findingmaximal efficiency
of PA with current-voltage pair (185) and (187) for prescribedsecond harmonic impedance can be reduced to the problemof finding voltage waveform of type (187) with maximal coef-ficient |119886
1V| for prescribed coefficients of second harmonic(see Section 5)
The following algorithm (analogous to Algorithm 22presented in [12]) provides the procedure for calculation ofmaximal efficiency with current-voltage pair (185) and (187)for prescribed second harmonic impedance The definitionof function atan 2(119910 119909) which appears in the step (iii) of thefollowing algorithm is given by (105)
Algorithm 32 (i) Choose 1199112= 1199032+1198951199092such that |119911
2| le 1|119886
2119894|
(ii) calculate 1198862V minus 1198951198872V = minus119911
21198862119894and 120582
2V = radic11988622V + 1198872
2V(iii) if 2120582
2V le 1 minus 1198862V then calculate 119886
1V = minus1 minus 1198862V and
1198871V = minus2119887
2V else calculate 1205821V = radic81205822V(1 minus 1205822V) 1205790V minus 120585V2 =
(12)atan2(1198872V 1198862V) 1198861V = minus120582
1V cos(1205790V minus 120585V2) and 1198871V =
minus1205821V sin(1205790V minus 120585V2)(iv) calculate efficiency 120578 = minus119886
11198941198861V2
(v) calculate 1199111= minus(119886
1V minus 1198951198871V)1198861119894 and 1199112119899 = 1199112Re119911
1
In this case study coefficients of fundamental and secondharmonic of current waveform are given by (186) Maximalefficiency of PA associated with the waveform pair (185)and (187) as a function of normalized second harmonicimpedance 119911
2119899= 1199112Re119911
1 is presented in Figure 16(a)
As can be seen from Figure 16(a) efficiency of 078 isachieved at the edge of Smith chart where second harmonicimpedance has small resistive part Corresponding PUFcalculated according to (179) is presented in Figure 16(b)Peak efficiency 120578 = 1205874 = 07854 and peak value of PUF = 1
are attained when second harmonic is short-circuited (whichcorresponds to ideal class-B operation [10 26])
For example for second harmonic impedance 1199112= 01 minus
11989505 and currentwaveform (185) fromAlgorithm32 it followsthat 2120582
2V le 1 minus 1198862V Furthermore according to step (iii)
of above algorithm maximal efficiency of PA is attained
Mathematical Problems in Engineering 23
+j2
+j5
+j1
+j2
+j5
minusj2
minusj5
minusj1
minusj2
minusj5
infin0
078
02
05
1 2 5
075 07 065
06 05
075
07
065
06
05
120578 lt 05
(a)
099
095
091
083 075
067
0 59 051
051
099
095
091
083
075 0
67 059
+j5
+j1
+j2
+j5
minusj5
minusj1
minusj2
minusj5
infin
+j2
minusj2
0 02
05
1 2 5
120578 lt 05
(b)
Figure 16 (a) Contours of maximal efficiency of PA and (b) contours of corresponding PUF associated with the waveform pair (185) and(187) as functions of normalized second harmonic impedance 119911
2119899= 1199112Re119911
1
2
3
4
1
0
2 3 410
Wav
efor
ms
Angle 120579120587
i(120579)
(120579)
Figure 17 Waveform pair (185) and (187) that provides maximalefficiency for 119911
2= 01 minus 11989505
with voltage waveform of type (187) with coefficients 1198862V =
minus00667 1198872V = minus03333 119886
1V = minus09333 and 1198871V = 06667 (see
Figure 17) Corresponding efficiency PUF and normalizedsecond harmonic impedance are 120578 = 07330 PUF = 07572and 1199112119899= 01683 minus 11989508415 respectively
On the other hand for second harmonic impedance 1199112=
01 minus 11989508 and current waveform (185) from Algorithm 32 itfollows that 2120582
2V gt 1 minus 1198862V Then according to step (iii) of
above algorithm maximal efficiency is attained with voltagewaveform of type (187) with coefficients 119886
2V = minus00667 1198872V =
minus05333 1198861V = minus09333 and 119887
1V = 10572 (see Figure 18)Efficiency PUF andnormalized secondharmonic impedanceare 120578 = 07330 PUF = 06332 and 119911
2119899= 01683 minus 11989513465
respectively
2
3
4
1
0
2 3 410
Wav
efor
ms
Angle 120579120587
i(120579)
(120579)
Figure 18 Waveform pair (185) and (187) that provides maximalefficiency for 119911
2= 01 minus 11989508
Case Study 72 As another case study let us consider the effi-ciency of PA providing that current waveform is nonnegativecosine waveform up to third harmonic with maximum valueof amplitude of fundamental harmonic [22] (see also [8])
119894 (120579) = 1 +1 + radic5
2cos 120579 + 2radic5
5cos 2120579 + 5 minus radic5
10cos 3120579
(188)
and voltage waveform is nonnegative waveform of type(187) Load impedances at fundamental second and thirdharmonic are 119911
1= minus(119886
1Vminus1198951198871V)1198861119894 1199112 = minus(1198862Vminus1198951198872V)1198862119894 and
1199113= 0 respectively According to our assumption the load
is passive and therefore Re1199111 gt 0 and Re119911
2 ge 0 which
further imply 11988611198941198861V lt 0 and 119886
21198941198862V le 0 respectively
Because current waveform (188) contains only cosineterms and voltage waveform is the same as in previous case
24 Mathematical Problems in Engineering
08
075
07
06
06
065
065
05
05
075 0
7
120578 lt 05
+j2
+j5
+j1
+j2
+j5
minusj2
minusj5
minusj1
minusj2
minusj5
infin0
02
05
1 2 5
(a)
085
08 0
75 07 0
6
05 04
08
075
0405
060
7
120578 lt 05
+j2
+j5
+j1
+j2
+j5
minusj2
minusj5
minusj1
minusj2
minusj5
infin0 02
05
1 2 5
(b)
Figure 19 (a) Contours of maximal efficiency of PA and (b) contours of corresponding PUF associated with the waveform pair (187)-(188)as functions of normalized second harmonic impedance 119911
2119899= 1199112Re119911
1
study the procedure for calculation of maximal efficiency ofPA with waveform pair (187)-(188) is the same as presentedin Algorithm 32 In this case study the coefficients of funda-mental and second harmonic of current waveform are 119886
1119894=
(1 + radic5)2 and 1198862119894= 2radic55 respectively
Maximal efficiency of PA associated with the waveformpair (187)-(188) as a function of normalized secondharmonicimpedance 119911
2119899= 1199112Re119911
1 is presented in Figure 19(a)
Efficiency of 08 is achieved at the edge of Smith chart wheresecond harmonic impedance has small resistive part Thetheoretical upper bound 120578 = (1 + radic5)4 asymp 08090 isattained when second harmonic is short-circuitedWhen thisupper bound is reached both second and third harmonic areshort-circuited which implies that we are dealing with finiteharmonic class-C [6 8] or dually when current and voltageinterchange their roles with finite harmonic inverse class-C[6 9] Corresponding PUF calculated according to (179) ispresented in Figure 19(b) Peak value of PUF asymp 08541 isattained when second harmonic is short-circuited
For example for second harmonic impedance 1199112
=
007 minus 11989504 and current waveform (188) from Algorithm 32 itfollows that 2120582
2V le 1 minus 1198862V Furthermore according to step
(iii) of Algorithm 32 maximal efficiency of PA is attainedwith voltage waveform of type (187) with coefficients 119886
2V =
minus00626 1198872V = minus03578 119886
1V = minus09374 and 1198871V = 07155 (see
Figure 20) Corresponding efficiency PUF and normalizedsecond harmonic impedance are 120578 = 07584 PUF = 06337and 1199112119899= 01208 minus 11989506904 respectively
On the other hand for 1199112= 005 minus 11989507 and current
waveform (187) it follows that 21205822V gt 1minus119886
2VThen accordingto step (iii) of Algorithm 32 themaximal efficiency is attainedwith voltage waveform of type (187) with coefficients 119886
2V =
minus00447 1198872V = minus06261 119886
1V = minus09318 and 1198871V = 10007 (see
Figure 21) Efficiency PUF and normalized second harmonic
2
3
4
1
0
2 3 410
Wav
efor
ms
Angle 120579120587
i(120579)
(120579)
Figure 20 Waveform pair (187)-(188) that provides maximal effi-ciency for 119911
2= 007 minus 11989504
impedance are 120578 = 07538 PUF = 05314 and 1199112119899= 00868minus
11989512156 respectively
72 Nonnegative Waveforms for 119896 = 3 in PA EfficiencyAnalysis In this subsection we provide another two casestudies of usage of description of nonnegative waveforms inPA efficiency analysis this time with fundamental and thirdharmonic (119896 = 3)
Case Study 73 Let us consider current-voltage pair such thatvoltage is nonnegative waveform with fundamental and thirdharmonic
V (120579) = 1 + 1198861V cos 120579 + 1198871V sin 120579 + 1198863V cos 3120579 + 1198873V sin 3120579
(189)
Mathematical Problems in Engineering 25
2
3
4
1
0
2 3 410
Wav
efor
ms
Angle 120579120587
i(120579)
(120579)
Figure 21 Waveform pair (187)-(188) that provides maximal effi-ciency for 119911
2= 005 minus 11989507
and current is nonnegative cosine waveform given by (188)Load impedances at fundamental second and third har-monic are 119911
1= minus(119886
1V minus 1198951198871V)1198861119894 1199112 = 0 and 119911
3= minus(119886
3V minus
1198951198873V)1198863119894 respectively According to our assumption the load
is passive and therefore Re1199111 gt 0 and Re119911
3 ge 0 which
further imply 11988611198941198861V lt 0 and 119886
31198941198863V le 0
In this subsection we consider the problem of findingmaximal efficiency of PA with waveform pair (188)-(189)for given third harmonic impedance As we mentionedearlier problem of finding maximal efficiency of PA withcurrent-voltage pair (188)-(189) for prescribed third har-monic impedance can be reduced to the problem of findingvoltage waveform of type (189) withmaximal coefficient |119886
1V|for prescribed coefficients of third harmonic (see Section 52)
The following algorithm provides the procedure forcalculation of maximal efficiency with current-voltage pair(188)-(189) The definition of function atan 2(119910 119909) whichappears in step (iii) of the following algorithm is given by(105)
Algorithm 33 (i) Choose 1199113= 1199033+1198951199093such that |119911
3| le 1|119886
3119894|
(ii) calculate 1198863V minus 1198951198873V = minus119911
31198863119894and 120582
3V = radic11988623V + 1198872
3V(iii) if 271205822
3V le (1 minus 21198863V)3 then calculate 119886
1V = minus1 minus 1198863V
and 1198871V = minus3119887
3V else calculate 1205821V = 3(3radic1205823V minus 120582
3V) 1205790V minus120585V3 = (13)atan 2(119887
3V 1198863V) 1198861V = minus1205821V cos(1205790V minus 120585V3) and
1198871V = minus120582
1V sin(1205790V minus 120585V3)(iv) calculate efficiency 120578 = minus119886
11198941198861V2
(v) calculate 1199111= minus(119886
1V minus 1198951198871V)1198861119894 and 1199113119899 = 1199113Re119911
1
In this case study coefficients of fundamental and thirdharmonic of current waveform are 119886
1119894= (1 + radic5)2 and
1198863119894= (5 minus radic5)10 respectively For the waveform pair (188)-
(189) maximal efficiency of PA as a function of normalizedthird harmonic impedance 119911
3119899= 1199113Re119911
1 is presented in
Figure 22 Efficiency of 08 is reached when third harmonicimpedance has small resistive part Peak efficiency 120578 = (1 +
radic5)4 asymp 08090 is achieved when third harmonic is short-circuited
+j2
+j5
+j1
+j2
+j5
minusj2
minusj5
minusj1
minusj2
minusj5
infin0
08
08
075
075
07
07
06
05
02
05
1 2
Figure 22 Contours of maximal efficiency of PA associated withthe waveform pair (188)-(189) as a function of normalized thirdharmonic impedance 119911
3119899= 1199113Re119911
1
For the present case study in what follows we show thatpower utilization factor is proportional to efficiency For volt-age waveform of type (189) it is easy to see that V(120579 + 120587) =
2minusV(120579) holdsThis relation along with the fact that waveformV(120579) that provides maximal efficiency has at least one zeroimplies that max[V(120579)] = 2 On the other hand currentwaveform (188) is cosine waveform with positive coefficientsand therefore max[119894(120579)] = 119894(0) = 2 + 4radic5 Consequentlyaccording to (179) the following relation holds
PUFCase study 73 = 2 (5 minus 2radic5) 120578 = 10557120578 (190)
Clearly the ratio PUF120578 is constant and therefore in this casestudy PUF can be easily calculated from the correspondingefficiency Accordingly peak efficiency and peak value ofPUFCase study 73 = 3radic52 minus 52 = 08541 are attained forthe same voltage waveform (when third harmonic is short-circuited)
In the first example current waveform (188) and 1199113=
02 minus 11989505 imply that 2712058223V le (1 minus 2119886
3V)3 Then according
to Algorithm 33 the voltage waveform of type (189) thatprovides maximal efficiency has the following coefficients1198863V = minus00553 119887
3V = minus01382 1198861V = minus09447 and 119887
1V =
04146 (see Figure 23) Efficiency PUF and normalized thirdharmonic impedance are 120578 = 07643 PUF = 08069 and1199113119899= 03425 minus 11989508564 respectivelyIn the second example current waveform (188) and 119911
3=
01 minus 11989511 imply that 2712058223V gt (1 minus 2119886
3V)3 Then according
to Algorithm 33 the voltage waveform of type (189) thatprovides maximal efficiency has the following coefficients1198863V = minus00276 119887
3V = minus03040 1198861V = minus09391 and 119887
1V =
05807 (see Figure 24) Efficiency PUF and normalized thirdharmonic impedance are 120578 = 07598 PUF = 08021 and1199113119899= 01723 minus 11989518952 respectively
26 Mathematical Problems in Engineering
2
3
4
1
0
2 3 410
Wav
efor
ms
Angle 120579120587
i(120579)
(120579)
Figure 23 Waveform pair (188)-(189) that provides maximal effi-ciency for 119911
3= 02 minus 11989505
2
3
4
1
0
2 3 410
Wav
efor
ms
Angle 120579120587
i(120579)
(120579)
Figure 24 Waveform pair (188)-(189) that provides maximal effi-ciency for 119911
3= 01 minus 11989511
Case Study 74 In this case study let us consider current-voltage pair where current is normalized waveform of type(180) with conduction angle 120572 = 115120587 (207∘) and voltageis nonnegative waveform of type (189) Substitution of 120572 =
115120587 and 119868dc = 1 into (182) leads to 119868119863
= 22535Furthermore substitution of 120572 = 115120587 and 119868
119863= 22535 into
(180) leads to
119894 (120579)
=
22535 [cos 120579 minus cos(1151205872
)] |120579| le115120587
2
0115120587
2le |120579| le 120587
(191)
Similarly substitution of 120572 = 115120587 and 119868119863
= 22535 into(183) and (184) for 119899 = 3 yields coefficients of fundamentaland third harmonic of waveform (191)
1198861119894= 14586 119886
3119894= minus01026 (192)
Because current waveform (191) contains only cosineterms and voltage waveform is the same as in previous case
+j2
+j5
+j1
+j2
+j5
minusj2
minusj5
minusj1
minusj2
minusj5
infin0 02
05
1 2
074
076
08
084
082
078
076
074
Figure 25 Contours of maximal efficiency of PA associated withthe waveform pair (189) and (191) as a function of normalized thirdharmonic impedance 119911
3119899= 1199113Re119911
1
study the procedure for calculation of maximal efficiencyof PA with waveform pair (189)ndash(191) is the same as thatpresented in Algorithm 33 In this case study the coefficientsof fundamental and third harmonic of current waveform aregiven by (192)
For the waveform pair (189) and (191) maximal efficiencyof PA as a function of normalized third harmonic impedance1199113119899= 1199113Re119911
1 is presented in Figure 25 Efficiency of 084
is obtained in vicinity of 1199113119899= 23685 (corresponding to 119911
3=
18750) Peak efficiency 120578 asymp 08421 is achieved for voltagewaveform of type (189) with coefficients 119886
1V = minus2radic3 1198863V =
radic39 and 1198871V = 1198873V = 0
In the course of finding power utilization factor noticethat currentwaveformof type (191) attains itsmaximumvaluefor 120579 = 0 Insertion of max[119894(120579)] = 119894(0) = 278 andmax[V(120579)] = 2 for voltage waveform of type (189) into (179)leads to
PUFCase study 74 = 1439120578 (193)
Again the ratio PUF120578 is constant and PUF can be easilycalculated from the corresponding efficiency Accordinglypeak value of PUFCase study 74 asymp 12118 and peak efficiencyare attained for the same voltage waveform
In the first example current waveform (191) and 1199113=
1 minus 11989502 imply that 2712058223V le (1 minus 2119886
3V)3 Then according to
Algorithm 33 voltage waveform of type (189) which providesmaximal efficiency has coefficients 119886
3V = 01026 1198873V =
00205 1198861V = minus11026 and 119887
1V = minus00616 (see Figure 26)Efficiency PUF and normalized third harmonic impedanceare 120578 = 08042 PUF = 11572 and 119911
3119899= 13228 minus 11989502646
respectivelyIn second example current waveform (191) and 119911
3=
15 minus 11989512 imply that 2712058223V gt (1 minus 2119886
3V)3 Then according to
Algorithm 33 voltage waveform of type (189) which providesmaximal efficiency has coefficients 119886
3V = 01540 1198873V =
Mathematical Problems in Engineering 27
2
3
4
1
0
2 3 410
Wav
efor
ms
Angle 120579120587
i(120579)
(120579)
Figure 26 Waveform pair (189) and (191) that provides maximalefficiency for 119911
3= 1 minus 11989502
2
3
4
1
0
2 3 410
Wav
efor
ms
Angle 120579120587
i(120579)
(120579)
Figure 27 Waveform pair (189) and (191) that provides maximalefficiency for 119911
3= 15 minus 11989512
01232 1198861V = minus11255 and 119887
1V = minus02575 (see Figure 27)Efficiency PUF and normalized third harmonic impedanceare 120578 = 08208 PUF = 11812 and 119911
3119899= 19439 minus 11989515552
respectively
8 Conclusion
In this paper we consider a problem of finding general de-scriptions of various classes of nonnegative waveforms withfundamental and 119896th harmonic These classes include non-negative waveforms with at least one zero nonnegative wave-forms with maximal amplitude of fundamental harmonic forprescribed amplitude of 119896th harmonic nonnegative wave-forms withmaximal coefficient of cosine part of fundamentalharmonic for prescribed coefficients of 119896th harmonic andnonnegative cosine waveforms with at least one zero Mainresults are stated in six propositions (Propositions 1 6 918 22 and 26) four corollaries (Corollaries 2ndash5) twentyremarks and three algorithms Four case studies of usageof closed form descriptions of nonnegative waveforms in PAefficiency analysis are considered in detail in Section 7
Appendices
Here we provide a list of finite sums of trigonometric func-tions used in this paper (Appendix A) and brief account ofthe Chebyshev polynomials (Appendix B)
A List of Some Finite Sums ofTrigonometric Functions
Dirichlet kernel (eg see [27]) is as follows
119863119896minus1
(120591) = 1 + 2
119896minus1
sum
119899=1
cos 119899120591 = sin ((2119896 minus 1) 1205912)sin (1205912)
(A1)
Fejer kernel (eg see [27]) can be expressed in the fol-lowing equivalent forms
119865119896minus1
(120591) =1
119896
119896minus1
sum
119899=0
119863119899(120591) = 1 +
2
119896
119896minus1
sum
119899=1
(119896 minus 119899) cos 119899120591
=(1 minus cos 119896120591)119896 (1 minus cos 120591)
(A2)
Lagrangersquos trigonometric identity (eg see [28]) is as fol-lows
1198781(120591) =
119896minus1
sum
119899=1
sin 119899120591 = sin (1198961205912) sin ((119896 minus 1) 1205912)sin (1205912)
(A3)
In what follows we show that the following three trigono-metric identities also hold
2
119896minus1
sum
119899=1
(119896 minus 119899) sin 119899120591 = 119896 sin 120591 minus sin 1198961205911 minus cos 120591
(A4)
119896minus1
sum
119899=1
cos (119896 minus 2119899) 120591 = sin (119896 minus 1) 120591sin 120591
(A5)
119896minus1
sum
119899=1
119899 (119896 minus 119899) cos (119896 minus 2119899) 120591
=sin (119896120591) cos 120591 minus 119896 cos (119896120591) sin 120591
2sin3120591
(A6)
Denote 1198782(120591) = 2sum
119896minus1
119899=1(119896 minus 119899) sin 119899120591 119878
3(120591) = sum
119896minus1
119899=1cos(119896 minus
2119899)120591 and 1198784(120591) = sum
119896minus1
119899=1119899(119896 minus 119899) cos(119896 minus 2119899)120591
Notice that 1198782(120591) = 2119896119878
1(120591) + 119889119863
119896minus1(120591)119889120591 which
immediately leads to (A4)
28 Mathematical Problems in Engineering
Identity (A5) can be obtained as follows
sin (119896 minus 1) 120591sin 120591
=119890119895(119896minus1)120591
minus 119890minus119895(119896minus1)120591
119890119895120591 minus 119890minus119895120591
= 119890119895119896120591
119890minus2119895120591
minus 119890minus2119895(119896minus1)120591
1 minus 119890minus2119895120591
= 119890119895119896120591
119896minus1
sum
119899=1
119890minus2119895119899120591
=
119896minus1
sum
119899=1
119890119895(119896minus2119899)120591
=
119896minus1
sum
119899=1
cos (119896 minus 2119899) 120591
(A7)
From 4119899(119896 minus 119899) = 1198962
minus (119896 minus 2119899)2 it follows that 4119878
4(120591) =
1198962
1198783(120591) + 119889
2
1198783(120591)119889120591
2 which leads to (A6)
B The Chebyshev Polynomials
The Chebyshev polynomials of the first kind 119881119899(119909) can be
defined by the following relation (eg see [29])
119881119899(119909) = cos 119899120591 when 119909 = cos 120591 (B1)
The Chebyshev polynomials of the second kind 119880119899(119909) can be
defined by the following relation (eg see [29])
119880119899(119909) =
sin (119899 + 1) 120591sin 120591
when 119909 = cos 120591 (B2)
The Chebyshev polynomials satisfy the following recur-rence relations (eg see [29])
1198810(119909) = 1 119881
1(119909) = 119909
119881119899+1
(119909) = 2119909119881119899(119909) minus 119881
119899minus1(119909)
1198800(119909) = 1 119880
1(119909) = 2119909
119880119899+1
(119909) = 2119909119880119899(119909) minus 119880
119899minus1(119909)
(B3)
The first few Chebyshev polynomials of the first and secondkind are 119881
2(119909) = 2119909
2
minus 1 1198813(119909) = 4119909
3
minus 3119909 1198814(119909) = 8119909
4
minus
81199092
+ 1 1198802(119909) = 4119909
2
minus 1 1198803(119909) = 8119909
3
minus 4119909 and 1198804(119909) =
161199094
minus 121199092
+ 1
Conflict of Interests
The authors declare that there is no conflict of interestsregarding the publication of this paper
Acknowledgment
This work is supported by the SerbianMinistry of EducationScience and Technology Development as a part of ProjectTP32016
References
[1] V I Arnolrsquod V S Afrajmovich Y S Ilrsquoyashenko and L PShilrsquonikov Dynamical Systems V Bifurcation Theory and Catas-trophe Theory Springer Berlin Germany 1994
[2] E Polak ldquoOn the mathematical foundations of nondifferen-tiable optimization in engineering designrdquo SIAM Review vol29 no 1 pp 21ndash89 1987
[3] N S Fuzik ldquoBiharmonic modes of a tuned RF power amplifierrdquoRadiotehnika vol 25 no 7 pp 62ndash71 1970 (Russian)
[4] P Colantonio F Giannini G Leuzzi and E Limiti ldquoClass Gapproach for low-voltage high-efficiency PA designrdquo Interna-tional Journal of RF and Microwave Computer-Aided Engineer-ing vol 10 no 6 pp 366ndash378 2000
[5] F H Raab ldquoMaximum efficiency and output of class-F poweramplifiersrdquo IEEE Transactions on Microwave Theory and Tech-niques vol 49 no 6 pp 1162ndash1166 2001
[6] FH Raab ldquoClass-E class-C and class-F power amplifiers basedupon a finite number of harmonicsrdquo IEEE Transactions onMicrowaveTheory and Techniques vol 49 no 8 pp 1462ndash14682001
[7] J D Rhodes ldquoOutput universality inmaximum efficiency linearpower amplifiersrdquo International Journal of Circuit Theory andApplications vol 31 no 4 pp 385ndash405 2003
[8] A Juhas and L A Novak ldquoComments on lsquoClass-E class-Cand classF power amplifier based upon a finite number ofharmonicsrsquordquo IEEE Transactions on Microwave Theory and Tech-niques vol 57 no 6 pp 1623ndash1625 2009
[9] M Roberg and Z Popovic ldquoAnalysis of high-efficiency poweramplifiers with arbitrary output harmonic terminationsrdquo IEEETransactions on Microwave Theory and Techniques vol 59 no8 pp 2037ndash2048 2011
[10] A Grebennikov N O Sokal and M J Franco Switchmode RFPower Amplifiers ElsevierAcademic Press San Diego CalifUSA 2nd edition 2012
[11] T Canning P J Tasker and S C Cripps ldquoContinuous modepower amplifier design using harmonic clipping contourstheory and practicerdquo IEEE Transactions on Microwave Theoryand Techniques vol 62 no 1 pp 100ndash110 2014
[12] A Juhas and L A Novak ldquoGeneral description of nonnegativewaveforms up to second harmonic for power amplifier mod-ellingrdquoMathematical Problems in Engineering vol 2014 ArticleID 709762 18 pages 2014
[13] V I Arnolrsquod V V Goryunov O V Lyashko and V A VasilrsquoevDynamical Systems VIIImdashSingularity Theory II ApplicationsSpringer Berlin Germany 1993
[14] D Siersma ldquoProperties of conflict sets in the planerdquo BanachCenter Publications Polish Academy of Sciences vol 50 no 1 pp267ndash276 1999 Proceedings of the Banach Center Symposiumon Geometry and Topology of Caustics (Caustics rsquo98) WarsawPoland
[15] M van Manen The geometry of conflict sets [Dissertation]Universiteit Utrecht UtrechtThe Netherlands 2003 httpigi-tur-archivelibraryuunldissertations2003-0912-123058c4pdf
[16] Y L Sachkov ldquoMaxwell strata and symmetries in the problemofoptimal rolling of a sphere over a planerdquo Sbornik Mathematicsvol 201 no 7-8 pp 1029ndash1051 2010
[17] I A Bogaevsky ldquoPerestroikas of shocks and singularities ofminimum functionsrdquoPhysicaDNonlinear Phenomena vol 173no 1-2 pp 1ndash28 2002
[18] Y L Sachkov ldquoMaxwell strata in the Euler elastic problemrdquoJournal of Dynamical andControl Systems vol 14 no 2 pp 169ndash234 2008
[19] M Siino and T Koike ldquoTopological classification of black holesgeneric Maxwell set and crease set of a horizonrdquo InternationalJournal ofModern Physics D Gravitation Astrophysics Cosmol-ogy vol 20 no 6 pp 1095ndash1122 2011
Mathematical Problems in Engineering 29
[20] F H Raab ldquoClass-F power amplifiers with maximally flat wave-formsrdquo IEEETransactions onMicrowaveTheory and Techniquesvol 45 no 11 pp 2007ndash2012 1997
[21] A Juhas and L A Novak ldquoMaximally flat waveforms with finitenumber of harmonics in class-F power amplifiersrdquo Mathemat-ical Problems in Engineering vol 2013 Article ID 169590 9pages 2013
[22] L Fejer ldquoUber trigonometrische polynomerdquo Journal fur dieReine und Angewandte Mathematik vol 1916 no 146 pp 53ndash82 1916 (German)
[23] S C Cripps ldquoBessel Waives [microwave bytes]rdquo IEEE Micro-wave Magazine vol 10 no 7 pp 30ndash36 117 2009
[24] L N Bryzgalova ldquoSingularities of the maximum of para-metrically dependent functionrdquo Functional Analysis and ItsApplications vol 11 no 1 pp 49ndash51 1977
[25] V I Arnold A A Davydov V A Vassiliev and V MZakalyukin Mathematical Models of Catastrophes Control ofCatastrophic Process Encyclopedia of Life Support Systems(EOLSS) EOLSS Publishers Oxford UK 2006
[26] S C Cripps RF Power Amplifiers for Wireless CommunicationsArtech House Norwood Mass USA 2nd edition 2006
[27] A Zygmund Trigonometric Series vol 1 Cambridge UniversityPress Cambridge UK 2nd edition 1959
[28] A Jeffrey and H DaiHandbook of Mathematical Formulas andIntegrals ElsevierAcademic Press San Diego Calif USA 4thedition 2008
[29] J C Mason and D C Handscomb Chebyshev PolynomialsChapman amp Hall CRC Press Boca Raton Fla USA 2003
Submit your manuscripts athttpwwwhindawicom
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical Problems in Engineering
Hindawi Publishing Corporationhttpwwwhindawicom
Differential EquationsInternational Journal of
Volume 2014
Applied MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Probability and StatisticsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical PhysicsAdvances in
Complex AnalysisJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
OptimizationJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
CombinatoricsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Operations ResearchAdvances in
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Function Spaces
Abstract and Applied AnalysisHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of Mathematics and Mathematical Sciences
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Algebra
Discrete Dynamics in Nature and Society
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Decision SciencesAdvances in
Discrete MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom
Volume 2014 Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Stochastic AnalysisInternational Journal of
Mathematical Problems in Engineering 7
2
1
0
0 1 2 3 4
Angle 120591120587
Wav
efor
ms
1205823 = radic2201205823 = radic281205823 = radic24
Figure 4 Nonnegative waveforms with at least one zero for 119896 = 31205910= 1205876 and 120585 = 31205874
Three examples of nonnegative waveforms with at leastone zero for 119896 = 3 are presented in Figure 4 (examples ofnonnegative waveformswith at least one zero for 119896 = 2 can befound in [12]) For all three waveforms presented in Figure 4we assume that 120591
0= 1205876 and 120585 = 31205874 From (40) it follows
that 1205823le radic24 Coefficients of waveform with 120582
3= radic220
(dotted line) are 1198861= minus08977 119887
1= minus03451 119886
3= 005
and 1198873= minus005 Coefficients of waveform with 120582
3= radic28
(dashed line) are 1198861= minus09453 119887
1= minus01127 119886
3= 0125
and 1198873= minus0125 Coefficients of waveform with 120582
3= radic24
(solid line) are 1198861= minus10245 119887
1= 02745 119886
3= 025 and
1198873= minus025 First two waveforms have one zero while third
waveform (presented with solid line) has two zeros
Proof of Proposition 6 Waveform of type (35) containingdc component fundamental and 119896th harmonic can be alsoexpressed in the form
119879119896(120591) = 1 + 120582
1cos (120591 + 120593
1) + 120582119896cos (119896120591 + 120593
119896) (47)
where1205821ge 0120582
119896ge 0120593
1isin (minus120587 120587] and120593
119896isin (minus120587 120587] It is easy
to see that relations between coefficient of (35) and param-eters of (47) read as follows
1198861= 1205821cos1205931 119887
1= minus1205821sin1205931 (48)
119886119896= 120582119896cos120593119896 119887
119896= minus120582119896sin120593119896 (49)
Let us introduce 120585 such that10038161003816100381610038161205851003816100381610038161003816 le 120587 120585 = (119896120591
0+ 120593119896) mod2120587 (50)
Using (50) coefficients (49) can be expressed as (45)-(46)Let us assume that 119879
119896(120591) is nonnegative waveform of type
(35) with at least one zero that is 119879119896(120591) ge 0 and 119879
119896(1205910) = 0
for some 1205910 Notice that conditions 119879
119896(120591) ge 0 and 119879
119896(1205910) = 0
imply that 1198791015840119896(1205910) = 0 From 119879
119896(1205910) = 0 and 1198791015840
119896(1205910) = 0 by
using (50) it follows that
1205821cos (120591
0+ 1205931) = minus (1 + 120582
119896cos 120585)
1205821sin (1205910+ 1205931) = minus119896120582
119896sin 120585
(51)
respectively On the other hand 1205821cos(120591 + 120593
1) can be rewrit-
ten as
1205821cos (120591 + 120593
1) = 1205821cos (120591
0+ 1205931) cos (120591 minus 120591
0)
minus 1205821sin (1205910+ 1205931) sin (120591 minus 120591
0)
(52)
Substitution of (51) into (52) yields
1205821cos (120591 + 120593
1) = minus (1 + 120582
119896cos 120585) cos (120591 minus 120591
0)
+ 119896120582119896sin 120585 sin (120591 minus 120591
0)
(53)
According to (50) it follows that cos(119896120591 + 120593119896) = cos(119896(120591 minus
1205910) + 120585) that is
cos (119896120591 + 120593119896) = cos 120585 cos 119896 (120591 minus 120591
0) minus sin 120585 sin 119896 (120591 minus 120591
0)
(54)
Furthermore substitution of (54) and (53) into (47) leads to
119879119896(120591) = [1 minus cos (120591 minus 120591
0)] [1 + 120582
119896cos 120585]
minus 120582119896[1 minus cos 119896 (120591 minus 120591
0)] cos 120585
+ 120582119896[119896 sin (120591 minus 120591
0) minus sin 119896 (120591 minus 120591
0)] sin 120585
(55)
According to (A2) and (A4) (see Appendices) there is com-mon factor [1minuscos(120591minus120591
0)] for all terms in (55) Consequently
(55) can be written in the form (38) where
119903119896(120591) = minus cos 120585 + [
1 minus cos 119896 (120591 minus 1205910)
1 minus cos (120591 minus 1205910)] cos 120585
minus [119896 sin (120591 minus 120591
0) minus sin 119896 (120591 minus 120591
0)
1 minus cos (120591 minus 1205910)
] sin 120585
(56)
From (56) by using (A2) (A4) and cos 120585 cos 119899(120591 minus 1205910) minus
sin 120585 sin 119899(120591 minus 1205910) = cos(119899(120591 minus 120591
0) + 120585) we obtain (39)
In what follows we are going to prove that (40) also holdsAccording to (38) 119879
119896(120591) is nonnegative if and only if
120582119896max120591
119903119896(120591) le 1 (57)
Let us first show that position of global maximumof 119903119896(120591)
belongs to the interval |120591 minus 1205910| le 2120587119896 Relation (56) can be
rewritten as
119903119896(120591) = 119903
119896(1205910minus2120585
119896) + 119902119896(120591) (58)
where
119903119896(1205910minus2120585
119896) = (119896 minus 1) cos 120585 + 119896 sin (120585 minus 120585119896)
sin (120585119896) (59)
119902119896(120591) =
1
1 minus cos (120591 minus 1205910)
sdot [cos 120585 minus cos(119896(120591 minus 1205910+120585
119896))
+119896 sin 120585sin (120585119896)
(cos(120591 minus 1205910+120585
119896) minus cos(120585
119896))]
(60)
8 Mathematical Problems in Engineering
For |120585| lt 120587 relation sin 120585 sin(120585119896) gt 0 obviously holds Fromcos 119905 gt cos 1199051015840 for |119905| le 120587119896 lt |119905
1015840
| le 120587 it follows that positionof global maximum of the function of type [119888 cos 119905 minus cos(119896119905)]for 119888 gt 0 belongs to interval |119905| le 120587119896 Therefore position ofglobal maximum of the expression in the square brackets in(60) for |120585| lt 120587 belongs to interval |120591 minus 120591
0+ 120585119896| le 120587119896 This
inequality together with |120585| lt 120587 leads to |120591minus1205910| lt 2120587119896 Since
[1 minus cos(120591 minus 1205910)]minus1 decreases with increasing |120591 minus 120591
0| le 120587 it
follows that 119902119896(120591) for |120585| lt 120587 has global maximum on interval
|120591minus1205910| lt 2120587119896 For |120585| = 120587 it is easy to show thatmax
120591119902119896(120591) =
119902119896(1205910plusmn 2120587119896) = 0 Since 119903
119896(120591) minus 119902
119896(120591) is constant (see (58))
it follows from previous considerations that 119903119896(120591) has global
maximum on interval |120591 minus 1205910| le 2120587119896
To find max120591119903119896(120591) let us consider first derivative of 119903
119896(120591)
with respect to 120591 Starting from (56) first derivative of 119903119896(120591)
can be expressed in the following form
119889119903119896(120591)
119889120591= minus119904 (120591) sdot sin(
119896 (120591 minus 1205910)
2+ 120585) (61)
where
119904 (120591) = [sin(119896 (120591 minus 120591
0)
2) cos(
120591 minus 1205910
2)
minus 119896 cos(119896 (120591 minus 120591
0)
2) sin(
120591 minus 1205910
2)]
sdot sinminus3 (120591 minus 1205910
2)
(62)
Using (A6) (see Appendices) (62) can be rewritten as
119904 (120591) = 2
119896minus1
sum
119899=1
119899 (119896 minus 119899) cos((119896 minus 2119899) (120591 minus 120591
0)
2) (63)
From 119899(119896 minus 119899) gt 0 and |119896 minus 2119899| lt 119896 119899 = 1 (119896 minus 1) itfollows that all summands in (63) decrease with increasing|120591 minus 1205910| providing that |120591 minus 120591
0| le 2120587119896 Therefore 119904(120591) ge 119904(120591
0plusmn
2120587119896) = 119896sin2(120587119896) gt 0 for |120591 minus 1205910| le 2120587119896 Consequently
119889119903119896(120591)119889120591 = 0 and |120591minus120591
0| le 2120587119896 imply that sin(119896(120591minus120591
0)2+
120585) = 0From |120585| le 120587 |120591minus120591
0| le 2120587119896 and sin(119896(120591minus120591
0)2+120585) = 0
it follows that 120591minus1205910+120585119896 = minus120585119896 or |120591minus120591
0+120585119896| = (2120587minus|120585|)119896
and therefore cos(119896(120591 minus 1205910+ 120585119896)) = cos 120585 Since cos(120585119896) ge
cos(2120587 minus |120585|)119896 it follows that max120591119902119896(120591) is attained for 120591 =
1205910minus2120585119896 Furthermore from (60) it follows that max
120591119902119896(120591) =
119902119896(1205910minus 2120585119896) = 0 which together with (58)-(59) leads to
max120591
119903119896(120591) = 119903
119896(1205910minus2120585
119896)
= (119896 minus 1) cos 120585 + 119896 sin (120585 minus 120585119896)sin (120585119896)
(64)
Both terms on the right hand side of (64) are even functionsof 120585 and decrease with increase of |120585| |120585| le 120587 Thereforemax120591119903119896(120591) attains its lowest value for |120585| = 120587 It is easy to
show that right hand side of (64) for |120585| = 120587 is equal to 1which further implies that
max120591
119903119896(120591) ge 1 (65)
From (65) it follows that (57) can be rewritten as 120582119896
le
[max120591119903119896(120591)]minus1 Finally substitution of (64) into 120582
119896le
[max120591119903119896(120591)]minus1 leads to (40) which completes the proof
32 Nonnegative Waveforms with Two Zeros Nonnegativewaveforms of type (35) with two zeros always possess twoglobal minima Such nonnegative waveforms are thereforerelated to the conflict set
In this subsection we provide general description of non-negative waveforms of type (35) for 119896 ge 2 and exactly twozeros According to Remark 7 120582
119896= 1 implies |120585| = 120587 and
119879119896(120591) = 1 minus cos 119896(120591 minus 120591
0) Number of zeros of 119879
119896(120591) = 1 minus
cos 119896(120591minus1205910) on fundamental period equals 119896 which is greater
than two for 119896 gt 2 and equal to two for 119896 = 2 In the followingproposition we exclude all waveforms with 120582
119896= 1 (the case
when 119896 = 2 and 1205822= 1 is going to be discussed in Remark 10)
Proposition 9 Every nonnegative waveform of type (35) withexactly two zeros can be expressed in the following form
119879119896(120591) = 120582
119896[1 minus cos (120591 minus 120591
0)] [1 minus cos(120591 minus 120591
0+2120585
119896)]
sdot [1198880+ 2
119896minus2
sum
119899=1
119888119899cos 119899(120591 minus 120591
0+120585
119896)]
(66)
where
119888119899= [sin(120585 minus 119899120585
119896) cos(120585
119896)
minus (119896 minus 119899) cos(120585 minus 119899120585
119896) sin(120585
119896)]
sdot sinminus3 (120585119896)
(67)
120582119896= [(119896 minus 1) cos 120585 + 119896 sin(120585 minus 120585119896)
sin(120585119896)]
minus1
(68)
0 lt10038161003816100381610038161205851003816100381610038161003816 lt 120587 (69)
Remark 10 For 119896 = 2 waveforms with 1205822= 1 also have
exactly two zeros These waveforms can be included in aboveproposition by substituting (69) with 0 lt |120585| le 120587
Remark 11 Apart from nonnegative waveforms of type (35)with two zeros there are another two types of nonnegativewaveforms which can be obtained from (66)ndash(68) These are
(i) nonnegative waveforms with 119896 zeros (correspondingto |120585| = 120587) and
(ii) maximally flat nonnegative waveforms (correspond-ing to 120585 = 0)
Notice that nonnegative waveforms of type (35) with120582119896= 1 can be obtained from (66)ndash(68) by setting |120585| =
120587 Substitution of 120582119896
= 1 and |120585| = 120587 into (66) alongwith execution of all multiplications and usage of (A2) (seeAppendices) leads to 119879
119896(120591) = 1 minus cos 119896(120591 minus 120591
0)
Mathematical Problems in Engineering 9
Also maximally flat nonnegative waveforms (they haveonly one zero [21]) can be obtained from (66)ndash(68) by setting120585 = 0 Thus substitution of 120585 = 0 into (66)ndash(68) leads tothe following form of maximally flat nonnegative waveformof type (35)
119879119896(120591) =
[1 minus cos (120591 minus 1205910)]2
3 (1198962 minus 1)
sdot [119896 (1198962
minus 1)
+ 2
119896minus2
sum
119899=1
(119896 minus 119899) ((119896 minus 119899)2
minus 1) cos 119899 (120591 minus 1205910)]
(70)
Maximally flat nonnegative waveforms of type (35) for 119896 le 4
can be expressed as
1198792(120591) =
2
3[1 minus cos(120591 minus 120591
0)]2
1198793(120591) =
1
2[1 minus cos(120591 minus 120591
0)]2
[2 + cos (120591 minus 1205910)]
1198794(120591) =
4
15[1 minus cos (120591 minus 120591
0)]2
sdot [5 + 4 cos (120591 minus 1205910) + cos 2 (120591 minus 120591
0)]
(71)
Remark 12 Every nonnegative waveform of type (35) withexactly one zero at nondegenerate critical point can bedescribed as in Proposition 6 providing that symbol ldquolerdquoin relation (40) is replaced with ldquoltrdquo This is an immediateconsequence of Propositions 6 and 9 and Remark 11
Remark 13 Identity [1minus cos(120591minus1205910)][1minus cos(120591minus120591
0+2120585119896)] =
[cos 120585119896 minus cos(120591 minus 1205910+ 120585119896)]
2 implies that (66) can be alsorewritten as
119879119896(120591) = 120582
119896[cos 120585119896
minus cos(120591 minus 1205910+120585
119896)]
2
sdot [1198880+ 2
119896minus2
sum
119899=1
119888119899cos 119899(120591 minus 120591
0+120585
119896)]
(72)
Furthermore substitution of (67) into (72) leads to
119879119896(120591)
= 120582119896[cos 120585119896
minus cos(120591 minus 1205910+120585
119896)]
sdot [(119896 minus 1) sin 120585sin (120585119896)
minus 2
119896minus1
sum
119899=1
sin (120585 minus 119899120585119896)sin (120585119896)
cos 119899(120591 minus 1205910+120585
119896)]
(73)
Remark 14 According to (A6) (see Appendices) it followsthat coefficients (67) can be expressed as
119888119899= 2
119896minus119899minus1
sum
119898=1
119898(119896 minus 119899 minus 119898) cos((119896 minus 119899 minus 2119898) 120585119896
) (74)
Furthermore from (74) it follows that coefficients 119888119896minus2
119888119896minus3
119888119896minus4
and 119888119896minus5
are equal to119888119896minus2
= 2 (75)
119888119896minus3
= 8 cos(120585119896) (76)
119888119896minus4
= 8 + 12 cos(2120585119896) (77)
119888119896minus5
= 24 cos(120585119896) + 16 cos(3120585
119896) (78)
For example for 119896 = 2 (75) and (68) lead to 1198880= 2 and
1205822= 1(2+cos 120585) respectively which from (72) further imply
that
1198792(120591) =
2 [cos(1205852) minus cos(120591 minus 1205910+ 1205852)]
2
[2 + cos 120585] (79)
Also for 119896 = 3 (75) (76) and (68) lead to 1198881= 2 119888
0=
8 cos(1205853) and 1205823= [2(3 cos(1205853) + cos 120585)]minus1 respectively
which from (72) further imply that
1198793(120591) =
2 [cos (1205853) minus cos (120591 minus 1205910+ 1205853)]
2
[3 cos (1205853) + cos 120585]
sdot [2 cos(1205853) + cos(120591 minus 120591
0+120585
3)]
(80)
Remark 15 According to (A5) (see Appendices) relation(68) can be rewritten as
120582119896= [(119896 minus 1) cos 120585 + 119896
119896minus1
sum
119899=1
cos((119896 minus 2119899)120585119896
)]
minus1
(81)
Clearly amplitude 120582119896of 119896th harmonic of nonnegative wave-
form of type (35) with exactly two zeros is even functionof 120585 Since cos((119896 minus 2119899)120585119896) 119899 = 0 (119896 minus 1) decreaseswith increase of |120585| on interval 0 le |120585| le 120587 it follows that120582119896monotonically increases with increase of |120585| Right hand
side of (68) is equal to 1(1198962
minus 1) for 120585 = 0 and to one for|120585| = 120587 Therefore for nonnegative waveforms of type (35)with exactly two zeros the following relation holds
1
1198962 minus 1lt 120582119896lt 1 (82)
The left boundary in (82) corresponds to maximally flatnonnegative waveforms (see Remark 11) The right boundaryin (82) corresponds to nonnegative waveforms with 119896 zeros(also see Remark 11)
Amplitude of 119896th harmonic of nonnegative waveform oftype (35) with two zeros as a function of parameter 120585 for 119896 le5 is presented in Figure 5
Remark 16 Nonnegative waveform of type (35) with twozeros can be also expressed in the following form
119879119896(120591) = 1 minus 120582
119896
119896 sin 120585sin (120585119896)
cos(120591 minus 1205910+120585
119896)
+ 120582119896cos (119896 (120591 minus 120591
0) + 120585)
(83)
10 Mathematical Problems in Engineering
1
08
06
04
02
0minus1 minus05 0 05
Am
plitu
de120582k
1
Parameter 120585120587
k = 2k = 3
k = 4k = 5
Figure 5 Amplitude of 119896th harmonic of nonnegative waveformwith two zeros as a function of parameter 120585
where 120582119896is given by (68) and 0 lt |120585| lt 120587 From (83) it follows
that coefficients of fundamental harmonic of nonnegativewaveform of type (35) with two zeros are
1198861= minus1205821cos(120591
0minus120585
119896) 119887
1= minus1205821sin(120591
0minus120585
119896) (84)
where 1205821is amplitude of fundamental harmonic
1205821=
119896 sin 120585sin (120585119896)
120582119896 (85)
Coefficients of 119896th harmonic are given by (45)-(46)Notice that (68) can be rewritten as
120582119896= [cos(120585
119896)
119896 sin 120585sin(120585119896)
minus cos 120585]minus1
(86)
By introducing new variable
119909 = cos(120585119896) (87)
and using the Chebyshev polynomials (eg see Appendices)relations (85) and (86) can be rewritten as
1205821= 119896120582119896119880119896minus1
(119909) (88)
120582119896=
1
119896119909119880119896minus1
(119909) minus 119881119896(119909)
(89)
where119881119896(119909) and119880
119896(119909) denote the Chebyshev polynomials of
the first and second kind respectively From (89) it followsthat
120582119896[119896119909119880119896minus1
(119909) minus 119881119896(119909)] minus 1 = 0 (90)
which is polynomial equation of 119896th degree in terms of var-iable 119909 From 0 lt |120585| lt 120587 and (87) it follows that
cos(120587119896) lt 119909 lt 1 (91)
Since 120582119896is monotonically increasing function of |120585| 0 lt |120585| lt
120587 it follows that 120582119896is monotonically decreasing function of
119909 This further implies that (90) has only one solution thatsatisfies (91) (For 119896 = 2 expression (91) reads cos(1205872) le
119909 lt 1) This solution for 119909 (which can be obtained at leastnumerically) according to (88) leads to amplitude 120582
1of
fundamental harmonicFor 119896 le 4 solutions of (90) and (91) are
119909 = radic1 minus 1205822
21205822
1
3lt 1205822le 1
119909 =1
23radic1205823
1
8lt 1205823lt 1
119909 = radic1
6(1 + radic
51205824+ 3
21205824
)1
15lt 1205824lt 1
(92)
Insertion of (92) into (88) leads to the following relationsbetween amplitude 120582
1of fundamental and amplitude 120582
119896of
119896th harmonic 119896 le 4
1205821= radic8120582
2(1 minus 120582
2)
1
3lt 1205822le 1 (93)
1205821= 3 (
3radic1205823minus 1205823)
1
8lt 1205823lt 1 (94)
1205821= radic
32
27(radic2120582
4(3 + 5120582
4)3
minus 21205824(9 + 7120582
4))
1
15lt 1205824lt 1
(95)
Proof of Proposition 9 As it has been shown earlier (seeProposition 6) nonnegative waveform of type (35) with atleast one zero can be represented in form (38) Since weexclude nonnegative waveforms with 120582
119896= 1 according to
Remark 7 it follows that we exclude case |120585| = 120587Therefore inthe quest for nonnegative waveforms of type (35) having twozeros we will start with waveforms of type (38) for |120585| lt 120587It is clear that nonnegative waveforms of type (38) have twozeros if and only if
120582119896= [max120591
119903119896(120591)]minus1
(96)
and max120591119903119896(120591) = 119903
119896(1205910) According to (64) max
120591119903119896(120591) =
119903119896(1205910) implies |120585| = 0 Therefore it is sufficient to consider
only the interval (69)Substituting (96) into (38) we obtain
119879119896(120591) =
[1 minus cos (120591 minus 1205910)] [max
120591119903119896(120591) minus 119903
119896(120591)]
max120591119903119896(120591)
(97)
Mathematical Problems in Engineering 11
Expression max120591119903119896(120591) minus 119903
119896(120591) according to (64) and (39)
equals
max120591
119903119896(120591) minus 119903
119896(120591) = 119896
sin ((119896 minus 1) 120585119896)sin (120585119896)
minus 2
119896minus1
sum
119899=1
(119896 minus 119899) cos (119899 (120591 minus 1205910) + 120585)
(98)
Comparison of (97) with (66) yields
max120591
119903119896(120591) minus 119903
119896(120591) = [1 minus cos(120591 minus 120591
0+2120585
119896)]
sdot [1198880+ 2
119896minus2
sum
119899=1
119888119899cos 119899(120591 minus 120591
0+120585
119896)]
(99)
where coefficients 119888119899 119899 = 0 119896 minus 2 are given by (67) In
what follows we are going to show that right hand sides of(98) and (99) are equal
From (67) it follows that
1198880minus 1198881cos(120585
119896) = 119896
sin (120585 minus 120585119896)sin (120585119896)
(100)
Also from (67) for 119899 = 1 119896minus3 it follows that the followingrelations hold
(119888119899minus1
+ 119888119899+1
) cos(120585119896) minus 2119888
119899= 2 (119896 minus 119899) cos(120585 minus 119899120585
119896)
(119888119899minus1
minus 119888119899+1
) sin(120585119896) = 2 (119896 minus 119899) sin(120585 minus 119899120585
119896)
(101)
From (99) by using (75) (76) (100)-(101) and trigonometricidentities
cos(120591 minus 1205910+2120585
119896) = cos(120585
119896) cos(120591 minus 120591
0+120585
119896)
minus sin(120585119896) sin(120591 minus 120591
0+120585
119896)
cos(120585 minus 119899120585
119896) cos(119899(120591 minus 120591
0+120585
119896))
minus sin(120585 minus 119899120585
119896) sin(119899(120591 minus 120591
0+120585
119896))
= cos (119899 (120591 minus 1205910) + 120585)
(102)
we obtain (98) Consequently (98) and (99) are equal whichcompletes the proof
33 Nonnegative Waveforms with Two Zeros and PrescribedCoefficients of 119896thHarmonic In this subsectionwe show thatfor prescribed coefficients 119886
119896and 119887119896 there are 119896 nonnegative
waveforms of type (35) with exactly two zeros According to
(37) and (82) coefficients 119886119896and 119887119896of nonnegative waveforms
of type (35) with exactly two zeros satisfy the followingrelation
1
1198962 minus 1lt radic1198862119896+ 1198872119896lt 1 (103)
According to Remark 16 the value of 119909 (see (87)) that cor-responds to 120582
119896= radic1198862
119896+ 1198872119896can be determined from (90)-
(91) As we mentioned earlier (90) has only one solutionthat satisfies (91) This value of 119909 according to (88) leadsto the amplitude 120582
1of fundamental harmonic (closed form
expressions for 1205821in terms of 120582
119896and 119896 le 4 are given by (93)ndash
(95))On the other hand from (45)-(46) it follows that
1198961205910minus 120585 = atan 2 (119887
119896 119886119896) + 2119902120587 119902 = 1 (119896 minus 1) (104)
where function atan 2(119910 119909) is defined as
atan 2 (119910 119909) =
arctan(119910
119909) if 119909 ge 0
arctan(119910
119909) + 120587 if 119909 lt 0 119910 ge 0
arctan(119910
119909) minus 120587 if 119909 lt 0 119910 lt 0
(105)
with the codomain (minus120587 120587] Furthermore according to (84)and (104) the coefficients of fundamental harmonic of non-negative waveforms with two zeros and prescribed coeffi-cients of 119896th harmonic are equal to
1198861= minus1205821cos[
atan 2 (119887119896 119886119896) + 2119902120587
119896]
1198871= minus1205821sin[
atan 2 (119887119896 119886119896) + 2119902120587
119896]
(106)
where 119902 = 0 (119896minus 1) For chosen 119902 according to (104) and(66) positions of zeros are
1205910=1
119896[120585 + atan 2 (119887
119896 119886119896) + 2119902120587]
1205910minus2120585
119896=1
119896[minus120585 + atan 2 (119887
119896 119886119896) + 2119902120587]
(107)
From (106) and 119902 = 0 (119896minus1) it follows that for prescribedcoefficients 119886
119896and 119887119896 there are 119896 nonnegative waveforms of
type (35) with exactly two zerosWe provide here an algorithm to facilitate calculation
of coefficients 1198861and 1198871of nonnegative waveforms of type
(35) with two zeros and prescribed coefficients 119886119896and 119887
119896
providing that 119886119896and 119887119896satisfy (103)
12 Mathematical Problems in Engineering
2
1
0
Wav
efor
ms
0 1 2 3 4
Angle 120591120587
q = 0
q = 1
q = 2
Figure 6 Nonnegative waveforms with two zeros for 119896 = 3 1198863=
minus015 and 1198873= minus02
Algorithm 17 (i) Calculate 120582119896= radic1198862119896+ 1198872119896
(ii) identify 119909 that satisfies both relations (90) and (91)(iii) calculate 120582
1according to (88)
(iv) choose integer 119902 such that 0 le 119902 le 119896 minus 1(v) calculate 119886
1and 1198871according to (106)
For 119896 le 4 by using (93) for 119896 = 2 (94) for 119896 = 3 and (95)for 119896 = 4 it is possible to calculate directly 120582
1from 120582
119896and
proceed to step (iv)For 119896 = 2 and prescribed coefficients 119886
2and 1198872 there are
two waveforms with two zeros one corresponding to 1198861lt 0
and the other corresponding to 1198861gt 0 (see also [12])
Let us take as an input 119896 = 3 1198863= minus015 and 119887
3= minus02
Execution of Algorithm 17 on this input yields 1205823= 025 and
1205821= 11399 (according to (94)) For 119902 = 0 we calculate
1198861= minus08432 and 119887
1= 07670 (corresponding waveform is
presented by solid line in Figure 6) for 119902 = 1 we calculate1198861= minus02426 and 119887
1= minus11138 (corresponding waveform is
presented by dashed line) for 119902 = 2 we calculate 1198861= 10859
and 1198871= 03468 (corresponding waveform is presented by
dotted line)As another example of the usage of Algorithm 17 let us
consider case 119896 = 4 and assume that1198864= minus015 and 119887
4= minus02
Consequently 1205824= 025 and 120582
1= 09861 (according to (95))
For 119902 = 0 3we calculate the following four pairs (1198861 1198871) of
coefficients of fundamental harmonic (minus08388 05184) for119902 = 0 (minus05184 minus08388) for 119902 = 1 (08388 minus05184) for 119902 =2 and (05184 08388) for 119902 = 3 Corresponding waveformsare presented in Figure 7
4 Nonnegative Waveforms with MaximalAmplitude of Fundamental Harmonic
In this section we provide general description of nonnegativewaveforms containing fundamental and 119896th harmonic withmaximal amplitude of fundamental harmonic for prescribedamplitude of 119896th harmonic
The main result of this section is presented in the fol-lowing proposition
3
2
1
0
Wav
efor
ms
0 1 2 3 4
Angle 120591120587
q = 0q = 1
q = 2q = 3
Figure 7 Nonnegative waveforms with two zeros for 119896 = 4 1198864=
minus015 and 1198874= minus02
Proposition 18 Every nonnegativewaveformof type (35)withmaximal amplitude 120582
1of fundamental harmonic and pre-
scribed amplitude 120582119896of 119896th harmonic can be expressed in the
following form
119879119896(120591) = [1 minus cos (120591 minus 120591
0)]
sdot [1 minus (119896 minus 1) 120582119896minus 2120582119896
119896minus1
sum
119899=1
(119896 minus 119899) cos (119899 (120591 minus 1205910))]
(108)
if 0 le 120582119896le 1(119896
2
minus 1) or
119879119896(120591) = 120582
119896[1 minus cos (120591 minus 120591
0)] [1 minus cos(120591 minus 120591
0+2120585
119896)]
sdot [1198880+ 2
119896minus2
sum
119899=1
119888119899cos 119899(120591 minus 120591
0+120585
119896)]
(109)
if 1(1198962 minus 1) le 120582119896le 1 providing that 119888
119899 119899 = 0 119896 minus 2 and
120582119896are related to 120585 via relations (67) and (68) respectively and
|120585| le 120587
Remark 19 Expression (108) can be obtained from (38) bysetting 120585 = 0 Furthermore insertion of 120585 = 0 into (43)ndash(46)leads to the following expressions for coefficients ofwaveformof type (108)
1198861= minus (1 + 120582
119896) cos 120591
0 119887
1= minus (1 + 120582
119896) sin 120591
0
119886119896= 120582119896cos (119896120591
0) 119887
119896= 120582119896sin (119896120591
0)
(110)
On the other hand (109) coincides with (66) Thereforethe expressions for coefficients of (109) and (66) also coincideThus expressions for coefficients of fundamental harmonic ofwaveform (109) are given by (84) where 120582
1is given by (85)
while expressions for coefficients of 119896th harmonic are givenby (45)-(46)
Waveforms described by (108) have exactly one zerowhile waveforms described by (109) for 1(1198962 minus 1) lt 120582
119896lt 1
Mathematical Problems in Engineering 13
14
12
1
08
06
04
02
00 05 1
Amplitude 120582k
Am
plitu
de1205821
k = 2
k = 3
k = 4
Figure 8 Maximal amplitude of fundamental harmonic as a func-tion of amplitude of 119896th harmonic
have exactly two zeros As we mentioned earlier waveforms(109) for 120582
119896= 1 have 119896 zeros
Remark 20 Maximal amplitude of fundamental harmonic ofnonnegative waveforms of type (35) for prescribed amplitudeof 119896th harmonic can be expressed as
1205821= 1 + 120582
119896 (111)
if 0 le 120582119896le 1(119896
2
minus 1) or
1205821=
119896 sin 120585119896 sin 120585 cos (120585119896) minus cos 120585 sin (120585119896)
(112)
if 1(1198962 minus 1) le 120582119896le 1 where 120585 is related to 120582
119896via (68) (or
(86)) and |120585| le 120587From (110) it follows that (111) holds Substitution of (86)
into (85) leads to (112)Notice that 120582
119896= 1(119896
2
minus 1) is the only common point ofthe intervals 0 le 120582
119896le 1(119896
2
minus 1) and 1(1198962
minus 1) le 120582119896le
1 According to (111) 120582119896= 1(119896
2
minus 1) corresponds to 1205821=
1198962
(1198962
minus1) It can be also obtained from (112) by setting 120585 = 0The waveforms corresponding to this pair of amplitudes aremaximally flat nonnegative waveforms
Maximal amplitude of fundamental harmonic of non-negative waveform of type (35) for 119896 le 4 as a function ofamplitude of 119896th harmonic is presented in Figure 8
Remark 21 Maximum value of amplitude of fundamentalharmonic of nonnegative waveform of type (35) is
1205821max =
1
cos (120587 (2119896)) (113)
This maximum value is attained for |120585| = 1205872 (see (112)) Thecorresponding value of amplitude of 119896th harmonic is 120582
119896=
(1119896) tan(120587(2119896)) Nonnegative waveforms of type (35) with1205821= 1205821max have two zeros at 1205910 and 1205910 minus 120587119896 for 120585 = 1205872 or
at 1205910and 1205910+ 120587119896 for 120585 = minus1205872
14
12
1
08
06
04
02
0minus1 minus05 0 05 1
Am
plitu
de1205821
Parameter 120585120587
k = 2k = 3k = 4
Figure 9 Maximal amplitude of fundamental harmonic as a func-tion of parameter 120585
To prove that (113) holds let us first show that the fol-lowing relation holds for 119896 ge 2
cos( 120587
2119896) lt 1 minus
1
1198962 (114)
From 119896 ge 2 it follows that sinc(120587(4119896)) gt sinc(1205874) wheresinc 119909 = (sin119909)119909 and therefore sin(120587(4119896)) gt 1(radic2119896)By using trigonometric identity cos 2119909 = 1 minus 2sin2119909 weimmediately obtain (114)
According to (111) and (112) it is clear that 1205821attains its
maximum value on the interval 1(1198962 minus 1) le 120582119896le 1 Since
120582119896is monotonic function of |120585| on interval |120585| le 120587 (see
Remark 15) it follows that 119889120582119896119889120585 = 0 for 0 lt |120585| lt 120587
Therefore to find critical points of 1205821as a function of 120582
119896
it is sufficient to find critical points of 1205821as a function of
|120585| 0 lt |120585| lt 120587 and consider its values at the end points120585 = 0 and |120585| = 120587 Plot of 120582
1as a function of parameter 120585
for 119896 le 4 is presented in Figure 9 According to (112) firstderivative of 120582
1with respect to 120585 is equal to zero if and only
if (119896 cos 120585 sin(120585119896) minus sin 120585 cos(120585119896)) cos 120585 = 0 On interval0 lt |120585| lt 120587 this is true if and only if |120585| = 1205872 Accordingto (112) 120582
1is equal to 119896
2
(1198962
minus 1) for 120585 = 0 equal to zerofor |120585| = 120587 and equal to 1 cos(120587(2119896)) for |120585| = 1205872 From(114) it follows that 1198962(1198962minus1) lt 1 cos(120587(2119896)) and thereforemaximum value of 120582
1is given by (113) Moreover maximum
value of 1205821is attained for |120585| = 1205872
According to above consideration all nonnegative wave-forms of type (35) having maximum value of amplitude offundamental harmonic can be obtained from (109) by setting|120585| = 1205872 Three of them corresponding to 119896 = 3 120585 = 1205872and three different values of 120591
0(01205876 and1205873) are presented
in Figure 10 Dotted line corresponds to 1205910= 0 (coefficients
of corresponding waveform are 1198861= minus1 119887
1= 1radic3 119886
3= 0
and 1198873= minusradic39) solid line to 120591
0= 1205876 (119886
1= minus2radic3 119887
1= 0
1198863= radic39 and 119887
3= 0) and dashed line to 120591
0= 1205873 (119886
1= minus1
1198871= minus1radic3 119886
3= 0 and 119887
3= radic39)
Proof of Proposition 18 As it has been shown earlier (Propo-sition 6) nonnegative waveform of type (35) with at least
14 Mathematical Problems in Engineering
2
1
0
0 1 2 3 4
Angle 120591120587
Wav
efor
ms
1205910 = 01205910 = 12058761205910 = 1205873
Figure 10 Nonnegative waveforms with maximum amplitude offundamental harmonic for 119896 = 3 and 120585 = 1205872
one zero can be represented in form (38) According to (43)(44) and (36) for amplitude 120582
1of fundamental harmonic of
waveforms of type (38) the following relation holds
1205821= radic(1 + 120582
119896cos 120585)2 + 11989621205822
119896sin2120585 (115)
where 120582119896satisfy (40) and |120585| le 120587
Because of (40) in the quest of finding maximal 1205821for
prescribed 120582119896 we have to consider the following two cases
(Case i)120582119896lt [(119896minus1) cos 120585 + 119896 sin(120585minus120585119896) sin(120585119896)]minus1
(Case ii)120582119896= [(119896minus1) cos 120585+119896 sin(120585minus120585119896) sin(120585119896)]minus1
Case i Since 120582119896lt [(119896 minus 1) cos 120585 + 119896 sin(120585 minus 120585119896) sin(120585119896)]minus1
implies 120582119896
= 1 according to (115) it follows that 1205821
= 0Hence 119889120582
1119889120585 = 0 implies
2120582119896sin 120585 [1 minus (1198962 minus 1) 120582
119896cos 120585] = 0 (116)
Therefore 1198891205821119889120585 = 0 if 120582
119896= 0 (Option 1) or sin 120585 = 0
(Option 2) or (1198962 minus 1)120582119896cos 120585 = 1 (Option 3)
Option 1 According to (115) 120582119896= 0 implies 120582
1= 1 (notice
that this implication shows that 1205821does not depend on 120585 and
therefore we can set 120585 to zero value)
Option 2 According to (115) sin 120585 = 0 implies 1205821= 1 +
120582119896cos 120585 which further leads to the conclusion that 120582
1is
maximal for 120585 = 0 For 120585 = 0 120582119896lt [(119896 minus 1) cos 120585 + 119896 sin(120585 minus
120585119896) sin(120585119896)]minus1 becomes 120582119896lt 1(119896
2
minus 1)
Option 3 This option leads to contradiction To show thatnotice that (119896
2
minus 1)120582119896cos 120585 = 1 and 120582
119896lt [(119896 minus
1) cos 120585 + 119896 sin(120585 minus 120585119896) sin(120585119896)]minus1 imply that (119896 minus 1) cos 120585 gtsin(120585minus120585119896) sin(120585119896) Using (A5) (see Appendices) the latestinequality can be rewritten assum119896minus1
119899=1[cos 120585minuscos((119896minus2119899)120585119896)] gt
0 But from |119896 minus 2119899| lt 119896 119899 = 1 (119896 minus 1) and |120585| le 120587
it follows that all summands are not positive and therefore(119896minus1) cos 120585 gt sin(120585minus120585119896) sin(120585119896) does not hold for |120585| le 120587
Consequently Case i implies 120585 = 0 and 120582119896lt 1(119896
2
minus 1)Finally substitution of 120585 = 0 into (38) leads to (108) whichproves that (108) holds for 120582
119896lt 1(119896
2
minus 1)
Case ii Relation120582119896= [(119896minus1) cos 120585+119896 sin(120585minus120585119896) sin(120585119896)]minus1
according to Proposition 9 and Remark 11 implies that cor-responding waveforms can be expressed via (66)ndash(68) for|120585| le 120587 Furthermore 120582
119896= [(119896 minus 1) cos 120585 + 119896 sin(120585 minus
120585119896) sin(120585119896)]minus1 and |120585| le 120587 imply 1(1198962 minus 1) le 120582119896le 1
This proves that (109) holds for 1(1198962 minus 1) le 120582119896le 1
Finally let us prove that (108) holds for 120582119896= 1(119896
2
minus
1) According to (68) (see also Remark 11) this value of 120582119896
corresponds to 120585 = 0 Furthermore substitution of 120582119896=
1(1198962
minus 1) and 120585 = 0 into (109) leads to (70) which can berewritten as
119879119896(120591) =
[1 minus cos (120591 minus 1205910)]
(1 minus 1198962)
sdot [119896 (119896 minus 1) minus 2
119896minus1
sum
119899=1
(119896 minus 119899) cos (119899 (120591 minus 1205910))]
(117)
Waveform (117) coincides with waveform (108) for 120582119896
=
1(1 minus 1198962
) Consequently (108) holds for 120582119896= 1(1 minus 119896
2
)which completes the proof
5 Nonnegative Waveforms with MaximalAbsolute Value of the Coefficient of CosineTerm of Fundamental Harmonic
In this sectionwe consider general description of nonnegativewaveforms of type (35) with maximal absolute value ofcoefficient 119886
1for prescribed coefficients of 119896th harmonicThis
type of waveform is of particular interest in PA efficiencyanalysis In a number of cases of practical interest eithercurrent or voltage waveform is prescribed In such casesthe problem of finding maximal efficiency of PA can bereduced to the problem of finding nonnegative waveformwith maximal coefficient 119886
1for prescribed coefficients of 119896th
harmonic (see also Section 7)In Section 51 we provide general description of nonneg-
ative waveforms of type (35) with maximal absolute value ofcoefficient 119886
1for prescribed coefficients of 119896th harmonic In
Section 52 we illustrate results of Section 51 for particularcase 119896 = 3
51 Nonnegative Waveforms with Maximal Absolute Value ofCoefficient 119886
1for 119896 ge 2 Waveforms 119879
119896(120591) of type (35) with
1198861ge 0 can be derived from those with 119886
1le 0 by shifting
by 120587 and therefore we can assume without loss of generalitythat 119886
1le 0 Notice that if 119896 is even then shifting 119879
119896(120591) by
120587 produces the same result as replacement of 1198861with minus119886
1
(119886119896remains the same) On the other hand if 119896 is odd then
shifting 119879119896(120591) by 120587 produces the same result as replacement
of 1198861with minus119886
1and 119886119896with minus119886
119896
According to (37) coefficients of 119896th harmonic can beexpressed as
119886119896= 120582119896cos 120575 119887
119896= 120582119896sin 120575 (118)
Mathematical Problems in Engineering 15
where
|120575| le 120587 (119)
Conversely for prescribed coefficients 119886119896and 119887
119896 120575 can be
determined as
120575 = atan 2 (119887119896 119886119896) (120)
where definition of function atan 2(119910 119909) is given by (105)The main result of this section is stated in the following
proposition
Proposition 22 Every nonnegative waveform of type (35)withmaximal absolute value of coefficient 119886
1le 0 for prescribed
coefficients 119886119896and 119887119896of 119896th harmonic can be represented as
119879119896(120591)
= [1 minus cos 120591]
sdot [1 minus (119896 minus 1) 119886119896minus 2
119896minus1
sum
119899=1
(119896 minus 119899) (119886119896cos 119899120591 + 119887
119896sin 119899120591)]
(121)
if 119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) le 1+119886
119896 where 120575 = atan 2(bk
119886119896) or
119879119896(120591) = 120582
119896[1 minus cos(120591 minus (120575 + 120585)
119896)]
sdot [1 minus cos(120591 minus (120575 minus 120585)
119896)]
sdot [1198880+ 2
119896minus2
sum
119899=1
119888119899cos 119899(120591 minus 120575
119896)]
(122)
if 119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) ge 1 + 119886
119896 where 119888
119899 119899 = 0
119896minus2 and 120582119896= radic1198862119896+ 1198872119896are related to 120585 via relations (67) and
(68) respectively and |120585| le 120587
Remark 23 Expression (121) can be obtained from (38) bysetting 120591
0= 0 and 120585 = minus120575 and then replacing 120582
119896cos 120575 with
119886119896(see (118)) and 120582
119896cos(119899120591 minus 120575) with 119886
119896cos 119899120591 + 119887
119896sin 119899120591
(see also (118)) Furthermore insertion of 1205910= 0 and 120585 =
minus120575 into (43)ndash(46) leads to the following relations betweenfundamental and 119896th harmonic coefficients of waveform(121)
1198861= minus (1 + 119886
119896) 119887
1= minus119896119887
119896 (123)
On the other hand expression (122) can be obtained from(66) by replacing 120591
0minus120585119896with 120575119896 Therefore substitution of
1205910minus 120585119896 = 120575119896 in (84) leads to
1198861= minus1205821cos(120575
119896) 119887
1= minus1205821sin(120575
119896) (124)
where 1205821is given by (85)
The fundamental harmonic coefficients 1198861and 1198871of wave-
form of type (35) with maximal absolute value of coefficient1198861le 0 satisfy both relations (123) and (124) if 119886
119896and 119887119896satisfy
1 + 119886119896= 119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) For such waveforms
relations 1205910= 0 and 120585 = minus120575 also hold
Remark 24 Amplitude of 119896th harmonic of nonnegativewaveform of type (35) with maximal absolute value of coeffi-cient 119886
1le 0 and coefficients 119886
119896 119887119896satisfying 1 + 119886
119896=
119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) is
120582119896=
sin (120575119896)119896 sin 120575 cos (120575119896) minus cos 120575 sin (120575119896)
(125)
To show that it is sufficient to substitute 119886119896= 120582119896cos 120575 (see
(118)) into 1 + 119886119896= 119896120582119896[sin 120575 sin(120575119896)] cos(120575119896)
Introducing new variable
119910 = cos(120575119896) (126)
and using the Chebyshev polynomials (eg see Appendices)relations 119886
119896= 120582119896cos 120575 and (125) can be rewritten as
119886119896= 120582119896119881119896(119910) (127)
120582119896=
1
119896119910119880119896minus1
(119910) minus 119881119896(119910)
(128)
where119881119896(119910) and119880
119896(119910) denote the Chebyshev polynomials of
the first and second kind respectively Substitution of (128)into (127) leads to
119886119896119896119910119880119896minus1
(119910) minus (1 + 119886119896) 119881119896(119910) = 0 (129)
which is polynomial equation of 119896th degree in terms of var-iable 119910 From |120575| le 120587 and (126) it follows that
cos(120587119896) le 119910 le 1 (130)
In what follows we show that 119886119896is monotonically increas-
ing function of 119910 on the interval (130) From 120585 = minus120575 (seeRemark 23) and (81) it follows that 120582minus1
119896= (119896 minus 1) cos 120575 +
119896sum119896minus1
119899=1cos((119896 minus 2119899)120575119896) ge 1 and therefore 119886
119896= 120582119896cos 120575 can
be rewritten as
119886119896=
cos 120575(119896 minus 1) cos 120575 + 119896sum119896minus1
119899=1cos ((119896 minus 2119899) 120575119896)
(131)
Obviously 119886119896is even function of 120575 and all cosines in (131)
are monotonically decreasing functions of |120575| on the interval|120575| le 120587 It is easy to show that cos((119896 minus 2119899)120575119896) 119899 =
1 (119896 minus 1) decreases slower than cos 120575 when |120575| increasesThis implies that denominator of the right hand side of(131) decreases slower than numerator Since denominator ispositive for |120575| le 120587 it further implies that 119886
119896is decreasing
function of |120575| on interval |120575| le 120587 Consequently 119886119896is
monotonically increasing function of 119910 on the interval (130)Thus we have shown that 119886
119896is monotonically increasing
function of 119910 on the interval (130) and therefore (129) hasonly one solution that satisfies (130) According to (128) thevalue of 119910 obtained from (129) and (130) either analyticallyor numerically leads to amplitude 120582
119896of 119896th harmonic
16 Mathematical Problems in Engineering
1
05
0
minus05
minus1
minus1 minus05 0 05 1
Coefficient ak
Coe
ffici
entb
k
radica2k+ b2
kle 1
k = 2k = 3k = 4
Figure 11 Plot of (119886119896 119887119896) satisfying 1 + 119886
119896= 119896120582
119896[sin 120575 sin(120575
119896)] cos(120575119896) for 119896 le 4
By solving (129) and (130) for 119896 le 4 we obtain
119910 = radic1 + 1198862
2 (1 minus 1198862) minus1 le 119886
2le1
3
119910 = radic3
4 (1 minus 21198863) minus1 le 119886
3le1
8
119910 =radicradic2 minus 4119886
4+ 1011988624minus 2 (1 minus 119886
4)
4 (1 minus 31198864)
minus1 le 1198864le
1
15
(132)
Insertion of (132) into (128) leads to the following explicitexpressions for the amplitude 120582
119896 119896 le 4
1205822=1
2(1 minus 119886
2) minus1 le 119886
2le1
3 (133)
1205822
3= [
1
3(1 minus 2119886
3)]
3
minus1 le 1198863le1
8 (134)
1205824=1
4(minus1 minus 119886
4+ radic2 minus 4119886
4+ 1011988624) minus1 le 119886
4le
1
15
(135)
Relations (133)ndash(135) define closed lines (see Figure 11) whichseparate points representing waveforms of type (121) frompoints representing waveforms of type (122) For given 119896points inside the corresponding curve refer to nonnegativewaveforms of type (121) whereas points outside curve (andradic1198862119896+ 1198872119896le 1) correspond to nonnegative waveforms of type
(122) Points on the respective curve correspond to the wave-forms which can be expressed in both forms (121) and (122)
Remark 25 Themaximum absolute value of coefficient 1198861of
nonnegative waveform of type (35) is
100381610038161003816100381611988611003816100381610038161003816max =
1
cos (120587 (2119896)) (136)
This maximum value is attained for |120585| = 1205872 and 120575 = 0
(see (124)) Notice that |1198861|max is equal to the maximum value
1205821max of amplitude of fundamental harmonic (see (113))
Coefficients of waveform with maximum absolute value ofcoefficient 119886
1 1198861lt 0 are
1198861= minus
1
cos (120587 (2119896)) 119886
119896=1
119896tan( 120587
(2119896))
1198871= 119887119896= 0
(137)
Waveformdescribed by (137) is cosinewaveformhaving zerosat 120587(2119896) and minus120587(2119896)
In the course of proving (136) notice first that |1198861|max le
1205821max holds According to (123) and (124) maximum of |119886
1|
occurs for 119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) ge 1 + 119886
119896 From (124)
it immediately follows that maximum value of |1198861| is attained
if and only if 1205821= 1205821max and 120575 = 0 which because of
120575119896 = 1205910minus120585119896 further implies 120591
0= 120585119896 Sincemaximumvalue
of 1205821is attained for |120585| = 1205872 it follows that corresponding
waveform has zeros at 120587(2119896) and minus120587(2119896)
Proof of Proposition 22 As it was mentioned earlier in thissection we can assume without loss of generality that 119886
1le 0
We consider waveforms119879119896(120591) of type (35) such that119879
119896(120591) ge 0
and119879119896(120591) = 0 for some 120591
0 Fromassumption that nonnegative
waveform 119879119896(120591) of type (35) has at least one zero it follows
that it can be expressed in form (38)Let us also assume that 120591
0is position of nondegenerate
critical point Therefore 119879119896(1205910) = 0 implies 1198791015840
119896(1205910) = 0 and
11987910158401015840
119896(1205910) gt 0 According to (55) second derivative of 119879
119896(120591) at
1205910can be expressed as 11987910158401015840
119896(1205910) = 1 minus 120582
119896(1198962
minus 1) cos 120585 Since11987910158401015840
119896(1205910) gt 0 it follows immediately that
1 minus 120582119896(1198962
minus 1) cos 120585 gt 0 (138)
Let us further assume that 119879119896(120591) has exactly one zeroThe
problem of finding maximum absolute value of 1198861is con-
nected to the problem of finding maximum of the minimumfunction (see Section 21) If waveforms possess unique globalminimum at nondegenerate critical point then correspond-ing minimum function is a smooth function of parameters[13] Consequently assumption that 119879
119896(120591) has exactly one
zero at nondegenerate critical point leads to the conclusionthat coefficient 119886
1is differentiable function of 120591
0 First
derivative of 1198861(see (43)) with respect to 120591
0 taking into
account that 1205971205851205971205910= 119896 (see (50)) can be expressed in the
following factorized form
1205971198861
1205971205910
= sin 1205910[1 minus 120582
119896(1198962
minus 1) cos 120585] (139)
Mathematical Problems in Engineering 17
From (138) and (139) it is clear that 12059711988611205971205910= 0 if and only if
sin 1205910= 0 According toRemark 12 assumption that119879
119896(120591)has
exactly one zero implies 120582119896lt 1 From (51) (48) and 120582
119896lt 1
it follows that 1198861cos 1205910+ 1198871sin 1205910lt 0 which together with
sin 1205910= 0 implies that 119886
1cos 1205910lt 0 Assumption 119886
1le 0
together with relations 1198861cos 1205910lt 0 and sin 120591
0= 0 further
implies 1198861
= 0 and
1205910= 0 (140)
Insertion of 1205910= 0 into (38) leads to
119879119896(120591)
= [1 minus cos 120591]
sdot [1 minus (119896 minus 1) 120582119896cos 120585 minus 2120582
119896
119896minus1
sum
119899=1
(119896 minus 119899) cos (119899120591 + 120585)]
(141)
Substitution of 1205910= 0 into (45) and (46) yields 119886
119896= 120582119896cos 120585
and 119887119896
= minus120582119896sin 120585 respectively Replacing 120582
119896cos 120585 with
119886119896and 120582
119896cos(119899120591 + 120585) with (119886
119896cos 119899120591 + 119887
119896sin 119899120591) in (141)
immediately leads to (121)Furthermore 119886
119896= 120582119896cos 120585 119887
119896= minus120582
119896sin 120585 and (118)
imply that
120575 = minus120585 (142)
According to (38)ndash(40) and (142) it follows that (141) is non-negative if and only if
120582119896[(119896 minus 1) cos 120575 + 119896 sin (120575 minus 120575119896)
sin (120575119896)] lt 1 (143)
Notice that 119886119896= 120582119896cos 120575 implies that the following relation
holds
120582119896[(119896 minus 1) cos 120575 + 119896 sin (120575 minus 120575119896)
sin (120575119896)]
= minus119886119896+ 119896120582119896
sin 120575sin (120575119896)
cos(120575119896)
(144)
Finally substitution of (144) into (143) leads to 119896120582119896[sin 120575
sin(120575119896)] cos(120575119896) lt 1 + 119886119896 which proves that (121) holds
when 119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) lt 1 + 119886
119896
Apart from nonnegative waveforms with exactly one zeroat nondegenerate critical point in what follows we will alsoconsider other types of nonnegative waveforms with at leastone zero According to Proposition 9 and Remark 11 thesewaveforms can be described by (66)ndash(68) providing that 0 le|120585| le 120587
According to (35) 119879119896(0) ge 0 implies 1 + 119886
1+ 119886119896ge 0
Consequently 1198861le 0 implies that |119886
1| le 1 + 119886
119896 On the other
hand according to (123) |1198861| = 1 + 119886
119896holds for waveforms
of type (121) The converse is also true 1198861le 0 and |119886
1| =
1 + 119886119896imply 119886
1= minus1 minus 119886
119896 which further from (35) implies
119879119896(0) = 0 Therefore in what follows it is enough to consider
only nonnegativewaveformswhich can be described by (66)ndash(68) and 0 le |120585| le 120587 with coefficients 119886
119896and 119887119896satisfying
119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) ge 1 + 119886
119896
For prescribed coefficients 119886119896and 119887119896 the amplitude 120582
119896=
radic1198862119896+ 1198872119896of 119896th harmonic is also prescribed According to
Remark 15 (see also Remark 16) 120582119896is monotonically
decreasing function of 119909 = cos(120585119896) The value of 119909 can beobtained by solving (90) subject to the constraint cos(120587119896) le119909 le 1 Then 120582
1can be determined from (88) From (106) it
immediately follows that maximal absolute value of 1198861le 0
corresponds to 119902 = 0 which from (104) and (120) furtherimplies that
120575 = 1198961205910minus 120585 (145)
Furthermore 119902 = 0 according to (107) implies that waveformzeros are
1205910=(120575 + 120585)
119896 120591
1015840
0= 1205910minus2120585
119896=(120575 minus 120585)
119896 (146)
Substitution of 1205910= (120575 + 120585)119896 into (66) yields (122) which
proves that (122) holds when 119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) ge
1 + 119886119896
In what follows we prove that (121) also holds when119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) = 1 + 119886
119896 Substitution of 119886
119896=
120582119896cos 120575 into 119896120582
119896[sin 120575 sin(120575119896)] cos(120575119896) = 1 + 119886
119896leads to
120582119896[(119896 minus 1) cos 120575 + 119896 sin (120575 minus 120575119896)
sin (120575119896)] = 1 (147)
As we mentioned earlier relation (142) holds for all wave-forms of type (121) Substituting (142) into (147) we obtain
120582119896[(119896 minus 1) cos 120585 + 119896 sin (120585 minus 120585119896)
sin (120585119896)] = 1 (148)
This expression can be rearranged as
120582119896
119896 sin ((119896 minus 1) 120585119896)sin 120585119896
= 1 minus (119896 minus 1) 120582119896cos 120585 (149)
On the other hand for waveforms of type (122) according to(68) relations (148) and (149) also hold Substitution of 120591
0=
(120575 + 120585)119896 (see (145)) and (67) into (122) leads to
119879119896(120591)
= 120582119896[1 minus cos (120591 minus 120591
0)]
sdot [119896 sin ((119896 minus 1) 120585119896)
sin 120585119896minus 2
119896minus1
sum
119899=1
(119896 minus 119899) cos (119899 (120591 minus 1205910) + 120585)]
(150)
Furthermore substitution of (142) into (145) implies that1205910
= 0 Finally substitution of 1205910
= 0 and (149) into(150) leads to (141) Therefore (141) holds when 119896120582
119896[sin 120575
sin(120575119896)] cos(120575119896) = 1 + 119886119896 which in turn shows that (121)
holds when 119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) = 1 + 119886
119896 This
completes the proof
18 Mathematical Problems in Engineering
52 Nonnegative Waveforms with Maximal Absolute Valueof Coefficient 119886
1for 119896 = 3 Nonnegative waveform of type
(35) for 119896 = 3 is widely used in PA design (eg see [10])In this subsection we illustrate results of Section 51 for thisparticular case The case 119896 = 2 is presented in detail in [12]
Coefficients of fundamental harmonic of nonnegativewaveform of type (35) with 119896 = 3 and maximal absolutevalue of coefficient 119886
1le 0 for prescribed coefficients 119886
3and
1198873(1205823= radic11988623+ 11988723) according to (123) (124) (134) (94) and
(120) are equal to
1198861= minus1 minus 119886
3 119887
1= minus3119887
3 (151)
if 12058223le [(1 minus 2119886
3)3]3
1198861= minus1205821cos(120575
3) 119887
1= minus1205821sin(120575
3) (152)
where 1205821= 3(
3radic1205823minus 1205823) and 120575 = atan 2(119887
3 1198863) if [(1 minus
21198863)3]3
le 1205822
3le 1The line 1205822
3= [(1minus2119886
3)3]3 (see case 119896 = 3
in Figure 11) separates points representing waveforms withcoefficients satisfying (151) from points representing wave-forms with coefficients satisfying (152) Waveforms describedby (151) for 1205822
3lt [(1 minus 2119886
3)3]3 have exactly one zero at
1205910= 0 Waveforms described by (151) and (152) for 1205822
3= [(1 minus
21198863)3]3 also have zero at 120591
0= 0 These waveforms as a rule
have exactly two zeros However there are two exceptionsone related to the maximally flat nonnegative waveform withcoefficients 119886
1= minus98 119886
3= 18 and 119887
1= 1198873= 0 which
has only one zero and the other related to the waveform withcoefficients 119886
1= 0 119886
3= minus1 and 119887
1= 1198873= 0 which has three
zerosWaveforms described by (152) for [(1minus21198863)3]3
lt 1205822
3lt
1 have two zeros Waveforms with 1205823= 1 have only third
harmonic (fundamental harmonic is zero)Plot of contours of maximal absolute value of coefficient
1198861 1198861le 0 for prescribed coefficients 119886
3and 1198873is presented
in Figure 12 According to Remark 25 the waveform withmaximum absolute value of 119886
1le 0 is fully described with
the following coefficients 1198861
= minus2radic3 1198863
= radic39 and1198871= 1198873= 0 This waveform has two zeros at plusmn1205876
Two examples of nonnegative waveforms for 119896 = 3
and maximal absolute value of coefficient 1198861 1198861le 0 with
prescribed coefficients 1198863and 1198873are presented in Figure 13
One waveform corresponds to the case 12058223lt [(1 minus 2119886
3)3]3
(solid line) and the other to the case 12058223gt [(1 minus 2119886
3)3]3
(dashed line)Thewaveform represented by solid line has onezero and its coefficients are 119886
3= minus01 119887
3= 01 119886
1= minus09
and 1198871= minus03 Dashed line corresponds to the waveform
having two zeros with coefficients 1198863= minus01 119887
3= 03 119886
1=
minus08844 and 1198871= minus06460 (case 1205822
3gt [(1 minus 2119886
3)3]3)
6 Nonnegative Cosine Waveforms withat Least One Zero
Nonnegative cosine waveforms have proved to be of impor-tance for waveform modelling in PA design (eg see [10])In this section we consider nonnegative cosine waveforms
1
05
0
minus05
minus1
minus1 minus05 0 05 1
Coefficient a3
Coe
ffici
entb
3
02
04
06
08
10
11
Figure 12 Contours ofmaximal absolute value of coefficient 1198861 1198861le
0 as a function of 1198863and 1198873
2
1
0
0 1 2 3 4
Angle 120591120587
Wav
efor
ms
a3 = minus01 b3 = 01
a3 = minus01 b3 = 03
Figure 13 Nonnegative waveforms for 119896 = 3 and maximal absolutevalue of 119886
1 1198861le 0 with prescribed coefficients 119886
3and 1198873
containing fundamental and 119896th harmonic with at least onezero
Cosine waveform with dc component fundamental and119896th harmonic can be obtained from (35) by setting 119887
1= 119887119896=
0 that is
119879119896(120591) = 1 + 119886
1cos 120591 + 119886
119896cos 119896120591 (153)
In Section 61 we provide general description of non-negative cosine waveforms of type (153) with at least onezero We show that nonnegative cosine waveforms with atleast one zero coincide with nonnegative cosine waveformswith maximal absolute value of coefficient 119886
1for prescribed
coefficient 119886119896 In Section 62 we illustrate results of Section 61
for particular case 119896 = 3
61 Nonnegative Cosine Waveforms with at Least One Zerofor 119896 ge 2 Amplitudes of fundamental and 119896th harmonic
Mathematical Problems in Engineering 19
of cosine waveform of type (153) are 1205821= |1198861| and 120582
119896=
|119886119896| respectively According to (42) for nonnegative cosine
waveforms of type (153) the following relation holds
minus1 le 119886119896le 1 (154)
This explains why 119896th harmonic coefficient 119886119896in Proposi-
tion 26 goes through interval [minus1 1]Waveforms (153) with 119886
1ge 0 can be obtained from
waveforms with 1198861le 0 by shifting by 120587 and therefore with-
out loss of generality we can assume that 1198861le 0
Proposition 26 Each nonnegative cosine waveform of type(153) with 119886
1le 0 and at least one zero can be represented as
119879119896(120591) = [1 minus cos 120591] [1 minus (119896 minus 1) 119886
119896minus 2119886119896
119896minus1
sum
119899=1
(119896 minus 119899) cos 119899120591]
(155)
if minus1 le 119886119896le 1(119896
2
minus 1) or
119879119896(120591) = 119886
119896[1 minus cos (120591 minus 120591
0)] [1 minus cos (120591 + 120591
0)]
sdot [1198880+ 2
119896minus2
sum
119899=1
119888119899cos 119899120591]
(156)
where
119888119899=sin ((119896 minus 119899) 120591
0) cos 120591
0minus (119896 minus 119899) cos ((119896 minus 119899) 120591
0) sin 120591
0
sin31205910
(157)
119886119896=
sin 1205910
119896 sin (1198961205910) cos 120591
0minus cos (119896120591
0) sin 120591
0
(158)
100381610038161003816100381612059101003816100381610038161003816 le
120587
119896 (159)
if 1(1198962 minus 1) le 119886119896le 1
Remark 27 Identity [1minuscos(120591minus1205910)][1minuscos(120591+120591
0)] = [cos 120591
0minus
cos 120591]2 implies that (156) can be rewritten as
119879119896(120591) = 119886
119896[cos 120591
0minus cos 120591]2 [119888
0+ 2
119896minus2
sum
119899=1
119888119899cos 119899120591] (160)
Furthermore substitution of (157) into (160) leads to
119879119896(120591) = 119886
119896[cos 120591
0minus cos 120591]
sdot [(119896 minus 1) sin 119896120591
0
sin 1205910
minus 2
119896minus1
sum
119899=1
sin ((119896 minus 119899) 1205910)
sin 1205910
cos 119899120591]
(161)
Remark 28 All nonnegative cosine waveforms of type (153)with at least one zero and 119886
1le 0 except one of them can be
represented either in form (155) or form (156)This exceptionis maximally flat cosine waveform with 119886
1lt 0 which can be
obtained from (155) for 119886119896= 1(119896
2
minus 1) or from (156) for 1205910=
0 Maximally flat cosine waveform with 1198861lt 0 can also be
obtained from (70) by setting 1205910= 0 Furthermore setting
1205910= 0 in (71) leads to maximally flat cosine waveforms for
119896 le 4 and 1198861lt 0
Remark 29 Nonnegative cosine waveform of type (155) with1198861lt 0 and minus1 lt 119886
119896le 1(119896
2
minus 1) has exactly one zero at120591 = 0 Nonnegative cosine waveform described by (156) with1198861lt 0 and 1(1198962 minus 1) lt 119886
119896lt 1 has two zeros at plusmn120591
0 where
0 lt |1205910| lt 120587119896 For 119886
119896= ∓1 nonnegative cosine waveform
of type (153) reduces to 119879119896(120591) = 1 ∓ cos 119896120591 (clearly these two
waveforms both have 119896 zeros)
Remark 30 Transformation of (155) into an additive formleads to the following relation
1198861= minus1 minus 119886
119896 (162)
where minus1 le 119886119896le 1(119896
2
minus1) Similarly transformation of (156)leads to the following relation
1198861= minus119886119896
119896 sin 1198961205910
sin 1205910
(163)
where 119886119896is given by (158) 1(1198962minus1) le 119886
119896le 1 and |120591
0| le 120587119896
Notice that coefficients of maximally flat cosine waveformnamely 119886
119896= 1(119896
2
minus1) and 1198861= minus1198962
(1198962
minus1) satisfy relation(162) They also satisfy relation (163) for 120591
0= 0
Remark 31 Nonnegative cosine waveforms of type (153) withat least one zero coincide with nonnegative cosine waveformswith maximal absolute value of coefficient 119886
1for prescribed
coefficient 119886119896
In proving that Remark 31 holds notice that expression(155) can be obtained from (121) by setting 119887
119896= 0 Fur-
thermore if 119886119896
ge 0 then 120582119896
= 119886119896 which together
with 119887119896
= 0 and (118) implies 120575 = 0 In this case119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) le 1 + 119886
119896becomes 1198962119886
119896le 1 +
119886119896 On the other hand if 119886
119896lt 0 then 120582
119896= minus119886
119896 which
together with 119887119896= 0 and (118) implies |120575| = 120587 In this case
119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) le 1 + 119886
119896becomes 0 le 1 + 119886
119896
Therefore every nonnegative cosine waveform of type (155)has maximal absolute value of coefficient 119886
1for prescribed
coefficient 119886119896 when minus1 le 119886
119896le 1(119896
2
minus 1)Let us now show that expression (156) can be obtained
from (122) by setting 119887119896= 0 and 119886
119896gt 0 For waveforms
of type (122) according to (118) 119887119896= 0 and 119886
119896gt 0 imply
120575 = 0 and 120582119896= 119886119896 Substitution of 120582
119896= 119886119896and 120575 = 0 into
119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) ge 1+119886
119896leads to 119886
119896ge 1(119896
2
minus1)Furthermore substitution of 120575 = 0 into (145) yields 120591
0= 120585119896
Insertion of 120582119896= 119886119896 120575 = 0 and 120591
0= 120585119896 into (122) leads
to (156) Therefore every nonnegative cosine waveform oftype (156) has maximal absolute value of coefficient 119886
1for
prescribed coefficient 119886119896 when 1(1198962 minus 1) le 119886
119896le 1
Proof of Proposition 26 Let us start with nonnegative cosinewaveform of type (153) with 120582
119896= |119886119896| = 1 According to
Remark 7 120582119896= |119886119896| = 1 implies that 120582
1= |1198861| = 0
Substitution of 119886119896
= minus1 into (155) and using (A2) (seeAppendices) lead to 119879
119896(120591) = 1 minus cos 119896120591 Consequently (155)
holds for 119886119896= minus1 On the other hand substitution of 119886
119896= 1
into (158) yields |1205910| = 120587119896 Furthermore substitution of
119886119896= 1 and 120591
0= 120587119896 (or 120591
0= minus120587119896) into (156) along
with performing all multiplications and using (A2) leads to
20 Mathematical Problems in Engineering
119879119896(120591) = 1 + cos 119896120591 Consequently (156)ndash(158) hold for 119886
119896= 1
and |1205910| = 120587119896
It is easy to see that 120582119896= |119886119896| lt 1 and 119879
119896(1205910) = 0 for some
1205910imply 120582
1= |1198861| = 0 Therefore in what follows we assume
that |119886119896| = 1 and 119886
1lt 0
Cosine waveforms are even functions of 120591 Therefore ifnonnegative cosine waveform has exactly one zero it has to beeither at 0 or at 120587 On the other hand if nonnegative cosinewaveform with 119886
1= 0 has exactly two zeros then these zeros
are placed at plusmn1205910 such that 120591
0is neither 0 nor 120587
In order to prove that (155) holds for minus1 lt 119886119896le 1(1minus119896
2
)let us start by referring to the description (38) of nonnegativewaveformswith at least one zero As wementioned earlier fornonnegative cosine waveformwith exactly one zero (denotedby 1205910) it is either 120591
0= 0 or 120591
0= 120587 Therefore in both cases
sin 1205910= 0 Substitution of sin 120591
0= 0 into (43) together with
1198861
= 0 and 120582119896= |119886119896| lt 1 leads to
1205910= 0 (164)
Clearly 1205910= 0 119887
1= 0 and 119887
119896= 0 according to (44) and (46)
imply 120582119896sin 120585 = 0 Since 120582
119896= |119886119896| it follows that |119886
119896| sin 120585 = 0
also holds which further implies 120582119896= 119886119896= 0 or sin 120585 =
0 In the case when 120582119896= 119886119896= 0 from (164) and (43) we
obtain 1198861= minus1 which further implies that 119879
119896(120591) = 1 minus cos 120591
Consequently (155) holds for 119886119896= 0 In the case when sin 120585 =
0 from (164) and (45) we obtain 119886119896= 120582119896if 120585 = 0 or 119886
119896= minus120582119896
if 120585 = 120587 Relations 119886119896= 120582119896and 120585 = 0 according to (40) imply
that 0 le 119886119896le 1(1 minus 119896
2
) Substitution of 120585 = 0 120582119896= 119886119896 and
(164) into (38) leads to (155) which proves that (155) holdsfor 0 le 119886
119896le 1(1 minus 119896
2
) On the other hand relations 119886119896=
minus120582119896and 120585 = 120587 according to (40) imply that minus1 lt 119886
119896le 0
Substitution of 120585 = 120587 120582119896= minus119886119896 and (164) into (38) also leads
to (155) which proves that (155) also holds for minus1 lt 119886119896le 0
Consequently (155) holds for minus1 lt 119886119896le 1(1 minus 119896
2
)In what follows we first prove that (156)-(157) hold for
1(1 minus 1198962
) lt 119886119896lt 1 For this purpose let us start with non-
negative waveforms with two zeros described by (66) As wementioned before nonnegative cosine waveforms with twozeros have zeros at 120591
0and minus120591
0 such that 120591
0= 0 and 120591
0= 120587
Relations 1198861
lt 0 and 1198871
= 0 according to (84) implycos(1205910minus 120585119896) = 1 and therefore
120585
119896= 1205910 (165)
From 120585119896 = 1205910and 0 lt |120585| lt 120587 it follows that 0 lt |120591
0| lt
120587119896 Insertion of 120585119896 = 1205910into (45) yields 119886
119896= 120582119896 Relations
119886119896= 120582119896and (82) imply that 1(1 minus 1198962) lt 119886
119896lt 1 Substitution
of 120582119896= 119886119896and 120585119896 = 120591
0into (66)ndash(68) leads to (156)ndash(158)
which proves that (156)ndash(158) hold for 1(1 minus 1198962
) lt 119886119896lt 1
and 0 lt |1205910| lt 120587119896
Finally substitution of 119886119896= 1(1 minus 119896
2
) and 1205910= 0 into
(161) leads to
119879119896(120591) =
[1 minus cos 120591](1 minus 1198962)
[119896 (119896 minus 1) minus 2
119896minus1
sum
119899=1
(119896 minus 119899) cos 119899120591]
(166)
Waveform (166) coincides with waveform (155) for 119886119896
=
1(1 minus 1198962
) which in turn proves that (156) holds for 119886119896=
1(1 minus 1198962
) and 1205910= 0 This completes the proof
62 Nonnegative CosineWaveforms with at Least One Zero for119896 = 3 In this subsection we consider nonnegative cosinewaveforms with at least one zero for 119896 = 3 (for case 119896 = 2
see [12])Cosine waveform with fundamental and third harmonic
reads
1198793(120591) = 1 + 119886
1cos 120591 + 119886
3cos 3120591 (167)
For 1198861le 0 and minus1 le 119886
3le 18 according to (155) non-
negative cosine waveform of type (167) with at least one zerocan be expressed as
1198793(120591) = (1 minus cos 120591) [1 minus 2119886
3(1 + 2 cos 120591 + cos 2120591)] (168)
From 1198793(120591 + 120587) = 2 minus 119879
3(120591) it immediately follows that for
1198861ge 0 and minus18 le 119886
3le 1 119879
3(120591) can be expressed as
1198793(120591) = (1 + cos 120591) [1 + 2119886
3(1 minus 2 cos 120591 + cos 2120591)] (169)
For 1198861le 0 and 18 le 119886
3le 1 from (158) it follows that 119886
3=
[8cos31205910]minus1 This relation along with (160) and (157) further
implies that 1198793(120591) can be expressed as
1198793(120591) =
[cos 1205910minus cos 120591]2 [2 cos 120591
0+ cos 120591]
2cos31205910
(170)
providing that |1205910| le 1205873 From 119879
3(120591 + 120587) = 2 minus 119879
3(120591) it
follows that (170) also holds for 1198861ge 0 and minus1 le 119886
3le minus18
providing that 1205910isin [21205873 41205873]
Maximally flat nonnegative cosinewaveformof type (167)with 119886
1lt 0 (minimum at 120591
0= 0) reads 119879
3(120591) = [1 minus
cos 120591]2[1 + (12) cos 120591] Dually maximally flat nonnegativecosine waveform with 119886
1gt 0 (minimum at 120591
0= 120587) reads
1198793(120591) = [1 + cos 120591]2[1 minus (12) cos 120591]In what follows we provide relations between coefficients
1198861and 1198863of nonnegative cosine waveforms of type (167) with
at least one zeroFor 1198861le 0 conversion of (168) into an additive form
immediately leads to the following relation
1198861= minus1 minus 119886
3for minus 1 le 119886
3le1
8 (171)
Conversion of (170) into an additive form leads to 1198861
=
minus31198863(2 cos 2120591
0+ 1) which can be also expressed as 119886
1=
minus31198863(4cos2120591
0minus 1) For 119886
1le 0 relations |120591
0| le 1205873 119886
1=
minus31198863(4cos2120591
0minus 1) and 119886
3= [8cos3120591
0]minus1 lead to
1198861= minus3 [ 3radic119886
3minus 1198863] for 1
8le 1198863le 1 (172)
Similarly for 1198861ge 0 conversion of (169) into an additive form
leads to the following relation
1198861= 1 minus 119886
3for minus
1
8le 1198863le 1 (173)
Mathematical Problems in Engineering 21
1
15
15
05
0
minus05
minus15minus15
minus1
minus1 minus05 0 05 1
Coefficient a1
Coe
ffici
enta
3 a1= minus
3(3radica3minus a3)
a1 =
minus1 minus a3
a1 =
1 minusa3
a1 =3(3radic|a3
| +a3)
Figure 14 Parameter space of cosine waveforms for 119896 = 3
For waveform of type (170) with 1198861ge 0 relations 120591
0isin [21205873
41205873] 1198861= minus3119886
3(4cos2120591
0minus 1) and 119886
3= [8cos3120591
0]minus1 lead to
1198861= 3 [
3radic10038161003816100381610038161198863
1003816100381610038161003816 + 1198863] for minus 1 le 1198863le minus
1
8 (174)
Every cosine waveform of type (167) corresponds to apair of real numbers (119886
1 1198863) and vice versa Points (119886
1 1198863)
in grey area in Figure 14 correspond to nonnegative cosinewaveforms for 119896 = 3 The points at the boundary of grey areacorrespond to nonnegative cosinewaveformswith at least onezero A number of shapes of nonnegative cosine waveformswith 119896 = 3 and at least one zero plotted on interval [minus120587 120587]are also presented in Figure 14 The boundary of grey area inFigure 14 consists of four line segments described by relations(171)ndash(174) The common point of line segments (172) and(173) is cusp point with coordinates 119886
1= 0 and 119886
3= 1
Another cusp point with coordinates 1198861= 0 and 119886
3= minus1
is the common point of line segments (171) and (174) Thecommon point of line segments (171)-(172) has coordinates(minus98 18) and common point of line segments (173)-(174)has coordinates (98 minus18) These points are representedby white circle dots and they correspond to maximallyflat cosine waveforms (eg see [21]) White triangle dotswith coordinates (2radic3 minusradic39) and (minus2radic3radic39) refer tothe nonnegative cosine waveforms with maximum value ofamplitude of fundamental harmonic
7 Four Case Studies of Usage of NonnegativeWaveforms in PA Efficiency Analysis
In this section we provide four case studies of usage ofdescription of nonnegative waveforms with fundamental and119896th harmonic in PA efficiency analysis In first two casestudies to be presented in Section 71 voltage is nonnegativewaveform with fundamental and second harmonic with atleast one zero In remaining two case studies to be consideredin Section 72 voltage waveform contains fundamental andthird harmonic
i
v Load
Vdc
Idc
Lch
vL
iL
Cb
+ +in
Figure 15 Generic PA circuit diagram
Let us consider generic PA circuit diagram as shown inFigure 15 We assume here that voltage and current wave-forms at the transistor output are
V (120579) = 1 + 1198861V cos 120579 + 1198871V sin 120579 + 119886119896V cos 119896120579 + 119887119896V sin 119896120579
119894 (120579) = 1 + 1198861119894cos 120579 +
infin
sum
119899=2
119886119899119894cos 119899120579
(175)
where 120579 stands for 120596119905 Both waveforms are normalized in thesense that dc components of voltage and current are 119881dc =
1 and 119868dc = 1 respectively Under assumption that blockingcapacitor 119862
119887behaves as short-circuit at the fundamental and
higher harmonics current and voltage waveforms at the loadare
V119871(120579) = 119886
1V cos 120579 + 1198871V sin 120579 + 119886119896V cos 119896120579 + 119887119896V sin 119896120579
119894119871(120579) = minus119886
1119894cos 120579 minus
infin
sum
119899=2
119886119899119894cos 119899120579
(176)
In terms of coefficients of voltage and current waveforms theload impedance at fundamental harmonic is 119911
1= minus(119886
1V minus
1198951198871V)1198861119894 whereas load impedance at 119896th harmonic is 119911
119896=
minus(119886119896Vminus119895119887119896V)119886119896119894 All other harmonics are short-circuited (119911
119899=
0 for 119899 = 1 and 119899 = 119896) Time average output power of PA (egsee [10]) with waveform pair (175) at fundamental frequencycan be expressed as
1198751= minus
11988611198941198861V
2 (177)
For normalized waveforms (175) with 119881dc = 1 and 119868dc = 1dc power is 119875dc = 1 Consequently PA efficiency 120578 = 119875
1119875dc
(eg see [10 26]) is equal to
120578 = minus11988611198941198861V
2 (178)
Thus time average output power 1198751of PA with pair of nor-
malized waveform (175) is equal to efficiency (178)Power utilization factor (PUF) is defined [26] as ldquothe
ratio of power delivered in a given situation to the power
22 Mathematical Problems in Engineering
delivered by the same device with the same supply voltagein Class A moderdquo Since the output power in class-A modeis 1198751class-A = max[V(120579)] sdotmax[119894(120579)]8 (eg see [9]) it follows
that power utilization factor PUF = 11987511198751class-A for PA with
pair of normalized waveforms (175) can be expressed as
PUF =8120578
max [V (120579)] sdotmax [119894 (120579)] (179)
71 NonnegativeWaveforms for 119896 = 2 in PAEfficiencyAnalysisIn this subsection we provide two case studies of usage ofdescription of nonnegative waveforms with fundamental andsecond harmonic (119896 = 2) in PA efficiency analysis For moreexamples of usage of descriptions of nonnegative waveformswith fundamental and second harmonic in PA efficiencyanalysis see [12]
Case Study 71 In this case study we consider efficiency of PAfor given second harmonic impedance providing that voltageis nonnegative waveform with fundamental and second har-monic and current is ldquohalf-sinerdquo waveform frequently used inefficiency analysis of classical PA operation (eg see [10])
Standard model of current waveform for classical PAoperation has the form (eg see [10 26])
119894119863(120579) =
119868119863[cos 120579 minus cos(120572
2)] |120579| le
120572
2
0120572
2le |120579| le 120587
(180)
where 120572 is conduction angle and 119868119863
gt 0 Since 119894119863(120579) is
even function it immediately follows that its Fourier seriescontains only dc component and cosine terms
119894119863(120579) = 119868dc +
infin
sum
119899=1
119868119899cos 119899120579 (181)
The dc component of the waveform (180) is
119868dc =119868119863120572
2120587[sinc(120572
2) minus cos(120572
2)] (182)
where sinc119909 = (sin119909)119909 The coefficient of the fundamentalharmonic component reads
1198681=119868119863120572
2120587(1 minus sinc120572) (183)
and the coefficient of 119899th harmonic component can bewrittenin the form
119868119899=119868119863
119899120587[sin ((119899 minus 1) 1205722)
(119899 minus 1)minussin ((119899 + 1) 1205722)
(119899 + 1)] 119899 ge 2
(184)
For ldquohalf-sinerdquo current waveform conduction angle is equalto 120587 (class-B conduction angle) According to (182) thisfurther implies that 119868dc = 119868
119863120587 To obtain normalized form of
waveform (180) we set 119868dc = 1which implies that 119868119863= 120587 Fur-
thermore substitution of 120572 = 120587 and 119868119863= 120587 in (180) leads to
119894 (120579) =
120587 cos 120579 |120579| lt120587
2
0120587
2lt |120579| le 120587
(185)
Similarly substitution of 119868119863= 120587 and 120572 = 120587 into (183) and
(184) leads to the coefficients of waveform (185) Coefficientsof fundamental and second harmonic respectively are
1198861119894=120587
2 119886
2119894=2
3 (186)
On the other hand voltage waveform of type (35) for 119896 =2 reads
V (120579) = 1 + 1198861V cos 120579 + 1198871V sin 120579 + 1198862V cos 2120579 + 1198872V sin 2120579
(187)
This waveform contains only fundamental and second har-monic and therefore all harmonics of order higher thantwo are short-circuited (119911
119899= 0 for 119899 gt 2) For current
voltage pair (185) and (187) load impedance at fundamentalharmonic is 119911
1= minus(119886
1V minus 1198951198871V)1198861119894 whereas load impedance
at second harmonic is 1199112= minus(119886
2Vminus1198951198872V)1198862119894 According to ourassumption the load is passive and therefore Re119911
1 gt 0 and
Re1199112 ge 0 which further imply 119886
11198941198861V lt 0 and 119886
21198941198862V le 0
respectivelyIt is easy to see that problem of findingmaximal efficiency
of PA with current-voltage pair (185) and (187) for prescribedsecond harmonic impedance can be reduced to the problemof finding voltage waveform of type (187) with maximal coef-ficient |119886
1V| for prescribed coefficients of second harmonic(see Section 5)
The following algorithm (analogous to Algorithm 22presented in [12]) provides the procedure for calculation ofmaximal efficiency with current-voltage pair (185) and (187)for prescribed second harmonic impedance The definitionof function atan 2(119910 119909) which appears in the step (iii) of thefollowing algorithm is given by (105)
Algorithm 32 (i) Choose 1199112= 1199032+1198951199092such that |119911
2| le 1|119886
2119894|
(ii) calculate 1198862V minus 1198951198872V = minus119911
21198862119894and 120582
2V = radic11988622V + 1198872
2V(iii) if 2120582
2V le 1 minus 1198862V then calculate 119886
1V = minus1 minus 1198862V and
1198871V = minus2119887
2V else calculate 1205821V = radic81205822V(1 minus 1205822V) 1205790V minus 120585V2 =
(12)atan2(1198872V 1198862V) 1198861V = minus120582
1V cos(1205790V minus 120585V2) and 1198871V =
minus1205821V sin(1205790V minus 120585V2)(iv) calculate efficiency 120578 = minus119886
11198941198861V2
(v) calculate 1199111= minus(119886
1V minus 1198951198871V)1198861119894 and 1199112119899 = 1199112Re119911
1
In this case study coefficients of fundamental and secondharmonic of current waveform are given by (186) Maximalefficiency of PA associated with the waveform pair (185)and (187) as a function of normalized second harmonicimpedance 119911
2119899= 1199112Re119911
1 is presented in Figure 16(a)
As can be seen from Figure 16(a) efficiency of 078 isachieved at the edge of Smith chart where second harmonicimpedance has small resistive part Corresponding PUFcalculated according to (179) is presented in Figure 16(b)Peak efficiency 120578 = 1205874 = 07854 and peak value of PUF = 1
are attained when second harmonic is short-circuited (whichcorresponds to ideal class-B operation [10 26])
For example for second harmonic impedance 1199112= 01 minus
11989505 and currentwaveform (185) fromAlgorithm32 it followsthat 2120582
2V le 1 minus 1198862V Furthermore according to step (iii)
of above algorithm maximal efficiency of PA is attained
Mathematical Problems in Engineering 23
+j2
+j5
+j1
+j2
+j5
minusj2
minusj5
minusj1
minusj2
minusj5
infin0
078
02
05
1 2 5
075 07 065
06 05
075
07
065
06
05
120578 lt 05
(a)
099
095
091
083 075
067
0 59 051
051
099
095
091
083
075 0
67 059
+j5
+j1
+j2
+j5
minusj5
minusj1
minusj2
minusj5
infin
+j2
minusj2
0 02
05
1 2 5
120578 lt 05
(b)
Figure 16 (a) Contours of maximal efficiency of PA and (b) contours of corresponding PUF associated with the waveform pair (185) and(187) as functions of normalized second harmonic impedance 119911
2119899= 1199112Re119911
1
2
3
4
1
0
2 3 410
Wav
efor
ms
Angle 120579120587
i(120579)
(120579)
Figure 17 Waveform pair (185) and (187) that provides maximalefficiency for 119911
2= 01 minus 11989505
with voltage waveform of type (187) with coefficients 1198862V =
minus00667 1198872V = minus03333 119886
1V = minus09333 and 1198871V = 06667 (see
Figure 17) Corresponding efficiency PUF and normalizedsecond harmonic impedance are 120578 = 07330 PUF = 07572and 1199112119899= 01683 minus 11989508415 respectively
On the other hand for second harmonic impedance 1199112=
01 minus 11989508 and current waveform (185) from Algorithm 32 itfollows that 2120582
2V gt 1 minus 1198862V Then according to step (iii) of
above algorithm maximal efficiency is attained with voltagewaveform of type (187) with coefficients 119886
2V = minus00667 1198872V =
minus05333 1198861V = minus09333 and 119887
1V = 10572 (see Figure 18)Efficiency PUF andnormalized secondharmonic impedanceare 120578 = 07330 PUF = 06332 and 119911
2119899= 01683 minus 11989513465
respectively
2
3
4
1
0
2 3 410
Wav
efor
ms
Angle 120579120587
i(120579)
(120579)
Figure 18 Waveform pair (185) and (187) that provides maximalefficiency for 119911
2= 01 minus 11989508
Case Study 72 As another case study let us consider the effi-ciency of PA providing that current waveform is nonnegativecosine waveform up to third harmonic with maximum valueof amplitude of fundamental harmonic [22] (see also [8])
119894 (120579) = 1 +1 + radic5
2cos 120579 + 2radic5
5cos 2120579 + 5 minus radic5
10cos 3120579
(188)
and voltage waveform is nonnegative waveform of type(187) Load impedances at fundamental second and thirdharmonic are 119911
1= minus(119886
1Vminus1198951198871V)1198861119894 1199112 = minus(1198862Vminus1198951198872V)1198862119894 and
1199113= 0 respectively According to our assumption the load
is passive and therefore Re1199111 gt 0 and Re119911
2 ge 0 which
further imply 11988611198941198861V lt 0 and 119886
21198941198862V le 0 respectively
Because current waveform (188) contains only cosineterms and voltage waveform is the same as in previous case
24 Mathematical Problems in Engineering
08
075
07
06
06
065
065
05
05
075 0
7
120578 lt 05
+j2
+j5
+j1
+j2
+j5
minusj2
minusj5
minusj1
minusj2
minusj5
infin0
02
05
1 2 5
(a)
085
08 0
75 07 0
6
05 04
08
075
0405
060
7
120578 lt 05
+j2
+j5
+j1
+j2
+j5
minusj2
minusj5
minusj1
minusj2
minusj5
infin0 02
05
1 2 5
(b)
Figure 19 (a) Contours of maximal efficiency of PA and (b) contours of corresponding PUF associated with the waveform pair (187)-(188)as functions of normalized second harmonic impedance 119911
2119899= 1199112Re119911
1
study the procedure for calculation of maximal efficiency ofPA with waveform pair (187)-(188) is the same as presentedin Algorithm 32 In this case study the coefficients of funda-mental and second harmonic of current waveform are 119886
1119894=
(1 + radic5)2 and 1198862119894= 2radic55 respectively
Maximal efficiency of PA associated with the waveformpair (187)-(188) as a function of normalized secondharmonicimpedance 119911
2119899= 1199112Re119911
1 is presented in Figure 19(a)
Efficiency of 08 is achieved at the edge of Smith chart wheresecond harmonic impedance has small resistive part Thetheoretical upper bound 120578 = (1 + radic5)4 asymp 08090 isattained when second harmonic is short-circuitedWhen thisupper bound is reached both second and third harmonic areshort-circuited which implies that we are dealing with finiteharmonic class-C [6 8] or dually when current and voltageinterchange their roles with finite harmonic inverse class-C[6 9] Corresponding PUF calculated according to (179) ispresented in Figure 19(b) Peak value of PUF asymp 08541 isattained when second harmonic is short-circuited
For example for second harmonic impedance 1199112
=
007 minus 11989504 and current waveform (188) from Algorithm 32 itfollows that 2120582
2V le 1 minus 1198862V Furthermore according to step
(iii) of Algorithm 32 maximal efficiency of PA is attainedwith voltage waveform of type (187) with coefficients 119886
2V =
minus00626 1198872V = minus03578 119886
1V = minus09374 and 1198871V = 07155 (see
Figure 20) Corresponding efficiency PUF and normalizedsecond harmonic impedance are 120578 = 07584 PUF = 06337and 1199112119899= 01208 minus 11989506904 respectively
On the other hand for 1199112= 005 minus 11989507 and current
waveform (187) it follows that 21205822V gt 1minus119886
2VThen accordingto step (iii) of Algorithm 32 themaximal efficiency is attainedwith voltage waveform of type (187) with coefficients 119886
2V =
minus00447 1198872V = minus06261 119886
1V = minus09318 and 1198871V = 10007 (see
Figure 21) Efficiency PUF and normalized second harmonic
2
3
4
1
0
2 3 410
Wav
efor
ms
Angle 120579120587
i(120579)
(120579)
Figure 20 Waveform pair (187)-(188) that provides maximal effi-ciency for 119911
2= 007 minus 11989504
impedance are 120578 = 07538 PUF = 05314 and 1199112119899= 00868minus
11989512156 respectively
72 Nonnegative Waveforms for 119896 = 3 in PA EfficiencyAnalysis In this subsection we provide another two casestudies of usage of description of nonnegative waveforms inPA efficiency analysis this time with fundamental and thirdharmonic (119896 = 3)
Case Study 73 Let us consider current-voltage pair such thatvoltage is nonnegative waveform with fundamental and thirdharmonic
V (120579) = 1 + 1198861V cos 120579 + 1198871V sin 120579 + 1198863V cos 3120579 + 1198873V sin 3120579
(189)
Mathematical Problems in Engineering 25
2
3
4
1
0
2 3 410
Wav
efor
ms
Angle 120579120587
i(120579)
(120579)
Figure 21 Waveform pair (187)-(188) that provides maximal effi-ciency for 119911
2= 005 minus 11989507
and current is nonnegative cosine waveform given by (188)Load impedances at fundamental second and third har-monic are 119911
1= minus(119886
1V minus 1198951198871V)1198861119894 1199112 = 0 and 119911
3= minus(119886
3V minus
1198951198873V)1198863119894 respectively According to our assumption the load
is passive and therefore Re1199111 gt 0 and Re119911
3 ge 0 which
further imply 11988611198941198861V lt 0 and 119886
31198941198863V le 0
In this subsection we consider the problem of findingmaximal efficiency of PA with waveform pair (188)-(189)for given third harmonic impedance As we mentionedearlier problem of finding maximal efficiency of PA withcurrent-voltage pair (188)-(189) for prescribed third har-monic impedance can be reduced to the problem of findingvoltage waveform of type (189) withmaximal coefficient |119886
1V|for prescribed coefficients of third harmonic (see Section 52)
The following algorithm provides the procedure forcalculation of maximal efficiency with current-voltage pair(188)-(189) The definition of function atan 2(119910 119909) whichappears in step (iii) of the following algorithm is given by(105)
Algorithm 33 (i) Choose 1199113= 1199033+1198951199093such that |119911
3| le 1|119886
3119894|
(ii) calculate 1198863V minus 1198951198873V = minus119911
31198863119894and 120582
3V = radic11988623V + 1198872
3V(iii) if 271205822
3V le (1 minus 21198863V)3 then calculate 119886
1V = minus1 minus 1198863V
and 1198871V = minus3119887
3V else calculate 1205821V = 3(3radic1205823V minus 120582
3V) 1205790V minus120585V3 = (13)atan 2(119887
3V 1198863V) 1198861V = minus1205821V cos(1205790V minus 120585V3) and
1198871V = minus120582
1V sin(1205790V minus 120585V3)(iv) calculate efficiency 120578 = minus119886
11198941198861V2
(v) calculate 1199111= minus(119886
1V minus 1198951198871V)1198861119894 and 1199113119899 = 1199113Re119911
1
In this case study coefficients of fundamental and thirdharmonic of current waveform are 119886
1119894= (1 + radic5)2 and
1198863119894= (5 minus radic5)10 respectively For the waveform pair (188)-
(189) maximal efficiency of PA as a function of normalizedthird harmonic impedance 119911
3119899= 1199113Re119911
1 is presented in
Figure 22 Efficiency of 08 is reached when third harmonicimpedance has small resistive part Peak efficiency 120578 = (1 +
radic5)4 asymp 08090 is achieved when third harmonic is short-circuited
+j2
+j5
+j1
+j2
+j5
minusj2
minusj5
minusj1
minusj2
minusj5
infin0
08
08
075
075
07
07
06
05
02
05
1 2
Figure 22 Contours of maximal efficiency of PA associated withthe waveform pair (188)-(189) as a function of normalized thirdharmonic impedance 119911
3119899= 1199113Re119911
1
For the present case study in what follows we show thatpower utilization factor is proportional to efficiency For volt-age waveform of type (189) it is easy to see that V(120579 + 120587) =
2minusV(120579) holdsThis relation along with the fact that waveformV(120579) that provides maximal efficiency has at least one zeroimplies that max[V(120579)] = 2 On the other hand currentwaveform (188) is cosine waveform with positive coefficientsand therefore max[119894(120579)] = 119894(0) = 2 + 4radic5 Consequentlyaccording to (179) the following relation holds
PUFCase study 73 = 2 (5 minus 2radic5) 120578 = 10557120578 (190)
Clearly the ratio PUF120578 is constant and therefore in this casestudy PUF can be easily calculated from the correspondingefficiency Accordingly peak efficiency and peak value ofPUFCase study 73 = 3radic52 minus 52 = 08541 are attained forthe same voltage waveform (when third harmonic is short-circuited)
In the first example current waveform (188) and 1199113=
02 minus 11989505 imply that 2712058223V le (1 minus 2119886
3V)3 Then according
to Algorithm 33 the voltage waveform of type (189) thatprovides maximal efficiency has the following coefficients1198863V = minus00553 119887
3V = minus01382 1198861V = minus09447 and 119887
1V =
04146 (see Figure 23) Efficiency PUF and normalized thirdharmonic impedance are 120578 = 07643 PUF = 08069 and1199113119899= 03425 minus 11989508564 respectivelyIn the second example current waveform (188) and 119911
3=
01 minus 11989511 imply that 2712058223V gt (1 minus 2119886
3V)3 Then according
to Algorithm 33 the voltage waveform of type (189) thatprovides maximal efficiency has the following coefficients1198863V = minus00276 119887
3V = minus03040 1198861V = minus09391 and 119887
1V =
05807 (see Figure 24) Efficiency PUF and normalized thirdharmonic impedance are 120578 = 07598 PUF = 08021 and1199113119899= 01723 minus 11989518952 respectively
26 Mathematical Problems in Engineering
2
3
4
1
0
2 3 410
Wav
efor
ms
Angle 120579120587
i(120579)
(120579)
Figure 23 Waveform pair (188)-(189) that provides maximal effi-ciency for 119911
3= 02 minus 11989505
2
3
4
1
0
2 3 410
Wav
efor
ms
Angle 120579120587
i(120579)
(120579)
Figure 24 Waveform pair (188)-(189) that provides maximal effi-ciency for 119911
3= 01 minus 11989511
Case Study 74 In this case study let us consider current-voltage pair where current is normalized waveform of type(180) with conduction angle 120572 = 115120587 (207∘) and voltageis nonnegative waveform of type (189) Substitution of 120572 =
115120587 and 119868dc = 1 into (182) leads to 119868119863
= 22535Furthermore substitution of 120572 = 115120587 and 119868
119863= 22535 into
(180) leads to
119894 (120579)
=
22535 [cos 120579 minus cos(1151205872
)] |120579| le115120587
2
0115120587
2le |120579| le 120587
(191)
Similarly substitution of 120572 = 115120587 and 119868119863
= 22535 into(183) and (184) for 119899 = 3 yields coefficients of fundamentaland third harmonic of waveform (191)
1198861119894= 14586 119886
3119894= minus01026 (192)
Because current waveform (191) contains only cosineterms and voltage waveform is the same as in previous case
+j2
+j5
+j1
+j2
+j5
minusj2
minusj5
minusj1
minusj2
minusj5
infin0 02
05
1 2
074
076
08
084
082
078
076
074
Figure 25 Contours of maximal efficiency of PA associated withthe waveform pair (189) and (191) as a function of normalized thirdharmonic impedance 119911
3119899= 1199113Re119911
1
study the procedure for calculation of maximal efficiencyof PA with waveform pair (189)ndash(191) is the same as thatpresented in Algorithm 33 In this case study the coefficientsof fundamental and third harmonic of current waveform aregiven by (192)
For the waveform pair (189) and (191) maximal efficiencyof PA as a function of normalized third harmonic impedance1199113119899= 1199113Re119911
1 is presented in Figure 25 Efficiency of 084
is obtained in vicinity of 1199113119899= 23685 (corresponding to 119911
3=
18750) Peak efficiency 120578 asymp 08421 is achieved for voltagewaveform of type (189) with coefficients 119886
1V = minus2radic3 1198863V =
radic39 and 1198871V = 1198873V = 0
In the course of finding power utilization factor noticethat currentwaveformof type (191) attains itsmaximumvaluefor 120579 = 0 Insertion of max[119894(120579)] = 119894(0) = 278 andmax[V(120579)] = 2 for voltage waveform of type (189) into (179)leads to
PUFCase study 74 = 1439120578 (193)
Again the ratio PUF120578 is constant and PUF can be easilycalculated from the corresponding efficiency Accordinglypeak value of PUFCase study 74 asymp 12118 and peak efficiencyare attained for the same voltage waveform
In the first example current waveform (191) and 1199113=
1 minus 11989502 imply that 2712058223V le (1 minus 2119886
3V)3 Then according to
Algorithm 33 voltage waveform of type (189) which providesmaximal efficiency has coefficients 119886
3V = 01026 1198873V =
00205 1198861V = minus11026 and 119887
1V = minus00616 (see Figure 26)Efficiency PUF and normalized third harmonic impedanceare 120578 = 08042 PUF = 11572 and 119911
3119899= 13228 minus 11989502646
respectivelyIn second example current waveform (191) and 119911
3=
15 minus 11989512 imply that 2712058223V gt (1 minus 2119886
3V)3 Then according to
Algorithm 33 voltage waveform of type (189) which providesmaximal efficiency has coefficients 119886
3V = 01540 1198873V =
Mathematical Problems in Engineering 27
2
3
4
1
0
2 3 410
Wav
efor
ms
Angle 120579120587
i(120579)
(120579)
Figure 26 Waveform pair (189) and (191) that provides maximalefficiency for 119911
3= 1 minus 11989502
2
3
4
1
0
2 3 410
Wav
efor
ms
Angle 120579120587
i(120579)
(120579)
Figure 27 Waveform pair (189) and (191) that provides maximalefficiency for 119911
3= 15 minus 11989512
01232 1198861V = minus11255 and 119887
1V = minus02575 (see Figure 27)Efficiency PUF and normalized third harmonic impedanceare 120578 = 08208 PUF = 11812 and 119911
3119899= 19439 minus 11989515552
respectively
8 Conclusion
In this paper we consider a problem of finding general de-scriptions of various classes of nonnegative waveforms withfundamental and 119896th harmonic These classes include non-negative waveforms with at least one zero nonnegative wave-forms with maximal amplitude of fundamental harmonic forprescribed amplitude of 119896th harmonic nonnegative wave-forms withmaximal coefficient of cosine part of fundamentalharmonic for prescribed coefficients of 119896th harmonic andnonnegative cosine waveforms with at least one zero Mainresults are stated in six propositions (Propositions 1 6 918 22 and 26) four corollaries (Corollaries 2ndash5) twentyremarks and three algorithms Four case studies of usageof closed form descriptions of nonnegative waveforms in PAefficiency analysis are considered in detail in Section 7
Appendices
Here we provide a list of finite sums of trigonometric func-tions used in this paper (Appendix A) and brief account ofthe Chebyshev polynomials (Appendix B)
A List of Some Finite Sums ofTrigonometric Functions
Dirichlet kernel (eg see [27]) is as follows
119863119896minus1
(120591) = 1 + 2
119896minus1
sum
119899=1
cos 119899120591 = sin ((2119896 minus 1) 1205912)sin (1205912)
(A1)
Fejer kernel (eg see [27]) can be expressed in the fol-lowing equivalent forms
119865119896minus1
(120591) =1
119896
119896minus1
sum
119899=0
119863119899(120591) = 1 +
2
119896
119896minus1
sum
119899=1
(119896 minus 119899) cos 119899120591
=(1 minus cos 119896120591)119896 (1 minus cos 120591)
(A2)
Lagrangersquos trigonometric identity (eg see [28]) is as fol-lows
1198781(120591) =
119896minus1
sum
119899=1
sin 119899120591 = sin (1198961205912) sin ((119896 minus 1) 1205912)sin (1205912)
(A3)
In what follows we show that the following three trigono-metric identities also hold
2
119896minus1
sum
119899=1
(119896 minus 119899) sin 119899120591 = 119896 sin 120591 minus sin 1198961205911 minus cos 120591
(A4)
119896minus1
sum
119899=1
cos (119896 minus 2119899) 120591 = sin (119896 minus 1) 120591sin 120591
(A5)
119896minus1
sum
119899=1
119899 (119896 minus 119899) cos (119896 minus 2119899) 120591
=sin (119896120591) cos 120591 minus 119896 cos (119896120591) sin 120591
2sin3120591
(A6)
Denote 1198782(120591) = 2sum
119896minus1
119899=1(119896 minus 119899) sin 119899120591 119878
3(120591) = sum
119896minus1
119899=1cos(119896 minus
2119899)120591 and 1198784(120591) = sum
119896minus1
119899=1119899(119896 minus 119899) cos(119896 minus 2119899)120591
Notice that 1198782(120591) = 2119896119878
1(120591) + 119889119863
119896minus1(120591)119889120591 which
immediately leads to (A4)
28 Mathematical Problems in Engineering
Identity (A5) can be obtained as follows
sin (119896 minus 1) 120591sin 120591
=119890119895(119896minus1)120591
minus 119890minus119895(119896minus1)120591
119890119895120591 minus 119890minus119895120591
= 119890119895119896120591
119890minus2119895120591
minus 119890minus2119895(119896minus1)120591
1 minus 119890minus2119895120591
= 119890119895119896120591
119896minus1
sum
119899=1
119890minus2119895119899120591
=
119896minus1
sum
119899=1
119890119895(119896minus2119899)120591
=
119896minus1
sum
119899=1
cos (119896 minus 2119899) 120591
(A7)
From 4119899(119896 minus 119899) = 1198962
minus (119896 minus 2119899)2 it follows that 4119878
4(120591) =
1198962
1198783(120591) + 119889
2
1198783(120591)119889120591
2 which leads to (A6)
B The Chebyshev Polynomials
The Chebyshev polynomials of the first kind 119881119899(119909) can be
defined by the following relation (eg see [29])
119881119899(119909) = cos 119899120591 when 119909 = cos 120591 (B1)
The Chebyshev polynomials of the second kind 119880119899(119909) can be
defined by the following relation (eg see [29])
119880119899(119909) =
sin (119899 + 1) 120591sin 120591
when 119909 = cos 120591 (B2)
The Chebyshev polynomials satisfy the following recur-rence relations (eg see [29])
1198810(119909) = 1 119881
1(119909) = 119909
119881119899+1
(119909) = 2119909119881119899(119909) minus 119881
119899minus1(119909)
1198800(119909) = 1 119880
1(119909) = 2119909
119880119899+1
(119909) = 2119909119880119899(119909) minus 119880
119899minus1(119909)
(B3)
The first few Chebyshev polynomials of the first and secondkind are 119881
2(119909) = 2119909
2
minus 1 1198813(119909) = 4119909
3
minus 3119909 1198814(119909) = 8119909
4
minus
81199092
+ 1 1198802(119909) = 4119909
2
minus 1 1198803(119909) = 8119909
3
minus 4119909 and 1198804(119909) =
161199094
minus 121199092
+ 1
Conflict of Interests
The authors declare that there is no conflict of interestsregarding the publication of this paper
Acknowledgment
This work is supported by the SerbianMinistry of EducationScience and Technology Development as a part of ProjectTP32016
References
[1] V I Arnolrsquod V S Afrajmovich Y S Ilrsquoyashenko and L PShilrsquonikov Dynamical Systems V Bifurcation Theory and Catas-trophe Theory Springer Berlin Germany 1994
[2] E Polak ldquoOn the mathematical foundations of nondifferen-tiable optimization in engineering designrdquo SIAM Review vol29 no 1 pp 21ndash89 1987
[3] N S Fuzik ldquoBiharmonic modes of a tuned RF power amplifierrdquoRadiotehnika vol 25 no 7 pp 62ndash71 1970 (Russian)
[4] P Colantonio F Giannini G Leuzzi and E Limiti ldquoClass Gapproach for low-voltage high-efficiency PA designrdquo Interna-tional Journal of RF and Microwave Computer-Aided Engineer-ing vol 10 no 6 pp 366ndash378 2000
[5] F H Raab ldquoMaximum efficiency and output of class-F poweramplifiersrdquo IEEE Transactions on Microwave Theory and Tech-niques vol 49 no 6 pp 1162ndash1166 2001
[6] FH Raab ldquoClass-E class-C and class-F power amplifiers basedupon a finite number of harmonicsrdquo IEEE Transactions onMicrowaveTheory and Techniques vol 49 no 8 pp 1462ndash14682001
[7] J D Rhodes ldquoOutput universality inmaximum efficiency linearpower amplifiersrdquo International Journal of Circuit Theory andApplications vol 31 no 4 pp 385ndash405 2003
[8] A Juhas and L A Novak ldquoComments on lsquoClass-E class-Cand classF power amplifier based upon a finite number ofharmonicsrsquordquo IEEE Transactions on Microwave Theory and Tech-niques vol 57 no 6 pp 1623ndash1625 2009
[9] M Roberg and Z Popovic ldquoAnalysis of high-efficiency poweramplifiers with arbitrary output harmonic terminationsrdquo IEEETransactions on Microwave Theory and Techniques vol 59 no8 pp 2037ndash2048 2011
[10] A Grebennikov N O Sokal and M J Franco Switchmode RFPower Amplifiers ElsevierAcademic Press San Diego CalifUSA 2nd edition 2012
[11] T Canning P J Tasker and S C Cripps ldquoContinuous modepower amplifier design using harmonic clipping contourstheory and practicerdquo IEEE Transactions on Microwave Theoryand Techniques vol 62 no 1 pp 100ndash110 2014
[12] A Juhas and L A Novak ldquoGeneral description of nonnegativewaveforms up to second harmonic for power amplifier mod-ellingrdquoMathematical Problems in Engineering vol 2014 ArticleID 709762 18 pages 2014
[13] V I Arnolrsquod V V Goryunov O V Lyashko and V A VasilrsquoevDynamical Systems VIIImdashSingularity Theory II ApplicationsSpringer Berlin Germany 1993
[14] D Siersma ldquoProperties of conflict sets in the planerdquo BanachCenter Publications Polish Academy of Sciences vol 50 no 1 pp267ndash276 1999 Proceedings of the Banach Center Symposiumon Geometry and Topology of Caustics (Caustics rsquo98) WarsawPoland
[15] M van Manen The geometry of conflict sets [Dissertation]Universiteit Utrecht UtrechtThe Netherlands 2003 httpigi-tur-archivelibraryuunldissertations2003-0912-123058c4pdf
[16] Y L Sachkov ldquoMaxwell strata and symmetries in the problemofoptimal rolling of a sphere over a planerdquo Sbornik Mathematicsvol 201 no 7-8 pp 1029ndash1051 2010
[17] I A Bogaevsky ldquoPerestroikas of shocks and singularities ofminimum functionsrdquoPhysicaDNonlinear Phenomena vol 173no 1-2 pp 1ndash28 2002
[18] Y L Sachkov ldquoMaxwell strata in the Euler elastic problemrdquoJournal of Dynamical andControl Systems vol 14 no 2 pp 169ndash234 2008
[19] M Siino and T Koike ldquoTopological classification of black holesgeneric Maxwell set and crease set of a horizonrdquo InternationalJournal ofModern Physics D Gravitation Astrophysics Cosmol-ogy vol 20 no 6 pp 1095ndash1122 2011
Mathematical Problems in Engineering 29
[20] F H Raab ldquoClass-F power amplifiers with maximally flat wave-formsrdquo IEEETransactions onMicrowaveTheory and Techniquesvol 45 no 11 pp 2007ndash2012 1997
[21] A Juhas and L A Novak ldquoMaximally flat waveforms with finitenumber of harmonics in class-F power amplifiersrdquo Mathemat-ical Problems in Engineering vol 2013 Article ID 169590 9pages 2013
[22] L Fejer ldquoUber trigonometrische polynomerdquo Journal fur dieReine und Angewandte Mathematik vol 1916 no 146 pp 53ndash82 1916 (German)
[23] S C Cripps ldquoBessel Waives [microwave bytes]rdquo IEEE Micro-wave Magazine vol 10 no 7 pp 30ndash36 117 2009
[24] L N Bryzgalova ldquoSingularities of the maximum of para-metrically dependent functionrdquo Functional Analysis and ItsApplications vol 11 no 1 pp 49ndash51 1977
[25] V I Arnold A A Davydov V A Vassiliev and V MZakalyukin Mathematical Models of Catastrophes Control ofCatastrophic Process Encyclopedia of Life Support Systems(EOLSS) EOLSS Publishers Oxford UK 2006
[26] S C Cripps RF Power Amplifiers for Wireless CommunicationsArtech House Norwood Mass USA 2nd edition 2006
[27] A Zygmund Trigonometric Series vol 1 Cambridge UniversityPress Cambridge UK 2nd edition 1959
[28] A Jeffrey and H DaiHandbook of Mathematical Formulas andIntegrals ElsevierAcademic Press San Diego Calif USA 4thedition 2008
[29] J C Mason and D C Handscomb Chebyshev PolynomialsChapman amp Hall CRC Press Boca Raton Fla USA 2003
Submit your manuscripts athttpwwwhindawicom
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical Problems in Engineering
Hindawi Publishing Corporationhttpwwwhindawicom
Differential EquationsInternational Journal of
Volume 2014
Applied MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Probability and StatisticsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical PhysicsAdvances in
Complex AnalysisJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
OptimizationJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
CombinatoricsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Operations ResearchAdvances in
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Function Spaces
Abstract and Applied AnalysisHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of Mathematics and Mathematical Sciences
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Algebra
Discrete Dynamics in Nature and Society
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Decision SciencesAdvances in
Discrete MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom
Volume 2014 Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Stochastic AnalysisInternational Journal of
8 Mathematical Problems in Engineering
For |120585| lt 120587 relation sin 120585 sin(120585119896) gt 0 obviously holds Fromcos 119905 gt cos 1199051015840 for |119905| le 120587119896 lt |119905
1015840
| le 120587 it follows that positionof global maximum of the function of type [119888 cos 119905 minus cos(119896119905)]for 119888 gt 0 belongs to interval |119905| le 120587119896 Therefore position ofglobal maximum of the expression in the square brackets in(60) for |120585| lt 120587 belongs to interval |120591 minus 120591
0+ 120585119896| le 120587119896 This
inequality together with |120585| lt 120587 leads to |120591minus1205910| lt 2120587119896 Since
[1 minus cos(120591 minus 1205910)]minus1 decreases with increasing |120591 minus 120591
0| le 120587 it
follows that 119902119896(120591) for |120585| lt 120587 has global maximum on interval
|120591minus1205910| lt 2120587119896 For |120585| = 120587 it is easy to show thatmax
120591119902119896(120591) =
119902119896(1205910plusmn 2120587119896) = 0 Since 119903
119896(120591) minus 119902
119896(120591) is constant (see (58))
it follows from previous considerations that 119903119896(120591) has global
maximum on interval |120591 minus 1205910| le 2120587119896
To find max120591119903119896(120591) let us consider first derivative of 119903
119896(120591)
with respect to 120591 Starting from (56) first derivative of 119903119896(120591)
can be expressed in the following form
119889119903119896(120591)
119889120591= minus119904 (120591) sdot sin(
119896 (120591 minus 1205910)
2+ 120585) (61)
where
119904 (120591) = [sin(119896 (120591 minus 120591
0)
2) cos(
120591 minus 1205910
2)
minus 119896 cos(119896 (120591 minus 120591
0)
2) sin(
120591 minus 1205910
2)]
sdot sinminus3 (120591 minus 1205910
2)
(62)
Using (A6) (see Appendices) (62) can be rewritten as
119904 (120591) = 2
119896minus1
sum
119899=1
119899 (119896 minus 119899) cos((119896 minus 2119899) (120591 minus 120591
0)
2) (63)
From 119899(119896 minus 119899) gt 0 and |119896 minus 2119899| lt 119896 119899 = 1 (119896 minus 1) itfollows that all summands in (63) decrease with increasing|120591 minus 1205910| providing that |120591 minus 120591
0| le 2120587119896 Therefore 119904(120591) ge 119904(120591
0plusmn
2120587119896) = 119896sin2(120587119896) gt 0 for |120591 minus 1205910| le 2120587119896 Consequently
119889119903119896(120591)119889120591 = 0 and |120591minus120591
0| le 2120587119896 imply that sin(119896(120591minus120591
0)2+
120585) = 0From |120585| le 120587 |120591minus120591
0| le 2120587119896 and sin(119896(120591minus120591
0)2+120585) = 0
it follows that 120591minus1205910+120585119896 = minus120585119896 or |120591minus120591
0+120585119896| = (2120587minus|120585|)119896
and therefore cos(119896(120591 minus 1205910+ 120585119896)) = cos 120585 Since cos(120585119896) ge
cos(2120587 minus |120585|)119896 it follows that max120591119902119896(120591) is attained for 120591 =
1205910minus2120585119896 Furthermore from (60) it follows that max
120591119902119896(120591) =
119902119896(1205910minus 2120585119896) = 0 which together with (58)-(59) leads to
max120591
119903119896(120591) = 119903
119896(1205910minus2120585
119896)
= (119896 minus 1) cos 120585 + 119896 sin (120585 minus 120585119896)sin (120585119896)
(64)
Both terms on the right hand side of (64) are even functionsof 120585 and decrease with increase of |120585| |120585| le 120587 Thereforemax120591119903119896(120591) attains its lowest value for |120585| = 120587 It is easy to
show that right hand side of (64) for |120585| = 120587 is equal to 1which further implies that
max120591
119903119896(120591) ge 1 (65)
From (65) it follows that (57) can be rewritten as 120582119896
le
[max120591119903119896(120591)]minus1 Finally substitution of (64) into 120582
119896le
[max120591119903119896(120591)]minus1 leads to (40) which completes the proof
32 Nonnegative Waveforms with Two Zeros Nonnegativewaveforms of type (35) with two zeros always possess twoglobal minima Such nonnegative waveforms are thereforerelated to the conflict set
In this subsection we provide general description of non-negative waveforms of type (35) for 119896 ge 2 and exactly twozeros According to Remark 7 120582
119896= 1 implies |120585| = 120587 and
119879119896(120591) = 1 minus cos 119896(120591 minus 120591
0) Number of zeros of 119879
119896(120591) = 1 minus
cos 119896(120591minus1205910) on fundamental period equals 119896 which is greater
than two for 119896 gt 2 and equal to two for 119896 = 2 In the followingproposition we exclude all waveforms with 120582
119896= 1 (the case
when 119896 = 2 and 1205822= 1 is going to be discussed in Remark 10)
Proposition 9 Every nonnegative waveform of type (35) withexactly two zeros can be expressed in the following form
119879119896(120591) = 120582
119896[1 minus cos (120591 minus 120591
0)] [1 minus cos(120591 minus 120591
0+2120585
119896)]
sdot [1198880+ 2
119896minus2
sum
119899=1
119888119899cos 119899(120591 minus 120591
0+120585
119896)]
(66)
where
119888119899= [sin(120585 minus 119899120585
119896) cos(120585
119896)
minus (119896 minus 119899) cos(120585 minus 119899120585
119896) sin(120585
119896)]
sdot sinminus3 (120585119896)
(67)
120582119896= [(119896 minus 1) cos 120585 + 119896 sin(120585 minus 120585119896)
sin(120585119896)]
minus1
(68)
0 lt10038161003816100381610038161205851003816100381610038161003816 lt 120587 (69)
Remark 10 For 119896 = 2 waveforms with 1205822= 1 also have
exactly two zeros These waveforms can be included in aboveproposition by substituting (69) with 0 lt |120585| le 120587
Remark 11 Apart from nonnegative waveforms of type (35)with two zeros there are another two types of nonnegativewaveforms which can be obtained from (66)ndash(68) These are
(i) nonnegative waveforms with 119896 zeros (correspondingto |120585| = 120587) and
(ii) maximally flat nonnegative waveforms (correspond-ing to 120585 = 0)
Notice that nonnegative waveforms of type (35) with120582119896= 1 can be obtained from (66)ndash(68) by setting |120585| =
120587 Substitution of 120582119896
= 1 and |120585| = 120587 into (66) alongwith execution of all multiplications and usage of (A2) (seeAppendices) leads to 119879
119896(120591) = 1 minus cos 119896(120591 minus 120591
0)
Mathematical Problems in Engineering 9
Also maximally flat nonnegative waveforms (they haveonly one zero [21]) can be obtained from (66)ndash(68) by setting120585 = 0 Thus substitution of 120585 = 0 into (66)ndash(68) leads tothe following form of maximally flat nonnegative waveformof type (35)
119879119896(120591) =
[1 minus cos (120591 minus 1205910)]2
3 (1198962 minus 1)
sdot [119896 (1198962
minus 1)
+ 2
119896minus2
sum
119899=1
(119896 minus 119899) ((119896 minus 119899)2
minus 1) cos 119899 (120591 minus 1205910)]
(70)
Maximally flat nonnegative waveforms of type (35) for 119896 le 4
can be expressed as
1198792(120591) =
2
3[1 minus cos(120591 minus 120591
0)]2
1198793(120591) =
1
2[1 minus cos(120591 minus 120591
0)]2
[2 + cos (120591 minus 1205910)]
1198794(120591) =
4
15[1 minus cos (120591 minus 120591
0)]2
sdot [5 + 4 cos (120591 minus 1205910) + cos 2 (120591 minus 120591
0)]
(71)
Remark 12 Every nonnegative waveform of type (35) withexactly one zero at nondegenerate critical point can bedescribed as in Proposition 6 providing that symbol ldquolerdquoin relation (40) is replaced with ldquoltrdquo This is an immediateconsequence of Propositions 6 and 9 and Remark 11
Remark 13 Identity [1minus cos(120591minus1205910)][1minus cos(120591minus120591
0+2120585119896)] =
[cos 120585119896 minus cos(120591 minus 1205910+ 120585119896)]
2 implies that (66) can be alsorewritten as
119879119896(120591) = 120582
119896[cos 120585119896
minus cos(120591 minus 1205910+120585
119896)]
2
sdot [1198880+ 2
119896minus2
sum
119899=1
119888119899cos 119899(120591 minus 120591
0+120585
119896)]
(72)
Furthermore substitution of (67) into (72) leads to
119879119896(120591)
= 120582119896[cos 120585119896
minus cos(120591 minus 1205910+120585
119896)]
sdot [(119896 minus 1) sin 120585sin (120585119896)
minus 2
119896minus1
sum
119899=1
sin (120585 minus 119899120585119896)sin (120585119896)
cos 119899(120591 minus 1205910+120585
119896)]
(73)
Remark 14 According to (A6) (see Appendices) it followsthat coefficients (67) can be expressed as
119888119899= 2
119896minus119899minus1
sum
119898=1
119898(119896 minus 119899 minus 119898) cos((119896 minus 119899 minus 2119898) 120585119896
) (74)
Furthermore from (74) it follows that coefficients 119888119896minus2
119888119896minus3
119888119896minus4
and 119888119896minus5
are equal to119888119896minus2
= 2 (75)
119888119896minus3
= 8 cos(120585119896) (76)
119888119896minus4
= 8 + 12 cos(2120585119896) (77)
119888119896minus5
= 24 cos(120585119896) + 16 cos(3120585
119896) (78)
For example for 119896 = 2 (75) and (68) lead to 1198880= 2 and
1205822= 1(2+cos 120585) respectively which from (72) further imply
that
1198792(120591) =
2 [cos(1205852) minus cos(120591 minus 1205910+ 1205852)]
2
[2 + cos 120585] (79)
Also for 119896 = 3 (75) (76) and (68) lead to 1198881= 2 119888
0=
8 cos(1205853) and 1205823= [2(3 cos(1205853) + cos 120585)]minus1 respectively
which from (72) further imply that
1198793(120591) =
2 [cos (1205853) minus cos (120591 minus 1205910+ 1205853)]
2
[3 cos (1205853) + cos 120585]
sdot [2 cos(1205853) + cos(120591 minus 120591
0+120585
3)]
(80)
Remark 15 According to (A5) (see Appendices) relation(68) can be rewritten as
120582119896= [(119896 minus 1) cos 120585 + 119896
119896minus1
sum
119899=1
cos((119896 minus 2119899)120585119896
)]
minus1
(81)
Clearly amplitude 120582119896of 119896th harmonic of nonnegative wave-
form of type (35) with exactly two zeros is even functionof 120585 Since cos((119896 minus 2119899)120585119896) 119899 = 0 (119896 minus 1) decreaseswith increase of |120585| on interval 0 le |120585| le 120587 it follows that120582119896monotonically increases with increase of |120585| Right hand
side of (68) is equal to 1(1198962
minus 1) for 120585 = 0 and to one for|120585| = 120587 Therefore for nonnegative waveforms of type (35)with exactly two zeros the following relation holds
1
1198962 minus 1lt 120582119896lt 1 (82)
The left boundary in (82) corresponds to maximally flatnonnegative waveforms (see Remark 11) The right boundaryin (82) corresponds to nonnegative waveforms with 119896 zeros(also see Remark 11)
Amplitude of 119896th harmonic of nonnegative waveform oftype (35) with two zeros as a function of parameter 120585 for 119896 le5 is presented in Figure 5
Remark 16 Nonnegative waveform of type (35) with twozeros can be also expressed in the following form
119879119896(120591) = 1 minus 120582
119896
119896 sin 120585sin (120585119896)
cos(120591 minus 1205910+120585
119896)
+ 120582119896cos (119896 (120591 minus 120591
0) + 120585)
(83)
10 Mathematical Problems in Engineering
1
08
06
04
02
0minus1 minus05 0 05
Am
plitu
de120582k
1
Parameter 120585120587
k = 2k = 3
k = 4k = 5
Figure 5 Amplitude of 119896th harmonic of nonnegative waveformwith two zeros as a function of parameter 120585
where 120582119896is given by (68) and 0 lt |120585| lt 120587 From (83) it follows
that coefficients of fundamental harmonic of nonnegativewaveform of type (35) with two zeros are
1198861= minus1205821cos(120591
0minus120585
119896) 119887
1= minus1205821sin(120591
0minus120585
119896) (84)
where 1205821is amplitude of fundamental harmonic
1205821=
119896 sin 120585sin (120585119896)
120582119896 (85)
Coefficients of 119896th harmonic are given by (45)-(46)Notice that (68) can be rewritten as
120582119896= [cos(120585
119896)
119896 sin 120585sin(120585119896)
minus cos 120585]minus1
(86)
By introducing new variable
119909 = cos(120585119896) (87)
and using the Chebyshev polynomials (eg see Appendices)relations (85) and (86) can be rewritten as
1205821= 119896120582119896119880119896minus1
(119909) (88)
120582119896=
1
119896119909119880119896minus1
(119909) minus 119881119896(119909)
(89)
where119881119896(119909) and119880
119896(119909) denote the Chebyshev polynomials of
the first and second kind respectively From (89) it followsthat
120582119896[119896119909119880119896minus1
(119909) minus 119881119896(119909)] minus 1 = 0 (90)
which is polynomial equation of 119896th degree in terms of var-iable 119909 From 0 lt |120585| lt 120587 and (87) it follows that
cos(120587119896) lt 119909 lt 1 (91)
Since 120582119896is monotonically increasing function of |120585| 0 lt |120585| lt
120587 it follows that 120582119896is monotonically decreasing function of
119909 This further implies that (90) has only one solution thatsatisfies (91) (For 119896 = 2 expression (91) reads cos(1205872) le
119909 lt 1) This solution for 119909 (which can be obtained at leastnumerically) according to (88) leads to amplitude 120582
1of
fundamental harmonicFor 119896 le 4 solutions of (90) and (91) are
119909 = radic1 minus 1205822
21205822
1
3lt 1205822le 1
119909 =1
23radic1205823
1
8lt 1205823lt 1
119909 = radic1
6(1 + radic
51205824+ 3
21205824
)1
15lt 1205824lt 1
(92)
Insertion of (92) into (88) leads to the following relationsbetween amplitude 120582
1of fundamental and amplitude 120582
119896of
119896th harmonic 119896 le 4
1205821= radic8120582
2(1 minus 120582
2)
1
3lt 1205822le 1 (93)
1205821= 3 (
3radic1205823minus 1205823)
1
8lt 1205823lt 1 (94)
1205821= radic
32
27(radic2120582
4(3 + 5120582
4)3
minus 21205824(9 + 7120582
4))
1
15lt 1205824lt 1
(95)
Proof of Proposition 9 As it has been shown earlier (seeProposition 6) nonnegative waveform of type (35) with atleast one zero can be represented in form (38) Since weexclude nonnegative waveforms with 120582
119896= 1 according to
Remark 7 it follows that we exclude case |120585| = 120587Therefore inthe quest for nonnegative waveforms of type (35) having twozeros we will start with waveforms of type (38) for |120585| lt 120587It is clear that nonnegative waveforms of type (38) have twozeros if and only if
120582119896= [max120591
119903119896(120591)]minus1
(96)
and max120591119903119896(120591) = 119903
119896(1205910) According to (64) max
120591119903119896(120591) =
119903119896(1205910) implies |120585| = 0 Therefore it is sufficient to consider
only the interval (69)Substituting (96) into (38) we obtain
119879119896(120591) =
[1 minus cos (120591 minus 1205910)] [max
120591119903119896(120591) minus 119903
119896(120591)]
max120591119903119896(120591)
(97)
Mathematical Problems in Engineering 11
Expression max120591119903119896(120591) minus 119903
119896(120591) according to (64) and (39)
equals
max120591
119903119896(120591) minus 119903
119896(120591) = 119896
sin ((119896 minus 1) 120585119896)sin (120585119896)
minus 2
119896minus1
sum
119899=1
(119896 minus 119899) cos (119899 (120591 minus 1205910) + 120585)
(98)
Comparison of (97) with (66) yields
max120591
119903119896(120591) minus 119903
119896(120591) = [1 minus cos(120591 minus 120591
0+2120585
119896)]
sdot [1198880+ 2
119896minus2
sum
119899=1
119888119899cos 119899(120591 minus 120591
0+120585
119896)]
(99)
where coefficients 119888119899 119899 = 0 119896 minus 2 are given by (67) In
what follows we are going to show that right hand sides of(98) and (99) are equal
From (67) it follows that
1198880minus 1198881cos(120585
119896) = 119896
sin (120585 minus 120585119896)sin (120585119896)
(100)
Also from (67) for 119899 = 1 119896minus3 it follows that the followingrelations hold
(119888119899minus1
+ 119888119899+1
) cos(120585119896) minus 2119888
119899= 2 (119896 minus 119899) cos(120585 minus 119899120585
119896)
(119888119899minus1
minus 119888119899+1
) sin(120585119896) = 2 (119896 minus 119899) sin(120585 minus 119899120585
119896)
(101)
From (99) by using (75) (76) (100)-(101) and trigonometricidentities
cos(120591 minus 1205910+2120585
119896) = cos(120585
119896) cos(120591 minus 120591
0+120585
119896)
minus sin(120585119896) sin(120591 minus 120591
0+120585
119896)
cos(120585 minus 119899120585
119896) cos(119899(120591 minus 120591
0+120585
119896))
minus sin(120585 minus 119899120585
119896) sin(119899(120591 minus 120591
0+120585
119896))
= cos (119899 (120591 minus 1205910) + 120585)
(102)
we obtain (98) Consequently (98) and (99) are equal whichcompletes the proof
33 Nonnegative Waveforms with Two Zeros and PrescribedCoefficients of 119896thHarmonic In this subsectionwe show thatfor prescribed coefficients 119886
119896and 119887119896 there are 119896 nonnegative
waveforms of type (35) with exactly two zeros According to
(37) and (82) coefficients 119886119896and 119887119896of nonnegative waveforms
of type (35) with exactly two zeros satisfy the followingrelation
1
1198962 minus 1lt radic1198862119896+ 1198872119896lt 1 (103)
According to Remark 16 the value of 119909 (see (87)) that cor-responds to 120582
119896= radic1198862
119896+ 1198872119896can be determined from (90)-
(91) As we mentioned earlier (90) has only one solutionthat satisfies (91) This value of 119909 according to (88) leadsto the amplitude 120582
1of fundamental harmonic (closed form
expressions for 1205821in terms of 120582
119896and 119896 le 4 are given by (93)ndash
(95))On the other hand from (45)-(46) it follows that
1198961205910minus 120585 = atan 2 (119887
119896 119886119896) + 2119902120587 119902 = 1 (119896 minus 1) (104)
where function atan 2(119910 119909) is defined as
atan 2 (119910 119909) =
arctan(119910
119909) if 119909 ge 0
arctan(119910
119909) + 120587 if 119909 lt 0 119910 ge 0
arctan(119910
119909) minus 120587 if 119909 lt 0 119910 lt 0
(105)
with the codomain (minus120587 120587] Furthermore according to (84)and (104) the coefficients of fundamental harmonic of non-negative waveforms with two zeros and prescribed coeffi-cients of 119896th harmonic are equal to
1198861= minus1205821cos[
atan 2 (119887119896 119886119896) + 2119902120587
119896]
1198871= minus1205821sin[
atan 2 (119887119896 119886119896) + 2119902120587
119896]
(106)
where 119902 = 0 (119896minus 1) For chosen 119902 according to (104) and(66) positions of zeros are
1205910=1
119896[120585 + atan 2 (119887
119896 119886119896) + 2119902120587]
1205910minus2120585
119896=1
119896[minus120585 + atan 2 (119887
119896 119886119896) + 2119902120587]
(107)
From (106) and 119902 = 0 (119896minus1) it follows that for prescribedcoefficients 119886
119896and 119887119896 there are 119896 nonnegative waveforms of
type (35) with exactly two zerosWe provide here an algorithm to facilitate calculation
of coefficients 1198861and 1198871of nonnegative waveforms of type
(35) with two zeros and prescribed coefficients 119886119896and 119887
119896
providing that 119886119896and 119887119896satisfy (103)
12 Mathematical Problems in Engineering
2
1
0
Wav
efor
ms
0 1 2 3 4
Angle 120591120587
q = 0
q = 1
q = 2
Figure 6 Nonnegative waveforms with two zeros for 119896 = 3 1198863=
minus015 and 1198873= minus02
Algorithm 17 (i) Calculate 120582119896= radic1198862119896+ 1198872119896
(ii) identify 119909 that satisfies both relations (90) and (91)(iii) calculate 120582
1according to (88)
(iv) choose integer 119902 such that 0 le 119902 le 119896 minus 1(v) calculate 119886
1and 1198871according to (106)
For 119896 le 4 by using (93) for 119896 = 2 (94) for 119896 = 3 and (95)for 119896 = 4 it is possible to calculate directly 120582
1from 120582
119896and
proceed to step (iv)For 119896 = 2 and prescribed coefficients 119886
2and 1198872 there are
two waveforms with two zeros one corresponding to 1198861lt 0
and the other corresponding to 1198861gt 0 (see also [12])
Let us take as an input 119896 = 3 1198863= minus015 and 119887
3= minus02
Execution of Algorithm 17 on this input yields 1205823= 025 and
1205821= 11399 (according to (94)) For 119902 = 0 we calculate
1198861= minus08432 and 119887
1= 07670 (corresponding waveform is
presented by solid line in Figure 6) for 119902 = 1 we calculate1198861= minus02426 and 119887
1= minus11138 (corresponding waveform is
presented by dashed line) for 119902 = 2 we calculate 1198861= 10859
and 1198871= 03468 (corresponding waveform is presented by
dotted line)As another example of the usage of Algorithm 17 let us
consider case 119896 = 4 and assume that1198864= minus015 and 119887
4= minus02
Consequently 1205824= 025 and 120582
1= 09861 (according to (95))
For 119902 = 0 3we calculate the following four pairs (1198861 1198871) of
coefficients of fundamental harmonic (minus08388 05184) for119902 = 0 (minus05184 minus08388) for 119902 = 1 (08388 minus05184) for 119902 =2 and (05184 08388) for 119902 = 3 Corresponding waveformsare presented in Figure 7
4 Nonnegative Waveforms with MaximalAmplitude of Fundamental Harmonic
In this section we provide general description of nonnegativewaveforms containing fundamental and 119896th harmonic withmaximal amplitude of fundamental harmonic for prescribedamplitude of 119896th harmonic
The main result of this section is presented in the fol-lowing proposition
3
2
1
0
Wav
efor
ms
0 1 2 3 4
Angle 120591120587
q = 0q = 1
q = 2q = 3
Figure 7 Nonnegative waveforms with two zeros for 119896 = 4 1198864=
minus015 and 1198874= minus02
Proposition 18 Every nonnegativewaveformof type (35)withmaximal amplitude 120582
1of fundamental harmonic and pre-
scribed amplitude 120582119896of 119896th harmonic can be expressed in the
following form
119879119896(120591) = [1 minus cos (120591 minus 120591
0)]
sdot [1 minus (119896 minus 1) 120582119896minus 2120582119896
119896minus1
sum
119899=1
(119896 minus 119899) cos (119899 (120591 minus 1205910))]
(108)
if 0 le 120582119896le 1(119896
2
minus 1) or
119879119896(120591) = 120582
119896[1 minus cos (120591 minus 120591
0)] [1 minus cos(120591 minus 120591
0+2120585
119896)]
sdot [1198880+ 2
119896minus2
sum
119899=1
119888119899cos 119899(120591 minus 120591
0+120585
119896)]
(109)
if 1(1198962 minus 1) le 120582119896le 1 providing that 119888
119899 119899 = 0 119896 minus 2 and
120582119896are related to 120585 via relations (67) and (68) respectively and
|120585| le 120587
Remark 19 Expression (108) can be obtained from (38) bysetting 120585 = 0 Furthermore insertion of 120585 = 0 into (43)ndash(46)leads to the following expressions for coefficients ofwaveformof type (108)
1198861= minus (1 + 120582
119896) cos 120591
0 119887
1= minus (1 + 120582
119896) sin 120591
0
119886119896= 120582119896cos (119896120591
0) 119887
119896= 120582119896sin (119896120591
0)
(110)
On the other hand (109) coincides with (66) Thereforethe expressions for coefficients of (109) and (66) also coincideThus expressions for coefficients of fundamental harmonic ofwaveform (109) are given by (84) where 120582
1is given by (85)
while expressions for coefficients of 119896th harmonic are givenby (45)-(46)
Waveforms described by (108) have exactly one zerowhile waveforms described by (109) for 1(1198962 minus 1) lt 120582
119896lt 1
Mathematical Problems in Engineering 13
14
12
1
08
06
04
02
00 05 1
Amplitude 120582k
Am
plitu
de1205821
k = 2
k = 3
k = 4
Figure 8 Maximal amplitude of fundamental harmonic as a func-tion of amplitude of 119896th harmonic
have exactly two zeros As we mentioned earlier waveforms(109) for 120582
119896= 1 have 119896 zeros
Remark 20 Maximal amplitude of fundamental harmonic ofnonnegative waveforms of type (35) for prescribed amplitudeof 119896th harmonic can be expressed as
1205821= 1 + 120582
119896 (111)
if 0 le 120582119896le 1(119896
2
minus 1) or
1205821=
119896 sin 120585119896 sin 120585 cos (120585119896) minus cos 120585 sin (120585119896)
(112)
if 1(1198962 minus 1) le 120582119896le 1 where 120585 is related to 120582
119896via (68) (or
(86)) and |120585| le 120587From (110) it follows that (111) holds Substitution of (86)
into (85) leads to (112)Notice that 120582
119896= 1(119896
2
minus 1) is the only common point ofthe intervals 0 le 120582
119896le 1(119896
2
minus 1) and 1(1198962
minus 1) le 120582119896le
1 According to (111) 120582119896= 1(119896
2
minus 1) corresponds to 1205821=
1198962
(1198962
minus1) It can be also obtained from (112) by setting 120585 = 0The waveforms corresponding to this pair of amplitudes aremaximally flat nonnegative waveforms
Maximal amplitude of fundamental harmonic of non-negative waveform of type (35) for 119896 le 4 as a function ofamplitude of 119896th harmonic is presented in Figure 8
Remark 21 Maximum value of amplitude of fundamentalharmonic of nonnegative waveform of type (35) is
1205821max =
1
cos (120587 (2119896)) (113)
This maximum value is attained for |120585| = 1205872 (see (112)) Thecorresponding value of amplitude of 119896th harmonic is 120582
119896=
(1119896) tan(120587(2119896)) Nonnegative waveforms of type (35) with1205821= 1205821max have two zeros at 1205910 and 1205910 minus 120587119896 for 120585 = 1205872 or
at 1205910and 1205910+ 120587119896 for 120585 = minus1205872
14
12
1
08
06
04
02
0minus1 minus05 0 05 1
Am
plitu
de1205821
Parameter 120585120587
k = 2k = 3k = 4
Figure 9 Maximal amplitude of fundamental harmonic as a func-tion of parameter 120585
To prove that (113) holds let us first show that the fol-lowing relation holds for 119896 ge 2
cos( 120587
2119896) lt 1 minus
1
1198962 (114)
From 119896 ge 2 it follows that sinc(120587(4119896)) gt sinc(1205874) wheresinc 119909 = (sin119909)119909 and therefore sin(120587(4119896)) gt 1(radic2119896)By using trigonometric identity cos 2119909 = 1 minus 2sin2119909 weimmediately obtain (114)
According to (111) and (112) it is clear that 1205821attains its
maximum value on the interval 1(1198962 minus 1) le 120582119896le 1 Since
120582119896is monotonic function of |120585| on interval |120585| le 120587 (see
Remark 15) it follows that 119889120582119896119889120585 = 0 for 0 lt |120585| lt 120587
Therefore to find critical points of 1205821as a function of 120582
119896
it is sufficient to find critical points of 1205821as a function of
|120585| 0 lt |120585| lt 120587 and consider its values at the end points120585 = 0 and |120585| = 120587 Plot of 120582
1as a function of parameter 120585
for 119896 le 4 is presented in Figure 9 According to (112) firstderivative of 120582
1with respect to 120585 is equal to zero if and only
if (119896 cos 120585 sin(120585119896) minus sin 120585 cos(120585119896)) cos 120585 = 0 On interval0 lt |120585| lt 120587 this is true if and only if |120585| = 1205872 Accordingto (112) 120582
1is equal to 119896
2
(1198962
minus 1) for 120585 = 0 equal to zerofor |120585| = 120587 and equal to 1 cos(120587(2119896)) for |120585| = 1205872 From(114) it follows that 1198962(1198962minus1) lt 1 cos(120587(2119896)) and thereforemaximum value of 120582
1is given by (113) Moreover maximum
value of 1205821is attained for |120585| = 1205872
According to above consideration all nonnegative wave-forms of type (35) having maximum value of amplitude offundamental harmonic can be obtained from (109) by setting|120585| = 1205872 Three of them corresponding to 119896 = 3 120585 = 1205872and three different values of 120591
0(01205876 and1205873) are presented
in Figure 10 Dotted line corresponds to 1205910= 0 (coefficients
of corresponding waveform are 1198861= minus1 119887
1= 1radic3 119886
3= 0
and 1198873= minusradic39) solid line to 120591
0= 1205876 (119886
1= minus2radic3 119887
1= 0
1198863= radic39 and 119887
3= 0) and dashed line to 120591
0= 1205873 (119886
1= minus1
1198871= minus1radic3 119886
3= 0 and 119887
3= radic39)
Proof of Proposition 18 As it has been shown earlier (Propo-sition 6) nonnegative waveform of type (35) with at least
14 Mathematical Problems in Engineering
2
1
0
0 1 2 3 4
Angle 120591120587
Wav
efor
ms
1205910 = 01205910 = 12058761205910 = 1205873
Figure 10 Nonnegative waveforms with maximum amplitude offundamental harmonic for 119896 = 3 and 120585 = 1205872
one zero can be represented in form (38) According to (43)(44) and (36) for amplitude 120582
1of fundamental harmonic of
waveforms of type (38) the following relation holds
1205821= radic(1 + 120582
119896cos 120585)2 + 11989621205822
119896sin2120585 (115)
where 120582119896satisfy (40) and |120585| le 120587
Because of (40) in the quest of finding maximal 1205821for
prescribed 120582119896 we have to consider the following two cases
(Case i)120582119896lt [(119896minus1) cos 120585 + 119896 sin(120585minus120585119896) sin(120585119896)]minus1
(Case ii)120582119896= [(119896minus1) cos 120585+119896 sin(120585minus120585119896) sin(120585119896)]minus1
Case i Since 120582119896lt [(119896 minus 1) cos 120585 + 119896 sin(120585 minus 120585119896) sin(120585119896)]minus1
implies 120582119896
= 1 according to (115) it follows that 1205821
= 0Hence 119889120582
1119889120585 = 0 implies
2120582119896sin 120585 [1 minus (1198962 minus 1) 120582
119896cos 120585] = 0 (116)
Therefore 1198891205821119889120585 = 0 if 120582
119896= 0 (Option 1) or sin 120585 = 0
(Option 2) or (1198962 minus 1)120582119896cos 120585 = 1 (Option 3)
Option 1 According to (115) 120582119896= 0 implies 120582
1= 1 (notice
that this implication shows that 1205821does not depend on 120585 and
therefore we can set 120585 to zero value)
Option 2 According to (115) sin 120585 = 0 implies 1205821= 1 +
120582119896cos 120585 which further leads to the conclusion that 120582
1is
maximal for 120585 = 0 For 120585 = 0 120582119896lt [(119896 minus 1) cos 120585 + 119896 sin(120585 minus
120585119896) sin(120585119896)]minus1 becomes 120582119896lt 1(119896
2
minus 1)
Option 3 This option leads to contradiction To show thatnotice that (119896
2
minus 1)120582119896cos 120585 = 1 and 120582
119896lt [(119896 minus
1) cos 120585 + 119896 sin(120585 minus 120585119896) sin(120585119896)]minus1 imply that (119896 minus 1) cos 120585 gtsin(120585minus120585119896) sin(120585119896) Using (A5) (see Appendices) the latestinequality can be rewritten assum119896minus1
119899=1[cos 120585minuscos((119896minus2119899)120585119896)] gt
0 But from |119896 minus 2119899| lt 119896 119899 = 1 (119896 minus 1) and |120585| le 120587
it follows that all summands are not positive and therefore(119896minus1) cos 120585 gt sin(120585minus120585119896) sin(120585119896) does not hold for |120585| le 120587
Consequently Case i implies 120585 = 0 and 120582119896lt 1(119896
2
minus 1)Finally substitution of 120585 = 0 into (38) leads to (108) whichproves that (108) holds for 120582
119896lt 1(119896
2
minus 1)
Case ii Relation120582119896= [(119896minus1) cos 120585+119896 sin(120585minus120585119896) sin(120585119896)]minus1
according to Proposition 9 and Remark 11 implies that cor-responding waveforms can be expressed via (66)ndash(68) for|120585| le 120587 Furthermore 120582
119896= [(119896 minus 1) cos 120585 + 119896 sin(120585 minus
120585119896) sin(120585119896)]minus1 and |120585| le 120587 imply 1(1198962 minus 1) le 120582119896le 1
This proves that (109) holds for 1(1198962 minus 1) le 120582119896le 1
Finally let us prove that (108) holds for 120582119896= 1(119896
2
minus
1) According to (68) (see also Remark 11) this value of 120582119896
corresponds to 120585 = 0 Furthermore substitution of 120582119896=
1(1198962
minus 1) and 120585 = 0 into (109) leads to (70) which can berewritten as
119879119896(120591) =
[1 minus cos (120591 minus 1205910)]
(1 minus 1198962)
sdot [119896 (119896 minus 1) minus 2
119896minus1
sum
119899=1
(119896 minus 119899) cos (119899 (120591 minus 1205910))]
(117)
Waveform (117) coincides with waveform (108) for 120582119896
=
1(1 minus 1198962
) Consequently (108) holds for 120582119896= 1(1 minus 119896
2
)which completes the proof
5 Nonnegative Waveforms with MaximalAbsolute Value of the Coefficient of CosineTerm of Fundamental Harmonic
In this sectionwe consider general description of nonnegativewaveforms of type (35) with maximal absolute value ofcoefficient 119886
1for prescribed coefficients of 119896th harmonicThis
type of waveform is of particular interest in PA efficiencyanalysis In a number of cases of practical interest eithercurrent or voltage waveform is prescribed In such casesthe problem of finding maximal efficiency of PA can bereduced to the problem of finding nonnegative waveformwith maximal coefficient 119886
1for prescribed coefficients of 119896th
harmonic (see also Section 7)In Section 51 we provide general description of nonneg-
ative waveforms of type (35) with maximal absolute value ofcoefficient 119886
1for prescribed coefficients of 119896th harmonic In
Section 52 we illustrate results of Section 51 for particularcase 119896 = 3
51 Nonnegative Waveforms with Maximal Absolute Value ofCoefficient 119886
1for 119896 ge 2 Waveforms 119879
119896(120591) of type (35) with
1198861ge 0 can be derived from those with 119886
1le 0 by shifting
by 120587 and therefore we can assume without loss of generalitythat 119886
1le 0 Notice that if 119896 is even then shifting 119879
119896(120591) by
120587 produces the same result as replacement of 1198861with minus119886
1
(119886119896remains the same) On the other hand if 119896 is odd then
shifting 119879119896(120591) by 120587 produces the same result as replacement
of 1198861with minus119886
1and 119886119896with minus119886
119896
According to (37) coefficients of 119896th harmonic can beexpressed as
119886119896= 120582119896cos 120575 119887
119896= 120582119896sin 120575 (118)
Mathematical Problems in Engineering 15
where
|120575| le 120587 (119)
Conversely for prescribed coefficients 119886119896and 119887
119896 120575 can be
determined as
120575 = atan 2 (119887119896 119886119896) (120)
where definition of function atan 2(119910 119909) is given by (105)The main result of this section is stated in the following
proposition
Proposition 22 Every nonnegative waveform of type (35)withmaximal absolute value of coefficient 119886
1le 0 for prescribed
coefficients 119886119896and 119887119896of 119896th harmonic can be represented as
119879119896(120591)
= [1 minus cos 120591]
sdot [1 minus (119896 minus 1) 119886119896minus 2
119896minus1
sum
119899=1
(119896 minus 119899) (119886119896cos 119899120591 + 119887
119896sin 119899120591)]
(121)
if 119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) le 1+119886
119896 where 120575 = atan 2(bk
119886119896) or
119879119896(120591) = 120582
119896[1 minus cos(120591 minus (120575 + 120585)
119896)]
sdot [1 minus cos(120591 minus (120575 minus 120585)
119896)]
sdot [1198880+ 2
119896minus2
sum
119899=1
119888119899cos 119899(120591 minus 120575
119896)]
(122)
if 119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) ge 1 + 119886
119896 where 119888
119899 119899 = 0
119896minus2 and 120582119896= radic1198862119896+ 1198872119896are related to 120585 via relations (67) and
(68) respectively and |120585| le 120587
Remark 23 Expression (121) can be obtained from (38) bysetting 120591
0= 0 and 120585 = minus120575 and then replacing 120582
119896cos 120575 with
119886119896(see (118)) and 120582
119896cos(119899120591 minus 120575) with 119886
119896cos 119899120591 + 119887
119896sin 119899120591
(see also (118)) Furthermore insertion of 1205910= 0 and 120585 =
minus120575 into (43)ndash(46) leads to the following relations betweenfundamental and 119896th harmonic coefficients of waveform(121)
1198861= minus (1 + 119886
119896) 119887
1= minus119896119887
119896 (123)
On the other hand expression (122) can be obtained from(66) by replacing 120591
0minus120585119896with 120575119896 Therefore substitution of
1205910minus 120585119896 = 120575119896 in (84) leads to
1198861= minus1205821cos(120575
119896) 119887
1= minus1205821sin(120575
119896) (124)
where 1205821is given by (85)
The fundamental harmonic coefficients 1198861and 1198871of wave-
form of type (35) with maximal absolute value of coefficient1198861le 0 satisfy both relations (123) and (124) if 119886
119896and 119887119896satisfy
1 + 119886119896= 119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) For such waveforms
relations 1205910= 0 and 120585 = minus120575 also hold
Remark 24 Amplitude of 119896th harmonic of nonnegativewaveform of type (35) with maximal absolute value of coeffi-cient 119886
1le 0 and coefficients 119886
119896 119887119896satisfying 1 + 119886
119896=
119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) is
120582119896=
sin (120575119896)119896 sin 120575 cos (120575119896) minus cos 120575 sin (120575119896)
(125)
To show that it is sufficient to substitute 119886119896= 120582119896cos 120575 (see
(118)) into 1 + 119886119896= 119896120582119896[sin 120575 sin(120575119896)] cos(120575119896)
Introducing new variable
119910 = cos(120575119896) (126)
and using the Chebyshev polynomials (eg see Appendices)relations 119886
119896= 120582119896cos 120575 and (125) can be rewritten as
119886119896= 120582119896119881119896(119910) (127)
120582119896=
1
119896119910119880119896minus1
(119910) minus 119881119896(119910)
(128)
where119881119896(119910) and119880
119896(119910) denote the Chebyshev polynomials of
the first and second kind respectively Substitution of (128)into (127) leads to
119886119896119896119910119880119896minus1
(119910) minus (1 + 119886119896) 119881119896(119910) = 0 (129)
which is polynomial equation of 119896th degree in terms of var-iable 119910 From |120575| le 120587 and (126) it follows that
cos(120587119896) le 119910 le 1 (130)
In what follows we show that 119886119896is monotonically increas-
ing function of 119910 on the interval (130) From 120585 = minus120575 (seeRemark 23) and (81) it follows that 120582minus1
119896= (119896 minus 1) cos 120575 +
119896sum119896minus1
119899=1cos((119896 minus 2119899)120575119896) ge 1 and therefore 119886
119896= 120582119896cos 120575 can
be rewritten as
119886119896=
cos 120575(119896 minus 1) cos 120575 + 119896sum119896minus1
119899=1cos ((119896 minus 2119899) 120575119896)
(131)
Obviously 119886119896is even function of 120575 and all cosines in (131)
are monotonically decreasing functions of |120575| on the interval|120575| le 120587 It is easy to show that cos((119896 minus 2119899)120575119896) 119899 =
1 (119896 minus 1) decreases slower than cos 120575 when |120575| increasesThis implies that denominator of the right hand side of(131) decreases slower than numerator Since denominator ispositive for |120575| le 120587 it further implies that 119886
119896is decreasing
function of |120575| on interval |120575| le 120587 Consequently 119886119896is
monotonically increasing function of 119910 on the interval (130)Thus we have shown that 119886
119896is monotonically increasing
function of 119910 on the interval (130) and therefore (129) hasonly one solution that satisfies (130) According to (128) thevalue of 119910 obtained from (129) and (130) either analyticallyor numerically leads to amplitude 120582
119896of 119896th harmonic
16 Mathematical Problems in Engineering
1
05
0
minus05
minus1
minus1 minus05 0 05 1
Coefficient ak
Coe
ffici
entb
k
radica2k+ b2
kle 1
k = 2k = 3k = 4
Figure 11 Plot of (119886119896 119887119896) satisfying 1 + 119886
119896= 119896120582
119896[sin 120575 sin(120575
119896)] cos(120575119896) for 119896 le 4
By solving (129) and (130) for 119896 le 4 we obtain
119910 = radic1 + 1198862
2 (1 minus 1198862) minus1 le 119886
2le1
3
119910 = radic3
4 (1 minus 21198863) minus1 le 119886
3le1
8
119910 =radicradic2 minus 4119886
4+ 1011988624minus 2 (1 minus 119886
4)
4 (1 minus 31198864)
minus1 le 1198864le
1
15
(132)
Insertion of (132) into (128) leads to the following explicitexpressions for the amplitude 120582
119896 119896 le 4
1205822=1
2(1 minus 119886
2) minus1 le 119886
2le1
3 (133)
1205822
3= [
1
3(1 minus 2119886
3)]
3
minus1 le 1198863le1
8 (134)
1205824=1
4(minus1 minus 119886
4+ radic2 minus 4119886
4+ 1011988624) minus1 le 119886
4le
1
15
(135)
Relations (133)ndash(135) define closed lines (see Figure 11) whichseparate points representing waveforms of type (121) frompoints representing waveforms of type (122) For given 119896points inside the corresponding curve refer to nonnegativewaveforms of type (121) whereas points outside curve (andradic1198862119896+ 1198872119896le 1) correspond to nonnegative waveforms of type
(122) Points on the respective curve correspond to the wave-forms which can be expressed in both forms (121) and (122)
Remark 25 Themaximum absolute value of coefficient 1198861of
nonnegative waveform of type (35) is
100381610038161003816100381611988611003816100381610038161003816max =
1
cos (120587 (2119896)) (136)
This maximum value is attained for |120585| = 1205872 and 120575 = 0
(see (124)) Notice that |1198861|max is equal to the maximum value
1205821max of amplitude of fundamental harmonic (see (113))
Coefficients of waveform with maximum absolute value ofcoefficient 119886
1 1198861lt 0 are
1198861= minus
1
cos (120587 (2119896)) 119886
119896=1
119896tan( 120587
(2119896))
1198871= 119887119896= 0
(137)
Waveformdescribed by (137) is cosinewaveformhaving zerosat 120587(2119896) and minus120587(2119896)
In the course of proving (136) notice first that |1198861|max le
1205821max holds According to (123) and (124) maximum of |119886
1|
occurs for 119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) ge 1 + 119886
119896 From (124)
it immediately follows that maximum value of |1198861| is attained
if and only if 1205821= 1205821max and 120575 = 0 which because of
120575119896 = 1205910minus120585119896 further implies 120591
0= 120585119896 Sincemaximumvalue
of 1205821is attained for |120585| = 1205872 it follows that corresponding
waveform has zeros at 120587(2119896) and minus120587(2119896)
Proof of Proposition 22 As it was mentioned earlier in thissection we can assume without loss of generality that 119886
1le 0
We consider waveforms119879119896(120591) of type (35) such that119879
119896(120591) ge 0
and119879119896(120591) = 0 for some 120591
0 Fromassumption that nonnegative
waveform 119879119896(120591) of type (35) has at least one zero it follows
that it can be expressed in form (38)Let us also assume that 120591
0is position of nondegenerate
critical point Therefore 119879119896(1205910) = 0 implies 1198791015840
119896(1205910) = 0 and
11987910158401015840
119896(1205910) gt 0 According to (55) second derivative of 119879
119896(120591) at
1205910can be expressed as 11987910158401015840
119896(1205910) = 1 minus 120582
119896(1198962
minus 1) cos 120585 Since11987910158401015840
119896(1205910) gt 0 it follows immediately that
1 minus 120582119896(1198962
minus 1) cos 120585 gt 0 (138)
Let us further assume that 119879119896(120591) has exactly one zeroThe
problem of finding maximum absolute value of 1198861is con-
nected to the problem of finding maximum of the minimumfunction (see Section 21) If waveforms possess unique globalminimum at nondegenerate critical point then correspond-ing minimum function is a smooth function of parameters[13] Consequently assumption that 119879
119896(120591) has exactly one
zero at nondegenerate critical point leads to the conclusionthat coefficient 119886
1is differentiable function of 120591
0 First
derivative of 1198861(see (43)) with respect to 120591
0 taking into
account that 1205971205851205971205910= 119896 (see (50)) can be expressed in the
following factorized form
1205971198861
1205971205910
= sin 1205910[1 minus 120582
119896(1198962
minus 1) cos 120585] (139)
Mathematical Problems in Engineering 17
From (138) and (139) it is clear that 12059711988611205971205910= 0 if and only if
sin 1205910= 0 According toRemark 12 assumption that119879
119896(120591)has
exactly one zero implies 120582119896lt 1 From (51) (48) and 120582
119896lt 1
it follows that 1198861cos 1205910+ 1198871sin 1205910lt 0 which together with
sin 1205910= 0 implies that 119886
1cos 1205910lt 0 Assumption 119886
1le 0
together with relations 1198861cos 1205910lt 0 and sin 120591
0= 0 further
implies 1198861
= 0 and
1205910= 0 (140)
Insertion of 1205910= 0 into (38) leads to
119879119896(120591)
= [1 minus cos 120591]
sdot [1 minus (119896 minus 1) 120582119896cos 120585 minus 2120582
119896
119896minus1
sum
119899=1
(119896 minus 119899) cos (119899120591 + 120585)]
(141)
Substitution of 1205910= 0 into (45) and (46) yields 119886
119896= 120582119896cos 120585
and 119887119896
= minus120582119896sin 120585 respectively Replacing 120582
119896cos 120585 with
119886119896and 120582
119896cos(119899120591 + 120585) with (119886
119896cos 119899120591 + 119887
119896sin 119899120591) in (141)
immediately leads to (121)Furthermore 119886
119896= 120582119896cos 120585 119887
119896= minus120582
119896sin 120585 and (118)
imply that
120575 = minus120585 (142)
According to (38)ndash(40) and (142) it follows that (141) is non-negative if and only if
120582119896[(119896 minus 1) cos 120575 + 119896 sin (120575 minus 120575119896)
sin (120575119896)] lt 1 (143)
Notice that 119886119896= 120582119896cos 120575 implies that the following relation
holds
120582119896[(119896 minus 1) cos 120575 + 119896 sin (120575 minus 120575119896)
sin (120575119896)]
= minus119886119896+ 119896120582119896
sin 120575sin (120575119896)
cos(120575119896)
(144)
Finally substitution of (144) into (143) leads to 119896120582119896[sin 120575
sin(120575119896)] cos(120575119896) lt 1 + 119886119896 which proves that (121) holds
when 119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) lt 1 + 119886
119896
Apart from nonnegative waveforms with exactly one zeroat nondegenerate critical point in what follows we will alsoconsider other types of nonnegative waveforms with at leastone zero According to Proposition 9 and Remark 11 thesewaveforms can be described by (66)ndash(68) providing that 0 le|120585| le 120587
According to (35) 119879119896(0) ge 0 implies 1 + 119886
1+ 119886119896ge 0
Consequently 1198861le 0 implies that |119886
1| le 1 + 119886
119896 On the other
hand according to (123) |1198861| = 1 + 119886
119896holds for waveforms
of type (121) The converse is also true 1198861le 0 and |119886
1| =
1 + 119886119896imply 119886
1= minus1 minus 119886
119896 which further from (35) implies
119879119896(0) = 0 Therefore in what follows it is enough to consider
only nonnegativewaveformswhich can be described by (66)ndash(68) and 0 le |120585| le 120587 with coefficients 119886
119896and 119887119896satisfying
119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) ge 1 + 119886
119896
For prescribed coefficients 119886119896and 119887119896 the amplitude 120582
119896=
radic1198862119896+ 1198872119896of 119896th harmonic is also prescribed According to
Remark 15 (see also Remark 16) 120582119896is monotonically
decreasing function of 119909 = cos(120585119896) The value of 119909 can beobtained by solving (90) subject to the constraint cos(120587119896) le119909 le 1 Then 120582
1can be determined from (88) From (106) it
immediately follows that maximal absolute value of 1198861le 0
corresponds to 119902 = 0 which from (104) and (120) furtherimplies that
120575 = 1198961205910minus 120585 (145)
Furthermore 119902 = 0 according to (107) implies that waveformzeros are
1205910=(120575 + 120585)
119896 120591
1015840
0= 1205910minus2120585
119896=(120575 minus 120585)
119896 (146)
Substitution of 1205910= (120575 + 120585)119896 into (66) yields (122) which
proves that (122) holds when 119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) ge
1 + 119886119896
In what follows we prove that (121) also holds when119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) = 1 + 119886
119896 Substitution of 119886
119896=
120582119896cos 120575 into 119896120582
119896[sin 120575 sin(120575119896)] cos(120575119896) = 1 + 119886
119896leads to
120582119896[(119896 minus 1) cos 120575 + 119896 sin (120575 minus 120575119896)
sin (120575119896)] = 1 (147)
As we mentioned earlier relation (142) holds for all wave-forms of type (121) Substituting (142) into (147) we obtain
120582119896[(119896 minus 1) cos 120585 + 119896 sin (120585 minus 120585119896)
sin (120585119896)] = 1 (148)
This expression can be rearranged as
120582119896
119896 sin ((119896 minus 1) 120585119896)sin 120585119896
= 1 minus (119896 minus 1) 120582119896cos 120585 (149)
On the other hand for waveforms of type (122) according to(68) relations (148) and (149) also hold Substitution of 120591
0=
(120575 + 120585)119896 (see (145)) and (67) into (122) leads to
119879119896(120591)
= 120582119896[1 minus cos (120591 minus 120591
0)]
sdot [119896 sin ((119896 minus 1) 120585119896)
sin 120585119896minus 2
119896minus1
sum
119899=1
(119896 minus 119899) cos (119899 (120591 minus 1205910) + 120585)]
(150)
Furthermore substitution of (142) into (145) implies that1205910
= 0 Finally substitution of 1205910
= 0 and (149) into(150) leads to (141) Therefore (141) holds when 119896120582
119896[sin 120575
sin(120575119896)] cos(120575119896) = 1 + 119886119896 which in turn shows that (121)
holds when 119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) = 1 + 119886
119896 This
completes the proof
18 Mathematical Problems in Engineering
52 Nonnegative Waveforms with Maximal Absolute Valueof Coefficient 119886
1for 119896 = 3 Nonnegative waveform of type
(35) for 119896 = 3 is widely used in PA design (eg see [10])In this subsection we illustrate results of Section 51 for thisparticular case The case 119896 = 2 is presented in detail in [12]
Coefficients of fundamental harmonic of nonnegativewaveform of type (35) with 119896 = 3 and maximal absolutevalue of coefficient 119886
1le 0 for prescribed coefficients 119886
3and
1198873(1205823= radic11988623+ 11988723) according to (123) (124) (134) (94) and
(120) are equal to
1198861= minus1 minus 119886
3 119887
1= minus3119887
3 (151)
if 12058223le [(1 minus 2119886
3)3]3
1198861= minus1205821cos(120575
3) 119887
1= minus1205821sin(120575
3) (152)
where 1205821= 3(
3radic1205823minus 1205823) and 120575 = atan 2(119887
3 1198863) if [(1 minus
21198863)3]3
le 1205822
3le 1The line 1205822
3= [(1minus2119886
3)3]3 (see case 119896 = 3
in Figure 11) separates points representing waveforms withcoefficients satisfying (151) from points representing wave-forms with coefficients satisfying (152) Waveforms describedby (151) for 1205822
3lt [(1 minus 2119886
3)3]3 have exactly one zero at
1205910= 0 Waveforms described by (151) and (152) for 1205822
3= [(1 minus
21198863)3]3 also have zero at 120591
0= 0 These waveforms as a rule
have exactly two zeros However there are two exceptionsone related to the maximally flat nonnegative waveform withcoefficients 119886
1= minus98 119886
3= 18 and 119887
1= 1198873= 0 which
has only one zero and the other related to the waveform withcoefficients 119886
1= 0 119886
3= minus1 and 119887
1= 1198873= 0 which has three
zerosWaveforms described by (152) for [(1minus21198863)3]3
lt 1205822
3lt
1 have two zeros Waveforms with 1205823= 1 have only third
harmonic (fundamental harmonic is zero)Plot of contours of maximal absolute value of coefficient
1198861 1198861le 0 for prescribed coefficients 119886
3and 1198873is presented
in Figure 12 According to Remark 25 the waveform withmaximum absolute value of 119886
1le 0 is fully described with
the following coefficients 1198861
= minus2radic3 1198863
= radic39 and1198871= 1198873= 0 This waveform has two zeros at plusmn1205876
Two examples of nonnegative waveforms for 119896 = 3
and maximal absolute value of coefficient 1198861 1198861le 0 with
prescribed coefficients 1198863and 1198873are presented in Figure 13
One waveform corresponds to the case 12058223lt [(1 minus 2119886
3)3]3
(solid line) and the other to the case 12058223gt [(1 minus 2119886
3)3]3
(dashed line)Thewaveform represented by solid line has onezero and its coefficients are 119886
3= minus01 119887
3= 01 119886
1= minus09
and 1198871= minus03 Dashed line corresponds to the waveform
having two zeros with coefficients 1198863= minus01 119887
3= 03 119886
1=
minus08844 and 1198871= minus06460 (case 1205822
3gt [(1 minus 2119886
3)3]3)
6 Nonnegative Cosine Waveforms withat Least One Zero
Nonnegative cosine waveforms have proved to be of impor-tance for waveform modelling in PA design (eg see [10])In this section we consider nonnegative cosine waveforms
1
05
0
minus05
minus1
minus1 minus05 0 05 1
Coefficient a3
Coe
ffici
entb
3
02
04
06
08
10
11
Figure 12 Contours ofmaximal absolute value of coefficient 1198861 1198861le
0 as a function of 1198863and 1198873
2
1
0
0 1 2 3 4
Angle 120591120587
Wav
efor
ms
a3 = minus01 b3 = 01
a3 = minus01 b3 = 03
Figure 13 Nonnegative waveforms for 119896 = 3 and maximal absolutevalue of 119886
1 1198861le 0 with prescribed coefficients 119886
3and 1198873
containing fundamental and 119896th harmonic with at least onezero
Cosine waveform with dc component fundamental and119896th harmonic can be obtained from (35) by setting 119887
1= 119887119896=
0 that is
119879119896(120591) = 1 + 119886
1cos 120591 + 119886
119896cos 119896120591 (153)
In Section 61 we provide general description of non-negative cosine waveforms of type (153) with at least onezero We show that nonnegative cosine waveforms with atleast one zero coincide with nonnegative cosine waveformswith maximal absolute value of coefficient 119886
1for prescribed
coefficient 119886119896 In Section 62 we illustrate results of Section 61
for particular case 119896 = 3
61 Nonnegative Cosine Waveforms with at Least One Zerofor 119896 ge 2 Amplitudes of fundamental and 119896th harmonic
Mathematical Problems in Engineering 19
of cosine waveform of type (153) are 1205821= |1198861| and 120582
119896=
|119886119896| respectively According to (42) for nonnegative cosine
waveforms of type (153) the following relation holds
minus1 le 119886119896le 1 (154)
This explains why 119896th harmonic coefficient 119886119896in Proposi-
tion 26 goes through interval [minus1 1]Waveforms (153) with 119886
1ge 0 can be obtained from
waveforms with 1198861le 0 by shifting by 120587 and therefore with-
out loss of generality we can assume that 1198861le 0
Proposition 26 Each nonnegative cosine waveform of type(153) with 119886
1le 0 and at least one zero can be represented as
119879119896(120591) = [1 minus cos 120591] [1 minus (119896 minus 1) 119886
119896minus 2119886119896
119896minus1
sum
119899=1
(119896 minus 119899) cos 119899120591]
(155)
if minus1 le 119886119896le 1(119896
2
minus 1) or
119879119896(120591) = 119886
119896[1 minus cos (120591 minus 120591
0)] [1 minus cos (120591 + 120591
0)]
sdot [1198880+ 2
119896minus2
sum
119899=1
119888119899cos 119899120591]
(156)
where
119888119899=sin ((119896 minus 119899) 120591
0) cos 120591
0minus (119896 minus 119899) cos ((119896 minus 119899) 120591
0) sin 120591
0
sin31205910
(157)
119886119896=
sin 1205910
119896 sin (1198961205910) cos 120591
0minus cos (119896120591
0) sin 120591
0
(158)
100381610038161003816100381612059101003816100381610038161003816 le
120587
119896 (159)
if 1(1198962 minus 1) le 119886119896le 1
Remark 27 Identity [1minuscos(120591minus1205910)][1minuscos(120591+120591
0)] = [cos 120591
0minus
cos 120591]2 implies that (156) can be rewritten as
119879119896(120591) = 119886
119896[cos 120591
0minus cos 120591]2 [119888
0+ 2
119896minus2
sum
119899=1
119888119899cos 119899120591] (160)
Furthermore substitution of (157) into (160) leads to
119879119896(120591) = 119886
119896[cos 120591
0minus cos 120591]
sdot [(119896 minus 1) sin 119896120591
0
sin 1205910
minus 2
119896minus1
sum
119899=1
sin ((119896 minus 119899) 1205910)
sin 1205910
cos 119899120591]
(161)
Remark 28 All nonnegative cosine waveforms of type (153)with at least one zero and 119886
1le 0 except one of them can be
represented either in form (155) or form (156)This exceptionis maximally flat cosine waveform with 119886
1lt 0 which can be
obtained from (155) for 119886119896= 1(119896
2
minus 1) or from (156) for 1205910=
0 Maximally flat cosine waveform with 1198861lt 0 can also be
obtained from (70) by setting 1205910= 0 Furthermore setting
1205910= 0 in (71) leads to maximally flat cosine waveforms for
119896 le 4 and 1198861lt 0
Remark 29 Nonnegative cosine waveform of type (155) with1198861lt 0 and minus1 lt 119886
119896le 1(119896
2
minus 1) has exactly one zero at120591 = 0 Nonnegative cosine waveform described by (156) with1198861lt 0 and 1(1198962 minus 1) lt 119886
119896lt 1 has two zeros at plusmn120591
0 where
0 lt |1205910| lt 120587119896 For 119886
119896= ∓1 nonnegative cosine waveform
of type (153) reduces to 119879119896(120591) = 1 ∓ cos 119896120591 (clearly these two
waveforms both have 119896 zeros)
Remark 30 Transformation of (155) into an additive formleads to the following relation
1198861= minus1 minus 119886
119896 (162)
where minus1 le 119886119896le 1(119896
2
minus1) Similarly transformation of (156)leads to the following relation
1198861= minus119886119896
119896 sin 1198961205910
sin 1205910
(163)
where 119886119896is given by (158) 1(1198962minus1) le 119886
119896le 1 and |120591
0| le 120587119896
Notice that coefficients of maximally flat cosine waveformnamely 119886
119896= 1(119896
2
minus1) and 1198861= minus1198962
(1198962
minus1) satisfy relation(162) They also satisfy relation (163) for 120591
0= 0
Remark 31 Nonnegative cosine waveforms of type (153) withat least one zero coincide with nonnegative cosine waveformswith maximal absolute value of coefficient 119886
1for prescribed
coefficient 119886119896
In proving that Remark 31 holds notice that expression(155) can be obtained from (121) by setting 119887
119896= 0 Fur-
thermore if 119886119896
ge 0 then 120582119896
= 119886119896 which together
with 119887119896
= 0 and (118) implies 120575 = 0 In this case119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) le 1 + 119886
119896becomes 1198962119886
119896le 1 +
119886119896 On the other hand if 119886
119896lt 0 then 120582
119896= minus119886
119896 which
together with 119887119896= 0 and (118) implies |120575| = 120587 In this case
119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) le 1 + 119886
119896becomes 0 le 1 + 119886
119896
Therefore every nonnegative cosine waveform of type (155)has maximal absolute value of coefficient 119886
1for prescribed
coefficient 119886119896 when minus1 le 119886
119896le 1(119896
2
minus 1)Let us now show that expression (156) can be obtained
from (122) by setting 119887119896= 0 and 119886
119896gt 0 For waveforms
of type (122) according to (118) 119887119896= 0 and 119886
119896gt 0 imply
120575 = 0 and 120582119896= 119886119896 Substitution of 120582
119896= 119886119896and 120575 = 0 into
119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) ge 1+119886
119896leads to 119886
119896ge 1(119896
2
minus1)Furthermore substitution of 120575 = 0 into (145) yields 120591
0= 120585119896
Insertion of 120582119896= 119886119896 120575 = 0 and 120591
0= 120585119896 into (122) leads
to (156) Therefore every nonnegative cosine waveform oftype (156) has maximal absolute value of coefficient 119886
1for
prescribed coefficient 119886119896 when 1(1198962 minus 1) le 119886
119896le 1
Proof of Proposition 26 Let us start with nonnegative cosinewaveform of type (153) with 120582
119896= |119886119896| = 1 According to
Remark 7 120582119896= |119886119896| = 1 implies that 120582
1= |1198861| = 0
Substitution of 119886119896
= minus1 into (155) and using (A2) (seeAppendices) lead to 119879
119896(120591) = 1 minus cos 119896120591 Consequently (155)
holds for 119886119896= minus1 On the other hand substitution of 119886
119896= 1
into (158) yields |1205910| = 120587119896 Furthermore substitution of
119886119896= 1 and 120591
0= 120587119896 (or 120591
0= minus120587119896) into (156) along
with performing all multiplications and using (A2) leads to
20 Mathematical Problems in Engineering
119879119896(120591) = 1 + cos 119896120591 Consequently (156)ndash(158) hold for 119886
119896= 1
and |1205910| = 120587119896
It is easy to see that 120582119896= |119886119896| lt 1 and 119879
119896(1205910) = 0 for some
1205910imply 120582
1= |1198861| = 0 Therefore in what follows we assume
that |119886119896| = 1 and 119886
1lt 0
Cosine waveforms are even functions of 120591 Therefore ifnonnegative cosine waveform has exactly one zero it has to beeither at 0 or at 120587 On the other hand if nonnegative cosinewaveform with 119886
1= 0 has exactly two zeros then these zeros
are placed at plusmn1205910 such that 120591
0is neither 0 nor 120587
In order to prove that (155) holds for minus1 lt 119886119896le 1(1minus119896
2
)let us start by referring to the description (38) of nonnegativewaveformswith at least one zero As wementioned earlier fornonnegative cosine waveformwith exactly one zero (denotedby 1205910) it is either 120591
0= 0 or 120591
0= 120587 Therefore in both cases
sin 1205910= 0 Substitution of sin 120591
0= 0 into (43) together with
1198861
= 0 and 120582119896= |119886119896| lt 1 leads to
1205910= 0 (164)
Clearly 1205910= 0 119887
1= 0 and 119887
119896= 0 according to (44) and (46)
imply 120582119896sin 120585 = 0 Since 120582
119896= |119886119896| it follows that |119886
119896| sin 120585 = 0
also holds which further implies 120582119896= 119886119896= 0 or sin 120585 =
0 In the case when 120582119896= 119886119896= 0 from (164) and (43) we
obtain 1198861= minus1 which further implies that 119879
119896(120591) = 1 minus cos 120591
Consequently (155) holds for 119886119896= 0 In the case when sin 120585 =
0 from (164) and (45) we obtain 119886119896= 120582119896if 120585 = 0 or 119886
119896= minus120582119896
if 120585 = 120587 Relations 119886119896= 120582119896and 120585 = 0 according to (40) imply
that 0 le 119886119896le 1(1 minus 119896
2
) Substitution of 120585 = 0 120582119896= 119886119896 and
(164) into (38) leads to (155) which proves that (155) holdsfor 0 le 119886
119896le 1(1 minus 119896
2
) On the other hand relations 119886119896=
minus120582119896and 120585 = 120587 according to (40) imply that minus1 lt 119886
119896le 0
Substitution of 120585 = 120587 120582119896= minus119886119896 and (164) into (38) also leads
to (155) which proves that (155) also holds for minus1 lt 119886119896le 0
Consequently (155) holds for minus1 lt 119886119896le 1(1 minus 119896
2
)In what follows we first prove that (156)-(157) hold for
1(1 minus 1198962
) lt 119886119896lt 1 For this purpose let us start with non-
negative waveforms with two zeros described by (66) As wementioned before nonnegative cosine waveforms with twozeros have zeros at 120591
0and minus120591
0 such that 120591
0= 0 and 120591
0= 120587
Relations 1198861
lt 0 and 1198871
= 0 according to (84) implycos(1205910minus 120585119896) = 1 and therefore
120585
119896= 1205910 (165)
From 120585119896 = 1205910and 0 lt |120585| lt 120587 it follows that 0 lt |120591
0| lt
120587119896 Insertion of 120585119896 = 1205910into (45) yields 119886
119896= 120582119896 Relations
119886119896= 120582119896and (82) imply that 1(1 minus 1198962) lt 119886
119896lt 1 Substitution
of 120582119896= 119886119896and 120585119896 = 120591
0into (66)ndash(68) leads to (156)ndash(158)
which proves that (156)ndash(158) hold for 1(1 minus 1198962
) lt 119886119896lt 1
and 0 lt |1205910| lt 120587119896
Finally substitution of 119886119896= 1(1 minus 119896
2
) and 1205910= 0 into
(161) leads to
119879119896(120591) =
[1 minus cos 120591](1 minus 1198962)
[119896 (119896 minus 1) minus 2
119896minus1
sum
119899=1
(119896 minus 119899) cos 119899120591]
(166)
Waveform (166) coincides with waveform (155) for 119886119896
=
1(1 minus 1198962
) which in turn proves that (156) holds for 119886119896=
1(1 minus 1198962
) and 1205910= 0 This completes the proof
62 Nonnegative CosineWaveforms with at Least One Zero for119896 = 3 In this subsection we consider nonnegative cosinewaveforms with at least one zero for 119896 = 3 (for case 119896 = 2
see [12])Cosine waveform with fundamental and third harmonic
reads
1198793(120591) = 1 + 119886
1cos 120591 + 119886
3cos 3120591 (167)
For 1198861le 0 and minus1 le 119886
3le 18 according to (155) non-
negative cosine waveform of type (167) with at least one zerocan be expressed as
1198793(120591) = (1 minus cos 120591) [1 minus 2119886
3(1 + 2 cos 120591 + cos 2120591)] (168)
From 1198793(120591 + 120587) = 2 minus 119879
3(120591) it immediately follows that for
1198861ge 0 and minus18 le 119886
3le 1 119879
3(120591) can be expressed as
1198793(120591) = (1 + cos 120591) [1 + 2119886
3(1 minus 2 cos 120591 + cos 2120591)] (169)
For 1198861le 0 and 18 le 119886
3le 1 from (158) it follows that 119886
3=
[8cos31205910]minus1 This relation along with (160) and (157) further
implies that 1198793(120591) can be expressed as
1198793(120591) =
[cos 1205910minus cos 120591]2 [2 cos 120591
0+ cos 120591]
2cos31205910
(170)
providing that |1205910| le 1205873 From 119879
3(120591 + 120587) = 2 minus 119879
3(120591) it
follows that (170) also holds for 1198861ge 0 and minus1 le 119886
3le minus18
providing that 1205910isin [21205873 41205873]
Maximally flat nonnegative cosinewaveformof type (167)with 119886
1lt 0 (minimum at 120591
0= 0) reads 119879
3(120591) = [1 minus
cos 120591]2[1 + (12) cos 120591] Dually maximally flat nonnegativecosine waveform with 119886
1gt 0 (minimum at 120591
0= 120587) reads
1198793(120591) = [1 + cos 120591]2[1 minus (12) cos 120591]In what follows we provide relations between coefficients
1198861and 1198863of nonnegative cosine waveforms of type (167) with
at least one zeroFor 1198861le 0 conversion of (168) into an additive form
immediately leads to the following relation
1198861= minus1 minus 119886
3for minus 1 le 119886
3le1
8 (171)
Conversion of (170) into an additive form leads to 1198861
=
minus31198863(2 cos 2120591
0+ 1) which can be also expressed as 119886
1=
minus31198863(4cos2120591
0minus 1) For 119886
1le 0 relations |120591
0| le 1205873 119886
1=
minus31198863(4cos2120591
0minus 1) and 119886
3= [8cos3120591
0]minus1 lead to
1198861= minus3 [ 3radic119886
3minus 1198863] for 1
8le 1198863le 1 (172)
Similarly for 1198861ge 0 conversion of (169) into an additive form
leads to the following relation
1198861= 1 minus 119886
3for minus
1
8le 1198863le 1 (173)
Mathematical Problems in Engineering 21
1
15
15
05
0
minus05
minus15minus15
minus1
minus1 minus05 0 05 1
Coefficient a1
Coe
ffici
enta
3 a1= minus
3(3radica3minus a3)
a1 =
minus1 minus a3
a1 =
1 minusa3
a1 =3(3radic|a3
| +a3)
Figure 14 Parameter space of cosine waveforms for 119896 = 3
For waveform of type (170) with 1198861ge 0 relations 120591
0isin [21205873
41205873] 1198861= minus3119886
3(4cos2120591
0minus 1) and 119886
3= [8cos3120591
0]minus1 lead to
1198861= 3 [
3radic10038161003816100381610038161198863
1003816100381610038161003816 + 1198863] for minus 1 le 1198863le minus
1
8 (174)
Every cosine waveform of type (167) corresponds to apair of real numbers (119886
1 1198863) and vice versa Points (119886
1 1198863)
in grey area in Figure 14 correspond to nonnegative cosinewaveforms for 119896 = 3 The points at the boundary of grey areacorrespond to nonnegative cosinewaveformswith at least onezero A number of shapes of nonnegative cosine waveformswith 119896 = 3 and at least one zero plotted on interval [minus120587 120587]are also presented in Figure 14 The boundary of grey area inFigure 14 consists of four line segments described by relations(171)ndash(174) The common point of line segments (172) and(173) is cusp point with coordinates 119886
1= 0 and 119886
3= 1
Another cusp point with coordinates 1198861= 0 and 119886
3= minus1
is the common point of line segments (171) and (174) Thecommon point of line segments (171)-(172) has coordinates(minus98 18) and common point of line segments (173)-(174)has coordinates (98 minus18) These points are representedby white circle dots and they correspond to maximallyflat cosine waveforms (eg see [21]) White triangle dotswith coordinates (2radic3 minusradic39) and (minus2radic3radic39) refer tothe nonnegative cosine waveforms with maximum value ofamplitude of fundamental harmonic
7 Four Case Studies of Usage of NonnegativeWaveforms in PA Efficiency Analysis
In this section we provide four case studies of usage ofdescription of nonnegative waveforms with fundamental and119896th harmonic in PA efficiency analysis In first two casestudies to be presented in Section 71 voltage is nonnegativewaveform with fundamental and second harmonic with atleast one zero In remaining two case studies to be consideredin Section 72 voltage waveform contains fundamental andthird harmonic
i
v Load
Vdc
Idc
Lch
vL
iL
Cb
+ +in
Figure 15 Generic PA circuit diagram
Let us consider generic PA circuit diagram as shown inFigure 15 We assume here that voltage and current wave-forms at the transistor output are
V (120579) = 1 + 1198861V cos 120579 + 1198871V sin 120579 + 119886119896V cos 119896120579 + 119887119896V sin 119896120579
119894 (120579) = 1 + 1198861119894cos 120579 +
infin
sum
119899=2
119886119899119894cos 119899120579
(175)
where 120579 stands for 120596119905 Both waveforms are normalized in thesense that dc components of voltage and current are 119881dc =
1 and 119868dc = 1 respectively Under assumption that blockingcapacitor 119862
119887behaves as short-circuit at the fundamental and
higher harmonics current and voltage waveforms at the loadare
V119871(120579) = 119886
1V cos 120579 + 1198871V sin 120579 + 119886119896V cos 119896120579 + 119887119896V sin 119896120579
119894119871(120579) = minus119886
1119894cos 120579 minus
infin
sum
119899=2
119886119899119894cos 119899120579
(176)
In terms of coefficients of voltage and current waveforms theload impedance at fundamental harmonic is 119911
1= minus(119886
1V minus
1198951198871V)1198861119894 whereas load impedance at 119896th harmonic is 119911
119896=
minus(119886119896Vminus119895119887119896V)119886119896119894 All other harmonics are short-circuited (119911
119899=
0 for 119899 = 1 and 119899 = 119896) Time average output power of PA (egsee [10]) with waveform pair (175) at fundamental frequencycan be expressed as
1198751= minus
11988611198941198861V
2 (177)
For normalized waveforms (175) with 119881dc = 1 and 119868dc = 1dc power is 119875dc = 1 Consequently PA efficiency 120578 = 119875
1119875dc
(eg see [10 26]) is equal to
120578 = minus11988611198941198861V
2 (178)
Thus time average output power 1198751of PA with pair of nor-
malized waveform (175) is equal to efficiency (178)Power utilization factor (PUF) is defined [26] as ldquothe
ratio of power delivered in a given situation to the power
22 Mathematical Problems in Engineering
delivered by the same device with the same supply voltagein Class A moderdquo Since the output power in class-A modeis 1198751class-A = max[V(120579)] sdotmax[119894(120579)]8 (eg see [9]) it follows
that power utilization factor PUF = 11987511198751class-A for PA with
pair of normalized waveforms (175) can be expressed as
PUF =8120578
max [V (120579)] sdotmax [119894 (120579)] (179)
71 NonnegativeWaveforms for 119896 = 2 in PAEfficiencyAnalysisIn this subsection we provide two case studies of usage ofdescription of nonnegative waveforms with fundamental andsecond harmonic (119896 = 2) in PA efficiency analysis For moreexamples of usage of descriptions of nonnegative waveformswith fundamental and second harmonic in PA efficiencyanalysis see [12]
Case Study 71 In this case study we consider efficiency of PAfor given second harmonic impedance providing that voltageis nonnegative waveform with fundamental and second har-monic and current is ldquohalf-sinerdquo waveform frequently used inefficiency analysis of classical PA operation (eg see [10])
Standard model of current waveform for classical PAoperation has the form (eg see [10 26])
119894119863(120579) =
119868119863[cos 120579 minus cos(120572
2)] |120579| le
120572
2
0120572
2le |120579| le 120587
(180)
where 120572 is conduction angle and 119868119863
gt 0 Since 119894119863(120579) is
even function it immediately follows that its Fourier seriescontains only dc component and cosine terms
119894119863(120579) = 119868dc +
infin
sum
119899=1
119868119899cos 119899120579 (181)
The dc component of the waveform (180) is
119868dc =119868119863120572
2120587[sinc(120572
2) minus cos(120572
2)] (182)
where sinc119909 = (sin119909)119909 The coefficient of the fundamentalharmonic component reads
1198681=119868119863120572
2120587(1 minus sinc120572) (183)
and the coefficient of 119899th harmonic component can bewrittenin the form
119868119899=119868119863
119899120587[sin ((119899 minus 1) 1205722)
(119899 minus 1)minussin ((119899 + 1) 1205722)
(119899 + 1)] 119899 ge 2
(184)
For ldquohalf-sinerdquo current waveform conduction angle is equalto 120587 (class-B conduction angle) According to (182) thisfurther implies that 119868dc = 119868
119863120587 To obtain normalized form of
waveform (180) we set 119868dc = 1which implies that 119868119863= 120587 Fur-
thermore substitution of 120572 = 120587 and 119868119863= 120587 in (180) leads to
119894 (120579) =
120587 cos 120579 |120579| lt120587
2
0120587
2lt |120579| le 120587
(185)
Similarly substitution of 119868119863= 120587 and 120572 = 120587 into (183) and
(184) leads to the coefficients of waveform (185) Coefficientsof fundamental and second harmonic respectively are
1198861119894=120587
2 119886
2119894=2
3 (186)
On the other hand voltage waveform of type (35) for 119896 =2 reads
V (120579) = 1 + 1198861V cos 120579 + 1198871V sin 120579 + 1198862V cos 2120579 + 1198872V sin 2120579
(187)
This waveform contains only fundamental and second har-monic and therefore all harmonics of order higher thantwo are short-circuited (119911
119899= 0 for 119899 gt 2) For current
voltage pair (185) and (187) load impedance at fundamentalharmonic is 119911
1= minus(119886
1V minus 1198951198871V)1198861119894 whereas load impedance
at second harmonic is 1199112= minus(119886
2Vminus1198951198872V)1198862119894 According to ourassumption the load is passive and therefore Re119911
1 gt 0 and
Re1199112 ge 0 which further imply 119886
11198941198861V lt 0 and 119886
21198941198862V le 0
respectivelyIt is easy to see that problem of findingmaximal efficiency
of PA with current-voltage pair (185) and (187) for prescribedsecond harmonic impedance can be reduced to the problemof finding voltage waveform of type (187) with maximal coef-ficient |119886
1V| for prescribed coefficients of second harmonic(see Section 5)
The following algorithm (analogous to Algorithm 22presented in [12]) provides the procedure for calculation ofmaximal efficiency with current-voltage pair (185) and (187)for prescribed second harmonic impedance The definitionof function atan 2(119910 119909) which appears in the step (iii) of thefollowing algorithm is given by (105)
Algorithm 32 (i) Choose 1199112= 1199032+1198951199092such that |119911
2| le 1|119886
2119894|
(ii) calculate 1198862V minus 1198951198872V = minus119911
21198862119894and 120582
2V = radic11988622V + 1198872
2V(iii) if 2120582
2V le 1 minus 1198862V then calculate 119886
1V = minus1 minus 1198862V and
1198871V = minus2119887
2V else calculate 1205821V = radic81205822V(1 minus 1205822V) 1205790V minus 120585V2 =
(12)atan2(1198872V 1198862V) 1198861V = minus120582
1V cos(1205790V minus 120585V2) and 1198871V =
minus1205821V sin(1205790V minus 120585V2)(iv) calculate efficiency 120578 = minus119886
11198941198861V2
(v) calculate 1199111= minus(119886
1V minus 1198951198871V)1198861119894 and 1199112119899 = 1199112Re119911
1
In this case study coefficients of fundamental and secondharmonic of current waveform are given by (186) Maximalefficiency of PA associated with the waveform pair (185)and (187) as a function of normalized second harmonicimpedance 119911
2119899= 1199112Re119911
1 is presented in Figure 16(a)
As can be seen from Figure 16(a) efficiency of 078 isachieved at the edge of Smith chart where second harmonicimpedance has small resistive part Corresponding PUFcalculated according to (179) is presented in Figure 16(b)Peak efficiency 120578 = 1205874 = 07854 and peak value of PUF = 1
are attained when second harmonic is short-circuited (whichcorresponds to ideal class-B operation [10 26])
For example for second harmonic impedance 1199112= 01 minus
11989505 and currentwaveform (185) fromAlgorithm32 it followsthat 2120582
2V le 1 minus 1198862V Furthermore according to step (iii)
of above algorithm maximal efficiency of PA is attained
Mathematical Problems in Engineering 23
+j2
+j5
+j1
+j2
+j5
minusj2
minusj5
minusj1
minusj2
minusj5
infin0
078
02
05
1 2 5
075 07 065
06 05
075
07
065
06
05
120578 lt 05
(a)
099
095
091
083 075
067
0 59 051
051
099
095
091
083
075 0
67 059
+j5
+j1
+j2
+j5
minusj5
minusj1
minusj2
minusj5
infin
+j2
minusj2
0 02
05
1 2 5
120578 lt 05
(b)
Figure 16 (a) Contours of maximal efficiency of PA and (b) contours of corresponding PUF associated with the waveform pair (185) and(187) as functions of normalized second harmonic impedance 119911
2119899= 1199112Re119911
1
2
3
4
1
0
2 3 410
Wav
efor
ms
Angle 120579120587
i(120579)
(120579)
Figure 17 Waveform pair (185) and (187) that provides maximalefficiency for 119911
2= 01 minus 11989505
with voltage waveform of type (187) with coefficients 1198862V =
minus00667 1198872V = minus03333 119886
1V = minus09333 and 1198871V = 06667 (see
Figure 17) Corresponding efficiency PUF and normalizedsecond harmonic impedance are 120578 = 07330 PUF = 07572and 1199112119899= 01683 minus 11989508415 respectively
On the other hand for second harmonic impedance 1199112=
01 minus 11989508 and current waveform (185) from Algorithm 32 itfollows that 2120582
2V gt 1 minus 1198862V Then according to step (iii) of
above algorithm maximal efficiency is attained with voltagewaveform of type (187) with coefficients 119886
2V = minus00667 1198872V =
minus05333 1198861V = minus09333 and 119887
1V = 10572 (see Figure 18)Efficiency PUF andnormalized secondharmonic impedanceare 120578 = 07330 PUF = 06332 and 119911
2119899= 01683 minus 11989513465
respectively
2
3
4
1
0
2 3 410
Wav
efor
ms
Angle 120579120587
i(120579)
(120579)
Figure 18 Waveform pair (185) and (187) that provides maximalefficiency for 119911
2= 01 minus 11989508
Case Study 72 As another case study let us consider the effi-ciency of PA providing that current waveform is nonnegativecosine waveform up to third harmonic with maximum valueof amplitude of fundamental harmonic [22] (see also [8])
119894 (120579) = 1 +1 + radic5
2cos 120579 + 2radic5
5cos 2120579 + 5 minus radic5
10cos 3120579
(188)
and voltage waveform is nonnegative waveform of type(187) Load impedances at fundamental second and thirdharmonic are 119911
1= minus(119886
1Vminus1198951198871V)1198861119894 1199112 = minus(1198862Vminus1198951198872V)1198862119894 and
1199113= 0 respectively According to our assumption the load
is passive and therefore Re1199111 gt 0 and Re119911
2 ge 0 which
further imply 11988611198941198861V lt 0 and 119886
21198941198862V le 0 respectively
Because current waveform (188) contains only cosineterms and voltage waveform is the same as in previous case
24 Mathematical Problems in Engineering
08
075
07
06
06
065
065
05
05
075 0
7
120578 lt 05
+j2
+j5
+j1
+j2
+j5
minusj2
minusj5
minusj1
minusj2
minusj5
infin0
02
05
1 2 5
(a)
085
08 0
75 07 0
6
05 04
08
075
0405
060
7
120578 lt 05
+j2
+j5
+j1
+j2
+j5
minusj2
minusj5
minusj1
minusj2
minusj5
infin0 02
05
1 2 5
(b)
Figure 19 (a) Contours of maximal efficiency of PA and (b) contours of corresponding PUF associated with the waveform pair (187)-(188)as functions of normalized second harmonic impedance 119911
2119899= 1199112Re119911
1
study the procedure for calculation of maximal efficiency ofPA with waveform pair (187)-(188) is the same as presentedin Algorithm 32 In this case study the coefficients of funda-mental and second harmonic of current waveform are 119886
1119894=
(1 + radic5)2 and 1198862119894= 2radic55 respectively
Maximal efficiency of PA associated with the waveformpair (187)-(188) as a function of normalized secondharmonicimpedance 119911
2119899= 1199112Re119911
1 is presented in Figure 19(a)
Efficiency of 08 is achieved at the edge of Smith chart wheresecond harmonic impedance has small resistive part Thetheoretical upper bound 120578 = (1 + radic5)4 asymp 08090 isattained when second harmonic is short-circuitedWhen thisupper bound is reached both second and third harmonic areshort-circuited which implies that we are dealing with finiteharmonic class-C [6 8] or dually when current and voltageinterchange their roles with finite harmonic inverse class-C[6 9] Corresponding PUF calculated according to (179) ispresented in Figure 19(b) Peak value of PUF asymp 08541 isattained when second harmonic is short-circuited
For example for second harmonic impedance 1199112
=
007 minus 11989504 and current waveform (188) from Algorithm 32 itfollows that 2120582
2V le 1 minus 1198862V Furthermore according to step
(iii) of Algorithm 32 maximal efficiency of PA is attainedwith voltage waveform of type (187) with coefficients 119886
2V =
minus00626 1198872V = minus03578 119886
1V = minus09374 and 1198871V = 07155 (see
Figure 20) Corresponding efficiency PUF and normalizedsecond harmonic impedance are 120578 = 07584 PUF = 06337and 1199112119899= 01208 minus 11989506904 respectively
On the other hand for 1199112= 005 minus 11989507 and current
waveform (187) it follows that 21205822V gt 1minus119886
2VThen accordingto step (iii) of Algorithm 32 themaximal efficiency is attainedwith voltage waveform of type (187) with coefficients 119886
2V =
minus00447 1198872V = minus06261 119886
1V = minus09318 and 1198871V = 10007 (see
Figure 21) Efficiency PUF and normalized second harmonic
2
3
4
1
0
2 3 410
Wav
efor
ms
Angle 120579120587
i(120579)
(120579)
Figure 20 Waveform pair (187)-(188) that provides maximal effi-ciency for 119911
2= 007 minus 11989504
impedance are 120578 = 07538 PUF = 05314 and 1199112119899= 00868minus
11989512156 respectively
72 Nonnegative Waveforms for 119896 = 3 in PA EfficiencyAnalysis In this subsection we provide another two casestudies of usage of description of nonnegative waveforms inPA efficiency analysis this time with fundamental and thirdharmonic (119896 = 3)
Case Study 73 Let us consider current-voltage pair such thatvoltage is nonnegative waveform with fundamental and thirdharmonic
V (120579) = 1 + 1198861V cos 120579 + 1198871V sin 120579 + 1198863V cos 3120579 + 1198873V sin 3120579
(189)
Mathematical Problems in Engineering 25
2
3
4
1
0
2 3 410
Wav
efor
ms
Angle 120579120587
i(120579)
(120579)
Figure 21 Waveform pair (187)-(188) that provides maximal effi-ciency for 119911
2= 005 minus 11989507
and current is nonnegative cosine waveform given by (188)Load impedances at fundamental second and third har-monic are 119911
1= minus(119886
1V minus 1198951198871V)1198861119894 1199112 = 0 and 119911
3= minus(119886
3V minus
1198951198873V)1198863119894 respectively According to our assumption the load
is passive and therefore Re1199111 gt 0 and Re119911
3 ge 0 which
further imply 11988611198941198861V lt 0 and 119886
31198941198863V le 0
In this subsection we consider the problem of findingmaximal efficiency of PA with waveform pair (188)-(189)for given third harmonic impedance As we mentionedearlier problem of finding maximal efficiency of PA withcurrent-voltage pair (188)-(189) for prescribed third har-monic impedance can be reduced to the problem of findingvoltage waveform of type (189) withmaximal coefficient |119886
1V|for prescribed coefficients of third harmonic (see Section 52)
The following algorithm provides the procedure forcalculation of maximal efficiency with current-voltage pair(188)-(189) The definition of function atan 2(119910 119909) whichappears in step (iii) of the following algorithm is given by(105)
Algorithm 33 (i) Choose 1199113= 1199033+1198951199093such that |119911
3| le 1|119886
3119894|
(ii) calculate 1198863V minus 1198951198873V = minus119911
31198863119894and 120582
3V = radic11988623V + 1198872
3V(iii) if 271205822
3V le (1 minus 21198863V)3 then calculate 119886
1V = minus1 minus 1198863V
and 1198871V = minus3119887
3V else calculate 1205821V = 3(3radic1205823V minus 120582
3V) 1205790V minus120585V3 = (13)atan 2(119887
3V 1198863V) 1198861V = minus1205821V cos(1205790V minus 120585V3) and
1198871V = minus120582
1V sin(1205790V minus 120585V3)(iv) calculate efficiency 120578 = minus119886
11198941198861V2
(v) calculate 1199111= minus(119886
1V minus 1198951198871V)1198861119894 and 1199113119899 = 1199113Re119911
1
In this case study coefficients of fundamental and thirdharmonic of current waveform are 119886
1119894= (1 + radic5)2 and
1198863119894= (5 minus radic5)10 respectively For the waveform pair (188)-
(189) maximal efficiency of PA as a function of normalizedthird harmonic impedance 119911
3119899= 1199113Re119911
1 is presented in
Figure 22 Efficiency of 08 is reached when third harmonicimpedance has small resistive part Peak efficiency 120578 = (1 +
radic5)4 asymp 08090 is achieved when third harmonic is short-circuited
+j2
+j5
+j1
+j2
+j5
minusj2
minusj5
minusj1
minusj2
minusj5
infin0
08
08
075
075
07
07
06
05
02
05
1 2
Figure 22 Contours of maximal efficiency of PA associated withthe waveform pair (188)-(189) as a function of normalized thirdharmonic impedance 119911
3119899= 1199113Re119911
1
For the present case study in what follows we show thatpower utilization factor is proportional to efficiency For volt-age waveform of type (189) it is easy to see that V(120579 + 120587) =
2minusV(120579) holdsThis relation along with the fact that waveformV(120579) that provides maximal efficiency has at least one zeroimplies that max[V(120579)] = 2 On the other hand currentwaveform (188) is cosine waveform with positive coefficientsand therefore max[119894(120579)] = 119894(0) = 2 + 4radic5 Consequentlyaccording to (179) the following relation holds
PUFCase study 73 = 2 (5 minus 2radic5) 120578 = 10557120578 (190)
Clearly the ratio PUF120578 is constant and therefore in this casestudy PUF can be easily calculated from the correspondingefficiency Accordingly peak efficiency and peak value ofPUFCase study 73 = 3radic52 minus 52 = 08541 are attained forthe same voltage waveform (when third harmonic is short-circuited)
In the first example current waveform (188) and 1199113=
02 minus 11989505 imply that 2712058223V le (1 minus 2119886
3V)3 Then according
to Algorithm 33 the voltage waveform of type (189) thatprovides maximal efficiency has the following coefficients1198863V = minus00553 119887
3V = minus01382 1198861V = minus09447 and 119887
1V =
04146 (see Figure 23) Efficiency PUF and normalized thirdharmonic impedance are 120578 = 07643 PUF = 08069 and1199113119899= 03425 minus 11989508564 respectivelyIn the second example current waveform (188) and 119911
3=
01 minus 11989511 imply that 2712058223V gt (1 minus 2119886
3V)3 Then according
to Algorithm 33 the voltage waveform of type (189) thatprovides maximal efficiency has the following coefficients1198863V = minus00276 119887
3V = minus03040 1198861V = minus09391 and 119887
1V =
05807 (see Figure 24) Efficiency PUF and normalized thirdharmonic impedance are 120578 = 07598 PUF = 08021 and1199113119899= 01723 minus 11989518952 respectively
26 Mathematical Problems in Engineering
2
3
4
1
0
2 3 410
Wav
efor
ms
Angle 120579120587
i(120579)
(120579)
Figure 23 Waveform pair (188)-(189) that provides maximal effi-ciency for 119911
3= 02 minus 11989505
2
3
4
1
0
2 3 410
Wav
efor
ms
Angle 120579120587
i(120579)
(120579)
Figure 24 Waveform pair (188)-(189) that provides maximal effi-ciency for 119911
3= 01 minus 11989511
Case Study 74 In this case study let us consider current-voltage pair where current is normalized waveform of type(180) with conduction angle 120572 = 115120587 (207∘) and voltageis nonnegative waveform of type (189) Substitution of 120572 =
115120587 and 119868dc = 1 into (182) leads to 119868119863
= 22535Furthermore substitution of 120572 = 115120587 and 119868
119863= 22535 into
(180) leads to
119894 (120579)
=
22535 [cos 120579 minus cos(1151205872
)] |120579| le115120587
2
0115120587
2le |120579| le 120587
(191)
Similarly substitution of 120572 = 115120587 and 119868119863
= 22535 into(183) and (184) for 119899 = 3 yields coefficients of fundamentaland third harmonic of waveform (191)
1198861119894= 14586 119886
3119894= minus01026 (192)
Because current waveform (191) contains only cosineterms and voltage waveform is the same as in previous case
+j2
+j5
+j1
+j2
+j5
minusj2
minusj5
minusj1
minusj2
minusj5
infin0 02
05
1 2
074
076
08
084
082
078
076
074
Figure 25 Contours of maximal efficiency of PA associated withthe waveform pair (189) and (191) as a function of normalized thirdharmonic impedance 119911
3119899= 1199113Re119911
1
study the procedure for calculation of maximal efficiencyof PA with waveform pair (189)ndash(191) is the same as thatpresented in Algorithm 33 In this case study the coefficientsof fundamental and third harmonic of current waveform aregiven by (192)
For the waveform pair (189) and (191) maximal efficiencyof PA as a function of normalized third harmonic impedance1199113119899= 1199113Re119911
1 is presented in Figure 25 Efficiency of 084
is obtained in vicinity of 1199113119899= 23685 (corresponding to 119911
3=
18750) Peak efficiency 120578 asymp 08421 is achieved for voltagewaveform of type (189) with coefficients 119886
1V = minus2radic3 1198863V =
radic39 and 1198871V = 1198873V = 0
In the course of finding power utilization factor noticethat currentwaveformof type (191) attains itsmaximumvaluefor 120579 = 0 Insertion of max[119894(120579)] = 119894(0) = 278 andmax[V(120579)] = 2 for voltage waveform of type (189) into (179)leads to
PUFCase study 74 = 1439120578 (193)
Again the ratio PUF120578 is constant and PUF can be easilycalculated from the corresponding efficiency Accordinglypeak value of PUFCase study 74 asymp 12118 and peak efficiencyare attained for the same voltage waveform
In the first example current waveform (191) and 1199113=
1 minus 11989502 imply that 2712058223V le (1 minus 2119886
3V)3 Then according to
Algorithm 33 voltage waveform of type (189) which providesmaximal efficiency has coefficients 119886
3V = 01026 1198873V =
00205 1198861V = minus11026 and 119887
1V = minus00616 (see Figure 26)Efficiency PUF and normalized third harmonic impedanceare 120578 = 08042 PUF = 11572 and 119911
3119899= 13228 minus 11989502646
respectivelyIn second example current waveform (191) and 119911
3=
15 minus 11989512 imply that 2712058223V gt (1 minus 2119886
3V)3 Then according to
Algorithm 33 voltage waveform of type (189) which providesmaximal efficiency has coefficients 119886
3V = 01540 1198873V =
Mathematical Problems in Engineering 27
2
3
4
1
0
2 3 410
Wav
efor
ms
Angle 120579120587
i(120579)
(120579)
Figure 26 Waveform pair (189) and (191) that provides maximalefficiency for 119911
3= 1 minus 11989502
2
3
4
1
0
2 3 410
Wav
efor
ms
Angle 120579120587
i(120579)
(120579)
Figure 27 Waveform pair (189) and (191) that provides maximalefficiency for 119911
3= 15 minus 11989512
01232 1198861V = minus11255 and 119887
1V = minus02575 (see Figure 27)Efficiency PUF and normalized third harmonic impedanceare 120578 = 08208 PUF = 11812 and 119911
3119899= 19439 minus 11989515552
respectively
8 Conclusion
In this paper we consider a problem of finding general de-scriptions of various classes of nonnegative waveforms withfundamental and 119896th harmonic These classes include non-negative waveforms with at least one zero nonnegative wave-forms with maximal amplitude of fundamental harmonic forprescribed amplitude of 119896th harmonic nonnegative wave-forms withmaximal coefficient of cosine part of fundamentalharmonic for prescribed coefficients of 119896th harmonic andnonnegative cosine waveforms with at least one zero Mainresults are stated in six propositions (Propositions 1 6 918 22 and 26) four corollaries (Corollaries 2ndash5) twentyremarks and three algorithms Four case studies of usageof closed form descriptions of nonnegative waveforms in PAefficiency analysis are considered in detail in Section 7
Appendices
Here we provide a list of finite sums of trigonometric func-tions used in this paper (Appendix A) and brief account ofthe Chebyshev polynomials (Appendix B)
A List of Some Finite Sums ofTrigonometric Functions
Dirichlet kernel (eg see [27]) is as follows
119863119896minus1
(120591) = 1 + 2
119896minus1
sum
119899=1
cos 119899120591 = sin ((2119896 minus 1) 1205912)sin (1205912)
(A1)
Fejer kernel (eg see [27]) can be expressed in the fol-lowing equivalent forms
119865119896minus1
(120591) =1
119896
119896minus1
sum
119899=0
119863119899(120591) = 1 +
2
119896
119896minus1
sum
119899=1
(119896 minus 119899) cos 119899120591
=(1 minus cos 119896120591)119896 (1 minus cos 120591)
(A2)
Lagrangersquos trigonometric identity (eg see [28]) is as fol-lows
1198781(120591) =
119896minus1
sum
119899=1
sin 119899120591 = sin (1198961205912) sin ((119896 minus 1) 1205912)sin (1205912)
(A3)
In what follows we show that the following three trigono-metric identities also hold
2
119896minus1
sum
119899=1
(119896 minus 119899) sin 119899120591 = 119896 sin 120591 minus sin 1198961205911 minus cos 120591
(A4)
119896minus1
sum
119899=1
cos (119896 minus 2119899) 120591 = sin (119896 minus 1) 120591sin 120591
(A5)
119896minus1
sum
119899=1
119899 (119896 minus 119899) cos (119896 minus 2119899) 120591
=sin (119896120591) cos 120591 minus 119896 cos (119896120591) sin 120591
2sin3120591
(A6)
Denote 1198782(120591) = 2sum
119896minus1
119899=1(119896 minus 119899) sin 119899120591 119878
3(120591) = sum
119896minus1
119899=1cos(119896 minus
2119899)120591 and 1198784(120591) = sum
119896minus1
119899=1119899(119896 minus 119899) cos(119896 minus 2119899)120591
Notice that 1198782(120591) = 2119896119878
1(120591) + 119889119863
119896minus1(120591)119889120591 which
immediately leads to (A4)
28 Mathematical Problems in Engineering
Identity (A5) can be obtained as follows
sin (119896 minus 1) 120591sin 120591
=119890119895(119896minus1)120591
minus 119890minus119895(119896minus1)120591
119890119895120591 minus 119890minus119895120591
= 119890119895119896120591
119890minus2119895120591
minus 119890minus2119895(119896minus1)120591
1 minus 119890minus2119895120591
= 119890119895119896120591
119896minus1
sum
119899=1
119890minus2119895119899120591
=
119896minus1
sum
119899=1
119890119895(119896minus2119899)120591
=
119896minus1
sum
119899=1
cos (119896 minus 2119899) 120591
(A7)
From 4119899(119896 minus 119899) = 1198962
minus (119896 minus 2119899)2 it follows that 4119878
4(120591) =
1198962
1198783(120591) + 119889
2
1198783(120591)119889120591
2 which leads to (A6)
B The Chebyshev Polynomials
The Chebyshev polynomials of the first kind 119881119899(119909) can be
defined by the following relation (eg see [29])
119881119899(119909) = cos 119899120591 when 119909 = cos 120591 (B1)
The Chebyshev polynomials of the second kind 119880119899(119909) can be
defined by the following relation (eg see [29])
119880119899(119909) =
sin (119899 + 1) 120591sin 120591
when 119909 = cos 120591 (B2)
The Chebyshev polynomials satisfy the following recur-rence relations (eg see [29])
1198810(119909) = 1 119881
1(119909) = 119909
119881119899+1
(119909) = 2119909119881119899(119909) minus 119881
119899minus1(119909)
1198800(119909) = 1 119880
1(119909) = 2119909
119880119899+1
(119909) = 2119909119880119899(119909) minus 119880
119899minus1(119909)
(B3)
The first few Chebyshev polynomials of the first and secondkind are 119881
2(119909) = 2119909
2
minus 1 1198813(119909) = 4119909
3
minus 3119909 1198814(119909) = 8119909
4
minus
81199092
+ 1 1198802(119909) = 4119909
2
minus 1 1198803(119909) = 8119909
3
minus 4119909 and 1198804(119909) =
161199094
minus 121199092
+ 1
Conflict of Interests
The authors declare that there is no conflict of interestsregarding the publication of this paper
Acknowledgment
This work is supported by the SerbianMinistry of EducationScience and Technology Development as a part of ProjectTP32016
References
[1] V I Arnolrsquod V S Afrajmovich Y S Ilrsquoyashenko and L PShilrsquonikov Dynamical Systems V Bifurcation Theory and Catas-trophe Theory Springer Berlin Germany 1994
[2] E Polak ldquoOn the mathematical foundations of nondifferen-tiable optimization in engineering designrdquo SIAM Review vol29 no 1 pp 21ndash89 1987
[3] N S Fuzik ldquoBiharmonic modes of a tuned RF power amplifierrdquoRadiotehnika vol 25 no 7 pp 62ndash71 1970 (Russian)
[4] P Colantonio F Giannini G Leuzzi and E Limiti ldquoClass Gapproach for low-voltage high-efficiency PA designrdquo Interna-tional Journal of RF and Microwave Computer-Aided Engineer-ing vol 10 no 6 pp 366ndash378 2000
[5] F H Raab ldquoMaximum efficiency and output of class-F poweramplifiersrdquo IEEE Transactions on Microwave Theory and Tech-niques vol 49 no 6 pp 1162ndash1166 2001
[6] FH Raab ldquoClass-E class-C and class-F power amplifiers basedupon a finite number of harmonicsrdquo IEEE Transactions onMicrowaveTheory and Techniques vol 49 no 8 pp 1462ndash14682001
[7] J D Rhodes ldquoOutput universality inmaximum efficiency linearpower amplifiersrdquo International Journal of Circuit Theory andApplications vol 31 no 4 pp 385ndash405 2003
[8] A Juhas and L A Novak ldquoComments on lsquoClass-E class-Cand classF power amplifier based upon a finite number ofharmonicsrsquordquo IEEE Transactions on Microwave Theory and Tech-niques vol 57 no 6 pp 1623ndash1625 2009
[9] M Roberg and Z Popovic ldquoAnalysis of high-efficiency poweramplifiers with arbitrary output harmonic terminationsrdquo IEEETransactions on Microwave Theory and Techniques vol 59 no8 pp 2037ndash2048 2011
[10] A Grebennikov N O Sokal and M J Franco Switchmode RFPower Amplifiers ElsevierAcademic Press San Diego CalifUSA 2nd edition 2012
[11] T Canning P J Tasker and S C Cripps ldquoContinuous modepower amplifier design using harmonic clipping contourstheory and practicerdquo IEEE Transactions on Microwave Theoryand Techniques vol 62 no 1 pp 100ndash110 2014
[12] A Juhas and L A Novak ldquoGeneral description of nonnegativewaveforms up to second harmonic for power amplifier mod-ellingrdquoMathematical Problems in Engineering vol 2014 ArticleID 709762 18 pages 2014
[13] V I Arnolrsquod V V Goryunov O V Lyashko and V A VasilrsquoevDynamical Systems VIIImdashSingularity Theory II ApplicationsSpringer Berlin Germany 1993
[14] D Siersma ldquoProperties of conflict sets in the planerdquo BanachCenter Publications Polish Academy of Sciences vol 50 no 1 pp267ndash276 1999 Proceedings of the Banach Center Symposiumon Geometry and Topology of Caustics (Caustics rsquo98) WarsawPoland
[15] M van Manen The geometry of conflict sets [Dissertation]Universiteit Utrecht UtrechtThe Netherlands 2003 httpigi-tur-archivelibraryuunldissertations2003-0912-123058c4pdf
[16] Y L Sachkov ldquoMaxwell strata and symmetries in the problemofoptimal rolling of a sphere over a planerdquo Sbornik Mathematicsvol 201 no 7-8 pp 1029ndash1051 2010
[17] I A Bogaevsky ldquoPerestroikas of shocks and singularities ofminimum functionsrdquoPhysicaDNonlinear Phenomena vol 173no 1-2 pp 1ndash28 2002
[18] Y L Sachkov ldquoMaxwell strata in the Euler elastic problemrdquoJournal of Dynamical andControl Systems vol 14 no 2 pp 169ndash234 2008
[19] M Siino and T Koike ldquoTopological classification of black holesgeneric Maxwell set and crease set of a horizonrdquo InternationalJournal ofModern Physics D Gravitation Astrophysics Cosmol-ogy vol 20 no 6 pp 1095ndash1122 2011
Mathematical Problems in Engineering 29
[20] F H Raab ldquoClass-F power amplifiers with maximally flat wave-formsrdquo IEEETransactions onMicrowaveTheory and Techniquesvol 45 no 11 pp 2007ndash2012 1997
[21] A Juhas and L A Novak ldquoMaximally flat waveforms with finitenumber of harmonics in class-F power amplifiersrdquo Mathemat-ical Problems in Engineering vol 2013 Article ID 169590 9pages 2013
[22] L Fejer ldquoUber trigonometrische polynomerdquo Journal fur dieReine und Angewandte Mathematik vol 1916 no 146 pp 53ndash82 1916 (German)
[23] S C Cripps ldquoBessel Waives [microwave bytes]rdquo IEEE Micro-wave Magazine vol 10 no 7 pp 30ndash36 117 2009
[24] L N Bryzgalova ldquoSingularities of the maximum of para-metrically dependent functionrdquo Functional Analysis and ItsApplications vol 11 no 1 pp 49ndash51 1977
[25] V I Arnold A A Davydov V A Vassiliev and V MZakalyukin Mathematical Models of Catastrophes Control ofCatastrophic Process Encyclopedia of Life Support Systems(EOLSS) EOLSS Publishers Oxford UK 2006
[26] S C Cripps RF Power Amplifiers for Wireless CommunicationsArtech House Norwood Mass USA 2nd edition 2006
[27] A Zygmund Trigonometric Series vol 1 Cambridge UniversityPress Cambridge UK 2nd edition 1959
[28] A Jeffrey and H DaiHandbook of Mathematical Formulas andIntegrals ElsevierAcademic Press San Diego Calif USA 4thedition 2008
[29] J C Mason and D C Handscomb Chebyshev PolynomialsChapman amp Hall CRC Press Boca Raton Fla USA 2003
Submit your manuscripts athttpwwwhindawicom
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical Problems in Engineering
Hindawi Publishing Corporationhttpwwwhindawicom
Differential EquationsInternational Journal of
Volume 2014
Applied MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Probability and StatisticsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical PhysicsAdvances in
Complex AnalysisJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
OptimizationJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
CombinatoricsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Operations ResearchAdvances in
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Function Spaces
Abstract and Applied AnalysisHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of Mathematics and Mathematical Sciences
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Algebra
Discrete Dynamics in Nature and Society
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Decision SciencesAdvances in
Discrete MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom
Volume 2014 Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Stochastic AnalysisInternational Journal of
Mathematical Problems in Engineering 9
Also maximally flat nonnegative waveforms (they haveonly one zero [21]) can be obtained from (66)ndash(68) by setting120585 = 0 Thus substitution of 120585 = 0 into (66)ndash(68) leads tothe following form of maximally flat nonnegative waveformof type (35)
119879119896(120591) =
[1 minus cos (120591 minus 1205910)]2
3 (1198962 minus 1)
sdot [119896 (1198962
minus 1)
+ 2
119896minus2
sum
119899=1
(119896 minus 119899) ((119896 minus 119899)2
minus 1) cos 119899 (120591 minus 1205910)]
(70)
Maximally flat nonnegative waveforms of type (35) for 119896 le 4
can be expressed as
1198792(120591) =
2
3[1 minus cos(120591 minus 120591
0)]2
1198793(120591) =
1
2[1 minus cos(120591 minus 120591
0)]2
[2 + cos (120591 minus 1205910)]
1198794(120591) =
4
15[1 minus cos (120591 minus 120591
0)]2
sdot [5 + 4 cos (120591 minus 1205910) + cos 2 (120591 minus 120591
0)]
(71)
Remark 12 Every nonnegative waveform of type (35) withexactly one zero at nondegenerate critical point can bedescribed as in Proposition 6 providing that symbol ldquolerdquoin relation (40) is replaced with ldquoltrdquo This is an immediateconsequence of Propositions 6 and 9 and Remark 11
Remark 13 Identity [1minus cos(120591minus1205910)][1minus cos(120591minus120591
0+2120585119896)] =
[cos 120585119896 minus cos(120591 minus 1205910+ 120585119896)]
2 implies that (66) can be alsorewritten as
119879119896(120591) = 120582
119896[cos 120585119896
minus cos(120591 minus 1205910+120585
119896)]
2
sdot [1198880+ 2
119896minus2
sum
119899=1
119888119899cos 119899(120591 minus 120591
0+120585
119896)]
(72)
Furthermore substitution of (67) into (72) leads to
119879119896(120591)
= 120582119896[cos 120585119896
minus cos(120591 minus 1205910+120585
119896)]
sdot [(119896 minus 1) sin 120585sin (120585119896)
minus 2
119896minus1
sum
119899=1
sin (120585 minus 119899120585119896)sin (120585119896)
cos 119899(120591 minus 1205910+120585
119896)]
(73)
Remark 14 According to (A6) (see Appendices) it followsthat coefficients (67) can be expressed as
119888119899= 2
119896minus119899minus1
sum
119898=1
119898(119896 minus 119899 minus 119898) cos((119896 minus 119899 minus 2119898) 120585119896
) (74)
Furthermore from (74) it follows that coefficients 119888119896minus2
119888119896minus3
119888119896minus4
and 119888119896minus5
are equal to119888119896minus2
= 2 (75)
119888119896minus3
= 8 cos(120585119896) (76)
119888119896minus4
= 8 + 12 cos(2120585119896) (77)
119888119896minus5
= 24 cos(120585119896) + 16 cos(3120585
119896) (78)
For example for 119896 = 2 (75) and (68) lead to 1198880= 2 and
1205822= 1(2+cos 120585) respectively which from (72) further imply
that
1198792(120591) =
2 [cos(1205852) minus cos(120591 minus 1205910+ 1205852)]
2
[2 + cos 120585] (79)
Also for 119896 = 3 (75) (76) and (68) lead to 1198881= 2 119888
0=
8 cos(1205853) and 1205823= [2(3 cos(1205853) + cos 120585)]minus1 respectively
which from (72) further imply that
1198793(120591) =
2 [cos (1205853) minus cos (120591 minus 1205910+ 1205853)]
2
[3 cos (1205853) + cos 120585]
sdot [2 cos(1205853) + cos(120591 minus 120591
0+120585
3)]
(80)
Remark 15 According to (A5) (see Appendices) relation(68) can be rewritten as
120582119896= [(119896 minus 1) cos 120585 + 119896
119896minus1
sum
119899=1
cos((119896 minus 2119899)120585119896
)]
minus1
(81)
Clearly amplitude 120582119896of 119896th harmonic of nonnegative wave-
form of type (35) with exactly two zeros is even functionof 120585 Since cos((119896 minus 2119899)120585119896) 119899 = 0 (119896 minus 1) decreaseswith increase of |120585| on interval 0 le |120585| le 120587 it follows that120582119896monotonically increases with increase of |120585| Right hand
side of (68) is equal to 1(1198962
minus 1) for 120585 = 0 and to one for|120585| = 120587 Therefore for nonnegative waveforms of type (35)with exactly two zeros the following relation holds
1
1198962 minus 1lt 120582119896lt 1 (82)
The left boundary in (82) corresponds to maximally flatnonnegative waveforms (see Remark 11) The right boundaryin (82) corresponds to nonnegative waveforms with 119896 zeros(also see Remark 11)
Amplitude of 119896th harmonic of nonnegative waveform oftype (35) with two zeros as a function of parameter 120585 for 119896 le5 is presented in Figure 5
Remark 16 Nonnegative waveform of type (35) with twozeros can be also expressed in the following form
119879119896(120591) = 1 minus 120582
119896
119896 sin 120585sin (120585119896)
cos(120591 minus 1205910+120585
119896)
+ 120582119896cos (119896 (120591 minus 120591
0) + 120585)
(83)
10 Mathematical Problems in Engineering
1
08
06
04
02
0minus1 minus05 0 05
Am
plitu
de120582k
1
Parameter 120585120587
k = 2k = 3
k = 4k = 5
Figure 5 Amplitude of 119896th harmonic of nonnegative waveformwith two zeros as a function of parameter 120585
where 120582119896is given by (68) and 0 lt |120585| lt 120587 From (83) it follows
that coefficients of fundamental harmonic of nonnegativewaveform of type (35) with two zeros are
1198861= minus1205821cos(120591
0minus120585
119896) 119887
1= minus1205821sin(120591
0minus120585
119896) (84)
where 1205821is amplitude of fundamental harmonic
1205821=
119896 sin 120585sin (120585119896)
120582119896 (85)
Coefficients of 119896th harmonic are given by (45)-(46)Notice that (68) can be rewritten as
120582119896= [cos(120585
119896)
119896 sin 120585sin(120585119896)
minus cos 120585]minus1
(86)
By introducing new variable
119909 = cos(120585119896) (87)
and using the Chebyshev polynomials (eg see Appendices)relations (85) and (86) can be rewritten as
1205821= 119896120582119896119880119896minus1
(119909) (88)
120582119896=
1
119896119909119880119896minus1
(119909) minus 119881119896(119909)
(89)
where119881119896(119909) and119880
119896(119909) denote the Chebyshev polynomials of
the first and second kind respectively From (89) it followsthat
120582119896[119896119909119880119896minus1
(119909) minus 119881119896(119909)] minus 1 = 0 (90)
which is polynomial equation of 119896th degree in terms of var-iable 119909 From 0 lt |120585| lt 120587 and (87) it follows that
cos(120587119896) lt 119909 lt 1 (91)
Since 120582119896is monotonically increasing function of |120585| 0 lt |120585| lt
120587 it follows that 120582119896is monotonically decreasing function of
119909 This further implies that (90) has only one solution thatsatisfies (91) (For 119896 = 2 expression (91) reads cos(1205872) le
119909 lt 1) This solution for 119909 (which can be obtained at leastnumerically) according to (88) leads to amplitude 120582
1of
fundamental harmonicFor 119896 le 4 solutions of (90) and (91) are
119909 = radic1 minus 1205822
21205822
1
3lt 1205822le 1
119909 =1
23radic1205823
1
8lt 1205823lt 1
119909 = radic1
6(1 + radic
51205824+ 3
21205824
)1
15lt 1205824lt 1
(92)
Insertion of (92) into (88) leads to the following relationsbetween amplitude 120582
1of fundamental and amplitude 120582
119896of
119896th harmonic 119896 le 4
1205821= radic8120582
2(1 minus 120582
2)
1
3lt 1205822le 1 (93)
1205821= 3 (
3radic1205823minus 1205823)
1
8lt 1205823lt 1 (94)
1205821= radic
32
27(radic2120582
4(3 + 5120582
4)3
minus 21205824(9 + 7120582
4))
1
15lt 1205824lt 1
(95)
Proof of Proposition 9 As it has been shown earlier (seeProposition 6) nonnegative waveform of type (35) with atleast one zero can be represented in form (38) Since weexclude nonnegative waveforms with 120582
119896= 1 according to
Remark 7 it follows that we exclude case |120585| = 120587Therefore inthe quest for nonnegative waveforms of type (35) having twozeros we will start with waveforms of type (38) for |120585| lt 120587It is clear that nonnegative waveforms of type (38) have twozeros if and only if
120582119896= [max120591
119903119896(120591)]minus1
(96)
and max120591119903119896(120591) = 119903
119896(1205910) According to (64) max
120591119903119896(120591) =
119903119896(1205910) implies |120585| = 0 Therefore it is sufficient to consider
only the interval (69)Substituting (96) into (38) we obtain
119879119896(120591) =
[1 minus cos (120591 minus 1205910)] [max
120591119903119896(120591) minus 119903
119896(120591)]
max120591119903119896(120591)
(97)
Mathematical Problems in Engineering 11
Expression max120591119903119896(120591) minus 119903
119896(120591) according to (64) and (39)
equals
max120591
119903119896(120591) minus 119903
119896(120591) = 119896
sin ((119896 minus 1) 120585119896)sin (120585119896)
minus 2
119896minus1
sum
119899=1
(119896 minus 119899) cos (119899 (120591 minus 1205910) + 120585)
(98)
Comparison of (97) with (66) yields
max120591
119903119896(120591) minus 119903
119896(120591) = [1 minus cos(120591 minus 120591
0+2120585
119896)]
sdot [1198880+ 2
119896minus2
sum
119899=1
119888119899cos 119899(120591 minus 120591
0+120585
119896)]
(99)
where coefficients 119888119899 119899 = 0 119896 minus 2 are given by (67) In
what follows we are going to show that right hand sides of(98) and (99) are equal
From (67) it follows that
1198880minus 1198881cos(120585
119896) = 119896
sin (120585 minus 120585119896)sin (120585119896)
(100)
Also from (67) for 119899 = 1 119896minus3 it follows that the followingrelations hold
(119888119899minus1
+ 119888119899+1
) cos(120585119896) minus 2119888
119899= 2 (119896 minus 119899) cos(120585 minus 119899120585
119896)
(119888119899minus1
minus 119888119899+1
) sin(120585119896) = 2 (119896 minus 119899) sin(120585 minus 119899120585
119896)
(101)
From (99) by using (75) (76) (100)-(101) and trigonometricidentities
cos(120591 minus 1205910+2120585
119896) = cos(120585
119896) cos(120591 minus 120591
0+120585
119896)
minus sin(120585119896) sin(120591 minus 120591
0+120585
119896)
cos(120585 minus 119899120585
119896) cos(119899(120591 minus 120591
0+120585
119896))
minus sin(120585 minus 119899120585
119896) sin(119899(120591 minus 120591
0+120585
119896))
= cos (119899 (120591 minus 1205910) + 120585)
(102)
we obtain (98) Consequently (98) and (99) are equal whichcompletes the proof
33 Nonnegative Waveforms with Two Zeros and PrescribedCoefficients of 119896thHarmonic In this subsectionwe show thatfor prescribed coefficients 119886
119896and 119887119896 there are 119896 nonnegative
waveforms of type (35) with exactly two zeros According to
(37) and (82) coefficients 119886119896and 119887119896of nonnegative waveforms
of type (35) with exactly two zeros satisfy the followingrelation
1
1198962 minus 1lt radic1198862119896+ 1198872119896lt 1 (103)
According to Remark 16 the value of 119909 (see (87)) that cor-responds to 120582
119896= radic1198862
119896+ 1198872119896can be determined from (90)-
(91) As we mentioned earlier (90) has only one solutionthat satisfies (91) This value of 119909 according to (88) leadsto the amplitude 120582
1of fundamental harmonic (closed form
expressions for 1205821in terms of 120582
119896and 119896 le 4 are given by (93)ndash
(95))On the other hand from (45)-(46) it follows that
1198961205910minus 120585 = atan 2 (119887
119896 119886119896) + 2119902120587 119902 = 1 (119896 minus 1) (104)
where function atan 2(119910 119909) is defined as
atan 2 (119910 119909) =
arctan(119910
119909) if 119909 ge 0
arctan(119910
119909) + 120587 if 119909 lt 0 119910 ge 0
arctan(119910
119909) minus 120587 if 119909 lt 0 119910 lt 0
(105)
with the codomain (minus120587 120587] Furthermore according to (84)and (104) the coefficients of fundamental harmonic of non-negative waveforms with two zeros and prescribed coeffi-cients of 119896th harmonic are equal to
1198861= minus1205821cos[
atan 2 (119887119896 119886119896) + 2119902120587
119896]
1198871= minus1205821sin[
atan 2 (119887119896 119886119896) + 2119902120587
119896]
(106)
where 119902 = 0 (119896minus 1) For chosen 119902 according to (104) and(66) positions of zeros are
1205910=1
119896[120585 + atan 2 (119887
119896 119886119896) + 2119902120587]
1205910minus2120585
119896=1
119896[minus120585 + atan 2 (119887
119896 119886119896) + 2119902120587]
(107)
From (106) and 119902 = 0 (119896minus1) it follows that for prescribedcoefficients 119886
119896and 119887119896 there are 119896 nonnegative waveforms of
type (35) with exactly two zerosWe provide here an algorithm to facilitate calculation
of coefficients 1198861and 1198871of nonnegative waveforms of type
(35) with two zeros and prescribed coefficients 119886119896and 119887
119896
providing that 119886119896and 119887119896satisfy (103)
12 Mathematical Problems in Engineering
2
1
0
Wav
efor
ms
0 1 2 3 4
Angle 120591120587
q = 0
q = 1
q = 2
Figure 6 Nonnegative waveforms with two zeros for 119896 = 3 1198863=
minus015 and 1198873= minus02
Algorithm 17 (i) Calculate 120582119896= radic1198862119896+ 1198872119896
(ii) identify 119909 that satisfies both relations (90) and (91)(iii) calculate 120582
1according to (88)
(iv) choose integer 119902 such that 0 le 119902 le 119896 minus 1(v) calculate 119886
1and 1198871according to (106)
For 119896 le 4 by using (93) for 119896 = 2 (94) for 119896 = 3 and (95)for 119896 = 4 it is possible to calculate directly 120582
1from 120582
119896and
proceed to step (iv)For 119896 = 2 and prescribed coefficients 119886
2and 1198872 there are
two waveforms with two zeros one corresponding to 1198861lt 0
and the other corresponding to 1198861gt 0 (see also [12])
Let us take as an input 119896 = 3 1198863= minus015 and 119887
3= minus02
Execution of Algorithm 17 on this input yields 1205823= 025 and
1205821= 11399 (according to (94)) For 119902 = 0 we calculate
1198861= minus08432 and 119887
1= 07670 (corresponding waveform is
presented by solid line in Figure 6) for 119902 = 1 we calculate1198861= minus02426 and 119887
1= minus11138 (corresponding waveform is
presented by dashed line) for 119902 = 2 we calculate 1198861= 10859
and 1198871= 03468 (corresponding waveform is presented by
dotted line)As another example of the usage of Algorithm 17 let us
consider case 119896 = 4 and assume that1198864= minus015 and 119887
4= minus02
Consequently 1205824= 025 and 120582
1= 09861 (according to (95))
For 119902 = 0 3we calculate the following four pairs (1198861 1198871) of
coefficients of fundamental harmonic (minus08388 05184) for119902 = 0 (minus05184 minus08388) for 119902 = 1 (08388 minus05184) for 119902 =2 and (05184 08388) for 119902 = 3 Corresponding waveformsare presented in Figure 7
4 Nonnegative Waveforms with MaximalAmplitude of Fundamental Harmonic
In this section we provide general description of nonnegativewaveforms containing fundamental and 119896th harmonic withmaximal amplitude of fundamental harmonic for prescribedamplitude of 119896th harmonic
The main result of this section is presented in the fol-lowing proposition
3
2
1
0
Wav
efor
ms
0 1 2 3 4
Angle 120591120587
q = 0q = 1
q = 2q = 3
Figure 7 Nonnegative waveforms with two zeros for 119896 = 4 1198864=
minus015 and 1198874= minus02
Proposition 18 Every nonnegativewaveformof type (35)withmaximal amplitude 120582
1of fundamental harmonic and pre-
scribed amplitude 120582119896of 119896th harmonic can be expressed in the
following form
119879119896(120591) = [1 minus cos (120591 minus 120591
0)]
sdot [1 minus (119896 minus 1) 120582119896minus 2120582119896
119896minus1
sum
119899=1
(119896 minus 119899) cos (119899 (120591 minus 1205910))]
(108)
if 0 le 120582119896le 1(119896
2
minus 1) or
119879119896(120591) = 120582
119896[1 minus cos (120591 minus 120591
0)] [1 minus cos(120591 minus 120591
0+2120585
119896)]
sdot [1198880+ 2
119896minus2
sum
119899=1
119888119899cos 119899(120591 minus 120591
0+120585
119896)]
(109)
if 1(1198962 minus 1) le 120582119896le 1 providing that 119888
119899 119899 = 0 119896 minus 2 and
120582119896are related to 120585 via relations (67) and (68) respectively and
|120585| le 120587
Remark 19 Expression (108) can be obtained from (38) bysetting 120585 = 0 Furthermore insertion of 120585 = 0 into (43)ndash(46)leads to the following expressions for coefficients ofwaveformof type (108)
1198861= minus (1 + 120582
119896) cos 120591
0 119887
1= minus (1 + 120582
119896) sin 120591
0
119886119896= 120582119896cos (119896120591
0) 119887
119896= 120582119896sin (119896120591
0)
(110)
On the other hand (109) coincides with (66) Thereforethe expressions for coefficients of (109) and (66) also coincideThus expressions for coefficients of fundamental harmonic ofwaveform (109) are given by (84) where 120582
1is given by (85)
while expressions for coefficients of 119896th harmonic are givenby (45)-(46)
Waveforms described by (108) have exactly one zerowhile waveforms described by (109) for 1(1198962 minus 1) lt 120582
119896lt 1
Mathematical Problems in Engineering 13
14
12
1
08
06
04
02
00 05 1
Amplitude 120582k
Am
plitu
de1205821
k = 2
k = 3
k = 4
Figure 8 Maximal amplitude of fundamental harmonic as a func-tion of amplitude of 119896th harmonic
have exactly two zeros As we mentioned earlier waveforms(109) for 120582
119896= 1 have 119896 zeros
Remark 20 Maximal amplitude of fundamental harmonic ofnonnegative waveforms of type (35) for prescribed amplitudeof 119896th harmonic can be expressed as
1205821= 1 + 120582
119896 (111)
if 0 le 120582119896le 1(119896
2
minus 1) or
1205821=
119896 sin 120585119896 sin 120585 cos (120585119896) minus cos 120585 sin (120585119896)
(112)
if 1(1198962 minus 1) le 120582119896le 1 where 120585 is related to 120582
119896via (68) (or
(86)) and |120585| le 120587From (110) it follows that (111) holds Substitution of (86)
into (85) leads to (112)Notice that 120582
119896= 1(119896
2
minus 1) is the only common point ofthe intervals 0 le 120582
119896le 1(119896
2
minus 1) and 1(1198962
minus 1) le 120582119896le
1 According to (111) 120582119896= 1(119896
2
minus 1) corresponds to 1205821=
1198962
(1198962
minus1) It can be also obtained from (112) by setting 120585 = 0The waveforms corresponding to this pair of amplitudes aremaximally flat nonnegative waveforms
Maximal amplitude of fundamental harmonic of non-negative waveform of type (35) for 119896 le 4 as a function ofamplitude of 119896th harmonic is presented in Figure 8
Remark 21 Maximum value of amplitude of fundamentalharmonic of nonnegative waveform of type (35) is
1205821max =
1
cos (120587 (2119896)) (113)
This maximum value is attained for |120585| = 1205872 (see (112)) Thecorresponding value of amplitude of 119896th harmonic is 120582
119896=
(1119896) tan(120587(2119896)) Nonnegative waveforms of type (35) with1205821= 1205821max have two zeros at 1205910 and 1205910 minus 120587119896 for 120585 = 1205872 or
at 1205910and 1205910+ 120587119896 for 120585 = minus1205872
14
12
1
08
06
04
02
0minus1 minus05 0 05 1
Am
plitu
de1205821
Parameter 120585120587
k = 2k = 3k = 4
Figure 9 Maximal amplitude of fundamental harmonic as a func-tion of parameter 120585
To prove that (113) holds let us first show that the fol-lowing relation holds for 119896 ge 2
cos( 120587
2119896) lt 1 minus
1
1198962 (114)
From 119896 ge 2 it follows that sinc(120587(4119896)) gt sinc(1205874) wheresinc 119909 = (sin119909)119909 and therefore sin(120587(4119896)) gt 1(radic2119896)By using trigonometric identity cos 2119909 = 1 minus 2sin2119909 weimmediately obtain (114)
According to (111) and (112) it is clear that 1205821attains its
maximum value on the interval 1(1198962 minus 1) le 120582119896le 1 Since
120582119896is monotonic function of |120585| on interval |120585| le 120587 (see
Remark 15) it follows that 119889120582119896119889120585 = 0 for 0 lt |120585| lt 120587
Therefore to find critical points of 1205821as a function of 120582
119896
it is sufficient to find critical points of 1205821as a function of
|120585| 0 lt |120585| lt 120587 and consider its values at the end points120585 = 0 and |120585| = 120587 Plot of 120582
1as a function of parameter 120585
for 119896 le 4 is presented in Figure 9 According to (112) firstderivative of 120582
1with respect to 120585 is equal to zero if and only
if (119896 cos 120585 sin(120585119896) minus sin 120585 cos(120585119896)) cos 120585 = 0 On interval0 lt |120585| lt 120587 this is true if and only if |120585| = 1205872 Accordingto (112) 120582
1is equal to 119896
2
(1198962
minus 1) for 120585 = 0 equal to zerofor |120585| = 120587 and equal to 1 cos(120587(2119896)) for |120585| = 1205872 From(114) it follows that 1198962(1198962minus1) lt 1 cos(120587(2119896)) and thereforemaximum value of 120582
1is given by (113) Moreover maximum
value of 1205821is attained for |120585| = 1205872
According to above consideration all nonnegative wave-forms of type (35) having maximum value of amplitude offundamental harmonic can be obtained from (109) by setting|120585| = 1205872 Three of them corresponding to 119896 = 3 120585 = 1205872and three different values of 120591
0(01205876 and1205873) are presented
in Figure 10 Dotted line corresponds to 1205910= 0 (coefficients
of corresponding waveform are 1198861= minus1 119887
1= 1radic3 119886
3= 0
and 1198873= minusradic39) solid line to 120591
0= 1205876 (119886
1= minus2radic3 119887
1= 0
1198863= radic39 and 119887
3= 0) and dashed line to 120591
0= 1205873 (119886
1= minus1
1198871= minus1radic3 119886
3= 0 and 119887
3= radic39)
Proof of Proposition 18 As it has been shown earlier (Propo-sition 6) nonnegative waveform of type (35) with at least
14 Mathematical Problems in Engineering
2
1
0
0 1 2 3 4
Angle 120591120587
Wav
efor
ms
1205910 = 01205910 = 12058761205910 = 1205873
Figure 10 Nonnegative waveforms with maximum amplitude offundamental harmonic for 119896 = 3 and 120585 = 1205872
one zero can be represented in form (38) According to (43)(44) and (36) for amplitude 120582
1of fundamental harmonic of
waveforms of type (38) the following relation holds
1205821= radic(1 + 120582
119896cos 120585)2 + 11989621205822
119896sin2120585 (115)
where 120582119896satisfy (40) and |120585| le 120587
Because of (40) in the quest of finding maximal 1205821for
prescribed 120582119896 we have to consider the following two cases
(Case i)120582119896lt [(119896minus1) cos 120585 + 119896 sin(120585minus120585119896) sin(120585119896)]minus1
(Case ii)120582119896= [(119896minus1) cos 120585+119896 sin(120585minus120585119896) sin(120585119896)]minus1
Case i Since 120582119896lt [(119896 minus 1) cos 120585 + 119896 sin(120585 minus 120585119896) sin(120585119896)]minus1
implies 120582119896
= 1 according to (115) it follows that 1205821
= 0Hence 119889120582
1119889120585 = 0 implies
2120582119896sin 120585 [1 minus (1198962 minus 1) 120582
119896cos 120585] = 0 (116)
Therefore 1198891205821119889120585 = 0 if 120582
119896= 0 (Option 1) or sin 120585 = 0
(Option 2) or (1198962 minus 1)120582119896cos 120585 = 1 (Option 3)
Option 1 According to (115) 120582119896= 0 implies 120582
1= 1 (notice
that this implication shows that 1205821does not depend on 120585 and
therefore we can set 120585 to zero value)
Option 2 According to (115) sin 120585 = 0 implies 1205821= 1 +
120582119896cos 120585 which further leads to the conclusion that 120582
1is
maximal for 120585 = 0 For 120585 = 0 120582119896lt [(119896 minus 1) cos 120585 + 119896 sin(120585 minus
120585119896) sin(120585119896)]minus1 becomes 120582119896lt 1(119896
2
minus 1)
Option 3 This option leads to contradiction To show thatnotice that (119896
2
minus 1)120582119896cos 120585 = 1 and 120582
119896lt [(119896 minus
1) cos 120585 + 119896 sin(120585 minus 120585119896) sin(120585119896)]minus1 imply that (119896 minus 1) cos 120585 gtsin(120585minus120585119896) sin(120585119896) Using (A5) (see Appendices) the latestinequality can be rewritten assum119896minus1
119899=1[cos 120585minuscos((119896minus2119899)120585119896)] gt
0 But from |119896 minus 2119899| lt 119896 119899 = 1 (119896 minus 1) and |120585| le 120587
it follows that all summands are not positive and therefore(119896minus1) cos 120585 gt sin(120585minus120585119896) sin(120585119896) does not hold for |120585| le 120587
Consequently Case i implies 120585 = 0 and 120582119896lt 1(119896
2
minus 1)Finally substitution of 120585 = 0 into (38) leads to (108) whichproves that (108) holds for 120582
119896lt 1(119896
2
minus 1)
Case ii Relation120582119896= [(119896minus1) cos 120585+119896 sin(120585minus120585119896) sin(120585119896)]minus1
according to Proposition 9 and Remark 11 implies that cor-responding waveforms can be expressed via (66)ndash(68) for|120585| le 120587 Furthermore 120582
119896= [(119896 minus 1) cos 120585 + 119896 sin(120585 minus
120585119896) sin(120585119896)]minus1 and |120585| le 120587 imply 1(1198962 minus 1) le 120582119896le 1
This proves that (109) holds for 1(1198962 minus 1) le 120582119896le 1
Finally let us prove that (108) holds for 120582119896= 1(119896
2
minus
1) According to (68) (see also Remark 11) this value of 120582119896
corresponds to 120585 = 0 Furthermore substitution of 120582119896=
1(1198962
minus 1) and 120585 = 0 into (109) leads to (70) which can berewritten as
119879119896(120591) =
[1 minus cos (120591 minus 1205910)]
(1 minus 1198962)
sdot [119896 (119896 minus 1) minus 2
119896minus1
sum
119899=1
(119896 minus 119899) cos (119899 (120591 minus 1205910))]
(117)
Waveform (117) coincides with waveform (108) for 120582119896
=
1(1 minus 1198962
) Consequently (108) holds for 120582119896= 1(1 minus 119896
2
)which completes the proof
5 Nonnegative Waveforms with MaximalAbsolute Value of the Coefficient of CosineTerm of Fundamental Harmonic
In this sectionwe consider general description of nonnegativewaveforms of type (35) with maximal absolute value ofcoefficient 119886
1for prescribed coefficients of 119896th harmonicThis
type of waveform is of particular interest in PA efficiencyanalysis In a number of cases of practical interest eithercurrent or voltage waveform is prescribed In such casesthe problem of finding maximal efficiency of PA can bereduced to the problem of finding nonnegative waveformwith maximal coefficient 119886
1for prescribed coefficients of 119896th
harmonic (see also Section 7)In Section 51 we provide general description of nonneg-
ative waveforms of type (35) with maximal absolute value ofcoefficient 119886
1for prescribed coefficients of 119896th harmonic In
Section 52 we illustrate results of Section 51 for particularcase 119896 = 3
51 Nonnegative Waveforms with Maximal Absolute Value ofCoefficient 119886
1for 119896 ge 2 Waveforms 119879
119896(120591) of type (35) with
1198861ge 0 can be derived from those with 119886
1le 0 by shifting
by 120587 and therefore we can assume without loss of generalitythat 119886
1le 0 Notice that if 119896 is even then shifting 119879
119896(120591) by
120587 produces the same result as replacement of 1198861with minus119886
1
(119886119896remains the same) On the other hand if 119896 is odd then
shifting 119879119896(120591) by 120587 produces the same result as replacement
of 1198861with minus119886
1and 119886119896with minus119886
119896
According to (37) coefficients of 119896th harmonic can beexpressed as
119886119896= 120582119896cos 120575 119887
119896= 120582119896sin 120575 (118)
Mathematical Problems in Engineering 15
where
|120575| le 120587 (119)
Conversely for prescribed coefficients 119886119896and 119887
119896 120575 can be
determined as
120575 = atan 2 (119887119896 119886119896) (120)
where definition of function atan 2(119910 119909) is given by (105)The main result of this section is stated in the following
proposition
Proposition 22 Every nonnegative waveform of type (35)withmaximal absolute value of coefficient 119886
1le 0 for prescribed
coefficients 119886119896and 119887119896of 119896th harmonic can be represented as
119879119896(120591)
= [1 minus cos 120591]
sdot [1 minus (119896 minus 1) 119886119896minus 2
119896minus1
sum
119899=1
(119896 minus 119899) (119886119896cos 119899120591 + 119887
119896sin 119899120591)]
(121)
if 119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) le 1+119886
119896 where 120575 = atan 2(bk
119886119896) or
119879119896(120591) = 120582
119896[1 minus cos(120591 minus (120575 + 120585)
119896)]
sdot [1 minus cos(120591 minus (120575 minus 120585)
119896)]
sdot [1198880+ 2
119896minus2
sum
119899=1
119888119899cos 119899(120591 minus 120575
119896)]
(122)
if 119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) ge 1 + 119886
119896 where 119888
119899 119899 = 0
119896minus2 and 120582119896= radic1198862119896+ 1198872119896are related to 120585 via relations (67) and
(68) respectively and |120585| le 120587
Remark 23 Expression (121) can be obtained from (38) bysetting 120591
0= 0 and 120585 = minus120575 and then replacing 120582
119896cos 120575 with
119886119896(see (118)) and 120582
119896cos(119899120591 minus 120575) with 119886
119896cos 119899120591 + 119887
119896sin 119899120591
(see also (118)) Furthermore insertion of 1205910= 0 and 120585 =
minus120575 into (43)ndash(46) leads to the following relations betweenfundamental and 119896th harmonic coefficients of waveform(121)
1198861= minus (1 + 119886
119896) 119887
1= minus119896119887
119896 (123)
On the other hand expression (122) can be obtained from(66) by replacing 120591
0minus120585119896with 120575119896 Therefore substitution of
1205910minus 120585119896 = 120575119896 in (84) leads to
1198861= minus1205821cos(120575
119896) 119887
1= minus1205821sin(120575
119896) (124)
where 1205821is given by (85)
The fundamental harmonic coefficients 1198861and 1198871of wave-
form of type (35) with maximal absolute value of coefficient1198861le 0 satisfy both relations (123) and (124) if 119886
119896and 119887119896satisfy
1 + 119886119896= 119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) For such waveforms
relations 1205910= 0 and 120585 = minus120575 also hold
Remark 24 Amplitude of 119896th harmonic of nonnegativewaveform of type (35) with maximal absolute value of coeffi-cient 119886
1le 0 and coefficients 119886
119896 119887119896satisfying 1 + 119886
119896=
119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) is
120582119896=
sin (120575119896)119896 sin 120575 cos (120575119896) minus cos 120575 sin (120575119896)
(125)
To show that it is sufficient to substitute 119886119896= 120582119896cos 120575 (see
(118)) into 1 + 119886119896= 119896120582119896[sin 120575 sin(120575119896)] cos(120575119896)
Introducing new variable
119910 = cos(120575119896) (126)
and using the Chebyshev polynomials (eg see Appendices)relations 119886
119896= 120582119896cos 120575 and (125) can be rewritten as
119886119896= 120582119896119881119896(119910) (127)
120582119896=
1
119896119910119880119896minus1
(119910) minus 119881119896(119910)
(128)
where119881119896(119910) and119880
119896(119910) denote the Chebyshev polynomials of
the first and second kind respectively Substitution of (128)into (127) leads to
119886119896119896119910119880119896minus1
(119910) minus (1 + 119886119896) 119881119896(119910) = 0 (129)
which is polynomial equation of 119896th degree in terms of var-iable 119910 From |120575| le 120587 and (126) it follows that
cos(120587119896) le 119910 le 1 (130)
In what follows we show that 119886119896is monotonically increas-
ing function of 119910 on the interval (130) From 120585 = minus120575 (seeRemark 23) and (81) it follows that 120582minus1
119896= (119896 minus 1) cos 120575 +
119896sum119896minus1
119899=1cos((119896 minus 2119899)120575119896) ge 1 and therefore 119886
119896= 120582119896cos 120575 can
be rewritten as
119886119896=
cos 120575(119896 minus 1) cos 120575 + 119896sum119896minus1
119899=1cos ((119896 minus 2119899) 120575119896)
(131)
Obviously 119886119896is even function of 120575 and all cosines in (131)
are monotonically decreasing functions of |120575| on the interval|120575| le 120587 It is easy to show that cos((119896 minus 2119899)120575119896) 119899 =
1 (119896 minus 1) decreases slower than cos 120575 when |120575| increasesThis implies that denominator of the right hand side of(131) decreases slower than numerator Since denominator ispositive for |120575| le 120587 it further implies that 119886
119896is decreasing
function of |120575| on interval |120575| le 120587 Consequently 119886119896is
monotonically increasing function of 119910 on the interval (130)Thus we have shown that 119886
119896is monotonically increasing
function of 119910 on the interval (130) and therefore (129) hasonly one solution that satisfies (130) According to (128) thevalue of 119910 obtained from (129) and (130) either analyticallyor numerically leads to amplitude 120582
119896of 119896th harmonic
16 Mathematical Problems in Engineering
1
05
0
minus05
minus1
minus1 minus05 0 05 1
Coefficient ak
Coe
ffici
entb
k
radica2k+ b2
kle 1
k = 2k = 3k = 4
Figure 11 Plot of (119886119896 119887119896) satisfying 1 + 119886
119896= 119896120582
119896[sin 120575 sin(120575
119896)] cos(120575119896) for 119896 le 4
By solving (129) and (130) for 119896 le 4 we obtain
119910 = radic1 + 1198862
2 (1 minus 1198862) minus1 le 119886
2le1
3
119910 = radic3
4 (1 minus 21198863) minus1 le 119886
3le1
8
119910 =radicradic2 minus 4119886
4+ 1011988624minus 2 (1 minus 119886
4)
4 (1 minus 31198864)
minus1 le 1198864le
1
15
(132)
Insertion of (132) into (128) leads to the following explicitexpressions for the amplitude 120582
119896 119896 le 4
1205822=1
2(1 minus 119886
2) minus1 le 119886
2le1
3 (133)
1205822
3= [
1
3(1 minus 2119886
3)]
3
minus1 le 1198863le1
8 (134)
1205824=1
4(minus1 minus 119886
4+ radic2 minus 4119886
4+ 1011988624) minus1 le 119886
4le
1
15
(135)
Relations (133)ndash(135) define closed lines (see Figure 11) whichseparate points representing waveforms of type (121) frompoints representing waveforms of type (122) For given 119896points inside the corresponding curve refer to nonnegativewaveforms of type (121) whereas points outside curve (andradic1198862119896+ 1198872119896le 1) correspond to nonnegative waveforms of type
(122) Points on the respective curve correspond to the wave-forms which can be expressed in both forms (121) and (122)
Remark 25 Themaximum absolute value of coefficient 1198861of
nonnegative waveform of type (35) is
100381610038161003816100381611988611003816100381610038161003816max =
1
cos (120587 (2119896)) (136)
This maximum value is attained for |120585| = 1205872 and 120575 = 0
(see (124)) Notice that |1198861|max is equal to the maximum value
1205821max of amplitude of fundamental harmonic (see (113))
Coefficients of waveform with maximum absolute value ofcoefficient 119886
1 1198861lt 0 are
1198861= minus
1
cos (120587 (2119896)) 119886
119896=1
119896tan( 120587
(2119896))
1198871= 119887119896= 0
(137)
Waveformdescribed by (137) is cosinewaveformhaving zerosat 120587(2119896) and minus120587(2119896)
In the course of proving (136) notice first that |1198861|max le
1205821max holds According to (123) and (124) maximum of |119886
1|
occurs for 119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) ge 1 + 119886
119896 From (124)
it immediately follows that maximum value of |1198861| is attained
if and only if 1205821= 1205821max and 120575 = 0 which because of
120575119896 = 1205910minus120585119896 further implies 120591
0= 120585119896 Sincemaximumvalue
of 1205821is attained for |120585| = 1205872 it follows that corresponding
waveform has zeros at 120587(2119896) and minus120587(2119896)
Proof of Proposition 22 As it was mentioned earlier in thissection we can assume without loss of generality that 119886
1le 0
We consider waveforms119879119896(120591) of type (35) such that119879
119896(120591) ge 0
and119879119896(120591) = 0 for some 120591
0 Fromassumption that nonnegative
waveform 119879119896(120591) of type (35) has at least one zero it follows
that it can be expressed in form (38)Let us also assume that 120591
0is position of nondegenerate
critical point Therefore 119879119896(1205910) = 0 implies 1198791015840
119896(1205910) = 0 and
11987910158401015840
119896(1205910) gt 0 According to (55) second derivative of 119879
119896(120591) at
1205910can be expressed as 11987910158401015840
119896(1205910) = 1 minus 120582
119896(1198962
minus 1) cos 120585 Since11987910158401015840
119896(1205910) gt 0 it follows immediately that
1 minus 120582119896(1198962
minus 1) cos 120585 gt 0 (138)
Let us further assume that 119879119896(120591) has exactly one zeroThe
problem of finding maximum absolute value of 1198861is con-
nected to the problem of finding maximum of the minimumfunction (see Section 21) If waveforms possess unique globalminimum at nondegenerate critical point then correspond-ing minimum function is a smooth function of parameters[13] Consequently assumption that 119879
119896(120591) has exactly one
zero at nondegenerate critical point leads to the conclusionthat coefficient 119886
1is differentiable function of 120591
0 First
derivative of 1198861(see (43)) with respect to 120591
0 taking into
account that 1205971205851205971205910= 119896 (see (50)) can be expressed in the
following factorized form
1205971198861
1205971205910
= sin 1205910[1 minus 120582
119896(1198962
minus 1) cos 120585] (139)
Mathematical Problems in Engineering 17
From (138) and (139) it is clear that 12059711988611205971205910= 0 if and only if
sin 1205910= 0 According toRemark 12 assumption that119879
119896(120591)has
exactly one zero implies 120582119896lt 1 From (51) (48) and 120582
119896lt 1
it follows that 1198861cos 1205910+ 1198871sin 1205910lt 0 which together with
sin 1205910= 0 implies that 119886
1cos 1205910lt 0 Assumption 119886
1le 0
together with relations 1198861cos 1205910lt 0 and sin 120591
0= 0 further
implies 1198861
= 0 and
1205910= 0 (140)
Insertion of 1205910= 0 into (38) leads to
119879119896(120591)
= [1 minus cos 120591]
sdot [1 minus (119896 minus 1) 120582119896cos 120585 minus 2120582
119896
119896minus1
sum
119899=1
(119896 minus 119899) cos (119899120591 + 120585)]
(141)
Substitution of 1205910= 0 into (45) and (46) yields 119886
119896= 120582119896cos 120585
and 119887119896
= minus120582119896sin 120585 respectively Replacing 120582
119896cos 120585 with
119886119896and 120582
119896cos(119899120591 + 120585) with (119886
119896cos 119899120591 + 119887
119896sin 119899120591) in (141)
immediately leads to (121)Furthermore 119886
119896= 120582119896cos 120585 119887
119896= minus120582
119896sin 120585 and (118)
imply that
120575 = minus120585 (142)
According to (38)ndash(40) and (142) it follows that (141) is non-negative if and only if
120582119896[(119896 minus 1) cos 120575 + 119896 sin (120575 minus 120575119896)
sin (120575119896)] lt 1 (143)
Notice that 119886119896= 120582119896cos 120575 implies that the following relation
holds
120582119896[(119896 minus 1) cos 120575 + 119896 sin (120575 minus 120575119896)
sin (120575119896)]
= minus119886119896+ 119896120582119896
sin 120575sin (120575119896)
cos(120575119896)
(144)
Finally substitution of (144) into (143) leads to 119896120582119896[sin 120575
sin(120575119896)] cos(120575119896) lt 1 + 119886119896 which proves that (121) holds
when 119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) lt 1 + 119886
119896
Apart from nonnegative waveforms with exactly one zeroat nondegenerate critical point in what follows we will alsoconsider other types of nonnegative waveforms with at leastone zero According to Proposition 9 and Remark 11 thesewaveforms can be described by (66)ndash(68) providing that 0 le|120585| le 120587
According to (35) 119879119896(0) ge 0 implies 1 + 119886
1+ 119886119896ge 0
Consequently 1198861le 0 implies that |119886
1| le 1 + 119886
119896 On the other
hand according to (123) |1198861| = 1 + 119886
119896holds for waveforms
of type (121) The converse is also true 1198861le 0 and |119886
1| =
1 + 119886119896imply 119886
1= minus1 minus 119886
119896 which further from (35) implies
119879119896(0) = 0 Therefore in what follows it is enough to consider
only nonnegativewaveformswhich can be described by (66)ndash(68) and 0 le |120585| le 120587 with coefficients 119886
119896and 119887119896satisfying
119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) ge 1 + 119886
119896
For prescribed coefficients 119886119896and 119887119896 the amplitude 120582
119896=
radic1198862119896+ 1198872119896of 119896th harmonic is also prescribed According to
Remark 15 (see also Remark 16) 120582119896is monotonically
decreasing function of 119909 = cos(120585119896) The value of 119909 can beobtained by solving (90) subject to the constraint cos(120587119896) le119909 le 1 Then 120582
1can be determined from (88) From (106) it
immediately follows that maximal absolute value of 1198861le 0
corresponds to 119902 = 0 which from (104) and (120) furtherimplies that
120575 = 1198961205910minus 120585 (145)
Furthermore 119902 = 0 according to (107) implies that waveformzeros are
1205910=(120575 + 120585)
119896 120591
1015840
0= 1205910minus2120585
119896=(120575 minus 120585)
119896 (146)
Substitution of 1205910= (120575 + 120585)119896 into (66) yields (122) which
proves that (122) holds when 119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) ge
1 + 119886119896
In what follows we prove that (121) also holds when119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) = 1 + 119886
119896 Substitution of 119886
119896=
120582119896cos 120575 into 119896120582
119896[sin 120575 sin(120575119896)] cos(120575119896) = 1 + 119886
119896leads to
120582119896[(119896 minus 1) cos 120575 + 119896 sin (120575 minus 120575119896)
sin (120575119896)] = 1 (147)
As we mentioned earlier relation (142) holds for all wave-forms of type (121) Substituting (142) into (147) we obtain
120582119896[(119896 minus 1) cos 120585 + 119896 sin (120585 minus 120585119896)
sin (120585119896)] = 1 (148)
This expression can be rearranged as
120582119896
119896 sin ((119896 minus 1) 120585119896)sin 120585119896
= 1 minus (119896 minus 1) 120582119896cos 120585 (149)
On the other hand for waveforms of type (122) according to(68) relations (148) and (149) also hold Substitution of 120591
0=
(120575 + 120585)119896 (see (145)) and (67) into (122) leads to
119879119896(120591)
= 120582119896[1 minus cos (120591 minus 120591
0)]
sdot [119896 sin ((119896 minus 1) 120585119896)
sin 120585119896minus 2
119896minus1
sum
119899=1
(119896 minus 119899) cos (119899 (120591 minus 1205910) + 120585)]
(150)
Furthermore substitution of (142) into (145) implies that1205910
= 0 Finally substitution of 1205910
= 0 and (149) into(150) leads to (141) Therefore (141) holds when 119896120582
119896[sin 120575
sin(120575119896)] cos(120575119896) = 1 + 119886119896 which in turn shows that (121)
holds when 119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) = 1 + 119886
119896 This
completes the proof
18 Mathematical Problems in Engineering
52 Nonnegative Waveforms with Maximal Absolute Valueof Coefficient 119886
1for 119896 = 3 Nonnegative waveform of type
(35) for 119896 = 3 is widely used in PA design (eg see [10])In this subsection we illustrate results of Section 51 for thisparticular case The case 119896 = 2 is presented in detail in [12]
Coefficients of fundamental harmonic of nonnegativewaveform of type (35) with 119896 = 3 and maximal absolutevalue of coefficient 119886
1le 0 for prescribed coefficients 119886
3and
1198873(1205823= radic11988623+ 11988723) according to (123) (124) (134) (94) and
(120) are equal to
1198861= minus1 minus 119886
3 119887
1= minus3119887
3 (151)
if 12058223le [(1 minus 2119886
3)3]3
1198861= minus1205821cos(120575
3) 119887
1= minus1205821sin(120575
3) (152)
where 1205821= 3(
3radic1205823minus 1205823) and 120575 = atan 2(119887
3 1198863) if [(1 minus
21198863)3]3
le 1205822
3le 1The line 1205822
3= [(1minus2119886
3)3]3 (see case 119896 = 3
in Figure 11) separates points representing waveforms withcoefficients satisfying (151) from points representing wave-forms with coefficients satisfying (152) Waveforms describedby (151) for 1205822
3lt [(1 minus 2119886
3)3]3 have exactly one zero at
1205910= 0 Waveforms described by (151) and (152) for 1205822
3= [(1 minus
21198863)3]3 also have zero at 120591
0= 0 These waveforms as a rule
have exactly two zeros However there are two exceptionsone related to the maximally flat nonnegative waveform withcoefficients 119886
1= minus98 119886
3= 18 and 119887
1= 1198873= 0 which
has only one zero and the other related to the waveform withcoefficients 119886
1= 0 119886
3= minus1 and 119887
1= 1198873= 0 which has three
zerosWaveforms described by (152) for [(1minus21198863)3]3
lt 1205822
3lt
1 have two zeros Waveforms with 1205823= 1 have only third
harmonic (fundamental harmonic is zero)Plot of contours of maximal absolute value of coefficient
1198861 1198861le 0 for prescribed coefficients 119886
3and 1198873is presented
in Figure 12 According to Remark 25 the waveform withmaximum absolute value of 119886
1le 0 is fully described with
the following coefficients 1198861
= minus2radic3 1198863
= radic39 and1198871= 1198873= 0 This waveform has two zeros at plusmn1205876
Two examples of nonnegative waveforms for 119896 = 3
and maximal absolute value of coefficient 1198861 1198861le 0 with
prescribed coefficients 1198863and 1198873are presented in Figure 13
One waveform corresponds to the case 12058223lt [(1 minus 2119886
3)3]3
(solid line) and the other to the case 12058223gt [(1 minus 2119886
3)3]3
(dashed line)Thewaveform represented by solid line has onezero and its coefficients are 119886
3= minus01 119887
3= 01 119886
1= minus09
and 1198871= minus03 Dashed line corresponds to the waveform
having two zeros with coefficients 1198863= minus01 119887
3= 03 119886
1=
minus08844 and 1198871= minus06460 (case 1205822
3gt [(1 minus 2119886
3)3]3)
6 Nonnegative Cosine Waveforms withat Least One Zero
Nonnegative cosine waveforms have proved to be of impor-tance for waveform modelling in PA design (eg see [10])In this section we consider nonnegative cosine waveforms
1
05
0
minus05
minus1
minus1 minus05 0 05 1
Coefficient a3
Coe
ffici
entb
3
02
04
06
08
10
11
Figure 12 Contours ofmaximal absolute value of coefficient 1198861 1198861le
0 as a function of 1198863and 1198873
2
1
0
0 1 2 3 4
Angle 120591120587
Wav
efor
ms
a3 = minus01 b3 = 01
a3 = minus01 b3 = 03
Figure 13 Nonnegative waveforms for 119896 = 3 and maximal absolutevalue of 119886
1 1198861le 0 with prescribed coefficients 119886
3and 1198873
containing fundamental and 119896th harmonic with at least onezero
Cosine waveform with dc component fundamental and119896th harmonic can be obtained from (35) by setting 119887
1= 119887119896=
0 that is
119879119896(120591) = 1 + 119886
1cos 120591 + 119886
119896cos 119896120591 (153)
In Section 61 we provide general description of non-negative cosine waveforms of type (153) with at least onezero We show that nonnegative cosine waveforms with atleast one zero coincide with nonnegative cosine waveformswith maximal absolute value of coefficient 119886
1for prescribed
coefficient 119886119896 In Section 62 we illustrate results of Section 61
for particular case 119896 = 3
61 Nonnegative Cosine Waveforms with at Least One Zerofor 119896 ge 2 Amplitudes of fundamental and 119896th harmonic
Mathematical Problems in Engineering 19
of cosine waveform of type (153) are 1205821= |1198861| and 120582
119896=
|119886119896| respectively According to (42) for nonnegative cosine
waveforms of type (153) the following relation holds
minus1 le 119886119896le 1 (154)
This explains why 119896th harmonic coefficient 119886119896in Proposi-
tion 26 goes through interval [minus1 1]Waveforms (153) with 119886
1ge 0 can be obtained from
waveforms with 1198861le 0 by shifting by 120587 and therefore with-
out loss of generality we can assume that 1198861le 0
Proposition 26 Each nonnegative cosine waveform of type(153) with 119886
1le 0 and at least one zero can be represented as
119879119896(120591) = [1 minus cos 120591] [1 minus (119896 minus 1) 119886
119896minus 2119886119896
119896minus1
sum
119899=1
(119896 minus 119899) cos 119899120591]
(155)
if minus1 le 119886119896le 1(119896
2
minus 1) or
119879119896(120591) = 119886
119896[1 minus cos (120591 minus 120591
0)] [1 minus cos (120591 + 120591
0)]
sdot [1198880+ 2
119896minus2
sum
119899=1
119888119899cos 119899120591]
(156)
where
119888119899=sin ((119896 minus 119899) 120591
0) cos 120591
0minus (119896 minus 119899) cos ((119896 minus 119899) 120591
0) sin 120591
0
sin31205910
(157)
119886119896=
sin 1205910
119896 sin (1198961205910) cos 120591
0minus cos (119896120591
0) sin 120591
0
(158)
100381610038161003816100381612059101003816100381610038161003816 le
120587
119896 (159)
if 1(1198962 minus 1) le 119886119896le 1
Remark 27 Identity [1minuscos(120591minus1205910)][1minuscos(120591+120591
0)] = [cos 120591
0minus
cos 120591]2 implies that (156) can be rewritten as
119879119896(120591) = 119886
119896[cos 120591
0minus cos 120591]2 [119888
0+ 2
119896minus2
sum
119899=1
119888119899cos 119899120591] (160)
Furthermore substitution of (157) into (160) leads to
119879119896(120591) = 119886
119896[cos 120591
0minus cos 120591]
sdot [(119896 minus 1) sin 119896120591
0
sin 1205910
minus 2
119896minus1
sum
119899=1
sin ((119896 minus 119899) 1205910)
sin 1205910
cos 119899120591]
(161)
Remark 28 All nonnegative cosine waveforms of type (153)with at least one zero and 119886
1le 0 except one of them can be
represented either in form (155) or form (156)This exceptionis maximally flat cosine waveform with 119886
1lt 0 which can be
obtained from (155) for 119886119896= 1(119896
2
minus 1) or from (156) for 1205910=
0 Maximally flat cosine waveform with 1198861lt 0 can also be
obtained from (70) by setting 1205910= 0 Furthermore setting
1205910= 0 in (71) leads to maximally flat cosine waveforms for
119896 le 4 and 1198861lt 0
Remark 29 Nonnegative cosine waveform of type (155) with1198861lt 0 and minus1 lt 119886
119896le 1(119896
2
minus 1) has exactly one zero at120591 = 0 Nonnegative cosine waveform described by (156) with1198861lt 0 and 1(1198962 minus 1) lt 119886
119896lt 1 has two zeros at plusmn120591
0 where
0 lt |1205910| lt 120587119896 For 119886
119896= ∓1 nonnegative cosine waveform
of type (153) reduces to 119879119896(120591) = 1 ∓ cos 119896120591 (clearly these two
waveforms both have 119896 zeros)
Remark 30 Transformation of (155) into an additive formleads to the following relation
1198861= minus1 minus 119886
119896 (162)
where minus1 le 119886119896le 1(119896
2
minus1) Similarly transformation of (156)leads to the following relation
1198861= minus119886119896
119896 sin 1198961205910
sin 1205910
(163)
where 119886119896is given by (158) 1(1198962minus1) le 119886
119896le 1 and |120591
0| le 120587119896
Notice that coefficients of maximally flat cosine waveformnamely 119886
119896= 1(119896
2
minus1) and 1198861= minus1198962
(1198962
minus1) satisfy relation(162) They also satisfy relation (163) for 120591
0= 0
Remark 31 Nonnegative cosine waveforms of type (153) withat least one zero coincide with nonnegative cosine waveformswith maximal absolute value of coefficient 119886
1for prescribed
coefficient 119886119896
In proving that Remark 31 holds notice that expression(155) can be obtained from (121) by setting 119887
119896= 0 Fur-
thermore if 119886119896
ge 0 then 120582119896
= 119886119896 which together
with 119887119896
= 0 and (118) implies 120575 = 0 In this case119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) le 1 + 119886
119896becomes 1198962119886
119896le 1 +
119886119896 On the other hand if 119886
119896lt 0 then 120582
119896= minus119886
119896 which
together with 119887119896= 0 and (118) implies |120575| = 120587 In this case
119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) le 1 + 119886
119896becomes 0 le 1 + 119886
119896
Therefore every nonnegative cosine waveform of type (155)has maximal absolute value of coefficient 119886
1for prescribed
coefficient 119886119896 when minus1 le 119886
119896le 1(119896
2
minus 1)Let us now show that expression (156) can be obtained
from (122) by setting 119887119896= 0 and 119886
119896gt 0 For waveforms
of type (122) according to (118) 119887119896= 0 and 119886
119896gt 0 imply
120575 = 0 and 120582119896= 119886119896 Substitution of 120582
119896= 119886119896and 120575 = 0 into
119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) ge 1+119886
119896leads to 119886
119896ge 1(119896
2
minus1)Furthermore substitution of 120575 = 0 into (145) yields 120591
0= 120585119896
Insertion of 120582119896= 119886119896 120575 = 0 and 120591
0= 120585119896 into (122) leads
to (156) Therefore every nonnegative cosine waveform oftype (156) has maximal absolute value of coefficient 119886
1for
prescribed coefficient 119886119896 when 1(1198962 minus 1) le 119886
119896le 1
Proof of Proposition 26 Let us start with nonnegative cosinewaveform of type (153) with 120582
119896= |119886119896| = 1 According to
Remark 7 120582119896= |119886119896| = 1 implies that 120582
1= |1198861| = 0
Substitution of 119886119896
= minus1 into (155) and using (A2) (seeAppendices) lead to 119879
119896(120591) = 1 minus cos 119896120591 Consequently (155)
holds for 119886119896= minus1 On the other hand substitution of 119886
119896= 1
into (158) yields |1205910| = 120587119896 Furthermore substitution of
119886119896= 1 and 120591
0= 120587119896 (or 120591
0= minus120587119896) into (156) along
with performing all multiplications and using (A2) leads to
20 Mathematical Problems in Engineering
119879119896(120591) = 1 + cos 119896120591 Consequently (156)ndash(158) hold for 119886
119896= 1
and |1205910| = 120587119896
It is easy to see that 120582119896= |119886119896| lt 1 and 119879
119896(1205910) = 0 for some
1205910imply 120582
1= |1198861| = 0 Therefore in what follows we assume
that |119886119896| = 1 and 119886
1lt 0
Cosine waveforms are even functions of 120591 Therefore ifnonnegative cosine waveform has exactly one zero it has to beeither at 0 or at 120587 On the other hand if nonnegative cosinewaveform with 119886
1= 0 has exactly two zeros then these zeros
are placed at plusmn1205910 such that 120591
0is neither 0 nor 120587
In order to prove that (155) holds for minus1 lt 119886119896le 1(1minus119896
2
)let us start by referring to the description (38) of nonnegativewaveformswith at least one zero As wementioned earlier fornonnegative cosine waveformwith exactly one zero (denotedby 1205910) it is either 120591
0= 0 or 120591
0= 120587 Therefore in both cases
sin 1205910= 0 Substitution of sin 120591
0= 0 into (43) together with
1198861
= 0 and 120582119896= |119886119896| lt 1 leads to
1205910= 0 (164)
Clearly 1205910= 0 119887
1= 0 and 119887
119896= 0 according to (44) and (46)
imply 120582119896sin 120585 = 0 Since 120582
119896= |119886119896| it follows that |119886
119896| sin 120585 = 0
also holds which further implies 120582119896= 119886119896= 0 or sin 120585 =
0 In the case when 120582119896= 119886119896= 0 from (164) and (43) we
obtain 1198861= minus1 which further implies that 119879
119896(120591) = 1 minus cos 120591
Consequently (155) holds for 119886119896= 0 In the case when sin 120585 =
0 from (164) and (45) we obtain 119886119896= 120582119896if 120585 = 0 or 119886
119896= minus120582119896
if 120585 = 120587 Relations 119886119896= 120582119896and 120585 = 0 according to (40) imply
that 0 le 119886119896le 1(1 minus 119896
2
) Substitution of 120585 = 0 120582119896= 119886119896 and
(164) into (38) leads to (155) which proves that (155) holdsfor 0 le 119886
119896le 1(1 minus 119896
2
) On the other hand relations 119886119896=
minus120582119896and 120585 = 120587 according to (40) imply that minus1 lt 119886
119896le 0
Substitution of 120585 = 120587 120582119896= minus119886119896 and (164) into (38) also leads
to (155) which proves that (155) also holds for minus1 lt 119886119896le 0
Consequently (155) holds for minus1 lt 119886119896le 1(1 minus 119896
2
)In what follows we first prove that (156)-(157) hold for
1(1 minus 1198962
) lt 119886119896lt 1 For this purpose let us start with non-
negative waveforms with two zeros described by (66) As wementioned before nonnegative cosine waveforms with twozeros have zeros at 120591
0and minus120591
0 such that 120591
0= 0 and 120591
0= 120587
Relations 1198861
lt 0 and 1198871
= 0 according to (84) implycos(1205910minus 120585119896) = 1 and therefore
120585
119896= 1205910 (165)
From 120585119896 = 1205910and 0 lt |120585| lt 120587 it follows that 0 lt |120591
0| lt
120587119896 Insertion of 120585119896 = 1205910into (45) yields 119886
119896= 120582119896 Relations
119886119896= 120582119896and (82) imply that 1(1 minus 1198962) lt 119886
119896lt 1 Substitution
of 120582119896= 119886119896and 120585119896 = 120591
0into (66)ndash(68) leads to (156)ndash(158)
which proves that (156)ndash(158) hold for 1(1 minus 1198962
) lt 119886119896lt 1
and 0 lt |1205910| lt 120587119896
Finally substitution of 119886119896= 1(1 minus 119896
2
) and 1205910= 0 into
(161) leads to
119879119896(120591) =
[1 minus cos 120591](1 minus 1198962)
[119896 (119896 minus 1) minus 2
119896minus1
sum
119899=1
(119896 minus 119899) cos 119899120591]
(166)
Waveform (166) coincides with waveform (155) for 119886119896
=
1(1 minus 1198962
) which in turn proves that (156) holds for 119886119896=
1(1 minus 1198962
) and 1205910= 0 This completes the proof
62 Nonnegative CosineWaveforms with at Least One Zero for119896 = 3 In this subsection we consider nonnegative cosinewaveforms with at least one zero for 119896 = 3 (for case 119896 = 2
see [12])Cosine waveform with fundamental and third harmonic
reads
1198793(120591) = 1 + 119886
1cos 120591 + 119886
3cos 3120591 (167)
For 1198861le 0 and minus1 le 119886
3le 18 according to (155) non-
negative cosine waveform of type (167) with at least one zerocan be expressed as
1198793(120591) = (1 minus cos 120591) [1 minus 2119886
3(1 + 2 cos 120591 + cos 2120591)] (168)
From 1198793(120591 + 120587) = 2 minus 119879
3(120591) it immediately follows that for
1198861ge 0 and minus18 le 119886
3le 1 119879
3(120591) can be expressed as
1198793(120591) = (1 + cos 120591) [1 + 2119886
3(1 minus 2 cos 120591 + cos 2120591)] (169)
For 1198861le 0 and 18 le 119886
3le 1 from (158) it follows that 119886
3=
[8cos31205910]minus1 This relation along with (160) and (157) further
implies that 1198793(120591) can be expressed as
1198793(120591) =
[cos 1205910minus cos 120591]2 [2 cos 120591
0+ cos 120591]
2cos31205910
(170)
providing that |1205910| le 1205873 From 119879
3(120591 + 120587) = 2 minus 119879
3(120591) it
follows that (170) also holds for 1198861ge 0 and minus1 le 119886
3le minus18
providing that 1205910isin [21205873 41205873]
Maximally flat nonnegative cosinewaveformof type (167)with 119886
1lt 0 (minimum at 120591
0= 0) reads 119879
3(120591) = [1 minus
cos 120591]2[1 + (12) cos 120591] Dually maximally flat nonnegativecosine waveform with 119886
1gt 0 (minimum at 120591
0= 120587) reads
1198793(120591) = [1 + cos 120591]2[1 minus (12) cos 120591]In what follows we provide relations between coefficients
1198861and 1198863of nonnegative cosine waveforms of type (167) with
at least one zeroFor 1198861le 0 conversion of (168) into an additive form
immediately leads to the following relation
1198861= minus1 minus 119886
3for minus 1 le 119886
3le1
8 (171)
Conversion of (170) into an additive form leads to 1198861
=
minus31198863(2 cos 2120591
0+ 1) which can be also expressed as 119886
1=
minus31198863(4cos2120591
0minus 1) For 119886
1le 0 relations |120591
0| le 1205873 119886
1=
minus31198863(4cos2120591
0minus 1) and 119886
3= [8cos3120591
0]minus1 lead to
1198861= minus3 [ 3radic119886
3minus 1198863] for 1
8le 1198863le 1 (172)
Similarly for 1198861ge 0 conversion of (169) into an additive form
leads to the following relation
1198861= 1 minus 119886
3for minus
1
8le 1198863le 1 (173)
Mathematical Problems in Engineering 21
1
15
15
05
0
minus05
minus15minus15
minus1
minus1 minus05 0 05 1
Coefficient a1
Coe
ffici
enta
3 a1= minus
3(3radica3minus a3)
a1 =
minus1 minus a3
a1 =
1 minusa3
a1 =3(3radic|a3
| +a3)
Figure 14 Parameter space of cosine waveforms for 119896 = 3
For waveform of type (170) with 1198861ge 0 relations 120591
0isin [21205873
41205873] 1198861= minus3119886
3(4cos2120591
0minus 1) and 119886
3= [8cos3120591
0]minus1 lead to
1198861= 3 [
3radic10038161003816100381610038161198863
1003816100381610038161003816 + 1198863] for minus 1 le 1198863le minus
1
8 (174)
Every cosine waveform of type (167) corresponds to apair of real numbers (119886
1 1198863) and vice versa Points (119886
1 1198863)
in grey area in Figure 14 correspond to nonnegative cosinewaveforms for 119896 = 3 The points at the boundary of grey areacorrespond to nonnegative cosinewaveformswith at least onezero A number of shapes of nonnegative cosine waveformswith 119896 = 3 and at least one zero plotted on interval [minus120587 120587]are also presented in Figure 14 The boundary of grey area inFigure 14 consists of four line segments described by relations(171)ndash(174) The common point of line segments (172) and(173) is cusp point with coordinates 119886
1= 0 and 119886
3= 1
Another cusp point with coordinates 1198861= 0 and 119886
3= minus1
is the common point of line segments (171) and (174) Thecommon point of line segments (171)-(172) has coordinates(minus98 18) and common point of line segments (173)-(174)has coordinates (98 minus18) These points are representedby white circle dots and they correspond to maximallyflat cosine waveforms (eg see [21]) White triangle dotswith coordinates (2radic3 minusradic39) and (minus2radic3radic39) refer tothe nonnegative cosine waveforms with maximum value ofamplitude of fundamental harmonic
7 Four Case Studies of Usage of NonnegativeWaveforms in PA Efficiency Analysis
In this section we provide four case studies of usage ofdescription of nonnegative waveforms with fundamental and119896th harmonic in PA efficiency analysis In first two casestudies to be presented in Section 71 voltage is nonnegativewaveform with fundamental and second harmonic with atleast one zero In remaining two case studies to be consideredin Section 72 voltage waveform contains fundamental andthird harmonic
i
v Load
Vdc
Idc
Lch
vL
iL
Cb
+ +in
Figure 15 Generic PA circuit diagram
Let us consider generic PA circuit diagram as shown inFigure 15 We assume here that voltage and current wave-forms at the transistor output are
V (120579) = 1 + 1198861V cos 120579 + 1198871V sin 120579 + 119886119896V cos 119896120579 + 119887119896V sin 119896120579
119894 (120579) = 1 + 1198861119894cos 120579 +
infin
sum
119899=2
119886119899119894cos 119899120579
(175)
where 120579 stands for 120596119905 Both waveforms are normalized in thesense that dc components of voltage and current are 119881dc =
1 and 119868dc = 1 respectively Under assumption that blockingcapacitor 119862
119887behaves as short-circuit at the fundamental and
higher harmonics current and voltage waveforms at the loadare
V119871(120579) = 119886
1V cos 120579 + 1198871V sin 120579 + 119886119896V cos 119896120579 + 119887119896V sin 119896120579
119894119871(120579) = minus119886
1119894cos 120579 minus
infin
sum
119899=2
119886119899119894cos 119899120579
(176)
In terms of coefficients of voltage and current waveforms theload impedance at fundamental harmonic is 119911
1= minus(119886
1V minus
1198951198871V)1198861119894 whereas load impedance at 119896th harmonic is 119911
119896=
minus(119886119896Vminus119895119887119896V)119886119896119894 All other harmonics are short-circuited (119911
119899=
0 for 119899 = 1 and 119899 = 119896) Time average output power of PA (egsee [10]) with waveform pair (175) at fundamental frequencycan be expressed as
1198751= minus
11988611198941198861V
2 (177)
For normalized waveforms (175) with 119881dc = 1 and 119868dc = 1dc power is 119875dc = 1 Consequently PA efficiency 120578 = 119875
1119875dc
(eg see [10 26]) is equal to
120578 = minus11988611198941198861V
2 (178)
Thus time average output power 1198751of PA with pair of nor-
malized waveform (175) is equal to efficiency (178)Power utilization factor (PUF) is defined [26] as ldquothe
ratio of power delivered in a given situation to the power
22 Mathematical Problems in Engineering
delivered by the same device with the same supply voltagein Class A moderdquo Since the output power in class-A modeis 1198751class-A = max[V(120579)] sdotmax[119894(120579)]8 (eg see [9]) it follows
that power utilization factor PUF = 11987511198751class-A for PA with
pair of normalized waveforms (175) can be expressed as
PUF =8120578
max [V (120579)] sdotmax [119894 (120579)] (179)
71 NonnegativeWaveforms for 119896 = 2 in PAEfficiencyAnalysisIn this subsection we provide two case studies of usage ofdescription of nonnegative waveforms with fundamental andsecond harmonic (119896 = 2) in PA efficiency analysis For moreexamples of usage of descriptions of nonnegative waveformswith fundamental and second harmonic in PA efficiencyanalysis see [12]
Case Study 71 In this case study we consider efficiency of PAfor given second harmonic impedance providing that voltageis nonnegative waveform with fundamental and second har-monic and current is ldquohalf-sinerdquo waveform frequently used inefficiency analysis of classical PA operation (eg see [10])
Standard model of current waveform for classical PAoperation has the form (eg see [10 26])
119894119863(120579) =
119868119863[cos 120579 minus cos(120572
2)] |120579| le
120572
2
0120572
2le |120579| le 120587
(180)
where 120572 is conduction angle and 119868119863
gt 0 Since 119894119863(120579) is
even function it immediately follows that its Fourier seriescontains only dc component and cosine terms
119894119863(120579) = 119868dc +
infin
sum
119899=1
119868119899cos 119899120579 (181)
The dc component of the waveform (180) is
119868dc =119868119863120572
2120587[sinc(120572
2) minus cos(120572
2)] (182)
where sinc119909 = (sin119909)119909 The coefficient of the fundamentalharmonic component reads
1198681=119868119863120572
2120587(1 minus sinc120572) (183)
and the coefficient of 119899th harmonic component can bewrittenin the form
119868119899=119868119863
119899120587[sin ((119899 minus 1) 1205722)
(119899 minus 1)minussin ((119899 + 1) 1205722)
(119899 + 1)] 119899 ge 2
(184)
For ldquohalf-sinerdquo current waveform conduction angle is equalto 120587 (class-B conduction angle) According to (182) thisfurther implies that 119868dc = 119868
119863120587 To obtain normalized form of
waveform (180) we set 119868dc = 1which implies that 119868119863= 120587 Fur-
thermore substitution of 120572 = 120587 and 119868119863= 120587 in (180) leads to
119894 (120579) =
120587 cos 120579 |120579| lt120587
2
0120587
2lt |120579| le 120587
(185)
Similarly substitution of 119868119863= 120587 and 120572 = 120587 into (183) and
(184) leads to the coefficients of waveform (185) Coefficientsof fundamental and second harmonic respectively are
1198861119894=120587
2 119886
2119894=2
3 (186)
On the other hand voltage waveform of type (35) for 119896 =2 reads
V (120579) = 1 + 1198861V cos 120579 + 1198871V sin 120579 + 1198862V cos 2120579 + 1198872V sin 2120579
(187)
This waveform contains only fundamental and second har-monic and therefore all harmonics of order higher thantwo are short-circuited (119911
119899= 0 for 119899 gt 2) For current
voltage pair (185) and (187) load impedance at fundamentalharmonic is 119911
1= minus(119886
1V minus 1198951198871V)1198861119894 whereas load impedance
at second harmonic is 1199112= minus(119886
2Vminus1198951198872V)1198862119894 According to ourassumption the load is passive and therefore Re119911
1 gt 0 and
Re1199112 ge 0 which further imply 119886
11198941198861V lt 0 and 119886
21198941198862V le 0
respectivelyIt is easy to see that problem of findingmaximal efficiency
of PA with current-voltage pair (185) and (187) for prescribedsecond harmonic impedance can be reduced to the problemof finding voltage waveform of type (187) with maximal coef-ficient |119886
1V| for prescribed coefficients of second harmonic(see Section 5)
The following algorithm (analogous to Algorithm 22presented in [12]) provides the procedure for calculation ofmaximal efficiency with current-voltage pair (185) and (187)for prescribed second harmonic impedance The definitionof function atan 2(119910 119909) which appears in the step (iii) of thefollowing algorithm is given by (105)
Algorithm 32 (i) Choose 1199112= 1199032+1198951199092such that |119911
2| le 1|119886
2119894|
(ii) calculate 1198862V minus 1198951198872V = minus119911
21198862119894and 120582
2V = radic11988622V + 1198872
2V(iii) if 2120582
2V le 1 minus 1198862V then calculate 119886
1V = minus1 minus 1198862V and
1198871V = minus2119887
2V else calculate 1205821V = radic81205822V(1 minus 1205822V) 1205790V minus 120585V2 =
(12)atan2(1198872V 1198862V) 1198861V = minus120582
1V cos(1205790V minus 120585V2) and 1198871V =
minus1205821V sin(1205790V minus 120585V2)(iv) calculate efficiency 120578 = minus119886
11198941198861V2
(v) calculate 1199111= minus(119886
1V minus 1198951198871V)1198861119894 and 1199112119899 = 1199112Re119911
1
In this case study coefficients of fundamental and secondharmonic of current waveform are given by (186) Maximalefficiency of PA associated with the waveform pair (185)and (187) as a function of normalized second harmonicimpedance 119911
2119899= 1199112Re119911
1 is presented in Figure 16(a)
As can be seen from Figure 16(a) efficiency of 078 isachieved at the edge of Smith chart where second harmonicimpedance has small resistive part Corresponding PUFcalculated according to (179) is presented in Figure 16(b)Peak efficiency 120578 = 1205874 = 07854 and peak value of PUF = 1
are attained when second harmonic is short-circuited (whichcorresponds to ideal class-B operation [10 26])
For example for second harmonic impedance 1199112= 01 minus
11989505 and currentwaveform (185) fromAlgorithm32 it followsthat 2120582
2V le 1 minus 1198862V Furthermore according to step (iii)
of above algorithm maximal efficiency of PA is attained
Mathematical Problems in Engineering 23
+j2
+j5
+j1
+j2
+j5
minusj2
minusj5
minusj1
minusj2
minusj5
infin0
078
02
05
1 2 5
075 07 065
06 05
075
07
065
06
05
120578 lt 05
(a)
099
095
091
083 075
067
0 59 051
051
099
095
091
083
075 0
67 059
+j5
+j1
+j2
+j5
minusj5
minusj1
minusj2
minusj5
infin
+j2
minusj2
0 02
05
1 2 5
120578 lt 05
(b)
Figure 16 (a) Contours of maximal efficiency of PA and (b) contours of corresponding PUF associated with the waveform pair (185) and(187) as functions of normalized second harmonic impedance 119911
2119899= 1199112Re119911
1
2
3
4
1
0
2 3 410
Wav
efor
ms
Angle 120579120587
i(120579)
(120579)
Figure 17 Waveform pair (185) and (187) that provides maximalefficiency for 119911
2= 01 minus 11989505
with voltage waveform of type (187) with coefficients 1198862V =
minus00667 1198872V = minus03333 119886
1V = minus09333 and 1198871V = 06667 (see
Figure 17) Corresponding efficiency PUF and normalizedsecond harmonic impedance are 120578 = 07330 PUF = 07572and 1199112119899= 01683 minus 11989508415 respectively
On the other hand for second harmonic impedance 1199112=
01 minus 11989508 and current waveform (185) from Algorithm 32 itfollows that 2120582
2V gt 1 minus 1198862V Then according to step (iii) of
above algorithm maximal efficiency is attained with voltagewaveform of type (187) with coefficients 119886
2V = minus00667 1198872V =
minus05333 1198861V = minus09333 and 119887
1V = 10572 (see Figure 18)Efficiency PUF andnormalized secondharmonic impedanceare 120578 = 07330 PUF = 06332 and 119911
2119899= 01683 minus 11989513465
respectively
2
3
4
1
0
2 3 410
Wav
efor
ms
Angle 120579120587
i(120579)
(120579)
Figure 18 Waveform pair (185) and (187) that provides maximalefficiency for 119911
2= 01 minus 11989508
Case Study 72 As another case study let us consider the effi-ciency of PA providing that current waveform is nonnegativecosine waveform up to third harmonic with maximum valueof amplitude of fundamental harmonic [22] (see also [8])
119894 (120579) = 1 +1 + radic5
2cos 120579 + 2radic5
5cos 2120579 + 5 minus radic5
10cos 3120579
(188)
and voltage waveform is nonnegative waveform of type(187) Load impedances at fundamental second and thirdharmonic are 119911
1= minus(119886
1Vminus1198951198871V)1198861119894 1199112 = minus(1198862Vminus1198951198872V)1198862119894 and
1199113= 0 respectively According to our assumption the load
is passive and therefore Re1199111 gt 0 and Re119911
2 ge 0 which
further imply 11988611198941198861V lt 0 and 119886
21198941198862V le 0 respectively
Because current waveform (188) contains only cosineterms and voltage waveform is the same as in previous case
24 Mathematical Problems in Engineering
08
075
07
06
06
065
065
05
05
075 0
7
120578 lt 05
+j2
+j5
+j1
+j2
+j5
minusj2
minusj5
minusj1
minusj2
minusj5
infin0
02
05
1 2 5
(a)
085
08 0
75 07 0
6
05 04
08
075
0405
060
7
120578 lt 05
+j2
+j5
+j1
+j2
+j5
minusj2
minusj5
minusj1
minusj2
minusj5
infin0 02
05
1 2 5
(b)
Figure 19 (a) Contours of maximal efficiency of PA and (b) contours of corresponding PUF associated with the waveform pair (187)-(188)as functions of normalized second harmonic impedance 119911
2119899= 1199112Re119911
1
study the procedure for calculation of maximal efficiency ofPA with waveform pair (187)-(188) is the same as presentedin Algorithm 32 In this case study the coefficients of funda-mental and second harmonic of current waveform are 119886
1119894=
(1 + radic5)2 and 1198862119894= 2radic55 respectively
Maximal efficiency of PA associated with the waveformpair (187)-(188) as a function of normalized secondharmonicimpedance 119911
2119899= 1199112Re119911
1 is presented in Figure 19(a)
Efficiency of 08 is achieved at the edge of Smith chart wheresecond harmonic impedance has small resistive part Thetheoretical upper bound 120578 = (1 + radic5)4 asymp 08090 isattained when second harmonic is short-circuitedWhen thisupper bound is reached both second and third harmonic areshort-circuited which implies that we are dealing with finiteharmonic class-C [6 8] or dually when current and voltageinterchange their roles with finite harmonic inverse class-C[6 9] Corresponding PUF calculated according to (179) ispresented in Figure 19(b) Peak value of PUF asymp 08541 isattained when second harmonic is short-circuited
For example for second harmonic impedance 1199112
=
007 minus 11989504 and current waveform (188) from Algorithm 32 itfollows that 2120582
2V le 1 minus 1198862V Furthermore according to step
(iii) of Algorithm 32 maximal efficiency of PA is attainedwith voltage waveform of type (187) with coefficients 119886
2V =
minus00626 1198872V = minus03578 119886
1V = minus09374 and 1198871V = 07155 (see
Figure 20) Corresponding efficiency PUF and normalizedsecond harmonic impedance are 120578 = 07584 PUF = 06337and 1199112119899= 01208 minus 11989506904 respectively
On the other hand for 1199112= 005 minus 11989507 and current
waveform (187) it follows that 21205822V gt 1minus119886
2VThen accordingto step (iii) of Algorithm 32 themaximal efficiency is attainedwith voltage waveform of type (187) with coefficients 119886
2V =
minus00447 1198872V = minus06261 119886
1V = minus09318 and 1198871V = 10007 (see
Figure 21) Efficiency PUF and normalized second harmonic
2
3
4
1
0
2 3 410
Wav
efor
ms
Angle 120579120587
i(120579)
(120579)
Figure 20 Waveform pair (187)-(188) that provides maximal effi-ciency for 119911
2= 007 minus 11989504
impedance are 120578 = 07538 PUF = 05314 and 1199112119899= 00868minus
11989512156 respectively
72 Nonnegative Waveforms for 119896 = 3 in PA EfficiencyAnalysis In this subsection we provide another two casestudies of usage of description of nonnegative waveforms inPA efficiency analysis this time with fundamental and thirdharmonic (119896 = 3)
Case Study 73 Let us consider current-voltage pair such thatvoltage is nonnegative waveform with fundamental and thirdharmonic
V (120579) = 1 + 1198861V cos 120579 + 1198871V sin 120579 + 1198863V cos 3120579 + 1198873V sin 3120579
(189)
Mathematical Problems in Engineering 25
2
3
4
1
0
2 3 410
Wav
efor
ms
Angle 120579120587
i(120579)
(120579)
Figure 21 Waveform pair (187)-(188) that provides maximal effi-ciency for 119911
2= 005 minus 11989507
and current is nonnegative cosine waveform given by (188)Load impedances at fundamental second and third har-monic are 119911
1= minus(119886
1V minus 1198951198871V)1198861119894 1199112 = 0 and 119911
3= minus(119886
3V minus
1198951198873V)1198863119894 respectively According to our assumption the load
is passive and therefore Re1199111 gt 0 and Re119911
3 ge 0 which
further imply 11988611198941198861V lt 0 and 119886
31198941198863V le 0
In this subsection we consider the problem of findingmaximal efficiency of PA with waveform pair (188)-(189)for given third harmonic impedance As we mentionedearlier problem of finding maximal efficiency of PA withcurrent-voltage pair (188)-(189) for prescribed third har-monic impedance can be reduced to the problem of findingvoltage waveform of type (189) withmaximal coefficient |119886
1V|for prescribed coefficients of third harmonic (see Section 52)
The following algorithm provides the procedure forcalculation of maximal efficiency with current-voltage pair(188)-(189) The definition of function atan 2(119910 119909) whichappears in step (iii) of the following algorithm is given by(105)
Algorithm 33 (i) Choose 1199113= 1199033+1198951199093such that |119911
3| le 1|119886
3119894|
(ii) calculate 1198863V minus 1198951198873V = minus119911
31198863119894and 120582
3V = radic11988623V + 1198872
3V(iii) if 271205822
3V le (1 minus 21198863V)3 then calculate 119886
1V = minus1 minus 1198863V
and 1198871V = minus3119887
3V else calculate 1205821V = 3(3radic1205823V minus 120582
3V) 1205790V minus120585V3 = (13)atan 2(119887
3V 1198863V) 1198861V = minus1205821V cos(1205790V minus 120585V3) and
1198871V = minus120582
1V sin(1205790V minus 120585V3)(iv) calculate efficiency 120578 = minus119886
11198941198861V2
(v) calculate 1199111= minus(119886
1V minus 1198951198871V)1198861119894 and 1199113119899 = 1199113Re119911
1
In this case study coefficients of fundamental and thirdharmonic of current waveform are 119886
1119894= (1 + radic5)2 and
1198863119894= (5 minus radic5)10 respectively For the waveform pair (188)-
(189) maximal efficiency of PA as a function of normalizedthird harmonic impedance 119911
3119899= 1199113Re119911
1 is presented in
Figure 22 Efficiency of 08 is reached when third harmonicimpedance has small resistive part Peak efficiency 120578 = (1 +
radic5)4 asymp 08090 is achieved when third harmonic is short-circuited
+j2
+j5
+j1
+j2
+j5
minusj2
minusj5
minusj1
minusj2
minusj5
infin0
08
08
075
075
07
07
06
05
02
05
1 2
Figure 22 Contours of maximal efficiency of PA associated withthe waveform pair (188)-(189) as a function of normalized thirdharmonic impedance 119911
3119899= 1199113Re119911
1
For the present case study in what follows we show thatpower utilization factor is proportional to efficiency For volt-age waveform of type (189) it is easy to see that V(120579 + 120587) =
2minusV(120579) holdsThis relation along with the fact that waveformV(120579) that provides maximal efficiency has at least one zeroimplies that max[V(120579)] = 2 On the other hand currentwaveform (188) is cosine waveform with positive coefficientsand therefore max[119894(120579)] = 119894(0) = 2 + 4radic5 Consequentlyaccording to (179) the following relation holds
PUFCase study 73 = 2 (5 minus 2radic5) 120578 = 10557120578 (190)
Clearly the ratio PUF120578 is constant and therefore in this casestudy PUF can be easily calculated from the correspondingefficiency Accordingly peak efficiency and peak value ofPUFCase study 73 = 3radic52 minus 52 = 08541 are attained forthe same voltage waveform (when third harmonic is short-circuited)
In the first example current waveform (188) and 1199113=
02 minus 11989505 imply that 2712058223V le (1 minus 2119886
3V)3 Then according
to Algorithm 33 the voltage waveform of type (189) thatprovides maximal efficiency has the following coefficients1198863V = minus00553 119887
3V = minus01382 1198861V = minus09447 and 119887
1V =
04146 (see Figure 23) Efficiency PUF and normalized thirdharmonic impedance are 120578 = 07643 PUF = 08069 and1199113119899= 03425 minus 11989508564 respectivelyIn the second example current waveform (188) and 119911
3=
01 minus 11989511 imply that 2712058223V gt (1 minus 2119886
3V)3 Then according
to Algorithm 33 the voltage waveform of type (189) thatprovides maximal efficiency has the following coefficients1198863V = minus00276 119887
3V = minus03040 1198861V = minus09391 and 119887
1V =
05807 (see Figure 24) Efficiency PUF and normalized thirdharmonic impedance are 120578 = 07598 PUF = 08021 and1199113119899= 01723 minus 11989518952 respectively
26 Mathematical Problems in Engineering
2
3
4
1
0
2 3 410
Wav
efor
ms
Angle 120579120587
i(120579)
(120579)
Figure 23 Waveform pair (188)-(189) that provides maximal effi-ciency for 119911
3= 02 minus 11989505
2
3
4
1
0
2 3 410
Wav
efor
ms
Angle 120579120587
i(120579)
(120579)
Figure 24 Waveform pair (188)-(189) that provides maximal effi-ciency for 119911
3= 01 minus 11989511
Case Study 74 In this case study let us consider current-voltage pair where current is normalized waveform of type(180) with conduction angle 120572 = 115120587 (207∘) and voltageis nonnegative waveform of type (189) Substitution of 120572 =
115120587 and 119868dc = 1 into (182) leads to 119868119863
= 22535Furthermore substitution of 120572 = 115120587 and 119868
119863= 22535 into
(180) leads to
119894 (120579)
=
22535 [cos 120579 minus cos(1151205872
)] |120579| le115120587
2
0115120587
2le |120579| le 120587
(191)
Similarly substitution of 120572 = 115120587 and 119868119863
= 22535 into(183) and (184) for 119899 = 3 yields coefficients of fundamentaland third harmonic of waveform (191)
1198861119894= 14586 119886
3119894= minus01026 (192)
Because current waveform (191) contains only cosineterms and voltage waveform is the same as in previous case
+j2
+j5
+j1
+j2
+j5
minusj2
minusj5
minusj1
minusj2
minusj5
infin0 02
05
1 2
074
076
08
084
082
078
076
074
Figure 25 Contours of maximal efficiency of PA associated withthe waveform pair (189) and (191) as a function of normalized thirdharmonic impedance 119911
3119899= 1199113Re119911
1
study the procedure for calculation of maximal efficiencyof PA with waveform pair (189)ndash(191) is the same as thatpresented in Algorithm 33 In this case study the coefficientsof fundamental and third harmonic of current waveform aregiven by (192)
For the waveform pair (189) and (191) maximal efficiencyof PA as a function of normalized third harmonic impedance1199113119899= 1199113Re119911
1 is presented in Figure 25 Efficiency of 084
is obtained in vicinity of 1199113119899= 23685 (corresponding to 119911
3=
18750) Peak efficiency 120578 asymp 08421 is achieved for voltagewaveform of type (189) with coefficients 119886
1V = minus2radic3 1198863V =
radic39 and 1198871V = 1198873V = 0
In the course of finding power utilization factor noticethat currentwaveformof type (191) attains itsmaximumvaluefor 120579 = 0 Insertion of max[119894(120579)] = 119894(0) = 278 andmax[V(120579)] = 2 for voltage waveform of type (189) into (179)leads to
PUFCase study 74 = 1439120578 (193)
Again the ratio PUF120578 is constant and PUF can be easilycalculated from the corresponding efficiency Accordinglypeak value of PUFCase study 74 asymp 12118 and peak efficiencyare attained for the same voltage waveform
In the first example current waveform (191) and 1199113=
1 minus 11989502 imply that 2712058223V le (1 minus 2119886
3V)3 Then according to
Algorithm 33 voltage waveform of type (189) which providesmaximal efficiency has coefficients 119886
3V = 01026 1198873V =
00205 1198861V = minus11026 and 119887
1V = minus00616 (see Figure 26)Efficiency PUF and normalized third harmonic impedanceare 120578 = 08042 PUF = 11572 and 119911
3119899= 13228 minus 11989502646
respectivelyIn second example current waveform (191) and 119911
3=
15 minus 11989512 imply that 2712058223V gt (1 minus 2119886
3V)3 Then according to
Algorithm 33 voltage waveform of type (189) which providesmaximal efficiency has coefficients 119886
3V = 01540 1198873V =
Mathematical Problems in Engineering 27
2
3
4
1
0
2 3 410
Wav
efor
ms
Angle 120579120587
i(120579)
(120579)
Figure 26 Waveform pair (189) and (191) that provides maximalefficiency for 119911
3= 1 minus 11989502
2
3
4
1
0
2 3 410
Wav
efor
ms
Angle 120579120587
i(120579)
(120579)
Figure 27 Waveform pair (189) and (191) that provides maximalefficiency for 119911
3= 15 minus 11989512
01232 1198861V = minus11255 and 119887
1V = minus02575 (see Figure 27)Efficiency PUF and normalized third harmonic impedanceare 120578 = 08208 PUF = 11812 and 119911
3119899= 19439 minus 11989515552
respectively
8 Conclusion
In this paper we consider a problem of finding general de-scriptions of various classes of nonnegative waveforms withfundamental and 119896th harmonic These classes include non-negative waveforms with at least one zero nonnegative wave-forms with maximal amplitude of fundamental harmonic forprescribed amplitude of 119896th harmonic nonnegative wave-forms withmaximal coefficient of cosine part of fundamentalharmonic for prescribed coefficients of 119896th harmonic andnonnegative cosine waveforms with at least one zero Mainresults are stated in six propositions (Propositions 1 6 918 22 and 26) four corollaries (Corollaries 2ndash5) twentyremarks and three algorithms Four case studies of usageof closed form descriptions of nonnegative waveforms in PAefficiency analysis are considered in detail in Section 7
Appendices
Here we provide a list of finite sums of trigonometric func-tions used in this paper (Appendix A) and brief account ofthe Chebyshev polynomials (Appendix B)
A List of Some Finite Sums ofTrigonometric Functions
Dirichlet kernel (eg see [27]) is as follows
119863119896minus1
(120591) = 1 + 2
119896minus1
sum
119899=1
cos 119899120591 = sin ((2119896 minus 1) 1205912)sin (1205912)
(A1)
Fejer kernel (eg see [27]) can be expressed in the fol-lowing equivalent forms
119865119896minus1
(120591) =1
119896
119896minus1
sum
119899=0
119863119899(120591) = 1 +
2
119896
119896minus1
sum
119899=1
(119896 minus 119899) cos 119899120591
=(1 minus cos 119896120591)119896 (1 minus cos 120591)
(A2)
Lagrangersquos trigonometric identity (eg see [28]) is as fol-lows
1198781(120591) =
119896minus1
sum
119899=1
sin 119899120591 = sin (1198961205912) sin ((119896 minus 1) 1205912)sin (1205912)
(A3)
In what follows we show that the following three trigono-metric identities also hold
2
119896minus1
sum
119899=1
(119896 minus 119899) sin 119899120591 = 119896 sin 120591 minus sin 1198961205911 minus cos 120591
(A4)
119896minus1
sum
119899=1
cos (119896 minus 2119899) 120591 = sin (119896 minus 1) 120591sin 120591
(A5)
119896minus1
sum
119899=1
119899 (119896 minus 119899) cos (119896 minus 2119899) 120591
=sin (119896120591) cos 120591 minus 119896 cos (119896120591) sin 120591
2sin3120591
(A6)
Denote 1198782(120591) = 2sum
119896minus1
119899=1(119896 minus 119899) sin 119899120591 119878
3(120591) = sum
119896minus1
119899=1cos(119896 minus
2119899)120591 and 1198784(120591) = sum
119896minus1
119899=1119899(119896 minus 119899) cos(119896 minus 2119899)120591
Notice that 1198782(120591) = 2119896119878
1(120591) + 119889119863
119896minus1(120591)119889120591 which
immediately leads to (A4)
28 Mathematical Problems in Engineering
Identity (A5) can be obtained as follows
sin (119896 minus 1) 120591sin 120591
=119890119895(119896minus1)120591
minus 119890minus119895(119896minus1)120591
119890119895120591 minus 119890minus119895120591
= 119890119895119896120591
119890minus2119895120591
minus 119890minus2119895(119896minus1)120591
1 minus 119890minus2119895120591
= 119890119895119896120591
119896minus1
sum
119899=1
119890minus2119895119899120591
=
119896minus1
sum
119899=1
119890119895(119896minus2119899)120591
=
119896minus1
sum
119899=1
cos (119896 minus 2119899) 120591
(A7)
From 4119899(119896 minus 119899) = 1198962
minus (119896 minus 2119899)2 it follows that 4119878
4(120591) =
1198962
1198783(120591) + 119889
2
1198783(120591)119889120591
2 which leads to (A6)
B The Chebyshev Polynomials
The Chebyshev polynomials of the first kind 119881119899(119909) can be
defined by the following relation (eg see [29])
119881119899(119909) = cos 119899120591 when 119909 = cos 120591 (B1)
The Chebyshev polynomials of the second kind 119880119899(119909) can be
defined by the following relation (eg see [29])
119880119899(119909) =
sin (119899 + 1) 120591sin 120591
when 119909 = cos 120591 (B2)
The Chebyshev polynomials satisfy the following recur-rence relations (eg see [29])
1198810(119909) = 1 119881
1(119909) = 119909
119881119899+1
(119909) = 2119909119881119899(119909) minus 119881
119899minus1(119909)
1198800(119909) = 1 119880
1(119909) = 2119909
119880119899+1
(119909) = 2119909119880119899(119909) minus 119880
119899minus1(119909)
(B3)
The first few Chebyshev polynomials of the first and secondkind are 119881
2(119909) = 2119909
2
minus 1 1198813(119909) = 4119909
3
minus 3119909 1198814(119909) = 8119909
4
minus
81199092
+ 1 1198802(119909) = 4119909
2
minus 1 1198803(119909) = 8119909
3
minus 4119909 and 1198804(119909) =
161199094
minus 121199092
+ 1
Conflict of Interests
The authors declare that there is no conflict of interestsregarding the publication of this paper
Acknowledgment
This work is supported by the SerbianMinistry of EducationScience and Technology Development as a part of ProjectTP32016
References
[1] V I Arnolrsquod V S Afrajmovich Y S Ilrsquoyashenko and L PShilrsquonikov Dynamical Systems V Bifurcation Theory and Catas-trophe Theory Springer Berlin Germany 1994
[2] E Polak ldquoOn the mathematical foundations of nondifferen-tiable optimization in engineering designrdquo SIAM Review vol29 no 1 pp 21ndash89 1987
[3] N S Fuzik ldquoBiharmonic modes of a tuned RF power amplifierrdquoRadiotehnika vol 25 no 7 pp 62ndash71 1970 (Russian)
[4] P Colantonio F Giannini G Leuzzi and E Limiti ldquoClass Gapproach for low-voltage high-efficiency PA designrdquo Interna-tional Journal of RF and Microwave Computer-Aided Engineer-ing vol 10 no 6 pp 366ndash378 2000
[5] F H Raab ldquoMaximum efficiency and output of class-F poweramplifiersrdquo IEEE Transactions on Microwave Theory and Tech-niques vol 49 no 6 pp 1162ndash1166 2001
[6] FH Raab ldquoClass-E class-C and class-F power amplifiers basedupon a finite number of harmonicsrdquo IEEE Transactions onMicrowaveTheory and Techniques vol 49 no 8 pp 1462ndash14682001
[7] J D Rhodes ldquoOutput universality inmaximum efficiency linearpower amplifiersrdquo International Journal of Circuit Theory andApplications vol 31 no 4 pp 385ndash405 2003
[8] A Juhas and L A Novak ldquoComments on lsquoClass-E class-Cand classF power amplifier based upon a finite number ofharmonicsrsquordquo IEEE Transactions on Microwave Theory and Tech-niques vol 57 no 6 pp 1623ndash1625 2009
[9] M Roberg and Z Popovic ldquoAnalysis of high-efficiency poweramplifiers with arbitrary output harmonic terminationsrdquo IEEETransactions on Microwave Theory and Techniques vol 59 no8 pp 2037ndash2048 2011
[10] A Grebennikov N O Sokal and M J Franco Switchmode RFPower Amplifiers ElsevierAcademic Press San Diego CalifUSA 2nd edition 2012
[11] T Canning P J Tasker and S C Cripps ldquoContinuous modepower amplifier design using harmonic clipping contourstheory and practicerdquo IEEE Transactions on Microwave Theoryand Techniques vol 62 no 1 pp 100ndash110 2014
[12] A Juhas and L A Novak ldquoGeneral description of nonnegativewaveforms up to second harmonic for power amplifier mod-ellingrdquoMathematical Problems in Engineering vol 2014 ArticleID 709762 18 pages 2014
[13] V I Arnolrsquod V V Goryunov O V Lyashko and V A VasilrsquoevDynamical Systems VIIImdashSingularity Theory II ApplicationsSpringer Berlin Germany 1993
[14] D Siersma ldquoProperties of conflict sets in the planerdquo BanachCenter Publications Polish Academy of Sciences vol 50 no 1 pp267ndash276 1999 Proceedings of the Banach Center Symposiumon Geometry and Topology of Caustics (Caustics rsquo98) WarsawPoland
[15] M van Manen The geometry of conflict sets [Dissertation]Universiteit Utrecht UtrechtThe Netherlands 2003 httpigi-tur-archivelibraryuunldissertations2003-0912-123058c4pdf
[16] Y L Sachkov ldquoMaxwell strata and symmetries in the problemofoptimal rolling of a sphere over a planerdquo Sbornik Mathematicsvol 201 no 7-8 pp 1029ndash1051 2010
[17] I A Bogaevsky ldquoPerestroikas of shocks and singularities ofminimum functionsrdquoPhysicaDNonlinear Phenomena vol 173no 1-2 pp 1ndash28 2002
[18] Y L Sachkov ldquoMaxwell strata in the Euler elastic problemrdquoJournal of Dynamical andControl Systems vol 14 no 2 pp 169ndash234 2008
[19] M Siino and T Koike ldquoTopological classification of black holesgeneric Maxwell set and crease set of a horizonrdquo InternationalJournal ofModern Physics D Gravitation Astrophysics Cosmol-ogy vol 20 no 6 pp 1095ndash1122 2011
Mathematical Problems in Engineering 29
[20] F H Raab ldquoClass-F power amplifiers with maximally flat wave-formsrdquo IEEETransactions onMicrowaveTheory and Techniquesvol 45 no 11 pp 2007ndash2012 1997
[21] A Juhas and L A Novak ldquoMaximally flat waveforms with finitenumber of harmonics in class-F power amplifiersrdquo Mathemat-ical Problems in Engineering vol 2013 Article ID 169590 9pages 2013
[22] L Fejer ldquoUber trigonometrische polynomerdquo Journal fur dieReine und Angewandte Mathematik vol 1916 no 146 pp 53ndash82 1916 (German)
[23] S C Cripps ldquoBessel Waives [microwave bytes]rdquo IEEE Micro-wave Magazine vol 10 no 7 pp 30ndash36 117 2009
[24] L N Bryzgalova ldquoSingularities of the maximum of para-metrically dependent functionrdquo Functional Analysis and ItsApplications vol 11 no 1 pp 49ndash51 1977
[25] V I Arnold A A Davydov V A Vassiliev and V MZakalyukin Mathematical Models of Catastrophes Control ofCatastrophic Process Encyclopedia of Life Support Systems(EOLSS) EOLSS Publishers Oxford UK 2006
[26] S C Cripps RF Power Amplifiers for Wireless CommunicationsArtech House Norwood Mass USA 2nd edition 2006
[27] A Zygmund Trigonometric Series vol 1 Cambridge UniversityPress Cambridge UK 2nd edition 1959
[28] A Jeffrey and H DaiHandbook of Mathematical Formulas andIntegrals ElsevierAcademic Press San Diego Calif USA 4thedition 2008
[29] J C Mason and D C Handscomb Chebyshev PolynomialsChapman amp Hall CRC Press Boca Raton Fla USA 2003
Submit your manuscripts athttpwwwhindawicom
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical Problems in Engineering
Hindawi Publishing Corporationhttpwwwhindawicom
Differential EquationsInternational Journal of
Volume 2014
Applied MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Probability and StatisticsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical PhysicsAdvances in
Complex AnalysisJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
OptimizationJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
CombinatoricsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Operations ResearchAdvances in
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Function Spaces
Abstract and Applied AnalysisHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of Mathematics and Mathematical Sciences
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Algebra
Discrete Dynamics in Nature and Society
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Decision SciencesAdvances in
Discrete MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom
Volume 2014 Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Stochastic AnalysisInternational Journal of
10 Mathematical Problems in Engineering
1
08
06
04
02
0minus1 minus05 0 05
Am
plitu
de120582k
1
Parameter 120585120587
k = 2k = 3
k = 4k = 5
Figure 5 Amplitude of 119896th harmonic of nonnegative waveformwith two zeros as a function of parameter 120585
where 120582119896is given by (68) and 0 lt |120585| lt 120587 From (83) it follows
that coefficients of fundamental harmonic of nonnegativewaveform of type (35) with two zeros are
1198861= minus1205821cos(120591
0minus120585
119896) 119887
1= minus1205821sin(120591
0minus120585
119896) (84)
where 1205821is amplitude of fundamental harmonic
1205821=
119896 sin 120585sin (120585119896)
120582119896 (85)
Coefficients of 119896th harmonic are given by (45)-(46)Notice that (68) can be rewritten as
120582119896= [cos(120585
119896)
119896 sin 120585sin(120585119896)
minus cos 120585]minus1
(86)
By introducing new variable
119909 = cos(120585119896) (87)
and using the Chebyshev polynomials (eg see Appendices)relations (85) and (86) can be rewritten as
1205821= 119896120582119896119880119896minus1
(119909) (88)
120582119896=
1
119896119909119880119896minus1
(119909) minus 119881119896(119909)
(89)
where119881119896(119909) and119880
119896(119909) denote the Chebyshev polynomials of
the first and second kind respectively From (89) it followsthat
120582119896[119896119909119880119896minus1
(119909) minus 119881119896(119909)] minus 1 = 0 (90)
which is polynomial equation of 119896th degree in terms of var-iable 119909 From 0 lt |120585| lt 120587 and (87) it follows that
cos(120587119896) lt 119909 lt 1 (91)
Since 120582119896is monotonically increasing function of |120585| 0 lt |120585| lt
120587 it follows that 120582119896is monotonically decreasing function of
119909 This further implies that (90) has only one solution thatsatisfies (91) (For 119896 = 2 expression (91) reads cos(1205872) le
119909 lt 1) This solution for 119909 (which can be obtained at leastnumerically) according to (88) leads to amplitude 120582
1of
fundamental harmonicFor 119896 le 4 solutions of (90) and (91) are
119909 = radic1 minus 1205822
21205822
1
3lt 1205822le 1
119909 =1
23radic1205823
1
8lt 1205823lt 1
119909 = radic1
6(1 + radic
51205824+ 3
21205824
)1
15lt 1205824lt 1
(92)
Insertion of (92) into (88) leads to the following relationsbetween amplitude 120582
1of fundamental and amplitude 120582
119896of
119896th harmonic 119896 le 4
1205821= radic8120582
2(1 minus 120582
2)
1
3lt 1205822le 1 (93)
1205821= 3 (
3radic1205823minus 1205823)
1
8lt 1205823lt 1 (94)
1205821= radic
32
27(radic2120582
4(3 + 5120582
4)3
minus 21205824(9 + 7120582
4))
1
15lt 1205824lt 1
(95)
Proof of Proposition 9 As it has been shown earlier (seeProposition 6) nonnegative waveform of type (35) with atleast one zero can be represented in form (38) Since weexclude nonnegative waveforms with 120582
119896= 1 according to
Remark 7 it follows that we exclude case |120585| = 120587Therefore inthe quest for nonnegative waveforms of type (35) having twozeros we will start with waveforms of type (38) for |120585| lt 120587It is clear that nonnegative waveforms of type (38) have twozeros if and only if
120582119896= [max120591
119903119896(120591)]minus1
(96)
and max120591119903119896(120591) = 119903
119896(1205910) According to (64) max
120591119903119896(120591) =
119903119896(1205910) implies |120585| = 0 Therefore it is sufficient to consider
only the interval (69)Substituting (96) into (38) we obtain
119879119896(120591) =
[1 minus cos (120591 minus 1205910)] [max
120591119903119896(120591) minus 119903
119896(120591)]
max120591119903119896(120591)
(97)
Mathematical Problems in Engineering 11
Expression max120591119903119896(120591) minus 119903
119896(120591) according to (64) and (39)
equals
max120591
119903119896(120591) minus 119903
119896(120591) = 119896
sin ((119896 minus 1) 120585119896)sin (120585119896)
minus 2
119896minus1
sum
119899=1
(119896 minus 119899) cos (119899 (120591 minus 1205910) + 120585)
(98)
Comparison of (97) with (66) yields
max120591
119903119896(120591) minus 119903
119896(120591) = [1 minus cos(120591 minus 120591
0+2120585
119896)]
sdot [1198880+ 2
119896minus2
sum
119899=1
119888119899cos 119899(120591 minus 120591
0+120585
119896)]
(99)
where coefficients 119888119899 119899 = 0 119896 minus 2 are given by (67) In
what follows we are going to show that right hand sides of(98) and (99) are equal
From (67) it follows that
1198880minus 1198881cos(120585
119896) = 119896
sin (120585 minus 120585119896)sin (120585119896)
(100)
Also from (67) for 119899 = 1 119896minus3 it follows that the followingrelations hold
(119888119899minus1
+ 119888119899+1
) cos(120585119896) minus 2119888
119899= 2 (119896 minus 119899) cos(120585 minus 119899120585
119896)
(119888119899minus1
minus 119888119899+1
) sin(120585119896) = 2 (119896 minus 119899) sin(120585 minus 119899120585
119896)
(101)
From (99) by using (75) (76) (100)-(101) and trigonometricidentities
cos(120591 minus 1205910+2120585
119896) = cos(120585
119896) cos(120591 minus 120591
0+120585
119896)
minus sin(120585119896) sin(120591 minus 120591
0+120585
119896)
cos(120585 minus 119899120585
119896) cos(119899(120591 minus 120591
0+120585
119896))
minus sin(120585 minus 119899120585
119896) sin(119899(120591 minus 120591
0+120585
119896))
= cos (119899 (120591 minus 1205910) + 120585)
(102)
we obtain (98) Consequently (98) and (99) are equal whichcompletes the proof
33 Nonnegative Waveforms with Two Zeros and PrescribedCoefficients of 119896thHarmonic In this subsectionwe show thatfor prescribed coefficients 119886
119896and 119887119896 there are 119896 nonnegative
waveforms of type (35) with exactly two zeros According to
(37) and (82) coefficients 119886119896and 119887119896of nonnegative waveforms
of type (35) with exactly two zeros satisfy the followingrelation
1
1198962 minus 1lt radic1198862119896+ 1198872119896lt 1 (103)
According to Remark 16 the value of 119909 (see (87)) that cor-responds to 120582
119896= radic1198862
119896+ 1198872119896can be determined from (90)-
(91) As we mentioned earlier (90) has only one solutionthat satisfies (91) This value of 119909 according to (88) leadsto the amplitude 120582
1of fundamental harmonic (closed form
expressions for 1205821in terms of 120582
119896and 119896 le 4 are given by (93)ndash
(95))On the other hand from (45)-(46) it follows that
1198961205910minus 120585 = atan 2 (119887
119896 119886119896) + 2119902120587 119902 = 1 (119896 minus 1) (104)
where function atan 2(119910 119909) is defined as
atan 2 (119910 119909) =
arctan(119910
119909) if 119909 ge 0
arctan(119910
119909) + 120587 if 119909 lt 0 119910 ge 0
arctan(119910
119909) minus 120587 if 119909 lt 0 119910 lt 0
(105)
with the codomain (minus120587 120587] Furthermore according to (84)and (104) the coefficients of fundamental harmonic of non-negative waveforms with two zeros and prescribed coeffi-cients of 119896th harmonic are equal to
1198861= minus1205821cos[
atan 2 (119887119896 119886119896) + 2119902120587
119896]
1198871= minus1205821sin[
atan 2 (119887119896 119886119896) + 2119902120587
119896]
(106)
where 119902 = 0 (119896minus 1) For chosen 119902 according to (104) and(66) positions of zeros are
1205910=1
119896[120585 + atan 2 (119887
119896 119886119896) + 2119902120587]
1205910minus2120585
119896=1
119896[minus120585 + atan 2 (119887
119896 119886119896) + 2119902120587]
(107)
From (106) and 119902 = 0 (119896minus1) it follows that for prescribedcoefficients 119886
119896and 119887119896 there are 119896 nonnegative waveforms of
type (35) with exactly two zerosWe provide here an algorithm to facilitate calculation
of coefficients 1198861and 1198871of nonnegative waveforms of type
(35) with two zeros and prescribed coefficients 119886119896and 119887
119896
providing that 119886119896and 119887119896satisfy (103)
12 Mathematical Problems in Engineering
2
1
0
Wav
efor
ms
0 1 2 3 4
Angle 120591120587
q = 0
q = 1
q = 2
Figure 6 Nonnegative waveforms with two zeros for 119896 = 3 1198863=
minus015 and 1198873= minus02
Algorithm 17 (i) Calculate 120582119896= radic1198862119896+ 1198872119896
(ii) identify 119909 that satisfies both relations (90) and (91)(iii) calculate 120582
1according to (88)
(iv) choose integer 119902 such that 0 le 119902 le 119896 minus 1(v) calculate 119886
1and 1198871according to (106)
For 119896 le 4 by using (93) for 119896 = 2 (94) for 119896 = 3 and (95)for 119896 = 4 it is possible to calculate directly 120582
1from 120582
119896and
proceed to step (iv)For 119896 = 2 and prescribed coefficients 119886
2and 1198872 there are
two waveforms with two zeros one corresponding to 1198861lt 0
and the other corresponding to 1198861gt 0 (see also [12])
Let us take as an input 119896 = 3 1198863= minus015 and 119887
3= minus02
Execution of Algorithm 17 on this input yields 1205823= 025 and
1205821= 11399 (according to (94)) For 119902 = 0 we calculate
1198861= minus08432 and 119887
1= 07670 (corresponding waveform is
presented by solid line in Figure 6) for 119902 = 1 we calculate1198861= minus02426 and 119887
1= minus11138 (corresponding waveform is
presented by dashed line) for 119902 = 2 we calculate 1198861= 10859
and 1198871= 03468 (corresponding waveform is presented by
dotted line)As another example of the usage of Algorithm 17 let us
consider case 119896 = 4 and assume that1198864= minus015 and 119887
4= minus02
Consequently 1205824= 025 and 120582
1= 09861 (according to (95))
For 119902 = 0 3we calculate the following four pairs (1198861 1198871) of
coefficients of fundamental harmonic (minus08388 05184) for119902 = 0 (minus05184 minus08388) for 119902 = 1 (08388 minus05184) for 119902 =2 and (05184 08388) for 119902 = 3 Corresponding waveformsare presented in Figure 7
4 Nonnegative Waveforms with MaximalAmplitude of Fundamental Harmonic
In this section we provide general description of nonnegativewaveforms containing fundamental and 119896th harmonic withmaximal amplitude of fundamental harmonic for prescribedamplitude of 119896th harmonic
The main result of this section is presented in the fol-lowing proposition
3
2
1
0
Wav
efor
ms
0 1 2 3 4
Angle 120591120587
q = 0q = 1
q = 2q = 3
Figure 7 Nonnegative waveforms with two zeros for 119896 = 4 1198864=
minus015 and 1198874= minus02
Proposition 18 Every nonnegativewaveformof type (35)withmaximal amplitude 120582
1of fundamental harmonic and pre-
scribed amplitude 120582119896of 119896th harmonic can be expressed in the
following form
119879119896(120591) = [1 minus cos (120591 minus 120591
0)]
sdot [1 minus (119896 minus 1) 120582119896minus 2120582119896
119896minus1
sum
119899=1
(119896 minus 119899) cos (119899 (120591 minus 1205910))]
(108)
if 0 le 120582119896le 1(119896
2
minus 1) or
119879119896(120591) = 120582
119896[1 minus cos (120591 minus 120591
0)] [1 minus cos(120591 minus 120591
0+2120585
119896)]
sdot [1198880+ 2
119896minus2
sum
119899=1
119888119899cos 119899(120591 minus 120591
0+120585
119896)]
(109)
if 1(1198962 minus 1) le 120582119896le 1 providing that 119888
119899 119899 = 0 119896 minus 2 and
120582119896are related to 120585 via relations (67) and (68) respectively and
|120585| le 120587
Remark 19 Expression (108) can be obtained from (38) bysetting 120585 = 0 Furthermore insertion of 120585 = 0 into (43)ndash(46)leads to the following expressions for coefficients ofwaveformof type (108)
1198861= minus (1 + 120582
119896) cos 120591
0 119887
1= minus (1 + 120582
119896) sin 120591
0
119886119896= 120582119896cos (119896120591
0) 119887
119896= 120582119896sin (119896120591
0)
(110)
On the other hand (109) coincides with (66) Thereforethe expressions for coefficients of (109) and (66) also coincideThus expressions for coefficients of fundamental harmonic ofwaveform (109) are given by (84) where 120582
1is given by (85)
while expressions for coefficients of 119896th harmonic are givenby (45)-(46)
Waveforms described by (108) have exactly one zerowhile waveforms described by (109) for 1(1198962 minus 1) lt 120582
119896lt 1
Mathematical Problems in Engineering 13
14
12
1
08
06
04
02
00 05 1
Amplitude 120582k
Am
plitu
de1205821
k = 2
k = 3
k = 4
Figure 8 Maximal amplitude of fundamental harmonic as a func-tion of amplitude of 119896th harmonic
have exactly two zeros As we mentioned earlier waveforms(109) for 120582
119896= 1 have 119896 zeros
Remark 20 Maximal amplitude of fundamental harmonic ofnonnegative waveforms of type (35) for prescribed amplitudeof 119896th harmonic can be expressed as
1205821= 1 + 120582
119896 (111)
if 0 le 120582119896le 1(119896
2
minus 1) or
1205821=
119896 sin 120585119896 sin 120585 cos (120585119896) minus cos 120585 sin (120585119896)
(112)
if 1(1198962 minus 1) le 120582119896le 1 where 120585 is related to 120582
119896via (68) (or
(86)) and |120585| le 120587From (110) it follows that (111) holds Substitution of (86)
into (85) leads to (112)Notice that 120582
119896= 1(119896
2
minus 1) is the only common point ofthe intervals 0 le 120582
119896le 1(119896
2
minus 1) and 1(1198962
minus 1) le 120582119896le
1 According to (111) 120582119896= 1(119896
2
minus 1) corresponds to 1205821=
1198962
(1198962
minus1) It can be also obtained from (112) by setting 120585 = 0The waveforms corresponding to this pair of amplitudes aremaximally flat nonnegative waveforms
Maximal amplitude of fundamental harmonic of non-negative waveform of type (35) for 119896 le 4 as a function ofamplitude of 119896th harmonic is presented in Figure 8
Remark 21 Maximum value of amplitude of fundamentalharmonic of nonnegative waveform of type (35) is
1205821max =
1
cos (120587 (2119896)) (113)
This maximum value is attained for |120585| = 1205872 (see (112)) Thecorresponding value of amplitude of 119896th harmonic is 120582
119896=
(1119896) tan(120587(2119896)) Nonnegative waveforms of type (35) with1205821= 1205821max have two zeros at 1205910 and 1205910 minus 120587119896 for 120585 = 1205872 or
at 1205910and 1205910+ 120587119896 for 120585 = minus1205872
14
12
1
08
06
04
02
0minus1 minus05 0 05 1
Am
plitu
de1205821
Parameter 120585120587
k = 2k = 3k = 4
Figure 9 Maximal amplitude of fundamental harmonic as a func-tion of parameter 120585
To prove that (113) holds let us first show that the fol-lowing relation holds for 119896 ge 2
cos( 120587
2119896) lt 1 minus
1
1198962 (114)
From 119896 ge 2 it follows that sinc(120587(4119896)) gt sinc(1205874) wheresinc 119909 = (sin119909)119909 and therefore sin(120587(4119896)) gt 1(radic2119896)By using trigonometric identity cos 2119909 = 1 minus 2sin2119909 weimmediately obtain (114)
According to (111) and (112) it is clear that 1205821attains its
maximum value on the interval 1(1198962 minus 1) le 120582119896le 1 Since
120582119896is monotonic function of |120585| on interval |120585| le 120587 (see
Remark 15) it follows that 119889120582119896119889120585 = 0 for 0 lt |120585| lt 120587
Therefore to find critical points of 1205821as a function of 120582
119896
it is sufficient to find critical points of 1205821as a function of
|120585| 0 lt |120585| lt 120587 and consider its values at the end points120585 = 0 and |120585| = 120587 Plot of 120582
1as a function of parameter 120585
for 119896 le 4 is presented in Figure 9 According to (112) firstderivative of 120582
1with respect to 120585 is equal to zero if and only
if (119896 cos 120585 sin(120585119896) minus sin 120585 cos(120585119896)) cos 120585 = 0 On interval0 lt |120585| lt 120587 this is true if and only if |120585| = 1205872 Accordingto (112) 120582
1is equal to 119896
2
(1198962
minus 1) for 120585 = 0 equal to zerofor |120585| = 120587 and equal to 1 cos(120587(2119896)) for |120585| = 1205872 From(114) it follows that 1198962(1198962minus1) lt 1 cos(120587(2119896)) and thereforemaximum value of 120582
1is given by (113) Moreover maximum
value of 1205821is attained for |120585| = 1205872
According to above consideration all nonnegative wave-forms of type (35) having maximum value of amplitude offundamental harmonic can be obtained from (109) by setting|120585| = 1205872 Three of them corresponding to 119896 = 3 120585 = 1205872and three different values of 120591
0(01205876 and1205873) are presented
in Figure 10 Dotted line corresponds to 1205910= 0 (coefficients
of corresponding waveform are 1198861= minus1 119887
1= 1radic3 119886
3= 0
and 1198873= minusradic39) solid line to 120591
0= 1205876 (119886
1= minus2radic3 119887
1= 0
1198863= radic39 and 119887
3= 0) and dashed line to 120591
0= 1205873 (119886
1= minus1
1198871= minus1radic3 119886
3= 0 and 119887
3= radic39)
Proof of Proposition 18 As it has been shown earlier (Propo-sition 6) nonnegative waveform of type (35) with at least
14 Mathematical Problems in Engineering
2
1
0
0 1 2 3 4
Angle 120591120587
Wav
efor
ms
1205910 = 01205910 = 12058761205910 = 1205873
Figure 10 Nonnegative waveforms with maximum amplitude offundamental harmonic for 119896 = 3 and 120585 = 1205872
one zero can be represented in form (38) According to (43)(44) and (36) for amplitude 120582
1of fundamental harmonic of
waveforms of type (38) the following relation holds
1205821= radic(1 + 120582
119896cos 120585)2 + 11989621205822
119896sin2120585 (115)
where 120582119896satisfy (40) and |120585| le 120587
Because of (40) in the quest of finding maximal 1205821for
prescribed 120582119896 we have to consider the following two cases
(Case i)120582119896lt [(119896minus1) cos 120585 + 119896 sin(120585minus120585119896) sin(120585119896)]minus1
(Case ii)120582119896= [(119896minus1) cos 120585+119896 sin(120585minus120585119896) sin(120585119896)]minus1
Case i Since 120582119896lt [(119896 minus 1) cos 120585 + 119896 sin(120585 minus 120585119896) sin(120585119896)]minus1
implies 120582119896
= 1 according to (115) it follows that 1205821
= 0Hence 119889120582
1119889120585 = 0 implies
2120582119896sin 120585 [1 minus (1198962 minus 1) 120582
119896cos 120585] = 0 (116)
Therefore 1198891205821119889120585 = 0 if 120582
119896= 0 (Option 1) or sin 120585 = 0
(Option 2) or (1198962 minus 1)120582119896cos 120585 = 1 (Option 3)
Option 1 According to (115) 120582119896= 0 implies 120582
1= 1 (notice
that this implication shows that 1205821does not depend on 120585 and
therefore we can set 120585 to zero value)
Option 2 According to (115) sin 120585 = 0 implies 1205821= 1 +
120582119896cos 120585 which further leads to the conclusion that 120582
1is
maximal for 120585 = 0 For 120585 = 0 120582119896lt [(119896 minus 1) cos 120585 + 119896 sin(120585 minus
120585119896) sin(120585119896)]minus1 becomes 120582119896lt 1(119896
2
minus 1)
Option 3 This option leads to contradiction To show thatnotice that (119896
2
minus 1)120582119896cos 120585 = 1 and 120582
119896lt [(119896 minus
1) cos 120585 + 119896 sin(120585 minus 120585119896) sin(120585119896)]minus1 imply that (119896 minus 1) cos 120585 gtsin(120585minus120585119896) sin(120585119896) Using (A5) (see Appendices) the latestinequality can be rewritten assum119896minus1
119899=1[cos 120585minuscos((119896minus2119899)120585119896)] gt
0 But from |119896 minus 2119899| lt 119896 119899 = 1 (119896 minus 1) and |120585| le 120587
it follows that all summands are not positive and therefore(119896minus1) cos 120585 gt sin(120585minus120585119896) sin(120585119896) does not hold for |120585| le 120587
Consequently Case i implies 120585 = 0 and 120582119896lt 1(119896
2
minus 1)Finally substitution of 120585 = 0 into (38) leads to (108) whichproves that (108) holds for 120582
119896lt 1(119896
2
minus 1)
Case ii Relation120582119896= [(119896minus1) cos 120585+119896 sin(120585minus120585119896) sin(120585119896)]minus1
according to Proposition 9 and Remark 11 implies that cor-responding waveforms can be expressed via (66)ndash(68) for|120585| le 120587 Furthermore 120582
119896= [(119896 minus 1) cos 120585 + 119896 sin(120585 minus
120585119896) sin(120585119896)]minus1 and |120585| le 120587 imply 1(1198962 minus 1) le 120582119896le 1
This proves that (109) holds for 1(1198962 minus 1) le 120582119896le 1
Finally let us prove that (108) holds for 120582119896= 1(119896
2
minus
1) According to (68) (see also Remark 11) this value of 120582119896
corresponds to 120585 = 0 Furthermore substitution of 120582119896=
1(1198962
minus 1) and 120585 = 0 into (109) leads to (70) which can berewritten as
119879119896(120591) =
[1 minus cos (120591 minus 1205910)]
(1 minus 1198962)
sdot [119896 (119896 minus 1) minus 2
119896minus1
sum
119899=1
(119896 minus 119899) cos (119899 (120591 minus 1205910))]
(117)
Waveform (117) coincides with waveform (108) for 120582119896
=
1(1 minus 1198962
) Consequently (108) holds for 120582119896= 1(1 minus 119896
2
)which completes the proof
5 Nonnegative Waveforms with MaximalAbsolute Value of the Coefficient of CosineTerm of Fundamental Harmonic
In this sectionwe consider general description of nonnegativewaveforms of type (35) with maximal absolute value ofcoefficient 119886
1for prescribed coefficients of 119896th harmonicThis
type of waveform is of particular interest in PA efficiencyanalysis In a number of cases of practical interest eithercurrent or voltage waveform is prescribed In such casesthe problem of finding maximal efficiency of PA can bereduced to the problem of finding nonnegative waveformwith maximal coefficient 119886
1for prescribed coefficients of 119896th
harmonic (see also Section 7)In Section 51 we provide general description of nonneg-
ative waveforms of type (35) with maximal absolute value ofcoefficient 119886
1for prescribed coefficients of 119896th harmonic In
Section 52 we illustrate results of Section 51 for particularcase 119896 = 3
51 Nonnegative Waveforms with Maximal Absolute Value ofCoefficient 119886
1for 119896 ge 2 Waveforms 119879
119896(120591) of type (35) with
1198861ge 0 can be derived from those with 119886
1le 0 by shifting
by 120587 and therefore we can assume without loss of generalitythat 119886
1le 0 Notice that if 119896 is even then shifting 119879
119896(120591) by
120587 produces the same result as replacement of 1198861with minus119886
1
(119886119896remains the same) On the other hand if 119896 is odd then
shifting 119879119896(120591) by 120587 produces the same result as replacement
of 1198861with minus119886
1and 119886119896with minus119886
119896
According to (37) coefficients of 119896th harmonic can beexpressed as
119886119896= 120582119896cos 120575 119887
119896= 120582119896sin 120575 (118)
Mathematical Problems in Engineering 15
where
|120575| le 120587 (119)
Conversely for prescribed coefficients 119886119896and 119887
119896 120575 can be
determined as
120575 = atan 2 (119887119896 119886119896) (120)
where definition of function atan 2(119910 119909) is given by (105)The main result of this section is stated in the following
proposition
Proposition 22 Every nonnegative waveform of type (35)withmaximal absolute value of coefficient 119886
1le 0 for prescribed
coefficients 119886119896and 119887119896of 119896th harmonic can be represented as
119879119896(120591)
= [1 minus cos 120591]
sdot [1 minus (119896 minus 1) 119886119896minus 2
119896minus1
sum
119899=1
(119896 minus 119899) (119886119896cos 119899120591 + 119887
119896sin 119899120591)]
(121)
if 119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) le 1+119886
119896 where 120575 = atan 2(bk
119886119896) or
119879119896(120591) = 120582
119896[1 minus cos(120591 minus (120575 + 120585)
119896)]
sdot [1 minus cos(120591 minus (120575 minus 120585)
119896)]
sdot [1198880+ 2
119896minus2
sum
119899=1
119888119899cos 119899(120591 minus 120575
119896)]
(122)
if 119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) ge 1 + 119886
119896 where 119888
119899 119899 = 0
119896minus2 and 120582119896= radic1198862119896+ 1198872119896are related to 120585 via relations (67) and
(68) respectively and |120585| le 120587
Remark 23 Expression (121) can be obtained from (38) bysetting 120591
0= 0 and 120585 = minus120575 and then replacing 120582
119896cos 120575 with
119886119896(see (118)) and 120582
119896cos(119899120591 minus 120575) with 119886
119896cos 119899120591 + 119887
119896sin 119899120591
(see also (118)) Furthermore insertion of 1205910= 0 and 120585 =
minus120575 into (43)ndash(46) leads to the following relations betweenfundamental and 119896th harmonic coefficients of waveform(121)
1198861= minus (1 + 119886
119896) 119887
1= minus119896119887
119896 (123)
On the other hand expression (122) can be obtained from(66) by replacing 120591
0minus120585119896with 120575119896 Therefore substitution of
1205910minus 120585119896 = 120575119896 in (84) leads to
1198861= minus1205821cos(120575
119896) 119887
1= minus1205821sin(120575
119896) (124)
where 1205821is given by (85)
The fundamental harmonic coefficients 1198861and 1198871of wave-
form of type (35) with maximal absolute value of coefficient1198861le 0 satisfy both relations (123) and (124) if 119886
119896and 119887119896satisfy
1 + 119886119896= 119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) For such waveforms
relations 1205910= 0 and 120585 = minus120575 also hold
Remark 24 Amplitude of 119896th harmonic of nonnegativewaveform of type (35) with maximal absolute value of coeffi-cient 119886
1le 0 and coefficients 119886
119896 119887119896satisfying 1 + 119886
119896=
119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) is
120582119896=
sin (120575119896)119896 sin 120575 cos (120575119896) minus cos 120575 sin (120575119896)
(125)
To show that it is sufficient to substitute 119886119896= 120582119896cos 120575 (see
(118)) into 1 + 119886119896= 119896120582119896[sin 120575 sin(120575119896)] cos(120575119896)
Introducing new variable
119910 = cos(120575119896) (126)
and using the Chebyshev polynomials (eg see Appendices)relations 119886
119896= 120582119896cos 120575 and (125) can be rewritten as
119886119896= 120582119896119881119896(119910) (127)
120582119896=
1
119896119910119880119896minus1
(119910) minus 119881119896(119910)
(128)
where119881119896(119910) and119880
119896(119910) denote the Chebyshev polynomials of
the first and second kind respectively Substitution of (128)into (127) leads to
119886119896119896119910119880119896minus1
(119910) minus (1 + 119886119896) 119881119896(119910) = 0 (129)
which is polynomial equation of 119896th degree in terms of var-iable 119910 From |120575| le 120587 and (126) it follows that
cos(120587119896) le 119910 le 1 (130)
In what follows we show that 119886119896is monotonically increas-
ing function of 119910 on the interval (130) From 120585 = minus120575 (seeRemark 23) and (81) it follows that 120582minus1
119896= (119896 minus 1) cos 120575 +
119896sum119896minus1
119899=1cos((119896 minus 2119899)120575119896) ge 1 and therefore 119886
119896= 120582119896cos 120575 can
be rewritten as
119886119896=
cos 120575(119896 minus 1) cos 120575 + 119896sum119896minus1
119899=1cos ((119896 minus 2119899) 120575119896)
(131)
Obviously 119886119896is even function of 120575 and all cosines in (131)
are monotonically decreasing functions of |120575| on the interval|120575| le 120587 It is easy to show that cos((119896 minus 2119899)120575119896) 119899 =
1 (119896 minus 1) decreases slower than cos 120575 when |120575| increasesThis implies that denominator of the right hand side of(131) decreases slower than numerator Since denominator ispositive for |120575| le 120587 it further implies that 119886
119896is decreasing
function of |120575| on interval |120575| le 120587 Consequently 119886119896is
monotonically increasing function of 119910 on the interval (130)Thus we have shown that 119886
119896is monotonically increasing
function of 119910 on the interval (130) and therefore (129) hasonly one solution that satisfies (130) According to (128) thevalue of 119910 obtained from (129) and (130) either analyticallyor numerically leads to amplitude 120582
119896of 119896th harmonic
16 Mathematical Problems in Engineering
1
05
0
minus05
minus1
minus1 minus05 0 05 1
Coefficient ak
Coe
ffici
entb
k
radica2k+ b2
kle 1
k = 2k = 3k = 4
Figure 11 Plot of (119886119896 119887119896) satisfying 1 + 119886
119896= 119896120582
119896[sin 120575 sin(120575
119896)] cos(120575119896) for 119896 le 4
By solving (129) and (130) for 119896 le 4 we obtain
119910 = radic1 + 1198862
2 (1 minus 1198862) minus1 le 119886
2le1
3
119910 = radic3
4 (1 minus 21198863) minus1 le 119886
3le1
8
119910 =radicradic2 minus 4119886
4+ 1011988624minus 2 (1 minus 119886
4)
4 (1 minus 31198864)
minus1 le 1198864le
1
15
(132)
Insertion of (132) into (128) leads to the following explicitexpressions for the amplitude 120582
119896 119896 le 4
1205822=1
2(1 minus 119886
2) minus1 le 119886
2le1
3 (133)
1205822
3= [
1
3(1 minus 2119886
3)]
3
minus1 le 1198863le1
8 (134)
1205824=1
4(minus1 minus 119886
4+ radic2 minus 4119886
4+ 1011988624) minus1 le 119886
4le
1
15
(135)
Relations (133)ndash(135) define closed lines (see Figure 11) whichseparate points representing waveforms of type (121) frompoints representing waveforms of type (122) For given 119896points inside the corresponding curve refer to nonnegativewaveforms of type (121) whereas points outside curve (andradic1198862119896+ 1198872119896le 1) correspond to nonnegative waveforms of type
(122) Points on the respective curve correspond to the wave-forms which can be expressed in both forms (121) and (122)
Remark 25 Themaximum absolute value of coefficient 1198861of
nonnegative waveform of type (35) is
100381610038161003816100381611988611003816100381610038161003816max =
1
cos (120587 (2119896)) (136)
This maximum value is attained for |120585| = 1205872 and 120575 = 0
(see (124)) Notice that |1198861|max is equal to the maximum value
1205821max of amplitude of fundamental harmonic (see (113))
Coefficients of waveform with maximum absolute value ofcoefficient 119886
1 1198861lt 0 are
1198861= minus
1
cos (120587 (2119896)) 119886
119896=1
119896tan( 120587
(2119896))
1198871= 119887119896= 0
(137)
Waveformdescribed by (137) is cosinewaveformhaving zerosat 120587(2119896) and minus120587(2119896)
In the course of proving (136) notice first that |1198861|max le
1205821max holds According to (123) and (124) maximum of |119886
1|
occurs for 119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) ge 1 + 119886
119896 From (124)
it immediately follows that maximum value of |1198861| is attained
if and only if 1205821= 1205821max and 120575 = 0 which because of
120575119896 = 1205910minus120585119896 further implies 120591
0= 120585119896 Sincemaximumvalue
of 1205821is attained for |120585| = 1205872 it follows that corresponding
waveform has zeros at 120587(2119896) and minus120587(2119896)
Proof of Proposition 22 As it was mentioned earlier in thissection we can assume without loss of generality that 119886
1le 0
We consider waveforms119879119896(120591) of type (35) such that119879
119896(120591) ge 0
and119879119896(120591) = 0 for some 120591
0 Fromassumption that nonnegative
waveform 119879119896(120591) of type (35) has at least one zero it follows
that it can be expressed in form (38)Let us also assume that 120591
0is position of nondegenerate
critical point Therefore 119879119896(1205910) = 0 implies 1198791015840
119896(1205910) = 0 and
11987910158401015840
119896(1205910) gt 0 According to (55) second derivative of 119879
119896(120591) at
1205910can be expressed as 11987910158401015840
119896(1205910) = 1 minus 120582
119896(1198962
minus 1) cos 120585 Since11987910158401015840
119896(1205910) gt 0 it follows immediately that
1 minus 120582119896(1198962
minus 1) cos 120585 gt 0 (138)
Let us further assume that 119879119896(120591) has exactly one zeroThe
problem of finding maximum absolute value of 1198861is con-
nected to the problem of finding maximum of the minimumfunction (see Section 21) If waveforms possess unique globalminimum at nondegenerate critical point then correspond-ing minimum function is a smooth function of parameters[13] Consequently assumption that 119879
119896(120591) has exactly one
zero at nondegenerate critical point leads to the conclusionthat coefficient 119886
1is differentiable function of 120591
0 First
derivative of 1198861(see (43)) with respect to 120591
0 taking into
account that 1205971205851205971205910= 119896 (see (50)) can be expressed in the
following factorized form
1205971198861
1205971205910
= sin 1205910[1 minus 120582
119896(1198962
minus 1) cos 120585] (139)
Mathematical Problems in Engineering 17
From (138) and (139) it is clear that 12059711988611205971205910= 0 if and only if
sin 1205910= 0 According toRemark 12 assumption that119879
119896(120591)has
exactly one zero implies 120582119896lt 1 From (51) (48) and 120582
119896lt 1
it follows that 1198861cos 1205910+ 1198871sin 1205910lt 0 which together with
sin 1205910= 0 implies that 119886
1cos 1205910lt 0 Assumption 119886
1le 0
together with relations 1198861cos 1205910lt 0 and sin 120591
0= 0 further
implies 1198861
= 0 and
1205910= 0 (140)
Insertion of 1205910= 0 into (38) leads to
119879119896(120591)
= [1 minus cos 120591]
sdot [1 minus (119896 minus 1) 120582119896cos 120585 minus 2120582
119896
119896minus1
sum
119899=1
(119896 minus 119899) cos (119899120591 + 120585)]
(141)
Substitution of 1205910= 0 into (45) and (46) yields 119886
119896= 120582119896cos 120585
and 119887119896
= minus120582119896sin 120585 respectively Replacing 120582
119896cos 120585 with
119886119896and 120582
119896cos(119899120591 + 120585) with (119886
119896cos 119899120591 + 119887
119896sin 119899120591) in (141)
immediately leads to (121)Furthermore 119886
119896= 120582119896cos 120585 119887
119896= minus120582
119896sin 120585 and (118)
imply that
120575 = minus120585 (142)
According to (38)ndash(40) and (142) it follows that (141) is non-negative if and only if
120582119896[(119896 minus 1) cos 120575 + 119896 sin (120575 minus 120575119896)
sin (120575119896)] lt 1 (143)
Notice that 119886119896= 120582119896cos 120575 implies that the following relation
holds
120582119896[(119896 minus 1) cos 120575 + 119896 sin (120575 minus 120575119896)
sin (120575119896)]
= minus119886119896+ 119896120582119896
sin 120575sin (120575119896)
cos(120575119896)
(144)
Finally substitution of (144) into (143) leads to 119896120582119896[sin 120575
sin(120575119896)] cos(120575119896) lt 1 + 119886119896 which proves that (121) holds
when 119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) lt 1 + 119886
119896
Apart from nonnegative waveforms with exactly one zeroat nondegenerate critical point in what follows we will alsoconsider other types of nonnegative waveforms with at leastone zero According to Proposition 9 and Remark 11 thesewaveforms can be described by (66)ndash(68) providing that 0 le|120585| le 120587
According to (35) 119879119896(0) ge 0 implies 1 + 119886
1+ 119886119896ge 0
Consequently 1198861le 0 implies that |119886
1| le 1 + 119886
119896 On the other
hand according to (123) |1198861| = 1 + 119886
119896holds for waveforms
of type (121) The converse is also true 1198861le 0 and |119886
1| =
1 + 119886119896imply 119886
1= minus1 minus 119886
119896 which further from (35) implies
119879119896(0) = 0 Therefore in what follows it is enough to consider
only nonnegativewaveformswhich can be described by (66)ndash(68) and 0 le |120585| le 120587 with coefficients 119886
119896and 119887119896satisfying
119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) ge 1 + 119886
119896
For prescribed coefficients 119886119896and 119887119896 the amplitude 120582
119896=
radic1198862119896+ 1198872119896of 119896th harmonic is also prescribed According to
Remark 15 (see also Remark 16) 120582119896is monotonically
decreasing function of 119909 = cos(120585119896) The value of 119909 can beobtained by solving (90) subject to the constraint cos(120587119896) le119909 le 1 Then 120582
1can be determined from (88) From (106) it
immediately follows that maximal absolute value of 1198861le 0
corresponds to 119902 = 0 which from (104) and (120) furtherimplies that
120575 = 1198961205910minus 120585 (145)
Furthermore 119902 = 0 according to (107) implies that waveformzeros are
1205910=(120575 + 120585)
119896 120591
1015840
0= 1205910minus2120585
119896=(120575 minus 120585)
119896 (146)
Substitution of 1205910= (120575 + 120585)119896 into (66) yields (122) which
proves that (122) holds when 119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) ge
1 + 119886119896
In what follows we prove that (121) also holds when119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) = 1 + 119886
119896 Substitution of 119886
119896=
120582119896cos 120575 into 119896120582
119896[sin 120575 sin(120575119896)] cos(120575119896) = 1 + 119886
119896leads to
120582119896[(119896 minus 1) cos 120575 + 119896 sin (120575 minus 120575119896)
sin (120575119896)] = 1 (147)
As we mentioned earlier relation (142) holds for all wave-forms of type (121) Substituting (142) into (147) we obtain
120582119896[(119896 minus 1) cos 120585 + 119896 sin (120585 minus 120585119896)
sin (120585119896)] = 1 (148)
This expression can be rearranged as
120582119896
119896 sin ((119896 minus 1) 120585119896)sin 120585119896
= 1 minus (119896 minus 1) 120582119896cos 120585 (149)
On the other hand for waveforms of type (122) according to(68) relations (148) and (149) also hold Substitution of 120591
0=
(120575 + 120585)119896 (see (145)) and (67) into (122) leads to
119879119896(120591)
= 120582119896[1 minus cos (120591 minus 120591
0)]
sdot [119896 sin ((119896 minus 1) 120585119896)
sin 120585119896minus 2
119896minus1
sum
119899=1
(119896 minus 119899) cos (119899 (120591 minus 1205910) + 120585)]
(150)
Furthermore substitution of (142) into (145) implies that1205910
= 0 Finally substitution of 1205910
= 0 and (149) into(150) leads to (141) Therefore (141) holds when 119896120582
119896[sin 120575
sin(120575119896)] cos(120575119896) = 1 + 119886119896 which in turn shows that (121)
holds when 119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) = 1 + 119886
119896 This
completes the proof
18 Mathematical Problems in Engineering
52 Nonnegative Waveforms with Maximal Absolute Valueof Coefficient 119886
1for 119896 = 3 Nonnegative waveform of type
(35) for 119896 = 3 is widely used in PA design (eg see [10])In this subsection we illustrate results of Section 51 for thisparticular case The case 119896 = 2 is presented in detail in [12]
Coefficients of fundamental harmonic of nonnegativewaveform of type (35) with 119896 = 3 and maximal absolutevalue of coefficient 119886
1le 0 for prescribed coefficients 119886
3and
1198873(1205823= radic11988623+ 11988723) according to (123) (124) (134) (94) and
(120) are equal to
1198861= minus1 minus 119886
3 119887
1= minus3119887
3 (151)
if 12058223le [(1 minus 2119886
3)3]3
1198861= minus1205821cos(120575
3) 119887
1= minus1205821sin(120575
3) (152)
where 1205821= 3(
3radic1205823minus 1205823) and 120575 = atan 2(119887
3 1198863) if [(1 minus
21198863)3]3
le 1205822
3le 1The line 1205822
3= [(1minus2119886
3)3]3 (see case 119896 = 3
in Figure 11) separates points representing waveforms withcoefficients satisfying (151) from points representing wave-forms with coefficients satisfying (152) Waveforms describedby (151) for 1205822
3lt [(1 minus 2119886
3)3]3 have exactly one zero at
1205910= 0 Waveforms described by (151) and (152) for 1205822
3= [(1 minus
21198863)3]3 also have zero at 120591
0= 0 These waveforms as a rule
have exactly two zeros However there are two exceptionsone related to the maximally flat nonnegative waveform withcoefficients 119886
1= minus98 119886
3= 18 and 119887
1= 1198873= 0 which
has only one zero and the other related to the waveform withcoefficients 119886
1= 0 119886
3= minus1 and 119887
1= 1198873= 0 which has three
zerosWaveforms described by (152) for [(1minus21198863)3]3
lt 1205822
3lt
1 have two zeros Waveforms with 1205823= 1 have only third
harmonic (fundamental harmonic is zero)Plot of contours of maximal absolute value of coefficient
1198861 1198861le 0 for prescribed coefficients 119886
3and 1198873is presented
in Figure 12 According to Remark 25 the waveform withmaximum absolute value of 119886
1le 0 is fully described with
the following coefficients 1198861
= minus2radic3 1198863
= radic39 and1198871= 1198873= 0 This waveform has two zeros at plusmn1205876
Two examples of nonnegative waveforms for 119896 = 3
and maximal absolute value of coefficient 1198861 1198861le 0 with
prescribed coefficients 1198863and 1198873are presented in Figure 13
One waveform corresponds to the case 12058223lt [(1 minus 2119886
3)3]3
(solid line) and the other to the case 12058223gt [(1 minus 2119886
3)3]3
(dashed line)Thewaveform represented by solid line has onezero and its coefficients are 119886
3= minus01 119887
3= 01 119886
1= minus09
and 1198871= minus03 Dashed line corresponds to the waveform
having two zeros with coefficients 1198863= minus01 119887
3= 03 119886
1=
minus08844 and 1198871= minus06460 (case 1205822
3gt [(1 minus 2119886
3)3]3)
6 Nonnegative Cosine Waveforms withat Least One Zero
Nonnegative cosine waveforms have proved to be of impor-tance for waveform modelling in PA design (eg see [10])In this section we consider nonnegative cosine waveforms
1
05
0
minus05
minus1
minus1 minus05 0 05 1
Coefficient a3
Coe
ffici
entb
3
02
04
06
08
10
11
Figure 12 Contours ofmaximal absolute value of coefficient 1198861 1198861le
0 as a function of 1198863and 1198873
2
1
0
0 1 2 3 4
Angle 120591120587
Wav
efor
ms
a3 = minus01 b3 = 01
a3 = minus01 b3 = 03
Figure 13 Nonnegative waveforms for 119896 = 3 and maximal absolutevalue of 119886
1 1198861le 0 with prescribed coefficients 119886
3and 1198873
containing fundamental and 119896th harmonic with at least onezero
Cosine waveform with dc component fundamental and119896th harmonic can be obtained from (35) by setting 119887
1= 119887119896=
0 that is
119879119896(120591) = 1 + 119886
1cos 120591 + 119886
119896cos 119896120591 (153)
In Section 61 we provide general description of non-negative cosine waveforms of type (153) with at least onezero We show that nonnegative cosine waveforms with atleast one zero coincide with nonnegative cosine waveformswith maximal absolute value of coefficient 119886
1for prescribed
coefficient 119886119896 In Section 62 we illustrate results of Section 61
for particular case 119896 = 3
61 Nonnegative Cosine Waveforms with at Least One Zerofor 119896 ge 2 Amplitudes of fundamental and 119896th harmonic
Mathematical Problems in Engineering 19
of cosine waveform of type (153) are 1205821= |1198861| and 120582
119896=
|119886119896| respectively According to (42) for nonnegative cosine
waveforms of type (153) the following relation holds
minus1 le 119886119896le 1 (154)
This explains why 119896th harmonic coefficient 119886119896in Proposi-
tion 26 goes through interval [minus1 1]Waveforms (153) with 119886
1ge 0 can be obtained from
waveforms with 1198861le 0 by shifting by 120587 and therefore with-
out loss of generality we can assume that 1198861le 0
Proposition 26 Each nonnegative cosine waveform of type(153) with 119886
1le 0 and at least one zero can be represented as
119879119896(120591) = [1 minus cos 120591] [1 minus (119896 minus 1) 119886
119896minus 2119886119896
119896minus1
sum
119899=1
(119896 minus 119899) cos 119899120591]
(155)
if minus1 le 119886119896le 1(119896
2
minus 1) or
119879119896(120591) = 119886
119896[1 minus cos (120591 minus 120591
0)] [1 minus cos (120591 + 120591
0)]
sdot [1198880+ 2
119896minus2
sum
119899=1
119888119899cos 119899120591]
(156)
where
119888119899=sin ((119896 minus 119899) 120591
0) cos 120591
0minus (119896 minus 119899) cos ((119896 minus 119899) 120591
0) sin 120591
0
sin31205910
(157)
119886119896=
sin 1205910
119896 sin (1198961205910) cos 120591
0minus cos (119896120591
0) sin 120591
0
(158)
100381610038161003816100381612059101003816100381610038161003816 le
120587
119896 (159)
if 1(1198962 minus 1) le 119886119896le 1
Remark 27 Identity [1minuscos(120591minus1205910)][1minuscos(120591+120591
0)] = [cos 120591
0minus
cos 120591]2 implies that (156) can be rewritten as
119879119896(120591) = 119886
119896[cos 120591
0minus cos 120591]2 [119888
0+ 2
119896minus2
sum
119899=1
119888119899cos 119899120591] (160)
Furthermore substitution of (157) into (160) leads to
119879119896(120591) = 119886
119896[cos 120591
0minus cos 120591]
sdot [(119896 minus 1) sin 119896120591
0
sin 1205910
minus 2
119896minus1
sum
119899=1
sin ((119896 minus 119899) 1205910)
sin 1205910
cos 119899120591]
(161)
Remark 28 All nonnegative cosine waveforms of type (153)with at least one zero and 119886
1le 0 except one of them can be
represented either in form (155) or form (156)This exceptionis maximally flat cosine waveform with 119886
1lt 0 which can be
obtained from (155) for 119886119896= 1(119896
2
minus 1) or from (156) for 1205910=
0 Maximally flat cosine waveform with 1198861lt 0 can also be
obtained from (70) by setting 1205910= 0 Furthermore setting
1205910= 0 in (71) leads to maximally flat cosine waveforms for
119896 le 4 and 1198861lt 0
Remark 29 Nonnegative cosine waveform of type (155) with1198861lt 0 and minus1 lt 119886
119896le 1(119896
2
minus 1) has exactly one zero at120591 = 0 Nonnegative cosine waveform described by (156) with1198861lt 0 and 1(1198962 minus 1) lt 119886
119896lt 1 has two zeros at plusmn120591
0 where
0 lt |1205910| lt 120587119896 For 119886
119896= ∓1 nonnegative cosine waveform
of type (153) reduces to 119879119896(120591) = 1 ∓ cos 119896120591 (clearly these two
waveforms both have 119896 zeros)
Remark 30 Transformation of (155) into an additive formleads to the following relation
1198861= minus1 minus 119886
119896 (162)
where minus1 le 119886119896le 1(119896
2
minus1) Similarly transformation of (156)leads to the following relation
1198861= minus119886119896
119896 sin 1198961205910
sin 1205910
(163)
where 119886119896is given by (158) 1(1198962minus1) le 119886
119896le 1 and |120591
0| le 120587119896
Notice that coefficients of maximally flat cosine waveformnamely 119886
119896= 1(119896
2
minus1) and 1198861= minus1198962
(1198962
minus1) satisfy relation(162) They also satisfy relation (163) for 120591
0= 0
Remark 31 Nonnegative cosine waveforms of type (153) withat least one zero coincide with nonnegative cosine waveformswith maximal absolute value of coefficient 119886
1for prescribed
coefficient 119886119896
In proving that Remark 31 holds notice that expression(155) can be obtained from (121) by setting 119887
119896= 0 Fur-
thermore if 119886119896
ge 0 then 120582119896
= 119886119896 which together
with 119887119896
= 0 and (118) implies 120575 = 0 In this case119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) le 1 + 119886
119896becomes 1198962119886
119896le 1 +
119886119896 On the other hand if 119886
119896lt 0 then 120582
119896= minus119886
119896 which
together with 119887119896= 0 and (118) implies |120575| = 120587 In this case
119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) le 1 + 119886
119896becomes 0 le 1 + 119886
119896
Therefore every nonnegative cosine waveform of type (155)has maximal absolute value of coefficient 119886
1for prescribed
coefficient 119886119896 when minus1 le 119886
119896le 1(119896
2
minus 1)Let us now show that expression (156) can be obtained
from (122) by setting 119887119896= 0 and 119886
119896gt 0 For waveforms
of type (122) according to (118) 119887119896= 0 and 119886
119896gt 0 imply
120575 = 0 and 120582119896= 119886119896 Substitution of 120582
119896= 119886119896and 120575 = 0 into
119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) ge 1+119886
119896leads to 119886
119896ge 1(119896
2
minus1)Furthermore substitution of 120575 = 0 into (145) yields 120591
0= 120585119896
Insertion of 120582119896= 119886119896 120575 = 0 and 120591
0= 120585119896 into (122) leads
to (156) Therefore every nonnegative cosine waveform oftype (156) has maximal absolute value of coefficient 119886
1for
prescribed coefficient 119886119896 when 1(1198962 minus 1) le 119886
119896le 1
Proof of Proposition 26 Let us start with nonnegative cosinewaveform of type (153) with 120582
119896= |119886119896| = 1 According to
Remark 7 120582119896= |119886119896| = 1 implies that 120582
1= |1198861| = 0
Substitution of 119886119896
= minus1 into (155) and using (A2) (seeAppendices) lead to 119879
119896(120591) = 1 minus cos 119896120591 Consequently (155)
holds for 119886119896= minus1 On the other hand substitution of 119886
119896= 1
into (158) yields |1205910| = 120587119896 Furthermore substitution of
119886119896= 1 and 120591
0= 120587119896 (or 120591
0= minus120587119896) into (156) along
with performing all multiplications and using (A2) leads to
20 Mathematical Problems in Engineering
119879119896(120591) = 1 + cos 119896120591 Consequently (156)ndash(158) hold for 119886
119896= 1
and |1205910| = 120587119896
It is easy to see that 120582119896= |119886119896| lt 1 and 119879
119896(1205910) = 0 for some
1205910imply 120582
1= |1198861| = 0 Therefore in what follows we assume
that |119886119896| = 1 and 119886
1lt 0
Cosine waveforms are even functions of 120591 Therefore ifnonnegative cosine waveform has exactly one zero it has to beeither at 0 or at 120587 On the other hand if nonnegative cosinewaveform with 119886
1= 0 has exactly two zeros then these zeros
are placed at plusmn1205910 such that 120591
0is neither 0 nor 120587
In order to prove that (155) holds for minus1 lt 119886119896le 1(1minus119896
2
)let us start by referring to the description (38) of nonnegativewaveformswith at least one zero As wementioned earlier fornonnegative cosine waveformwith exactly one zero (denotedby 1205910) it is either 120591
0= 0 or 120591
0= 120587 Therefore in both cases
sin 1205910= 0 Substitution of sin 120591
0= 0 into (43) together with
1198861
= 0 and 120582119896= |119886119896| lt 1 leads to
1205910= 0 (164)
Clearly 1205910= 0 119887
1= 0 and 119887
119896= 0 according to (44) and (46)
imply 120582119896sin 120585 = 0 Since 120582
119896= |119886119896| it follows that |119886
119896| sin 120585 = 0
also holds which further implies 120582119896= 119886119896= 0 or sin 120585 =
0 In the case when 120582119896= 119886119896= 0 from (164) and (43) we
obtain 1198861= minus1 which further implies that 119879
119896(120591) = 1 minus cos 120591
Consequently (155) holds for 119886119896= 0 In the case when sin 120585 =
0 from (164) and (45) we obtain 119886119896= 120582119896if 120585 = 0 or 119886
119896= minus120582119896
if 120585 = 120587 Relations 119886119896= 120582119896and 120585 = 0 according to (40) imply
that 0 le 119886119896le 1(1 minus 119896
2
) Substitution of 120585 = 0 120582119896= 119886119896 and
(164) into (38) leads to (155) which proves that (155) holdsfor 0 le 119886
119896le 1(1 minus 119896
2
) On the other hand relations 119886119896=
minus120582119896and 120585 = 120587 according to (40) imply that minus1 lt 119886
119896le 0
Substitution of 120585 = 120587 120582119896= minus119886119896 and (164) into (38) also leads
to (155) which proves that (155) also holds for minus1 lt 119886119896le 0
Consequently (155) holds for minus1 lt 119886119896le 1(1 minus 119896
2
)In what follows we first prove that (156)-(157) hold for
1(1 minus 1198962
) lt 119886119896lt 1 For this purpose let us start with non-
negative waveforms with two zeros described by (66) As wementioned before nonnegative cosine waveforms with twozeros have zeros at 120591
0and minus120591
0 such that 120591
0= 0 and 120591
0= 120587
Relations 1198861
lt 0 and 1198871
= 0 according to (84) implycos(1205910minus 120585119896) = 1 and therefore
120585
119896= 1205910 (165)
From 120585119896 = 1205910and 0 lt |120585| lt 120587 it follows that 0 lt |120591
0| lt
120587119896 Insertion of 120585119896 = 1205910into (45) yields 119886
119896= 120582119896 Relations
119886119896= 120582119896and (82) imply that 1(1 minus 1198962) lt 119886
119896lt 1 Substitution
of 120582119896= 119886119896and 120585119896 = 120591
0into (66)ndash(68) leads to (156)ndash(158)
which proves that (156)ndash(158) hold for 1(1 minus 1198962
) lt 119886119896lt 1
and 0 lt |1205910| lt 120587119896
Finally substitution of 119886119896= 1(1 minus 119896
2
) and 1205910= 0 into
(161) leads to
119879119896(120591) =
[1 minus cos 120591](1 minus 1198962)
[119896 (119896 minus 1) minus 2
119896minus1
sum
119899=1
(119896 minus 119899) cos 119899120591]
(166)
Waveform (166) coincides with waveform (155) for 119886119896
=
1(1 minus 1198962
) which in turn proves that (156) holds for 119886119896=
1(1 minus 1198962
) and 1205910= 0 This completes the proof
62 Nonnegative CosineWaveforms with at Least One Zero for119896 = 3 In this subsection we consider nonnegative cosinewaveforms with at least one zero for 119896 = 3 (for case 119896 = 2
see [12])Cosine waveform with fundamental and third harmonic
reads
1198793(120591) = 1 + 119886
1cos 120591 + 119886
3cos 3120591 (167)
For 1198861le 0 and minus1 le 119886
3le 18 according to (155) non-
negative cosine waveform of type (167) with at least one zerocan be expressed as
1198793(120591) = (1 minus cos 120591) [1 minus 2119886
3(1 + 2 cos 120591 + cos 2120591)] (168)
From 1198793(120591 + 120587) = 2 minus 119879
3(120591) it immediately follows that for
1198861ge 0 and minus18 le 119886
3le 1 119879
3(120591) can be expressed as
1198793(120591) = (1 + cos 120591) [1 + 2119886
3(1 minus 2 cos 120591 + cos 2120591)] (169)
For 1198861le 0 and 18 le 119886
3le 1 from (158) it follows that 119886
3=
[8cos31205910]minus1 This relation along with (160) and (157) further
implies that 1198793(120591) can be expressed as
1198793(120591) =
[cos 1205910minus cos 120591]2 [2 cos 120591
0+ cos 120591]
2cos31205910
(170)
providing that |1205910| le 1205873 From 119879
3(120591 + 120587) = 2 minus 119879
3(120591) it
follows that (170) also holds for 1198861ge 0 and minus1 le 119886
3le minus18
providing that 1205910isin [21205873 41205873]
Maximally flat nonnegative cosinewaveformof type (167)with 119886
1lt 0 (minimum at 120591
0= 0) reads 119879
3(120591) = [1 minus
cos 120591]2[1 + (12) cos 120591] Dually maximally flat nonnegativecosine waveform with 119886
1gt 0 (minimum at 120591
0= 120587) reads
1198793(120591) = [1 + cos 120591]2[1 minus (12) cos 120591]In what follows we provide relations between coefficients
1198861and 1198863of nonnegative cosine waveforms of type (167) with
at least one zeroFor 1198861le 0 conversion of (168) into an additive form
immediately leads to the following relation
1198861= minus1 minus 119886
3for minus 1 le 119886
3le1
8 (171)
Conversion of (170) into an additive form leads to 1198861
=
minus31198863(2 cos 2120591
0+ 1) which can be also expressed as 119886
1=
minus31198863(4cos2120591
0minus 1) For 119886
1le 0 relations |120591
0| le 1205873 119886
1=
minus31198863(4cos2120591
0minus 1) and 119886
3= [8cos3120591
0]minus1 lead to
1198861= minus3 [ 3radic119886
3minus 1198863] for 1
8le 1198863le 1 (172)
Similarly for 1198861ge 0 conversion of (169) into an additive form
leads to the following relation
1198861= 1 minus 119886
3for minus
1
8le 1198863le 1 (173)
Mathematical Problems in Engineering 21
1
15
15
05
0
minus05
minus15minus15
minus1
minus1 minus05 0 05 1
Coefficient a1
Coe
ffici
enta
3 a1= minus
3(3radica3minus a3)
a1 =
minus1 minus a3
a1 =
1 minusa3
a1 =3(3radic|a3
| +a3)
Figure 14 Parameter space of cosine waveforms for 119896 = 3
For waveform of type (170) with 1198861ge 0 relations 120591
0isin [21205873
41205873] 1198861= minus3119886
3(4cos2120591
0minus 1) and 119886
3= [8cos3120591
0]minus1 lead to
1198861= 3 [
3radic10038161003816100381610038161198863
1003816100381610038161003816 + 1198863] for minus 1 le 1198863le minus
1
8 (174)
Every cosine waveform of type (167) corresponds to apair of real numbers (119886
1 1198863) and vice versa Points (119886
1 1198863)
in grey area in Figure 14 correspond to nonnegative cosinewaveforms for 119896 = 3 The points at the boundary of grey areacorrespond to nonnegative cosinewaveformswith at least onezero A number of shapes of nonnegative cosine waveformswith 119896 = 3 and at least one zero plotted on interval [minus120587 120587]are also presented in Figure 14 The boundary of grey area inFigure 14 consists of four line segments described by relations(171)ndash(174) The common point of line segments (172) and(173) is cusp point with coordinates 119886
1= 0 and 119886
3= 1
Another cusp point with coordinates 1198861= 0 and 119886
3= minus1
is the common point of line segments (171) and (174) Thecommon point of line segments (171)-(172) has coordinates(minus98 18) and common point of line segments (173)-(174)has coordinates (98 minus18) These points are representedby white circle dots and they correspond to maximallyflat cosine waveforms (eg see [21]) White triangle dotswith coordinates (2radic3 minusradic39) and (minus2radic3radic39) refer tothe nonnegative cosine waveforms with maximum value ofamplitude of fundamental harmonic
7 Four Case Studies of Usage of NonnegativeWaveforms in PA Efficiency Analysis
In this section we provide four case studies of usage ofdescription of nonnegative waveforms with fundamental and119896th harmonic in PA efficiency analysis In first two casestudies to be presented in Section 71 voltage is nonnegativewaveform with fundamental and second harmonic with atleast one zero In remaining two case studies to be consideredin Section 72 voltage waveform contains fundamental andthird harmonic
i
v Load
Vdc
Idc
Lch
vL
iL
Cb
+ +in
Figure 15 Generic PA circuit diagram
Let us consider generic PA circuit diagram as shown inFigure 15 We assume here that voltage and current wave-forms at the transistor output are
V (120579) = 1 + 1198861V cos 120579 + 1198871V sin 120579 + 119886119896V cos 119896120579 + 119887119896V sin 119896120579
119894 (120579) = 1 + 1198861119894cos 120579 +
infin
sum
119899=2
119886119899119894cos 119899120579
(175)
where 120579 stands for 120596119905 Both waveforms are normalized in thesense that dc components of voltage and current are 119881dc =
1 and 119868dc = 1 respectively Under assumption that blockingcapacitor 119862
119887behaves as short-circuit at the fundamental and
higher harmonics current and voltage waveforms at the loadare
V119871(120579) = 119886
1V cos 120579 + 1198871V sin 120579 + 119886119896V cos 119896120579 + 119887119896V sin 119896120579
119894119871(120579) = minus119886
1119894cos 120579 minus
infin
sum
119899=2
119886119899119894cos 119899120579
(176)
In terms of coefficients of voltage and current waveforms theload impedance at fundamental harmonic is 119911
1= minus(119886
1V minus
1198951198871V)1198861119894 whereas load impedance at 119896th harmonic is 119911
119896=
minus(119886119896Vminus119895119887119896V)119886119896119894 All other harmonics are short-circuited (119911
119899=
0 for 119899 = 1 and 119899 = 119896) Time average output power of PA (egsee [10]) with waveform pair (175) at fundamental frequencycan be expressed as
1198751= minus
11988611198941198861V
2 (177)
For normalized waveforms (175) with 119881dc = 1 and 119868dc = 1dc power is 119875dc = 1 Consequently PA efficiency 120578 = 119875
1119875dc
(eg see [10 26]) is equal to
120578 = minus11988611198941198861V
2 (178)
Thus time average output power 1198751of PA with pair of nor-
malized waveform (175) is equal to efficiency (178)Power utilization factor (PUF) is defined [26] as ldquothe
ratio of power delivered in a given situation to the power
22 Mathematical Problems in Engineering
delivered by the same device with the same supply voltagein Class A moderdquo Since the output power in class-A modeis 1198751class-A = max[V(120579)] sdotmax[119894(120579)]8 (eg see [9]) it follows
that power utilization factor PUF = 11987511198751class-A for PA with
pair of normalized waveforms (175) can be expressed as
PUF =8120578
max [V (120579)] sdotmax [119894 (120579)] (179)
71 NonnegativeWaveforms for 119896 = 2 in PAEfficiencyAnalysisIn this subsection we provide two case studies of usage ofdescription of nonnegative waveforms with fundamental andsecond harmonic (119896 = 2) in PA efficiency analysis For moreexamples of usage of descriptions of nonnegative waveformswith fundamental and second harmonic in PA efficiencyanalysis see [12]
Case Study 71 In this case study we consider efficiency of PAfor given second harmonic impedance providing that voltageis nonnegative waveform with fundamental and second har-monic and current is ldquohalf-sinerdquo waveform frequently used inefficiency analysis of classical PA operation (eg see [10])
Standard model of current waveform for classical PAoperation has the form (eg see [10 26])
119894119863(120579) =
119868119863[cos 120579 minus cos(120572
2)] |120579| le
120572
2
0120572
2le |120579| le 120587
(180)
where 120572 is conduction angle and 119868119863
gt 0 Since 119894119863(120579) is
even function it immediately follows that its Fourier seriescontains only dc component and cosine terms
119894119863(120579) = 119868dc +
infin
sum
119899=1
119868119899cos 119899120579 (181)
The dc component of the waveform (180) is
119868dc =119868119863120572
2120587[sinc(120572
2) minus cos(120572
2)] (182)
where sinc119909 = (sin119909)119909 The coefficient of the fundamentalharmonic component reads
1198681=119868119863120572
2120587(1 minus sinc120572) (183)
and the coefficient of 119899th harmonic component can bewrittenin the form
119868119899=119868119863
119899120587[sin ((119899 minus 1) 1205722)
(119899 minus 1)minussin ((119899 + 1) 1205722)
(119899 + 1)] 119899 ge 2
(184)
For ldquohalf-sinerdquo current waveform conduction angle is equalto 120587 (class-B conduction angle) According to (182) thisfurther implies that 119868dc = 119868
119863120587 To obtain normalized form of
waveform (180) we set 119868dc = 1which implies that 119868119863= 120587 Fur-
thermore substitution of 120572 = 120587 and 119868119863= 120587 in (180) leads to
119894 (120579) =
120587 cos 120579 |120579| lt120587
2
0120587
2lt |120579| le 120587
(185)
Similarly substitution of 119868119863= 120587 and 120572 = 120587 into (183) and
(184) leads to the coefficients of waveform (185) Coefficientsof fundamental and second harmonic respectively are
1198861119894=120587
2 119886
2119894=2
3 (186)
On the other hand voltage waveform of type (35) for 119896 =2 reads
V (120579) = 1 + 1198861V cos 120579 + 1198871V sin 120579 + 1198862V cos 2120579 + 1198872V sin 2120579
(187)
This waveform contains only fundamental and second har-monic and therefore all harmonics of order higher thantwo are short-circuited (119911
119899= 0 for 119899 gt 2) For current
voltage pair (185) and (187) load impedance at fundamentalharmonic is 119911
1= minus(119886
1V minus 1198951198871V)1198861119894 whereas load impedance
at second harmonic is 1199112= minus(119886
2Vminus1198951198872V)1198862119894 According to ourassumption the load is passive and therefore Re119911
1 gt 0 and
Re1199112 ge 0 which further imply 119886
11198941198861V lt 0 and 119886
21198941198862V le 0
respectivelyIt is easy to see that problem of findingmaximal efficiency
of PA with current-voltage pair (185) and (187) for prescribedsecond harmonic impedance can be reduced to the problemof finding voltage waveform of type (187) with maximal coef-ficient |119886
1V| for prescribed coefficients of second harmonic(see Section 5)
The following algorithm (analogous to Algorithm 22presented in [12]) provides the procedure for calculation ofmaximal efficiency with current-voltage pair (185) and (187)for prescribed second harmonic impedance The definitionof function atan 2(119910 119909) which appears in the step (iii) of thefollowing algorithm is given by (105)
Algorithm 32 (i) Choose 1199112= 1199032+1198951199092such that |119911
2| le 1|119886
2119894|
(ii) calculate 1198862V minus 1198951198872V = minus119911
21198862119894and 120582
2V = radic11988622V + 1198872
2V(iii) if 2120582
2V le 1 minus 1198862V then calculate 119886
1V = minus1 minus 1198862V and
1198871V = minus2119887
2V else calculate 1205821V = radic81205822V(1 minus 1205822V) 1205790V minus 120585V2 =
(12)atan2(1198872V 1198862V) 1198861V = minus120582
1V cos(1205790V minus 120585V2) and 1198871V =
minus1205821V sin(1205790V minus 120585V2)(iv) calculate efficiency 120578 = minus119886
11198941198861V2
(v) calculate 1199111= minus(119886
1V minus 1198951198871V)1198861119894 and 1199112119899 = 1199112Re119911
1
In this case study coefficients of fundamental and secondharmonic of current waveform are given by (186) Maximalefficiency of PA associated with the waveform pair (185)and (187) as a function of normalized second harmonicimpedance 119911
2119899= 1199112Re119911
1 is presented in Figure 16(a)
As can be seen from Figure 16(a) efficiency of 078 isachieved at the edge of Smith chart where second harmonicimpedance has small resistive part Corresponding PUFcalculated according to (179) is presented in Figure 16(b)Peak efficiency 120578 = 1205874 = 07854 and peak value of PUF = 1
are attained when second harmonic is short-circuited (whichcorresponds to ideal class-B operation [10 26])
For example for second harmonic impedance 1199112= 01 minus
11989505 and currentwaveform (185) fromAlgorithm32 it followsthat 2120582
2V le 1 minus 1198862V Furthermore according to step (iii)
of above algorithm maximal efficiency of PA is attained
Mathematical Problems in Engineering 23
+j2
+j5
+j1
+j2
+j5
minusj2
minusj5
minusj1
minusj2
minusj5
infin0
078
02
05
1 2 5
075 07 065
06 05
075
07
065
06
05
120578 lt 05
(a)
099
095
091
083 075
067
0 59 051
051
099
095
091
083
075 0
67 059
+j5
+j1
+j2
+j5
minusj5
minusj1
minusj2
minusj5
infin
+j2
minusj2
0 02
05
1 2 5
120578 lt 05
(b)
Figure 16 (a) Contours of maximal efficiency of PA and (b) contours of corresponding PUF associated with the waveform pair (185) and(187) as functions of normalized second harmonic impedance 119911
2119899= 1199112Re119911
1
2
3
4
1
0
2 3 410
Wav
efor
ms
Angle 120579120587
i(120579)
(120579)
Figure 17 Waveform pair (185) and (187) that provides maximalefficiency for 119911
2= 01 minus 11989505
with voltage waveform of type (187) with coefficients 1198862V =
minus00667 1198872V = minus03333 119886
1V = minus09333 and 1198871V = 06667 (see
Figure 17) Corresponding efficiency PUF and normalizedsecond harmonic impedance are 120578 = 07330 PUF = 07572and 1199112119899= 01683 minus 11989508415 respectively
On the other hand for second harmonic impedance 1199112=
01 minus 11989508 and current waveform (185) from Algorithm 32 itfollows that 2120582
2V gt 1 minus 1198862V Then according to step (iii) of
above algorithm maximal efficiency is attained with voltagewaveform of type (187) with coefficients 119886
2V = minus00667 1198872V =
minus05333 1198861V = minus09333 and 119887
1V = 10572 (see Figure 18)Efficiency PUF andnormalized secondharmonic impedanceare 120578 = 07330 PUF = 06332 and 119911
2119899= 01683 minus 11989513465
respectively
2
3
4
1
0
2 3 410
Wav
efor
ms
Angle 120579120587
i(120579)
(120579)
Figure 18 Waveform pair (185) and (187) that provides maximalefficiency for 119911
2= 01 minus 11989508
Case Study 72 As another case study let us consider the effi-ciency of PA providing that current waveform is nonnegativecosine waveform up to third harmonic with maximum valueof amplitude of fundamental harmonic [22] (see also [8])
119894 (120579) = 1 +1 + radic5
2cos 120579 + 2radic5
5cos 2120579 + 5 minus radic5
10cos 3120579
(188)
and voltage waveform is nonnegative waveform of type(187) Load impedances at fundamental second and thirdharmonic are 119911
1= minus(119886
1Vminus1198951198871V)1198861119894 1199112 = minus(1198862Vminus1198951198872V)1198862119894 and
1199113= 0 respectively According to our assumption the load
is passive and therefore Re1199111 gt 0 and Re119911
2 ge 0 which
further imply 11988611198941198861V lt 0 and 119886
21198941198862V le 0 respectively
Because current waveform (188) contains only cosineterms and voltage waveform is the same as in previous case
24 Mathematical Problems in Engineering
08
075
07
06
06
065
065
05
05
075 0
7
120578 lt 05
+j2
+j5
+j1
+j2
+j5
minusj2
minusj5
minusj1
minusj2
minusj5
infin0
02
05
1 2 5
(a)
085
08 0
75 07 0
6
05 04
08
075
0405
060
7
120578 lt 05
+j2
+j5
+j1
+j2
+j5
minusj2
minusj5
minusj1
minusj2
minusj5
infin0 02
05
1 2 5
(b)
Figure 19 (a) Contours of maximal efficiency of PA and (b) contours of corresponding PUF associated with the waveform pair (187)-(188)as functions of normalized second harmonic impedance 119911
2119899= 1199112Re119911
1
study the procedure for calculation of maximal efficiency ofPA with waveform pair (187)-(188) is the same as presentedin Algorithm 32 In this case study the coefficients of funda-mental and second harmonic of current waveform are 119886
1119894=
(1 + radic5)2 and 1198862119894= 2radic55 respectively
Maximal efficiency of PA associated with the waveformpair (187)-(188) as a function of normalized secondharmonicimpedance 119911
2119899= 1199112Re119911
1 is presented in Figure 19(a)
Efficiency of 08 is achieved at the edge of Smith chart wheresecond harmonic impedance has small resistive part Thetheoretical upper bound 120578 = (1 + radic5)4 asymp 08090 isattained when second harmonic is short-circuitedWhen thisupper bound is reached both second and third harmonic areshort-circuited which implies that we are dealing with finiteharmonic class-C [6 8] or dually when current and voltageinterchange their roles with finite harmonic inverse class-C[6 9] Corresponding PUF calculated according to (179) ispresented in Figure 19(b) Peak value of PUF asymp 08541 isattained when second harmonic is short-circuited
For example for second harmonic impedance 1199112
=
007 minus 11989504 and current waveform (188) from Algorithm 32 itfollows that 2120582
2V le 1 minus 1198862V Furthermore according to step
(iii) of Algorithm 32 maximal efficiency of PA is attainedwith voltage waveform of type (187) with coefficients 119886
2V =
minus00626 1198872V = minus03578 119886
1V = minus09374 and 1198871V = 07155 (see
Figure 20) Corresponding efficiency PUF and normalizedsecond harmonic impedance are 120578 = 07584 PUF = 06337and 1199112119899= 01208 minus 11989506904 respectively
On the other hand for 1199112= 005 minus 11989507 and current
waveform (187) it follows that 21205822V gt 1minus119886
2VThen accordingto step (iii) of Algorithm 32 themaximal efficiency is attainedwith voltage waveform of type (187) with coefficients 119886
2V =
minus00447 1198872V = minus06261 119886
1V = minus09318 and 1198871V = 10007 (see
Figure 21) Efficiency PUF and normalized second harmonic
2
3
4
1
0
2 3 410
Wav
efor
ms
Angle 120579120587
i(120579)
(120579)
Figure 20 Waveform pair (187)-(188) that provides maximal effi-ciency for 119911
2= 007 minus 11989504
impedance are 120578 = 07538 PUF = 05314 and 1199112119899= 00868minus
11989512156 respectively
72 Nonnegative Waveforms for 119896 = 3 in PA EfficiencyAnalysis In this subsection we provide another two casestudies of usage of description of nonnegative waveforms inPA efficiency analysis this time with fundamental and thirdharmonic (119896 = 3)
Case Study 73 Let us consider current-voltage pair such thatvoltage is nonnegative waveform with fundamental and thirdharmonic
V (120579) = 1 + 1198861V cos 120579 + 1198871V sin 120579 + 1198863V cos 3120579 + 1198873V sin 3120579
(189)
Mathematical Problems in Engineering 25
2
3
4
1
0
2 3 410
Wav
efor
ms
Angle 120579120587
i(120579)
(120579)
Figure 21 Waveform pair (187)-(188) that provides maximal effi-ciency for 119911
2= 005 minus 11989507
and current is nonnegative cosine waveform given by (188)Load impedances at fundamental second and third har-monic are 119911
1= minus(119886
1V minus 1198951198871V)1198861119894 1199112 = 0 and 119911
3= minus(119886
3V minus
1198951198873V)1198863119894 respectively According to our assumption the load
is passive and therefore Re1199111 gt 0 and Re119911
3 ge 0 which
further imply 11988611198941198861V lt 0 and 119886
31198941198863V le 0
In this subsection we consider the problem of findingmaximal efficiency of PA with waveform pair (188)-(189)for given third harmonic impedance As we mentionedearlier problem of finding maximal efficiency of PA withcurrent-voltage pair (188)-(189) for prescribed third har-monic impedance can be reduced to the problem of findingvoltage waveform of type (189) withmaximal coefficient |119886
1V|for prescribed coefficients of third harmonic (see Section 52)
The following algorithm provides the procedure forcalculation of maximal efficiency with current-voltage pair(188)-(189) The definition of function atan 2(119910 119909) whichappears in step (iii) of the following algorithm is given by(105)
Algorithm 33 (i) Choose 1199113= 1199033+1198951199093such that |119911
3| le 1|119886
3119894|
(ii) calculate 1198863V minus 1198951198873V = minus119911
31198863119894and 120582
3V = radic11988623V + 1198872
3V(iii) if 271205822
3V le (1 minus 21198863V)3 then calculate 119886
1V = minus1 minus 1198863V
and 1198871V = minus3119887
3V else calculate 1205821V = 3(3radic1205823V minus 120582
3V) 1205790V minus120585V3 = (13)atan 2(119887
3V 1198863V) 1198861V = minus1205821V cos(1205790V minus 120585V3) and
1198871V = minus120582
1V sin(1205790V minus 120585V3)(iv) calculate efficiency 120578 = minus119886
11198941198861V2
(v) calculate 1199111= minus(119886
1V minus 1198951198871V)1198861119894 and 1199113119899 = 1199113Re119911
1
In this case study coefficients of fundamental and thirdharmonic of current waveform are 119886
1119894= (1 + radic5)2 and
1198863119894= (5 minus radic5)10 respectively For the waveform pair (188)-
(189) maximal efficiency of PA as a function of normalizedthird harmonic impedance 119911
3119899= 1199113Re119911
1 is presented in
Figure 22 Efficiency of 08 is reached when third harmonicimpedance has small resistive part Peak efficiency 120578 = (1 +
radic5)4 asymp 08090 is achieved when third harmonic is short-circuited
+j2
+j5
+j1
+j2
+j5
minusj2
minusj5
minusj1
minusj2
minusj5
infin0
08
08
075
075
07
07
06
05
02
05
1 2
Figure 22 Contours of maximal efficiency of PA associated withthe waveform pair (188)-(189) as a function of normalized thirdharmonic impedance 119911
3119899= 1199113Re119911
1
For the present case study in what follows we show thatpower utilization factor is proportional to efficiency For volt-age waveform of type (189) it is easy to see that V(120579 + 120587) =
2minusV(120579) holdsThis relation along with the fact that waveformV(120579) that provides maximal efficiency has at least one zeroimplies that max[V(120579)] = 2 On the other hand currentwaveform (188) is cosine waveform with positive coefficientsand therefore max[119894(120579)] = 119894(0) = 2 + 4radic5 Consequentlyaccording to (179) the following relation holds
PUFCase study 73 = 2 (5 minus 2radic5) 120578 = 10557120578 (190)
Clearly the ratio PUF120578 is constant and therefore in this casestudy PUF can be easily calculated from the correspondingefficiency Accordingly peak efficiency and peak value ofPUFCase study 73 = 3radic52 minus 52 = 08541 are attained forthe same voltage waveform (when third harmonic is short-circuited)
In the first example current waveform (188) and 1199113=
02 minus 11989505 imply that 2712058223V le (1 minus 2119886
3V)3 Then according
to Algorithm 33 the voltage waveform of type (189) thatprovides maximal efficiency has the following coefficients1198863V = minus00553 119887
3V = minus01382 1198861V = minus09447 and 119887
1V =
04146 (see Figure 23) Efficiency PUF and normalized thirdharmonic impedance are 120578 = 07643 PUF = 08069 and1199113119899= 03425 minus 11989508564 respectivelyIn the second example current waveform (188) and 119911
3=
01 minus 11989511 imply that 2712058223V gt (1 minus 2119886
3V)3 Then according
to Algorithm 33 the voltage waveform of type (189) thatprovides maximal efficiency has the following coefficients1198863V = minus00276 119887
3V = minus03040 1198861V = minus09391 and 119887
1V =
05807 (see Figure 24) Efficiency PUF and normalized thirdharmonic impedance are 120578 = 07598 PUF = 08021 and1199113119899= 01723 minus 11989518952 respectively
26 Mathematical Problems in Engineering
2
3
4
1
0
2 3 410
Wav
efor
ms
Angle 120579120587
i(120579)
(120579)
Figure 23 Waveform pair (188)-(189) that provides maximal effi-ciency for 119911
3= 02 minus 11989505
2
3
4
1
0
2 3 410
Wav
efor
ms
Angle 120579120587
i(120579)
(120579)
Figure 24 Waveform pair (188)-(189) that provides maximal effi-ciency for 119911
3= 01 minus 11989511
Case Study 74 In this case study let us consider current-voltage pair where current is normalized waveform of type(180) with conduction angle 120572 = 115120587 (207∘) and voltageis nonnegative waveform of type (189) Substitution of 120572 =
115120587 and 119868dc = 1 into (182) leads to 119868119863
= 22535Furthermore substitution of 120572 = 115120587 and 119868
119863= 22535 into
(180) leads to
119894 (120579)
=
22535 [cos 120579 minus cos(1151205872
)] |120579| le115120587
2
0115120587
2le |120579| le 120587
(191)
Similarly substitution of 120572 = 115120587 and 119868119863
= 22535 into(183) and (184) for 119899 = 3 yields coefficients of fundamentaland third harmonic of waveform (191)
1198861119894= 14586 119886
3119894= minus01026 (192)
Because current waveform (191) contains only cosineterms and voltage waveform is the same as in previous case
+j2
+j5
+j1
+j2
+j5
minusj2
minusj5
minusj1
minusj2
minusj5
infin0 02
05
1 2
074
076
08
084
082
078
076
074
Figure 25 Contours of maximal efficiency of PA associated withthe waveform pair (189) and (191) as a function of normalized thirdharmonic impedance 119911
3119899= 1199113Re119911
1
study the procedure for calculation of maximal efficiencyof PA with waveform pair (189)ndash(191) is the same as thatpresented in Algorithm 33 In this case study the coefficientsof fundamental and third harmonic of current waveform aregiven by (192)
For the waveform pair (189) and (191) maximal efficiencyof PA as a function of normalized third harmonic impedance1199113119899= 1199113Re119911
1 is presented in Figure 25 Efficiency of 084
is obtained in vicinity of 1199113119899= 23685 (corresponding to 119911
3=
18750) Peak efficiency 120578 asymp 08421 is achieved for voltagewaveform of type (189) with coefficients 119886
1V = minus2radic3 1198863V =
radic39 and 1198871V = 1198873V = 0
In the course of finding power utilization factor noticethat currentwaveformof type (191) attains itsmaximumvaluefor 120579 = 0 Insertion of max[119894(120579)] = 119894(0) = 278 andmax[V(120579)] = 2 for voltage waveform of type (189) into (179)leads to
PUFCase study 74 = 1439120578 (193)
Again the ratio PUF120578 is constant and PUF can be easilycalculated from the corresponding efficiency Accordinglypeak value of PUFCase study 74 asymp 12118 and peak efficiencyare attained for the same voltage waveform
In the first example current waveform (191) and 1199113=
1 minus 11989502 imply that 2712058223V le (1 minus 2119886
3V)3 Then according to
Algorithm 33 voltage waveform of type (189) which providesmaximal efficiency has coefficients 119886
3V = 01026 1198873V =
00205 1198861V = minus11026 and 119887
1V = minus00616 (see Figure 26)Efficiency PUF and normalized third harmonic impedanceare 120578 = 08042 PUF = 11572 and 119911
3119899= 13228 minus 11989502646
respectivelyIn second example current waveform (191) and 119911
3=
15 minus 11989512 imply that 2712058223V gt (1 minus 2119886
3V)3 Then according to
Algorithm 33 voltage waveform of type (189) which providesmaximal efficiency has coefficients 119886
3V = 01540 1198873V =
Mathematical Problems in Engineering 27
2
3
4
1
0
2 3 410
Wav
efor
ms
Angle 120579120587
i(120579)
(120579)
Figure 26 Waveform pair (189) and (191) that provides maximalefficiency for 119911
3= 1 minus 11989502
2
3
4
1
0
2 3 410
Wav
efor
ms
Angle 120579120587
i(120579)
(120579)
Figure 27 Waveform pair (189) and (191) that provides maximalefficiency for 119911
3= 15 minus 11989512
01232 1198861V = minus11255 and 119887
1V = minus02575 (see Figure 27)Efficiency PUF and normalized third harmonic impedanceare 120578 = 08208 PUF = 11812 and 119911
3119899= 19439 minus 11989515552
respectively
8 Conclusion
In this paper we consider a problem of finding general de-scriptions of various classes of nonnegative waveforms withfundamental and 119896th harmonic These classes include non-negative waveforms with at least one zero nonnegative wave-forms with maximal amplitude of fundamental harmonic forprescribed amplitude of 119896th harmonic nonnegative wave-forms withmaximal coefficient of cosine part of fundamentalharmonic for prescribed coefficients of 119896th harmonic andnonnegative cosine waveforms with at least one zero Mainresults are stated in six propositions (Propositions 1 6 918 22 and 26) four corollaries (Corollaries 2ndash5) twentyremarks and three algorithms Four case studies of usageof closed form descriptions of nonnegative waveforms in PAefficiency analysis are considered in detail in Section 7
Appendices
Here we provide a list of finite sums of trigonometric func-tions used in this paper (Appendix A) and brief account ofthe Chebyshev polynomials (Appendix B)
A List of Some Finite Sums ofTrigonometric Functions
Dirichlet kernel (eg see [27]) is as follows
119863119896minus1
(120591) = 1 + 2
119896minus1
sum
119899=1
cos 119899120591 = sin ((2119896 minus 1) 1205912)sin (1205912)
(A1)
Fejer kernel (eg see [27]) can be expressed in the fol-lowing equivalent forms
119865119896minus1
(120591) =1
119896
119896minus1
sum
119899=0
119863119899(120591) = 1 +
2
119896
119896minus1
sum
119899=1
(119896 minus 119899) cos 119899120591
=(1 minus cos 119896120591)119896 (1 minus cos 120591)
(A2)
Lagrangersquos trigonometric identity (eg see [28]) is as fol-lows
1198781(120591) =
119896minus1
sum
119899=1
sin 119899120591 = sin (1198961205912) sin ((119896 minus 1) 1205912)sin (1205912)
(A3)
In what follows we show that the following three trigono-metric identities also hold
2
119896minus1
sum
119899=1
(119896 minus 119899) sin 119899120591 = 119896 sin 120591 minus sin 1198961205911 minus cos 120591
(A4)
119896minus1
sum
119899=1
cos (119896 minus 2119899) 120591 = sin (119896 minus 1) 120591sin 120591
(A5)
119896minus1
sum
119899=1
119899 (119896 minus 119899) cos (119896 minus 2119899) 120591
=sin (119896120591) cos 120591 minus 119896 cos (119896120591) sin 120591
2sin3120591
(A6)
Denote 1198782(120591) = 2sum
119896minus1
119899=1(119896 minus 119899) sin 119899120591 119878
3(120591) = sum
119896minus1
119899=1cos(119896 minus
2119899)120591 and 1198784(120591) = sum
119896minus1
119899=1119899(119896 minus 119899) cos(119896 minus 2119899)120591
Notice that 1198782(120591) = 2119896119878
1(120591) + 119889119863
119896minus1(120591)119889120591 which
immediately leads to (A4)
28 Mathematical Problems in Engineering
Identity (A5) can be obtained as follows
sin (119896 minus 1) 120591sin 120591
=119890119895(119896minus1)120591
minus 119890minus119895(119896minus1)120591
119890119895120591 minus 119890minus119895120591
= 119890119895119896120591
119890minus2119895120591
minus 119890minus2119895(119896minus1)120591
1 minus 119890minus2119895120591
= 119890119895119896120591
119896minus1
sum
119899=1
119890minus2119895119899120591
=
119896minus1
sum
119899=1
119890119895(119896minus2119899)120591
=
119896minus1
sum
119899=1
cos (119896 minus 2119899) 120591
(A7)
From 4119899(119896 minus 119899) = 1198962
minus (119896 minus 2119899)2 it follows that 4119878
4(120591) =
1198962
1198783(120591) + 119889
2
1198783(120591)119889120591
2 which leads to (A6)
B The Chebyshev Polynomials
The Chebyshev polynomials of the first kind 119881119899(119909) can be
defined by the following relation (eg see [29])
119881119899(119909) = cos 119899120591 when 119909 = cos 120591 (B1)
The Chebyshev polynomials of the second kind 119880119899(119909) can be
defined by the following relation (eg see [29])
119880119899(119909) =
sin (119899 + 1) 120591sin 120591
when 119909 = cos 120591 (B2)
The Chebyshev polynomials satisfy the following recur-rence relations (eg see [29])
1198810(119909) = 1 119881
1(119909) = 119909
119881119899+1
(119909) = 2119909119881119899(119909) minus 119881
119899minus1(119909)
1198800(119909) = 1 119880
1(119909) = 2119909
119880119899+1
(119909) = 2119909119880119899(119909) minus 119880
119899minus1(119909)
(B3)
The first few Chebyshev polynomials of the first and secondkind are 119881
2(119909) = 2119909
2
minus 1 1198813(119909) = 4119909
3
minus 3119909 1198814(119909) = 8119909
4
minus
81199092
+ 1 1198802(119909) = 4119909
2
minus 1 1198803(119909) = 8119909
3
minus 4119909 and 1198804(119909) =
161199094
minus 121199092
+ 1
Conflict of Interests
The authors declare that there is no conflict of interestsregarding the publication of this paper
Acknowledgment
This work is supported by the SerbianMinistry of EducationScience and Technology Development as a part of ProjectTP32016
References
[1] V I Arnolrsquod V S Afrajmovich Y S Ilrsquoyashenko and L PShilrsquonikov Dynamical Systems V Bifurcation Theory and Catas-trophe Theory Springer Berlin Germany 1994
[2] E Polak ldquoOn the mathematical foundations of nondifferen-tiable optimization in engineering designrdquo SIAM Review vol29 no 1 pp 21ndash89 1987
[3] N S Fuzik ldquoBiharmonic modes of a tuned RF power amplifierrdquoRadiotehnika vol 25 no 7 pp 62ndash71 1970 (Russian)
[4] P Colantonio F Giannini G Leuzzi and E Limiti ldquoClass Gapproach for low-voltage high-efficiency PA designrdquo Interna-tional Journal of RF and Microwave Computer-Aided Engineer-ing vol 10 no 6 pp 366ndash378 2000
[5] F H Raab ldquoMaximum efficiency and output of class-F poweramplifiersrdquo IEEE Transactions on Microwave Theory and Tech-niques vol 49 no 6 pp 1162ndash1166 2001
[6] FH Raab ldquoClass-E class-C and class-F power amplifiers basedupon a finite number of harmonicsrdquo IEEE Transactions onMicrowaveTheory and Techniques vol 49 no 8 pp 1462ndash14682001
[7] J D Rhodes ldquoOutput universality inmaximum efficiency linearpower amplifiersrdquo International Journal of Circuit Theory andApplications vol 31 no 4 pp 385ndash405 2003
[8] A Juhas and L A Novak ldquoComments on lsquoClass-E class-Cand classF power amplifier based upon a finite number ofharmonicsrsquordquo IEEE Transactions on Microwave Theory and Tech-niques vol 57 no 6 pp 1623ndash1625 2009
[9] M Roberg and Z Popovic ldquoAnalysis of high-efficiency poweramplifiers with arbitrary output harmonic terminationsrdquo IEEETransactions on Microwave Theory and Techniques vol 59 no8 pp 2037ndash2048 2011
[10] A Grebennikov N O Sokal and M J Franco Switchmode RFPower Amplifiers ElsevierAcademic Press San Diego CalifUSA 2nd edition 2012
[11] T Canning P J Tasker and S C Cripps ldquoContinuous modepower amplifier design using harmonic clipping contourstheory and practicerdquo IEEE Transactions on Microwave Theoryand Techniques vol 62 no 1 pp 100ndash110 2014
[12] A Juhas and L A Novak ldquoGeneral description of nonnegativewaveforms up to second harmonic for power amplifier mod-ellingrdquoMathematical Problems in Engineering vol 2014 ArticleID 709762 18 pages 2014
[13] V I Arnolrsquod V V Goryunov O V Lyashko and V A VasilrsquoevDynamical Systems VIIImdashSingularity Theory II ApplicationsSpringer Berlin Germany 1993
[14] D Siersma ldquoProperties of conflict sets in the planerdquo BanachCenter Publications Polish Academy of Sciences vol 50 no 1 pp267ndash276 1999 Proceedings of the Banach Center Symposiumon Geometry and Topology of Caustics (Caustics rsquo98) WarsawPoland
[15] M van Manen The geometry of conflict sets [Dissertation]Universiteit Utrecht UtrechtThe Netherlands 2003 httpigi-tur-archivelibraryuunldissertations2003-0912-123058c4pdf
[16] Y L Sachkov ldquoMaxwell strata and symmetries in the problemofoptimal rolling of a sphere over a planerdquo Sbornik Mathematicsvol 201 no 7-8 pp 1029ndash1051 2010
[17] I A Bogaevsky ldquoPerestroikas of shocks and singularities ofminimum functionsrdquoPhysicaDNonlinear Phenomena vol 173no 1-2 pp 1ndash28 2002
[18] Y L Sachkov ldquoMaxwell strata in the Euler elastic problemrdquoJournal of Dynamical andControl Systems vol 14 no 2 pp 169ndash234 2008
[19] M Siino and T Koike ldquoTopological classification of black holesgeneric Maxwell set and crease set of a horizonrdquo InternationalJournal ofModern Physics D Gravitation Astrophysics Cosmol-ogy vol 20 no 6 pp 1095ndash1122 2011
Mathematical Problems in Engineering 29
[20] F H Raab ldquoClass-F power amplifiers with maximally flat wave-formsrdquo IEEETransactions onMicrowaveTheory and Techniquesvol 45 no 11 pp 2007ndash2012 1997
[21] A Juhas and L A Novak ldquoMaximally flat waveforms with finitenumber of harmonics in class-F power amplifiersrdquo Mathemat-ical Problems in Engineering vol 2013 Article ID 169590 9pages 2013
[22] L Fejer ldquoUber trigonometrische polynomerdquo Journal fur dieReine und Angewandte Mathematik vol 1916 no 146 pp 53ndash82 1916 (German)
[23] S C Cripps ldquoBessel Waives [microwave bytes]rdquo IEEE Micro-wave Magazine vol 10 no 7 pp 30ndash36 117 2009
[24] L N Bryzgalova ldquoSingularities of the maximum of para-metrically dependent functionrdquo Functional Analysis and ItsApplications vol 11 no 1 pp 49ndash51 1977
[25] V I Arnold A A Davydov V A Vassiliev and V MZakalyukin Mathematical Models of Catastrophes Control ofCatastrophic Process Encyclopedia of Life Support Systems(EOLSS) EOLSS Publishers Oxford UK 2006
[26] S C Cripps RF Power Amplifiers for Wireless CommunicationsArtech House Norwood Mass USA 2nd edition 2006
[27] A Zygmund Trigonometric Series vol 1 Cambridge UniversityPress Cambridge UK 2nd edition 1959
[28] A Jeffrey and H DaiHandbook of Mathematical Formulas andIntegrals ElsevierAcademic Press San Diego Calif USA 4thedition 2008
[29] J C Mason and D C Handscomb Chebyshev PolynomialsChapman amp Hall CRC Press Boca Raton Fla USA 2003
Submit your manuscripts athttpwwwhindawicom
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical Problems in Engineering
Hindawi Publishing Corporationhttpwwwhindawicom
Differential EquationsInternational Journal of
Volume 2014
Applied MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Probability and StatisticsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical PhysicsAdvances in
Complex AnalysisJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
OptimizationJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
CombinatoricsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Operations ResearchAdvances in
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Function Spaces
Abstract and Applied AnalysisHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of Mathematics and Mathematical Sciences
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Algebra
Discrete Dynamics in Nature and Society
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Decision SciencesAdvances in
Discrete MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom
Volume 2014 Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Stochastic AnalysisInternational Journal of
Mathematical Problems in Engineering 11
Expression max120591119903119896(120591) minus 119903
119896(120591) according to (64) and (39)
equals
max120591
119903119896(120591) minus 119903
119896(120591) = 119896
sin ((119896 minus 1) 120585119896)sin (120585119896)
minus 2
119896minus1
sum
119899=1
(119896 minus 119899) cos (119899 (120591 minus 1205910) + 120585)
(98)
Comparison of (97) with (66) yields
max120591
119903119896(120591) minus 119903
119896(120591) = [1 minus cos(120591 minus 120591
0+2120585
119896)]
sdot [1198880+ 2
119896minus2
sum
119899=1
119888119899cos 119899(120591 minus 120591
0+120585
119896)]
(99)
where coefficients 119888119899 119899 = 0 119896 minus 2 are given by (67) In
what follows we are going to show that right hand sides of(98) and (99) are equal
From (67) it follows that
1198880minus 1198881cos(120585
119896) = 119896
sin (120585 minus 120585119896)sin (120585119896)
(100)
Also from (67) for 119899 = 1 119896minus3 it follows that the followingrelations hold
(119888119899minus1
+ 119888119899+1
) cos(120585119896) minus 2119888
119899= 2 (119896 minus 119899) cos(120585 minus 119899120585
119896)
(119888119899minus1
minus 119888119899+1
) sin(120585119896) = 2 (119896 minus 119899) sin(120585 minus 119899120585
119896)
(101)
From (99) by using (75) (76) (100)-(101) and trigonometricidentities
cos(120591 minus 1205910+2120585
119896) = cos(120585
119896) cos(120591 minus 120591
0+120585
119896)
minus sin(120585119896) sin(120591 minus 120591
0+120585
119896)
cos(120585 minus 119899120585
119896) cos(119899(120591 minus 120591
0+120585
119896))
minus sin(120585 minus 119899120585
119896) sin(119899(120591 minus 120591
0+120585
119896))
= cos (119899 (120591 minus 1205910) + 120585)
(102)
we obtain (98) Consequently (98) and (99) are equal whichcompletes the proof
33 Nonnegative Waveforms with Two Zeros and PrescribedCoefficients of 119896thHarmonic In this subsectionwe show thatfor prescribed coefficients 119886
119896and 119887119896 there are 119896 nonnegative
waveforms of type (35) with exactly two zeros According to
(37) and (82) coefficients 119886119896and 119887119896of nonnegative waveforms
of type (35) with exactly two zeros satisfy the followingrelation
1
1198962 minus 1lt radic1198862119896+ 1198872119896lt 1 (103)
According to Remark 16 the value of 119909 (see (87)) that cor-responds to 120582
119896= radic1198862
119896+ 1198872119896can be determined from (90)-
(91) As we mentioned earlier (90) has only one solutionthat satisfies (91) This value of 119909 according to (88) leadsto the amplitude 120582
1of fundamental harmonic (closed form
expressions for 1205821in terms of 120582
119896and 119896 le 4 are given by (93)ndash
(95))On the other hand from (45)-(46) it follows that
1198961205910minus 120585 = atan 2 (119887
119896 119886119896) + 2119902120587 119902 = 1 (119896 minus 1) (104)
where function atan 2(119910 119909) is defined as
atan 2 (119910 119909) =
arctan(119910
119909) if 119909 ge 0
arctan(119910
119909) + 120587 if 119909 lt 0 119910 ge 0
arctan(119910
119909) minus 120587 if 119909 lt 0 119910 lt 0
(105)
with the codomain (minus120587 120587] Furthermore according to (84)and (104) the coefficients of fundamental harmonic of non-negative waveforms with two zeros and prescribed coeffi-cients of 119896th harmonic are equal to
1198861= minus1205821cos[
atan 2 (119887119896 119886119896) + 2119902120587
119896]
1198871= minus1205821sin[
atan 2 (119887119896 119886119896) + 2119902120587
119896]
(106)
where 119902 = 0 (119896minus 1) For chosen 119902 according to (104) and(66) positions of zeros are
1205910=1
119896[120585 + atan 2 (119887
119896 119886119896) + 2119902120587]
1205910minus2120585
119896=1
119896[minus120585 + atan 2 (119887
119896 119886119896) + 2119902120587]
(107)
From (106) and 119902 = 0 (119896minus1) it follows that for prescribedcoefficients 119886
119896and 119887119896 there are 119896 nonnegative waveforms of
type (35) with exactly two zerosWe provide here an algorithm to facilitate calculation
of coefficients 1198861and 1198871of nonnegative waveforms of type
(35) with two zeros and prescribed coefficients 119886119896and 119887
119896
providing that 119886119896and 119887119896satisfy (103)
12 Mathematical Problems in Engineering
2
1
0
Wav
efor
ms
0 1 2 3 4
Angle 120591120587
q = 0
q = 1
q = 2
Figure 6 Nonnegative waveforms with two zeros for 119896 = 3 1198863=
minus015 and 1198873= minus02
Algorithm 17 (i) Calculate 120582119896= radic1198862119896+ 1198872119896
(ii) identify 119909 that satisfies both relations (90) and (91)(iii) calculate 120582
1according to (88)
(iv) choose integer 119902 such that 0 le 119902 le 119896 minus 1(v) calculate 119886
1and 1198871according to (106)
For 119896 le 4 by using (93) for 119896 = 2 (94) for 119896 = 3 and (95)for 119896 = 4 it is possible to calculate directly 120582
1from 120582
119896and
proceed to step (iv)For 119896 = 2 and prescribed coefficients 119886
2and 1198872 there are
two waveforms with two zeros one corresponding to 1198861lt 0
and the other corresponding to 1198861gt 0 (see also [12])
Let us take as an input 119896 = 3 1198863= minus015 and 119887
3= minus02
Execution of Algorithm 17 on this input yields 1205823= 025 and
1205821= 11399 (according to (94)) For 119902 = 0 we calculate
1198861= minus08432 and 119887
1= 07670 (corresponding waveform is
presented by solid line in Figure 6) for 119902 = 1 we calculate1198861= minus02426 and 119887
1= minus11138 (corresponding waveform is
presented by dashed line) for 119902 = 2 we calculate 1198861= 10859
and 1198871= 03468 (corresponding waveform is presented by
dotted line)As another example of the usage of Algorithm 17 let us
consider case 119896 = 4 and assume that1198864= minus015 and 119887
4= minus02
Consequently 1205824= 025 and 120582
1= 09861 (according to (95))
For 119902 = 0 3we calculate the following four pairs (1198861 1198871) of
coefficients of fundamental harmonic (minus08388 05184) for119902 = 0 (minus05184 minus08388) for 119902 = 1 (08388 minus05184) for 119902 =2 and (05184 08388) for 119902 = 3 Corresponding waveformsare presented in Figure 7
4 Nonnegative Waveforms with MaximalAmplitude of Fundamental Harmonic
In this section we provide general description of nonnegativewaveforms containing fundamental and 119896th harmonic withmaximal amplitude of fundamental harmonic for prescribedamplitude of 119896th harmonic
The main result of this section is presented in the fol-lowing proposition
3
2
1
0
Wav
efor
ms
0 1 2 3 4
Angle 120591120587
q = 0q = 1
q = 2q = 3
Figure 7 Nonnegative waveforms with two zeros for 119896 = 4 1198864=
minus015 and 1198874= minus02
Proposition 18 Every nonnegativewaveformof type (35)withmaximal amplitude 120582
1of fundamental harmonic and pre-
scribed amplitude 120582119896of 119896th harmonic can be expressed in the
following form
119879119896(120591) = [1 minus cos (120591 minus 120591
0)]
sdot [1 minus (119896 minus 1) 120582119896minus 2120582119896
119896minus1
sum
119899=1
(119896 minus 119899) cos (119899 (120591 minus 1205910))]
(108)
if 0 le 120582119896le 1(119896
2
minus 1) or
119879119896(120591) = 120582
119896[1 minus cos (120591 minus 120591
0)] [1 minus cos(120591 minus 120591
0+2120585
119896)]
sdot [1198880+ 2
119896minus2
sum
119899=1
119888119899cos 119899(120591 minus 120591
0+120585
119896)]
(109)
if 1(1198962 minus 1) le 120582119896le 1 providing that 119888
119899 119899 = 0 119896 minus 2 and
120582119896are related to 120585 via relations (67) and (68) respectively and
|120585| le 120587
Remark 19 Expression (108) can be obtained from (38) bysetting 120585 = 0 Furthermore insertion of 120585 = 0 into (43)ndash(46)leads to the following expressions for coefficients ofwaveformof type (108)
1198861= minus (1 + 120582
119896) cos 120591
0 119887
1= minus (1 + 120582
119896) sin 120591
0
119886119896= 120582119896cos (119896120591
0) 119887
119896= 120582119896sin (119896120591
0)
(110)
On the other hand (109) coincides with (66) Thereforethe expressions for coefficients of (109) and (66) also coincideThus expressions for coefficients of fundamental harmonic ofwaveform (109) are given by (84) where 120582
1is given by (85)
while expressions for coefficients of 119896th harmonic are givenby (45)-(46)
Waveforms described by (108) have exactly one zerowhile waveforms described by (109) for 1(1198962 minus 1) lt 120582
119896lt 1
Mathematical Problems in Engineering 13
14
12
1
08
06
04
02
00 05 1
Amplitude 120582k
Am
plitu
de1205821
k = 2
k = 3
k = 4
Figure 8 Maximal amplitude of fundamental harmonic as a func-tion of amplitude of 119896th harmonic
have exactly two zeros As we mentioned earlier waveforms(109) for 120582
119896= 1 have 119896 zeros
Remark 20 Maximal amplitude of fundamental harmonic ofnonnegative waveforms of type (35) for prescribed amplitudeof 119896th harmonic can be expressed as
1205821= 1 + 120582
119896 (111)
if 0 le 120582119896le 1(119896
2
minus 1) or
1205821=
119896 sin 120585119896 sin 120585 cos (120585119896) minus cos 120585 sin (120585119896)
(112)
if 1(1198962 minus 1) le 120582119896le 1 where 120585 is related to 120582
119896via (68) (or
(86)) and |120585| le 120587From (110) it follows that (111) holds Substitution of (86)
into (85) leads to (112)Notice that 120582
119896= 1(119896
2
minus 1) is the only common point ofthe intervals 0 le 120582
119896le 1(119896
2
minus 1) and 1(1198962
minus 1) le 120582119896le
1 According to (111) 120582119896= 1(119896
2
minus 1) corresponds to 1205821=
1198962
(1198962
minus1) It can be also obtained from (112) by setting 120585 = 0The waveforms corresponding to this pair of amplitudes aremaximally flat nonnegative waveforms
Maximal amplitude of fundamental harmonic of non-negative waveform of type (35) for 119896 le 4 as a function ofamplitude of 119896th harmonic is presented in Figure 8
Remark 21 Maximum value of amplitude of fundamentalharmonic of nonnegative waveform of type (35) is
1205821max =
1
cos (120587 (2119896)) (113)
This maximum value is attained for |120585| = 1205872 (see (112)) Thecorresponding value of amplitude of 119896th harmonic is 120582
119896=
(1119896) tan(120587(2119896)) Nonnegative waveforms of type (35) with1205821= 1205821max have two zeros at 1205910 and 1205910 minus 120587119896 for 120585 = 1205872 or
at 1205910and 1205910+ 120587119896 for 120585 = minus1205872
14
12
1
08
06
04
02
0minus1 minus05 0 05 1
Am
plitu
de1205821
Parameter 120585120587
k = 2k = 3k = 4
Figure 9 Maximal amplitude of fundamental harmonic as a func-tion of parameter 120585
To prove that (113) holds let us first show that the fol-lowing relation holds for 119896 ge 2
cos( 120587
2119896) lt 1 minus
1
1198962 (114)
From 119896 ge 2 it follows that sinc(120587(4119896)) gt sinc(1205874) wheresinc 119909 = (sin119909)119909 and therefore sin(120587(4119896)) gt 1(radic2119896)By using trigonometric identity cos 2119909 = 1 minus 2sin2119909 weimmediately obtain (114)
According to (111) and (112) it is clear that 1205821attains its
maximum value on the interval 1(1198962 minus 1) le 120582119896le 1 Since
120582119896is monotonic function of |120585| on interval |120585| le 120587 (see
Remark 15) it follows that 119889120582119896119889120585 = 0 for 0 lt |120585| lt 120587
Therefore to find critical points of 1205821as a function of 120582
119896
it is sufficient to find critical points of 1205821as a function of
|120585| 0 lt |120585| lt 120587 and consider its values at the end points120585 = 0 and |120585| = 120587 Plot of 120582
1as a function of parameter 120585
for 119896 le 4 is presented in Figure 9 According to (112) firstderivative of 120582
1with respect to 120585 is equal to zero if and only
if (119896 cos 120585 sin(120585119896) minus sin 120585 cos(120585119896)) cos 120585 = 0 On interval0 lt |120585| lt 120587 this is true if and only if |120585| = 1205872 Accordingto (112) 120582
1is equal to 119896
2
(1198962
minus 1) for 120585 = 0 equal to zerofor |120585| = 120587 and equal to 1 cos(120587(2119896)) for |120585| = 1205872 From(114) it follows that 1198962(1198962minus1) lt 1 cos(120587(2119896)) and thereforemaximum value of 120582
1is given by (113) Moreover maximum
value of 1205821is attained for |120585| = 1205872
According to above consideration all nonnegative wave-forms of type (35) having maximum value of amplitude offundamental harmonic can be obtained from (109) by setting|120585| = 1205872 Three of them corresponding to 119896 = 3 120585 = 1205872and three different values of 120591
0(01205876 and1205873) are presented
in Figure 10 Dotted line corresponds to 1205910= 0 (coefficients
of corresponding waveform are 1198861= minus1 119887
1= 1radic3 119886
3= 0
and 1198873= minusradic39) solid line to 120591
0= 1205876 (119886
1= minus2radic3 119887
1= 0
1198863= radic39 and 119887
3= 0) and dashed line to 120591
0= 1205873 (119886
1= minus1
1198871= minus1radic3 119886
3= 0 and 119887
3= radic39)
Proof of Proposition 18 As it has been shown earlier (Propo-sition 6) nonnegative waveform of type (35) with at least
14 Mathematical Problems in Engineering
2
1
0
0 1 2 3 4
Angle 120591120587
Wav
efor
ms
1205910 = 01205910 = 12058761205910 = 1205873
Figure 10 Nonnegative waveforms with maximum amplitude offundamental harmonic for 119896 = 3 and 120585 = 1205872
one zero can be represented in form (38) According to (43)(44) and (36) for amplitude 120582
1of fundamental harmonic of
waveforms of type (38) the following relation holds
1205821= radic(1 + 120582
119896cos 120585)2 + 11989621205822
119896sin2120585 (115)
where 120582119896satisfy (40) and |120585| le 120587
Because of (40) in the quest of finding maximal 1205821for
prescribed 120582119896 we have to consider the following two cases
(Case i)120582119896lt [(119896minus1) cos 120585 + 119896 sin(120585minus120585119896) sin(120585119896)]minus1
(Case ii)120582119896= [(119896minus1) cos 120585+119896 sin(120585minus120585119896) sin(120585119896)]minus1
Case i Since 120582119896lt [(119896 minus 1) cos 120585 + 119896 sin(120585 minus 120585119896) sin(120585119896)]minus1
implies 120582119896
= 1 according to (115) it follows that 1205821
= 0Hence 119889120582
1119889120585 = 0 implies
2120582119896sin 120585 [1 minus (1198962 minus 1) 120582
119896cos 120585] = 0 (116)
Therefore 1198891205821119889120585 = 0 if 120582
119896= 0 (Option 1) or sin 120585 = 0
(Option 2) or (1198962 minus 1)120582119896cos 120585 = 1 (Option 3)
Option 1 According to (115) 120582119896= 0 implies 120582
1= 1 (notice
that this implication shows that 1205821does not depend on 120585 and
therefore we can set 120585 to zero value)
Option 2 According to (115) sin 120585 = 0 implies 1205821= 1 +
120582119896cos 120585 which further leads to the conclusion that 120582
1is
maximal for 120585 = 0 For 120585 = 0 120582119896lt [(119896 minus 1) cos 120585 + 119896 sin(120585 minus
120585119896) sin(120585119896)]minus1 becomes 120582119896lt 1(119896
2
minus 1)
Option 3 This option leads to contradiction To show thatnotice that (119896
2
minus 1)120582119896cos 120585 = 1 and 120582
119896lt [(119896 minus
1) cos 120585 + 119896 sin(120585 minus 120585119896) sin(120585119896)]minus1 imply that (119896 minus 1) cos 120585 gtsin(120585minus120585119896) sin(120585119896) Using (A5) (see Appendices) the latestinequality can be rewritten assum119896minus1
119899=1[cos 120585minuscos((119896minus2119899)120585119896)] gt
0 But from |119896 minus 2119899| lt 119896 119899 = 1 (119896 minus 1) and |120585| le 120587
it follows that all summands are not positive and therefore(119896minus1) cos 120585 gt sin(120585minus120585119896) sin(120585119896) does not hold for |120585| le 120587
Consequently Case i implies 120585 = 0 and 120582119896lt 1(119896
2
minus 1)Finally substitution of 120585 = 0 into (38) leads to (108) whichproves that (108) holds for 120582
119896lt 1(119896
2
minus 1)
Case ii Relation120582119896= [(119896minus1) cos 120585+119896 sin(120585minus120585119896) sin(120585119896)]minus1
according to Proposition 9 and Remark 11 implies that cor-responding waveforms can be expressed via (66)ndash(68) for|120585| le 120587 Furthermore 120582
119896= [(119896 minus 1) cos 120585 + 119896 sin(120585 minus
120585119896) sin(120585119896)]minus1 and |120585| le 120587 imply 1(1198962 minus 1) le 120582119896le 1
This proves that (109) holds for 1(1198962 minus 1) le 120582119896le 1
Finally let us prove that (108) holds for 120582119896= 1(119896
2
minus
1) According to (68) (see also Remark 11) this value of 120582119896
corresponds to 120585 = 0 Furthermore substitution of 120582119896=
1(1198962
minus 1) and 120585 = 0 into (109) leads to (70) which can berewritten as
119879119896(120591) =
[1 minus cos (120591 minus 1205910)]
(1 minus 1198962)
sdot [119896 (119896 minus 1) minus 2
119896minus1
sum
119899=1
(119896 minus 119899) cos (119899 (120591 minus 1205910))]
(117)
Waveform (117) coincides with waveform (108) for 120582119896
=
1(1 minus 1198962
) Consequently (108) holds for 120582119896= 1(1 minus 119896
2
)which completes the proof
5 Nonnegative Waveforms with MaximalAbsolute Value of the Coefficient of CosineTerm of Fundamental Harmonic
In this sectionwe consider general description of nonnegativewaveforms of type (35) with maximal absolute value ofcoefficient 119886
1for prescribed coefficients of 119896th harmonicThis
type of waveform is of particular interest in PA efficiencyanalysis In a number of cases of practical interest eithercurrent or voltage waveform is prescribed In such casesthe problem of finding maximal efficiency of PA can bereduced to the problem of finding nonnegative waveformwith maximal coefficient 119886
1for prescribed coefficients of 119896th
harmonic (see also Section 7)In Section 51 we provide general description of nonneg-
ative waveforms of type (35) with maximal absolute value ofcoefficient 119886
1for prescribed coefficients of 119896th harmonic In
Section 52 we illustrate results of Section 51 for particularcase 119896 = 3
51 Nonnegative Waveforms with Maximal Absolute Value ofCoefficient 119886
1for 119896 ge 2 Waveforms 119879
119896(120591) of type (35) with
1198861ge 0 can be derived from those with 119886
1le 0 by shifting
by 120587 and therefore we can assume without loss of generalitythat 119886
1le 0 Notice that if 119896 is even then shifting 119879
119896(120591) by
120587 produces the same result as replacement of 1198861with minus119886
1
(119886119896remains the same) On the other hand if 119896 is odd then
shifting 119879119896(120591) by 120587 produces the same result as replacement
of 1198861with minus119886
1and 119886119896with minus119886
119896
According to (37) coefficients of 119896th harmonic can beexpressed as
119886119896= 120582119896cos 120575 119887
119896= 120582119896sin 120575 (118)
Mathematical Problems in Engineering 15
where
|120575| le 120587 (119)
Conversely for prescribed coefficients 119886119896and 119887
119896 120575 can be
determined as
120575 = atan 2 (119887119896 119886119896) (120)
where definition of function atan 2(119910 119909) is given by (105)The main result of this section is stated in the following
proposition
Proposition 22 Every nonnegative waveform of type (35)withmaximal absolute value of coefficient 119886
1le 0 for prescribed
coefficients 119886119896and 119887119896of 119896th harmonic can be represented as
119879119896(120591)
= [1 minus cos 120591]
sdot [1 minus (119896 minus 1) 119886119896minus 2
119896minus1
sum
119899=1
(119896 minus 119899) (119886119896cos 119899120591 + 119887
119896sin 119899120591)]
(121)
if 119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) le 1+119886
119896 where 120575 = atan 2(bk
119886119896) or
119879119896(120591) = 120582
119896[1 minus cos(120591 minus (120575 + 120585)
119896)]
sdot [1 minus cos(120591 minus (120575 minus 120585)
119896)]
sdot [1198880+ 2
119896minus2
sum
119899=1
119888119899cos 119899(120591 minus 120575
119896)]
(122)
if 119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) ge 1 + 119886
119896 where 119888
119899 119899 = 0
119896minus2 and 120582119896= radic1198862119896+ 1198872119896are related to 120585 via relations (67) and
(68) respectively and |120585| le 120587
Remark 23 Expression (121) can be obtained from (38) bysetting 120591
0= 0 and 120585 = minus120575 and then replacing 120582
119896cos 120575 with
119886119896(see (118)) and 120582
119896cos(119899120591 minus 120575) with 119886
119896cos 119899120591 + 119887
119896sin 119899120591
(see also (118)) Furthermore insertion of 1205910= 0 and 120585 =
minus120575 into (43)ndash(46) leads to the following relations betweenfundamental and 119896th harmonic coefficients of waveform(121)
1198861= minus (1 + 119886
119896) 119887
1= minus119896119887
119896 (123)
On the other hand expression (122) can be obtained from(66) by replacing 120591
0minus120585119896with 120575119896 Therefore substitution of
1205910minus 120585119896 = 120575119896 in (84) leads to
1198861= minus1205821cos(120575
119896) 119887
1= minus1205821sin(120575
119896) (124)
where 1205821is given by (85)
The fundamental harmonic coefficients 1198861and 1198871of wave-
form of type (35) with maximal absolute value of coefficient1198861le 0 satisfy both relations (123) and (124) if 119886
119896and 119887119896satisfy
1 + 119886119896= 119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) For such waveforms
relations 1205910= 0 and 120585 = minus120575 also hold
Remark 24 Amplitude of 119896th harmonic of nonnegativewaveform of type (35) with maximal absolute value of coeffi-cient 119886
1le 0 and coefficients 119886
119896 119887119896satisfying 1 + 119886
119896=
119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) is
120582119896=
sin (120575119896)119896 sin 120575 cos (120575119896) minus cos 120575 sin (120575119896)
(125)
To show that it is sufficient to substitute 119886119896= 120582119896cos 120575 (see
(118)) into 1 + 119886119896= 119896120582119896[sin 120575 sin(120575119896)] cos(120575119896)
Introducing new variable
119910 = cos(120575119896) (126)
and using the Chebyshev polynomials (eg see Appendices)relations 119886
119896= 120582119896cos 120575 and (125) can be rewritten as
119886119896= 120582119896119881119896(119910) (127)
120582119896=
1
119896119910119880119896minus1
(119910) minus 119881119896(119910)
(128)
where119881119896(119910) and119880
119896(119910) denote the Chebyshev polynomials of
the first and second kind respectively Substitution of (128)into (127) leads to
119886119896119896119910119880119896minus1
(119910) minus (1 + 119886119896) 119881119896(119910) = 0 (129)
which is polynomial equation of 119896th degree in terms of var-iable 119910 From |120575| le 120587 and (126) it follows that
cos(120587119896) le 119910 le 1 (130)
In what follows we show that 119886119896is monotonically increas-
ing function of 119910 on the interval (130) From 120585 = minus120575 (seeRemark 23) and (81) it follows that 120582minus1
119896= (119896 minus 1) cos 120575 +
119896sum119896minus1
119899=1cos((119896 minus 2119899)120575119896) ge 1 and therefore 119886
119896= 120582119896cos 120575 can
be rewritten as
119886119896=
cos 120575(119896 minus 1) cos 120575 + 119896sum119896minus1
119899=1cos ((119896 minus 2119899) 120575119896)
(131)
Obviously 119886119896is even function of 120575 and all cosines in (131)
are monotonically decreasing functions of |120575| on the interval|120575| le 120587 It is easy to show that cos((119896 minus 2119899)120575119896) 119899 =
1 (119896 minus 1) decreases slower than cos 120575 when |120575| increasesThis implies that denominator of the right hand side of(131) decreases slower than numerator Since denominator ispositive for |120575| le 120587 it further implies that 119886
119896is decreasing
function of |120575| on interval |120575| le 120587 Consequently 119886119896is
monotonically increasing function of 119910 on the interval (130)Thus we have shown that 119886
119896is monotonically increasing
function of 119910 on the interval (130) and therefore (129) hasonly one solution that satisfies (130) According to (128) thevalue of 119910 obtained from (129) and (130) either analyticallyor numerically leads to amplitude 120582
119896of 119896th harmonic
16 Mathematical Problems in Engineering
1
05
0
minus05
minus1
minus1 minus05 0 05 1
Coefficient ak
Coe
ffici
entb
k
radica2k+ b2
kle 1
k = 2k = 3k = 4
Figure 11 Plot of (119886119896 119887119896) satisfying 1 + 119886
119896= 119896120582
119896[sin 120575 sin(120575
119896)] cos(120575119896) for 119896 le 4
By solving (129) and (130) for 119896 le 4 we obtain
119910 = radic1 + 1198862
2 (1 minus 1198862) minus1 le 119886
2le1
3
119910 = radic3
4 (1 minus 21198863) minus1 le 119886
3le1
8
119910 =radicradic2 minus 4119886
4+ 1011988624minus 2 (1 minus 119886
4)
4 (1 minus 31198864)
minus1 le 1198864le
1
15
(132)
Insertion of (132) into (128) leads to the following explicitexpressions for the amplitude 120582
119896 119896 le 4
1205822=1
2(1 minus 119886
2) minus1 le 119886
2le1
3 (133)
1205822
3= [
1
3(1 minus 2119886
3)]
3
minus1 le 1198863le1
8 (134)
1205824=1
4(minus1 minus 119886
4+ radic2 minus 4119886
4+ 1011988624) minus1 le 119886
4le
1
15
(135)
Relations (133)ndash(135) define closed lines (see Figure 11) whichseparate points representing waveforms of type (121) frompoints representing waveforms of type (122) For given 119896points inside the corresponding curve refer to nonnegativewaveforms of type (121) whereas points outside curve (andradic1198862119896+ 1198872119896le 1) correspond to nonnegative waveforms of type
(122) Points on the respective curve correspond to the wave-forms which can be expressed in both forms (121) and (122)
Remark 25 Themaximum absolute value of coefficient 1198861of
nonnegative waveform of type (35) is
100381610038161003816100381611988611003816100381610038161003816max =
1
cos (120587 (2119896)) (136)
This maximum value is attained for |120585| = 1205872 and 120575 = 0
(see (124)) Notice that |1198861|max is equal to the maximum value
1205821max of amplitude of fundamental harmonic (see (113))
Coefficients of waveform with maximum absolute value ofcoefficient 119886
1 1198861lt 0 are
1198861= minus
1
cos (120587 (2119896)) 119886
119896=1
119896tan( 120587
(2119896))
1198871= 119887119896= 0
(137)
Waveformdescribed by (137) is cosinewaveformhaving zerosat 120587(2119896) and minus120587(2119896)
In the course of proving (136) notice first that |1198861|max le
1205821max holds According to (123) and (124) maximum of |119886
1|
occurs for 119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) ge 1 + 119886
119896 From (124)
it immediately follows that maximum value of |1198861| is attained
if and only if 1205821= 1205821max and 120575 = 0 which because of
120575119896 = 1205910minus120585119896 further implies 120591
0= 120585119896 Sincemaximumvalue
of 1205821is attained for |120585| = 1205872 it follows that corresponding
waveform has zeros at 120587(2119896) and minus120587(2119896)
Proof of Proposition 22 As it was mentioned earlier in thissection we can assume without loss of generality that 119886
1le 0
We consider waveforms119879119896(120591) of type (35) such that119879
119896(120591) ge 0
and119879119896(120591) = 0 for some 120591
0 Fromassumption that nonnegative
waveform 119879119896(120591) of type (35) has at least one zero it follows
that it can be expressed in form (38)Let us also assume that 120591
0is position of nondegenerate
critical point Therefore 119879119896(1205910) = 0 implies 1198791015840
119896(1205910) = 0 and
11987910158401015840
119896(1205910) gt 0 According to (55) second derivative of 119879
119896(120591) at
1205910can be expressed as 11987910158401015840
119896(1205910) = 1 minus 120582
119896(1198962
minus 1) cos 120585 Since11987910158401015840
119896(1205910) gt 0 it follows immediately that
1 minus 120582119896(1198962
minus 1) cos 120585 gt 0 (138)
Let us further assume that 119879119896(120591) has exactly one zeroThe
problem of finding maximum absolute value of 1198861is con-
nected to the problem of finding maximum of the minimumfunction (see Section 21) If waveforms possess unique globalminimum at nondegenerate critical point then correspond-ing minimum function is a smooth function of parameters[13] Consequently assumption that 119879
119896(120591) has exactly one
zero at nondegenerate critical point leads to the conclusionthat coefficient 119886
1is differentiable function of 120591
0 First
derivative of 1198861(see (43)) with respect to 120591
0 taking into
account that 1205971205851205971205910= 119896 (see (50)) can be expressed in the
following factorized form
1205971198861
1205971205910
= sin 1205910[1 minus 120582
119896(1198962
minus 1) cos 120585] (139)
Mathematical Problems in Engineering 17
From (138) and (139) it is clear that 12059711988611205971205910= 0 if and only if
sin 1205910= 0 According toRemark 12 assumption that119879
119896(120591)has
exactly one zero implies 120582119896lt 1 From (51) (48) and 120582
119896lt 1
it follows that 1198861cos 1205910+ 1198871sin 1205910lt 0 which together with
sin 1205910= 0 implies that 119886
1cos 1205910lt 0 Assumption 119886
1le 0
together with relations 1198861cos 1205910lt 0 and sin 120591
0= 0 further
implies 1198861
= 0 and
1205910= 0 (140)
Insertion of 1205910= 0 into (38) leads to
119879119896(120591)
= [1 minus cos 120591]
sdot [1 minus (119896 minus 1) 120582119896cos 120585 minus 2120582
119896
119896minus1
sum
119899=1
(119896 minus 119899) cos (119899120591 + 120585)]
(141)
Substitution of 1205910= 0 into (45) and (46) yields 119886
119896= 120582119896cos 120585
and 119887119896
= minus120582119896sin 120585 respectively Replacing 120582
119896cos 120585 with
119886119896and 120582
119896cos(119899120591 + 120585) with (119886
119896cos 119899120591 + 119887
119896sin 119899120591) in (141)
immediately leads to (121)Furthermore 119886
119896= 120582119896cos 120585 119887
119896= minus120582
119896sin 120585 and (118)
imply that
120575 = minus120585 (142)
According to (38)ndash(40) and (142) it follows that (141) is non-negative if and only if
120582119896[(119896 minus 1) cos 120575 + 119896 sin (120575 minus 120575119896)
sin (120575119896)] lt 1 (143)
Notice that 119886119896= 120582119896cos 120575 implies that the following relation
holds
120582119896[(119896 minus 1) cos 120575 + 119896 sin (120575 minus 120575119896)
sin (120575119896)]
= minus119886119896+ 119896120582119896
sin 120575sin (120575119896)
cos(120575119896)
(144)
Finally substitution of (144) into (143) leads to 119896120582119896[sin 120575
sin(120575119896)] cos(120575119896) lt 1 + 119886119896 which proves that (121) holds
when 119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) lt 1 + 119886
119896
Apart from nonnegative waveforms with exactly one zeroat nondegenerate critical point in what follows we will alsoconsider other types of nonnegative waveforms with at leastone zero According to Proposition 9 and Remark 11 thesewaveforms can be described by (66)ndash(68) providing that 0 le|120585| le 120587
According to (35) 119879119896(0) ge 0 implies 1 + 119886
1+ 119886119896ge 0
Consequently 1198861le 0 implies that |119886
1| le 1 + 119886
119896 On the other
hand according to (123) |1198861| = 1 + 119886
119896holds for waveforms
of type (121) The converse is also true 1198861le 0 and |119886
1| =
1 + 119886119896imply 119886
1= minus1 minus 119886
119896 which further from (35) implies
119879119896(0) = 0 Therefore in what follows it is enough to consider
only nonnegativewaveformswhich can be described by (66)ndash(68) and 0 le |120585| le 120587 with coefficients 119886
119896and 119887119896satisfying
119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) ge 1 + 119886
119896
For prescribed coefficients 119886119896and 119887119896 the amplitude 120582
119896=
radic1198862119896+ 1198872119896of 119896th harmonic is also prescribed According to
Remark 15 (see also Remark 16) 120582119896is monotonically
decreasing function of 119909 = cos(120585119896) The value of 119909 can beobtained by solving (90) subject to the constraint cos(120587119896) le119909 le 1 Then 120582
1can be determined from (88) From (106) it
immediately follows that maximal absolute value of 1198861le 0
corresponds to 119902 = 0 which from (104) and (120) furtherimplies that
120575 = 1198961205910minus 120585 (145)
Furthermore 119902 = 0 according to (107) implies that waveformzeros are
1205910=(120575 + 120585)
119896 120591
1015840
0= 1205910minus2120585
119896=(120575 minus 120585)
119896 (146)
Substitution of 1205910= (120575 + 120585)119896 into (66) yields (122) which
proves that (122) holds when 119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) ge
1 + 119886119896
In what follows we prove that (121) also holds when119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) = 1 + 119886
119896 Substitution of 119886
119896=
120582119896cos 120575 into 119896120582
119896[sin 120575 sin(120575119896)] cos(120575119896) = 1 + 119886
119896leads to
120582119896[(119896 minus 1) cos 120575 + 119896 sin (120575 minus 120575119896)
sin (120575119896)] = 1 (147)
As we mentioned earlier relation (142) holds for all wave-forms of type (121) Substituting (142) into (147) we obtain
120582119896[(119896 minus 1) cos 120585 + 119896 sin (120585 minus 120585119896)
sin (120585119896)] = 1 (148)
This expression can be rearranged as
120582119896
119896 sin ((119896 minus 1) 120585119896)sin 120585119896
= 1 minus (119896 minus 1) 120582119896cos 120585 (149)
On the other hand for waveforms of type (122) according to(68) relations (148) and (149) also hold Substitution of 120591
0=
(120575 + 120585)119896 (see (145)) and (67) into (122) leads to
119879119896(120591)
= 120582119896[1 minus cos (120591 minus 120591
0)]
sdot [119896 sin ((119896 minus 1) 120585119896)
sin 120585119896minus 2
119896minus1
sum
119899=1
(119896 minus 119899) cos (119899 (120591 minus 1205910) + 120585)]
(150)
Furthermore substitution of (142) into (145) implies that1205910
= 0 Finally substitution of 1205910
= 0 and (149) into(150) leads to (141) Therefore (141) holds when 119896120582
119896[sin 120575
sin(120575119896)] cos(120575119896) = 1 + 119886119896 which in turn shows that (121)
holds when 119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) = 1 + 119886
119896 This
completes the proof
18 Mathematical Problems in Engineering
52 Nonnegative Waveforms with Maximal Absolute Valueof Coefficient 119886
1for 119896 = 3 Nonnegative waveform of type
(35) for 119896 = 3 is widely used in PA design (eg see [10])In this subsection we illustrate results of Section 51 for thisparticular case The case 119896 = 2 is presented in detail in [12]
Coefficients of fundamental harmonic of nonnegativewaveform of type (35) with 119896 = 3 and maximal absolutevalue of coefficient 119886
1le 0 for prescribed coefficients 119886
3and
1198873(1205823= radic11988623+ 11988723) according to (123) (124) (134) (94) and
(120) are equal to
1198861= minus1 minus 119886
3 119887
1= minus3119887
3 (151)
if 12058223le [(1 minus 2119886
3)3]3
1198861= minus1205821cos(120575
3) 119887
1= minus1205821sin(120575
3) (152)
where 1205821= 3(
3radic1205823minus 1205823) and 120575 = atan 2(119887
3 1198863) if [(1 minus
21198863)3]3
le 1205822
3le 1The line 1205822
3= [(1minus2119886
3)3]3 (see case 119896 = 3
in Figure 11) separates points representing waveforms withcoefficients satisfying (151) from points representing wave-forms with coefficients satisfying (152) Waveforms describedby (151) for 1205822
3lt [(1 minus 2119886
3)3]3 have exactly one zero at
1205910= 0 Waveforms described by (151) and (152) for 1205822
3= [(1 minus
21198863)3]3 also have zero at 120591
0= 0 These waveforms as a rule
have exactly two zeros However there are two exceptionsone related to the maximally flat nonnegative waveform withcoefficients 119886
1= minus98 119886
3= 18 and 119887
1= 1198873= 0 which
has only one zero and the other related to the waveform withcoefficients 119886
1= 0 119886
3= minus1 and 119887
1= 1198873= 0 which has three
zerosWaveforms described by (152) for [(1minus21198863)3]3
lt 1205822
3lt
1 have two zeros Waveforms with 1205823= 1 have only third
harmonic (fundamental harmonic is zero)Plot of contours of maximal absolute value of coefficient
1198861 1198861le 0 for prescribed coefficients 119886
3and 1198873is presented
in Figure 12 According to Remark 25 the waveform withmaximum absolute value of 119886
1le 0 is fully described with
the following coefficients 1198861
= minus2radic3 1198863
= radic39 and1198871= 1198873= 0 This waveform has two zeros at plusmn1205876
Two examples of nonnegative waveforms for 119896 = 3
and maximal absolute value of coefficient 1198861 1198861le 0 with
prescribed coefficients 1198863and 1198873are presented in Figure 13
One waveform corresponds to the case 12058223lt [(1 minus 2119886
3)3]3
(solid line) and the other to the case 12058223gt [(1 minus 2119886
3)3]3
(dashed line)Thewaveform represented by solid line has onezero and its coefficients are 119886
3= minus01 119887
3= 01 119886
1= minus09
and 1198871= minus03 Dashed line corresponds to the waveform
having two zeros with coefficients 1198863= minus01 119887
3= 03 119886
1=
minus08844 and 1198871= minus06460 (case 1205822
3gt [(1 minus 2119886
3)3]3)
6 Nonnegative Cosine Waveforms withat Least One Zero
Nonnegative cosine waveforms have proved to be of impor-tance for waveform modelling in PA design (eg see [10])In this section we consider nonnegative cosine waveforms
1
05
0
minus05
minus1
minus1 minus05 0 05 1
Coefficient a3
Coe
ffici
entb
3
02
04
06
08
10
11
Figure 12 Contours ofmaximal absolute value of coefficient 1198861 1198861le
0 as a function of 1198863and 1198873
2
1
0
0 1 2 3 4
Angle 120591120587
Wav
efor
ms
a3 = minus01 b3 = 01
a3 = minus01 b3 = 03
Figure 13 Nonnegative waveforms for 119896 = 3 and maximal absolutevalue of 119886
1 1198861le 0 with prescribed coefficients 119886
3and 1198873
containing fundamental and 119896th harmonic with at least onezero
Cosine waveform with dc component fundamental and119896th harmonic can be obtained from (35) by setting 119887
1= 119887119896=
0 that is
119879119896(120591) = 1 + 119886
1cos 120591 + 119886
119896cos 119896120591 (153)
In Section 61 we provide general description of non-negative cosine waveforms of type (153) with at least onezero We show that nonnegative cosine waveforms with atleast one zero coincide with nonnegative cosine waveformswith maximal absolute value of coefficient 119886
1for prescribed
coefficient 119886119896 In Section 62 we illustrate results of Section 61
for particular case 119896 = 3
61 Nonnegative Cosine Waveforms with at Least One Zerofor 119896 ge 2 Amplitudes of fundamental and 119896th harmonic
Mathematical Problems in Engineering 19
of cosine waveform of type (153) are 1205821= |1198861| and 120582
119896=
|119886119896| respectively According to (42) for nonnegative cosine
waveforms of type (153) the following relation holds
minus1 le 119886119896le 1 (154)
This explains why 119896th harmonic coefficient 119886119896in Proposi-
tion 26 goes through interval [minus1 1]Waveforms (153) with 119886
1ge 0 can be obtained from
waveforms with 1198861le 0 by shifting by 120587 and therefore with-
out loss of generality we can assume that 1198861le 0
Proposition 26 Each nonnegative cosine waveform of type(153) with 119886
1le 0 and at least one zero can be represented as
119879119896(120591) = [1 minus cos 120591] [1 minus (119896 minus 1) 119886
119896minus 2119886119896
119896minus1
sum
119899=1
(119896 minus 119899) cos 119899120591]
(155)
if minus1 le 119886119896le 1(119896
2
minus 1) or
119879119896(120591) = 119886
119896[1 minus cos (120591 minus 120591
0)] [1 minus cos (120591 + 120591
0)]
sdot [1198880+ 2
119896minus2
sum
119899=1
119888119899cos 119899120591]
(156)
where
119888119899=sin ((119896 minus 119899) 120591
0) cos 120591
0minus (119896 minus 119899) cos ((119896 minus 119899) 120591
0) sin 120591
0
sin31205910
(157)
119886119896=
sin 1205910
119896 sin (1198961205910) cos 120591
0minus cos (119896120591
0) sin 120591
0
(158)
100381610038161003816100381612059101003816100381610038161003816 le
120587
119896 (159)
if 1(1198962 minus 1) le 119886119896le 1
Remark 27 Identity [1minuscos(120591minus1205910)][1minuscos(120591+120591
0)] = [cos 120591
0minus
cos 120591]2 implies that (156) can be rewritten as
119879119896(120591) = 119886
119896[cos 120591
0minus cos 120591]2 [119888
0+ 2
119896minus2
sum
119899=1
119888119899cos 119899120591] (160)
Furthermore substitution of (157) into (160) leads to
119879119896(120591) = 119886
119896[cos 120591
0minus cos 120591]
sdot [(119896 minus 1) sin 119896120591
0
sin 1205910
minus 2
119896minus1
sum
119899=1
sin ((119896 minus 119899) 1205910)
sin 1205910
cos 119899120591]
(161)
Remark 28 All nonnegative cosine waveforms of type (153)with at least one zero and 119886
1le 0 except one of them can be
represented either in form (155) or form (156)This exceptionis maximally flat cosine waveform with 119886
1lt 0 which can be
obtained from (155) for 119886119896= 1(119896
2
minus 1) or from (156) for 1205910=
0 Maximally flat cosine waveform with 1198861lt 0 can also be
obtained from (70) by setting 1205910= 0 Furthermore setting
1205910= 0 in (71) leads to maximally flat cosine waveforms for
119896 le 4 and 1198861lt 0
Remark 29 Nonnegative cosine waveform of type (155) with1198861lt 0 and minus1 lt 119886
119896le 1(119896
2
minus 1) has exactly one zero at120591 = 0 Nonnegative cosine waveform described by (156) with1198861lt 0 and 1(1198962 minus 1) lt 119886
119896lt 1 has two zeros at plusmn120591
0 where
0 lt |1205910| lt 120587119896 For 119886
119896= ∓1 nonnegative cosine waveform
of type (153) reduces to 119879119896(120591) = 1 ∓ cos 119896120591 (clearly these two
waveforms both have 119896 zeros)
Remark 30 Transformation of (155) into an additive formleads to the following relation
1198861= minus1 minus 119886
119896 (162)
where minus1 le 119886119896le 1(119896
2
minus1) Similarly transformation of (156)leads to the following relation
1198861= minus119886119896
119896 sin 1198961205910
sin 1205910
(163)
where 119886119896is given by (158) 1(1198962minus1) le 119886
119896le 1 and |120591
0| le 120587119896
Notice that coefficients of maximally flat cosine waveformnamely 119886
119896= 1(119896
2
minus1) and 1198861= minus1198962
(1198962
minus1) satisfy relation(162) They also satisfy relation (163) for 120591
0= 0
Remark 31 Nonnegative cosine waveforms of type (153) withat least one zero coincide with nonnegative cosine waveformswith maximal absolute value of coefficient 119886
1for prescribed
coefficient 119886119896
In proving that Remark 31 holds notice that expression(155) can be obtained from (121) by setting 119887
119896= 0 Fur-
thermore if 119886119896
ge 0 then 120582119896
= 119886119896 which together
with 119887119896
= 0 and (118) implies 120575 = 0 In this case119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) le 1 + 119886
119896becomes 1198962119886
119896le 1 +
119886119896 On the other hand if 119886
119896lt 0 then 120582
119896= minus119886
119896 which
together with 119887119896= 0 and (118) implies |120575| = 120587 In this case
119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) le 1 + 119886
119896becomes 0 le 1 + 119886
119896
Therefore every nonnegative cosine waveform of type (155)has maximal absolute value of coefficient 119886
1for prescribed
coefficient 119886119896 when minus1 le 119886
119896le 1(119896
2
minus 1)Let us now show that expression (156) can be obtained
from (122) by setting 119887119896= 0 and 119886
119896gt 0 For waveforms
of type (122) according to (118) 119887119896= 0 and 119886
119896gt 0 imply
120575 = 0 and 120582119896= 119886119896 Substitution of 120582
119896= 119886119896and 120575 = 0 into
119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) ge 1+119886
119896leads to 119886
119896ge 1(119896
2
minus1)Furthermore substitution of 120575 = 0 into (145) yields 120591
0= 120585119896
Insertion of 120582119896= 119886119896 120575 = 0 and 120591
0= 120585119896 into (122) leads
to (156) Therefore every nonnegative cosine waveform oftype (156) has maximal absolute value of coefficient 119886
1for
prescribed coefficient 119886119896 when 1(1198962 minus 1) le 119886
119896le 1
Proof of Proposition 26 Let us start with nonnegative cosinewaveform of type (153) with 120582
119896= |119886119896| = 1 According to
Remark 7 120582119896= |119886119896| = 1 implies that 120582
1= |1198861| = 0
Substitution of 119886119896
= minus1 into (155) and using (A2) (seeAppendices) lead to 119879
119896(120591) = 1 minus cos 119896120591 Consequently (155)
holds for 119886119896= minus1 On the other hand substitution of 119886
119896= 1
into (158) yields |1205910| = 120587119896 Furthermore substitution of
119886119896= 1 and 120591
0= 120587119896 (or 120591
0= minus120587119896) into (156) along
with performing all multiplications and using (A2) leads to
20 Mathematical Problems in Engineering
119879119896(120591) = 1 + cos 119896120591 Consequently (156)ndash(158) hold for 119886
119896= 1
and |1205910| = 120587119896
It is easy to see that 120582119896= |119886119896| lt 1 and 119879
119896(1205910) = 0 for some
1205910imply 120582
1= |1198861| = 0 Therefore in what follows we assume
that |119886119896| = 1 and 119886
1lt 0
Cosine waveforms are even functions of 120591 Therefore ifnonnegative cosine waveform has exactly one zero it has to beeither at 0 or at 120587 On the other hand if nonnegative cosinewaveform with 119886
1= 0 has exactly two zeros then these zeros
are placed at plusmn1205910 such that 120591
0is neither 0 nor 120587
In order to prove that (155) holds for minus1 lt 119886119896le 1(1minus119896
2
)let us start by referring to the description (38) of nonnegativewaveformswith at least one zero As wementioned earlier fornonnegative cosine waveformwith exactly one zero (denotedby 1205910) it is either 120591
0= 0 or 120591
0= 120587 Therefore in both cases
sin 1205910= 0 Substitution of sin 120591
0= 0 into (43) together with
1198861
= 0 and 120582119896= |119886119896| lt 1 leads to
1205910= 0 (164)
Clearly 1205910= 0 119887
1= 0 and 119887
119896= 0 according to (44) and (46)
imply 120582119896sin 120585 = 0 Since 120582
119896= |119886119896| it follows that |119886
119896| sin 120585 = 0
also holds which further implies 120582119896= 119886119896= 0 or sin 120585 =
0 In the case when 120582119896= 119886119896= 0 from (164) and (43) we
obtain 1198861= minus1 which further implies that 119879
119896(120591) = 1 minus cos 120591
Consequently (155) holds for 119886119896= 0 In the case when sin 120585 =
0 from (164) and (45) we obtain 119886119896= 120582119896if 120585 = 0 or 119886
119896= minus120582119896
if 120585 = 120587 Relations 119886119896= 120582119896and 120585 = 0 according to (40) imply
that 0 le 119886119896le 1(1 minus 119896
2
) Substitution of 120585 = 0 120582119896= 119886119896 and
(164) into (38) leads to (155) which proves that (155) holdsfor 0 le 119886
119896le 1(1 minus 119896
2
) On the other hand relations 119886119896=
minus120582119896and 120585 = 120587 according to (40) imply that minus1 lt 119886
119896le 0
Substitution of 120585 = 120587 120582119896= minus119886119896 and (164) into (38) also leads
to (155) which proves that (155) also holds for minus1 lt 119886119896le 0
Consequently (155) holds for minus1 lt 119886119896le 1(1 minus 119896
2
)In what follows we first prove that (156)-(157) hold for
1(1 minus 1198962
) lt 119886119896lt 1 For this purpose let us start with non-
negative waveforms with two zeros described by (66) As wementioned before nonnegative cosine waveforms with twozeros have zeros at 120591
0and minus120591
0 such that 120591
0= 0 and 120591
0= 120587
Relations 1198861
lt 0 and 1198871
= 0 according to (84) implycos(1205910minus 120585119896) = 1 and therefore
120585
119896= 1205910 (165)
From 120585119896 = 1205910and 0 lt |120585| lt 120587 it follows that 0 lt |120591
0| lt
120587119896 Insertion of 120585119896 = 1205910into (45) yields 119886
119896= 120582119896 Relations
119886119896= 120582119896and (82) imply that 1(1 minus 1198962) lt 119886
119896lt 1 Substitution
of 120582119896= 119886119896and 120585119896 = 120591
0into (66)ndash(68) leads to (156)ndash(158)
which proves that (156)ndash(158) hold for 1(1 minus 1198962
) lt 119886119896lt 1
and 0 lt |1205910| lt 120587119896
Finally substitution of 119886119896= 1(1 minus 119896
2
) and 1205910= 0 into
(161) leads to
119879119896(120591) =
[1 minus cos 120591](1 minus 1198962)
[119896 (119896 minus 1) minus 2
119896minus1
sum
119899=1
(119896 minus 119899) cos 119899120591]
(166)
Waveform (166) coincides with waveform (155) for 119886119896
=
1(1 minus 1198962
) which in turn proves that (156) holds for 119886119896=
1(1 minus 1198962
) and 1205910= 0 This completes the proof
62 Nonnegative CosineWaveforms with at Least One Zero for119896 = 3 In this subsection we consider nonnegative cosinewaveforms with at least one zero for 119896 = 3 (for case 119896 = 2
see [12])Cosine waveform with fundamental and third harmonic
reads
1198793(120591) = 1 + 119886
1cos 120591 + 119886
3cos 3120591 (167)
For 1198861le 0 and minus1 le 119886
3le 18 according to (155) non-
negative cosine waveform of type (167) with at least one zerocan be expressed as
1198793(120591) = (1 minus cos 120591) [1 minus 2119886
3(1 + 2 cos 120591 + cos 2120591)] (168)
From 1198793(120591 + 120587) = 2 minus 119879
3(120591) it immediately follows that for
1198861ge 0 and minus18 le 119886
3le 1 119879
3(120591) can be expressed as
1198793(120591) = (1 + cos 120591) [1 + 2119886
3(1 minus 2 cos 120591 + cos 2120591)] (169)
For 1198861le 0 and 18 le 119886
3le 1 from (158) it follows that 119886
3=
[8cos31205910]minus1 This relation along with (160) and (157) further
implies that 1198793(120591) can be expressed as
1198793(120591) =
[cos 1205910minus cos 120591]2 [2 cos 120591
0+ cos 120591]
2cos31205910
(170)
providing that |1205910| le 1205873 From 119879
3(120591 + 120587) = 2 minus 119879
3(120591) it
follows that (170) also holds for 1198861ge 0 and minus1 le 119886
3le minus18
providing that 1205910isin [21205873 41205873]
Maximally flat nonnegative cosinewaveformof type (167)with 119886
1lt 0 (minimum at 120591
0= 0) reads 119879
3(120591) = [1 minus
cos 120591]2[1 + (12) cos 120591] Dually maximally flat nonnegativecosine waveform with 119886
1gt 0 (minimum at 120591
0= 120587) reads
1198793(120591) = [1 + cos 120591]2[1 minus (12) cos 120591]In what follows we provide relations between coefficients
1198861and 1198863of nonnegative cosine waveforms of type (167) with
at least one zeroFor 1198861le 0 conversion of (168) into an additive form
immediately leads to the following relation
1198861= minus1 minus 119886
3for minus 1 le 119886
3le1
8 (171)
Conversion of (170) into an additive form leads to 1198861
=
minus31198863(2 cos 2120591
0+ 1) which can be also expressed as 119886
1=
minus31198863(4cos2120591
0minus 1) For 119886
1le 0 relations |120591
0| le 1205873 119886
1=
minus31198863(4cos2120591
0minus 1) and 119886
3= [8cos3120591
0]minus1 lead to
1198861= minus3 [ 3radic119886
3minus 1198863] for 1
8le 1198863le 1 (172)
Similarly for 1198861ge 0 conversion of (169) into an additive form
leads to the following relation
1198861= 1 minus 119886
3for minus
1
8le 1198863le 1 (173)
Mathematical Problems in Engineering 21
1
15
15
05
0
minus05
minus15minus15
minus1
minus1 minus05 0 05 1
Coefficient a1
Coe
ffici
enta
3 a1= minus
3(3radica3minus a3)
a1 =
minus1 minus a3
a1 =
1 minusa3
a1 =3(3radic|a3
| +a3)
Figure 14 Parameter space of cosine waveforms for 119896 = 3
For waveform of type (170) with 1198861ge 0 relations 120591
0isin [21205873
41205873] 1198861= minus3119886
3(4cos2120591
0minus 1) and 119886
3= [8cos3120591
0]minus1 lead to
1198861= 3 [
3radic10038161003816100381610038161198863
1003816100381610038161003816 + 1198863] for minus 1 le 1198863le minus
1
8 (174)
Every cosine waveform of type (167) corresponds to apair of real numbers (119886
1 1198863) and vice versa Points (119886
1 1198863)
in grey area in Figure 14 correspond to nonnegative cosinewaveforms for 119896 = 3 The points at the boundary of grey areacorrespond to nonnegative cosinewaveformswith at least onezero A number of shapes of nonnegative cosine waveformswith 119896 = 3 and at least one zero plotted on interval [minus120587 120587]are also presented in Figure 14 The boundary of grey area inFigure 14 consists of four line segments described by relations(171)ndash(174) The common point of line segments (172) and(173) is cusp point with coordinates 119886
1= 0 and 119886
3= 1
Another cusp point with coordinates 1198861= 0 and 119886
3= minus1
is the common point of line segments (171) and (174) Thecommon point of line segments (171)-(172) has coordinates(minus98 18) and common point of line segments (173)-(174)has coordinates (98 minus18) These points are representedby white circle dots and they correspond to maximallyflat cosine waveforms (eg see [21]) White triangle dotswith coordinates (2radic3 minusradic39) and (minus2radic3radic39) refer tothe nonnegative cosine waveforms with maximum value ofamplitude of fundamental harmonic
7 Four Case Studies of Usage of NonnegativeWaveforms in PA Efficiency Analysis
In this section we provide four case studies of usage ofdescription of nonnegative waveforms with fundamental and119896th harmonic in PA efficiency analysis In first two casestudies to be presented in Section 71 voltage is nonnegativewaveform with fundamental and second harmonic with atleast one zero In remaining two case studies to be consideredin Section 72 voltage waveform contains fundamental andthird harmonic
i
v Load
Vdc
Idc
Lch
vL
iL
Cb
+ +in
Figure 15 Generic PA circuit diagram
Let us consider generic PA circuit diagram as shown inFigure 15 We assume here that voltage and current wave-forms at the transistor output are
V (120579) = 1 + 1198861V cos 120579 + 1198871V sin 120579 + 119886119896V cos 119896120579 + 119887119896V sin 119896120579
119894 (120579) = 1 + 1198861119894cos 120579 +
infin
sum
119899=2
119886119899119894cos 119899120579
(175)
where 120579 stands for 120596119905 Both waveforms are normalized in thesense that dc components of voltage and current are 119881dc =
1 and 119868dc = 1 respectively Under assumption that blockingcapacitor 119862
119887behaves as short-circuit at the fundamental and
higher harmonics current and voltage waveforms at the loadare
V119871(120579) = 119886
1V cos 120579 + 1198871V sin 120579 + 119886119896V cos 119896120579 + 119887119896V sin 119896120579
119894119871(120579) = minus119886
1119894cos 120579 minus
infin
sum
119899=2
119886119899119894cos 119899120579
(176)
In terms of coefficients of voltage and current waveforms theload impedance at fundamental harmonic is 119911
1= minus(119886
1V minus
1198951198871V)1198861119894 whereas load impedance at 119896th harmonic is 119911
119896=
minus(119886119896Vminus119895119887119896V)119886119896119894 All other harmonics are short-circuited (119911
119899=
0 for 119899 = 1 and 119899 = 119896) Time average output power of PA (egsee [10]) with waveform pair (175) at fundamental frequencycan be expressed as
1198751= minus
11988611198941198861V
2 (177)
For normalized waveforms (175) with 119881dc = 1 and 119868dc = 1dc power is 119875dc = 1 Consequently PA efficiency 120578 = 119875
1119875dc
(eg see [10 26]) is equal to
120578 = minus11988611198941198861V
2 (178)
Thus time average output power 1198751of PA with pair of nor-
malized waveform (175) is equal to efficiency (178)Power utilization factor (PUF) is defined [26] as ldquothe
ratio of power delivered in a given situation to the power
22 Mathematical Problems in Engineering
delivered by the same device with the same supply voltagein Class A moderdquo Since the output power in class-A modeis 1198751class-A = max[V(120579)] sdotmax[119894(120579)]8 (eg see [9]) it follows
that power utilization factor PUF = 11987511198751class-A for PA with
pair of normalized waveforms (175) can be expressed as
PUF =8120578
max [V (120579)] sdotmax [119894 (120579)] (179)
71 NonnegativeWaveforms for 119896 = 2 in PAEfficiencyAnalysisIn this subsection we provide two case studies of usage ofdescription of nonnegative waveforms with fundamental andsecond harmonic (119896 = 2) in PA efficiency analysis For moreexamples of usage of descriptions of nonnegative waveformswith fundamental and second harmonic in PA efficiencyanalysis see [12]
Case Study 71 In this case study we consider efficiency of PAfor given second harmonic impedance providing that voltageis nonnegative waveform with fundamental and second har-monic and current is ldquohalf-sinerdquo waveform frequently used inefficiency analysis of classical PA operation (eg see [10])
Standard model of current waveform for classical PAoperation has the form (eg see [10 26])
119894119863(120579) =
119868119863[cos 120579 minus cos(120572
2)] |120579| le
120572
2
0120572
2le |120579| le 120587
(180)
where 120572 is conduction angle and 119868119863
gt 0 Since 119894119863(120579) is
even function it immediately follows that its Fourier seriescontains only dc component and cosine terms
119894119863(120579) = 119868dc +
infin
sum
119899=1
119868119899cos 119899120579 (181)
The dc component of the waveform (180) is
119868dc =119868119863120572
2120587[sinc(120572
2) minus cos(120572
2)] (182)
where sinc119909 = (sin119909)119909 The coefficient of the fundamentalharmonic component reads
1198681=119868119863120572
2120587(1 minus sinc120572) (183)
and the coefficient of 119899th harmonic component can bewrittenin the form
119868119899=119868119863
119899120587[sin ((119899 minus 1) 1205722)
(119899 minus 1)minussin ((119899 + 1) 1205722)
(119899 + 1)] 119899 ge 2
(184)
For ldquohalf-sinerdquo current waveform conduction angle is equalto 120587 (class-B conduction angle) According to (182) thisfurther implies that 119868dc = 119868
119863120587 To obtain normalized form of
waveform (180) we set 119868dc = 1which implies that 119868119863= 120587 Fur-
thermore substitution of 120572 = 120587 and 119868119863= 120587 in (180) leads to
119894 (120579) =
120587 cos 120579 |120579| lt120587
2
0120587
2lt |120579| le 120587
(185)
Similarly substitution of 119868119863= 120587 and 120572 = 120587 into (183) and
(184) leads to the coefficients of waveform (185) Coefficientsof fundamental and second harmonic respectively are
1198861119894=120587
2 119886
2119894=2
3 (186)
On the other hand voltage waveform of type (35) for 119896 =2 reads
V (120579) = 1 + 1198861V cos 120579 + 1198871V sin 120579 + 1198862V cos 2120579 + 1198872V sin 2120579
(187)
This waveform contains only fundamental and second har-monic and therefore all harmonics of order higher thantwo are short-circuited (119911
119899= 0 for 119899 gt 2) For current
voltage pair (185) and (187) load impedance at fundamentalharmonic is 119911
1= minus(119886
1V minus 1198951198871V)1198861119894 whereas load impedance
at second harmonic is 1199112= minus(119886
2Vminus1198951198872V)1198862119894 According to ourassumption the load is passive and therefore Re119911
1 gt 0 and
Re1199112 ge 0 which further imply 119886
11198941198861V lt 0 and 119886
21198941198862V le 0
respectivelyIt is easy to see that problem of findingmaximal efficiency
of PA with current-voltage pair (185) and (187) for prescribedsecond harmonic impedance can be reduced to the problemof finding voltage waveform of type (187) with maximal coef-ficient |119886
1V| for prescribed coefficients of second harmonic(see Section 5)
The following algorithm (analogous to Algorithm 22presented in [12]) provides the procedure for calculation ofmaximal efficiency with current-voltage pair (185) and (187)for prescribed second harmonic impedance The definitionof function atan 2(119910 119909) which appears in the step (iii) of thefollowing algorithm is given by (105)
Algorithm 32 (i) Choose 1199112= 1199032+1198951199092such that |119911
2| le 1|119886
2119894|
(ii) calculate 1198862V minus 1198951198872V = minus119911
21198862119894and 120582
2V = radic11988622V + 1198872
2V(iii) if 2120582
2V le 1 minus 1198862V then calculate 119886
1V = minus1 minus 1198862V and
1198871V = minus2119887
2V else calculate 1205821V = radic81205822V(1 minus 1205822V) 1205790V minus 120585V2 =
(12)atan2(1198872V 1198862V) 1198861V = minus120582
1V cos(1205790V minus 120585V2) and 1198871V =
minus1205821V sin(1205790V minus 120585V2)(iv) calculate efficiency 120578 = minus119886
11198941198861V2
(v) calculate 1199111= minus(119886
1V minus 1198951198871V)1198861119894 and 1199112119899 = 1199112Re119911
1
In this case study coefficients of fundamental and secondharmonic of current waveform are given by (186) Maximalefficiency of PA associated with the waveform pair (185)and (187) as a function of normalized second harmonicimpedance 119911
2119899= 1199112Re119911
1 is presented in Figure 16(a)
As can be seen from Figure 16(a) efficiency of 078 isachieved at the edge of Smith chart where second harmonicimpedance has small resistive part Corresponding PUFcalculated according to (179) is presented in Figure 16(b)Peak efficiency 120578 = 1205874 = 07854 and peak value of PUF = 1
are attained when second harmonic is short-circuited (whichcorresponds to ideal class-B operation [10 26])
For example for second harmonic impedance 1199112= 01 minus
11989505 and currentwaveform (185) fromAlgorithm32 it followsthat 2120582
2V le 1 minus 1198862V Furthermore according to step (iii)
of above algorithm maximal efficiency of PA is attained
Mathematical Problems in Engineering 23
+j2
+j5
+j1
+j2
+j5
minusj2
minusj5
minusj1
minusj2
minusj5
infin0
078
02
05
1 2 5
075 07 065
06 05
075
07
065
06
05
120578 lt 05
(a)
099
095
091
083 075
067
0 59 051
051
099
095
091
083
075 0
67 059
+j5
+j1
+j2
+j5
minusj5
minusj1
minusj2
minusj5
infin
+j2
minusj2
0 02
05
1 2 5
120578 lt 05
(b)
Figure 16 (a) Contours of maximal efficiency of PA and (b) contours of corresponding PUF associated with the waveform pair (185) and(187) as functions of normalized second harmonic impedance 119911
2119899= 1199112Re119911
1
2
3
4
1
0
2 3 410
Wav
efor
ms
Angle 120579120587
i(120579)
(120579)
Figure 17 Waveform pair (185) and (187) that provides maximalefficiency for 119911
2= 01 minus 11989505
with voltage waveform of type (187) with coefficients 1198862V =
minus00667 1198872V = minus03333 119886
1V = minus09333 and 1198871V = 06667 (see
Figure 17) Corresponding efficiency PUF and normalizedsecond harmonic impedance are 120578 = 07330 PUF = 07572and 1199112119899= 01683 minus 11989508415 respectively
On the other hand for second harmonic impedance 1199112=
01 minus 11989508 and current waveform (185) from Algorithm 32 itfollows that 2120582
2V gt 1 minus 1198862V Then according to step (iii) of
above algorithm maximal efficiency is attained with voltagewaveform of type (187) with coefficients 119886
2V = minus00667 1198872V =
minus05333 1198861V = minus09333 and 119887
1V = 10572 (see Figure 18)Efficiency PUF andnormalized secondharmonic impedanceare 120578 = 07330 PUF = 06332 and 119911
2119899= 01683 minus 11989513465
respectively
2
3
4
1
0
2 3 410
Wav
efor
ms
Angle 120579120587
i(120579)
(120579)
Figure 18 Waveform pair (185) and (187) that provides maximalefficiency for 119911
2= 01 minus 11989508
Case Study 72 As another case study let us consider the effi-ciency of PA providing that current waveform is nonnegativecosine waveform up to third harmonic with maximum valueof amplitude of fundamental harmonic [22] (see also [8])
119894 (120579) = 1 +1 + radic5
2cos 120579 + 2radic5
5cos 2120579 + 5 minus radic5
10cos 3120579
(188)
and voltage waveform is nonnegative waveform of type(187) Load impedances at fundamental second and thirdharmonic are 119911
1= minus(119886
1Vminus1198951198871V)1198861119894 1199112 = minus(1198862Vminus1198951198872V)1198862119894 and
1199113= 0 respectively According to our assumption the load
is passive and therefore Re1199111 gt 0 and Re119911
2 ge 0 which
further imply 11988611198941198861V lt 0 and 119886
21198941198862V le 0 respectively
Because current waveform (188) contains only cosineterms and voltage waveform is the same as in previous case
24 Mathematical Problems in Engineering
08
075
07
06
06
065
065
05
05
075 0
7
120578 lt 05
+j2
+j5
+j1
+j2
+j5
minusj2
minusj5
minusj1
minusj2
minusj5
infin0
02
05
1 2 5
(a)
085
08 0
75 07 0
6
05 04
08
075
0405
060
7
120578 lt 05
+j2
+j5
+j1
+j2
+j5
minusj2
minusj5
minusj1
minusj2
minusj5
infin0 02
05
1 2 5
(b)
Figure 19 (a) Contours of maximal efficiency of PA and (b) contours of corresponding PUF associated with the waveform pair (187)-(188)as functions of normalized second harmonic impedance 119911
2119899= 1199112Re119911
1
study the procedure for calculation of maximal efficiency ofPA with waveform pair (187)-(188) is the same as presentedin Algorithm 32 In this case study the coefficients of funda-mental and second harmonic of current waveform are 119886
1119894=
(1 + radic5)2 and 1198862119894= 2radic55 respectively
Maximal efficiency of PA associated with the waveformpair (187)-(188) as a function of normalized secondharmonicimpedance 119911
2119899= 1199112Re119911
1 is presented in Figure 19(a)
Efficiency of 08 is achieved at the edge of Smith chart wheresecond harmonic impedance has small resistive part Thetheoretical upper bound 120578 = (1 + radic5)4 asymp 08090 isattained when second harmonic is short-circuitedWhen thisupper bound is reached both second and third harmonic areshort-circuited which implies that we are dealing with finiteharmonic class-C [6 8] or dually when current and voltageinterchange their roles with finite harmonic inverse class-C[6 9] Corresponding PUF calculated according to (179) ispresented in Figure 19(b) Peak value of PUF asymp 08541 isattained when second harmonic is short-circuited
For example for second harmonic impedance 1199112
=
007 minus 11989504 and current waveform (188) from Algorithm 32 itfollows that 2120582
2V le 1 minus 1198862V Furthermore according to step
(iii) of Algorithm 32 maximal efficiency of PA is attainedwith voltage waveform of type (187) with coefficients 119886
2V =
minus00626 1198872V = minus03578 119886
1V = minus09374 and 1198871V = 07155 (see
Figure 20) Corresponding efficiency PUF and normalizedsecond harmonic impedance are 120578 = 07584 PUF = 06337and 1199112119899= 01208 minus 11989506904 respectively
On the other hand for 1199112= 005 minus 11989507 and current
waveform (187) it follows that 21205822V gt 1minus119886
2VThen accordingto step (iii) of Algorithm 32 themaximal efficiency is attainedwith voltage waveform of type (187) with coefficients 119886
2V =
minus00447 1198872V = minus06261 119886
1V = minus09318 and 1198871V = 10007 (see
Figure 21) Efficiency PUF and normalized second harmonic
2
3
4
1
0
2 3 410
Wav
efor
ms
Angle 120579120587
i(120579)
(120579)
Figure 20 Waveform pair (187)-(188) that provides maximal effi-ciency for 119911
2= 007 minus 11989504
impedance are 120578 = 07538 PUF = 05314 and 1199112119899= 00868minus
11989512156 respectively
72 Nonnegative Waveforms for 119896 = 3 in PA EfficiencyAnalysis In this subsection we provide another two casestudies of usage of description of nonnegative waveforms inPA efficiency analysis this time with fundamental and thirdharmonic (119896 = 3)
Case Study 73 Let us consider current-voltage pair such thatvoltage is nonnegative waveform with fundamental and thirdharmonic
V (120579) = 1 + 1198861V cos 120579 + 1198871V sin 120579 + 1198863V cos 3120579 + 1198873V sin 3120579
(189)
Mathematical Problems in Engineering 25
2
3
4
1
0
2 3 410
Wav
efor
ms
Angle 120579120587
i(120579)
(120579)
Figure 21 Waveform pair (187)-(188) that provides maximal effi-ciency for 119911
2= 005 minus 11989507
and current is nonnegative cosine waveform given by (188)Load impedances at fundamental second and third har-monic are 119911
1= minus(119886
1V minus 1198951198871V)1198861119894 1199112 = 0 and 119911
3= minus(119886
3V minus
1198951198873V)1198863119894 respectively According to our assumption the load
is passive and therefore Re1199111 gt 0 and Re119911
3 ge 0 which
further imply 11988611198941198861V lt 0 and 119886
31198941198863V le 0
In this subsection we consider the problem of findingmaximal efficiency of PA with waveform pair (188)-(189)for given third harmonic impedance As we mentionedearlier problem of finding maximal efficiency of PA withcurrent-voltage pair (188)-(189) for prescribed third har-monic impedance can be reduced to the problem of findingvoltage waveform of type (189) withmaximal coefficient |119886
1V|for prescribed coefficients of third harmonic (see Section 52)
The following algorithm provides the procedure forcalculation of maximal efficiency with current-voltage pair(188)-(189) The definition of function atan 2(119910 119909) whichappears in step (iii) of the following algorithm is given by(105)
Algorithm 33 (i) Choose 1199113= 1199033+1198951199093such that |119911
3| le 1|119886
3119894|
(ii) calculate 1198863V minus 1198951198873V = minus119911
31198863119894and 120582
3V = radic11988623V + 1198872
3V(iii) if 271205822
3V le (1 minus 21198863V)3 then calculate 119886
1V = minus1 minus 1198863V
and 1198871V = minus3119887
3V else calculate 1205821V = 3(3radic1205823V minus 120582
3V) 1205790V minus120585V3 = (13)atan 2(119887
3V 1198863V) 1198861V = minus1205821V cos(1205790V minus 120585V3) and
1198871V = minus120582
1V sin(1205790V minus 120585V3)(iv) calculate efficiency 120578 = minus119886
11198941198861V2
(v) calculate 1199111= minus(119886
1V minus 1198951198871V)1198861119894 and 1199113119899 = 1199113Re119911
1
In this case study coefficients of fundamental and thirdharmonic of current waveform are 119886
1119894= (1 + radic5)2 and
1198863119894= (5 minus radic5)10 respectively For the waveform pair (188)-
(189) maximal efficiency of PA as a function of normalizedthird harmonic impedance 119911
3119899= 1199113Re119911
1 is presented in
Figure 22 Efficiency of 08 is reached when third harmonicimpedance has small resistive part Peak efficiency 120578 = (1 +
radic5)4 asymp 08090 is achieved when third harmonic is short-circuited
+j2
+j5
+j1
+j2
+j5
minusj2
minusj5
minusj1
minusj2
minusj5
infin0
08
08
075
075
07
07
06
05
02
05
1 2
Figure 22 Contours of maximal efficiency of PA associated withthe waveform pair (188)-(189) as a function of normalized thirdharmonic impedance 119911
3119899= 1199113Re119911
1
For the present case study in what follows we show thatpower utilization factor is proportional to efficiency For volt-age waveform of type (189) it is easy to see that V(120579 + 120587) =
2minusV(120579) holdsThis relation along with the fact that waveformV(120579) that provides maximal efficiency has at least one zeroimplies that max[V(120579)] = 2 On the other hand currentwaveform (188) is cosine waveform with positive coefficientsand therefore max[119894(120579)] = 119894(0) = 2 + 4radic5 Consequentlyaccording to (179) the following relation holds
PUFCase study 73 = 2 (5 minus 2radic5) 120578 = 10557120578 (190)
Clearly the ratio PUF120578 is constant and therefore in this casestudy PUF can be easily calculated from the correspondingefficiency Accordingly peak efficiency and peak value ofPUFCase study 73 = 3radic52 minus 52 = 08541 are attained forthe same voltage waveform (when third harmonic is short-circuited)
In the first example current waveform (188) and 1199113=
02 minus 11989505 imply that 2712058223V le (1 minus 2119886
3V)3 Then according
to Algorithm 33 the voltage waveform of type (189) thatprovides maximal efficiency has the following coefficients1198863V = minus00553 119887
3V = minus01382 1198861V = minus09447 and 119887
1V =
04146 (see Figure 23) Efficiency PUF and normalized thirdharmonic impedance are 120578 = 07643 PUF = 08069 and1199113119899= 03425 minus 11989508564 respectivelyIn the second example current waveform (188) and 119911
3=
01 minus 11989511 imply that 2712058223V gt (1 minus 2119886
3V)3 Then according
to Algorithm 33 the voltage waveform of type (189) thatprovides maximal efficiency has the following coefficients1198863V = minus00276 119887
3V = minus03040 1198861V = minus09391 and 119887
1V =
05807 (see Figure 24) Efficiency PUF and normalized thirdharmonic impedance are 120578 = 07598 PUF = 08021 and1199113119899= 01723 minus 11989518952 respectively
26 Mathematical Problems in Engineering
2
3
4
1
0
2 3 410
Wav
efor
ms
Angle 120579120587
i(120579)
(120579)
Figure 23 Waveform pair (188)-(189) that provides maximal effi-ciency for 119911
3= 02 minus 11989505
2
3
4
1
0
2 3 410
Wav
efor
ms
Angle 120579120587
i(120579)
(120579)
Figure 24 Waveform pair (188)-(189) that provides maximal effi-ciency for 119911
3= 01 minus 11989511
Case Study 74 In this case study let us consider current-voltage pair where current is normalized waveform of type(180) with conduction angle 120572 = 115120587 (207∘) and voltageis nonnegative waveform of type (189) Substitution of 120572 =
115120587 and 119868dc = 1 into (182) leads to 119868119863
= 22535Furthermore substitution of 120572 = 115120587 and 119868
119863= 22535 into
(180) leads to
119894 (120579)
=
22535 [cos 120579 minus cos(1151205872
)] |120579| le115120587
2
0115120587
2le |120579| le 120587
(191)
Similarly substitution of 120572 = 115120587 and 119868119863
= 22535 into(183) and (184) for 119899 = 3 yields coefficients of fundamentaland third harmonic of waveform (191)
1198861119894= 14586 119886
3119894= minus01026 (192)
Because current waveform (191) contains only cosineterms and voltage waveform is the same as in previous case
+j2
+j5
+j1
+j2
+j5
minusj2
minusj5
minusj1
minusj2
minusj5
infin0 02
05
1 2
074
076
08
084
082
078
076
074
Figure 25 Contours of maximal efficiency of PA associated withthe waveform pair (189) and (191) as a function of normalized thirdharmonic impedance 119911
3119899= 1199113Re119911
1
study the procedure for calculation of maximal efficiencyof PA with waveform pair (189)ndash(191) is the same as thatpresented in Algorithm 33 In this case study the coefficientsof fundamental and third harmonic of current waveform aregiven by (192)
For the waveform pair (189) and (191) maximal efficiencyof PA as a function of normalized third harmonic impedance1199113119899= 1199113Re119911
1 is presented in Figure 25 Efficiency of 084
is obtained in vicinity of 1199113119899= 23685 (corresponding to 119911
3=
18750) Peak efficiency 120578 asymp 08421 is achieved for voltagewaveform of type (189) with coefficients 119886
1V = minus2radic3 1198863V =
radic39 and 1198871V = 1198873V = 0
In the course of finding power utilization factor noticethat currentwaveformof type (191) attains itsmaximumvaluefor 120579 = 0 Insertion of max[119894(120579)] = 119894(0) = 278 andmax[V(120579)] = 2 for voltage waveform of type (189) into (179)leads to
PUFCase study 74 = 1439120578 (193)
Again the ratio PUF120578 is constant and PUF can be easilycalculated from the corresponding efficiency Accordinglypeak value of PUFCase study 74 asymp 12118 and peak efficiencyare attained for the same voltage waveform
In the first example current waveform (191) and 1199113=
1 minus 11989502 imply that 2712058223V le (1 minus 2119886
3V)3 Then according to
Algorithm 33 voltage waveform of type (189) which providesmaximal efficiency has coefficients 119886
3V = 01026 1198873V =
00205 1198861V = minus11026 and 119887
1V = minus00616 (see Figure 26)Efficiency PUF and normalized third harmonic impedanceare 120578 = 08042 PUF = 11572 and 119911
3119899= 13228 minus 11989502646
respectivelyIn second example current waveform (191) and 119911
3=
15 minus 11989512 imply that 2712058223V gt (1 minus 2119886
3V)3 Then according to
Algorithm 33 voltage waveform of type (189) which providesmaximal efficiency has coefficients 119886
3V = 01540 1198873V =
Mathematical Problems in Engineering 27
2
3
4
1
0
2 3 410
Wav
efor
ms
Angle 120579120587
i(120579)
(120579)
Figure 26 Waveform pair (189) and (191) that provides maximalefficiency for 119911
3= 1 minus 11989502
2
3
4
1
0
2 3 410
Wav
efor
ms
Angle 120579120587
i(120579)
(120579)
Figure 27 Waveform pair (189) and (191) that provides maximalefficiency for 119911
3= 15 minus 11989512
01232 1198861V = minus11255 and 119887
1V = minus02575 (see Figure 27)Efficiency PUF and normalized third harmonic impedanceare 120578 = 08208 PUF = 11812 and 119911
3119899= 19439 minus 11989515552
respectively
8 Conclusion
In this paper we consider a problem of finding general de-scriptions of various classes of nonnegative waveforms withfundamental and 119896th harmonic These classes include non-negative waveforms with at least one zero nonnegative wave-forms with maximal amplitude of fundamental harmonic forprescribed amplitude of 119896th harmonic nonnegative wave-forms withmaximal coefficient of cosine part of fundamentalharmonic for prescribed coefficients of 119896th harmonic andnonnegative cosine waveforms with at least one zero Mainresults are stated in six propositions (Propositions 1 6 918 22 and 26) four corollaries (Corollaries 2ndash5) twentyremarks and three algorithms Four case studies of usageof closed form descriptions of nonnegative waveforms in PAefficiency analysis are considered in detail in Section 7
Appendices
Here we provide a list of finite sums of trigonometric func-tions used in this paper (Appendix A) and brief account ofthe Chebyshev polynomials (Appendix B)
A List of Some Finite Sums ofTrigonometric Functions
Dirichlet kernel (eg see [27]) is as follows
119863119896minus1
(120591) = 1 + 2
119896minus1
sum
119899=1
cos 119899120591 = sin ((2119896 minus 1) 1205912)sin (1205912)
(A1)
Fejer kernel (eg see [27]) can be expressed in the fol-lowing equivalent forms
119865119896minus1
(120591) =1
119896
119896minus1
sum
119899=0
119863119899(120591) = 1 +
2
119896
119896minus1
sum
119899=1
(119896 minus 119899) cos 119899120591
=(1 minus cos 119896120591)119896 (1 minus cos 120591)
(A2)
Lagrangersquos trigonometric identity (eg see [28]) is as fol-lows
1198781(120591) =
119896minus1
sum
119899=1
sin 119899120591 = sin (1198961205912) sin ((119896 minus 1) 1205912)sin (1205912)
(A3)
In what follows we show that the following three trigono-metric identities also hold
2
119896minus1
sum
119899=1
(119896 minus 119899) sin 119899120591 = 119896 sin 120591 minus sin 1198961205911 minus cos 120591
(A4)
119896minus1
sum
119899=1
cos (119896 minus 2119899) 120591 = sin (119896 minus 1) 120591sin 120591
(A5)
119896minus1
sum
119899=1
119899 (119896 minus 119899) cos (119896 minus 2119899) 120591
=sin (119896120591) cos 120591 minus 119896 cos (119896120591) sin 120591
2sin3120591
(A6)
Denote 1198782(120591) = 2sum
119896minus1
119899=1(119896 minus 119899) sin 119899120591 119878
3(120591) = sum
119896minus1
119899=1cos(119896 minus
2119899)120591 and 1198784(120591) = sum
119896minus1
119899=1119899(119896 minus 119899) cos(119896 minus 2119899)120591
Notice that 1198782(120591) = 2119896119878
1(120591) + 119889119863
119896minus1(120591)119889120591 which
immediately leads to (A4)
28 Mathematical Problems in Engineering
Identity (A5) can be obtained as follows
sin (119896 minus 1) 120591sin 120591
=119890119895(119896minus1)120591
minus 119890minus119895(119896minus1)120591
119890119895120591 minus 119890minus119895120591
= 119890119895119896120591
119890minus2119895120591
minus 119890minus2119895(119896minus1)120591
1 minus 119890minus2119895120591
= 119890119895119896120591
119896minus1
sum
119899=1
119890minus2119895119899120591
=
119896minus1
sum
119899=1
119890119895(119896minus2119899)120591
=
119896minus1
sum
119899=1
cos (119896 minus 2119899) 120591
(A7)
From 4119899(119896 minus 119899) = 1198962
minus (119896 minus 2119899)2 it follows that 4119878
4(120591) =
1198962
1198783(120591) + 119889
2
1198783(120591)119889120591
2 which leads to (A6)
B The Chebyshev Polynomials
The Chebyshev polynomials of the first kind 119881119899(119909) can be
defined by the following relation (eg see [29])
119881119899(119909) = cos 119899120591 when 119909 = cos 120591 (B1)
The Chebyshev polynomials of the second kind 119880119899(119909) can be
defined by the following relation (eg see [29])
119880119899(119909) =
sin (119899 + 1) 120591sin 120591
when 119909 = cos 120591 (B2)
The Chebyshev polynomials satisfy the following recur-rence relations (eg see [29])
1198810(119909) = 1 119881
1(119909) = 119909
119881119899+1
(119909) = 2119909119881119899(119909) minus 119881
119899minus1(119909)
1198800(119909) = 1 119880
1(119909) = 2119909
119880119899+1
(119909) = 2119909119880119899(119909) minus 119880
119899minus1(119909)
(B3)
The first few Chebyshev polynomials of the first and secondkind are 119881
2(119909) = 2119909
2
minus 1 1198813(119909) = 4119909
3
minus 3119909 1198814(119909) = 8119909
4
minus
81199092
+ 1 1198802(119909) = 4119909
2
minus 1 1198803(119909) = 8119909
3
minus 4119909 and 1198804(119909) =
161199094
minus 121199092
+ 1
Conflict of Interests
The authors declare that there is no conflict of interestsregarding the publication of this paper
Acknowledgment
This work is supported by the SerbianMinistry of EducationScience and Technology Development as a part of ProjectTP32016
References
[1] V I Arnolrsquod V S Afrajmovich Y S Ilrsquoyashenko and L PShilrsquonikov Dynamical Systems V Bifurcation Theory and Catas-trophe Theory Springer Berlin Germany 1994
[2] E Polak ldquoOn the mathematical foundations of nondifferen-tiable optimization in engineering designrdquo SIAM Review vol29 no 1 pp 21ndash89 1987
[3] N S Fuzik ldquoBiharmonic modes of a tuned RF power amplifierrdquoRadiotehnika vol 25 no 7 pp 62ndash71 1970 (Russian)
[4] P Colantonio F Giannini G Leuzzi and E Limiti ldquoClass Gapproach for low-voltage high-efficiency PA designrdquo Interna-tional Journal of RF and Microwave Computer-Aided Engineer-ing vol 10 no 6 pp 366ndash378 2000
[5] F H Raab ldquoMaximum efficiency and output of class-F poweramplifiersrdquo IEEE Transactions on Microwave Theory and Tech-niques vol 49 no 6 pp 1162ndash1166 2001
[6] FH Raab ldquoClass-E class-C and class-F power amplifiers basedupon a finite number of harmonicsrdquo IEEE Transactions onMicrowaveTheory and Techniques vol 49 no 8 pp 1462ndash14682001
[7] J D Rhodes ldquoOutput universality inmaximum efficiency linearpower amplifiersrdquo International Journal of Circuit Theory andApplications vol 31 no 4 pp 385ndash405 2003
[8] A Juhas and L A Novak ldquoComments on lsquoClass-E class-Cand classF power amplifier based upon a finite number ofharmonicsrsquordquo IEEE Transactions on Microwave Theory and Tech-niques vol 57 no 6 pp 1623ndash1625 2009
[9] M Roberg and Z Popovic ldquoAnalysis of high-efficiency poweramplifiers with arbitrary output harmonic terminationsrdquo IEEETransactions on Microwave Theory and Techniques vol 59 no8 pp 2037ndash2048 2011
[10] A Grebennikov N O Sokal and M J Franco Switchmode RFPower Amplifiers ElsevierAcademic Press San Diego CalifUSA 2nd edition 2012
[11] T Canning P J Tasker and S C Cripps ldquoContinuous modepower amplifier design using harmonic clipping contourstheory and practicerdquo IEEE Transactions on Microwave Theoryand Techniques vol 62 no 1 pp 100ndash110 2014
[12] A Juhas and L A Novak ldquoGeneral description of nonnegativewaveforms up to second harmonic for power amplifier mod-ellingrdquoMathematical Problems in Engineering vol 2014 ArticleID 709762 18 pages 2014
[13] V I Arnolrsquod V V Goryunov O V Lyashko and V A VasilrsquoevDynamical Systems VIIImdashSingularity Theory II ApplicationsSpringer Berlin Germany 1993
[14] D Siersma ldquoProperties of conflict sets in the planerdquo BanachCenter Publications Polish Academy of Sciences vol 50 no 1 pp267ndash276 1999 Proceedings of the Banach Center Symposiumon Geometry and Topology of Caustics (Caustics rsquo98) WarsawPoland
[15] M van Manen The geometry of conflict sets [Dissertation]Universiteit Utrecht UtrechtThe Netherlands 2003 httpigi-tur-archivelibraryuunldissertations2003-0912-123058c4pdf
[16] Y L Sachkov ldquoMaxwell strata and symmetries in the problemofoptimal rolling of a sphere over a planerdquo Sbornik Mathematicsvol 201 no 7-8 pp 1029ndash1051 2010
[17] I A Bogaevsky ldquoPerestroikas of shocks and singularities ofminimum functionsrdquoPhysicaDNonlinear Phenomena vol 173no 1-2 pp 1ndash28 2002
[18] Y L Sachkov ldquoMaxwell strata in the Euler elastic problemrdquoJournal of Dynamical andControl Systems vol 14 no 2 pp 169ndash234 2008
[19] M Siino and T Koike ldquoTopological classification of black holesgeneric Maxwell set and crease set of a horizonrdquo InternationalJournal ofModern Physics D Gravitation Astrophysics Cosmol-ogy vol 20 no 6 pp 1095ndash1122 2011
Mathematical Problems in Engineering 29
[20] F H Raab ldquoClass-F power amplifiers with maximally flat wave-formsrdquo IEEETransactions onMicrowaveTheory and Techniquesvol 45 no 11 pp 2007ndash2012 1997
[21] A Juhas and L A Novak ldquoMaximally flat waveforms with finitenumber of harmonics in class-F power amplifiersrdquo Mathemat-ical Problems in Engineering vol 2013 Article ID 169590 9pages 2013
[22] L Fejer ldquoUber trigonometrische polynomerdquo Journal fur dieReine und Angewandte Mathematik vol 1916 no 146 pp 53ndash82 1916 (German)
[23] S C Cripps ldquoBessel Waives [microwave bytes]rdquo IEEE Micro-wave Magazine vol 10 no 7 pp 30ndash36 117 2009
[24] L N Bryzgalova ldquoSingularities of the maximum of para-metrically dependent functionrdquo Functional Analysis and ItsApplications vol 11 no 1 pp 49ndash51 1977
[25] V I Arnold A A Davydov V A Vassiliev and V MZakalyukin Mathematical Models of Catastrophes Control ofCatastrophic Process Encyclopedia of Life Support Systems(EOLSS) EOLSS Publishers Oxford UK 2006
[26] S C Cripps RF Power Amplifiers for Wireless CommunicationsArtech House Norwood Mass USA 2nd edition 2006
[27] A Zygmund Trigonometric Series vol 1 Cambridge UniversityPress Cambridge UK 2nd edition 1959
[28] A Jeffrey and H DaiHandbook of Mathematical Formulas andIntegrals ElsevierAcademic Press San Diego Calif USA 4thedition 2008
[29] J C Mason and D C Handscomb Chebyshev PolynomialsChapman amp Hall CRC Press Boca Raton Fla USA 2003
Submit your manuscripts athttpwwwhindawicom
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical Problems in Engineering
Hindawi Publishing Corporationhttpwwwhindawicom
Differential EquationsInternational Journal of
Volume 2014
Applied MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Probability and StatisticsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical PhysicsAdvances in
Complex AnalysisJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
OptimizationJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
CombinatoricsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Operations ResearchAdvances in
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Function Spaces
Abstract and Applied AnalysisHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of Mathematics and Mathematical Sciences
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Algebra
Discrete Dynamics in Nature and Society
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Decision SciencesAdvances in
Discrete MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom
Volume 2014 Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Stochastic AnalysisInternational Journal of
12 Mathematical Problems in Engineering
2
1
0
Wav
efor
ms
0 1 2 3 4
Angle 120591120587
q = 0
q = 1
q = 2
Figure 6 Nonnegative waveforms with two zeros for 119896 = 3 1198863=
minus015 and 1198873= minus02
Algorithm 17 (i) Calculate 120582119896= radic1198862119896+ 1198872119896
(ii) identify 119909 that satisfies both relations (90) and (91)(iii) calculate 120582
1according to (88)
(iv) choose integer 119902 such that 0 le 119902 le 119896 minus 1(v) calculate 119886
1and 1198871according to (106)
For 119896 le 4 by using (93) for 119896 = 2 (94) for 119896 = 3 and (95)for 119896 = 4 it is possible to calculate directly 120582
1from 120582
119896and
proceed to step (iv)For 119896 = 2 and prescribed coefficients 119886
2and 1198872 there are
two waveforms with two zeros one corresponding to 1198861lt 0
and the other corresponding to 1198861gt 0 (see also [12])
Let us take as an input 119896 = 3 1198863= minus015 and 119887
3= minus02
Execution of Algorithm 17 on this input yields 1205823= 025 and
1205821= 11399 (according to (94)) For 119902 = 0 we calculate
1198861= minus08432 and 119887
1= 07670 (corresponding waveform is
presented by solid line in Figure 6) for 119902 = 1 we calculate1198861= minus02426 and 119887
1= minus11138 (corresponding waveform is
presented by dashed line) for 119902 = 2 we calculate 1198861= 10859
and 1198871= 03468 (corresponding waveform is presented by
dotted line)As another example of the usage of Algorithm 17 let us
consider case 119896 = 4 and assume that1198864= minus015 and 119887
4= minus02
Consequently 1205824= 025 and 120582
1= 09861 (according to (95))
For 119902 = 0 3we calculate the following four pairs (1198861 1198871) of
coefficients of fundamental harmonic (minus08388 05184) for119902 = 0 (minus05184 minus08388) for 119902 = 1 (08388 minus05184) for 119902 =2 and (05184 08388) for 119902 = 3 Corresponding waveformsare presented in Figure 7
4 Nonnegative Waveforms with MaximalAmplitude of Fundamental Harmonic
In this section we provide general description of nonnegativewaveforms containing fundamental and 119896th harmonic withmaximal amplitude of fundamental harmonic for prescribedamplitude of 119896th harmonic
The main result of this section is presented in the fol-lowing proposition
3
2
1
0
Wav
efor
ms
0 1 2 3 4
Angle 120591120587
q = 0q = 1
q = 2q = 3
Figure 7 Nonnegative waveforms with two zeros for 119896 = 4 1198864=
minus015 and 1198874= minus02
Proposition 18 Every nonnegativewaveformof type (35)withmaximal amplitude 120582
1of fundamental harmonic and pre-
scribed amplitude 120582119896of 119896th harmonic can be expressed in the
following form
119879119896(120591) = [1 minus cos (120591 minus 120591
0)]
sdot [1 minus (119896 minus 1) 120582119896minus 2120582119896
119896minus1
sum
119899=1
(119896 minus 119899) cos (119899 (120591 minus 1205910))]
(108)
if 0 le 120582119896le 1(119896
2
minus 1) or
119879119896(120591) = 120582
119896[1 minus cos (120591 minus 120591
0)] [1 minus cos(120591 minus 120591
0+2120585
119896)]
sdot [1198880+ 2
119896minus2
sum
119899=1
119888119899cos 119899(120591 minus 120591
0+120585
119896)]
(109)
if 1(1198962 minus 1) le 120582119896le 1 providing that 119888
119899 119899 = 0 119896 minus 2 and
120582119896are related to 120585 via relations (67) and (68) respectively and
|120585| le 120587
Remark 19 Expression (108) can be obtained from (38) bysetting 120585 = 0 Furthermore insertion of 120585 = 0 into (43)ndash(46)leads to the following expressions for coefficients ofwaveformof type (108)
1198861= minus (1 + 120582
119896) cos 120591
0 119887
1= minus (1 + 120582
119896) sin 120591
0
119886119896= 120582119896cos (119896120591
0) 119887
119896= 120582119896sin (119896120591
0)
(110)
On the other hand (109) coincides with (66) Thereforethe expressions for coefficients of (109) and (66) also coincideThus expressions for coefficients of fundamental harmonic ofwaveform (109) are given by (84) where 120582
1is given by (85)
while expressions for coefficients of 119896th harmonic are givenby (45)-(46)
Waveforms described by (108) have exactly one zerowhile waveforms described by (109) for 1(1198962 minus 1) lt 120582
119896lt 1
Mathematical Problems in Engineering 13
14
12
1
08
06
04
02
00 05 1
Amplitude 120582k
Am
plitu
de1205821
k = 2
k = 3
k = 4
Figure 8 Maximal amplitude of fundamental harmonic as a func-tion of amplitude of 119896th harmonic
have exactly two zeros As we mentioned earlier waveforms(109) for 120582
119896= 1 have 119896 zeros
Remark 20 Maximal amplitude of fundamental harmonic ofnonnegative waveforms of type (35) for prescribed amplitudeof 119896th harmonic can be expressed as
1205821= 1 + 120582
119896 (111)
if 0 le 120582119896le 1(119896
2
minus 1) or
1205821=
119896 sin 120585119896 sin 120585 cos (120585119896) minus cos 120585 sin (120585119896)
(112)
if 1(1198962 minus 1) le 120582119896le 1 where 120585 is related to 120582
119896via (68) (or
(86)) and |120585| le 120587From (110) it follows that (111) holds Substitution of (86)
into (85) leads to (112)Notice that 120582
119896= 1(119896
2
minus 1) is the only common point ofthe intervals 0 le 120582
119896le 1(119896
2
minus 1) and 1(1198962
minus 1) le 120582119896le
1 According to (111) 120582119896= 1(119896
2
minus 1) corresponds to 1205821=
1198962
(1198962
minus1) It can be also obtained from (112) by setting 120585 = 0The waveforms corresponding to this pair of amplitudes aremaximally flat nonnegative waveforms
Maximal amplitude of fundamental harmonic of non-negative waveform of type (35) for 119896 le 4 as a function ofamplitude of 119896th harmonic is presented in Figure 8
Remark 21 Maximum value of amplitude of fundamentalharmonic of nonnegative waveform of type (35) is
1205821max =
1
cos (120587 (2119896)) (113)
This maximum value is attained for |120585| = 1205872 (see (112)) Thecorresponding value of amplitude of 119896th harmonic is 120582
119896=
(1119896) tan(120587(2119896)) Nonnegative waveforms of type (35) with1205821= 1205821max have two zeros at 1205910 and 1205910 minus 120587119896 for 120585 = 1205872 or
at 1205910and 1205910+ 120587119896 for 120585 = minus1205872
14
12
1
08
06
04
02
0minus1 minus05 0 05 1
Am
plitu
de1205821
Parameter 120585120587
k = 2k = 3k = 4
Figure 9 Maximal amplitude of fundamental harmonic as a func-tion of parameter 120585
To prove that (113) holds let us first show that the fol-lowing relation holds for 119896 ge 2
cos( 120587
2119896) lt 1 minus
1
1198962 (114)
From 119896 ge 2 it follows that sinc(120587(4119896)) gt sinc(1205874) wheresinc 119909 = (sin119909)119909 and therefore sin(120587(4119896)) gt 1(radic2119896)By using trigonometric identity cos 2119909 = 1 minus 2sin2119909 weimmediately obtain (114)
According to (111) and (112) it is clear that 1205821attains its
maximum value on the interval 1(1198962 minus 1) le 120582119896le 1 Since
120582119896is monotonic function of |120585| on interval |120585| le 120587 (see
Remark 15) it follows that 119889120582119896119889120585 = 0 for 0 lt |120585| lt 120587
Therefore to find critical points of 1205821as a function of 120582
119896
it is sufficient to find critical points of 1205821as a function of
|120585| 0 lt |120585| lt 120587 and consider its values at the end points120585 = 0 and |120585| = 120587 Plot of 120582
1as a function of parameter 120585
for 119896 le 4 is presented in Figure 9 According to (112) firstderivative of 120582
1with respect to 120585 is equal to zero if and only
if (119896 cos 120585 sin(120585119896) minus sin 120585 cos(120585119896)) cos 120585 = 0 On interval0 lt |120585| lt 120587 this is true if and only if |120585| = 1205872 Accordingto (112) 120582
1is equal to 119896
2
(1198962
minus 1) for 120585 = 0 equal to zerofor |120585| = 120587 and equal to 1 cos(120587(2119896)) for |120585| = 1205872 From(114) it follows that 1198962(1198962minus1) lt 1 cos(120587(2119896)) and thereforemaximum value of 120582
1is given by (113) Moreover maximum
value of 1205821is attained for |120585| = 1205872
According to above consideration all nonnegative wave-forms of type (35) having maximum value of amplitude offundamental harmonic can be obtained from (109) by setting|120585| = 1205872 Three of them corresponding to 119896 = 3 120585 = 1205872and three different values of 120591
0(01205876 and1205873) are presented
in Figure 10 Dotted line corresponds to 1205910= 0 (coefficients
of corresponding waveform are 1198861= minus1 119887
1= 1radic3 119886
3= 0
and 1198873= minusradic39) solid line to 120591
0= 1205876 (119886
1= minus2radic3 119887
1= 0
1198863= radic39 and 119887
3= 0) and dashed line to 120591
0= 1205873 (119886
1= minus1
1198871= minus1radic3 119886
3= 0 and 119887
3= radic39)
Proof of Proposition 18 As it has been shown earlier (Propo-sition 6) nonnegative waveform of type (35) with at least
14 Mathematical Problems in Engineering
2
1
0
0 1 2 3 4
Angle 120591120587
Wav
efor
ms
1205910 = 01205910 = 12058761205910 = 1205873
Figure 10 Nonnegative waveforms with maximum amplitude offundamental harmonic for 119896 = 3 and 120585 = 1205872
one zero can be represented in form (38) According to (43)(44) and (36) for amplitude 120582
1of fundamental harmonic of
waveforms of type (38) the following relation holds
1205821= radic(1 + 120582
119896cos 120585)2 + 11989621205822
119896sin2120585 (115)
where 120582119896satisfy (40) and |120585| le 120587
Because of (40) in the quest of finding maximal 1205821for
prescribed 120582119896 we have to consider the following two cases
(Case i)120582119896lt [(119896minus1) cos 120585 + 119896 sin(120585minus120585119896) sin(120585119896)]minus1
(Case ii)120582119896= [(119896minus1) cos 120585+119896 sin(120585minus120585119896) sin(120585119896)]minus1
Case i Since 120582119896lt [(119896 minus 1) cos 120585 + 119896 sin(120585 minus 120585119896) sin(120585119896)]minus1
implies 120582119896
= 1 according to (115) it follows that 1205821
= 0Hence 119889120582
1119889120585 = 0 implies
2120582119896sin 120585 [1 minus (1198962 minus 1) 120582
119896cos 120585] = 0 (116)
Therefore 1198891205821119889120585 = 0 if 120582
119896= 0 (Option 1) or sin 120585 = 0
(Option 2) or (1198962 minus 1)120582119896cos 120585 = 1 (Option 3)
Option 1 According to (115) 120582119896= 0 implies 120582
1= 1 (notice
that this implication shows that 1205821does not depend on 120585 and
therefore we can set 120585 to zero value)
Option 2 According to (115) sin 120585 = 0 implies 1205821= 1 +
120582119896cos 120585 which further leads to the conclusion that 120582
1is
maximal for 120585 = 0 For 120585 = 0 120582119896lt [(119896 minus 1) cos 120585 + 119896 sin(120585 minus
120585119896) sin(120585119896)]minus1 becomes 120582119896lt 1(119896
2
minus 1)
Option 3 This option leads to contradiction To show thatnotice that (119896
2
minus 1)120582119896cos 120585 = 1 and 120582
119896lt [(119896 minus
1) cos 120585 + 119896 sin(120585 minus 120585119896) sin(120585119896)]minus1 imply that (119896 minus 1) cos 120585 gtsin(120585minus120585119896) sin(120585119896) Using (A5) (see Appendices) the latestinequality can be rewritten assum119896minus1
119899=1[cos 120585minuscos((119896minus2119899)120585119896)] gt
0 But from |119896 minus 2119899| lt 119896 119899 = 1 (119896 minus 1) and |120585| le 120587
it follows that all summands are not positive and therefore(119896minus1) cos 120585 gt sin(120585minus120585119896) sin(120585119896) does not hold for |120585| le 120587
Consequently Case i implies 120585 = 0 and 120582119896lt 1(119896
2
minus 1)Finally substitution of 120585 = 0 into (38) leads to (108) whichproves that (108) holds for 120582
119896lt 1(119896
2
minus 1)
Case ii Relation120582119896= [(119896minus1) cos 120585+119896 sin(120585minus120585119896) sin(120585119896)]minus1
according to Proposition 9 and Remark 11 implies that cor-responding waveforms can be expressed via (66)ndash(68) for|120585| le 120587 Furthermore 120582
119896= [(119896 minus 1) cos 120585 + 119896 sin(120585 minus
120585119896) sin(120585119896)]minus1 and |120585| le 120587 imply 1(1198962 minus 1) le 120582119896le 1
This proves that (109) holds for 1(1198962 minus 1) le 120582119896le 1
Finally let us prove that (108) holds for 120582119896= 1(119896
2
minus
1) According to (68) (see also Remark 11) this value of 120582119896
corresponds to 120585 = 0 Furthermore substitution of 120582119896=
1(1198962
minus 1) and 120585 = 0 into (109) leads to (70) which can berewritten as
119879119896(120591) =
[1 minus cos (120591 minus 1205910)]
(1 minus 1198962)
sdot [119896 (119896 minus 1) minus 2
119896minus1
sum
119899=1
(119896 minus 119899) cos (119899 (120591 minus 1205910))]
(117)
Waveform (117) coincides with waveform (108) for 120582119896
=
1(1 minus 1198962
) Consequently (108) holds for 120582119896= 1(1 minus 119896
2
)which completes the proof
5 Nonnegative Waveforms with MaximalAbsolute Value of the Coefficient of CosineTerm of Fundamental Harmonic
In this sectionwe consider general description of nonnegativewaveforms of type (35) with maximal absolute value ofcoefficient 119886
1for prescribed coefficients of 119896th harmonicThis
type of waveform is of particular interest in PA efficiencyanalysis In a number of cases of practical interest eithercurrent or voltage waveform is prescribed In such casesthe problem of finding maximal efficiency of PA can bereduced to the problem of finding nonnegative waveformwith maximal coefficient 119886
1for prescribed coefficients of 119896th
harmonic (see also Section 7)In Section 51 we provide general description of nonneg-
ative waveforms of type (35) with maximal absolute value ofcoefficient 119886
1for prescribed coefficients of 119896th harmonic In
Section 52 we illustrate results of Section 51 for particularcase 119896 = 3
51 Nonnegative Waveforms with Maximal Absolute Value ofCoefficient 119886
1for 119896 ge 2 Waveforms 119879
119896(120591) of type (35) with
1198861ge 0 can be derived from those with 119886
1le 0 by shifting
by 120587 and therefore we can assume without loss of generalitythat 119886
1le 0 Notice that if 119896 is even then shifting 119879
119896(120591) by
120587 produces the same result as replacement of 1198861with minus119886
1
(119886119896remains the same) On the other hand if 119896 is odd then
shifting 119879119896(120591) by 120587 produces the same result as replacement
of 1198861with minus119886
1and 119886119896with minus119886
119896
According to (37) coefficients of 119896th harmonic can beexpressed as
119886119896= 120582119896cos 120575 119887
119896= 120582119896sin 120575 (118)
Mathematical Problems in Engineering 15
where
|120575| le 120587 (119)
Conversely for prescribed coefficients 119886119896and 119887
119896 120575 can be
determined as
120575 = atan 2 (119887119896 119886119896) (120)
where definition of function atan 2(119910 119909) is given by (105)The main result of this section is stated in the following
proposition
Proposition 22 Every nonnegative waveform of type (35)withmaximal absolute value of coefficient 119886
1le 0 for prescribed
coefficients 119886119896and 119887119896of 119896th harmonic can be represented as
119879119896(120591)
= [1 minus cos 120591]
sdot [1 minus (119896 minus 1) 119886119896minus 2
119896minus1
sum
119899=1
(119896 minus 119899) (119886119896cos 119899120591 + 119887
119896sin 119899120591)]
(121)
if 119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) le 1+119886
119896 where 120575 = atan 2(bk
119886119896) or
119879119896(120591) = 120582
119896[1 minus cos(120591 minus (120575 + 120585)
119896)]
sdot [1 minus cos(120591 minus (120575 minus 120585)
119896)]
sdot [1198880+ 2
119896minus2
sum
119899=1
119888119899cos 119899(120591 minus 120575
119896)]
(122)
if 119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) ge 1 + 119886
119896 where 119888
119899 119899 = 0
119896minus2 and 120582119896= radic1198862119896+ 1198872119896are related to 120585 via relations (67) and
(68) respectively and |120585| le 120587
Remark 23 Expression (121) can be obtained from (38) bysetting 120591
0= 0 and 120585 = minus120575 and then replacing 120582
119896cos 120575 with
119886119896(see (118)) and 120582
119896cos(119899120591 minus 120575) with 119886
119896cos 119899120591 + 119887
119896sin 119899120591
(see also (118)) Furthermore insertion of 1205910= 0 and 120585 =
minus120575 into (43)ndash(46) leads to the following relations betweenfundamental and 119896th harmonic coefficients of waveform(121)
1198861= minus (1 + 119886
119896) 119887
1= minus119896119887
119896 (123)
On the other hand expression (122) can be obtained from(66) by replacing 120591
0minus120585119896with 120575119896 Therefore substitution of
1205910minus 120585119896 = 120575119896 in (84) leads to
1198861= minus1205821cos(120575
119896) 119887
1= minus1205821sin(120575
119896) (124)
where 1205821is given by (85)
The fundamental harmonic coefficients 1198861and 1198871of wave-
form of type (35) with maximal absolute value of coefficient1198861le 0 satisfy both relations (123) and (124) if 119886
119896and 119887119896satisfy
1 + 119886119896= 119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) For such waveforms
relations 1205910= 0 and 120585 = minus120575 also hold
Remark 24 Amplitude of 119896th harmonic of nonnegativewaveform of type (35) with maximal absolute value of coeffi-cient 119886
1le 0 and coefficients 119886
119896 119887119896satisfying 1 + 119886
119896=
119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) is
120582119896=
sin (120575119896)119896 sin 120575 cos (120575119896) minus cos 120575 sin (120575119896)
(125)
To show that it is sufficient to substitute 119886119896= 120582119896cos 120575 (see
(118)) into 1 + 119886119896= 119896120582119896[sin 120575 sin(120575119896)] cos(120575119896)
Introducing new variable
119910 = cos(120575119896) (126)
and using the Chebyshev polynomials (eg see Appendices)relations 119886
119896= 120582119896cos 120575 and (125) can be rewritten as
119886119896= 120582119896119881119896(119910) (127)
120582119896=
1
119896119910119880119896minus1
(119910) minus 119881119896(119910)
(128)
where119881119896(119910) and119880
119896(119910) denote the Chebyshev polynomials of
the first and second kind respectively Substitution of (128)into (127) leads to
119886119896119896119910119880119896minus1
(119910) minus (1 + 119886119896) 119881119896(119910) = 0 (129)
which is polynomial equation of 119896th degree in terms of var-iable 119910 From |120575| le 120587 and (126) it follows that
cos(120587119896) le 119910 le 1 (130)
In what follows we show that 119886119896is monotonically increas-
ing function of 119910 on the interval (130) From 120585 = minus120575 (seeRemark 23) and (81) it follows that 120582minus1
119896= (119896 minus 1) cos 120575 +
119896sum119896minus1
119899=1cos((119896 minus 2119899)120575119896) ge 1 and therefore 119886
119896= 120582119896cos 120575 can
be rewritten as
119886119896=
cos 120575(119896 minus 1) cos 120575 + 119896sum119896minus1
119899=1cos ((119896 minus 2119899) 120575119896)
(131)
Obviously 119886119896is even function of 120575 and all cosines in (131)
are monotonically decreasing functions of |120575| on the interval|120575| le 120587 It is easy to show that cos((119896 minus 2119899)120575119896) 119899 =
1 (119896 minus 1) decreases slower than cos 120575 when |120575| increasesThis implies that denominator of the right hand side of(131) decreases slower than numerator Since denominator ispositive for |120575| le 120587 it further implies that 119886
119896is decreasing
function of |120575| on interval |120575| le 120587 Consequently 119886119896is
monotonically increasing function of 119910 on the interval (130)Thus we have shown that 119886
119896is monotonically increasing
function of 119910 on the interval (130) and therefore (129) hasonly one solution that satisfies (130) According to (128) thevalue of 119910 obtained from (129) and (130) either analyticallyor numerically leads to amplitude 120582
119896of 119896th harmonic
16 Mathematical Problems in Engineering
1
05
0
minus05
minus1
minus1 minus05 0 05 1
Coefficient ak
Coe
ffici
entb
k
radica2k+ b2
kle 1
k = 2k = 3k = 4
Figure 11 Plot of (119886119896 119887119896) satisfying 1 + 119886
119896= 119896120582
119896[sin 120575 sin(120575
119896)] cos(120575119896) for 119896 le 4
By solving (129) and (130) for 119896 le 4 we obtain
119910 = radic1 + 1198862
2 (1 minus 1198862) minus1 le 119886
2le1
3
119910 = radic3
4 (1 minus 21198863) minus1 le 119886
3le1
8
119910 =radicradic2 minus 4119886
4+ 1011988624minus 2 (1 minus 119886
4)
4 (1 minus 31198864)
minus1 le 1198864le
1
15
(132)
Insertion of (132) into (128) leads to the following explicitexpressions for the amplitude 120582
119896 119896 le 4
1205822=1
2(1 minus 119886
2) minus1 le 119886
2le1
3 (133)
1205822
3= [
1
3(1 minus 2119886
3)]
3
minus1 le 1198863le1
8 (134)
1205824=1
4(minus1 minus 119886
4+ radic2 minus 4119886
4+ 1011988624) minus1 le 119886
4le
1
15
(135)
Relations (133)ndash(135) define closed lines (see Figure 11) whichseparate points representing waveforms of type (121) frompoints representing waveforms of type (122) For given 119896points inside the corresponding curve refer to nonnegativewaveforms of type (121) whereas points outside curve (andradic1198862119896+ 1198872119896le 1) correspond to nonnegative waveforms of type
(122) Points on the respective curve correspond to the wave-forms which can be expressed in both forms (121) and (122)
Remark 25 Themaximum absolute value of coefficient 1198861of
nonnegative waveform of type (35) is
100381610038161003816100381611988611003816100381610038161003816max =
1
cos (120587 (2119896)) (136)
This maximum value is attained for |120585| = 1205872 and 120575 = 0
(see (124)) Notice that |1198861|max is equal to the maximum value
1205821max of amplitude of fundamental harmonic (see (113))
Coefficients of waveform with maximum absolute value ofcoefficient 119886
1 1198861lt 0 are
1198861= minus
1
cos (120587 (2119896)) 119886
119896=1
119896tan( 120587
(2119896))
1198871= 119887119896= 0
(137)
Waveformdescribed by (137) is cosinewaveformhaving zerosat 120587(2119896) and minus120587(2119896)
In the course of proving (136) notice first that |1198861|max le
1205821max holds According to (123) and (124) maximum of |119886
1|
occurs for 119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) ge 1 + 119886
119896 From (124)
it immediately follows that maximum value of |1198861| is attained
if and only if 1205821= 1205821max and 120575 = 0 which because of
120575119896 = 1205910minus120585119896 further implies 120591
0= 120585119896 Sincemaximumvalue
of 1205821is attained for |120585| = 1205872 it follows that corresponding
waveform has zeros at 120587(2119896) and minus120587(2119896)
Proof of Proposition 22 As it was mentioned earlier in thissection we can assume without loss of generality that 119886
1le 0
We consider waveforms119879119896(120591) of type (35) such that119879
119896(120591) ge 0
and119879119896(120591) = 0 for some 120591
0 Fromassumption that nonnegative
waveform 119879119896(120591) of type (35) has at least one zero it follows
that it can be expressed in form (38)Let us also assume that 120591
0is position of nondegenerate
critical point Therefore 119879119896(1205910) = 0 implies 1198791015840
119896(1205910) = 0 and
11987910158401015840
119896(1205910) gt 0 According to (55) second derivative of 119879
119896(120591) at
1205910can be expressed as 11987910158401015840
119896(1205910) = 1 minus 120582
119896(1198962
minus 1) cos 120585 Since11987910158401015840
119896(1205910) gt 0 it follows immediately that
1 minus 120582119896(1198962
minus 1) cos 120585 gt 0 (138)
Let us further assume that 119879119896(120591) has exactly one zeroThe
problem of finding maximum absolute value of 1198861is con-
nected to the problem of finding maximum of the minimumfunction (see Section 21) If waveforms possess unique globalminimum at nondegenerate critical point then correspond-ing minimum function is a smooth function of parameters[13] Consequently assumption that 119879
119896(120591) has exactly one
zero at nondegenerate critical point leads to the conclusionthat coefficient 119886
1is differentiable function of 120591
0 First
derivative of 1198861(see (43)) with respect to 120591
0 taking into
account that 1205971205851205971205910= 119896 (see (50)) can be expressed in the
following factorized form
1205971198861
1205971205910
= sin 1205910[1 minus 120582
119896(1198962
minus 1) cos 120585] (139)
Mathematical Problems in Engineering 17
From (138) and (139) it is clear that 12059711988611205971205910= 0 if and only if
sin 1205910= 0 According toRemark 12 assumption that119879
119896(120591)has
exactly one zero implies 120582119896lt 1 From (51) (48) and 120582
119896lt 1
it follows that 1198861cos 1205910+ 1198871sin 1205910lt 0 which together with
sin 1205910= 0 implies that 119886
1cos 1205910lt 0 Assumption 119886
1le 0
together with relations 1198861cos 1205910lt 0 and sin 120591
0= 0 further
implies 1198861
= 0 and
1205910= 0 (140)
Insertion of 1205910= 0 into (38) leads to
119879119896(120591)
= [1 minus cos 120591]
sdot [1 minus (119896 minus 1) 120582119896cos 120585 minus 2120582
119896
119896minus1
sum
119899=1
(119896 minus 119899) cos (119899120591 + 120585)]
(141)
Substitution of 1205910= 0 into (45) and (46) yields 119886
119896= 120582119896cos 120585
and 119887119896
= minus120582119896sin 120585 respectively Replacing 120582
119896cos 120585 with
119886119896and 120582
119896cos(119899120591 + 120585) with (119886
119896cos 119899120591 + 119887
119896sin 119899120591) in (141)
immediately leads to (121)Furthermore 119886
119896= 120582119896cos 120585 119887
119896= minus120582
119896sin 120585 and (118)
imply that
120575 = minus120585 (142)
According to (38)ndash(40) and (142) it follows that (141) is non-negative if and only if
120582119896[(119896 minus 1) cos 120575 + 119896 sin (120575 minus 120575119896)
sin (120575119896)] lt 1 (143)
Notice that 119886119896= 120582119896cos 120575 implies that the following relation
holds
120582119896[(119896 minus 1) cos 120575 + 119896 sin (120575 minus 120575119896)
sin (120575119896)]
= minus119886119896+ 119896120582119896
sin 120575sin (120575119896)
cos(120575119896)
(144)
Finally substitution of (144) into (143) leads to 119896120582119896[sin 120575
sin(120575119896)] cos(120575119896) lt 1 + 119886119896 which proves that (121) holds
when 119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) lt 1 + 119886
119896
Apart from nonnegative waveforms with exactly one zeroat nondegenerate critical point in what follows we will alsoconsider other types of nonnegative waveforms with at leastone zero According to Proposition 9 and Remark 11 thesewaveforms can be described by (66)ndash(68) providing that 0 le|120585| le 120587
According to (35) 119879119896(0) ge 0 implies 1 + 119886
1+ 119886119896ge 0
Consequently 1198861le 0 implies that |119886
1| le 1 + 119886
119896 On the other
hand according to (123) |1198861| = 1 + 119886
119896holds for waveforms
of type (121) The converse is also true 1198861le 0 and |119886
1| =
1 + 119886119896imply 119886
1= minus1 minus 119886
119896 which further from (35) implies
119879119896(0) = 0 Therefore in what follows it is enough to consider
only nonnegativewaveformswhich can be described by (66)ndash(68) and 0 le |120585| le 120587 with coefficients 119886
119896and 119887119896satisfying
119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) ge 1 + 119886
119896
For prescribed coefficients 119886119896and 119887119896 the amplitude 120582
119896=
radic1198862119896+ 1198872119896of 119896th harmonic is also prescribed According to
Remark 15 (see also Remark 16) 120582119896is monotonically
decreasing function of 119909 = cos(120585119896) The value of 119909 can beobtained by solving (90) subject to the constraint cos(120587119896) le119909 le 1 Then 120582
1can be determined from (88) From (106) it
immediately follows that maximal absolute value of 1198861le 0
corresponds to 119902 = 0 which from (104) and (120) furtherimplies that
120575 = 1198961205910minus 120585 (145)
Furthermore 119902 = 0 according to (107) implies that waveformzeros are
1205910=(120575 + 120585)
119896 120591
1015840
0= 1205910minus2120585
119896=(120575 minus 120585)
119896 (146)
Substitution of 1205910= (120575 + 120585)119896 into (66) yields (122) which
proves that (122) holds when 119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) ge
1 + 119886119896
In what follows we prove that (121) also holds when119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) = 1 + 119886
119896 Substitution of 119886
119896=
120582119896cos 120575 into 119896120582
119896[sin 120575 sin(120575119896)] cos(120575119896) = 1 + 119886
119896leads to
120582119896[(119896 minus 1) cos 120575 + 119896 sin (120575 minus 120575119896)
sin (120575119896)] = 1 (147)
As we mentioned earlier relation (142) holds for all wave-forms of type (121) Substituting (142) into (147) we obtain
120582119896[(119896 minus 1) cos 120585 + 119896 sin (120585 minus 120585119896)
sin (120585119896)] = 1 (148)
This expression can be rearranged as
120582119896
119896 sin ((119896 minus 1) 120585119896)sin 120585119896
= 1 minus (119896 minus 1) 120582119896cos 120585 (149)
On the other hand for waveforms of type (122) according to(68) relations (148) and (149) also hold Substitution of 120591
0=
(120575 + 120585)119896 (see (145)) and (67) into (122) leads to
119879119896(120591)
= 120582119896[1 minus cos (120591 minus 120591
0)]
sdot [119896 sin ((119896 minus 1) 120585119896)
sin 120585119896minus 2
119896minus1
sum
119899=1
(119896 minus 119899) cos (119899 (120591 minus 1205910) + 120585)]
(150)
Furthermore substitution of (142) into (145) implies that1205910
= 0 Finally substitution of 1205910
= 0 and (149) into(150) leads to (141) Therefore (141) holds when 119896120582
119896[sin 120575
sin(120575119896)] cos(120575119896) = 1 + 119886119896 which in turn shows that (121)
holds when 119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) = 1 + 119886
119896 This
completes the proof
18 Mathematical Problems in Engineering
52 Nonnegative Waveforms with Maximal Absolute Valueof Coefficient 119886
1for 119896 = 3 Nonnegative waveform of type
(35) for 119896 = 3 is widely used in PA design (eg see [10])In this subsection we illustrate results of Section 51 for thisparticular case The case 119896 = 2 is presented in detail in [12]
Coefficients of fundamental harmonic of nonnegativewaveform of type (35) with 119896 = 3 and maximal absolutevalue of coefficient 119886
1le 0 for prescribed coefficients 119886
3and
1198873(1205823= radic11988623+ 11988723) according to (123) (124) (134) (94) and
(120) are equal to
1198861= minus1 minus 119886
3 119887
1= minus3119887
3 (151)
if 12058223le [(1 minus 2119886
3)3]3
1198861= minus1205821cos(120575
3) 119887
1= minus1205821sin(120575
3) (152)
where 1205821= 3(
3radic1205823minus 1205823) and 120575 = atan 2(119887
3 1198863) if [(1 minus
21198863)3]3
le 1205822
3le 1The line 1205822
3= [(1minus2119886
3)3]3 (see case 119896 = 3
in Figure 11) separates points representing waveforms withcoefficients satisfying (151) from points representing wave-forms with coefficients satisfying (152) Waveforms describedby (151) for 1205822
3lt [(1 minus 2119886
3)3]3 have exactly one zero at
1205910= 0 Waveforms described by (151) and (152) for 1205822
3= [(1 minus
21198863)3]3 also have zero at 120591
0= 0 These waveforms as a rule
have exactly two zeros However there are two exceptionsone related to the maximally flat nonnegative waveform withcoefficients 119886
1= minus98 119886
3= 18 and 119887
1= 1198873= 0 which
has only one zero and the other related to the waveform withcoefficients 119886
1= 0 119886
3= minus1 and 119887
1= 1198873= 0 which has three
zerosWaveforms described by (152) for [(1minus21198863)3]3
lt 1205822
3lt
1 have two zeros Waveforms with 1205823= 1 have only third
harmonic (fundamental harmonic is zero)Plot of contours of maximal absolute value of coefficient
1198861 1198861le 0 for prescribed coefficients 119886
3and 1198873is presented
in Figure 12 According to Remark 25 the waveform withmaximum absolute value of 119886
1le 0 is fully described with
the following coefficients 1198861
= minus2radic3 1198863
= radic39 and1198871= 1198873= 0 This waveform has two zeros at plusmn1205876
Two examples of nonnegative waveforms for 119896 = 3
and maximal absolute value of coefficient 1198861 1198861le 0 with
prescribed coefficients 1198863and 1198873are presented in Figure 13
One waveform corresponds to the case 12058223lt [(1 minus 2119886
3)3]3
(solid line) and the other to the case 12058223gt [(1 minus 2119886
3)3]3
(dashed line)Thewaveform represented by solid line has onezero and its coefficients are 119886
3= minus01 119887
3= 01 119886
1= minus09
and 1198871= minus03 Dashed line corresponds to the waveform
having two zeros with coefficients 1198863= minus01 119887
3= 03 119886
1=
minus08844 and 1198871= minus06460 (case 1205822
3gt [(1 minus 2119886
3)3]3)
6 Nonnegative Cosine Waveforms withat Least One Zero
Nonnegative cosine waveforms have proved to be of impor-tance for waveform modelling in PA design (eg see [10])In this section we consider nonnegative cosine waveforms
1
05
0
minus05
minus1
minus1 minus05 0 05 1
Coefficient a3
Coe
ffici
entb
3
02
04
06
08
10
11
Figure 12 Contours ofmaximal absolute value of coefficient 1198861 1198861le
0 as a function of 1198863and 1198873
2
1
0
0 1 2 3 4
Angle 120591120587
Wav
efor
ms
a3 = minus01 b3 = 01
a3 = minus01 b3 = 03
Figure 13 Nonnegative waveforms for 119896 = 3 and maximal absolutevalue of 119886
1 1198861le 0 with prescribed coefficients 119886
3and 1198873
containing fundamental and 119896th harmonic with at least onezero
Cosine waveform with dc component fundamental and119896th harmonic can be obtained from (35) by setting 119887
1= 119887119896=
0 that is
119879119896(120591) = 1 + 119886
1cos 120591 + 119886
119896cos 119896120591 (153)
In Section 61 we provide general description of non-negative cosine waveforms of type (153) with at least onezero We show that nonnegative cosine waveforms with atleast one zero coincide with nonnegative cosine waveformswith maximal absolute value of coefficient 119886
1for prescribed
coefficient 119886119896 In Section 62 we illustrate results of Section 61
for particular case 119896 = 3
61 Nonnegative Cosine Waveforms with at Least One Zerofor 119896 ge 2 Amplitudes of fundamental and 119896th harmonic
Mathematical Problems in Engineering 19
of cosine waveform of type (153) are 1205821= |1198861| and 120582
119896=
|119886119896| respectively According to (42) for nonnegative cosine
waveforms of type (153) the following relation holds
minus1 le 119886119896le 1 (154)
This explains why 119896th harmonic coefficient 119886119896in Proposi-
tion 26 goes through interval [minus1 1]Waveforms (153) with 119886
1ge 0 can be obtained from
waveforms with 1198861le 0 by shifting by 120587 and therefore with-
out loss of generality we can assume that 1198861le 0
Proposition 26 Each nonnegative cosine waveform of type(153) with 119886
1le 0 and at least one zero can be represented as
119879119896(120591) = [1 minus cos 120591] [1 minus (119896 minus 1) 119886
119896minus 2119886119896
119896minus1
sum
119899=1
(119896 minus 119899) cos 119899120591]
(155)
if minus1 le 119886119896le 1(119896
2
minus 1) or
119879119896(120591) = 119886
119896[1 minus cos (120591 minus 120591
0)] [1 minus cos (120591 + 120591
0)]
sdot [1198880+ 2
119896minus2
sum
119899=1
119888119899cos 119899120591]
(156)
where
119888119899=sin ((119896 minus 119899) 120591
0) cos 120591
0minus (119896 minus 119899) cos ((119896 minus 119899) 120591
0) sin 120591
0
sin31205910
(157)
119886119896=
sin 1205910
119896 sin (1198961205910) cos 120591
0minus cos (119896120591
0) sin 120591
0
(158)
100381610038161003816100381612059101003816100381610038161003816 le
120587
119896 (159)
if 1(1198962 minus 1) le 119886119896le 1
Remark 27 Identity [1minuscos(120591minus1205910)][1minuscos(120591+120591
0)] = [cos 120591
0minus
cos 120591]2 implies that (156) can be rewritten as
119879119896(120591) = 119886
119896[cos 120591
0minus cos 120591]2 [119888
0+ 2
119896minus2
sum
119899=1
119888119899cos 119899120591] (160)
Furthermore substitution of (157) into (160) leads to
119879119896(120591) = 119886
119896[cos 120591
0minus cos 120591]
sdot [(119896 minus 1) sin 119896120591
0
sin 1205910
minus 2
119896minus1
sum
119899=1
sin ((119896 minus 119899) 1205910)
sin 1205910
cos 119899120591]
(161)
Remark 28 All nonnegative cosine waveforms of type (153)with at least one zero and 119886
1le 0 except one of them can be
represented either in form (155) or form (156)This exceptionis maximally flat cosine waveform with 119886
1lt 0 which can be
obtained from (155) for 119886119896= 1(119896
2
minus 1) or from (156) for 1205910=
0 Maximally flat cosine waveform with 1198861lt 0 can also be
obtained from (70) by setting 1205910= 0 Furthermore setting
1205910= 0 in (71) leads to maximally flat cosine waveforms for
119896 le 4 and 1198861lt 0
Remark 29 Nonnegative cosine waveform of type (155) with1198861lt 0 and minus1 lt 119886
119896le 1(119896
2
minus 1) has exactly one zero at120591 = 0 Nonnegative cosine waveform described by (156) with1198861lt 0 and 1(1198962 minus 1) lt 119886
119896lt 1 has two zeros at plusmn120591
0 where
0 lt |1205910| lt 120587119896 For 119886
119896= ∓1 nonnegative cosine waveform
of type (153) reduces to 119879119896(120591) = 1 ∓ cos 119896120591 (clearly these two
waveforms both have 119896 zeros)
Remark 30 Transformation of (155) into an additive formleads to the following relation
1198861= minus1 minus 119886
119896 (162)
where minus1 le 119886119896le 1(119896
2
minus1) Similarly transformation of (156)leads to the following relation
1198861= minus119886119896
119896 sin 1198961205910
sin 1205910
(163)
where 119886119896is given by (158) 1(1198962minus1) le 119886
119896le 1 and |120591
0| le 120587119896
Notice that coefficients of maximally flat cosine waveformnamely 119886
119896= 1(119896
2
minus1) and 1198861= minus1198962
(1198962
minus1) satisfy relation(162) They also satisfy relation (163) for 120591
0= 0
Remark 31 Nonnegative cosine waveforms of type (153) withat least one zero coincide with nonnegative cosine waveformswith maximal absolute value of coefficient 119886
1for prescribed
coefficient 119886119896
In proving that Remark 31 holds notice that expression(155) can be obtained from (121) by setting 119887
119896= 0 Fur-
thermore if 119886119896
ge 0 then 120582119896
= 119886119896 which together
with 119887119896
= 0 and (118) implies 120575 = 0 In this case119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) le 1 + 119886
119896becomes 1198962119886
119896le 1 +
119886119896 On the other hand if 119886
119896lt 0 then 120582
119896= minus119886
119896 which
together with 119887119896= 0 and (118) implies |120575| = 120587 In this case
119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) le 1 + 119886
119896becomes 0 le 1 + 119886
119896
Therefore every nonnegative cosine waveform of type (155)has maximal absolute value of coefficient 119886
1for prescribed
coefficient 119886119896 when minus1 le 119886
119896le 1(119896
2
minus 1)Let us now show that expression (156) can be obtained
from (122) by setting 119887119896= 0 and 119886
119896gt 0 For waveforms
of type (122) according to (118) 119887119896= 0 and 119886
119896gt 0 imply
120575 = 0 and 120582119896= 119886119896 Substitution of 120582
119896= 119886119896and 120575 = 0 into
119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) ge 1+119886
119896leads to 119886
119896ge 1(119896
2
minus1)Furthermore substitution of 120575 = 0 into (145) yields 120591
0= 120585119896
Insertion of 120582119896= 119886119896 120575 = 0 and 120591
0= 120585119896 into (122) leads
to (156) Therefore every nonnegative cosine waveform oftype (156) has maximal absolute value of coefficient 119886
1for
prescribed coefficient 119886119896 when 1(1198962 minus 1) le 119886
119896le 1
Proof of Proposition 26 Let us start with nonnegative cosinewaveform of type (153) with 120582
119896= |119886119896| = 1 According to
Remark 7 120582119896= |119886119896| = 1 implies that 120582
1= |1198861| = 0
Substitution of 119886119896
= minus1 into (155) and using (A2) (seeAppendices) lead to 119879
119896(120591) = 1 minus cos 119896120591 Consequently (155)
holds for 119886119896= minus1 On the other hand substitution of 119886
119896= 1
into (158) yields |1205910| = 120587119896 Furthermore substitution of
119886119896= 1 and 120591
0= 120587119896 (or 120591
0= minus120587119896) into (156) along
with performing all multiplications and using (A2) leads to
20 Mathematical Problems in Engineering
119879119896(120591) = 1 + cos 119896120591 Consequently (156)ndash(158) hold for 119886
119896= 1
and |1205910| = 120587119896
It is easy to see that 120582119896= |119886119896| lt 1 and 119879
119896(1205910) = 0 for some
1205910imply 120582
1= |1198861| = 0 Therefore in what follows we assume
that |119886119896| = 1 and 119886
1lt 0
Cosine waveforms are even functions of 120591 Therefore ifnonnegative cosine waveform has exactly one zero it has to beeither at 0 or at 120587 On the other hand if nonnegative cosinewaveform with 119886
1= 0 has exactly two zeros then these zeros
are placed at plusmn1205910 such that 120591
0is neither 0 nor 120587
In order to prove that (155) holds for minus1 lt 119886119896le 1(1minus119896
2
)let us start by referring to the description (38) of nonnegativewaveformswith at least one zero As wementioned earlier fornonnegative cosine waveformwith exactly one zero (denotedby 1205910) it is either 120591
0= 0 or 120591
0= 120587 Therefore in both cases
sin 1205910= 0 Substitution of sin 120591
0= 0 into (43) together with
1198861
= 0 and 120582119896= |119886119896| lt 1 leads to
1205910= 0 (164)
Clearly 1205910= 0 119887
1= 0 and 119887
119896= 0 according to (44) and (46)
imply 120582119896sin 120585 = 0 Since 120582
119896= |119886119896| it follows that |119886
119896| sin 120585 = 0
also holds which further implies 120582119896= 119886119896= 0 or sin 120585 =
0 In the case when 120582119896= 119886119896= 0 from (164) and (43) we
obtain 1198861= minus1 which further implies that 119879
119896(120591) = 1 minus cos 120591
Consequently (155) holds for 119886119896= 0 In the case when sin 120585 =
0 from (164) and (45) we obtain 119886119896= 120582119896if 120585 = 0 or 119886
119896= minus120582119896
if 120585 = 120587 Relations 119886119896= 120582119896and 120585 = 0 according to (40) imply
that 0 le 119886119896le 1(1 minus 119896
2
) Substitution of 120585 = 0 120582119896= 119886119896 and
(164) into (38) leads to (155) which proves that (155) holdsfor 0 le 119886
119896le 1(1 minus 119896
2
) On the other hand relations 119886119896=
minus120582119896and 120585 = 120587 according to (40) imply that minus1 lt 119886
119896le 0
Substitution of 120585 = 120587 120582119896= minus119886119896 and (164) into (38) also leads
to (155) which proves that (155) also holds for minus1 lt 119886119896le 0
Consequently (155) holds for minus1 lt 119886119896le 1(1 minus 119896
2
)In what follows we first prove that (156)-(157) hold for
1(1 minus 1198962
) lt 119886119896lt 1 For this purpose let us start with non-
negative waveforms with two zeros described by (66) As wementioned before nonnegative cosine waveforms with twozeros have zeros at 120591
0and minus120591
0 such that 120591
0= 0 and 120591
0= 120587
Relations 1198861
lt 0 and 1198871
= 0 according to (84) implycos(1205910minus 120585119896) = 1 and therefore
120585
119896= 1205910 (165)
From 120585119896 = 1205910and 0 lt |120585| lt 120587 it follows that 0 lt |120591
0| lt
120587119896 Insertion of 120585119896 = 1205910into (45) yields 119886
119896= 120582119896 Relations
119886119896= 120582119896and (82) imply that 1(1 minus 1198962) lt 119886
119896lt 1 Substitution
of 120582119896= 119886119896and 120585119896 = 120591
0into (66)ndash(68) leads to (156)ndash(158)
which proves that (156)ndash(158) hold for 1(1 minus 1198962
) lt 119886119896lt 1
and 0 lt |1205910| lt 120587119896
Finally substitution of 119886119896= 1(1 minus 119896
2
) and 1205910= 0 into
(161) leads to
119879119896(120591) =
[1 minus cos 120591](1 minus 1198962)
[119896 (119896 minus 1) minus 2
119896minus1
sum
119899=1
(119896 minus 119899) cos 119899120591]
(166)
Waveform (166) coincides with waveform (155) for 119886119896
=
1(1 minus 1198962
) which in turn proves that (156) holds for 119886119896=
1(1 minus 1198962
) and 1205910= 0 This completes the proof
62 Nonnegative CosineWaveforms with at Least One Zero for119896 = 3 In this subsection we consider nonnegative cosinewaveforms with at least one zero for 119896 = 3 (for case 119896 = 2
see [12])Cosine waveform with fundamental and third harmonic
reads
1198793(120591) = 1 + 119886
1cos 120591 + 119886
3cos 3120591 (167)
For 1198861le 0 and minus1 le 119886
3le 18 according to (155) non-
negative cosine waveform of type (167) with at least one zerocan be expressed as
1198793(120591) = (1 minus cos 120591) [1 minus 2119886
3(1 + 2 cos 120591 + cos 2120591)] (168)
From 1198793(120591 + 120587) = 2 minus 119879
3(120591) it immediately follows that for
1198861ge 0 and minus18 le 119886
3le 1 119879
3(120591) can be expressed as
1198793(120591) = (1 + cos 120591) [1 + 2119886
3(1 minus 2 cos 120591 + cos 2120591)] (169)
For 1198861le 0 and 18 le 119886
3le 1 from (158) it follows that 119886
3=
[8cos31205910]minus1 This relation along with (160) and (157) further
implies that 1198793(120591) can be expressed as
1198793(120591) =
[cos 1205910minus cos 120591]2 [2 cos 120591
0+ cos 120591]
2cos31205910
(170)
providing that |1205910| le 1205873 From 119879
3(120591 + 120587) = 2 minus 119879
3(120591) it
follows that (170) also holds for 1198861ge 0 and minus1 le 119886
3le minus18
providing that 1205910isin [21205873 41205873]
Maximally flat nonnegative cosinewaveformof type (167)with 119886
1lt 0 (minimum at 120591
0= 0) reads 119879
3(120591) = [1 minus
cos 120591]2[1 + (12) cos 120591] Dually maximally flat nonnegativecosine waveform with 119886
1gt 0 (minimum at 120591
0= 120587) reads
1198793(120591) = [1 + cos 120591]2[1 minus (12) cos 120591]In what follows we provide relations between coefficients
1198861and 1198863of nonnegative cosine waveforms of type (167) with
at least one zeroFor 1198861le 0 conversion of (168) into an additive form
immediately leads to the following relation
1198861= minus1 minus 119886
3for minus 1 le 119886
3le1
8 (171)
Conversion of (170) into an additive form leads to 1198861
=
minus31198863(2 cos 2120591
0+ 1) which can be also expressed as 119886
1=
minus31198863(4cos2120591
0minus 1) For 119886
1le 0 relations |120591
0| le 1205873 119886
1=
minus31198863(4cos2120591
0minus 1) and 119886
3= [8cos3120591
0]minus1 lead to
1198861= minus3 [ 3radic119886
3minus 1198863] for 1
8le 1198863le 1 (172)
Similarly for 1198861ge 0 conversion of (169) into an additive form
leads to the following relation
1198861= 1 minus 119886
3for minus
1
8le 1198863le 1 (173)
Mathematical Problems in Engineering 21
1
15
15
05
0
minus05
minus15minus15
minus1
minus1 minus05 0 05 1
Coefficient a1
Coe
ffici
enta
3 a1= minus
3(3radica3minus a3)
a1 =
minus1 minus a3
a1 =
1 minusa3
a1 =3(3radic|a3
| +a3)
Figure 14 Parameter space of cosine waveforms for 119896 = 3
For waveform of type (170) with 1198861ge 0 relations 120591
0isin [21205873
41205873] 1198861= minus3119886
3(4cos2120591
0minus 1) and 119886
3= [8cos3120591
0]minus1 lead to
1198861= 3 [
3radic10038161003816100381610038161198863
1003816100381610038161003816 + 1198863] for minus 1 le 1198863le minus
1
8 (174)
Every cosine waveform of type (167) corresponds to apair of real numbers (119886
1 1198863) and vice versa Points (119886
1 1198863)
in grey area in Figure 14 correspond to nonnegative cosinewaveforms for 119896 = 3 The points at the boundary of grey areacorrespond to nonnegative cosinewaveformswith at least onezero A number of shapes of nonnegative cosine waveformswith 119896 = 3 and at least one zero plotted on interval [minus120587 120587]are also presented in Figure 14 The boundary of grey area inFigure 14 consists of four line segments described by relations(171)ndash(174) The common point of line segments (172) and(173) is cusp point with coordinates 119886
1= 0 and 119886
3= 1
Another cusp point with coordinates 1198861= 0 and 119886
3= minus1
is the common point of line segments (171) and (174) Thecommon point of line segments (171)-(172) has coordinates(minus98 18) and common point of line segments (173)-(174)has coordinates (98 minus18) These points are representedby white circle dots and they correspond to maximallyflat cosine waveforms (eg see [21]) White triangle dotswith coordinates (2radic3 minusradic39) and (minus2radic3radic39) refer tothe nonnegative cosine waveforms with maximum value ofamplitude of fundamental harmonic
7 Four Case Studies of Usage of NonnegativeWaveforms in PA Efficiency Analysis
In this section we provide four case studies of usage ofdescription of nonnegative waveforms with fundamental and119896th harmonic in PA efficiency analysis In first two casestudies to be presented in Section 71 voltage is nonnegativewaveform with fundamental and second harmonic with atleast one zero In remaining two case studies to be consideredin Section 72 voltage waveform contains fundamental andthird harmonic
i
v Load
Vdc
Idc
Lch
vL
iL
Cb
+ +in
Figure 15 Generic PA circuit diagram
Let us consider generic PA circuit diagram as shown inFigure 15 We assume here that voltage and current wave-forms at the transistor output are
V (120579) = 1 + 1198861V cos 120579 + 1198871V sin 120579 + 119886119896V cos 119896120579 + 119887119896V sin 119896120579
119894 (120579) = 1 + 1198861119894cos 120579 +
infin
sum
119899=2
119886119899119894cos 119899120579
(175)
where 120579 stands for 120596119905 Both waveforms are normalized in thesense that dc components of voltage and current are 119881dc =
1 and 119868dc = 1 respectively Under assumption that blockingcapacitor 119862
119887behaves as short-circuit at the fundamental and
higher harmonics current and voltage waveforms at the loadare
V119871(120579) = 119886
1V cos 120579 + 1198871V sin 120579 + 119886119896V cos 119896120579 + 119887119896V sin 119896120579
119894119871(120579) = minus119886
1119894cos 120579 minus
infin
sum
119899=2
119886119899119894cos 119899120579
(176)
In terms of coefficients of voltage and current waveforms theload impedance at fundamental harmonic is 119911
1= minus(119886
1V minus
1198951198871V)1198861119894 whereas load impedance at 119896th harmonic is 119911
119896=
minus(119886119896Vminus119895119887119896V)119886119896119894 All other harmonics are short-circuited (119911
119899=
0 for 119899 = 1 and 119899 = 119896) Time average output power of PA (egsee [10]) with waveform pair (175) at fundamental frequencycan be expressed as
1198751= minus
11988611198941198861V
2 (177)
For normalized waveforms (175) with 119881dc = 1 and 119868dc = 1dc power is 119875dc = 1 Consequently PA efficiency 120578 = 119875
1119875dc
(eg see [10 26]) is equal to
120578 = minus11988611198941198861V
2 (178)
Thus time average output power 1198751of PA with pair of nor-
malized waveform (175) is equal to efficiency (178)Power utilization factor (PUF) is defined [26] as ldquothe
ratio of power delivered in a given situation to the power
22 Mathematical Problems in Engineering
delivered by the same device with the same supply voltagein Class A moderdquo Since the output power in class-A modeis 1198751class-A = max[V(120579)] sdotmax[119894(120579)]8 (eg see [9]) it follows
that power utilization factor PUF = 11987511198751class-A for PA with
pair of normalized waveforms (175) can be expressed as
PUF =8120578
max [V (120579)] sdotmax [119894 (120579)] (179)
71 NonnegativeWaveforms for 119896 = 2 in PAEfficiencyAnalysisIn this subsection we provide two case studies of usage ofdescription of nonnegative waveforms with fundamental andsecond harmonic (119896 = 2) in PA efficiency analysis For moreexamples of usage of descriptions of nonnegative waveformswith fundamental and second harmonic in PA efficiencyanalysis see [12]
Case Study 71 In this case study we consider efficiency of PAfor given second harmonic impedance providing that voltageis nonnegative waveform with fundamental and second har-monic and current is ldquohalf-sinerdquo waveform frequently used inefficiency analysis of classical PA operation (eg see [10])
Standard model of current waveform for classical PAoperation has the form (eg see [10 26])
119894119863(120579) =
119868119863[cos 120579 minus cos(120572
2)] |120579| le
120572
2
0120572
2le |120579| le 120587
(180)
where 120572 is conduction angle and 119868119863
gt 0 Since 119894119863(120579) is
even function it immediately follows that its Fourier seriescontains only dc component and cosine terms
119894119863(120579) = 119868dc +
infin
sum
119899=1
119868119899cos 119899120579 (181)
The dc component of the waveform (180) is
119868dc =119868119863120572
2120587[sinc(120572
2) minus cos(120572
2)] (182)
where sinc119909 = (sin119909)119909 The coefficient of the fundamentalharmonic component reads
1198681=119868119863120572
2120587(1 minus sinc120572) (183)
and the coefficient of 119899th harmonic component can bewrittenin the form
119868119899=119868119863
119899120587[sin ((119899 minus 1) 1205722)
(119899 minus 1)minussin ((119899 + 1) 1205722)
(119899 + 1)] 119899 ge 2
(184)
For ldquohalf-sinerdquo current waveform conduction angle is equalto 120587 (class-B conduction angle) According to (182) thisfurther implies that 119868dc = 119868
119863120587 To obtain normalized form of
waveform (180) we set 119868dc = 1which implies that 119868119863= 120587 Fur-
thermore substitution of 120572 = 120587 and 119868119863= 120587 in (180) leads to
119894 (120579) =
120587 cos 120579 |120579| lt120587
2
0120587
2lt |120579| le 120587
(185)
Similarly substitution of 119868119863= 120587 and 120572 = 120587 into (183) and
(184) leads to the coefficients of waveform (185) Coefficientsof fundamental and second harmonic respectively are
1198861119894=120587
2 119886
2119894=2
3 (186)
On the other hand voltage waveform of type (35) for 119896 =2 reads
V (120579) = 1 + 1198861V cos 120579 + 1198871V sin 120579 + 1198862V cos 2120579 + 1198872V sin 2120579
(187)
This waveform contains only fundamental and second har-monic and therefore all harmonics of order higher thantwo are short-circuited (119911
119899= 0 for 119899 gt 2) For current
voltage pair (185) and (187) load impedance at fundamentalharmonic is 119911
1= minus(119886
1V minus 1198951198871V)1198861119894 whereas load impedance
at second harmonic is 1199112= minus(119886
2Vminus1198951198872V)1198862119894 According to ourassumption the load is passive and therefore Re119911
1 gt 0 and
Re1199112 ge 0 which further imply 119886
11198941198861V lt 0 and 119886
21198941198862V le 0
respectivelyIt is easy to see that problem of findingmaximal efficiency
of PA with current-voltage pair (185) and (187) for prescribedsecond harmonic impedance can be reduced to the problemof finding voltage waveform of type (187) with maximal coef-ficient |119886
1V| for prescribed coefficients of second harmonic(see Section 5)
The following algorithm (analogous to Algorithm 22presented in [12]) provides the procedure for calculation ofmaximal efficiency with current-voltage pair (185) and (187)for prescribed second harmonic impedance The definitionof function atan 2(119910 119909) which appears in the step (iii) of thefollowing algorithm is given by (105)
Algorithm 32 (i) Choose 1199112= 1199032+1198951199092such that |119911
2| le 1|119886
2119894|
(ii) calculate 1198862V minus 1198951198872V = minus119911
21198862119894and 120582
2V = radic11988622V + 1198872
2V(iii) if 2120582
2V le 1 minus 1198862V then calculate 119886
1V = minus1 minus 1198862V and
1198871V = minus2119887
2V else calculate 1205821V = radic81205822V(1 minus 1205822V) 1205790V minus 120585V2 =
(12)atan2(1198872V 1198862V) 1198861V = minus120582
1V cos(1205790V minus 120585V2) and 1198871V =
minus1205821V sin(1205790V minus 120585V2)(iv) calculate efficiency 120578 = minus119886
11198941198861V2
(v) calculate 1199111= minus(119886
1V minus 1198951198871V)1198861119894 and 1199112119899 = 1199112Re119911
1
In this case study coefficients of fundamental and secondharmonic of current waveform are given by (186) Maximalefficiency of PA associated with the waveform pair (185)and (187) as a function of normalized second harmonicimpedance 119911
2119899= 1199112Re119911
1 is presented in Figure 16(a)
As can be seen from Figure 16(a) efficiency of 078 isachieved at the edge of Smith chart where second harmonicimpedance has small resistive part Corresponding PUFcalculated according to (179) is presented in Figure 16(b)Peak efficiency 120578 = 1205874 = 07854 and peak value of PUF = 1
are attained when second harmonic is short-circuited (whichcorresponds to ideal class-B operation [10 26])
For example for second harmonic impedance 1199112= 01 minus
11989505 and currentwaveform (185) fromAlgorithm32 it followsthat 2120582
2V le 1 minus 1198862V Furthermore according to step (iii)
of above algorithm maximal efficiency of PA is attained
Mathematical Problems in Engineering 23
+j2
+j5
+j1
+j2
+j5
minusj2
minusj5
minusj1
minusj2
minusj5
infin0
078
02
05
1 2 5
075 07 065
06 05
075
07
065
06
05
120578 lt 05
(a)
099
095
091
083 075
067
0 59 051
051
099
095
091
083
075 0
67 059
+j5
+j1
+j2
+j5
minusj5
minusj1
minusj2
minusj5
infin
+j2
minusj2
0 02
05
1 2 5
120578 lt 05
(b)
Figure 16 (a) Contours of maximal efficiency of PA and (b) contours of corresponding PUF associated with the waveform pair (185) and(187) as functions of normalized second harmonic impedance 119911
2119899= 1199112Re119911
1
2
3
4
1
0
2 3 410
Wav
efor
ms
Angle 120579120587
i(120579)
(120579)
Figure 17 Waveform pair (185) and (187) that provides maximalefficiency for 119911
2= 01 minus 11989505
with voltage waveform of type (187) with coefficients 1198862V =
minus00667 1198872V = minus03333 119886
1V = minus09333 and 1198871V = 06667 (see
Figure 17) Corresponding efficiency PUF and normalizedsecond harmonic impedance are 120578 = 07330 PUF = 07572and 1199112119899= 01683 minus 11989508415 respectively
On the other hand for second harmonic impedance 1199112=
01 minus 11989508 and current waveform (185) from Algorithm 32 itfollows that 2120582
2V gt 1 minus 1198862V Then according to step (iii) of
above algorithm maximal efficiency is attained with voltagewaveform of type (187) with coefficients 119886
2V = minus00667 1198872V =
minus05333 1198861V = minus09333 and 119887
1V = 10572 (see Figure 18)Efficiency PUF andnormalized secondharmonic impedanceare 120578 = 07330 PUF = 06332 and 119911
2119899= 01683 minus 11989513465
respectively
2
3
4
1
0
2 3 410
Wav
efor
ms
Angle 120579120587
i(120579)
(120579)
Figure 18 Waveform pair (185) and (187) that provides maximalefficiency for 119911
2= 01 minus 11989508
Case Study 72 As another case study let us consider the effi-ciency of PA providing that current waveform is nonnegativecosine waveform up to third harmonic with maximum valueof amplitude of fundamental harmonic [22] (see also [8])
119894 (120579) = 1 +1 + radic5
2cos 120579 + 2radic5
5cos 2120579 + 5 minus radic5
10cos 3120579
(188)
and voltage waveform is nonnegative waveform of type(187) Load impedances at fundamental second and thirdharmonic are 119911
1= minus(119886
1Vminus1198951198871V)1198861119894 1199112 = minus(1198862Vminus1198951198872V)1198862119894 and
1199113= 0 respectively According to our assumption the load
is passive and therefore Re1199111 gt 0 and Re119911
2 ge 0 which
further imply 11988611198941198861V lt 0 and 119886
21198941198862V le 0 respectively
Because current waveform (188) contains only cosineterms and voltage waveform is the same as in previous case
24 Mathematical Problems in Engineering
08
075
07
06
06
065
065
05
05
075 0
7
120578 lt 05
+j2
+j5
+j1
+j2
+j5
minusj2
minusj5
minusj1
minusj2
minusj5
infin0
02
05
1 2 5
(a)
085
08 0
75 07 0
6
05 04
08
075
0405
060
7
120578 lt 05
+j2
+j5
+j1
+j2
+j5
minusj2
minusj5
minusj1
minusj2
minusj5
infin0 02
05
1 2 5
(b)
Figure 19 (a) Contours of maximal efficiency of PA and (b) contours of corresponding PUF associated with the waveform pair (187)-(188)as functions of normalized second harmonic impedance 119911
2119899= 1199112Re119911
1
study the procedure for calculation of maximal efficiency ofPA with waveform pair (187)-(188) is the same as presentedin Algorithm 32 In this case study the coefficients of funda-mental and second harmonic of current waveform are 119886
1119894=
(1 + radic5)2 and 1198862119894= 2radic55 respectively
Maximal efficiency of PA associated with the waveformpair (187)-(188) as a function of normalized secondharmonicimpedance 119911
2119899= 1199112Re119911
1 is presented in Figure 19(a)
Efficiency of 08 is achieved at the edge of Smith chart wheresecond harmonic impedance has small resistive part Thetheoretical upper bound 120578 = (1 + radic5)4 asymp 08090 isattained when second harmonic is short-circuitedWhen thisupper bound is reached both second and third harmonic areshort-circuited which implies that we are dealing with finiteharmonic class-C [6 8] or dually when current and voltageinterchange their roles with finite harmonic inverse class-C[6 9] Corresponding PUF calculated according to (179) ispresented in Figure 19(b) Peak value of PUF asymp 08541 isattained when second harmonic is short-circuited
For example for second harmonic impedance 1199112
=
007 minus 11989504 and current waveform (188) from Algorithm 32 itfollows that 2120582
2V le 1 minus 1198862V Furthermore according to step
(iii) of Algorithm 32 maximal efficiency of PA is attainedwith voltage waveform of type (187) with coefficients 119886
2V =
minus00626 1198872V = minus03578 119886
1V = minus09374 and 1198871V = 07155 (see
Figure 20) Corresponding efficiency PUF and normalizedsecond harmonic impedance are 120578 = 07584 PUF = 06337and 1199112119899= 01208 minus 11989506904 respectively
On the other hand for 1199112= 005 minus 11989507 and current
waveform (187) it follows that 21205822V gt 1minus119886
2VThen accordingto step (iii) of Algorithm 32 themaximal efficiency is attainedwith voltage waveform of type (187) with coefficients 119886
2V =
minus00447 1198872V = minus06261 119886
1V = minus09318 and 1198871V = 10007 (see
Figure 21) Efficiency PUF and normalized second harmonic
2
3
4
1
0
2 3 410
Wav
efor
ms
Angle 120579120587
i(120579)
(120579)
Figure 20 Waveform pair (187)-(188) that provides maximal effi-ciency for 119911
2= 007 minus 11989504
impedance are 120578 = 07538 PUF = 05314 and 1199112119899= 00868minus
11989512156 respectively
72 Nonnegative Waveforms for 119896 = 3 in PA EfficiencyAnalysis In this subsection we provide another two casestudies of usage of description of nonnegative waveforms inPA efficiency analysis this time with fundamental and thirdharmonic (119896 = 3)
Case Study 73 Let us consider current-voltage pair such thatvoltage is nonnegative waveform with fundamental and thirdharmonic
V (120579) = 1 + 1198861V cos 120579 + 1198871V sin 120579 + 1198863V cos 3120579 + 1198873V sin 3120579
(189)
Mathematical Problems in Engineering 25
2
3
4
1
0
2 3 410
Wav
efor
ms
Angle 120579120587
i(120579)
(120579)
Figure 21 Waveform pair (187)-(188) that provides maximal effi-ciency for 119911
2= 005 minus 11989507
and current is nonnegative cosine waveform given by (188)Load impedances at fundamental second and third har-monic are 119911
1= minus(119886
1V minus 1198951198871V)1198861119894 1199112 = 0 and 119911
3= minus(119886
3V minus
1198951198873V)1198863119894 respectively According to our assumption the load
is passive and therefore Re1199111 gt 0 and Re119911
3 ge 0 which
further imply 11988611198941198861V lt 0 and 119886
31198941198863V le 0
In this subsection we consider the problem of findingmaximal efficiency of PA with waveform pair (188)-(189)for given third harmonic impedance As we mentionedearlier problem of finding maximal efficiency of PA withcurrent-voltage pair (188)-(189) for prescribed third har-monic impedance can be reduced to the problem of findingvoltage waveform of type (189) withmaximal coefficient |119886
1V|for prescribed coefficients of third harmonic (see Section 52)
The following algorithm provides the procedure forcalculation of maximal efficiency with current-voltage pair(188)-(189) The definition of function atan 2(119910 119909) whichappears in step (iii) of the following algorithm is given by(105)
Algorithm 33 (i) Choose 1199113= 1199033+1198951199093such that |119911
3| le 1|119886
3119894|
(ii) calculate 1198863V minus 1198951198873V = minus119911
31198863119894and 120582
3V = radic11988623V + 1198872
3V(iii) if 271205822
3V le (1 minus 21198863V)3 then calculate 119886
1V = minus1 minus 1198863V
and 1198871V = minus3119887
3V else calculate 1205821V = 3(3radic1205823V minus 120582
3V) 1205790V minus120585V3 = (13)atan 2(119887
3V 1198863V) 1198861V = minus1205821V cos(1205790V minus 120585V3) and
1198871V = minus120582
1V sin(1205790V minus 120585V3)(iv) calculate efficiency 120578 = minus119886
11198941198861V2
(v) calculate 1199111= minus(119886
1V minus 1198951198871V)1198861119894 and 1199113119899 = 1199113Re119911
1
In this case study coefficients of fundamental and thirdharmonic of current waveform are 119886
1119894= (1 + radic5)2 and
1198863119894= (5 minus radic5)10 respectively For the waveform pair (188)-
(189) maximal efficiency of PA as a function of normalizedthird harmonic impedance 119911
3119899= 1199113Re119911
1 is presented in
Figure 22 Efficiency of 08 is reached when third harmonicimpedance has small resistive part Peak efficiency 120578 = (1 +
radic5)4 asymp 08090 is achieved when third harmonic is short-circuited
+j2
+j5
+j1
+j2
+j5
minusj2
minusj5
minusj1
minusj2
minusj5
infin0
08
08
075
075
07
07
06
05
02
05
1 2
Figure 22 Contours of maximal efficiency of PA associated withthe waveform pair (188)-(189) as a function of normalized thirdharmonic impedance 119911
3119899= 1199113Re119911
1
For the present case study in what follows we show thatpower utilization factor is proportional to efficiency For volt-age waveform of type (189) it is easy to see that V(120579 + 120587) =
2minusV(120579) holdsThis relation along with the fact that waveformV(120579) that provides maximal efficiency has at least one zeroimplies that max[V(120579)] = 2 On the other hand currentwaveform (188) is cosine waveform with positive coefficientsand therefore max[119894(120579)] = 119894(0) = 2 + 4radic5 Consequentlyaccording to (179) the following relation holds
PUFCase study 73 = 2 (5 minus 2radic5) 120578 = 10557120578 (190)
Clearly the ratio PUF120578 is constant and therefore in this casestudy PUF can be easily calculated from the correspondingefficiency Accordingly peak efficiency and peak value ofPUFCase study 73 = 3radic52 minus 52 = 08541 are attained forthe same voltage waveform (when third harmonic is short-circuited)
In the first example current waveform (188) and 1199113=
02 minus 11989505 imply that 2712058223V le (1 minus 2119886
3V)3 Then according
to Algorithm 33 the voltage waveform of type (189) thatprovides maximal efficiency has the following coefficients1198863V = minus00553 119887
3V = minus01382 1198861V = minus09447 and 119887
1V =
04146 (see Figure 23) Efficiency PUF and normalized thirdharmonic impedance are 120578 = 07643 PUF = 08069 and1199113119899= 03425 minus 11989508564 respectivelyIn the second example current waveform (188) and 119911
3=
01 minus 11989511 imply that 2712058223V gt (1 minus 2119886
3V)3 Then according
to Algorithm 33 the voltage waveform of type (189) thatprovides maximal efficiency has the following coefficients1198863V = minus00276 119887
3V = minus03040 1198861V = minus09391 and 119887
1V =
05807 (see Figure 24) Efficiency PUF and normalized thirdharmonic impedance are 120578 = 07598 PUF = 08021 and1199113119899= 01723 minus 11989518952 respectively
26 Mathematical Problems in Engineering
2
3
4
1
0
2 3 410
Wav
efor
ms
Angle 120579120587
i(120579)
(120579)
Figure 23 Waveform pair (188)-(189) that provides maximal effi-ciency for 119911
3= 02 minus 11989505
2
3
4
1
0
2 3 410
Wav
efor
ms
Angle 120579120587
i(120579)
(120579)
Figure 24 Waveform pair (188)-(189) that provides maximal effi-ciency for 119911
3= 01 minus 11989511
Case Study 74 In this case study let us consider current-voltage pair where current is normalized waveform of type(180) with conduction angle 120572 = 115120587 (207∘) and voltageis nonnegative waveform of type (189) Substitution of 120572 =
115120587 and 119868dc = 1 into (182) leads to 119868119863
= 22535Furthermore substitution of 120572 = 115120587 and 119868
119863= 22535 into
(180) leads to
119894 (120579)
=
22535 [cos 120579 minus cos(1151205872
)] |120579| le115120587
2
0115120587
2le |120579| le 120587
(191)
Similarly substitution of 120572 = 115120587 and 119868119863
= 22535 into(183) and (184) for 119899 = 3 yields coefficients of fundamentaland third harmonic of waveform (191)
1198861119894= 14586 119886
3119894= minus01026 (192)
Because current waveform (191) contains only cosineterms and voltage waveform is the same as in previous case
+j2
+j5
+j1
+j2
+j5
minusj2
minusj5
minusj1
minusj2
minusj5
infin0 02
05
1 2
074
076
08
084
082
078
076
074
Figure 25 Contours of maximal efficiency of PA associated withthe waveform pair (189) and (191) as a function of normalized thirdharmonic impedance 119911
3119899= 1199113Re119911
1
study the procedure for calculation of maximal efficiencyof PA with waveform pair (189)ndash(191) is the same as thatpresented in Algorithm 33 In this case study the coefficientsof fundamental and third harmonic of current waveform aregiven by (192)
For the waveform pair (189) and (191) maximal efficiencyof PA as a function of normalized third harmonic impedance1199113119899= 1199113Re119911
1 is presented in Figure 25 Efficiency of 084
is obtained in vicinity of 1199113119899= 23685 (corresponding to 119911
3=
18750) Peak efficiency 120578 asymp 08421 is achieved for voltagewaveform of type (189) with coefficients 119886
1V = minus2radic3 1198863V =
radic39 and 1198871V = 1198873V = 0
In the course of finding power utilization factor noticethat currentwaveformof type (191) attains itsmaximumvaluefor 120579 = 0 Insertion of max[119894(120579)] = 119894(0) = 278 andmax[V(120579)] = 2 for voltage waveform of type (189) into (179)leads to
PUFCase study 74 = 1439120578 (193)
Again the ratio PUF120578 is constant and PUF can be easilycalculated from the corresponding efficiency Accordinglypeak value of PUFCase study 74 asymp 12118 and peak efficiencyare attained for the same voltage waveform
In the first example current waveform (191) and 1199113=
1 minus 11989502 imply that 2712058223V le (1 minus 2119886
3V)3 Then according to
Algorithm 33 voltage waveform of type (189) which providesmaximal efficiency has coefficients 119886
3V = 01026 1198873V =
00205 1198861V = minus11026 and 119887
1V = minus00616 (see Figure 26)Efficiency PUF and normalized third harmonic impedanceare 120578 = 08042 PUF = 11572 and 119911
3119899= 13228 minus 11989502646
respectivelyIn second example current waveform (191) and 119911
3=
15 minus 11989512 imply that 2712058223V gt (1 minus 2119886
3V)3 Then according to
Algorithm 33 voltage waveform of type (189) which providesmaximal efficiency has coefficients 119886
3V = 01540 1198873V =
Mathematical Problems in Engineering 27
2
3
4
1
0
2 3 410
Wav
efor
ms
Angle 120579120587
i(120579)
(120579)
Figure 26 Waveform pair (189) and (191) that provides maximalefficiency for 119911
3= 1 minus 11989502
2
3
4
1
0
2 3 410
Wav
efor
ms
Angle 120579120587
i(120579)
(120579)
Figure 27 Waveform pair (189) and (191) that provides maximalefficiency for 119911
3= 15 minus 11989512
01232 1198861V = minus11255 and 119887
1V = minus02575 (see Figure 27)Efficiency PUF and normalized third harmonic impedanceare 120578 = 08208 PUF = 11812 and 119911
3119899= 19439 minus 11989515552
respectively
8 Conclusion
In this paper we consider a problem of finding general de-scriptions of various classes of nonnegative waveforms withfundamental and 119896th harmonic These classes include non-negative waveforms with at least one zero nonnegative wave-forms with maximal amplitude of fundamental harmonic forprescribed amplitude of 119896th harmonic nonnegative wave-forms withmaximal coefficient of cosine part of fundamentalharmonic for prescribed coefficients of 119896th harmonic andnonnegative cosine waveforms with at least one zero Mainresults are stated in six propositions (Propositions 1 6 918 22 and 26) four corollaries (Corollaries 2ndash5) twentyremarks and three algorithms Four case studies of usageof closed form descriptions of nonnegative waveforms in PAefficiency analysis are considered in detail in Section 7
Appendices
Here we provide a list of finite sums of trigonometric func-tions used in this paper (Appendix A) and brief account ofthe Chebyshev polynomials (Appendix B)
A List of Some Finite Sums ofTrigonometric Functions
Dirichlet kernel (eg see [27]) is as follows
119863119896minus1
(120591) = 1 + 2
119896minus1
sum
119899=1
cos 119899120591 = sin ((2119896 minus 1) 1205912)sin (1205912)
(A1)
Fejer kernel (eg see [27]) can be expressed in the fol-lowing equivalent forms
119865119896minus1
(120591) =1
119896
119896minus1
sum
119899=0
119863119899(120591) = 1 +
2
119896
119896minus1
sum
119899=1
(119896 minus 119899) cos 119899120591
=(1 minus cos 119896120591)119896 (1 minus cos 120591)
(A2)
Lagrangersquos trigonometric identity (eg see [28]) is as fol-lows
1198781(120591) =
119896minus1
sum
119899=1
sin 119899120591 = sin (1198961205912) sin ((119896 minus 1) 1205912)sin (1205912)
(A3)
In what follows we show that the following three trigono-metric identities also hold
2
119896minus1
sum
119899=1
(119896 minus 119899) sin 119899120591 = 119896 sin 120591 minus sin 1198961205911 minus cos 120591
(A4)
119896minus1
sum
119899=1
cos (119896 minus 2119899) 120591 = sin (119896 minus 1) 120591sin 120591
(A5)
119896minus1
sum
119899=1
119899 (119896 minus 119899) cos (119896 minus 2119899) 120591
=sin (119896120591) cos 120591 minus 119896 cos (119896120591) sin 120591
2sin3120591
(A6)
Denote 1198782(120591) = 2sum
119896minus1
119899=1(119896 minus 119899) sin 119899120591 119878
3(120591) = sum
119896minus1
119899=1cos(119896 minus
2119899)120591 and 1198784(120591) = sum
119896minus1
119899=1119899(119896 minus 119899) cos(119896 minus 2119899)120591
Notice that 1198782(120591) = 2119896119878
1(120591) + 119889119863
119896minus1(120591)119889120591 which
immediately leads to (A4)
28 Mathematical Problems in Engineering
Identity (A5) can be obtained as follows
sin (119896 minus 1) 120591sin 120591
=119890119895(119896minus1)120591
minus 119890minus119895(119896minus1)120591
119890119895120591 minus 119890minus119895120591
= 119890119895119896120591
119890minus2119895120591
minus 119890minus2119895(119896minus1)120591
1 minus 119890minus2119895120591
= 119890119895119896120591
119896minus1
sum
119899=1
119890minus2119895119899120591
=
119896minus1
sum
119899=1
119890119895(119896minus2119899)120591
=
119896minus1
sum
119899=1
cos (119896 minus 2119899) 120591
(A7)
From 4119899(119896 minus 119899) = 1198962
minus (119896 minus 2119899)2 it follows that 4119878
4(120591) =
1198962
1198783(120591) + 119889
2
1198783(120591)119889120591
2 which leads to (A6)
B The Chebyshev Polynomials
The Chebyshev polynomials of the first kind 119881119899(119909) can be
defined by the following relation (eg see [29])
119881119899(119909) = cos 119899120591 when 119909 = cos 120591 (B1)
The Chebyshev polynomials of the second kind 119880119899(119909) can be
defined by the following relation (eg see [29])
119880119899(119909) =
sin (119899 + 1) 120591sin 120591
when 119909 = cos 120591 (B2)
The Chebyshev polynomials satisfy the following recur-rence relations (eg see [29])
1198810(119909) = 1 119881
1(119909) = 119909
119881119899+1
(119909) = 2119909119881119899(119909) minus 119881
119899minus1(119909)
1198800(119909) = 1 119880
1(119909) = 2119909
119880119899+1
(119909) = 2119909119880119899(119909) minus 119880
119899minus1(119909)
(B3)
The first few Chebyshev polynomials of the first and secondkind are 119881
2(119909) = 2119909
2
minus 1 1198813(119909) = 4119909
3
minus 3119909 1198814(119909) = 8119909
4
minus
81199092
+ 1 1198802(119909) = 4119909
2
minus 1 1198803(119909) = 8119909
3
minus 4119909 and 1198804(119909) =
161199094
minus 121199092
+ 1
Conflict of Interests
The authors declare that there is no conflict of interestsregarding the publication of this paper
Acknowledgment
This work is supported by the SerbianMinistry of EducationScience and Technology Development as a part of ProjectTP32016
References
[1] V I Arnolrsquod V S Afrajmovich Y S Ilrsquoyashenko and L PShilrsquonikov Dynamical Systems V Bifurcation Theory and Catas-trophe Theory Springer Berlin Germany 1994
[2] E Polak ldquoOn the mathematical foundations of nondifferen-tiable optimization in engineering designrdquo SIAM Review vol29 no 1 pp 21ndash89 1987
[3] N S Fuzik ldquoBiharmonic modes of a tuned RF power amplifierrdquoRadiotehnika vol 25 no 7 pp 62ndash71 1970 (Russian)
[4] P Colantonio F Giannini G Leuzzi and E Limiti ldquoClass Gapproach for low-voltage high-efficiency PA designrdquo Interna-tional Journal of RF and Microwave Computer-Aided Engineer-ing vol 10 no 6 pp 366ndash378 2000
[5] F H Raab ldquoMaximum efficiency and output of class-F poweramplifiersrdquo IEEE Transactions on Microwave Theory and Tech-niques vol 49 no 6 pp 1162ndash1166 2001
[6] FH Raab ldquoClass-E class-C and class-F power amplifiers basedupon a finite number of harmonicsrdquo IEEE Transactions onMicrowaveTheory and Techniques vol 49 no 8 pp 1462ndash14682001
[7] J D Rhodes ldquoOutput universality inmaximum efficiency linearpower amplifiersrdquo International Journal of Circuit Theory andApplications vol 31 no 4 pp 385ndash405 2003
[8] A Juhas and L A Novak ldquoComments on lsquoClass-E class-Cand classF power amplifier based upon a finite number ofharmonicsrsquordquo IEEE Transactions on Microwave Theory and Tech-niques vol 57 no 6 pp 1623ndash1625 2009
[9] M Roberg and Z Popovic ldquoAnalysis of high-efficiency poweramplifiers with arbitrary output harmonic terminationsrdquo IEEETransactions on Microwave Theory and Techniques vol 59 no8 pp 2037ndash2048 2011
[10] A Grebennikov N O Sokal and M J Franco Switchmode RFPower Amplifiers ElsevierAcademic Press San Diego CalifUSA 2nd edition 2012
[11] T Canning P J Tasker and S C Cripps ldquoContinuous modepower amplifier design using harmonic clipping contourstheory and practicerdquo IEEE Transactions on Microwave Theoryand Techniques vol 62 no 1 pp 100ndash110 2014
[12] A Juhas and L A Novak ldquoGeneral description of nonnegativewaveforms up to second harmonic for power amplifier mod-ellingrdquoMathematical Problems in Engineering vol 2014 ArticleID 709762 18 pages 2014
[13] V I Arnolrsquod V V Goryunov O V Lyashko and V A VasilrsquoevDynamical Systems VIIImdashSingularity Theory II ApplicationsSpringer Berlin Germany 1993
[14] D Siersma ldquoProperties of conflict sets in the planerdquo BanachCenter Publications Polish Academy of Sciences vol 50 no 1 pp267ndash276 1999 Proceedings of the Banach Center Symposiumon Geometry and Topology of Caustics (Caustics rsquo98) WarsawPoland
[15] M van Manen The geometry of conflict sets [Dissertation]Universiteit Utrecht UtrechtThe Netherlands 2003 httpigi-tur-archivelibraryuunldissertations2003-0912-123058c4pdf
[16] Y L Sachkov ldquoMaxwell strata and symmetries in the problemofoptimal rolling of a sphere over a planerdquo Sbornik Mathematicsvol 201 no 7-8 pp 1029ndash1051 2010
[17] I A Bogaevsky ldquoPerestroikas of shocks and singularities ofminimum functionsrdquoPhysicaDNonlinear Phenomena vol 173no 1-2 pp 1ndash28 2002
[18] Y L Sachkov ldquoMaxwell strata in the Euler elastic problemrdquoJournal of Dynamical andControl Systems vol 14 no 2 pp 169ndash234 2008
[19] M Siino and T Koike ldquoTopological classification of black holesgeneric Maxwell set and crease set of a horizonrdquo InternationalJournal ofModern Physics D Gravitation Astrophysics Cosmol-ogy vol 20 no 6 pp 1095ndash1122 2011
Mathematical Problems in Engineering 29
[20] F H Raab ldquoClass-F power amplifiers with maximally flat wave-formsrdquo IEEETransactions onMicrowaveTheory and Techniquesvol 45 no 11 pp 2007ndash2012 1997
[21] A Juhas and L A Novak ldquoMaximally flat waveforms with finitenumber of harmonics in class-F power amplifiersrdquo Mathemat-ical Problems in Engineering vol 2013 Article ID 169590 9pages 2013
[22] L Fejer ldquoUber trigonometrische polynomerdquo Journal fur dieReine und Angewandte Mathematik vol 1916 no 146 pp 53ndash82 1916 (German)
[23] S C Cripps ldquoBessel Waives [microwave bytes]rdquo IEEE Micro-wave Magazine vol 10 no 7 pp 30ndash36 117 2009
[24] L N Bryzgalova ldquoSingularities of the maximum of para-metrically dependent functionrdquo Functional Analysis and ItsApplications vol 11 no 1 pp 49ndash51 1977
[25] V I Arnold A A Davydov V A Vassiliev and V MZakalyukin Mathematical Models of Catastrophes Control ofCatastrophic Process Encyclopedia of Life Support Systems(EOLSS) EOLSS Publishers Oxford UK 2006
[26] S C Cripps RF Power Amplifiers for Wireless CommunicationsArtech House Norwood Mass USA 2nd edition 2006
[27] A Zygmund Trigonometric Series vol 1 Cambridge UniversityPress Cambridge UK 2nd edition 1959
[28] A Jeffrey and H DaiHandbook of Mathematical Formulas andIntegrals ElsevierAcademic Press San Diego Calif USA 4thedition 2008
[29] J C Mason and D C Handscomb Chebyshev PolynomialsChapman amp Hall CRC Press Boca Raton Fla USA 2003
Submit your manuscripts athttpwwwhindawicom
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical Problems in Engineering
Hindawi Publishing Corporationhttpwwwhindawicom
Differential EquationsInternational Journal of
Volume 2014
Applied MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Probability and StatisticsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical PhysicsAdvances in
Complex AnalysisJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
OptimizationJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
CombinatoricsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Operations ResearchAdvances in
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Function Spaces
Abstract and Applied AnalysisHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of Mathematics and Mathematical Sciences
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Algebra
Discrete Dynamics in Nature and Society
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Decision SciencesAdvances in
Discrete MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom
Volume 2014 Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Stochastic AnalysisInternational Journal of
Mathematical Problems in Engineering 13
14
12
1
08
06
04
02
00 05 1
Amplitude 120582k
Am
plitu
de1205821
k = 2
k = 3
k = 4
Figure 8 Maximal amplitude of fundamental harmonic as a func-tion of amplitude of 119896th harmonic
have exactly two zeros As we mentioned earlier waveforms(109) for 120582
119896= 1 have 119896 zeros
Remark 20 Maximal amplitude of fundamental harmonic ofnonnegative waveforms of type (35) for prescribed amplitudeof 119896th harmonic can be expressed as
1205821= 1 + 120582
119896 (111)
if 0 le 120582119896le 1(119896
2
minus 1) or
1205821=
119896 sin 120585119896 sin 120585 cos (120585119896) minus cos 120585 sin (120585119896)
(112)
if 1(1198962 minus 1) le 120582119896le 1 where 120585 is related to 120582
119896via (68) (or
(86)) and |120585| le 120587From (110) it follows that (111) holds Substitution of (86)
into (85) leads to (112)Notice that 120582
119896= 1(119896
2
minus 1) is the only common point ofthe intervals 0 le 120582
119896le 1(119896
2
minus 1) and 1(1198962
minus 1) le 120582119896le
1 According to (111) 120582119896= 1(119896
2
minus 1) corresponds to 1205821=
1198962
(1198962
minus1) It can be also obtained from (112) by setting 120585 = 0The waveforms corresponding to this pair of amplitudes aremaximally flat nonnegative waveforms
Maximal amplitude of fundamental harmonic of non-negative waveform of type (35) for 119896 le 4 as a function ofamplitude of 119896th harmonic is presented in Figure 8
Remark 21 Maximum value of amplitude of fundamentalharmonic of nonnegative waveform of type (35) is
1205821max =
1
cos (120587 (2119896)) (113)
This maximum value is attained for |120585| = 1205872 (see (112)) Thecorresponding value of amplitude of 119896th harmonic is 120582
119896=
(1119896) tan(120587(2119896)) Nonnegative waveforms of type (35) with1205821= 1205821max have two zeros at 1205910 and 1205910 minus 120587119896 for 120585 = 1205872 or
at 1205910and 1205910+ 120587119896 for 120585 = minus1205872
14
12
1
08
06
04
02
0minus1 minus05 0 05 1
Am
plitu
de1205821
Parameter 120585120587
k = 2k = 3k = 4
Figure 9 Maximal amplitude of fundamental harmonic as a func-tion of parameter 120585
To prove that (113) holds let us first show that the fol-lowing relation holds for 119896 ge 2
cos( 120587
2119896) lt 1 minus
1
1198962 (114)
From 119896 ge 2 it follows that sinc(120587(4119896)) gt sinc(1205874) wheresinc 119909 = (sin119909)119909 and therefore sin(120587(4119896)) gt 1(radic2119896)By using trigonometric identity cos 2119909 = 1 minus 2sin2119909 weimmediately obtain (114)
According to (111) and (112) it is clear that 1205821attains its
maximum value on the interval 1(1198962 minus 1) le 120582119896le 1 Since
120582119896is monotonic function of |120585| on interval |120585| le 120587 (see
Remark 15) it follows that 119889120582119896119889120585 = 0 for 0 lt |120585| lt 120587
Therefore to find critical points of 1205821as a function of 120582
119896
it is sufficient to find critical points of 1205821as a function of
|120585| 0 lt |120585| lt 120587 and consider its values at the end points120585 = 0 and |120585| = 120587 Plot of 120582
1as a function of parameter 120585
for 119896 le 4 is presented in Figure 9 According to (112) firstderivative of 120582
1with respect to 120585 is equal to zero if and only
if (119896 cos 120585 sin(120585119896) minus sin 120585 cos(120585119896)) cos 120585 = 0 On interval0 lt |120585| lt 120587 this is true if and only if |120585| = 1205872 Accordingto (112) 120582
1is equal to 119896
2
(1198962
minus 1) for 120585 = 0 equal to zerofor |120585| = 120587 and equal to 1 cos(120587(2119896)) for |120585| = 1205872 From(114) it follows that 1198962(1198962minus1) lt 1 cos(120587(2119896)) and thereforemaximum value of 120582
1is given by (113) Moreover maximum
value of 1205821is attained for |120585| = 1205872
According to above consideration all nonnegative wave-forms of type (35) having maximum value of amplitude offundamental harmonic can be obtained from (109) by setting|120585| = 1205872 Three of them corresponding to 119896 = 3 120585 = 1205872and three different values of 120591
0(01205876 and1205873) are presented
in Figure 10 Dotted line corresponds to 1205910= 0 (coefficients
of corresponding waveform are 1198861= minus1 119887
1= 1radic3 119886
3= 0
and 1198873= minusradic39) solid line to 120591
0= 1205876 (119886
1= minus2radic3 119887
1= 0
1198863= radic39 and 119887
3= 0) and dashed line to 120591
0= 1205873 (119886
1= minus1
1198871= minus1radic3 119886
3= 0 and 119887
3= radic39)
Proof of Proposition 18 As it has been shown earlier (Propo-sition 6) nonnegative waveform of type (35) with at least
14 Mathematical Problems in Engineering
2
1
0
0 1 2 3 4
Angle 120591120587
Wav
efor
ms
1205910 = 01205910 = 12058761205910 = 1205873
Figure 10 Nonnegative waveforms with maximum amplitude offundamental harmonic for 119896 = 3 and 120585 = 1205872
one zero can be represented in form (38) According to (43)(44) and (36) for amplitude 120582
1of fundamental harmonic of
waveforms of type (38) the following relation holds
1205821= radic(1 + 120582
119896cos 120585)2 + 11989621205822
119896sin2120585 (115)
where 120582119896satisfy (40) and |120585| le 120587
Because of (40) in the quest of finding maximal 1205821for
prescribed 120582119896 we have to consider the following two cases
(Case i)120582119896lt [(119896minus1) cos 120585 + 119896 sin(120585minus120585119896) sin(120585119896)]minus1
(Case ii)120582119896= [(119896minus1) cos 120585+119896 sin(120585minus120585119896) sin(120585119896)]minus1
Case i Since 120582119896lt [(119896 minus 1) cos 120585 + 119896 sin(120585 minus 120585119896) sin(120585119896)]minus1
implies 120582119896
= 1 according to (115) it follows that 1205821
= 0Hence 119889120582
1119889120585 = 0 implies
2120582119896sin 120585 [1 minus (1198962 minus 1) 120582
119896cos 120585] = 0 (116)
Therefore 1198891205821119889120585 = 0 if 120582
119896= 0 (Option 1) or sin 120585 = 0
(Option 2) or (1198962 minus 1)120582119896cos 120585 = 1 (Option 3)
Option 1 According to (115) 120582119896= 0 implies 120582
1= 1 (notice
that this implication shows that 1205821does not depend on 120585 and
therefore we can set 120585 to zero value)
Option 2 According to (115) sin 120585 = 0 implies 1205821= 1 +
120582119896cos 120585 which further leads to the conclusion that 120582
1is
maximal for 120585 = 0 For 120585 = 0 120582119896lt [(119896 minus 1) cos 120585 + 119896 sin(120585 minus
120585119896) sin(120585119896)]minus1 becomes 120582119896lt 1(119896
2
minus 1)
Option 3 This option leads to contradiction To show thatnotice that (119896
2
minus 1)120582119896cos 120585 = 1 and 120582
119896lt [(119896 minus
1) cos 120585 + 119896 sin(120585 minus 120585119896) sin(120585119896)]minus1 imply that (119896 minus 1) cos 120585 gtsin(120585minus120585119896) sin(120585119896) Using (A5) (see Appendices) the latestinequality can be rewritten assum119896minus1
119899=1[cos 120585minuscos((119896minus2119899)120585119896)] gt
0 But from |119896 minus 2119899| lt 119896 119899 = 1 (119896 minus 1) and |120585| le 120587
it follows that all summands are not positive and therefore(119896minus1) cos 120585 gt sin(120585minus120585119896) sin(120585119896) does not hold for |120585| le 120587
Consequently Case i implies 120585 = 0 and 120582119896lt 1(119896
2
minus 1)Finally substitution of 120585 = 0 into (38) leads to (108) whichproves that (108) holds for 120582
119896lt 1(119896
2
minus 1)
Case ii Relation120582119896= [(119896minus1) cos 120585+119896 sin(120585minus120585119896) sin(120585119896)]minus1
according to Proposition 9 and Remark 11 implies that cor-responding waveforms can be expressed via (66)ndash(68) for|120585| le 120587 Furthermore 120582
119896= [(119896 minus 1) cos 120585 + 119896 sin(120585 minus
120585119896) sin(120585119896)]minus1 and |120585| le 120587 imply 1(1198962 minus 1) le 120582119896le 1
This proves that (109) holds for 1(1198962 minus 1) le 120582119896le 1
Finally let us prove that (108) holds for 120582119896= 1(119896
2
minus
1) According to (68) (see also Remark 11) this value of 120582119896
corresponds to 120585 = 0 Furthermore substitution of 120582119896=
1(1198962
minus 1) and 120585 = 0 into (109) leads to (70) which can berewritten as
119879119896(120591) =
[1 minus cos (120591 minus 1205910)]
(1 minus 1198962)
sdot [119896 (119896 minus 1) minus 2
119896minus1
sum
119899=1
(119896 minus 119899) cos (119899 (120591 minus 1205910))]
(117)
Waveform (117) coincides with waveform (108) for 120582119896
=
1(1 minus 1198962
) Consequently (108) holds for 120582119896= 1(1 minus 119896
2
)which completes the proof
5 Nonnegative Waveforms with MaximalAbsolute Value of the Coefficient of CosineTerm of Fundamental Harmonic
In this sectionwe consider general description of nonnegativewaveforms of type (35) with maximal absolute value ofcoefficient 119886
1for prescribed coefficients of 119896th harmonicThis
type of waveform is of particular interest in PA efficiencyanalysis In a number of cases of practical interest eithercurrent or voltage waveform is prescribed In such casesthe problem of finding maximal efficiency of PA can bereduced to the problem of finding nonnegative waveformwith maximal coefficient 119886
1for prescribed coefficients of 119896th
harmonic (see also Section 7)In Section 51 we provide general description of nonneg-
ative waveforms of type (35) with maximal absolute value ofcoefficient 119886
1for prescribed coefficients of 119896th harmonic In
Section 52 we illustrate results of Section 51 for particularcase 119896 = 3
51 Nonnegative Waveforms with Maximal Absolute Value ofCoefficient 119886
1for 119896 ge 2 Waveforms 119879
119896(120591) of type (35) with
1198861ge 0 can be derived from those with 119886
1le 0 by shifting
by 120587 and therefore we can assume without loss of generalitythat 119886
1le 0 Notice that if 119896 is even then shifting 119879
119896(120591) by
120587 produces the same result as replacement of 1198861with minus119886
1
(119886119896remains the same) On the other hand if 119896 is odd then
shifting 119879119896(120591) by 120587 produces the same result as replacement
of 1198861with minus119886
1and 119886119896with minus119886
119896
According to (37) coefficients of 119896th harmonic can beexpressed as
119886119896= 120582119896cos 120575 119887
119896= 120582119896sin 120575 (118)
Mathematical Problems in Engineering 15
where
|120575| le 120587 (119)
Conversely for prescribed coefficients 119886119896and 119887
119896 120575 can be
determined as
120575 = atan 2 (119887119896 119886119896) (120)
where definition of function atan 2(119910 119909) is given by (105)The main result of this section is stated in the following
proposition
Proposition 22 Every nonnegative waveform of type (35)withmaximal absolute value of coefficient 119886
1le 0 for prescribed
coefficients 119886119896and 119887119896of 119896th harmonic can be represented as
119879119896(120591)
= [1 minus cos 120591]
sdot [1 minus (119896 minus 1) 119886119896minus 2
119896minus1
sum
119899=1
(119896 minus 119899) (119886119896cos 119899120591 + 119887
119896sin 119899120591)]
(121)
if 119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) le 1+119886
119896 where 120575 = atan 2(bk
119886119896) or
119879119896(120591) = 120582
119896[1 minus cos(120591 minus (120575 + 120585)
119896)]
sdot [1 minus cos(120591 minus (120575 minus 120585)
119896)]
sdot [1198880+ 2
119896minus2
sum
119899=1
119888119899cos 119899(120591 minus 120575
119896)]
(122)
if 119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) ge 1 + 119886
119896 where 119888
119899 119899 = 0
119896minus2 and 120582119896= radic1198862119896+ 1198872119896are related to 120585 via relations (67) and
(68) respectively and |120585| le 120587
Remark 23 Expression (121) can be obtained from (38) bysetting 120591
0= 0 and 120585 = minus120575 and then replacing 120582
119896cos 120575 with
119886119896(see (118)) and 120582
119896cos(119899120591 minus 120575) with 119886
119896cos 119899120591 + 119887
119896sin 119899120591
(see also (118)) Furthermore insertion of 1205910= 0 and 120585 =
minus120575 into (43)ndash(46) leads to the following relations betweenfundamental and 119896th harmonic coefficients of waveform(121)
1198861= minus (1 + 119886
119896) 119887
1= minus119896119887
119896 (123)
On the other hand expression (122) can be obtained from(66) by replacing 120591
0minus120585119896with 120575119896 Therefore substitution of
1205910minus 120585119896 = 120575119896 in (84) leads to
1198861= minus1205821cos(120575
119896) 119887
1= minus1205821sin(120575
119896) (124)
where 1205821is given by (85)
The fundamental harmonic coefficients 1198861and 1198871of wave-
form of type (35) with maximal absolute value of coefficient1198861le 0 satisfy both relations (123) and (124) if 119886
119896and 119887119896satisfy
1 + 119886119896= 119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) For such waveforms
relations 1205910= 0 and 120585 = minus120575 also hold
Remark 24 Amplitude of 119896th harmonic of nonnegativewaveform of type (35) with maximal absolute value of coeffi-cient 119886
1le 0 and coefficients 119886
119896 119887119896satisfying 1 + 119886
119896=
119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) is
120582119896=
sin (120575119896)119896 sin 120575 cos (120575119896) minus cos 120575 sin (120575119896)
(125)
To show that it is sufficient to substitute 119886119896= 120582119896cos 120575 (see
(118)) into 1 + 119886119896= 119896120582119896[sin 120575 sin(120575119896)] cos(120575119896)
Introducing new variable
119910 = cos(120575119896) (126)
and using the Chebyshev polynomials (eg see Appendices)relations 119886
119896= 120582119896cos 120575 and (125) can be rewritten as
119886119896= 120582119896119881119896(119910) (127)
120582119896=
1
119896119910119880119896minus1
(119910) minus 119881119896(119910)
(128)
where119881119896(119910) and119880
119896(119910) denote the Chebyshev polynomials of
the first and second kind respectively Substitution of (128)into (127) leads to
119886119896119896119910119880119896minus1
(119910) minus (1 + 119886119896) 119881119896(119910) = 0 (129)
which is polynomial equation of 119896th degree in terms of var-iable 119910 From |120575| le 120587 and (126) it follows that
cos(120587119896) le 119910 le 1 (130)
In what follows we show that 119886119896is monotonically increas-
ing function of 119910 on the interval (130) From 120585 = minus120575 (seeRemark 23) and (81) it follows that 120582minus1
119896= (119896 minus 1) cos 120575 +
119896sum119896minus1
119899=1cos((119896 minus 2119899)120575119896) ge 1 and therefore 119886
119896= 120582119896cos 120575 can
be rewritten as
119886119896=
cos 120575(119896 minus 1) cos 120575 + 119896sum119896minus1
119899=1cos ((119896 minus 2119899) 120575119896)
(131)
Obviously 119886119896is even function of 120575 and all cosines in (131)
are monotonically decreasing functions of |120575| on the interval|120575| le 120587 It is easy to show that cos((119896 minus 2119899)120575119896) 119899 =
1 (119896 minus 1) decreases slower than cos 120575 when |120575| increasesThis implies that denominator of the right hand side of(131) decreases slower than numerator Since denominator ispositive for |120575| le 120587 it further implies that 119886
119896is decreasing
function of |120575| on interval |120575| le 120587 Consequently 119886119896is
monotonically increasing function of 119910 on the interval (130)Thus we have shown that 119886
119896is monotonically increasing
function of 119910 on the interval (130) and therefore (129) hasonly one solution that satisfies (130) According to (128) thevalue of 119910 obtained from (129) and (130) either analyticallyor numerically leads to amplitude 120582
119896of 119896th harmonic
16 Mathematical Problems in Engineering
1
05
0
minus05
minus1
minus1 minus05 0 05 1
Coefficient ak
Coe
ffici
entb
k
radica2k+ b2
kle 1
k = 2k = 3k = 4
Figure 11 Plot of (119886119896 119887119896) satisfying 1 + 119886
119896= 119896120582
119896[sin 120575 sin(120575
119896)] cos(120575119896) for 119896 le 4
By solving (129) and (130) for 119896 le 4 we obtain
119910 = radic1 + 1198862
2 (1 minus 1198862) minus1 le 119886
2le1
3
119910 = radic3
4 (1 minus 21198863) minus1 le 119886
3le1
8
119910 =radicradic2 minus 4119886
4+ 1011988624minus 2 (1 minus 119886
4)
4 (1 minus 31198864)
minus1 le 1198864le
1
15
(132)
Insertion of (132) into (128) leads to the following explicitexpressions for the amplitude 120582
119896 119896 le 4
1205822=1
2(1 minus 119886
2) minus1 le 119886
2le1
3 (133)
1205822
3= [
1
3(1 minus 2119886
3)]
3
minus1 le 1198863le1
8 (134)
1205824=1
4(minus1 minus 119886
4+ radic2 minus 4119886
4+ 1011988624) minus1 le 119886
4le
1
15
(135)
Relations (133)ndash(135) define closed lines (see Figure 11) whichseparate points representing waveforms of type (121) frompoints representing waveforms of type (122) For given 119896points inside the corresponding curve refer to nonnegativewaveforms of type (121) whereas points outside curve (andradic1198862119896+ 1198872119896le 1) correspond to nonnegative waveforms of type
(122) Points on the respective curve correspond to the wave-forms which can be expressed in both forms (121) and (122)
Remark 25 Themaximum absolute value of coefficient 1198861of
nonnegative waveform of type (35) is
100381610038161003816100381611988611003816100381610038161003816max =
1
cos (120587 (2119896)) (136)
This maximum value is attained for |120585| = 1205872 and 120575 = 0
(see (124)) Notice that |1198861|max is equal to the maximum value
1205821max of amplitude of fundamental harmonic (see (113))
Coefficients of waveform with maximum absolute value ofcoefficient 119886
1 1198861lt 0 are
1198861= minus
1
cos (120587 (2119896)) 119886
119896=1
119896tan( 120587
(2119896))
1198871= 119887119896= 0
(137)
Waveformdescribed by (137) is cosinewaveformhaving zerosat 120587(2119896) and minus120587(2119896)
In the course of proving (136) notice first that |1198861|max le
1205821max holds According to (123) and (124) maximum of |119886
1|
occurs for 119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) ge 1 + 119886
119896 From (124)
it immediately follows that maximum value of |1198861| is attained
if and only if 1205821= 1205821max and 120575 = 0 which because of
120575119896 = 1205910minus120585119896 further implies 120591
0= 120585119896 Sincemaximumvalue
of 1205821is attained for |120585| = 1205872 it follows that corresponding
waveform has zeros at 120587(2119896) and minus120587(2119896)
Proof of Proposition 22 As it was mentioned earlier in thissection we can assume without loss of generality that 119886
1le 0
We consider waveforms119879119896(120591) of type (35) such that119879
119896(120591) ge 0
and119879119896(120591) = 0 for some 120591
0 Fromassumption that nonnegative
waveform 119879119896(120591) of type (35) has at least one zero it follows
that it can be expressed in form (38)Let us also assume that 120591
0is position of nondegenerate
critical point Therefore 119879119896(1205910) = 0 implies 1198791015840
119896(1205910) = 0 and
11987910158401015840
119896(1205910) gt 0 According to (55) second derivative of 119879
119896(120591) at
1205910can be expressed as 11987910158401015840
119896(1205910) = 1 minus 120582
119896(1198962
minus 1) cos 120585 Since11987910158401015840
119896(1205910) gt 0 it follows immediately that
1 minus 120582119896(1198962
minus 1) cos 120585 gt 0 (138)
Let us further assume that 119879119896(120591) has exactly one zeroThe
problem of finding maximum absolute value of 1198861is con-
nected to the problem of finding maximum of the minimumfunction (see Section 21) If waveforms possess unique globalminimum at nondegenerate critical point then correspond-ing minimum function is a smooth function of parameters[13] Consequently assumption that 119879
119896(120591) has exactly one
zero at nondegenerate critical point leads to the conclusionthat coefficient 119886
1is differentiable function of 120591
0 First
derivative of 1198861(see (43)) with respect to 120591
0 taking into
account that 1205971205851205971205910= 119896 (see (50)) can be expressed in the
following factorized form
1205971198861
1205971205910
= sin 1205910[1 minus 120582
119896(1198962
minus 1) cos 120585] (139)
Mathematical Problems in Engineering 17
From (138) and (139) it is clear that 12059711988611205971205910= 0 if and only if
sin 1205910= 0 According toRemark 12 assumption that119879
119896(120591)has
exactly one zero implies 120582119896lt 1 From (51) (48) and 120582
119896lt 1
it follows that 1198861cos 1205910+ 1198871sin 1205910lt 0 which together with
sin 1205910= 0 implies that 119886
1cos 1205910lt 0 Assumption 119886
1le 0
together with relations 1198861cos 1205910lt 0 and sin 120591
0= 0 further
implies 1198861
= 0 and
1205910= 0 (140)
Insertion of 1205910= 0 into (38) leads to
119879119896(120591)
= [1 minus cos 120591]
sdot [1 minus (119896 minus 1) 120582119896cos 120585 minus 2120582
119896
119896minus1
sum
119899=1
(119896 minus 119899) cos (119899120591 + 120585)]
(141)
Substitution of 1205910= 0 into (45) and (46) yields 119886
119896= 120582119896cos 120585
and 119887119896
= minus120582119896sin 120585 respectively Replacing 120582
119896cos 120585 with
119886119896and 120582
119896cos(119899120591 + 120585) with (119886
119896cos 119899120591 + 119887
119896sin 119899120591) in (141)
immediately leads to (121)Furthermore 119886
119896= 120582119896cos 120585 119887
119896= minus120582
119896sin 120585 and (118)
imply that
120575 = minus120585 (142)
According to (38)ndash(40) and (142) it follows that (141) is non-negative if and only if
120582119896[(119896 minus 1) cos 120575 + 119896 sin (120575 minus 120575119896)
sin (120575119896)] lt 1 (143)
Notice that 119886119896= 120582119896cos 120575 implies that the following relation
holds
120582119896[(119896 minus 1) cos 120575 + 119896 sin (120575 minus 120575119896)
sin (120575119896)]
= minus119886119896+ 119896120582119896
sin 120575sin (120575119896)
cos(120575119896)
(144)
Finally substitution of (144) into (143) leads to 119896120582119896[sin 120575
sin(120575119896)] cos(120575119896) lt 1 + 119886119896 which proves that (121) holds
when 119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) lt 1 + 119886
119896
Apart from nonnegative waveforms with exactly one zeroat nondegenerate critical point in what follows we will alsoconsider other types of nonnegative waveforms with at leastone zero According to Proposition 9 and Remark 11 thesewaveforms can be described by (66)ndash(68) providing that 0 le|120585| le 120587
According to (35) 119879119896(0) ge 0 implies 1 + 119886
1+ 119886119896ge 0
Consequently 1198861le 0 implies that |119886
1| le 1 + 119886
119896 On the other
hand according to (123) |1198861| = 1 + 119886
119896holds for waveforms
of type (121) The converse is also true 1198861le 0 and |119886
1| =
1 + 119886119896imply 119886
1= minus1 minus 119886
119896 which further from (35) implies
119879119896(0) = 0 Therefore in what follows it is enough to consider
only nonnegativewaveformswhich can be described by (66)ndash(68) and 0 le |120585| le 120587 with coefficients 119886
119896and 119887119896satisfying
119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) ge 1 + 119886
119896
For prescribed coefficients 119886119896and 119887119896 the amplitude 120582
119896=
radic1198862119896+ 1198872119896of 119896th harmonic is also prescribed According to
Remark 15 (see also Remark 16) 120582119896is monotonically
decreasing function of 119909 = cos(120585119896) The value of 119909 can beobtained by solving (90) subject to the constraint cos(120587119896) le119909 le 1 Then 120582
1can be determined from (88) From (106) it
immediately follows that maximal absolute value of 1198861le 0
corresponds to 119902 = 0 which from (104) and (120) furtherimplies that
120575 = 1198961205910minus 120585 (145)
Furthermore 119902 = 0 according to (107) implies that waveformzeros are
1205910=(120575 + 120585)
119896 120591
1015840
0= 1205910minus2120585
119896=(120575 minus 120585)
119896 (146)
Substitution of 1205910= (120575 + 120585)119896 into (66) yields (122) which
proves that (122) holds when 119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) ge
1 + 119886119896
In what follows we prove that (121) also holds when119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) = 1 + 119886
119896 Substitution of 119886
119896=
120582119896cos 120575 into 119896120582
119896[sin 120575 sin(120575119896)] cos(120575119896) = 1 + 119886
119896leads to
120582119896[(119896 minus 1) cos 120575 + 119896 sin (120575 minus 120575119896)
sin (120575119896)] = 1 (147)
As we mentioned earlier relation (142) holds for all wave-forms of type (121) Substituting (142) into (147) we obtain
120582119896[(119896 minus 1) cos 120585 + 119896 sin (120585 minus 120585119896)
sin (120585119896)] = 1 (148)
This expression can be rearranged as
120582119896
119896 sin ((119896 minus 1) 120585119896)sin 120585119896
= 1 minus (119896 minus 1) 120582119896cos 120585 (149)
On the other hand for waveforms of type (122) according to(68) relations (148) and (149) also hold Substitution of 120591
0=
(120575 + 120585)119896 (see (145)) and (67) into (122) leads to
119879119896(120591)
= 120582119896[1 minus cos (120591 minus 120591
0)]
sdot [119896 sin ((119896 minus 1) 120585119896)
sin 120585119896minus 2
119896minus1
sum
119899=1
(119896 minus 119899) cos (119899 (120591 minus 1205910) + 120585)]
(150)
Furthermore substitution of (142) into (145) implies that1205910
= 0 Finally substitution of 1205910
= 0 and (149) into(150) leads to (141) Therefore (141) holds when 119896120582
119896[sin 120575
sin(120575119896)] cos(120575119896) = 1 + 119886119896 which in turn shows that (121)
holds when 119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) = 1 + 119886
119896 This
completes the proof
18 Mathematical Problems in Engineering
52 Nonnegative Waveforms with Maximal Absolute Valueof Coefficient 119886
1for 119896 = 3 Nonnegative waveform of type
(35) for 119896 = 3 is widely used in PA design (eg see [10])In this subsection we illustrate results of Section 51 for thisparticular case The case 119896 = 2 is presented in detail in [12]
Coefficients of fundamental harmonic of nonnegativewaveform of type (35) with 119896 = 3 and maximal absolutevalue of coefficient 119886
1le 0 for prescribed coefficients 119886
3and
1198873(1205823= radic11988623+ 11988723) according to (123) (124) (134) (94) and
(120) are equal to
1198861= minus1 minus 119886
3 119887
1= minus3119887
3 (151)
if 12058223le [(1 minus 2119886
3)3]3
1198861= minus1205821cos(120575
3) 119887
1= minus1205821sin(120575
3) (152)
where 1205821= 3(
3radic1205823minus 1205823) and 120575 = atan 2(119887
3 1198863) if [(1 minus
21198863)3]3
le 1205822
3le 1The line 1205822
3= [(1minus2119886
3)3]3 (see case 119896 = 3
in Figure 11) separates points representing waveforms withcoefficients satisfying (151) from points representing wave-forms with coefficients satisfying (152) Waveforms describedby (151) for 1205822
3lt [(1 minus 2119886
3)3]3 have exactly one zero at
1205910= 0 Waveforms described by (151) and (152) for 1205822
3= [(1 minus
21198863)3]3 also have zero at 120591
0= 0 These waveforms as a rule
have exactly two zeros However there are two exceptionsone related to the maximally flat nonnegative waveform withcoefficients 119886
1= minus98 119886
3= 18 and 119887
1= 1198873= 0 which
has only one zero and the other related to the waveform withcoefficients 119886
1= 0 119886
3= minus1 and 119887
1= 1198873= 0 which has three
zerosWaveforms described by (152) for [(1minus21198863)3]3
lt 1205822
3lt
1 have two zeros Waveforms with 1205823= 1 have only third
harmonic (fundamental harmonic is zero)Plot of contours of maximal absolute value of coefficient
1198861 1198861le 0 for prescribed coefficients 119886
3and 1198873is presented
in Figure 12 According to Remark 25 the waveform withmaximum absolute value of 119886
1le 0 is fully described with
the following coefficients 1198861
= minus2radic3 1198863
= radic39 and1198871= 1198873= 0 This waveform has two zeros at plusmn1205876
Two examples of nonnegative waveforms for 119896 = 3
and maximal absolute value of coefficient 1198861 1198861le 0 with
prescribed coefficients 1198863and 1198873are presented in Figure 13
One waveform corresponds to the case 12058223lt [(1 minus 2119886
3)3]3
(solid line) and the other to the case 12058223gt [(1 minus 2119886
3)3]3
(dashed line)Thewaveform represented by solid line has onezero and its coefficients are 119886
3= minus01 119887
3= 01 119886
1= minus09
and 1198871= minus03 Dashed line corresponds to the waveform
having two zeros with coefficients 1198863= minus01 119887
3= 03 119886
1=
minus08844 and 1198871= minus06460 (case 1205822
3gt [(1 minus 2119886
3)3]3)
6 Nonnegative Cosine Waveforms withat Least One Zero
Nonnegative cosine waveforms have proved to be of impor-tance for waveform modelling in PA design (eg see [10])In this section we consider nonnegative cosine waveforms
1
05
0
minus05
minus1
minus1 minus05 0 05 1
Coefficient a3
Coe
ffici
entb
3
02
04
06
08
10
11
Figure 12 Contours ofmaximal absolute value of coefficient 1198861 1198861le
0 as a function of 1198863and 1198873
2
1
0
0 1 2 3 4
Angle 120591120587
Wav
efor
ms
a3 = minus01 b3 = 01
a3 = minus01 b3 = 03
Figure 13 Nonnegative waveforms for 119896 = 3 and maximal absolutevalue of 119886
1 1198861le 0 with prescribed coefficients 119886
3and 1198873
containing fundamental and 119896th harmonic with at least onezero
Cosine waveform with dc component fundamental and119896th harmonic can be obtained from (35) by setting 119887
1= 119887119896=
0 that is
119879119896(120591) = 1 + 119886
1cos 120591 + 119886
119896cos 119896120591 (153)
In Section 61 we provide general description of non-negative cosine waveforms of type (153) with at least onezero We show that nonnegative cosine waveforms with atleast one zero coincide with nonnegative cosine waveformswith maximal absolute value of coefficient 119886
1for prescribed
coefficient 119886119896 In Section 62 we illustrate results of Section 61
for particular case 119896 = 3
61 Nonnegative Cosine Waveforms with at Least One Zerofor 119896 ge 2 Amplitudes of fundamental and 119896th harmonic
Mathematical Problems in Engineering 19
of cosine waveform of type (153) are 1205821= |1198861| and 120582
119896=
|119886119896| respectively According to (42) for nonnegative cosine
waveforms of type (153) the following relation holds
minus1 le 119886119896le 1 (154)
This explains why 119896th harmonic coefficient 119886119896in Proposi-
tion 26 goes through interval [minus1 1]Waveforms (153) with 119886
1ge 0 can be obtained from
waveforms with 1198861le 0 by shifting by 120587 and therefore with-
out loss of generality we can assume that 1198861le 0
Proposition 26 Each nonnegative cosine waveform of type(153) with 119886
1le 0 and at least one zero can be represented as
119879119896(120591) = [1 minus cos 120591] [1 minus (119896 minus 1) 119886
119896minus 2119886119896
119896minus1
sum
119899=1
(119896 minus 119899) cos 119899120591]
(155)
if minus1 le 119886119896le 1(119896
2
minus 1) or
119879119896(120591) = 119886
119896[1 minus cos (120591 minus 120591
0)] [1 minus cos (120591 + 120591
0)]
sdot [1198880+ 2
119896minus2
sum
119899=1
119888119899cos 119899120591]
(156)
where
119888119899=sin ((119896 minus 119899) 120591
0) cos 120591
0minus (119896 minus 119899) cos ((119896 minus 119899) 120591
0) sin 120591
0
sin31205910
(157)
119886119896=
sin 1205910
119896 sin (1198961205910) cos 120591
0minus cos (119896120591
0) sin 120591
0
(158)
100381610038161003816100381612059101003816100381610038161003816 le
120587
119896 (159)
if 1(1198962 minus 1) le 119886119896le 1
Remark 27 Identity [1minuscos(120591minus1205910)][1minuscos(120591+120591
0)] = [cos 120591
0minus
cos 120591]2 implies that (156) can be rewritten as
119879119896(120591) = 119886
119896[cos 120591
0minus cos 120591]2 [119888
0+ 2
119896minus2
sum
119899=1
119888119899cos 119899120591] (160)
Furthermore substitution of (157) into (160) leads to
119879119896(120591) = 119886
119896[cos 120591
0minus cos 120591]
sdot [(119896 minus 1) sin 119896120591
0
sin 1205910
minus 2
119896minus1
sum
119899=1
sin ((119896 minus 119899) 1205910)
sin 1205910
cos 119899120591]
(161)
Remark 28 All nonnegative cosine waveforms of type (153)with at least one zero and 119886
1le 0 except one of them can be
represented either in form (155) or form (156)This exceptionis maximally flat cosine waveform with 119886
1lt 0 which can be
obtained from (155) for 119886119896= 1(119896
2
minus 1) or from (156) for 1205910=
0 Maximally flat cosine waveform with 1198861lt 0 can also be
obtained from (70) by setting 1205910= 0 Furthermore setting
1205910= 0 in (71) leads to maximally flat cosine waveforms for
119896 le 4 and 1198861lt 0
Remark 29 Nonnegative cosine waveform of type (155) with1198861lt 0 and minus1 lt 119886
119896le 1(119896
2
minus 1) has exactly one zero at120591 = 0 Nonnegative cosine waveform described by (156) with1198861lt 0 and 1(1198962 minus 1) lt 119886
119896lt 1 has two zeros at plusmn120591
0 where
0 lt |1205910| lt 120587119896 For 119886
119896= ∓1 nonnegative cosine waveform
of type (153) reduces to 119879119896(120591) = 1 ∓ cos 119896120591 (clearly these two
waveforms both have 119896 zeros)
Remark 30 Transformation of (155) into an additive formleads to the following relation
1198861= minus1 minus 119886
119896 (162)
where minus1 le 119886119896le 1(119896
2
minus1) Similarly transformation of (156)leads to the following relation
1198861= minus119886119896
119896 sin 1198961205910
sin 1205910
(163)
where 119886119896is given by (158) 1(1198962minus1) le 119886
119896le 1 and |120591
0| le 120587119896
Notice that coefficients of maximally flat cosine waveformnamely 119886
119896= 1(119896
2
minus1) and 1198861= minus1198962
(1198962
minus1) satisfy relation(162) They also satisfy relation (163) for 120591
0= 0
Remark 31 Nonnegative cosine waveforms of type (153) withat least one zero coincide with nonnegative cosine waveformswith maximal absolute value of coefficient 119886
1for prescribed
coefficient 119886119896
In proving that Remark 31 holds notice that expression(155) can be obtained from (121) by setting 119887
119896= 0 Fur-
thermore if 119886119896
ge 0 then 120582119896
= 119886119896 which together
with 119887119896
= 0 and (118) implies 120575 = 0 In this case119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) le 1 + 119886
119896becomes 1198962119886
119896le 1 +
119886119896 On the other hand if 119886
119896lt 0 then 120582
119896= minus119886
119896 which
together with 119887119896= 0 and (118) implies |120575| = 120587 In this case
119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) le 1 + 119886
119896becomes 0 le 1 + 119886
119896
Therefore every nonnegative cosine waveform of type (155)has maximal absolute value of coefficient 119886
1for prescribed
coefficient 119886119896 when minus1 le 119886
119896le 1(119896
2
minus 1)Let us now show that expression (156) can be obtained
from (122) by setting 119887119896= 0 and 119886
119896gt 0 For waveforms
of type (122) according to (118) 119887119896= 0 and 119886
119896gt 0 imply
120575 = 0 and 120582119896= 119886119896 Substitution of 120582
119896= 119886119896and 120575 = 0 into
119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) ge 1+119886
119896leads to 119886
119896ge 1(119896
2
minus1)Furthermore substitution of 120575 = 0 into (145) yields 120591
0= 120585119896
Insertion of 120582119896= 119886119896 120575 = 0 and 120591
0= 120585119896 into (122) leads
to (156) Therefore every nonnegative cosine waveform oftype (156) has maximal absolute value of coefficient 119886
1for
prescribed coefficient 119886119896 when 1(1198962 minus 1) le 119886
119896le 1
Proof of Proposition 26 Let us start with nonnegative cosinewaveform of type (153) with 120582
119896= |119886119896| = 1 According to
Remark 7 120582119896= |119886119896| = 1 implies that 120582
1= |1198861| = 0
Substitution of 119886119896
= minus1 into (155) and using (A2) (seeAppendices) lead to 119879
119896(120591) = 1 minus cos 119896120591 Consequently (155)
holds for 119886119896= minus1 On the other hand substitution of 119886
119896= 1
into (158) yields |1205910| = 120587119896 Furthermore substitution of
119886119896= 1 and 120591
0= 120587119896 (or 120591
0= minus120587119896) into (156) along
with performing all multiplications and using (A2) leads to
20 Mathematical Problems in Engineering
119879119896(120591) = 1 + cos 119896120591 Consequently (156)ndash(158) hold for 119886
119896= 1
and |1205910| = 120587119896
It is easy to see that 120582119896= |119886119896| lt 1 and 119879
119896(1205910) = 0 for some
1205910imply 120582
1= |1198861| = 0 Therefore in what follows we assume
that |119886119896| = 1 and 119886
1lt 0
Cosine waveforms are even functions of 120591 Therefore ifnonnegative cosine waveform has exactly one zero it has to beeither at 0 or at 120587 On the other hand if nonnegative cosinewaveform with 119886
1= 0 has exactly two zeros then these zeros
are placed at plusmn1205910 such that 120591
0is neither 0 nor 120587
In order to prove that (155) holds for minus1 lt 119886119896le 1(1minus119896
2
)let us start by referring to the description (38) of nonnegativewaveformswith at least one zero As wementioned earlier fornonnegative cosine waveformwith exactly one zero (denotedby 1205910) it is either 120591
0= 0 or 120591
0= 120587 Therefore in both cases
sin 1205910= 0 Substitution of sin 120591
0= 0 into (43) together with
1198861
= 0 and 120582119896= |119886119896| lt 1 leads to
1205910= 0 (164)
Clearly 1205910= 0 119887
1= 0 and 119887
119896= 0 according to (44) and (46)
imply 120582119896sin 120585 = 0 Since 120582
119896= |119886119896| it follows that |119886
119896| sin 120585 = 0
also holds which further implies 120582119896= 119886119896= 0 or sin 120585 =
0 In the case when 120582119896= 119886119896= 0 from (164) and (43) we
obtain 1198861= minus1 which further implies that 119879
119896(120591) = 1 minus cos 120591
Consequently (155) holds for 119886119896= 0 In the case when sin 120585 =
0 from (164) and (45) we obtain 119886119896= 120582119896if 120585 = 0 or 119886
119896= minus120582119896
if 120585 = 120587 Relations 119886119896= 120582119896and 120585 = 0 according to (40) imply
that 0 le 119886119896le 1(1 minus 119896
2
) Substitution of 120585 = 0 120582119896= 119886119896 and
(164) into (38) leads to (155) which proves that (155) holdsfor 0 le 119886
119896le 1(1 minus 119896
2
) On the other hand relations 119886119896=
minus120582119896and 120585 = 120587 according to (40) imply that minus1 lt 119886
119896le 0
Substitution of 120585 = 120587 120582119896= minus119886119896 and (164) into (38) also leads
to (155) which proves that (155) also holds for minus1 lt 119886119896le 0
Consequently (155) holds for minus1 lt 119886119896le 1(1 minus 119896
2
)In what follows we first prove that (156)-(157) hold for
1(1 minus 1198962
) lt 119886119896lt 1 For this purpose let us start with non-
negative waveforms with two zeros described by (66) As wementioned before nonnegative cosine waveforms with twozeros have zeros at 120591
0and minus120591
0 such that 120591
0= 0 and 120591
0= 120587
Relations 1198861
lt 0 and 1198871
= 0 according to (84) implycos(1205910minus 120585119896) = 1 and therefore
120585
119896= 1205910 (165)
From 120585119896 = 1205910and 0 lt |120585| lt 120587 it follows that 0 lt |120591
0| lt
120587119896 Insertion of 120585119896 = 1205910into (45) yields 119886
119896= 120582119896 Relations
119886119896= 120582119896and (82) imply that 1(1 minus 1198962) lt 119886
119896lt 1 Substitution
of 120582119896= 119886119896and 120585119896 = 120591
0into (66)ndash(68) leads to (156)ndash(158)
which proves that (156)ndash(158) hold for 1(1 minus 1198962
) lt 119886119896lt 1
and 0 lt |1205910| lt 120587119896
Finally substitution of 119886119896= 1(1 minus 119896
2
) and 1205910= 0 into
(161) leads to
119879119896(120591) =
[1 minus cos 120591](1 minus 1198962)
[119896 (119896 minus 1) minus 2
119896minus1
sum
119899=1
(119896 minus 119899) cos 119899120591]
(166)
Waveform (166) coincides with waveform (155) for 119886119896
=
1(1 minus 1198962
) which in turn proves that (156) holds for 119886119896=
1(1 minus 1198962
) and 1205910= 0 This completes the proof
62 Nonnegative CosineWaveforms with at Least One Zero for119896 = 3 In this subsection we consider nonnegative cosinewaveforms with at least one zero for 119896 = 3 (for case 119896 = 2
see [12])Cosine waveform with fundamental and third harmonic
reads
1198793(120591) = 1 + 119886
1cos 120591 + 119886
3cos 3120591 (167)
For 1198861le 0 and minus1 le 119886
3le 18 according to (155) non-
negative cosine waveform of type (167) with at least one zerocan be expressed as
1198793(120591) = (1 minus cos 120591) [1 minus 2119886
3(1 + 2 cos 120591 + cos 2120591)] (168)
From 1198793(120591 + 120587) = 2 minus 119879
3(120591) it immediately follows that for
1198861ge 0 and minus18 le 119886
3le 1 119879
3(120591) can be expressed as
1198793(120591) = (1 + cos 120591) [1 + 2119886
3(1 minus 2 cos 120591 + cos 2120591)] (169)
For 1198861le 0 and 18 le 119886
3le 1 from (158) it follows that 119886
3=
[8cos31205910]minus1 This relation along with (160) and (157) further
implies that 1198793(120591) can be expressed as
1198793(120591) =
[cos 1205910minus cos 120591]2 [2 cos 120591
0+ cos 120591]
2cos31205910
(170)
providing that |1205910| le 1205873 From 119879
3(120591 + 120587) = 2 minus 119879
3(120591) it
follows that (170) also holds for 1198861ge 0 and minus1 le 119886
3le minus18
providing that 1205910isin [21205873 41205873]
Maximally flat nonnegative cosinewaveformof type (167)with 119886
1lt 0 (minimum at 120591
0= 0) reads 119879
3(120591) = [1 minus
cos 120591]2[1 + (12) cos 120591] Dually maximally flat nonnegativecosine waveform with 119886
1gt 0 (minimum at 120591
0= 120587) reads
1198793(120591) = [1 + cos 120591]2[1 minus (12) cos 120591]In what follows we provide relations between coefficients
1198861and 1198863of nonnegative cosine waveforms of type (167) with
at least one zeroFor 1198861le 0 conversion of (168) into an additive form
immediately leads to the following relation
1198861= minus1 minus 119886
3for minus 1 le 119886
3le1
8 (171)
Conversion of (170) into an additive form leads to 1198861
=
minus31198863(2 cos 2120591
0+ 1) which can be also expressed as 119886
1=
minus31198863(4cos2120591
0minus 1) For 119886
1le 0 relations |120591
0| le 1205873 119886
1=
minus31198863(4cos2120591
0minus 1) and 119886
3= [8cos3120591
0]minus1 lead to
1198861= minus3 [ 3radic119886
3minus 1198863] for 1
8le 1198863le 1 (172)
Similarly for 1198861ge 0 conversion of (169) into an additive form
leads to the following relation
1198861= 1 minus 119886
3for minus
1
8le 1198863le 1 (173)
Mathematical Problems in Engineering 21
1
15
15
05
0
minus05
minus15minus15
minus1
minus1 minus05 0 05 1
Coefficient a1
Coe
ffici
enta
3 a1= minus
3(3radica3minus a3)
a1 =
minus1 minus a3
a1 =
1 minusa3
a1 =3(3radic|a3
| +a3)
Figure 14 Parameter space of cosine waveforms for 119896 = 3
For waveform of type (170) with 1198861ge 0 relations 120591
0isin [21205873
41205873] 1198861= minus3119886
3(4cos2120591
0minus 1) and 119886
3= [8cos3120591
0]minus1 lead to
1198861= 3 [
3radic10038161003816100381610038161198863
1003816100381610038161003816 + 1198863] for minus 1 le 1198863le minus
1
8 (174)
Every cosine waveform of type (167) corresponds to apair of real numbers (119886
1 1198863) and vice versa Points (119886
1 1198863)
in grey area in Figure 14 correspond to nonnegative cosinewaveforms for 119896 = 3 The points at the boundary of grey areacorrespond to nonnegative cosinewaveformswith at least onezero A number of shapes of nonnegative cosine waveformswith 119896 = 3 and at least one zero plotted on interval [minus120587 120587]are also presented in Figure 14 The boundary of grey area inFigure 14 consists of four line segments described by relations(171)ndash(174) The common point of line segments (172) and(173) is cusp point with coordinates 119886
1= 0 and 119886
3= 1
Another cusp point with coordinates 1198861= 0 and 119886
3= minus1
is the common point of line segments (171) and (174) Thecommon point of line segments (171)-(172) has coordinates(minus98 18) and common point of line segments (173)-(174)has coordinates (98 minus18) These points are representedby white circle dots and they correspond to maximallyflat cosine waveforms (eg see [21]) White triangle dotswith coordinates (2radic3 minusradic39) and (minus2radic3radic39) refer tothe nonnegative cosine waveforms with maximum value ofamplitude of fundamental harmonic
7 Four Case Studies of Usage of NonnegativeWaveforms in PA Efficiency Analysis
In this section we provide four case studies of usage ofdescription of nonnegative waveforms with fundamental and119896th harmonic in PA efficiency analysis In first two casestudies to be presented in Section 71 voltage is nonnegativewaveform with fundamental and second harmonic with atleast one zero In remaining two case studies to be consideredin Section 72 voltage waveform contains fundamental andthird harmonic
i
v Load
Vdc
Idc
Lch
vL
iL
Cb
+ +in
Figure 15 Generic PA circuit diagram
Let us consider generic PA circuit diagram as shown inFigure 15 We assume here that voltage and current wave-forms at the transistor output are
V (120579) = 1 + 1198861V cos 120579 + 1198871V sin 120579 + 119886119896V cos 119896120579 + 119887119896V sin 119896120579
119894 (120579) = 1 + 1198861119894cos 120579 +
infin
sum
119899=2
119886119899119894cos 119899120579
(175)
where 120579 stands for 120596119905 Both waveforms are normalized in thesense that dc components of voltage and current are 119881dc =
1 and 119868dc = 1 respectively Under assumption that blockingcapacitor 119862
119887behaves as short-circuit at the fundamental and
higher harmonics current and voltage waveforms at the loadare
V119871(120579) = 119886
1V cos 120579 + 1198871V sin 120579 + 119886119896V cos 119896120579 + 119887119896V sin 119896120579
119894119871(120579) = minus119886
1119894cos 120579 minus
infin
sum
119899=2
119886119899119894cos 119899120579
(176)
In terms of coefficients of voltage and current waveforms theload impedance at fundamental harmonic is 119911
1= minus(119886
1V minus
1198951198871V)1198861119894 whereas load impedance at 119896th harmonic is 119911
119896=
minus(119886119896Vminus119895119887119896V)119886119896119894 All other harmonics are short-circuited (119911
119899=
0 for 119899 = 1 and 119899 = 119896) Time average output power of PA (egsee [10]) with waveform pair (175) at fundamental frequencycan be expressed as
1198751= minus
11988611198941198861V
2 (177)
For normalized waveforms (175) with 119881dc = 1 and 119868dc = 1dc power is 119875dc = 1 Consequently PA efficiency 120578 = 119875
1119875dc
(eg see [10 26]) is equal to
120578 = minus11988611198941198861V
2 (178)
Thus time average output power 1198751of PA with pair of nor-
malized waveform (175) is equal to efficiency (178)Power utilization factor (PUF) is defined [26] as ldquothe
ratio of power delivered in a given situation to the power
22 Mathematical Problems in Engineering
delivered by the same device with the same supply voltagein Class A moderdquo Since the output power in class-A modeis 1198751class-A = max[V(120579)] sdotmax[119894(120579)]8 (eg see [9]) it follows
that power utilization factor PUF = 11987511198751class-A for PA with
pair of normalized waveforms (175) can be expressed as
PUF =8120578
max [V (120579)] sdotmax [119894 (120579)] (179)
71 NonnegativeWaveforms for 119896 = 2 in PAEfficiencyAnalysisIn this subsection we provide two case studies of usage ofdescription of nonnegative waveforms with fundamental andsecond harmonic (119896 = 2) in PA efficiency analysis For moreexamples of usage of descriptions of nonnegative waveformswith fundamental and second harmonic in PA efficiencyanalysis see [12]
Case Study 71 In this case study we consider efficiency of PAfor given second harmonic impedance providing that voltageis nonnegative waveform with fundamental and second har-monic and current is ldquohalf-sinerdquo waveform frequently used inefficiency analysis of classical PA operation (eg see [10])
Standard model of current waveform for classical PAoperation has the form (eg see [10 26])
119894119863(120579) =
119868119863[cos 120579 minus cos(120572
2)] |120579| le
120572
2
0120572
2le |120579| le 120587
(180)
where 120572 is conduction angle and 119868119863
gt 0 Since 119894119863(120579) is
even function it immediately follows that its Fourier seriescontains only dc component and cosine terms
119894119863(120579) = 119868dc +
infin
sum
119899=1
119868119899cos 119899120579 (181)
The dc component of the waveform (180) is
119868dc =119868119863120572
2120587[sinc(120572
2) minus cos(120572
2)] (182)
where sinc119909 = (sin119909)119909 The coefficient of the fundamentalharmonic component reads
1198681=119868119863120572
2120587(1 minus sinc120572) (183)
and the coefficient of 119899th harmonic component can bewrittenin the form
119868119899=119868119863
119899120587[sin ((119899 minus 1) 1205722)
(119899 minus 1)minussin ((119899 + 1) 1205722)
(119899 + 1)] 119899 ge 2
(184)
For ldquohalf-sinerdquo current waveform conduction angle is equalto 120587 (class-B conduction angle) According to (182) thisfurther implies that 119868dc = 119868
119863120587 To obtain normalized form of
waveform (180) we set 119868dc = 1which implies that 119868119863= 120587 Fur-
thermore substitution of 120572 = 120587 and 119868119863= 120587 in (180) leads to
119894 (120579) =
120587 cos 120579 |120579| lt120587
2
0120587
2lt |120579| le 120587
(185)
Similarly substitution of 119868119863= 120587 and 120572 = 120587 into (183) and
(184) leads to the coefficients of waveform (185) Coefficientsof fundamental and second harmonic respectively are
1198861119894=120587
2 119886
2119894=2
3 (186)
On the other hand voltage waveform of type (35) for 119896 =2 reads
V (120579) = 1 + 1198861V cos 120579 + 1198871V sin 120579 + 1198862V cos 2120579 + 1198872V sin 2120579
(187)
This waveform contains only fundamental and second har-monic and therefore all harmonics of order higher thantwo are short-circuited (119911
119899= 0 for 119899 gt 2) For current
voltage pair (185) and (187) load impedance at fundamentalharmonic is 119911
1= minus(119886
1V minus 1198951198871V)1198861119894 whereas load impedance
at second harmonic is 1199112= minus(119886
2Vminus1198951198872V)1198862119894 According to ourassumption the load is passive and therefore Re119911
1 gt 0 and
Re1199112 ge 0 which further imply 119886
11198941198861V lt 0 and 119886
21198941198862V le 0
respectivelyIt is easy to see that problem of findingmaximal efficiency
of PA with current-voltage pair (185) and (187) for prescribedsecond harmonic impedance can be reduced to the problemof finding voltage waveform of type (187) with maximal coef-ficient |119886
1V| for prescribed coefficients of second harmonic(see Section 5)
The following algorithm (analogous to Algorithm 22presented in [12]) provides the procedure for calculation ofmaximal efficiency with current-voltage pair (185) and (187)for prescribed second harmonic impedance The definitionof function atan 2(119910 119909) which appears in the step (iii) of thefollowing algorithm is given by (105)
Algorithm 32 (i) Choose 1199112= 1199032+1198951199092such that |119911
2| le 1|119886
2119894|
(ii) calculate 1198862V minus 1198951198872V = minus119911
21198862119894and 120582
2V = radic11988622V + 1198872
2V(iii) if 2120582
2V le 1 minus 1198862V then calculate 119886
1V = minus1 minus 1198862V and
1198871V = minus2119887
2V else calculate 1205821V = radic81205822V(1 minus 1205822V) 1205790V minus 120585V2 =
(12)atan2(1198872V 1198862V) 1198861V = minus120582
1V cos(1205790V minus 120585V2) and 1198871V =
minus1205821V sin(1205790V minus 120585V2)(iv) calculate efficiency 120578 = minus119886
11198941198861V2
(v) calculate 1199111= minus(119886
1V minus 1198951198871V)1198861119894 and 1199112119899 = 1199112Re119911
1
In this case study coefficients of fundamental and secondharmonic of current waveform are given by (186) Maximalefficiency of PA associated with the waveform pair (185)and (187) as a function of normalized second harmonicimpedance 119911
2119899= 1199112Re119911
1 is presented in Figure 16(a)
As can be seen from Figure 16(a) efficiency of 078 isachieved at the edge of Smith chart where second harmonicimpedance has small resistive part Corresponding PUFcalculated according to (179) is presented in Figure 16(b)Peak efficiency 120578 = 1205874 = 07854 and peak value of PUF = 1
are attained when second harmonic is short-circuited (whichcorresponds to ideal class-B operation [10 26])
For example for second harmonic impedance 1199112= 01 minus
11989505 and currentwaveform (185) fromAlgorithm32 it followsthat 2120582
2V le 1 minus 1198862V Furthermore according to step (iii)
of above algorithm maximal efficiency of PA is attained
Mathematical Problems in Engineering 23
+j2
+j5
+j1
+j2
+j5
minusj2
minusj5
minusj1
minusj2
minusj5
infin0
078
02
05
1 2 5
075 07 065
06 05
075
07
065
06
05
120578 lt 05
(a)
099
095
091
083 075
067
0 59 051
051
099
095
091
083
075 0
67 059
+j5
+j1
+j2
+j5
minusj5
minusj1
minusj2
minusj5
infin
+j2
minusj2
0 02
05
1 2 5
120578 lt 05
(b)
Figure 16 (a) Contours of maximal efficiency of PA and (b) contours of corresponding PUF associated with the waveform pair (185) and(187) as functions of normalized second harmonic impedance 119911
2119899= 1199112Re119911
1
2
3
4
1
0
2 3 410
Wav
efor
ms
Angle 120579120587
i(120579)
(120579)
Figure 17 Waveform pair (185) and (187) that provides maximalefficiency for 119911
2= 01 minus 11989505
with voltage waveform of type (187) with coefficients 1198862V =
minus00667 1198872V = minus03333 119886
1V = minus09333 and 1198871V = 06667 (see
Figure 17) Corresponding efficiency PUF and normalizedsecond harmonic impedance are 120578 = 07330 PUF = 07572and 1199112119899= 01683 minus 11989508415 respectively
On the other hand for second harmonic impedance 1199112=
01 minus 11989508 and current waveform (185) from Algorithm 32 itfollows that 2120582
2V gt 1 minus 1198862V Then according to step (iii) of
above algorithm maximal efficiency is attained with voltagewaveform of type (187) with coefficients 119886
2V = minus00667 1198872V =
minus05333 1198861V = minus09333 and 119887
1V = 10572 (see Figure 18)Efficiency PUF andnormalized secondharmonic impedanceare 120578 = 07330 PUF = 06332 and 119911
2119899= 01683 minus 11989513465
respectively
2
3
4
1
0
2 3 410
Wav
efor
ms
Angle 120579120587
i(120579)
(120579)
Figure 18 Waveform pair (185) and (187) that provides maximalefficiency for 119911
2= 01 minus 11989508
Case Study 72 As another case study let us consider the effi-ciency of PA providing that current waveform is nonnegativecosine waveform up to third harmonic with maximum valueof amplitude of fundamental harmonic [22] (see also [8])
119894 (120579) = 1 +1 + radic5
2cos 120579 + 2radic5
5cos 2120579 + 5 minus radic5
10cos 3120579
(188)
and voltage waveform is nonnegative waveform of type(187) Load impedances at fundamental second and thirdharmonic are 119911
1= minus(119886
1Vminus1198951198871V)1198861119894 1199112 = minus(1198862Vminus1198951198872V)1198862119894 and
1199113= 0 respectively According to our assumption the load
is passive and therefore Re1199111 gt 0 and Re119911
2 ge 0 which
further imply 11988611198941198861V lt 0 and 119886
21198941198862V le 0 respectively
Because current waveform (188) contains only cosineterms and voltage waveform is the same as in previous case
24 Mathematical Problems in Engineering
08
075
07
06
06
065
065
05
05
075 0
7
120578 lt 05
+j2
+j5
+j1
+j2
+j5
minusj2
minusj5
minusj1
minusj2
minusj5
infin0
02
05
1 2 5
(a)
085
08 0
75 07 0
6
05 04
08
075
0405
060
7
120578 lt 05
+j2
+j5
+j1
+j2
+j5
minusj2
minusj5
minusj1
minusj2
minusj5
infin0 02
05
1 2 5
(b)
Figure 19 (a) Contours of maximal efficiency of PA and (b) contours of corresponding PUF associated with the waveform pair (187)-(188)as functions of normalized second harmonic impedance 119911
2119899= 1199112Re119911
1
study the procedure for calculation of maximal efficiency ofPA with waveform pair (187)-(188) is the same as presentedin Algorithm 32 In this case study the coefficients of funda-mental and second harmonic of current waveform are 119886
1119894=
(1 + radic5)2 and 1198862119894= 2radic55 respectively
Maximal efficiency of PA associated with the waveformpair (187)-(188) as a function of normalized secondharmonicimpedance 119911
2119899= 1199112Re119911
1 is presented in Figure 19(a)
Efficiency of 08 is achieved at the edge of Smith chart wheresecond harmonic impedance has small resistive part Thetheoretical upper bound 120578 = (1 + radic5)4 asymp 08090 isattained when second harmonic is short-circuitedWhen thisupper bound is reached both second and third harmonic areshort-circuited which implies that we are dealing with finiteharmonic class-C [6 8] or dually when current and voltageinterchange their roles with finite harmonic inverse class-C[6 9] Corresponding PUF calculated according to (179) ispresented in Figure 19(b) Peak value of PUF asymp 08541 isattained when second harmonic is short-circuited
For example for second harmonic impedance 1199112
=
007 minus 11989504 and current waveform (188) from Algorithm 32 itfollows that 2120582
2V le 1 minus 1198862V Furthermore according to step
(iii) of Algorithm 32 maximal efficiency of PA is attainedwith voltage waveform of type (187) with coefficients 119886
2V =
minus00626 1198872V = minus03578 119886
1V = minus09374 and 1198871V = 07155 (see
Figure 20) Corresponding efficiency PUF and normalizedsecond harmonic impedance are 120578 = 07584 PUF = 06337and 1199112119899= 01208 minus 11989506904 respectively
On the other hand for 1199112= 005 minus 11989507 and current
waveform (187) it follows that 21205822V gt 1minus119886
2VThen accordingto step (iii) of Algorithm 32 themaximal efficiency is attainedwith voltage waveform of type (187) with coefficients 119886
2V =
minus00447 1198872V = minus06261 119886
1V = minus09318 and 1198871V = 10007 (see
Figure 21) Efficiency PUF and normalized second harmonic
2
3
4
1
0
2 3 410
Wav
efor
ms
Angle 120579120587
i(120579)
(120579)
Figure 20 Waveform pair (187)-(188) that provides maximal effi-ciency for 119911
2= 007 minus 11989504
impedance are 120578 = 07538 PUF = 05314 and 1199112119899= 00868minus
11989512156 respectively
72 Nonnegative Waveforms for 119896 = 3 in PA EfficiencyAnalysis In this subsection we provide another two casestudies of usage of description of nonnegative waveforms inPA efficiency analysis this time with fundamental and thirdharmonic (119896 = 3)
Case Study 73 Let us consider current-voltage pair such thatvoltage is nonnegative waveform with fundamental and thirdharmonic
V (120579) = 1 + 1198861V cos 120579 + 1198871V sin 120579 + 1198863V cos 3120579 + 1198873V sin 3120579
(189)
Mathematical Problems in Engineering 25
2
3
4
1
0
2 3 410
Wav
efor
ms
Angle 120579120587
i(120579)
(120579)
Figure 21 Waveform pair (187)-(188) that provides maximal effi-ciency for 119911
2= 005 minus 11989507
and current is nonnegative cosine waveform given by (188)Load impedances at fundamental second and third har-monic are 119911
1= minus(119886
1V minus 1198951198871V)1198861119894 1199112 = 0 and 119911
3= minus(119886
3V minus
1198951198873V)1198863119894 respectively According to our assumption the load
is passive and therefore Re1199111 gt 0 and Re119911
3 ge 0 which
further imply 11988611198941198861V lt 0 and 119886
31198941198863V le 0
In this subsection we consider the problem of findingmaximal efficiency of PA with waveform pair (188)-(189)for given third harmonic impedance As we mentionedearlier problem of finding maximal efficiency of PA withcurrent-voltage pair (188)-(189) for prescribed third har-monic impedance can be reduced to the problem of findingvoltage waveform of type (189) withmaximal coefficient |119886
1V|for prescribed coefficients of third harmonic (see Section 52)
The following algorithm provides the procedure forcalculation of maximal efficiency with current-voltage pair(188)-(189) The definition of function atan 2(119910 119909) whichappears in step (iii) of the following algorithm is given by(105)
Algorithm 33 (i) Choose 1199113= 1199033+1198951199093such that |119911
3| le 1|119886
3119894|
(ii) calculate 1198863V minus 1198951198873V = minus119911
31198863119894and 120582
3V = radic11988623V + 1198872
3V(iii) if 271205822
3V le (1 minus 21198863V)3 then calculate 119886
1V = minus1 minus 1198863V
and 1198871V = minus3119887
3V else calculate 1205821V = 3(3radic1205823V minus 120582
3V) 1205790V minus120585V3 = (13)atan 2(119887
3V 1198863V) 1198861V = minus1205821V cos(1205790V minus 120585V3) and
1198871V = minus120582
1V sin(1205790V minus 120585V3)(iv) calculate efficiency 120578 = minus119886
11198941198861V2
(v) calculate 1199111= minus(119886
1V minus 1198951198871V)1198861119894 and 1199113119899 = 1199113Re119911
1
In this case study coefficients of fundamental and thirdharmonic of current waveform are 119886
1119894= (1 + radic5)2 and
1198863119894= (5 minus radic5)10 respectively For the waveform pair (188)-
(189) maximal efficiency of PA as a function of normalizedthird harmonic impedance 119911
3119899= 1199113Re119911
1 is presented in
Figure 22 Efficiency of 08 is reached when third harmonicimpedance has small resistive part Peak efficiency 120578 = (1 +
radic5)4 asymp 08090 is achieved when third harmonic is short-circuited
+j2
+j5
+j1
+j2
+j5
minusj2
minusj5
minusj1
minusj2
minusj5
infin0
08
08
075
075
07
07
06
05
02
05
1 2
Figure 22 Contours of maximal efficiency of PA associated withthe waveform pair (188)-(189) as a function of normalized thirdharmonic impedance 119911
3119899= 1199113Re119911
1
For the present case study in what follows we show thatpower utilization factor is proportional to efficiency For volt-age waveform of type (189) it is easy to see that V(120579 + 120587) =
2minusV(120579) holdsThis relation along with the fact that waveformV(120579) that provides maximal efficiency has at least one zeroimplies that max[V(120579)] = 2 On the other hand currentwaveform (188) is cosine waveform with positive coefficientsand therefore max[119894(120579)] = 119894(0) = 2 + 4radic5 Consequentlyaccording to (179) the following relation holds
PUFCase study 73 = 2 (5 minus 2radic5) 120578 = 10557120578 (190)
Clearly the ratio PUF120578 is constant and therefore in this casestudy PUF can be easily calculated from the correspondingefficiency Accordingly peak efficiency and peak value ofPUFCase study 73 = 3radic52 minus 52 = 08541 are attained forthe same voltage waveform (when third harmonic is short-circuited)
In the first example current waveform (188) and 1199113=
02 minus 11989505 imply that 2712058223V le (1 minus 2119886
3V)3 Then according
to Algorithm 33 the voltage waveform of type (189) thatprovides maximal efficiency has the following coefficients1198863V = minus00553 119887
3V = minus01382 1198861V = minus09447 and 119887
1V =
04146 (see Figure 23) Efficiency PUF and normalized thirdharmonic impedance are 120578 = 07643 PUF = 08069 and1199113119899= 03425 minus 11989508564 respectivelyIn the second example current waveform (188) and 119911
3=
01 minus 11989511 imply that 2712058223V gt (1 minus 2119886
3V)3 Then according
to Algorithm 33 the voltage waveform of type (189) thatprovides maximal efficiency has the following coefficients1198863V = minus00276 119887
3V = minus03040 1198861V = minus09391 and 119887
1V =
05807 (see Figure 24) Efficiency PUF and normalized thirdharmonic impedance are 120578 = 07598 PUF = 08021 and1199113119899= 01723 minus 11989518952 respectively
26 Mathematical Problems in Engineering
2
3
4
1
0
2 3 410
Wav
efor
ms
Angle 120579120587
i(120579)
(120579)
Figure 23 Waveform pair (188)-(189) that provides maximal effi-ciency for 119911
3= 02 minus 11989505
2
3
4
1
0
2 3 410
Wav
efor
ms
Angle 120579120587
i(120579)
(120579)
Figure 24 Waveform pair (188)-(189) that provides maximal effi-ciency for 119911
3= 01 minus 11989511
Case Study 74 In this case study let us consider current-voltage pair where current is normalized waveform of type(180) with conduction angle 120572 = 115120587 (207∘) and voltageis nonnegative waveform of type (189) Substitution of 120572 =
115120587 and 119868dc = 1 into (182) leads to 119868119863
= 22535Furthermore substitution of 120572 = 115120587 and 119868
119863= 22535 into
(180) leads to
119894 (120579)
=
22535 [cos 120579 minus cos(1151205872
)] |120579| le115120587
2
0115120587
2le |120579| le 120587
(191)
Similarly substitution of 120572 = 115120587 and 119868119863
= 22535 into(183) and (184) for 119899 = 3 yields coefficients of fundamentaland third harmonic of waveform (191)
1198861119894= 14586 119886
3119894= minus01026 (192)
Because current waveform (191) contains only cosineterms and voltage waveform is the same as in previous case
+j2
+j5
+j1
+j2
+j5
minusj2
minusj5
minusj1
minusj2
minusj5
infin0 02
05
1 2
074
076
08
084
082
078
076
074
Figure 25 Contours of maximal efficiency of PA associated withthe waveform pair (189) and (191) as a function of normalized thirdharmonic impedance 119911
3119899= 1199113Re119911
1
study the procedure for calculation of maximal efficiencyof PA with waveform pair (189)ndash(191) is the same as thatpresented in Algorithm 33 In this case study the coefficientsof fundamental and third harmonic of current waveform aregiven by (192)
For the waveform pair (189) and (191) maximal efficiencyof PA as a function of normalized third harmonic impedance1199113119899= 1199113Re119911
1 is presented in Figure 25 Efficiency of 084
is obtained in vicinity of 1199113119899= 23685 (corresponding to 119911
3=
18750) Peak efficiency 120578 asymp 08421 is achieved for voltagewaveform of type (189) with coefficients 119886
1V = minus2radic3 1198863V =
radic39 and 1198871V = 1198873V = 0
In the course of finding power utilization factor noticethat currentwaveformof type (191) attains itsmaximumvaluefor 120579 = 0 Insertion of max[119894(120579)] = 119894(0) = 278 andmax[V(120579)] = 2 for voltage waveform of type (189) into (179)leads to
PUFCase study 74 = 1439120578 (193)
Again the ratio PUF120578 is constant and PUF can be easilycalculated from the corresponding efficiency Accordinglypeak value of PUFCase study 74 asymp 12118 and peak efficiencyare attained for the same voltage waveform
In the first example current waveform (191) and 1199113=
1 minus 11989502 imply that 2712058223V le (1 minus 2119886
3V)3 Then according to
Algorithm 33 voltage waveform of type (189) which providesmaximal efficiency has coefficients 119886
3V = 01026 1198873V =
00205 1198861V = minus11026 and 119887
1V = minus00616 (see Figure 26)Efficiency PUF and normalized third harmonic impedanceare 120578 = 08042 PUF = 11572 and 119911
3119899= 13228 minus 11989502646
respectivelyIn second example current waveform (191) and 119911
3=
15 minus 11989512 imply that 2712058223V gt (1 minus 2119886
3V)3 Then according to
Algorithm 33 voltage waveform of type (189) which providesmaximal efficiency has coefficients 119886
3V = 01540 1198873V =
Mathematical Problems in Engineering 27
2
3
4
1
0
2 3 410
Wav
efor
ms
Angle 120579120587
i(120579)
(120579)
Figure 26 Waveform pair (189) and (191) that provides maximalefficiency for 119911
3= 1 minus 11989502
2
3
4
1
0
2 3 410
Wav
efor
ms
Angle 120579120587
i(120579)
(120579)
Figure 27 Waveform pair (189) and (191) that provides maximalefficiency for 119911
3= 15 minus 11989512
01232 1198861V = minus11255 and 119887
1V = minus02575 (see Figure 27)Efficiency PUF and normalized third harmonic impedanceare 120578 = 08208 PUF = 11812 and 119911
3119899= 19439 minus 11989515552
respectively
8 Conclusion
In this paper we consider a problem of finding general de-scriptions of various classes of nonnegative waveforms withfundamental and 119896th harmonic These classes include non-negative waveforms with at least one zero nonnegative wave-forms with maximal amplitude of fundamental harmonic forprescribed amplitude of 119896th harmonic nonnegative wave-forms withmaximal coefficient of cosine part of fundamentalharmonic for prescribed coefficients of 119896th harmonic andnonnegative cosine waveforms with at least one zero Mainresults are stated in six propositions (Propositions 1 6 918 22 and 26) four corollaries (Corollaries 2ndash5) twentyremarks and three algorithms Four case studies of usageof closed form descriptions of nonnegative waveforms in PAefficiency analysis are considered in detail in Section 7
Appendices
Here we provide a list of finite sums of trigonometric func-tions used in this paper (Appendix A) and brief account ofthe Chebyshev polynomials (Appendix B)
A List of Some Finite Sums ofTrigonometric Functions
Dirichlet kernel (eg see [27]) is as follows
119863119896minus1
(120591) = 1 + 2
119896minus1
sum
119899=1
cos 119899120591 = sin ((2119896 minus 1) 1205912)sin (1205912)
(A1)
Fejer kernel (eg see [27]) can be expressed in the fol-lowing equivalent forms
119865119896minus1
(120591) =1
119896
119896minus1
sum
119899=0
119863119899(120591) = 1 +
2
119896
119896minus1
sum
119899=1
(119896 minus 119899) cos 119899120591
=(1 minus cos 119896120591)119896 (1 minus cos 120591)
(A2)
Lagrangersquos trigonometric identity (eg see [28]) is as fol-lows
1198781(120591) =
119896minus1
sum
119899=1
sin 119899120591 = sin (1198961205912) sin ((119896 minus 1) 1205912)sin (1205912)
(A3)
In what follows we show that the following three trigono-metric identities also hold
2
119896minus1
sum
119899=1
(119896 minus 119899) sin 119899120591 = 119896 sin 120591 minus sin 1198961205911 minus cos 120591
(A4)
119896minus1
sum
119899=1
cos (119896 minus 2119899) 120591 = sin (119896 minus 1) 120591sin 120591
(A5)
119896minus1
sum
119899=1
119899 (119896 minus 119899) cos (119896 minus 2119899) 120591
=sin (119896120591) cos 120591 minus 119896 cos (119896120591) sin 120591
2sin3120591
(A6)
Denote 1198782(120591) = 2sum
119896minus1
119899=1(119896 minus 119899) sin 119899120591 119878
3(120591) = sum
119896minus1
119899=1cos(119896 minus
2119899)120591 and 1198784(120591) = sum
119896minus1
119899=1119899(119896 minus 119899) cos(119896 minus 2119899)120591
Notice that 1198782(120591) = 2119896119878
1(120591) + 119889119863
119896minus1(120591)119889120591 which
immediately leads to (A4)
28 Mathematical Problems in Engineering
Identity (A5) can be obtained as follows
sin (119896 minus 1) 120591sin 120591
=119890119895(119896minus1)120591
minus 119890minus119895(119896minus1)120591
119890119895120591 minus 119890minus119895120591
= 119890119895119896120591
119890minus2119895120591
minus 119890minus2119895(119896minus1)120591
1 minus 119890minus2119895120591
= 119890119895119896120591
119896minus1
sum
119899=1
119890minus2119895119899120591
=
119896minus1
sum
119899=1
119890119895(119896minus2119899)120591
=
119896minus1
sum
119899=1
cos (119896 minus 2119899) 120591
(A7)
From 4119899(119896 minus 119899) = 1198962
minus (119896 minus 2119899)2 it follows that 4119878
4(120591) =
1198962
1198783(120591) + 119889
2
1198783(120591)119889120591
2 which leads to (A6)
B The Chebyshev Polynomials
The Chebyshev polynomials of the first kind 119881119899(119909) can be
defined by the following relation (eg see [29])
119881119899(119909) = cos 119899120591 when 119909 = cos 120591 (B1)
The Chebyshev polynomials of the second kind 119880119899(119909) can be
defined by the following relation (eg see [29])
119880119899(119909) =
sin (119899 + 1) 120591sin 120591
when 119909 = cos 120591 (B2)
The Chebyshev polynomials satisfy the following recur-rence relations (eg see [29])
1198810(119909) = 1 119881
1(119909) = 119909
119881119899+1
(119909) = 2119909119881119899(119909) minus 119881
119899minus1(119909)
1198800(119909) = 1 119880
1(119909) = 2119909
119880119899+1
(119909) = 2119909119880119899(119909) minus 119880
119899minus1(119909)
(B3)
The first few Chebyshev polynomials of the first and secondkind are 119881
2(119909) = 2119909
2
minus 1 1198813(119909) = 4119909
3
minus 3119909 1198814(119909) = 8119909
4
minus
81199092
+ 1 1198802(119909) = 4119909
2
minus 1 1198803(119909) = 8119909
3
minus 4119909 and 1198804(119909) =
161199094
minus 121199092
+ 1
Conflict of Interests
The authors declare that there is no conflict of interestsregarding the publication of this paper
Acknowledgment
This work is supported by the SerbianMinistry of EducationScience and Technology Development as a part of ProjectTP32016
References
[1] V I Arnolrsquod V S Afrajmovich Y S Ilrsquoyashenko and L PShilrsquonikov Dynamical Systems V Bifurcation Theory and Catas-trophe Theory Springer Berlin Germany 1994
[2] E Polak ldquoOn the mathematical foundations of nondifferen-tiable optimization in engineering designrdquo SIAM Review vol29 no 1 pp 21ndash89 1987
[3] N S Fuzik ldquoBiharmonic modes of a tuned RF power amplifierrdquoRadiotehnika vol 25 no 7 pp 62ndash71 1970 (Russian)
[4] P Colantonio F Giannini G Leuzzi and E Limiti ldquoClass Gapproach for low-voltage high-efficiency PA designrdquo Interna-tional Journal of RF and Microwave Computer-Aided Engineer-ing vol 10 no 6 pp 366ndash378 2000
[5] F H Raab ldquoMaximum efficiency and output of class-F poweramplifiersrdquo IEEE Transactions on Microwave Theory and Tech-niques vol 49 no 6 pp 1162ndash1166 2001
[6] FH Raab ldquoClass-E class-C and class-F power amplifiers basedupon a finite number of harmonicsrdquo IEEE Transactions onMicrowaveTheory and Techniques vol 49 no 8 pp 1462ndash14682001
[7] J D Rhodes ldquoOutput universality inmaximum efficiency linearpower amplifiersrdquo International Journal of Circuit Theory andApplications vol 31 no 4 pp 385ndash405 2003
[8] A Juhas and L A Novak ldquoComments on lsquoClass-E class-Cand classF power amplifier based upon a finite number ofharmonicsrsquordquo IEEE Transactions on Microwave Theory and Tech-niques vol 57 no 6 pp 1623ndash1625 2009
[9] M Roberg and Z Popovic ldquoAnalysis of high-efficiency poweramplifiers with arbitrary output harmonic terminationsrdquo IEEETransactions on Microwave Theory and Techniques vol 59 no8 pp 2037ndash2048 2011
[10] A Grebennikov N O Sokal and M J Franco Switchmode RFPower Amplifiers ElsevierAcademic Press San Diego CalifUSA 2nd edition 2012
[11] T Canning P J Tasker and S C Cripps ldquoContinuous modepower amplifier design using harmonic clipping contourstheory and practicerdquo IEEE Transactions on Microwave Theoryand Techniques vol 62 no 1 pp 100ndash110 2014
[12] A Juhas and L A Novak ldquoGeneral description of nonnegativewaveforms up to second harmonic for power amplifier mod-ellingrdquoMathematical Problems in Engineering vol 2014 ArticleID 709762 18 pages 2014
[13] V I Arnolrsquod V V Goryunov O V Lyashko and V A VasilrsquoevDynamical Systems VIIImdashSingularity Theory II ApplicationsSpringer Berlin Germany 1993
[14] D Siersma ldquoProperties of conflict sets in the planerdquo BanachCenter Publications Polish Academy of Sciences vol 50 no 1 pp267ndash276 1999 Proceedings of the Banach Center Symposiumon Geometry and Topology of Caustics (Caustics rsquo98) WarsawPoland
[15] M van Manen The geometry of conflict sets [Dissertation]Universiteit Utrecht UtrechtThe Netherlands 2003 httpigi-tur-archivelibraryuunldissertations2003-0912-123058c4pdf
[16] Y L Sachkov ldquoMaxwell strata and symmetries in the problemofoptimal rolling of a sphere over a planerdquo Sbornik Mathematicsvol 201 no 7-8 pp 1029ndash1051 2010
[17] I A Bogaevsky ldquoPerestroikas of shocks and singularities ofminimum functionsrdquoPhysicaDNonlinear Phenomena vol 173no 1-2 pp 1ndash28 2002
[18] Y L Sachkov ldquoMaxwell strata in the Euler elastic problemrdquoJournal of Dynamical andControl Systems vol 14 no 2 pp 169ndash234 2008
[19] M Siino and T Koike ldquoTopological classification of black holesgeneric Maxwell set and crease set of a horizonrdquo InternationalJournal ofModern Physics D Gravitation Astrophysics Cosmol-ogy vol 20 no 6 pp 1095ndash1122 2011
Mathematical Problems in Engineering 29
[20] F H Raab ldquoClass-F power amplifiers with maximally flat wave-formsrdquo IEEETransactions onMicrowaveTheory and Techniquesvol 45 no 11 pp 2007ndash2012 1997
[21] A Juhas and L A Novak ldquoMaximally flat waveforms with finitenumber of harmonics in class-F power amplifiersrdquo Mathemat-ical Problems in Engineering vol 2013 Article ID 169590 9pages 2013
[22] L Fejer ldquoUber trigonometrische polynomerdquo Journal fur dieReine und Angewandte Mathematik vol 1916 no 146 pp 53ndash82 1916 (German)
[23] S C Cripps ldquoBessel Waives [microwave bytes]rdquo IEEE Micro-wave Magazine vol 10 no 7 pp 30ndash36 117 2009
[24] L N Bryzgalova ldquoSingularities of the maximum of para-metrically dependent functionrdquo Functional Analysis and ItsApplications vol 11 no 1 pp 49ndash51 1977
[25] V I Arnold A A Davydov V A Vassiliev and V MZakalyukin Mathematical Models of Catastrophes Control ofCatastrophic Process Encyclopedia of Life Support Systems(EOLSS) EOLSS Publishers Oxford UK 2006
[26] S C Cripps RF Power Amplifiers for Wireless CommunicationsArtech House Norwood Mass USA 2nd edition 2006
[27] A Zygmund Trigonometric Series vol 1 Cambridge UniversityPress Cambridge UK 2nd edition 1959
[28] A Jeffrey and H DaiHandbook of Mathematical Formulas andIntegrals ElsevierAcademic Press San Diego Calif USA 4thedition 2008
[29] J C Mason and D C Handscomb Chebyshev PolynomialsChapman amp Hall CRC Press Boca Raton Fla USA 2003
Submit your manuscripts athttpwwwhindawicom
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical Problems in Engineering
Hindawi Publishing Corporationhttpwwwhindawicom
Differential EquationsInternational Journal of
Volume 2014
Applied MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Probability and StatisticsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical PhysicsAdvances in
Complex AnalysisJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
OptimizationJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
CombinatoricsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Operations ResearchAdvances in
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Function Spaces
Abstract and Applied AnalysisHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of Mathematics and Mathematical Sciences
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Algebra
Discrete Dynamics in Nature and Society
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Decision SciencesAdvances in
Discrete MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom
Volume 2014 Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Stochastic AnalysisInternational Journal of
14 Mathematical Problems in Engineering
2
1
0
0 1 2 3 4
Angle 120591120587
Wav
efor
ms
1205910 = 01205910 = 12058761205910 = 1205873
Figure 10 Nonnegative waveforms with maximum amplitude offundamental harmonic for 119896 = 3 and 120585 = 1205872
one zero can be represented in form (38) According to (43)(44) and (36) for amplitude 120582
1of fundamental harmonic of
waveforms of type (38) the following relation holds
1205821= radic(1 + 120582
119896cos 120585)2 + 11989621205822
119896sin2120585 (115)
where 120582119896satisfy (40) and |120585| le 120587
Because of (40) in the quest of finding maximal 1205821for
prescribed 120582119896 we have to consider the following two cases
(Case i)120582119896lt [(119896minus1) cos 120585 + 119896 sin(120585minus120585119896) sin(120585119896)]minus1
(Case ii)120582119896= [(119896minus1) cos 120585+119896 sin(120585minus120585119896) sin(120585119896)]minus1
Case i Since 120582119896lt [(119896 minus 1) cos 120585 + 119896 sin(120585 minus 120585119896) sin(120585119896)]minus1
implies 120582119896
= 1 according to (115) it follows that 1205821
= 0Hence 119889120582
1119889120585 = 0 implies
2120582119896sin 120585 [1 minus (1198962 minus 1) 120582
119896cos 120585] = 0 (116)
Therefore 1198891205821119889120585 = 0 if 120582
119896= 0 (Option 1) or sin 120585 = 0
(Option 2) or (1198962 minus 1)120582119896cos 120585 = 1 (Option 3)
Option 1 According to (115) 120582119896= 0 implies 120582
1= 1 (notice
that this implication shows that 1205821does not depend on 120585 and
therefore we can set 120585 to zero value)
Option 2 According to (115) sin 120585 = 0 implies 1205821= 1 +
120582119896cos 120585 which further leads to the conclusion that 120582
1is
maximal for 120585 = 0 For 120585 = 0 120582119896lt [(119896 minus 1) cos 120585 + 119896 sin(120585 minus
120585119896) sin(120585119896)]minus1 becomes 120582119896lt 1(119896
2
minus 1)
Option 3 This option leads to contradiction To show thatnotice that (119896
2
minus 1)120582119896cos 120585 = 1 and 120582
119896lt [(119896 minus
1) cos 120585 + 119896 sin(120585 minus 120585119896) sin(120585119896)]minus1 imply that (119896 minus 1) cos 120585 gtsin(120585minus120585119896) sin(120585119896) Using (A5) (see Appendices) the latestinequality can be rewritten assum119896minus1
119899=1[cos 120585minuscos((119896minus2119899)120585119896)] gt
0 But from |119896 minus 2119899| lt 119896 119899 = 1 (119896 minus 1) and |120585| le 120587
it follows that all summands are not positive and therefore(119896minus1) cos 120585 gt sin(120585minus120585119896) sin(120585119896) does not hold for |120585| le 120587
Consequently Case i implies 120585 = 0 and 120582119896lt 1(119896
2
minus 1)Finally substitution of 120585 = 0 into (38) leads to (108) whichproves that (108) holds for 120582
119896lt 1(119896
2
minus 1)
Case ii Relation120582119896= [(119896minus1) cos 120585+119896 sin(120585minus120585119896) sin(120585119896)]minus1
according to Proposition 9 and Remark 11 implies that cor-responding waveforms can be expressed via (66)ndash(68) for|120585| le 120587 Furthermore 120582
119896= [(119896 minus 1) cos 120585 + 119896 sin(120585 minus
120585119896) sin(120585119896)]minus1 and |120585| le 120587 imply 1(1198962 minus 1) le 120582119896le 1
This proves that (109) holds for 1(1198962 minus 1) le 120582119896le 1
Finally let us prove that (108) holds for 120582119896= 1(119896
2
minus
1) According to (68) (see also Remark 11) this value of 120582119896
corresponds to 120585 = 0 Furthermore substitution of 120582119896=
1(1198962
minus 1) and 120585 = 0 into (109) leads to (70) which can berewritten as
119879119896(120591) =
[1 minus cos (120591 minus 1205910)]
(1 minus 1198962)
sdot [119896 (119896 minus 1) minus 2
119896minus1
sum
119899=1
(119896 minus 119899) cos (119899 (120591 minus 1205910))]
(117)
Waveform (117) coincides with waveform (108) for 120582119896
=
1(1 minus 1198962
) Consequently (108) holds for 120582119896= 1(1 minus 119896
2
)which completes the proof
5 Nonnegative Waveforms with MaximalAbsolute Value of the Coefficient of CosineTerm of Fundamental Harmonic
In this sectionwe consider general description of nonnegativewaveforms of type (35) with maximal absolute value ofcoefficient 119886
1for prescribed coefficients of 119896th harmonicThis
type of waveform is of particular interest in PA efficiencyanalysis In a number of cases of practical interest eithercurrent or voltage waveform is prescribed In such casesthe problem of finding maximal efficiency of PA can bereduced to the problem of finding nonnegative waveformwith maximal coefficient 119886
1for prescribed coefficients of 119896th
harmonic (see also Section 7)In Section 51 we provide general description of nonneg-
ative waveforms of type (35) with maximal absolute value ofcoefficient 119886
1for prescribed coefficients of 119896th harmonic In
Section 52 we illustrate results of Section 51 for particularcase 119896 = 3
51 Nonnegative Waveforms with Maximal Absolute Value ofCoefficient 119886
1for 119896 ge 2 Waveforms 119879
119896(120591) of type (35) with
1198861ge 0 can be derived from those with 119886
1le 0 by shifting
by 120587 and therefore we can assume without loss of generalitythat 119886
1le 0 Notice that if 119896 is even then shifting 119879
119896(120591) by
120587 produces the same result as replacement of 1198861with minus119886
1
(119886119896remains the same) On the other hand if 119896 is odd then
shifting 119879119896(120591) by 120587 produces the same result as replacement
of 1198861with minus119886
1and 119886119896with minus119886
119896
According to (37) coefficients of 119896th harmonic can beexpressed as
119886119896= 120582119896cos 120575 119887
119896= 120582119896sin 120575 (118)
Mathematical Problems in Engineering 15
where
|120575| le 120587 (119)
Conversely for prescribed coefficients 119886119896and 119887
119896 120575 can be
determined as
120575 = atan 2 (119887119896 119886119896) (120)
where definition of function atan 2(119910 119909) is given by (105)The main result of this section is stated in the following
proposition
Proposition 22 Every nonnegative waveform of type (35)withmaximal absolute value of coefficient 119886
1le 0 for prescribed
coefficients 119886119896and 119887119896of 119896th harmonic can be represented as
119879119896(120591)
= [1 minus cos 120591]
sdot [1 minus (119896 minus 1) 119886119896minus 2
119896minus1
sum
119899=1
(119896 minus 119899) (119886119896cos 119899120591 + 119887
119896sin 119899120591)]
(121)
if 119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) le 1+119886
119896 where 120575 = atan 2(bk
119886119896) or
119879119896(120591) = 120582
119896[1 minus cos(120591 minus (120575 + 120585)
119896)]
sdot [1 minus cos(120591 minus (120575 minus 120585)
119896)]
sdot [1198880+ 2
119896minus2
sum
119899=1
119888119899cos 119899(120591 minus 120575
119896)]
(122)
if 119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) ge 1 + 119886
119896 where 119888
119899 119899 = 0
119896minus2 and 120582119896= radic1198862119896+ 1198872119896are related to 120585 via relations (67) and
(68) respectively and |120585| le 120587
Remark 23 Expression (121) can be obtained from (38) bysetting 120591
0= 0 and 120585 = minus120575 and then replacing 120582
119896cos 120575 with
119886119896(see (118)) and 120582
119896cos(119899120591 minus 120575) with 119886
119896cos 119899120591 + 119887
119896sin 119899120591
(see also (118)) Furthermore insertion of 1205910= 0 and 120585 =
minus120575 into (43)ndash(46) leads to the following relations betweenfundamental and 119896th harmonic coefficients of waveform(121)
1198861= minus (1 + 119886
119896) 119887
1= minus119896119887
119896 (123)
On the other hand expression (122) can be obtained from(66) by replacing 120591
0minus120585119896with 120575119896 Therefore substitution of
1205910minus 120585119896 = 120575119896 in (84) leads to
1198861= minus1205821cos(120575
119896) 119887
1= minus1205821sin(120575
119896) (124)
where 1205821is given by (85)
The fundamental harmonic coefficients 1198861and 1198871of wave-
form of type (35) with maximal absolute value of coefficient1198861le 0 satisfy both relations (123) and (124) if 119886
119896and 119887119896satisfy
1 + 119886119896= 119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) For such waveforms
relations 1205910= 0 and 120585 = minus120575 also hold
Remark 24 Amplitude of 119896th harmonic of nonnegativewaveform of type (35) with maximal absolute value of coeffi-cient 119886
1le 0 and coefficients 119886
119896 119887119896satisfying 1 + 119886
119896=
119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) is
120582119896=
sin (120575119896)119896 sin 120575 cos (120575119896) minus cos 120575 sin (120575119896)
(125)
To show that it is sufficient to substitute 119886119896= 120582119896cos 120575 (see
(118)) into 1 + 119886119896= 119896120582119896[sin 120575 sin(120575119896)] cos(120575119896)
Introducing new variable
119910 = cos(120575119896) (126)
and using the Chebyshev polynomials (eg see Appendices)relations 119886
119896= 120582119896cos 120575 and (125) can be rewritten as
119886119896= 120582119896119881119896(119910) (127)
120582119896=
1
119896119910119880119896minus1
(119910) minus 119881119896(119910)
(128)
where119881119896(119910) and119880
119896(119910) denote the Chebyshev polynomials of
the first and second kind respectively Substitution of (128)into (127) leads to
119886119896119896119910119880119896minus1
(119910) minus (1 + 119886119896) 119881119896(119910) = 0 (129)
which is polynomial equation of 119896th degree in terms of var-iable 119910 From |120575| le 120587 and (126) it follows that
cos(120587119896) le 119910 le 1 (130)
In what follows we show that 119886119896is monotonically increas-
ing function of 119910 on the interval (130) From 120585 = minus120575 (seeRemark 23) and (81) it follows that 120582minus1
119896= (119896 minus 1) cos 120575 +
119896sum119896minus1
119899=1cos((119896 minus 2119899)120575119896) ge 1 and therefore 119886
119896= 120582119896cos 120575 can
be rewritten as
119886119896=
cos 120575(119896 minus 1) cos 120575 + 119896sum119896minus1
119899=1cos ((119896 minus 2119899) 120575119896)
(131)
Obviously 119886119896is even function of 120575 and all cosines in (131)
are monotonically decreasing functions of |120575| on the interval|120575| le 120587 It is easy to show that cos((119896 minus 2119899)120575119896) 119899 =
1 (119896 minus 1) decreases slower than cos 120575 when |120575| increasesThis implies that denominator of the right hand side of(131) decreases slower than numerator Since denominator ispositive for |120575| le 120587 it further implies that 119886
119896is decreasing
function of |120575| on interval |120575| le 120587 Consequently 119886119896is
monotonically increasing function of 119910 on the interval (130)Thus we have shown that 119886
119896is monotonically increasing
function of 119910 on the interval (130) and therefore (129) hasonly one solution that satisfies (130) According to (128) thevalue of 119910 obtained from (129) and (130) either analyticallyor numerically leads to amplitude 120582
119896of 119896th harmonic
16 Mathematical Problems in Engineering
1
05
0
minus05
minus1
minus1 minus05 0 05 1
Coefficient ak
Coe
ffici
entb
k
radica2k+ b2
kle 1
k = 2k = 3k = 4
Figure 11 Plot of (119886119896 119887119896) satisfying 1 + 119886
119896= 119896120582
119896[sin 120575 sin(120575
119896)] cos(120575119896) for 119896 le 4
By solving (129) and (130) for 119896 le 4 we obtain
119910 = radic1 + 1198862
2 (1 minus 1198862) minus1 le 119886
2le1
3
119910 = radic3
4 (1 minus 21198863) minus1 le 119886
3le1
8
119910 =radicradic2 minus 4119886
4+ 1011988624minus 2 (1 minus 119886
4)
4 (1 minus 31198864)
minus1 le 1198864le
1
15
(132)
Insertion of (132) into (128) leads to the following explicitexpressions for the amplitude 120582
119896 119896 le 4
1205822=1
2(1 minus 119886
2) minus1 le 119886
2le1
3 (133)
1205822
3= [
1
3(1 minus 2119886
3)]
3
minus1 le 1198863le1
8 (134)
1205824=1
4(minus1 minus 119886
4+ radic2 minus 4119886
4+ 1011988624) minus1 le 119886
4le
1
15
(135)
Relations (133)ndash(135) define closed lines (see Figure 11) whichseparate points representing waveforms of type (121) frompoints representing waveforms of type (122) For given 119896points inside the corresponding curve refer to nonnegativewaveforms of type (121) whereas points outside curve (andradic1198862119896+ 1198872119896le 1) correspond to nonnegative waveforms of type
(122) Points on the respective curve correspond to the wave-forms which can be expressed in both forms (121) and (122)
Remark 25 Themaximum absolute value of coefficient 1198861of
nonnegative waveform of type (35) is
100381610038161003816100381611988611003816100381610038161003816max =
1
cos (120587 (2119896)) (136)
This maximum value is attained for |120585| = 1205872 and 120575 = 0
(see (124)) Notice that |1198861|max is equal to the maximum value
1205821max of amplitude of fundamental harmonic (see (113))
Coefficients of waveform with maximum absolute value ofcoefficient 119886
1 1198861lt 0 are
1198861= minus
1
cos (120587 (2119896)) 119886
119896=1
119896tan( 120587
(2119896))
1198871= 119887119896= 0
(137)
Waveformdescribed by (137) is cosinewaveformhaving zerosat 120587(2119896) and minus120587(2119896)
In the course of proving (136) notice first that |1198861|max le
1205821max holds According to (123) and (124) maximum of |119886
1|
occurs for 119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) ge 1 + 119886
119896 From (124)
it immediately follows that maximum value of |1198861| is attained
if and only if 1205821= 1205821max and 120575 = 0 which because of
120575119896 = 1205910minus120585119896 further implies 120591
0= 120585119896 Sincemaximumvalue
of 1205821is attained for |120585| = 1205872 it follows that corresponding
waveform has zeros at 120587(2119896) and minus120587(2119896)
Proof of Proposition 22 As it was mentioned earlier in thissection we can assume without loss of generality that 119886
1le 0
We consider waveforms119879119896(120591) of type (35) such that119879
119896(120591) ge 0
and119879119896(120591) = 0 for some 120591
0 Fromassumption that nonnegative
waveform 119879119896(120591) of type (35) has at least one zero it follows
that it can be expressed in form (38)Let us also assume that 120591
0is position of nondegenerate
critical point Therefore 119879119896(1205910) = 0 implies 1198791015840
119896(1205910) = 0 and
11987910158401015840
119896(1205910) gt 0 According to (55) second derivative of 119879
119896(120591) at
1205910can be expressed as 11987910158401015840
119896(1205910) = 1 minus 120582
119896(1198962
minus 1) cos 120585 Since11987910158401015840
119896(1205910) gt 0 it follows immediately that
1 minus 120582119896(1198962
minus 1) cos 120585 gt 0 (138)
Let us further assume that 119879119896(120591) has exactly one zeroThe
problem of finding maximum absolute value of 1198861is con-
nected to the problem of finding maximum of the minimumfunction (see Section 21) If waveforms possess unique globalminimum at nondegenerate critical point then correspond-ing minimum function is a smooth function of parameters[13] Consequently assumption that 119879
119896(120591) has exactly one
zero at nondegenerate critical point leads to the conclusionthat coefficient 119886
1is differentiable function of 120591
0 First
derivative of 1198861(see (43)) with respect to 120591
0 taking into
account that 1205971205851205971205910= 119896 (see (50)) can be expressed in the
following factorized form
1205971198861
1205971205910
= sin 1205910[1 minus 120582
119896(1198962
minus 1) cos 120585] (139)
Mathematical Problems in Engineering 17
From (138) and (139) it is clear that 12059711988611205971205910= 0 if and only if
sin 1205910= 0 According toRemark 12 assumption that119879
119896(120591)has
exactly one zero implies 120582119896lt 1 From (51) (48) and 120582
119896lt 1
it follows that 1198861cos 1205910+ 1198871sin 1205910lt 0 which together with
sin 1205910= 0 implies that 119886
1cos 1205910lt 0 Assumption 119886
1le 0
together with relations 1198861cos 1205910lt 0 and sin 120591
0= 0 further
implies 1198861
= 0 and
1205910= 0 (140)
Insertion of 1205910= 0 into (38) leads to
119879119896(120591)
= [1 minus cos 120591]
sdot [1 minus (119896 minus 1) 120582119896cos 120585 minus 2120582
119896
119896minus1
sum
119899=1
(119896 minus 119899) cos (119899120591 + 120585)]
(141)
Substitution of 1205910= 0 into (45) and (46) yields 119886
119896= 120582119896cos 120585
and 119887119896
= minus120582119896sin 120585 respectively Replacing 120582
119896cos 120585 with
119886119896and 120582
119896cos(119899120591 + 120585) with (119886
119896cos 119899120591 + 119887
119896sin 119899120591) in (141)
immediately leads to (121)Furthermore 119886
119896= 120582119896cos 120585 119887
119896= minus120582
119896sin 120585 and (118)
imply that
120575 = minus120585 (142)
According to (38)ndash(40) and (142) it follows that (141) is non-negative if and only if
120582119896[(119896 minus 1) cos 120575 + 119896 sin (120575 minus 120575119896)
sin (120575119896)] lt 1 (143)
Notice that 119886119896= 120582119896cos 120575 implies that the following relation
holds
120582119896[(119896 minus 1) cos 120575 + 119896 sin (120575 minus 120575119896)
sin (120575119896)]
= minus119886119896+ 119896120582119896
sin 120575sin (120575119896)
cos(120575119896)
(144)
Finally substitution of (144) into (143) leads to 119896120582119896[sin 120575
sin(120575119896)] cos(120575119896) lt 1 + 119886119896 which proves that (121) holds
when 119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) lt 1 + 119886
119896
Apart from nonnegative waveforms with exactly one zeroat nondegenerate critical point in what follows we will alsoconsider other types of nonnegative waveforms with at leastone zero According to Proposition 9 and Remark 11 thesewaveforms can be described by (66)ndash(68) providing that 0 le|120585| le 120587
According to (35) 119879119896(0) ge 0 implies 1 + 119886
1+ 119886119896ge 0
Consequently 1198861le 0 implies that |119886
1| le 1 + 119886
119896 On the other
hand according to (123) |1198861| = 1 + 119886
119896holds for waveforms
of type (121) The converse is also true 1198861le 0 and |119886
1| =
1 + 119886119896imply 119886
1= minus1 minus 119886
119896 which further from (35) implies
119879119896(0) = 0 Therefore in what follows it is enough to consider
only nonnegativewaveformswhich can be described by (66)ndash(68) and 0 le |120585| le 120587 with coefficients 119886
119896and 119887119896satisfying
119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) ge 1 + 119886
119896
For prescribed coefficients 119886119896and 119887119896 the amplitude 120582
119896=
radic1198862119896+ 1198872119896of 119896th harmonic is also prescribed According to
Remark 15 (see also Remark 16) 120582119896is monotonically
decreasing function of 119909 = cos(120585119896) The value of 119909 can beobtained by solving (90) subject to the constraint cos(120587119896) le119909 le 1 Then 120582
1can be determined from (88) From (106) it
immediately follows that maximal absolute value of 1198861le 0
corresponds to 119902 = 0 which from (104) and (120) furtherimplies that
120575 = 1198961205910minus 120585 (145)
Furthermore 119902 = 0 according to (107) implies that waveformzeros are
1205910=(120575 + 120585)
119896 120591
1015840
0= 1205910minus2120585
119896=(120575 minus 120585)
119896 (146)
Substitution of 1205910= (120575 + 120585)119896 into (66) yields (122) which
proves that (122) holds when 119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) ge
1 + 119886119896
In what follows we prove that (121) also holds when119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) = 1 + 119886
119896 Substitution of 119886
119896=
120582119896cos 120575 into 119896120582
119896[sin 120575 sin(120575119896)] cos(120575119896) = 1 + 119886
119896leads to
120582119896[(119896 minus 1) cos 120575 + 119896 sin (120575 minus 120575119896)
sin (120575119896)] = 1 (147)
As we mentioned earlier relation (142) holds for all wave-forms of type (121) Substituting (142) into (147) we obtain
120582119896[(119896 minus 1) cos 120585 + 119896 sin (120585 minus 120585119896)
sin (120585119896)] = 1 (148)
This expression can be rearranged as
120582119896
119896 sin ((119896 minus 1) 120585119896)sin 120585119896
= 1 minus (119896 minus 1) 120582119896cos 120585 (149)
On the other hand for waveforms of type (122) according to(68) relations (148) and (149) also hold Substitution of 120591
0=
(120575 + 120585)119896 (see (145)) and (67) into (122) leads to
119879119896(120591)
= 120582119896[1 minus cos (120591 minus 120591
0)]
sdot [119896 sin ((119896 minus 1) 120585119896)
sin 120585119896minus 2
119896minus1
sum
119899=1
(119896 minus 119899) cos (119899 (120591 minus 1205910) + 120585)]
(150)
Furthermore substitution of (142) into (145) implies that1205910
= 0 Finally substitution of 1205910
= 0 and (149) into(150) leads to (141) Therefore (141) holds when 119896120582
119896[sin 120575
sin(120575119896)] cos(120575119896) = 1 + 119886119896 which in turn shows that (121)
holds when 119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) = 1 + 119886
119896 This
completes the proof
18 Mathematical Problems in Engineering
52 Nonnegative Waveforms with Maximal Absolute Valueof Coefficient 119886
1for 119896 = 3 Nonnegative waveform of type
(35) for 119896 = 3 is widely used in PA design (eg see [10])In this subsection we illustrate results of Section 51 for thisparticular case The case 119896 = 2 is presented in detail in [12]
Coefficients of fundamental harmonic of nonnegativewaveform of type (35) with 119896 = 3 and maximal absolutevalue of coefficient 119886
1le 0 for prescribed coefficients 119886
3and
1198873(1205823= radic11988623+ 11988723) according to (123) (124) (134) (94) and
(120) are equal to
1198861= minus1 minus 119886
3 119887
1= minus3119887
3 (151)
if 12058223le [(1 minus 2119886
3)3]3
1198861= minus1205821cos(120575
3) 119887
1= minus1205821sin(120575
3) (152)
where 1205821= 3(
3radic1205823minus 1205823) and 120575 = atan 2(119887
3 1198863) if [(1 minus
21198863)3]3
le 1205822
3le 1The line 1205822
3= [(1minus2119886
3)3]3 (see case 119896 = 3
in Figure 11) separates points representing waveforms withcoefficients satisfying (151) from points representing wave-forms with coefficients satisfying (152) Waveforms describedby (151) for 1205822
3lt [(1 minus 2119886
3)3]3 have exactly one zero at
1205910= 0 Waveforms described by (151) and (152) for 1205822
3= [(1 minus
21198863)3]3 also have zero at 120591
0= 0 These waveforms as a rule
have exactly two zeros However there are two exceptionsone related to the maximally flat nonnegative waveform withcoefficients 119886
1= minus98 119886
3= 18 and 119887
1= 1198873= 0 which
has only one zero and the other related to the waveform withcoefficients 119886
1= 0 119886
3= minus1 and 119887
1= 1198873= 0 which has three
zerosWaveforms described by (152) for [(1minus21198863)3]3
lt 1205822
3lt
1 have two zeros Waveforms with 1205823= 1 have only third
harmonic (fundamental harmonic is zero)Plot of contours of maximal absolute value of coefficient
1198861 1198861le 0 for prescribed coefficients 119886
3and 1198873is presented
in Figure 12 According to Remark 25 the waveform withmaximum absolute value of 119886
1le 0 is fully described with
the following coefficients 1198861
= minus2radic3 1198863
= radic39 and1198871= 1198873= 0 This waveform has two zeros at plusmn1205876
Two examples of nonnegative waveforms for 119896 = 3
and maximal absolute value of coefficient 1198861 1198861le 0 with
prescribed coefficients 1198863and 1198873are presented in Figure 13
One waveform corresponds to the case 12058223lt [(1 minus 2119886
3)3]3
(solid line) and the other to the case 12058223gt [(1 minus 2119886
3)3]3
(dashed line)Thewaveform represented by solid line has onezero and its coefficients are 119886
3= minus01 119887
3= 01 119886
1= minus09
and 1198871= minus03 Dashed line corresponds to the waveform
having two zeros with coefficients 1198863= minus01 119887
3= 03 119886
1=
minus08844 and 1198871= minus06460 (case 1205822
3gt [(1 minus 2119886
3)3]3)
6 Nonnegative Cosine Waveforms withat Least One Zero
Nonnegative cosine waveforms have proved to be of impor-tance for waveform modelling in PA design (eg see [10])In this section we consider nonnegative cosine waveforms
1
05
0
minus05
minus1
minus1 minus05 0 05 1
Coefficient a3
Coe
ffici
entb
3
02
04
06
08
10
11
Figure 12 Contours ofmaximal absolute value of coefficient 1198861 1198861le
0 as a function of 1198863and 1198873
2
1
0
0 1 2 3 4
Angle 120591120587
Wav
efor
ms
a3 = minus01 b3 = 01
a3 = minus01 b3 = 03
Figure 13 Nonnegative waveforms for 119896 = 3 and maximal absolutevalue of 119886
1 1198861le 0 with prescribed coefficients 119886
3and 1198873
containing fundamental and 119896th harmonic with at least onezero
Cosine waveform with dc component fundamental and119896th harmonic can be obtained from (35) by setting 119887
1= 119887119896=
0 that is
119879119896(120591) = 1 + 119886
1cos 120591 + 119886
119896cos 119896120591 (153)
In Section 61 we provide general description of non-negative cosine waveforms of type (153) with at least onezero We show that nonnegative cosine waveforms with atleast one zero coincide with nonnegative cosine waveformswith maximal absolute value of coefficient 119886
1for prescribed
coefficient 119886119896 In Section 62 we illustrate results of Section 61
for particular case 119896 = 3
61 Nonnegative Cosine Waveforms with at Least One Zerofor 119896 ge 2 Amplitudes of fundamental and 119896th harmonic
Mathematical Problems in Engineering 19
of cosine waveform of type (153) are 1205821= |1198861| and 120582
119896=
|119886119896| respectively According to (42) for nonnegative cosine
waveforms of type (153) the following relation holds
minus1 le 119886119896le 1 (154)
This explains why 119896th harmonic coefficient 119886119896in Proposi-
tion 26 goes through interval [minus1 1]Waveforms (153) with 119886
1ge 0 can be obtained from
waveforms with 1198861le 0 by shifting by 120587 and therefore with-
out loss of generality we can assume that 1198861le 0
Proposition 26 Each nonnegative cosine waveform of type(153) with 119886
1le 0 and at least one zero can be represented as
119879119896(120591) = [1 minus cos 120591] [1 minus (119896 minus 1) 119886
119896minus 2119886119896
119896minus1
sum
119899=1
(119896 minus 119899) cos 119899120591]
(155)
if minus1 le 119886119896le 1(119896
2
minus 1) or
119879119896(120591) = 119886
119896[1 minus cos (120591 minus 120591
0)] [1 minus cos (120591 + 120591
0)]
sdot [1198880+ 2
119896minus2
sum
119899=1
119888119899cos 119899120591]
(156)
where
119888119899=sin ((119896 minus 119899) 120591
0) cos 120591
0minus (119896 minus 119899) cos ((119896 minus 119899) 120591
0) sin 120591
0
sin31205910
(157)
119886119896=
sin 1205910
119896 sin (1198961205910) cos 120591
0minus cos (119896120591
0) sin 120591
0
(158)
100381610038161003816100381612059101003816100381610038161003816 le
120587
119896 (159)
if 1(1198962 minus 1) le 119886119896le 1
Remark 27 Identity [1minuscos(120591minus1205910)][1minuscos(120591+120591
0)] = [cos 120591
0minus
cos 120591]2 implies that (156) can be rewritten as
119879119896(120591) = 119886
119896[cos 120591
0minus cos 120591]2 [119888
0+ 2
119896minus2
sum
119899=1
119888119899cos 119899120591] (160)
Furthermore substitution of (157) into (160) leads to
119879119896(120591) = 119886
119896[cos 120591
0minus cos 120591]
sdot [(119896 minus 1) sin 119896120591
0
sin 1205910
minus 2
119896minus1
sum
119899=1
sin ((119896 minus 119899) 1205910)
sin 1205910
cos 119899120591]
(161)
Remark 28 All nonnegative cosine waveforms of type (153)with at least one zero and 119886
1le 0 except one of them can be
represented either in form (155) or form (156)This exceptionis maximally flat cosine waveform with 119886
1lt 0 which can be
obtained from (155) for 119886119896= 1(119896
2
minus 1) or from (156) for 1205910=
0 Maximally flat cosine waveform with 1198861lt 0 can also be
obtained from (70) by setting 1205910= 0 Furthermore setting
1205910= 0 in (71) leads to maximally flat cosine waveforms for
119896 le 4 and 1198861lt 0
Remark 29 Nonnegative cosine waveform of type (155) with1198861lt 0 and minus1 lt 119886
119896le 1(119896
2
minus 1) has exactly one zero at120591 = 0 Nonnegative cosine waveform described by (156) with1198861lt 0 and 1(1198962 minus 1) lt 119886
119896lt 1 has two zeros at plusmn120591
0 where
0 lt |1205910| lt 120587119896 For 119886
119896= ∓1 nonnegative cosine waveform
of type (153) reduces to 119879119896(120591) = 1 ∓ cos 119896120591 (clearly these two
waveforms both have 119896 zeros)
Remark 30 Transformation of (155) into an additive formleads to the following relation
1198861= minus1 minus 119886
119896 (162)
where minus1 le 119886119896le 1(119896
2
minus1) Similarly transformation of (156)leads to the following relation
1198861= minus119886119896
119896 sin 1198961205910
sin 1205910
(163)
where 119886119896is given by (158) 1(1198962minus1) le 119886
119896le 1 and |120591
0| le 120587119896
Notice that coefficients of maximally flat cosine waveformnamely 119886
119896= 1(119896
2
minus1) and 1198861= minus1198962
(1198962
minus1) satisfy relation(162) They also satisfy relation (163) for 120591
0= 0
Remark 31 Nonnegative cosine waveforms of type (153) withat least one zero coincide with nonnegative cosine waveformswith maximal absolute value of coefficient 119886
1for prescribed
coefficient 119886119896
In proving that Remark 31 holds notice that expression(155) can be obtained from (121) by setting 119887
119896= 0 Fur-
thermore if 119886119896
ge 0 then 120582119896
= 119886119896 which together
with 119887119896
= 0 and (118) implies 120575 = 0 In this case119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) le 1 + 119886
119896becomes 1198962119886
119896le 1 +
119886119896 On the other hand if 119886
119896lt 0 then 120582
119896= minus119886
119896 which
together with 119887119896= 0 and (118) implies |120575| = 120587 In this case
119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) le 1 + 119886
119896becomes 0 le 1 + 119886
119896
Therefore every nonnegative cosine waveform of type (155)has maximal absolute value of coefficient 119886
1for prescribed
coefficient 119886119896 when minus1 le 119886
119896le 1(119896
2
minus 1)Let us now show that expression (156) can be obtained
from (122) by setting 119887119896= 0 and 119886
119896gt 0 For waveforms
of type (122) according to (118) 119887119896= 0 and 119886
119896gt 0 imply
120575 = 0 and 120582119896= 119886119896 Substitution of 120582
119896= 119886119896and 120575 = 0 into
119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) ge 1+119886
119896leads to 119886
119896ge 1(119896
2
minus1)Furthermore substitution of 120575 = 0 into (145) yields 120591
0= 120585119896
Insertion of 120582119896= 119886119896 120575 = 0 and 120591
0= 120585119896 into (122) leads
to (156) Therefore every nonnegative cosine waveform oftype (156) has maximal absolute value of coefficient 119886
1for
prescribed coefficient 119886119896 when 1(1198962 minus 1) le 119886
119896le 1
Proof of Proposition 26 Let us start with nonnegative cosinewaveform of type (153) with 120582
119896= |119886119896| = 1 According to
Remark 7 120582119896= |119886119896| = 1 implies that 120582
1= |1198861| = 0
Substitution of 119886119896
= minus1 into (155) and using (A2) (seeAppendices) lead to 119879
119896(120591) = 1 minus cos 119896120591 Consequently (155)
holds for 119886119896= minus1 On the other hand substitution of 119886
119896= 1
into (158) yields |1205910| = 120587119896 Furthermore substitution of
119886119896= 1 and 120591
0= 120587119896 (or 120591
0= minus120587119896) into (156) along
with performing all multiplications and using (A2) leads to
20 Mathematical Problems in Engineering
119879119896(120591) = 1 + cos 119896120591 Consequently (156)ndash(158) hold for 119886
119896= 1
and |1205910| = 120587119896
It is easy to see that 120582119896= |119886119896| lt 1 and 119879
119896(1205910) = 0 for some
1205910imply 120582
1= |1198861| = 0 Therefore in what follows we assume
that |119886119896| = 1 and 119886
1lt 0
Cosine waveforms are even functions of 120591 Therefore ifnonnegative cosine waveform has exactly one zero it has to beeither at 0 or at 120587 On the other hand if nonnegative cosinewaveform with 119886
1= 0 has exactly two zeros then these zeros
are placed at plusmn1205910 such that 120591
0is neither 0 nor 120587
In order to prove that (155) holds for minus1 lt 119886119896le 1(1minus119896
2
)let us start by referring to the description (38) of nonnegativewaveformswith at least one zero As wementioned earlier fornonnegative cosine waveformwith exactly one zero (denotedby 1205910) it is either 120591
0= 0 or 120591
0= 120587 Therefore in both cases
sin 1205910= 0 Substitution of sin 120591
0= 0 into (43) together with
1198861
= 0 and 120582119896= |119886119896| lt 1 leads to
1205910= 0 (164)
Clearly 1205910= 0 119887
1= 0 and 119887
119896= 0 according to (44) and (46)
imply 120582119896sin 120585 = 0 Since 120582
119896= |119886119896| it follows that |119886
119896| sin 120585 = 0
also holds which further implies 120582119896= 119886119896= 0 or sin 120585 =
0 In the case when 120582119896= 119886119896= 0 from (164) and (43) we
obtain 1198861= minus1 which further implies that 119879
119896(120591) = 1 minus cos 120591
Consequently (155) holds for 119886119896= 0 In the case when sin 120585 =
0 from (164) and (45) we obtain 119886119896= 120582119896if 120585 = 0 or 119886
119896= minus120582119896
if 120585 = 120587 Relations 119886119896= 120582119896and 120585 = 0 according to (40) imply
that 0 le 119886119896le 1(1 minus 119896
2
) Substitution of 120585 = 0 120582119896= 119886119896 and
(164) into (38) leads to (155) which proves that (155) holdsfor 0 le 119886
119896le 1(1 minus 119896
2
) On the other hand relations 119886119896=
minus120582119896and 120585 = 120587 according to (40) imply that minus1 lt 119886
119896le 0
Substitution of 120585 = 120587 120582119896= minus119886119896 and (164) into (38) also leads
to (155) which proves that (155) also holds for minus1 lt 119886119896le 0
Consequently (155) holds for minus1 lt 119886119896le 1(1 minus 119896
2
)In what follows we first prove that (156)-(157) hold for
1(1 minus 1198962
) lt 119886119896lt 1 For this purpose let us start with non-
negative waveforms with two zeros described by (66) As wementioned before nonnegative cosine waveforms with twozeros have zeros at 120591
0and minus120591
0 such that 120591
0= 0 and 120591
0= 120587
Relations 1198861
lt 0 and 1198871
= 0 according to (84) implycos(1205910minus 120585119896) = 1 and therefore
120585
119896= 1205910 (165)
From 120585119896 = 1205910and 0 lt |120585| lt 120587 it follows that 0 lt |120591
0| lt
120587119896 Insertion of 120585119896 = 1205910into (45) yields 119886
119896= 120582119896 Relations
119886119896= 120582119896and (82) imply that 1(1 minus 1198962) lt 119886
119896lt 1 Substitution
of 120582119896= 119886119896and 120585119896 = 120591
0into (66)ndash(68) leads to (156)ndash(158)
which proves that (156)ndash(158) hold for 1(1 minus 1198962
) lt 119886119896lt 1
and 0 lt |1205910| lt 120587119896
Finally substitution of 119886119896= 1(1 minus 119896
2
) and 1205910= 0 into
(161) leads to
119879119896(120591) =
[1 minus cos 120591](1 minus 1198962)
[119896 (119896 minus 1) minus 2
119896minus1
sum
119899=1
(119896 minus 119899) cos 119899120591]
(166)
Waveform (166) coincides with waveform (155) for 119886119896
=
1(1 minus 1198962
) which in turn proves that (156) holds for 119886119896=
1(1 minus 1198962
) and 1205910= 0 This completes the proof
62 Nonnegative CosineWaveforms with at Least One Zero for119896 = 3 In this subsection we consider nonnegative cosinewaveforms with at least one zero for 119896 = 3 (for case 119896 = 2
see [12])Cosine waveform with fundamental and third harmonic
reads
1198793(120591) = 1 + 119886
1cos 120591 + 119886
3cos 3120591 (167)
For 1198861le 0 and minus1 le 119886
3le 18 according to (155) non-
negative cosine waveform of type (167) with at least one zerocan be expressed as
1198793(120591) = (1 minus cos 120591) [1 minus 2119886
3(1 + 2 cos 120591 + cos 2120591)] (168)
From 1198793(120591 + 120587) = 2 minus 119879
3(120591) it immediately follows that for
1198861ge 0 and minus18 le 119886
3le 1 119879
3(120591) can be expressed as
1198793(120591) = (1 + cos 120591) [1 + 2119886
3(1 minus 2 cos 120591 + cos 2120591)] (169)
For 1198861le 0 and 18 le 119886
3le 1 from (158) it follows that 119886
3=
[8cos31205910]minus1 This relation along with (160) and (157) further
implies that 1198793(120591) can be expressed as
1198793(120591) =
[cos 1205910minus cos 120591]2 [2 cos 120591
0+ cos 120591]
2cos31205910
(170)
providing that |1205910| le 1205873 From 119879
3(120591 + 120587) = 2 minus 119879
3(120591) it
follows that (170) also holds for 1198861ge 0 and minus1 le 119886
3le minus18
providing that 1205910isin [21205873 41205873]
Maximally flat nonnegative cosinewaveformof type (167)with 119886
1lt 0 (minimum at 120591
0= 0) reads 119879
3(120591) = [1 minus
cos 120591]2[1 + (12) cos 120591] Dually maximally flat nonnegativecosine waveform with 119886
1gt 0 (minimum at 120591
0= 120587) reads
1198793(120591) = [1 + cos 120591]2[1 minus (12) cos 120591]In what follows we provide relations between coefficients
1198861and 1198863of nonnegative cosine waveforms of type (167) with
at least one zeroFor 1198861le 0 conversion of (168) into an additive form
immediately leads to the following relation
1198861= minus1 minus 119886
3for minus 1 le 119886
3le1
8 (171)
Conversion of (170) into an additive form leads to 1198861
=
minus31198863(2 cos 2120591
0+ 1) which can be also expressed as 119886
1=
minus31198863(4cos2120591
0minus 1) For 119886
1le 0 relations |120591
0| le 1205873 119886
1=
minus31198863(4cos2120591
0minus 1) and 119886
3= [8cos3120591
0]minus1 lead to
1198861= minus3 [ 3radic119886
3minus 1198863] for 1
8le 1198863le 1 (172)
Similarly for 1198861ge 0 conversion of (169) into an additive form
leads to the following relation
1198861= 1 minus 119886
3for minus
1
8le 1198863le 1 (173)
Mathematical Problems in Engineering 21
1
15
15
05
0
minus05
minus15minus15
minus1
minus1 minus05 0 05 1
Coefficient a1
Coe
ffici
enta
3 a1= minus
3(3radica3minus a3)
a1 =
minus1 minus a3
a1 =
1 minusa3
a1 =3(3radic|a3
| +a3)
Figure 14 Parameter space of cosine waveforms for 119896 = 3
For waveform of type (170) with 1198861ge 0 relations 120591
0isin [21205873
41205873] 1198861= minus3119886
3(4cos2120591
0minus 1) and 119886
3= [8cos3120591
0]minus1 lead to
1198861= 3 [
3radic10038161003816100381610038161198863
1003816100381610038161003816 + 1198863] for minus 1 le 1198863le minus
1
8 (174)
Every cosine waveform of type (167) corresponds to apair of real numbers (119886
1 1198863) and vice versa Points (119886
1 1198863)
in grey area in Figure 14 correspond to nonnegative cosinewaveforms for 119896 = 3 The points at the boundary of grey areacorrespond to nonnegative cosinewaveformswith at least onezero A number of shapes of nonnegative cosine waveformswith 119896 = 3 and at least one zero plotted on interval [minus120587 120587]are also presented in Figure 14 The boundary of grey area inFigure 14 consists of four line segments described by relations(171)ndash(174) The common point of line segments (172) and(173) is cusp point with coordinates 119886
1= 0 and 119886
3= 1
Another cusp point with coordinates 1198861= 0 and 119886
3= minus1
is the common point of line segments (171) and (174) Thecommon point of line segments (171)-(172) has coordinates(minus98 18) and common point of line segments (173)-(174)has coordinates (98 minus18) These points are representedby white circle dots and they correspond to maximallyflat cosine waveforms (eg see [21]) White triangle dotswith coordinates (2radic3 minusradic39) and (minus2radic3radic39) refer tothe nonnegative cosine waveforms with maximum value ofamplitude of fundamental harmonic
7 Four Case Studies of Usage of NonnegativeWaveforms in PA Efficiency Analysis
In this section we provide four case studies of usage ofdescription of nonnegative waveforms with fundamental and119896th harmonic in PA efficiency analysis In first two casestudies to be presented in Section 71 voltage is nonnegativewaveform with fundamental and second harmonic with atleast one zero In remaining two case studies to be consideredin Section 72 voltage waveform contains fundamental andthird harmonic
i
v Load
Vdc
Idc
Lch
vL
iL
Cb
+ +in
Figure 15 Generic PA circuit diagram
Let us consider generic PA circuit diagram as shown inFigure 15 We assume here that voltage and current wave-forms at the transistor output are
V (120579) = 1 + 1198861V cos 120579 + 1198871V sin 120579 + 119886119896V cos 119896120579 + 119887119896V sin 119896120579
119894 (120579) = 1 + 1198861119894cos 120579 +
infin
sum
119899=2
119886119899119894cos 119899120579
(175)
where 120579 stands for 120596119905 Both waveforms are normalized in thesense that dc components of voltage and current are 119881dc =
1 and 119868dc = 1 respectively Under assumption that blockingcapacitor 119862
119887behaves as short-circuit at the fundamental and
higher harmonics current and voltage waveforms at the loadare
V119871(120579) = 119886
1V cos 120579 + 1198871V sin 120579 + 119886119896V cos 119896120579 + 119887119896V sin 119896120579
119894119871(120579) = minus119886
1119894cos 120579 minus
infin
sum
119899=2
119886119899119894cos 119899120579
(176)
In terms of coefficients of voltage and current waveforms theload impedance at fundamental harmonic is 119911
1= minus(119886
1V minus
1198951198871V)1198861119894 whereas load impedance at 119896th harmonic is 119911
119896=
minus(119886119896Vminus119895119887119896V)119886119896119894 All other harmonics are short-circuited (119911
119899=
0 for 119899 = 1 and 119899 = 119896) Time average output power of PA (egsee [10]) with waveform pair (175) at fundamental frequencycan be expressed as
1198751= minus
11988611198941198861V
2 (177)
For normalized waveforms (175) with 119881dc = 1 and 119868dc = 1dc power is 119875dc = 1 Consequently PA efficiency 120578 = 119875
1119875dc
(eg see [10 26]) is equal to
120578 = minus11988611198941198861V
2 (178)
Thus time average output power 1198751of PA with pair of nor-
malized waveform (175) is equal to efficiency (178)Power utilization factor (PUF) is defined [26] as ldquothe
ratio of power delivered in a given situation to the power
22 Mathematical Problems in Engineering
delivered by the same device with the same supply voltagein Class A moderdquo Since the output power in class-A modeis 1198751class-A = max[V(120579)] sdotmax[119894(120579)]8 (eg see [9]) it follows
that power utilization factor PUF = 11987511198751class-A for PA with
pair of normalized waveforms (175) can be expressed as
PUF =8120578
max [V (120579)] sdotmax [119894 (120579)] (179)
71 NonnegativeWaveforms for 119896 = 2 in PAEfficiencyAnalysisIn this subsection we provide two case studies of usage ofdescription of nonnegative waveforms with fundamental andsecond harmonic (119896 = 2) in PA efficiency analysis For moreexamples of usage of descriptions of nonnegative waveformswith fundamental and second harmonic in PA efficiencyanalysis see [12]
Case Study 71 In this case study we consider efficiency of PAfor given second harmonic impedance providing that voltageis nonnegative waveform with fundamental and second har-monic and current is ldquohalf-sinerdquo waveform frequently used inefficiency analysis of classical PA operation (eg see [10])
Standard model of current waveform for classical PAoperation has the form (eg see [10 26])
119894119863(120579) =
119868119863[cos 120579 minus cos(120572
2)] |120579| le
120572
2
0120572
2le |120579| le 120587
(180)
where 120572 is conduction angle and 119868119863
gt 0 Since 119894119863(120579) is
even function it immediately follows that its Fourier seriescontains only dc component and cosine terms
119894119863(120579) = 119868dc +
infin
sum
119899=1
119868119899cos 119899120579 (181)
The dc component of the waveform (180) is
119868dc =119868119863120572
2120587[sinc(120572
2) minus cos(120572
2)] (182)
where sinc119909 = (sin119909)119909 The coefficient of the fundamentalharmonic component reads
1198681=119868119863120572
2120587(1 minus sinc120572) (183)
and the coefficient of 119899th harmonic component can bewrittenin the form
119868119899=119868119863
119899120587[sin ((119899 minus 1) 1205722)
(119899 minus 1)minussin ((119899 + 1) 1205722)
(119899 + 1)] 119899 ge 2
(184)
For ldquohalf-sinerdquo current waveform conduction angle is equalto 120587 (class-B conduction angle) According to (182) thisfurther implies that 119868dc = 119868
119863120587 To obtain normalized form of
waveform (180) we set 119868dc = 1which implies that 119868119863= 120587 Fur-
thermore substitution of 120572 = 120587 and 119868119863= 120587 in (180) leads to
119894 (120579) =
120587 cos 120579 |120579| lt120587
2
0120587
2lt |120579| le 120587
(185)
Similarly substitution of 119868119863= 120587 and 120572 = 120587 into (183) and
(184) leads to the coefficients of waveform (185) Coefficientsof fundamental and second harmonic respectively are
1198861119894=120587
2 119886
2119894=2
3 (186)
On the other hand voltage waveform of type (35) for 119896 =2 reads
V (120579) = 1 + 1198861V cos 120579 + 1198871V sin 120579 + 1198862V cos 2120579 + 1198872V sin 2120579
(187)
This waveform contains only fundamental and second har-monic and therefore all harmonics of order higher thantwo are short-circuited (119911
119899= 0 for 119899 gt 2) For current
voltage pair (185) and (187) load impedance at fundamentalharmonic is 119911
1= minus(119886
1V minus 1198951198871V)1198861119894 whereas load impedance
at second harmonic is 1199112= minus(119886
2Vminus1198951198872V)1198862119894 According to ourassumption the load is passive and therefore Re119911
1 gt 0 and
Re1199112 ge 0 which further imply 119886
11198941198861V lt 0 and 119886
21198941198862V le 0
respectivelyIt is easy to see that problem of findingmaximal efficiency
of PA with current-voltage pair (185) and (187) for prescribedsecond harmonic impedance can be reduced to the problemof finding voltage waveform of type (187) with maximal coef-ficient |119886
1V| for prescribed coefficients of second harmonic(see Section 5)
The following algorithm (analogous to Algorithm 22presented in [12]) provides the procedure for calculation ofmaximal efficiency with current-voltage pair (185) and (187)for prescribed second harmonic impedance The definitionof function atan 2(119910 119909) which appears in the step (iii) of thefollowing algorithm is given by (105)
Algorithm 32 (i) Choose 1199112= 1199032+1198951199092such that |119911
2| le 1|119886
2119894|
(ii) calculate 1198862V minus 1198951198872V = minus119911
21198862119894and 120582
2V = radic11988622V + 1198872
2V(iii) if 2120582
2V le 1 minus 1198862V then calculate 119886
1V = minus1 minus 1198862V and
1198871V = minus2119887
2V else calculate 1205821V = radic81205822V(1 minus 1205822V) 1205790V minus 120585V2 =
(12)atan2(1198872V 1198862V) 1198861V = minus120582
1V cos(1205790V minus 120585V2) and 1198871V =
minus1205821V sin(1205790V minus 120585V2)(iv) calculate efficiency 120578 = minus119886
11198941198861V2
(v) calculate 1199111= minus(119886
1V minus 1198951198871V)1198861119894 and 1199112119899 = 1199112Re119911
1
In this case study coefficients of fundamental and secondharmonic of current waveform are given by (186) Maximalefficiency of PA associated with the waveform pair (185)and (187) as a function of normalized second harmonicimpedance 119911
2119899= 1199112Re119911
1 is presented in Figure 16(a)
As can be seen from Figure 16(a) efficiency of 078 isachieved at the edge of Smith chart where second harmonicimpedance has small resistive part Corresponding PUFcalculated according to (179) is presented in Figure 16(b)Peak efficiency 120578 = 1205874 = 07854 and peak value of PUF = 1
are attained when second harmonic is short-circuited (whichcorresponds to ideal class-B operation [10 26])
For example for second harmonic impedance 1199112= 01 minus
11989505 and currentwaveform (185) fromAlgorithm32 it followsthat 2120582
2V le 1 minus 1198862V Furthermore according to step (iii)
of above algorithm maximal efficiency of PA is attained
Mathematical Problems in Engineering 23
+j2
+j5
+j1
+j2
+j5
minusj2
minusj5
minusj1
minusj2
minusj5
infin0
078
02
05
1 2 5
075 07 065
06 05
075
07
065
06
05
120578 lt 05
(a)
099
095
091
083 075
067
0 59 051
051
099
095
091
083
075 0
67 059
+j5
+j1
+j2
+j5
minusj5
minusj1
minusj2
minusj5
infin
+j2
minusj2
0 02
05
1 2 5
120578 lt 05
(b)
Figure 16 (a) Contours of maximal efficiency of PA and (b) contours of corresponding PUF associated with the waveform pair (185) and(187) as functions of normalized second harmonic impedance 119911
2119899= 1199112Re119911
1
2
3
4
1
0
2 3 410
Wav
efor
ms
Angle 120579120587
i(120579)
(120579)
Figure 17 Waveform pair (185) and (187) that provides maximalefficiency for 119911
2= 01 minus 11989505
with voltage waveform of type (187) with coefficients 1198862V =
minus00667 1198872V = minus03333 119886
1V = minus09333 and 1198871V = 06667 (see
Figure 17) Corresponding efficiency PUF and normalizedsecond harmonic impedance are 120578 = 07330 PUF = 07572and 1199112119899= 01683 minus 11989508415 respectively
On the other hand for second harmonic impedance 1199112=
01 minus 11989508 and current waveform (185) from Algorithm 32 itfollows that 2120582
2V gt 1 minus 1198862V Then according to step (iii) of
above algorithm maximal efficiency is attained with voltagewaveform of type (187) with coefficients 119886
2V = minus00667 1198872V =
minus05333 1198861V = minus09333 and 119887
1V = 10572 (see Figure 18)Efficiency PUF andnormalized secondharmonic impedanceare 120578 = 07330 PUF = 06332 and 119911
2119899= 01683 minus 11989513465
respectively
2
3
4
1
0
2 3 410
Wav
efor
ms
Angle 120579120587
i(120579)
(120579)
Figure 18 Waveform pair (185) and (187) that provides maximalefficiency for 119911
2= 01 minus 11989508
Case Study 72 As another case study let us consider the effi-ciency of PA providing that current waveform is nonnegativecosine waveform up to third harmonic with maximum valueof amplitude of fundamental harmonic [22] (see also [8])
119894 (120579) = 1 +1 + radic5
2cos 120579 + 2radic5
5cos 2120579 + 5 minus radic5
10cos 3120579
(188)
and voltage waveform is nonnegative waveform of type(187) Load impedances at fundamental second and thirdharmonic are 119911
1= minus(119886
1Vminus1198951198871V)1198861119894 1199112 = minus(1198862Vminus1198951198872V)1198862119894 and
1199113= 0 respectively According to our assumption the load
is passive and therefore Re1199111 gt 0 and Re119911
2 ge 0 which
further imply 11988611198941198861V lt 0 and 119886
21198941198862V le 0 respectively
Because current waveform (188) contains only cosineterms and voltage waveform is the same as in previous case
24 Mathematical Problems in Engineering
08
075
07
06
06
065
065
05
05
075 0
7
120578 lt 05
+j2
+j5
+j1
+j2
+j5
minusj2
minusj5
minusj1
minusj2
minusj5
infin0
02
05
1 2 5
(a)
085
08 0
75 07 0
6
05 04
08
075
0405
060
7
120578 lt 05
+j2
+j5
+j1
+j2
+j5
minusj2
minusj5
minusj1
minusj2
minusj5
infin0 02
05
1 2 5
(b)
Figure 19 (a) Contours of maximal efficiency of PA and (b) contours of corresponding PUF associated with the waveform pair (187)-(188)as functions of normalized second harmonic impedance 119911
2119899= 1199112Re119911
1
study the procedure for calculation of maximal efficiency ofPA with waveform pair (187)-(188) is the same as presentedin Algorithm 32 In this case study the coefficients of funda-mental and second harmonic of current waveform are 119886
1119894=
(1 + radic5)2 and 1198862119894= 2radic55 respectively
Maximal efficiency of PA associated with the waveformpair (187)-(188) as a function of normalized secondharmonicimpedance 119911
2119899= 1199112Re119911
1 is presented in Figure 19(a)
Efficiency of 08 is achieved at the edge of Smith chart wheresecond harmonic impedance has small resistive part Thetheoretical upper bound 120578 = (1 + radic5)4 asymp 08090 isattained when second harmonic is short-circuitedWhen thisupper bound is reached both second and third harmonic areshort-circuited which implies that we are dealing with finiteharmonic class-C [6 8] or dually when current and voltageinterchange their roles with finite harmonic inverse class-C[6 9] Corresponding PUF calculated according to (179) ispresented in Figure 19(b) Peak value of PUF asymp 08541 isattained when second harmonic is short-circuited
For example for second harmonic impedance 1199112
=
007 minus 11989504 and current waveform (188) from Algorithm 32 itfollows that 2120582
2V le 1 minus 1198862V Furthermore according to step
(iii) of Algorithm 32 maximal efficiency of PA is attainedwith voltage waveform of type (187) with coefficients 119886
2V =
minus00626 1198872V = minus03578 119886
1V = minus09374 and 1198871V = 07155 (see
Figure 20) Corresponding efficiency PUF and normalizedsecond harmonic impedance are 120578 = 07584 PUF = 06337and 1199112119899= 01208 minus 11989506904 respectively
On the other hand for 1199112= 005 minus 11989507 and current
waveform (187) it follows that 21205822V gt 1minus119886
2VThen accordingto step (iii) of Algorithm 32 themaximal efficiency is attainedwith voltage waveform of type (187) with coefficients 119886
2V =
minus00447 1198872V = minus06261 119886
1V = minus09318 and 1198871V = 10007 (see
Figure 21) Efficiency PUF and normalized second harmonic
2
3
4
1
0
2 3 410
Wav
efor
ms
Angle 120579120587
i(120579)
(120579)
Figure 20 Waveform pair (187)-(188) that provides maximal effi-ciency for 119911
2= 007 minus 11989504
impedance are 120578 = 07538 PUF = 05314 and 1199112119899= 00868minus
11989512156 respectively
72 Nonnegative Waveforms for 119896 = 3 in PA EfficiencyAnalysis In this subsection we provide another two casestudies of usage of description of nonnegative waveforms inPA efficiency analysis this time with fundamental and thirdharmonic (119896 = 3)
Case Study 73 Let us consider current-voltage pair such thatvoltage is nonnegative waveform with fundamental and thirdharmonic
V (120579) = 1 + 1198861V cos 120579 + 1198871V sin 120579 + 1198863V cos 3120579 + 1198873V sin 3120579
(189)
Mathematical Problems in Engineering 25
2
3
4
1
0
2 3 410
Wav
efor
ms
Angle 120579120587
i(120579)
(120579)
Figure 21 Waveform pair (187)-(188) that provides maximal effi-ciency for 119911
2= 005 minus 11989507
and current is nonnegative cosine waveform given by (188)Load impedances at fundamental second and third har-monic are 119911
1= minus(119886
1V minus 1198951198871V)1198861119894 1199112 = 0 and 119911
3= minus(119886
3V minus
1198951198873V)1198863119894 respectively According to our assumption the load
is passive and therefore Re1199111 gt 0 and Re119911
3 ge 0 which
further imply 11988611198941198861V lt 0 and 119886
31198941198863V le 0
In this subsection we consider the problem of findingmaximal efficiency of PA with waveform pair (188)-(189)for given third harmonic impedance As we mentionedearlier problem of finding maximal efficiency of PA withcurrent-voltage pair (188)-(189) for prescribed third har-monic impedance can be reduced to the problem of findingvoltage waveform of type (189) withmaximal coefficient |119886
1V|for prescribed coefficients of third harmonic (see Section 52)
The following algorithm provides the procedure forcalculation of maximal efficiency with current-voltage pair(188)-(189) The definition of function atan 2(119910 119909) whichappears in step (iii) of the following algorithm is given by(105)
Algorithm 33 (i) Choose 1199113= 1199033+1198951199093such that |119911
3| le 1|119886
3119894|
(ii) calculate 1198863V minus 1198951198873V = minus119911
31198863119894and 120582
3V = radic11988623V + 1198872
3V(iii) if 271205822
3V le (1 minus 21198863V)3 then calculate 119886
1V = minus1 minus 1198863V
and 1198871V = minus3119887
3V else calculate 1205821V = 3(3radic1205823V minus 120582
3V) 1205790V minus120585V3 = (13)atan 2(119887
3V 1198863V) 1198861V = minus1205821V cos(1205790V minus 120585V3) and
1198871V = minus120582
1V sin(1205790V minus 120585V3)(iv) calculate efficiency 120578 = minus119886
11198941198861V2
(v) calculate 1199111= minus(119886
1V minus 1198951198871V)1198861119894 and 1199113119899 = 1199113Re119911
1
In this case study coefficients of fundamental and thirdharmonic of current waveform are 119886
1119894= (1 + radic5)2 and
1198863119894= (5 minus radic5)10 respectively For the waveform pair (188)-
(189) maximal efficiency of PA as a function of normalizedthird harmonic impedance 119911
3119899= 1199113Re119911
1 is presented in
Figure 22 Efficiency of 08 is reached when third harmonicimpedance has small resistive part Peak efficiency 120578 = (1 +
radic5)4 asymp 08090 is achieved when third harmonic is short-circuited
+j2
+j5
+j1
+j2
+j5
minusj2
minusj5
minusj1
minusj2
minusj5
infin0
08
08
075
075
07
07
06
05
02
05
1 2
Figure 22 Contours of maximal efficiency of PA associated withthe waveform pair (188)-(189) as a function of normalized thirdharmonic impedance 119911
3119899= 1199113Re119911
1
For the present case study in what follows we show thatpower utilization factor is proportional to efficiency For volt-age waveform of type (189) it is easy to see that V(120579 + 120587) =
2minusV(120579) holdsThis relation along with the fact that waveformV(120579) that provides maximal efficiency has at least one zeroimplies that max[V(120579)] = 2 On the other hand currentwaveform (188) is cosine waveform with positive coefficientsand therefore max[119894(120579)] = 119894(0) = 2 + 4radic5 Consequentlyaccording to (179) the following relation holds
PUFCase study 73 = 2 (5 minus 2radic5) 120578 = 10557120578 (190)
Clearly the ratio PUF120578 is constant and therefore in this casestudy PUF can be easily calculated from the correspondingefficiency Accordingly peak efficiency and peak value ofPUFCase study 73 = 3radic52 minus 52 = 08541 are attained forthe same voltage waveform (when third harmonic is short-circuited)
In the first example current waveform (188) and 1199113=
02 minus 11989505 imply that 2712058223V le (1 minus 2119886
3V)3 Then according
to Algorithm 33 the voltage waveform of type (189) thatprovides maximal efficiency has the following coefficients1198863V = minus00553 119887
3V = minus01382 1198861V = minus09447 and 119887
1V =
04146 (see Figure 23) Efficiency PUF and normalized thirdharmonic impedance are 120578 = 07643 PUF = 08069 and1199113119899= 03425 minus 11989508564 respectivelyIn the second example current waveform (188) and 119911
3=
01 minus 11989511 imply that 2712058223V gt (1 minus 2119886
3V)3 Then according
to Algorithm 33 the voltage waveform of type (189) thatprovides maximal efficiency has the following coefficients1198863V = minus00276 119887
3V = minus03040 1198861V = minus09391 and 119887
1V =
05807 (see Figure 24) Efficiency PUF and normalized thirdharmonic impedance are 120578 = 07598 PUF = 08021 and1199113119899= 01723 minus 11989518952 respectively
26 Mathematical Problems in Engineering
2
3
4
1
0
2 3 410
Wav
efor
ms
Angle 120579120587
i(120579)
(120579)
Figure 23 Waveform pair (188)-(189) that provides maximal effi-ciency for 119911
3= 02 minus 11989505
2
3
4
1
0
2 3 410
Wav
efor
ms
Angle 120579120587
i(120579)
(120579)
Figure 24 Waveform pair (188)-(189) that provides maximal effi-ciency for 119911
3= 01 minus 11989511
Case Study 74 In this case study let us consider current-voltage pair where current is normalized waveform of type(180) with conduction angle 120572 = 115120587 (207∘) and voltageis nonnegative waveform of type (189) Substitution of 120572 =
115120587 and 119868dc = 1 into (182) leads to 119868119863
= 22535Furthermore substitution of 120572 = 115120587 and 119868
119863= 22535 into
(180) leads to
119894 (120579)
=
22535 [cos 120579 minus cos(1151205872
)] |120579| le115120587
2
0115120587
2le |120579| le 120587
(191)
Similarly substitution of 120572 = 115120587 and 119868119863
= 22535 into(183) and (184) for 119899 = 3 yields coefficients of fundamentaland third harmonic of waveform (191)
1198861119894= 14586 119886
3119894= minus01026 (192)
Because current waveform (191) contains only cosineterms and voltage waveform is the same as in previous case
+j2
+j5
+j1
+j2
+j5
minusj2
minusj5
minusj1
minusj2
minusj5
infin0 02
05
1 2
074
076
08
084
082
078
076
074
Figure 25 Contours of maximal efficiency of PA associated withthe waveform pair (189) and (191) as a function of normalized thirdharmonic impedance 119911
3119899= 1199113Re119911
1
study the procedure for calculation of maximal efficiencyof PA with waveform pair (189)ndash(191) is the same as thatpresented in Algorithm 33 In this case study the coefficientsof fundamental and third harmonic of current waveform aregiven by (192)
For the waveform pair (189) and (191) maximal efficiencyof PA as a function of normalized third harmonic impedance1199113119899= 1199113Re119911
1 is presented in Figure 25 Efficiency of 084
is obtained in vicinity of 1199113119899= 23685 (corresponding to 119911
3=
18750) Peak efficiency 120578 asymp 08421 is achieved for voltagewaveform of type (189) with coefficients 119886
1V = minus2radic3 1198863V =
radic39 and 1198871V = 1198873V = 0
In the course of finding power utilization factor noticethat currentwaveformof type (191) attains itsmaximumvaluefor 120579 = 0 Insertion of max[119894(120579)] = 119894(0) = 278 andmax[V(120579)] = 2 for voltage waveform of type (189) into (179)leads to
PUFCase study 74 = 1439120578 (193)
Again the ratio PUF120578 is constant and PUF can be easilycalculated from the corresponding efficiency Accordinglypeak value of PUFCase study 74 asymp 12118 and peak efficiencyare attained for the same voltage waveform
In the first example current waveform (191) and 1199113=
1 minus 11989502 imply that 2712058223V le (1 minus 2119886
3V)3 Then according to
Algorithm 33 voltage waveform of type (189) which providesmaximal efficiency has coefficients 119886
3V = 01026 1198873V =
00205 1198861V = minus11026 and 119887
1V = minus00616 (see Figure 26)Efficiency PUF and normalized third harmonic impedanceare 120578 = 08042 PUF = 11572 and 119911
3119899= 13228 minus 11989502646
respectivelyIn second example current waveform (191) and 119911
3=
15 minus 11989512 imply that 2712058223V gt (1 minus 2119886
3V)3 Then according to
Algorithm 33 voltage waveform of type (189) which providesmaximal efficiency has coefficients 119886
3V = 01540 1198873V =
Mathematical Problems in Engineering 27
2
3
4
1
0
2 3 410
Wav
efor
ms
Angle 120579120587
i(120579)
(120579)
Figure 26 Waveform pair (189) and (191) that provides maximalefficiency for 119911
3= 1 minus 11989502
2
3
4
1
0
2 3 410
Wav
efor
ms
Angle 120579120587
i(120579)
(120579)
Figure 27 Waveform pair (189) and (191) that provides maximalefficiency for 119911
3= 15 minus 11989512
01232 1198861V = minus11255 and 119887
1V = minus02575 (see Figure 27)Efficiency PUF and normalized third harmonic impedanceare 120578 = 08208 PUF = 11812 and 119911
3119899= 19439 minus 11989515552
respectively
8 Conclusion
In this paper we consider a problem of finding general de-scriptions of various classes of nonnegative waveforms withfundamental and 119896th harmonic These classes include non-negative waveforms with at least one zero nonnegative wave-forms with maximal amplitude of fundamental harmonic forprescribed amplitude of 119896th harmonic nonnegative wave-forms withmaximal coefficient of cosine part of fundamentalharmonic for prescribed coefficients of 119896th harmonic andnonnegative cosine waveforms with at least one zero Mainresults are stated in six propositions (Propositions 1 6 918 22 and 26) four corollaries (Corollaries 2ndash5) twentyremarks and three algorithms Four case studies of usageof closed form descriptions of nonnegative waveforms in PAefficiency analysis are considered in detail in Section 7
Appendices
Here we provide a list of finite sums of trigonometric func-tions used in this paper (Appendix A) and brief account ofthe Chebyshev polynomials (Appendix B)
A List of Some Finite Sums ofTrigonometric Functions
Dirichlet kernel (eg see [27]) is as follows
119863119896minus1
(120591) = 1 + 2
119896minus1
sum
119899=1
cos 119899120591 = sin ((2119896 minus 1) 1205912)sin (1205912)
(A1)
Fejer kernel (eg see [27]) can be expressed in the fol-lowing equivalent forms
119865119896minus1
(120591) =1
119896
119896minus1
sum
119899=0
119863119899(120591) = 1 +
2
119896
119896minus1
sum
119899=1
(119896 minus 119899) cos 119899120591
=(1 minus cos 119896120591)119896 (1 minus cos 120591)
(A2)
Lagrangersquos trigonometric identity (eg see [28]) is as fol-lows
1198781(120591) =
119896minus1
sum
119899=1
sin 119899120591 = sin (1198961205912) sin ((119896 minus 1) 1205912)sin (1205912)
(A3)
In what follows we show that the following three trigono-metric identities also hold
2
119896minus1
sum
119899=1
(119896 minus 119899) sin 119899120591 = 119896 sin 120591 minus sin 1198961205911 minus cos 120591
(A4)
119896minus1
sum
119899=1
cos (119896 minus 2119899) 120591 = sin (119896 minus 1) 120591sin 120591
(A5)
119896minus1
sum
119899=1
119899 (119896 minus 119899) cos (119896 minus 2119899) 120591
=sin (119896120591) cos 120591 minus 119896 cos (119896120591) sin 120591
2sin3120591
(A6)
Denote 1198782(120591) = 2sum
119896minus1
119899=1(119896 minus 119899) sin 119899120591 119878
3(120591) = sum
119896minus1
119899=1cos(119896 minus
2119899)120591 and 1198784(120591) = sum
119896minus1
119899=1119899(119896 minus 119899) cos(119896 minus 2119899)120591
Notice that 1198782(120591) = 2119896119878
1(120591) + 119889119863
119896minus1(120591)119889120591 which
immediately leads to (A4)
28 Mathematical Problems in Engineering
Identity (A5) can be obtained as follows
sin (119896 minus 1) 120591sin 120591
=119890119895(119896minus1)120591
minus 119890minus119895(119896minus1)120591
119890119895120591 minus 119890minus119895120591
= 119890119895119896120591
119890minus2119895120591
minus 119890minus2119895(119896minus1)120591
1 minus 119890minus2119895120591
= 119890119895119896120591
119896minus1
sum
119899=1
119890minus2119895119899120591
=
119896minus1
sum
119899=1
119890119895(119896minus2119899)120591
=
119896minus1
sum
119899=1
cos (119896 minus 2119899) 120591
(A7)
From 4119899(119896 minus 119899) = 1198962
minus (119896 minus 2119899)2 it follows that 4119878
4(120591) =
1198962
1198783(120591) + 119889
2
1198783(120591)119889120591
2 which leads to (A6)
B The Chebyshev Polynomials
The Chebyshev polynomials of the first kind 119881119899(119909) can be
defined by the following relation (eg see [29])
119881119899(119909) = cos 119899120591 when 119909 = cos 120591 (B1)
The Chebyshev polynomials of the second kind 119880119899(119909) can be
defined by the following relation (eg see [29])
119880119899(119909) =
sin (119899 + 1) 120591sin 120591
when 119909 = cos 120591 (B2)
The Chebyshev polynomials satisfy the following recur-rence relations (eg see [29])
1198810(119909) = 1 119881
1(119909) = 119909
119881119899+1
(119909) = 2119909119881119899(119909) minus 119881
119899minus1(119909)
1198800(119909) = 1 119880
1(119909) = 2119909
119880119899+1
(119909) = 2119909119880119899(119909) minus 119880
119899minus1(119909)
(B3)
The first few Chebyshev polynomials of the first and secondkind are 119881
2(119909) = 2119909
2
minus 1 1198813(119909) = 4119909
3
minus 3119909 1198814(119909) = 8119909
4
minus
81199092
+ 1 1198802(119909) = 4119909
2
minus 1 1198803(119909) = 8119909
3
minus 4119909 and 1198804(119909) =
161199094
minus 121199092
+ 1
Conflict of Interests
The authors declare that there is no conflict of interestsregarding the publication of this paper
Acknowledgment
This work is supported by the SerbianMinistry of EducationScience and Technology Development as a part of ProjectTP32016
References
[1] V I Arnolrsquod V S Afrajmovich Y S Ilrsquoyashenko and L PShilrsquonikov Dynamical Systems V Bifurcation Theory and Catas-trophe Theory Springer Berlin Germany 1994
[2] E Polak ldquoOn the mathematical foundations of nondifferen-tiable optimization in engineering designrdquo SIAM Review vol29 no 1 pp 21ndash89 1987
[3] N S Fuzik ldquoBiharmonic modes of a tuned RF power amplifierrdquoRadiotehnika vol 25 no 7 pp 62ndash71 1970 (Russian)
[4] P Colantonio F Giannini G Leuzzi and E Limiti ldquoClass Gapproach for low-voltage high-efficiency PA designrdquo Interna-tional Journal of RF and Microwave Computer-Aided Engineer-ing vol 10 no 6 pp 366ndash378 2000
[5] F H Raab ldquoMaximum efficiency and output of class-F poweramplifiersrdquo IEEE Transactions on Microwave Theory and Tech-niques vol 49 no 6 pp 1162ndash1166 2001
[6] FH Raab ldquoClass-E class-C and class-F power amplifiers basedupon a finite number of harmonicsrdquo IEEE Transactions onMicrowaveTheory and Techniques vol 49 no 8 pp 1462ndash14682001
[7] J D Rhodes ldquoOutput universality inmaximum efficiency linearpower amplifiersrdquo International Journal of Circuit Theory andApplications vol 31 no 4 pp 385ndash405 2003
[8] A Juhas and L A Novak ldquoComments on lsquoClass-E class-Cand classF power amplifier based upon a finite number ofharmonicsrsquordquo IEEE Transactions on Microwave Theory and Tech-niques vol 57 no 6 pp 1623ndash1625 2009
[9] M Roberg and Z Popovic ldquoAnalysis of high-efficiency poweramplifiers with arbitrary output harmonic terminationsrdquo IEEETransactions on Microwave Theory and Techniques vol 59 no8 pp 2037ndash2048 2011
[10] A Grebennikov N O Sokal and M J Franco Switchmode RFPower Amplifiers ElsevierAcademic Press San Diego CalifUSA 2nd edition 2012
[11] T Canning P J Tasker and S C Cripps ldquoContinuous modepower amplifier design using harmonic clipping contourstheory and practicerdquo IEEE Transactions on Microwave Theoryand Techniques vol 62 no 1 pp 100ndash110 2014
[12] A Juhas and L A Novak ldquoGeneral description of nonnegativewaveforms up to second harmonic for power amplifier mod-ellingrdquoMathematical Problems in Engineering vol 2014 ArticleID 709762 18 pages 2014
[13] V I Arnolrsquod V V Goryunov O V Lyashko and V A VasilrsquoevDynamical Systems VIIImdashSingularity Theory II ApplicationsSpringer Berlin Germany 1993
[14] D Siersma ldquoProperties of conflict sets in the planerdquo BanachCenter Publications Polish Academy of Sciences vol 50 no 1 pp267ndash276 1999 Proceedings of the Banach Center Symposiumon Geometry and Topology of Caustics (Caustics rsquo98) WarsawPoland
[15] M van Manen The geometry of conflict sets [Dissertation]Universiteit Utrecht UtrechtThe Netherlands 2003 httpigi-tur-archivelibraryuunldissertations2003-0912-123058c4pdf
[16] Y L Sachkov ldquoMaxwell strata and symmetries in the problemofoptimal rolling of a sphere over a planerdquo Sbornik Mathematicsvol 201 no 7-8 pp 1029ndash1051 2010
[17] I A Bogaevsky ldquoPerestroikas of shocks and singularities ofminimum functionsrdquoPhysicaDNonlinear Phenomena vol 173no 1-2 pp 1ndash28 2002
[18] Y L Sachkov ldquoMaxwell strata in the Euler elastic problemrdquoJournal of Dynamical andControl Systems vol 14 no 2 pp 169ndash234 2008
[19] M Siino and T Koike ldquoTopological classification of black holesgeneric Maxwell set and crease set of a horizonrdquo InternationalJournal ofModern Physics D Gravitation Astrophysics Cosmol-ogy vol 20 no 6 pp 1095ndash1122 2011
Mathematical Problems in Engineering 29
[20] F H Raab ldquoClass-F power amplifiers with maximally flat wave-formsrdquo IEEETransactions onMicrowaveTheory and Techniquesvol 45 no 11 pp 2007ndash2012 1997
[21] A Juhas and L A Novak ldquoMaximally flat waveforms with finitenumber of harmonics in class-F power amplifiersrdquo Mathemat-ical Problems in Engineering vol 2013 Article ID 169590 9pages 2013
[22] L Fejer ldquoUber trigonometrische polynomerdquo Journal fur dieReine und Angewandte Mathematik vol 1916 no 146 pp 53ndash82 1916 (German)
[23] S C Cripps ldquoBessel Waives [microwave bytes]rdquo IEEE Micro-wave Magazine vol 10 no 7 pp 30ndash36 117 2009
[24] L N Bryzgalova ldquoSingularities of the maximum of para-metrically dependent functionrdquo Functional Analysis and ItsApplications vol 11 no 1 pp 49ndash51 1977
[25] V I Arnold A A Davydov V A Vassiliev and V MZakalyukin Mathematical Models of Catastrophes Control ofCatastrophic Process Encyclopedia of Life Support Systems(EOLSS) EOLSS Publishers Oxford UK 2006
[26] S C Cripps RF Power Amplifiers for Wireless CommunicationsArtech House Norwood Mass USA 2nd edition 2006
[27] A Zygmund Trigonometric Series vol 1 Cambridge UniversityPress Cambridge UK 2nd edition 1959
[28] A Jeffrey and H DaiHandbook of Mathematical Formulas andIntegrals ElsevierAcademic Press San Diego Calif USA 4thedition 2008
[29] J C Mason and D C Handscomb Chebyshev PolynomialsChapman amp Hall CRC Press Boca Raton Fla USA 2003
Submit your manuscripts athttpwwwhindawicom
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical Problems in Engineering
Hindawi Publishing Corporationhttpwwwhindawicom
Differential EquationsInternational Journal of
Volume 2014
Applied MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Probability and StatisticsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical PhysicsAdvances in
Complex AnalysisJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
OptimizationJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
CombinatoricsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Operations ResearchAdvances in
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Function Spaces
Abstract and Applied AnalysisHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of Mathematics and Mathematical Sciences
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Algebra
Discrete Dynamics in Nature and Society
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Decision SciencesAdvances in
Discrete MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom
Volume 2014 Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Stochastic AnalysisInternational Journal of
Mathematical Problems in Engineering 15
where
|120575| le 120587 (119)
Conversely for prescribed coefficients 119886119896and 119887
119896 120575 can be
determined as
120575 = atan 2 (119887119896 119886119896) (120)
where definition of function atan 2(119910 119909) is given by (105)The main result of this section is stated in the following
proposition
Proposition 22 Every nonnegative waveform of type (35)withmaximal absolute value of coefficient 119886
1le 0 for prescribed
coefficients 119886119896and 119887119896of 119896th harmonic can be represented as
119879119896(120591)
= [1 minus cos 120591]
sdot [1 minus (119896 minus 1) 119886119896minus 2
119896minus1
sum
119899=1
(119896 minus 119899) (119886119896cos 119899120591 + 119887
119896sin 119899120591)]
(121)
if 119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) le 1+119886
119896 where 120575 = atan 2(bk
119886119896) or
119879119896(120591) = 120582
119896[1 minus cos(120591 minus (120575 + 120585)
119896)]
sdot [1 minus cos(120591 minus (120575 minus 120585)
119896)]
sdot [1198880+ 2
119896minus2
sum
119899=1
119888119899cos 119899(120591 minus 120575
119896)]
(122)
if 119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) ge 1 + 119886
119896 where 119888
119899 119899 = 0
119896minus2 and 120582119896= radic1198862119896+ 1198872119896are related to 120585 via relations (67) and
(68) respectively and |120585| le 120587
Remark 23 Expression (121) can be obtained from (38) bysetting 120591
0= 0 and 120585 = minus120575 and then replacing 120582
119896cos 120575 with
119886119896(see (118)) and 120582
119896cos(119899120591 minus 120575) with 119886
119896cos 119899120591 + 119887
119896sin 119899120591
(see also (118)) Furthermore insertion of 1205910= 0 and 120585 =
minus120575 into (43)ndash(46) leads to the following relations betweenfundamental and 119896th harmonic coefficients of waveform(121)
1198861= minus (1 + 119886
119896) 119887
1= minus119896119887
119896 (123)
On the other hand expression (122) can be obtained from(66) by replacing 120591
0minus120585119896with 120575119896 Therefore substitution of
1205910minus 120585119896 = 120575119896 in (84) leads to
1198861= minus1205821cos(120575
119896) 119887
1= minus1205821sin(120575
119896) (124)
where 1205821is given by (85)
The fundamental harmonic coefficients 1198861and 1198871of wave-
form of type (35) with maximal absolute value of coefficient1198861le 0 satisfy both relations (123) and (124) if 119886
119896and 119887119896satisfy
1 + 119886119896= 119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) For such waveforms
relations 1205910= 0 and 120585 = minus120575 also hold
Remark 24 Amplitude of 119896th harmonic of nonnegativewaveform of type (35) with maximal absolute value of coeffi-cient 119886
1le 0 and coefficients 119886
119896 119887119896satisfying 1 + 119886
119896=
119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) is
120582119896=
sin (120575119896)119896 sin 120575 cos (120575119896) minus cos 120575 sin (120575119896)
(125)
To show that it is sufficient to substitute 119886119896= 120582119896cos 120575 (see
(118)) into 1 + 119886119896= 119896120582119896[sin 120575 sin(120575119896)] cos(120575119896)
Introducing new variable
119910 = cos(120575119896) (126)
and using the Chebyshev polynomials (eg see Appendices)relations 119886
119896= 120582119896cos 120575 and (125) can be rewritten as
119886119896= 120582119896119881119896(119910) (127)
120582119896=
1
119896119910119880119896minus1
(119910) minus 119881119896(119910)
(128)
where119881119896(119910) and119880
119896(119910) denote the Chebyshev polynomials of
the first and second kind respectively Substitution of (128)into (127) leads to
119886119896119896119910119880119896minus1
(119910) minus (1 + 119886119896) 119881119896(119910) = 0 (129)
which is polynomial equation of 119896th degree in terms of var-iable 119910 From |120575| le 120587 and (126) it follows that
cos(120587119896) le 119910 le 1 (130)
In what follows we show that 119886119896is monotonically increas-
ing function of 119910 on the interval (130) From 120585 = minus120575 (seeRemark 23) and (81) it follows that 120582minus1
119896= (119896 minus 1) cos 120575 +
119896sum119896minus1
119899=1cos((119896 minus 2119899)120575119896) ge 1 and therefore 119886
119896= 120582119896cos 120575 can
be rewritten as
119886119896=
cos 120575(119896 minus 1) cos 120575 + 119896sum119896minus1
119899=1cos ((119896 minus 2119899) 120575119896)
(131)
Obviously 119886119896is even function of 120575 and all cosines in (131)
are monotonically decreasing functions of |120575| on the interval|120575| le 120587 It is easy to show that cos((119896 minus 2119899)120575119896) 119899 =
1 (119896 minus 1) decreases slower than cos 120575 when |120575| increasesThis implies that denominator of the right hand side of(131) decreases slower than numerator Since denominator ispositive for |120575| le 120587 it further implies that 119886
119896is decreasing
function of |120575| on interval |120575| le 120587 Consequently 119886119896is
monotonically increasing function of 119910 on the interval (130)Thus we have shown that 119886
119896is monotonically increasing
function of 119910 on the interval (130) and therefore (129) hasonly one solution that satisfies (130) According to (128) thevalue of 119910 obtained from (129) and (130) either analyticallyor numerically leads to amplitude 120582
119896of 119896th harmonic
16 Mathematical Problems in Engineering
1
05
0
minus05
minus1
minus1 minus05 0 05 1
Coefficient ak
Coe
ffici
entb
k
radica2k+ b2
kle 1
k = 2k = 3k = 4
Figure 11 Plot of (119886119896 119887119896) satisfying 1 + 119886
119896= 119896120582
119896[sin 120575 sin(120575
119896)] cos(120575119896) for 119896 le 4
By solving (129) and (130) for 119896 le 4 we obtain
119910 = radic1 + 1198862
2 (1 minus 1198862) minus1 le 119886
2le1
3
119910 = radic3
4 (1 minus 21198863) minus1 le 119886
3le1
8
119910 =radicradic2 minus 4119886
4+ 1011988624minus 2 (1 minus 119886
4)
4 (1 minus 31198864)
minus1 le 1198864le
1
15
(132)
Insertion of (132) into (128) leads to the following explicitexpressions for the amplitude 120582
119896 119896 le 4
1205822=1
2(1 minus 119886
2) minus1 le 119886
2le1
3 (133)
1205822
3= [
1
3(1 minus 2119886
3)]
3
minus1 le 1198863le1
8 (134)
1205824=1
4(minus1 minus 119886
4+ radic2 minus 4119886
4+ 1011988624) minus1 le 119886
4le
1
15
(135)
Relations (133)ndash(135) define closed lines (see Figure 11) whichseparate points representing waveforms of type (121) frompoints representing waveforms of type (122) For given 119896points inside the corresponding curve refer to nonnegativewaveforms of type (121) whereas points outside curve (andradic1198862119896+ 1198872119896le 1) correspond to nonnegative waveforms of type
(122) Points on the respective curve correspond to the wave-forms which can be expressed in both forms (121) and (122)
Remark 25 Themaximum absolute value of coefficient 1198861of
nonnegative waveform of type (35) is
100381610038161003816100381611988611003816100381610038161003816max =
1
cos (120587 (2119896)) (136)
This maximum value is attained for |120585| = 1205872 and 120575 = 0
(see (124)) Notice that |1198861|max is equal to the maximum value
1205821max of amplitude of fundamental harmonic (see (113))
Coefficients of waveform with maximum absolute value ofcoefficient 119886
1 1198861lt 0 are
1198861= minus
1
cos (120587 (2119896)) 119886
119896=1
119896tan( 120587
(2119896))
1198871= 119887119896= 0
(137)
Waveformdescribed by (137) is cosinewaveformhaving zerosat 120587(2119896) and minus120587(2119896)
In the course of proving (136) notice first that |1198861|max le
1205821max holds According to (123) and (124) maximum of |119886
1|
occurs for 119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) ge 1 + 119886
119896 From (124)
it immediately follows that maximum value of |1198861| is attained
if and only if 1205821= 1205821max and 120575 = 0 which because of
120575119896 = 1205910minus120585119896 further implies 120591
0= 120585119896 Sincemaximumvalue
of 1205821is attained for |120585| = 1205872 it follows that corresponding
waveform has zeros at 120587(2119896) and minus120587(2119896)
Proof of Proposition 22 As it was mentioned earlier in thissection we can assume without loss of generality that 119886
1le 0
We consider waveforms119879119896(120591) of type (35) such that119879
119896(120591) ge 0
and119879119896(120591) = 0 for some 120591
0 Fromassumption that nonnegative
waveform 119879119896(120591) of type (35) has at least one zero it follows
that it can be expressed in form (38)Let us also assume that 120591
0is position of nondegenerate
critical point Therefore 119879119896(1205910) = 0 implies 1198791015840
119896(1205910) = 0 and
11987910158401015840
119896(1205910) gt 0 According to (55) second derivative of 119879
119896(120591) at
1205910can be expressed as 11987910158401015840
119896(1205910) = 1 minus 120582
119896(1198962
minus 1) cos 120585 Since11987910158401015840
119896(1205910) gt 0 it follows immediately that
1 minus 120582119896(1198962
minus 1) cos 120585 gt 0 (138)
Let us further assume that 119879119896(120591) has exactly one zeroThe
problem of finding maximum absolute value of 1198861is con-
nected to the problem of finding maximum of the minimumfunction (see Section 21) If waveforms possess unique globalminimum at nondegenerate critical point then correspond-ing minimum function is a smooth function of parameters[13] Consequently assumption that 119879
119896(120591) has exactly one
zero at nondegenerate critical point leads to the conclusionthat coefficient 119886
1is differentiable function of 120591
0 First
derivative of 1198861(see (43)) with respect to 120591
0 taking into
account that 1205971205851205971205910= 119896 (see (50)) can be expressed in the
following factorized form
1205971198861
1205971205910
= sin 1205910[1 minus 120582
119896(1198962
minus 1) cos 120585] (139)
Mathematical Problems in Engineering 17
From (138) and (139) it is clear that 12059711988611205971205910= 0 if and only if
sin 1205910= 0 According toRemark 12 assumption that119879
119896(120591)has
exactly one zero implies 120582119896lt 1 From (51) (48) and 120582
119896lt 1
it follows that 1198861cos 1205910+ 1198871sin 1205910lt 0 which together with
sin 1205910= 0 implies that 119886
1cos 1205910lt 0 Assumption 119886
1le 0
together with relations 1198861cos 1205910lt 0 and sin 120591
0= 0 further
implies 1198861
= 0 and
1205910= 0 (140)
Insertion of 1205910= 0 into (38) leads to
119879119896(120591)
= [1 minus cos 120591]
sdot [1 minus (119896 minus 1) 120582119896cos 120585 minus 2120582
119896
119896minus1
sum
119899=1
(119896 minus 119899) cos (119899120591 + 120585)]
(141)
Substitution of 1205910= 0 into (45) and (46) yields 119886
119896= 120582119896cos 120585
and 119887119896
= minus120582119896sin 120585 respectively Replacing 120582
119896cos 120585 with
119886119896and 120582
119896cos(119899120591 + 120585) with (119886
119896cos 119899120591 + 119887
119896sin 119899120591) in (141)
immediately leads to (121)Furthermore 119886
119896= 120582119896cos 120585 119887
119896= minus120582
119896sin 120585 and (118)
imply that
120575 = minus120585 (142)
According to (38)ndash(40) and (142) it follows that (141) is non-negative if and only if
120582119896[(119896 minus 1) cos 120575 + 119896 sin (120575 minus 120575119896)
sin (120575119896)] lt 1 (143)
Notice that 119886119896= 120582119896cos 120575 implies that the following relation
holds
120582119896[(119896 minus 1) cos 120575 + 119896 sin (120575 minus 120575119896)
sin (120575119896)]
= minus119886119896+ 119896120582119896
sin 120575sin (120575119896)
cos(120575119896)
(144)
Finally substitution of (144) into (143) leads to 119896120582119896[sin 120575
sin(120575119896)] cos(120575119896) lt 1 + 119886119896 which proves that (121) holds
when 119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) lt 1 + 119886
119896
Apart from nonnegative waveforms with exactly one zeroat nondegenerate critical point in what follows we will alsoconsider other types of nonnegative waveforms with at leastone zero According to Proposition 9 and Remark 11 thesewaveforms can be described by (66)ndash(68) providing that 0 le|120585| le 120587
According to (35) 119879119896(0) ge 0 implies 1 + 119886
1+ 119886119896ge 0
Consequently 1198861le 0 implies that |119886
1| le 1 + 119886
119896 On the other
hand according to (123) |1198861| = 1 + 119886
119896holds for waveforms
of type (121) The converse is also true 1198861le 0 and |119886
1| =
1 + 119886119896imply 119886
1= minus1 minus 119886
119896 which further from (35) implies
119879119896(0) = 0 Therefore in what follows it is enough to consider
only nonnegativewaveformswhich can be described by (66)ndash(68) and 0 le |120585| le 120587 with coefficients 119886
119896and 119887119896satisfying
119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) ge 1 + 119886
119896
For prescribed coefficients 119886119896and 119887119896 the amplitude 120582
119896=
radic1198862119896+ 1198872119896of 119896th harmonic is also prescribed According to
Remark 15 (see also Remark 16) 120582119896is monotonically
decreasing function of 119909 = cos(120585119896) The value of 119909 can beobtained by solving (90) subject to the constraint cos(120587119896) le119909 le 1 Then 120582
1can be determined from (88) From (106) it
immediately follows that maximal absolute value of 1198861le 0
corresponds to 119902 = 0 which from (104) and (120) furtherimplies that
120575 = 1198961205910minus 120585 (145)
Furthermore 119902 = 0 according to (107) implies that waveformzeros are
1205910=(120575 + 120585)
119896 120591
1015840
0= 1205910minus2120585
119896=(120575 minus 120585)
119896 (146)
Substitution of 1205910= (120575 + 120585)119896 into (66) yields (122) which
proves that (122) holds when 119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) ge
1 + 119886119896
In what follows we prove that (121) also holds when119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) = 1 + 119886
119896 Substitution of 119886
119896=
120582119896cos 120575 into 119896120582
119896[sin 120575 sin(120575119896)] cos(120575119896) = 1 + 119886
119896leads to
120582119896[(119896 minus 1) cos 120575 + 119896 sin (120575 minus 120575119896)
sin (120575119896)] = 1 (147)
As we mentioned earlier relation (142) holds for all wave-forms of type (121) Substituting (142) into (147) we obtain
120582119896[(119896 minus 1) cos 120585 + 119896 sin (120585 minus 120585119896)
sin (120585119896)] = 1 (148)
This expression can be rearranged as
120582119896
119896 sin ((119896 minus 1) 120585119896)sin 120585119896
= 1 minus (119896 minus 1) 120582119896cos 120585 (149)
On the other hand for waveforms of type (122) according to(68) relations (148) and (149) also hold Substitution of 120591
0=
(120575 + 120585)119896 (see (145)) and (67) into (122) leads to
119879119896(120591)
= 120582119896[1 minus cos (120591 minus 120591
0)]
sdot [119896 sin ((119896 minus 1) 120585119896)
sin 120585119896minus 2
119896minus1
sum
119899=1
(119896 minus 119899) cos (119899 (120591 minus 1205910) + 120585)]
(150)
Furthermore substitution of (142) into (145) implies that1205910
= 0 Finally substitution of 1205910
= 0 and (149) into(150) leads to (141) Therefore (141) holds when 119896120582
119896[sin 120575
sin(120575119896)] cos(120575119896) = 1 + 119886119896 which in turn shows that (121)
holds when 119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) = 1 + 119886
119896 This
completes the proof
18 Mathematical Problems in Engineering
52 Nonnegative Waveforms with Maximal Absolute Valueof Coefficient 119886
1for 119896 = 3 Nonnegative waveform of type
(35) for 119896 = 3 is widely used in PA design (eg see [10])In this subsection we illustrate results of Section 51 for thisparticular case The case 119896 = 2 is presented in detail in [12]
Coefficients of fundamental harmonic of nonnegativewaveform of type (35) with 119896 = 3 and maximal absolutevalue of coefficient 119886
1le 0 for prescribed coefficients 119886
3and
1198873(1205823= radic11988623+ 11988723) according to (123) (124) (134) (94) and
(120) are equal to
1198861= minus1 minus 119886
3 119887
1= minus3119887
3 (151)
if 12058223le [(1 minus 2119886
3)3]3
1198861= minus1205821cos(120575
3) 119887
1= minus1205821sin(120575
3) (152)
where 1205821= 3(
3radic1205823minus 1205823) and 120575 = atan 2(119887
3 1198863) if [(1 minus
21198863)3]3
le 1205822
3le 1The line 1205822
3= [(1minus2119886
3)3]3 (see case 119896 = 3
in Figure 11) separates points representing waveforms withcoefficients satisfying (151) from points representing wave-forms with coefficients satisfying (152) Waveforms describedby (151) for 1205822
3lt [(1 minus 2119886
3)3]3 have exactly one zero at
1205910= 0 Waveforms described by (151) and (152) for 1205822
3= [(1 minus
21198863)3]3 also have zero at 120591
0= 0 These waveforms as a rule
have exactly two zeros However there are two exceptionsone related to the maximally flat nonnegative waveform withcoefficients 119886
1= minus98 119886
3= 18 and 119887
1= 1198873= 0 which
has only one zero and the other related to the waveform withcoefficients 119886
1= 0 119886
3= minus1 and 119887
1= 1198873= 0 which has three
zerosWaveforms described by (152) for [(1minus21198863)3]3
lt 1205822
3lt
1 have two zeros Waveforms with 1205823= 1 have only third
harmonic (fundamental harmonic is zero)Plot of contours of maximal absolute value of coefficient
1198861 1198861le 0 for prescribed coefficients 119886
3and 1198873is presented
in Figure 12 According to Remark 25 the waveform withmaximum absolute value of 119886
1le 0 is fully described with
the following coefficients 1198861
= minus2radic3 1198863
= radic39 and1198871= 1198873= 0 This waveform has two zeros at plusmn1205876
Two examples of nonnegative waveforms for 119896 = 3
and maximal absolute value of coefficient 1198861 1198861le 0 with
prescribed coefficients 1198863and 1198873are presented in Figure 13
One waveform corresponds to the case 12058223lt [(1 minus 2119886
3)3]3
(solid line) and the other to the case 12058223gt [(1 minus 2119886
3)3]3
(dashed line)Thewaveform represented by solid line has onezero and its coefficients are 119886
3= minus01 119887
3= 01 119886
1= minus09
and 1198871= minus03 Dashed line corresponds to the waveform
having two zeros with coefficients 1198863= minus01 119887
3= 03 119886
1=
minus08844 and 1198871= minus06460 (case 1205822
3gt [(1 minus 2119886
3)3]3)
6 Nonnegative Cosine Waveforms withat Least One Zero
Nonnegative cosine waveforms have proved to be of impor-tance for waveform modelling in PA design (eg see [10])In this section we consider nonnegative cosine waveforms
1
05
0
minus05
minus1
minus1 minus05 0 05 1
Coefficient a3
Coe
ffici
entb
3
02
04
06
08
10
11
Figure 12 Contours ofmaximal absolute value of coefficient 1198861 1198861le
0 as a function of 1198863and 1198873
2
1
0
0 1 2 3 4
Angle 120591120587
Wav
efor
ms
a3 = minus01 b3 = 01
a3 = minus01 b3 = 03
Figure 13 Nonnegative waveforms for 119896 = 3 and maximal absolutevalue of 119886
1 1198861le 0 with prescribed coefficients 119886
3and 1198873
containing fundamental and 119896th harmonic with at least onezero
Cosine waveform with dc component fundamental and119896th harmonic can be obtained from (35) by setting 119887
1= 119887119896=
0 that is
119879119896(120591) = 1 + 119886
1cos 120591 + 119886
119896cos 119896120591 (153)
In Section 61 we provide general description of non-negative cosine waveforms of type (153) with at least onezero We show that nonnegative cosine waveforms with atleast one zero coincide with nonnegative cosine waveformswith maximal absolute value of coefficient 119886
1for prescribed
coefficient 119886119896 In Section 62 we illustrate results of Section 61
for particular case 119896 = 3
61 Nonnegative Cosine Waveforms with at Least One Zerofor 119896 ge 2 Amplitudes of fundamental and 119896th harmonic
Mathematical Problems in Engineering 19
of cosine waveform of type (153) are 1205821= |1198861| and 120582
119896=
|119886119896| respectively According to (42) for nonnegative cosine
waveforms of type (153) the following relation holds
minus1 le 119886119896le 1 (154)
This explains why 119896th harmonic coefficient 119886119896in Proposi-
tion 26 goes through interval [minus1 1]Waveforms (153) with 119886
1ge 0 can be obtained from
waveforms with 1198861le 0 by shifting by 120587 and therefore with-
out loss of generality we can assume that 1198861le 0
Proposition 26 Each nonnegative cosine waveform of type(153) with 119886
1le 0 and at least one zero can be represented as
119879119896(120591) = [1 minus cos 120591] [1 minus (119896 minus 1) 119886
119896minus 2119886119896
119896minus1
sum
119899=1
(119896 minus 119899) cos 119899120591]
(155)
if minus1 le 119886119896le 1(119896
2
minus 1) or
119879119896(120591) = 119886
119896[1 minus cos (120591 minus 120591
0)] [1 minus cos (120591 + 120591
0)]
sdot [1198880+ 2
119896minus2
sum
119899=1
119888119899cos 119899120591]
(156)
where
119888119899=sin ((119896 minus 119899) 120591
0) cos 120591
0minus (119896 minus 119899) cos ((119896 minus 119899) 120591
0) sin 120591
0
sin31205910
(157)
119886119896=
sin 1205910
119896 sin (1198961205910) cos 120591
0minus cos (119896120591
0) sin 120591
0
(158)
100381610038161003816100381612059101003816100381610038161003816 le
120587
119896 (159)
if 1(1198962 minus 1) le 119886119896le 1
Remark 27 Identity [1minuscos(120591minus1205910)][1minuscos(120591+120591
0)] = [cos 120591
0minus
cos 120591]2 implies that (156) can be rewritten as
119879119896(120591) = 119886
119896[cos 120591
0minus cos 120591]2 [119888
0+ 2
119896minus2
sum
119899=1
119888119899cos 119899120591] (160)
Furthermore substitution of (157) into (160) leads to
119879119896(120591) = 119886
119896[cos 120591
0minus cos 120591]
sdot [(119896 minus 1) sin 119896120591
0
sin 1205910
minus 2
119896minus1
sum
119899=1
sin ((119896 minus 119899) 1205910)
sin 1205910
cos 119899120591]
(161)
Remark 28 All nonnegative cosine waveforms of type (153)with at least one zero and 119886
1le 0 except one of them can be
represented either in form (155) or form (156)This exceptionis maximally flat cosine waveform with 119886
1lt 0 which can be
obtained from (155) for 119886119896= 1(119896
2
minus 1) or from (156) for 1205910=
0 Maximally flat cosine waveform with 1198861lt 0 can also be
obtained from (70) by setting 1205910= 0 Furthermore setting
1205910= 0 in (71) leads to maximally flat cosine waveforms for
119896 le 4 and 1198861lt 0
Remark 29 Nonnegative cosine waveform of type (155) with1198861lt 0 and minus1 lt 119886
119896le 1(119896
2
minus 1) has exactly one zero at120591 = 0 Nonnegative cosine waveform described by (156) with1198861lt 0 and 1(1198962 minus 1) lt 119886
119896lt 1 has two zeros at plusmn120591
0 where
0 lt |1205910| lt 120587119896 For 119886
119896= ∓1 nonnegative cosine waveform
of type (153) reduces to 119879119896(120591) = 1 ∓ cos 119896120591 (clearly these two
waveforms both have 119896 zeros)
Remark 30 Transformation of (155) into an additive formleads to the following relation
1198861= minus1 minus 119886
119896 (162)
where minus1 le 119886119896le 1(119896
2
minus1) Similarly transformation of (156)leads to the following relation
1198861= minus119886119896
119896 sin 1198961205910
sin 1205910
(163)
where 119886119896is given by (158) 1(1198962minus1) le 119886
119896le 1 and |120591
0| le 120587119896
Notice that coefficients of maximally flat cosine waveformnamely 119886
119896= 1(119896
2
minus1) and 1198861= minus1198962
(1198962
minus1) satisfy relation(162) They also satisfy relation (163) for 120591
0= 0
Remark 31 Nonnegative cosine waveforms of type (153) withat least one zero coincide with nonnegative cosine waveformswith maximal absolute value of coefficient 119886
1for prescribed
coefficient 119886119896
In proving that Remark 31 holds notice that expression(155) can be obtained from (121) by setting 119887
119896= 0 Fur-
thermore if 119886119896
ge 0 then 120582119896
= 119886119896 which together
with 119887119896
= 0 and (118) implies 120575 = 0 In this case119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) le 1 + 119886
119896becomes 1198962119886
119896le 1 +
119886119896 On the other hand if 119886
119896lt 0 then 120582
119896= minus119886
119896 which
together with 119887119896= 0 and (118) implies |120575| = 120587 In this case
119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) le 1 + 119886
119896becomes 0 le 1 + 119886
119896
Therefore every nonnegative cosine waveform of type (155)has maximal absolute value of coefficient 119886
1for prescribed
coefficient 119886119896 when minus1 le 119886
119896le 1(119896
2
minus 1)Let us now show that expression (156) can be obtained
from (122) by setting 119887119896= 0 and 119886
119896gt 0 For waveforms
of type (122) according to (118) 119887119896= 0 and 119886
119896gt 0 imply
120575 = 0 and 120582119896= 119886119896 Substitution of 120582
119896= 119886119896and 120575 = 0 into
119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) ge 1+119886
119896leads to 119886
119896ge 1(119896
2
minus1)Furthermore substitution of 120575 = 0 into (145) yields 120591
0= 120585119896
Insertion of 120582119896= 119886119896 120575 = 0 and 120591
0= 120585119896 into (122) leads
to (156) Therefore every nonnegative cosine waveform oftype (156) has maximal absolute value of coefficient 119886
1for
prescribed coefficient 119886119896 when 1(1198962 minus 1) le 119886
119896le 1
Proof of Proposition 26 Let us start with nonnegative cosinewaveform of type (153) with 120582
119896= |119886119896| = 1 According to
Remark 7 120582119896= |119886119896| = 1 implies that 120582
1= |1198861| = 0
Substitution of 119886119896
= minus1 into (155) and using (A2) (seeAppendices) lead to 119879
119896(120591) = 1 minus cos 119896120591 Consequently (155)
holds for 119886119896= minus1 On the other hand substitution of 119886
119896= 1
into (158) yields |1205910| = 120587119896 Furthermore substitution of
119886119896= 1 and 120591
0= 120587119896 (or 120591
0= minus120587119896) into (156) along
with performing all multiplications and using (A2) leads to
20 Mathematical Problems in Engineering
119879119896(120591) = 1 + cos 119896120591 Consequently (156)ndash(158) hold for 119886
119896= 1
and |1205910| = 120587119896
It is easy to see that 120582119896= |119886119896| lt 1 and 119879
119896(1205910) = 0 for some
1205910imply 120582
1= |1198861| = 0 Therefore in what follows we assume
that |119886119896| = 1 and 119886
1lt 0
Cosine waveforms are even functions of 120591 Therefore ifnonnegative cosine waveform has exactly one zero it has to beeither at 0 or at 120587 On the other hand if nonnegative cosinewaveform with 119886
1= 0 has exactly two zeros then these zeros
are placed at plusmn1205910 such that 120591
0is neither 0 nor 120587
In order to prove that (155) holds for minus1 lt 119886119896le 1(1minus119896
2
)let us start by referring to the description (38) of nonnegativewaveformswith at least one zero As wementioned earlier fornonnegative cosine waveformwith exactly one zero (denotedby 1205910) it is either 120591
0= 0 or 120591
0= 120587 Therefore in both cases
sin 1205910= 0 Substitution of sin 120591
0= 0 into (43) together with
1198861
= 0 and 120582119896= |119886119896| lt 1 leads to
1205910= 0 (164)
Clearly 1205910= 0 119887
1= 0 and 119887
119896= 0 according to (44) and (46)
imply 120582119896sin 120585 = 0 Since 120582
119896= |119886119896| it follows that |119886
119896| sin 120585 = 0
also holds which further implies 120582119896= 119886119896= 0 or sin 120585 =
0 In the case when 120582119896= 119886119896= 0 from (164) and (43) we
obtain 1198861= minus1 which further implies that 119879
119896(120591) = 1 minus cos 120591
Consequently (155) holds for 119886119896= 0 In the case when sin 120585 =
0 from (164) and (45) we obtain 119886119896= 120582119896if 120585 = 0 or 119886
119896= minus120582119896
if 120585 = 120587 Relations 119886119896= 120582119896and 120585 = 0 according to (40) imply
that 0 le 119886119896le 1(1 minus 119896
2
) Substitution of 120585 = 0 120582119896= 119886119896 and
(164) into (38) leads to (155) which proves that (155) holdsfor 0 le 119886
119896le 1(1 minus 119896
2
) On the other hand relations 119886119896=
minus120582119896and 120585 = 120587 according to (40) imply that minus1 lt 119886
119896le 0
Substitution of 120585 = 120587 120582119896= minus119886119896 and (164) into (38) also leads
to (155) which proves that (155) also holds for minus1 lt 119886119896le 0
Consequently (155) holds for minus1 lt 119886119896le 1(1 minus 119896
2
)In what follows we first prove that (156)-(157) hold for
1(1 minus 1198962
) lt 119886119896lt 1 For this purpose let us start with non-
negative waveforms with two zeros described by (66) As wementioned before nonnegative cosine waveforms with twozeros have zeros at 120591
0and minus120591
0 such that 120591
0= 0 and 120591
0= 120587
Relations 1198861
lt 0 and 1198871
= 0 according to (84) implycos(1205910minus 120585119896) = 1 and therefore
120585
119896= 1205910 (165)
From 120585119896 = 1205910and 0 lt |120585| lt 120587 it follows that 0 lt |120591
0| lt
120587119896 Insertion of 120585119896 = 1205910into (45) yields 119886
119896= 120582119896 Relations
119886119896= 120582119896and (82) imply that 1(1 minus 1198962) lt 119886
119896lt 1 Substitution
of 120582119896= 119886119896and 120585119896 = 120591
0into (66)ndash(68) leads to (156)ndash(158)
which proves that (156)ndash(158) hold for 1(1 minus 1198962
) lt 119886119896lt 1
and 0 lt |1205910| lt 120587119896
Finally substitution of 119886119896= 1(1 minus 119896
2
) and 1205910= 0 into
(161) leads to
119879119896(120591) =
[1 minus cos 120591](1 minus 1198962)
[119896 (119896 minus 1) minus 2
119896minus1
sum
119899=1
(119896 minus 119899) cos 119899120591]
(166)
Waveform (166) coincides with waveform (155) for 119886119896
=
1(1 minus 1198962
) which in turn proves that (156) holds for 119886119896=
1(1 minus 1198962
) and 1205910= 0 This completes the proof
62 Nonnegative CosineWaveforms with at Least One Zero for119896 = 3 In this subsection we consider nonnegative cosinewaveforms with at least one zero for 119896 = 3 (for case 119896 = 2
see [12])Cosine waveform with fundamental and third harmonic
reads
1198793(120591) = 1 + 119886
1cos 120591 + 119886
3cos 3120591 (167)
For 1198861le 0 and minus1 le 119886
3le 18 according to (155) non-
negative cosine waveform of type (167) with at least one zerocan be expressed as
1198793(120591) = (1 minus cos 120591) [1 minus 2119886
3(1 + 2 cos 120591 + cos 2120591)] (168)
From 1198793(120591 + 120587) = 2 minus 119879
3(120591) it immediately follows that for
1198861ge 0 and minus18 le 119886
3le 1 119879
3(120591) can be expressed as
1198793(120591) = (1 + cos 120591) [1 + 2119886
3(1 minus 2 cos 120591 + cos 2120591)] (169)
For 1198861le 0 and 18 le 119886
3le 1 from (158) it follows that 119886
3=
[8cos31205910]minus1 This relation along with (160) and (157) further
implies that 1198793(120591) can be expressed as
1198793(120591) =
[cos 1205910minus cos 120591]2 [2 cos 120591
0+ cos 120591]
2cos31205910
(170)
providing that |1205910| le 1205873 From 119879
3(120591 + 120587) = 2 minus 119879
3(120591) it
follows that (170) also holds for 1198861ge 0 and minus1 le 119886
3le minus18
providing that 1205910isin [21205873 41205873]
Maximally flat nonnegative cosinewaveformof type (167)with 119886
1lt 0 (minimum at 120591
0= 0) reads 119879
3(120591) = [1 minus
cos 120591]2[1 + (12) cos 120591] Dually maximally flat nonnegativecosine waveform with 119886
1gt 0 (minimum at 120591
0= 120587) reads
1198793(120591) = [1 + cos 120591]2[1 minus (12) cos 120591]In what follows we provide relations between coefficients
1198861and 1198863of nonnegative cosine waveforms of type (167) with
at least one zeroFor 1198861le 0 conversion of (168) into an additive form
immediately leads to the following relation
1198861= minus1 minus 119886
3for minus 1 le 119886
3le1
8 (171)
Conversion of (170) into an additive form leads to 1198861
=
minus31198863(2 cos 2120591
0+ 1) which can be also expressed as 119886
1=
minus31198863(4cos2120591
0minus 1) For 119886
1le 0 relations |120591
0| le 1205873 119886
1=
minus31198863(4cos2120591
0minus 1) and 119886
3= [8cos3120591
0]minus1 lead to
1198861= minus3 [ 3radic119886
3minus 1198863] for 1
8le 1198863le 1 (172)
Similarly for 1198861ge 0 conversion of (169) into an additive form
leads to the following relation
1198861= 1 minus 119886
3for minus
1
8le 1198863le 1 (173)
Mathematical Problems in Engineering 21
1
15
15
05
0
minus05
minus15minus15
minus1
minus1 minus05 0 05 1
Coefficient a1
Coe
ffici
enta
3 a1= minus
3(3radica3minus a3)
a1 =
minus1 minus a3
a1 =
1 minusa3
a1 =3(3radic|a3
| +a3)
Figure 14 Parameter space of cosine waveforms for 119896 = 3
For waveform of type (170) with 1198861ge 0 relations 120591
0isin [21205873
41205873] 1198861= minus3119886
3(4cos2120591
0minus 1) and 119886
3= [8cos3120591
0]minus1 lead to
1198861= 3 [
3radic10038161003816100381610038161198863
1003816100381610038161003816 + 1198863] for minus 1 le 1198863le minus
1
8 (174)
Every cosine waveform of type (167) corresponds to apair of real numbers (119886
1 1198863) and vice versa Points (119886
1 1198863)
in grey area in Figure 14 correspond to nonnegative cosinewaveforms for 119896 = 3 The points at the boundary of grey areacorrespond to nonnegative cosinewaveformswith at least onezero A number of shapes of nonnegative cosine waveformswith 119896 = 3 and at least one zero plotted on interval [minus120587 120587]are also presented in Figure 14 The boundary of grey area inFigure 14 consists of four line segments described by relations(171)ndash(174) The common point of line segments (172) and(173) is cusp point with coordinates 119886
1= 0 and 119886
3= 1
Another cusp point with coordinates 1198861= 0 and 119886
3= minus1
is the common point of line segments (171) and (174) Thecommon point of line segments (171)-(172) has coordinates(minus98 18) and common point of line segments (173)-(174)has coordinates (98 minus18) These points are representedby white circle dots and they correspond to maximallyflat cosine waveforms (eg see [21]) White triangle dotswith coordinates (2radic3 minusradic39) and (minus2radic3radic39) refer tothe nonnegative cosine waveforms with maximum value ofamplitude of fundamental harmonic
7 Four Case Studies of Usage of NonnegativeWaveforms in PA Efficiency Analysis
In this section we provide four case studies of usage ofdescription of nonnegative waveforms with fundamental and119896th harmonic in PA efficiency analysis In first two casestudies to be presented in Section 71 voltage is nonnegativewaveform with fundamental and second harmonic with atleast one zero In remaining two case studies to be consideredin Section 72 voltage waveform contains fundamental andthird harmonic
i
v Load
Vdc
Idc
Lch
vL
iL
Cb
+ +in
Figure 15 Generic PA circuit diagram
Let us consider generic PA circuit diagram as shown inFigure 15 We assume here that voltage and current wave-forms at the transistor output are
V (120579) = 1 + 1198861V cos 120579 + 1198871V sin 120579 + 119886119896V cos 119896120579 + 119887119896V sin 119896120579
119894 (120579) = 1 + 1198861119894cos 120579 +
infin
sum
119899=2
119886119899119894cos 119899120579
(175)
where 120579 stands for 120596119905 Both waveforms are normalized in thesense that dc components of voltage and current are 119881dc =
1 and 119868dc = 1 respectively Under assumption that blockingcapacitor 119862
119887behaves as short-circuit at the fundamental and
higher harmonics current and voltage waveforms at the loadare
V119871(120579) = 119886
1V cos 120579 + 1198871V sin 120579 + 119886119896V cos 119896120579 + 119887119896V sin 119896120579
119894119871(120579) = minus119886
1119894cos 120579 minus
infin
sum
119899=2
119886119899119894cos 119899120579
(176)
In terms of coefficients of voltage and current waveforms theload impedance at fundamental harmonic is 119911
1= minus(119886
1V minus
1198951198871V)1198861119894 whereas load impedance at 119896th harmonic is 119911
119896=
minus(119886119896Vminus119895119887119896V)119886119896119894 All other harmonics are short-circuited (119911
119899=
0 for 119899 = 1 and 119899 = 119896) Time average output power of PA (egsee [10]) with waveform pair (175) at fundamental frequencycan be expressed as
1198751= minus
11988611198941198861V
2 (177)
For normalized waveforms (175) with 119881dc = 1 and 119868dc = 1dc power is 119875dc = 1 Consequently PA efficiency 120578 = 119875
1119875dc
(eg see [10 26]) is equal to
120578 = minus11988611198941198861V
2 (178)
Thus time average output power 1198751of PA with pair of nor-
malized waveform (175) is equal to efficiency (178)Power utilization factor (PUF) is defined [26] as ldquothe
ratio of power delivered in a given situation to the power
22 Mathematical Problems in Engineering
delivered by the same device with the same supply voltagein Class A moderdquo Since the output power in class-A modeis 1198751class-A = max[V(120579)] sdotmax[119894(120579)]8 (eg see [9]) it follows
that power utilization factor PUF = 11987511198751class-A for PA with
pair of normalized waveforms (175) can be expressed as
PUF =8120578
max [V (120579)] sdotmax [119894 (120579)] (179)
71 NonnegativeWaveforms for 119896 = 2 in PAEfficiencyAnalysisIn this subsection we provide two case studies of usage ofdescription of nonnegative waveforms with fundamental andsecond harmonic (119896 = 2) in PA efficiency analysis For moreexamples of usage of descriptions of nonnegative waveformswith fundamental and second harmonic in PA efficiencyanalysis see [12]
Case Study 71 In this case study we consider efficiency of PAfor given second harmonic impedance providing that voltageis nonnegative waveform with fundamental and second har-monic and current is ldquohalf-sinerdquo waveform frequently used inefficiency analysis of classical PA operation (eg see [10])
Standard model of current waveform for classical PAoperation has the form (eg see [10 26])
119894119863(120579) =
119868119863[cos 120579 minus cos(120572
2)] |120579| le
120572
2
0120572
2le |120579| le 120587
(180)
where 120572 is conduction angle and 119868119863
gt 0 Since 119894119863(120579) is
even function it immediately follows that its Fourier seriescontains only dc component and cosine terms
119894119863(120579) = 119868dc +
infin
sum
119899=1
119868119899cos 119899120579 (181)
The dc component of the waveform (180) is
119868dc =119868119863120572
2120587[sinc(120572
2) minus cos(120572
2)] (182)
where sinc119909 = (sin119909)119909 The coefficient of the fundamentalharmonic component reads
1198681=119868119863120572
2120587(1 minus sinc120572) (183)
and the coefficient of 119899th harmonic component can bewrittenin the form
119868119899=119868119863
119899120587[sin ((119899 minus 1) 1205722)
(119899 minus 1)minussin ((119899 + 1) 1205722)
(119899 + 1)] 119899 ge 2
(184)
For ldquohalf-sinerdquo current waveform conduction angle is equalto 120587 (class-B conduction angle) According to (182) thisfurther implies that 119868dc = 119868
119863120587 To obtain normalized form of
waveform (180) we set 119868dc = 1which implies that 119868119863= 120587 Fur-
thermore substitution of 120572 = 120587 and 119868119863= 120587 in (180) leads to
119894 (120579) =
120587 cos 120579 |120579| lt120587
2
0120587
2lt |120579| le 120587
(185)
Similarly substitution of 119868119863= 120587 and 120572 = 120587 into (183) and
(184) leads to the coefficients of waveform (185) Coefficientsof fundamental and second harmonic respectively are
1198861119894=120587
2 119886
2119894=2
3 (186)
On the other hand voltage waveform of type (35) for 119896 =2 reads
V (120579) = 1 + 1198861V cos 120579 + 1198871V sin 120579 + 1198862V cos 2120579 + 1198872V sin 2120579
(187)
This waveform contains only fundamental and second har-monic and therefore all harmonics of order higher thantwo are short-circuited (119911
119899= 0 for 119899 gt 2) For current
voltage pair (185) and (187) load impedance at fundamentalharmonic is 119911
1= minus(119886
1V minus 1198951198871V)1198861119894 whereas load impedance
at second harmonic is 1199112= minus(119886
2Vminus1198951198872V)1198862119894 According to ourassumption the load is passive and therefore Re119911
1 gt 0 and
Re1199112 ge 0 which further imply 119886
11198941198861V lt 0 and 119886
21198941198862V le 0
respectivelyIt is easy to see that problem of findingmaximal efficiency
of PA with current-voltage pair (185) and (187) for prescribedsecond harmonic impedance can be reduced to the problemof finding voltage waveform of type (187) with maximal coef-ficient |119886
1V| for prescribed coefficients of second harmonic(see Section 5)
The following algorithm (analogous to Algorithm 22presented in [12]) provides the procedure for calculation ofmaximal efficiency with current-voltage pair (185) and (187)for prescribed second harmonic impedance The definitionof function atan 2(119910 119909) which appears in the step (iii) of thefollowing algorithm is given by (105)
Algorithm 32 (i) Choose 1199112= 1199032+1198951199092such that |119911
2| le 1|119886
2119894|
(ii) calculate 1198862V minus 1198951198872V = minus119911
21198862119894and 120582
2V = radic11988622V + 1198872
2V(iii) if 2120582
2V le 1 minus 1198862V then calculate 119886
1V = minus1 minus 1198862V and
1198871V = minus2119887
2V else calculate 1205821V = radic81205822V(1 minus 1205822V) 1205790V minus 120585V2 =
(12)atan2(1198872V 1198862V) 1198861V = minus120582
1V cos(1205790V minus 120585V2) and 1198871V =
minus1205821V sin(1205790V minus 120585V2)(iv) calculate efficiency 120578 = minus119886
11198941198861V2
(v) calculate 1199111= minus(119886
1V minus 1198951198871V)1198861119894 and 1199112119899 = 1199112Re119911
1
In this case study coefficients of fundamental and secondharmonic of current waveform are given by (186) Maximalefficiency of PA associated with the waveform pair (185)and (187) as a function of normalized second harmonicimpedance 119911
2119899= 1199112Re119911
1 is presented in Figure 16(a)
As can be seen from Figure 16(a) efficiency of 078 isachieved at the edge of Smith chart where second harmonicimpedance has small resistive part Corresponding PUFcalculated according to (179) is presented in Figure 16(b)Peak efficiency 120578 = 1205874 = 07854 and peak value of PUF = 1
are attained when second harmonic is short-circuited (whichcorresponds to ideal class-B operation [10 26])
For example for second harmonic impedance 1199112= 01 minus
11989505 and currentwaveform (185) fromAlgorithm32 it followsthat 2120582
2V le 1 minus 1198862V Furthermore according to step (iii)
of above algorithm maximal efficiency of PA is attained
Mathematical Problems in Engineering 23
+j2
+j5
+j1
+j2
+j5
minusj2
minusj5
minusj1
minusj2
minusj5
infin0
078
02
05
1 2 5
075 07 065
06 05
075
07
065
06
05
120578 lt 05
(a)
099
095
091
083 075
067
0 59 051
051
099
095
091
083
075 0
67 059
+j5
+j1
+j2
+j5
minusj5
minusj1
minusj2
minusj5
infin
+j2
minusj2
0 02
05
1 2 5
120578 lt 05
(b)
Figure 16 (a) Contours of maximal efficiency of PA and (b) contours of corresponding PUF associated with the waveform pair (185) and(187) as functions of normalized second harmonic impedance 119911
2119899= 1199112Re119911
1
2
3
4
1
0
2 3 410
Wav
efor
ms
Angle 120579120587
i(120579)
(120579)
Figure 17 Waveform pair (185) and (187) that provides maximalefficiency for 119911
2= 01 minus 11989505
with voltage waveform of type (187) with coefficients 1198862V =
minus00667 1198872V = minus03333 119886
1V = minus09333 and 1198871V = 06667 (see
Figure 17) Corresponding efficiency PUF and normalizedsecond harmonic impedance are 120578 = 07330 PUF = 07572and 1199112119899= 01683 minus 11989508415 respectively
On the other hand for second harmonic impedance 1199112=
01 minus 11989508 and current waveform (185) from Algorithm 32 itfollows that 2120582
2V gt 1 minus 1198862V Then according to step (iii) of
above algorithm maximal efficiency is attained with voltagewaveform of type (187) with coefficients 119886
2V = minus00667 1198872V =
minus05333 1198861V = minus09333 and 119887
1V = 10572 (see Figure 18)Efficiency PUF andnormalized secondharmonic impedanceare 120578 = 07330 PUF = 06332 and 119911
2119899= 01683 minus 11989513465
respectively
2
3
4
1
0
2 3 410
Wav
efor
ms
Angle 120579120587
i(120579)
(120579)
Figure 18 Waveform pair (185) and (187) that provides maximalefficiency for 119911
2= 01 minus 11989508
Case Study 72 As another case study let us consider the effi-ciency of PA providing that current waveform is nonnegativecosine waveform up to third harmonic with maximum valueof amplitude of fundamental harmonic [22] (see also [8])
119894 (120579) = 1 +1 + radic5
2cos 120579 + 2radic5
5cos 2120579 + 5 minus radic5
10cos 3120579
(188)
and voltage waveform is nonnegative waveform of type(187) Load impedances at fundamental second and thirdharmonic are 119911
1= minus(119886
1Vminus1198951198871V)1198861119894 1199112 = minus(1198862Vminus1198951198872V)1198862119894 and
1199113= 0 respectively According to our assumption the load
is passive and therefore Re1199111 gt 0 and Re119911
2 ge 0 which
further imply 11988611198941198861V lt 0 and 119886
21198941198862V le 0 respectively
Because current waveform (188) contains only cosineterms and voltage waveform is the same as in previous case
24 Mathematical Problems in Engineering
08
075
07
06
06
065
065
05
05
075 0
7
120578 lt 05
+j2
+j5
+j1
+j2
+j5
minusj2
minusj5
minusj1
minusj2
minusj5
infin0
02
05
1 2 5
(a)
085
08 0
75 07 0
6
05 04
08
075
0405
060
7
120578 lt 05
+j2
+j5
+j1
+j2
+j5
minusj2
minusj5
minusj1
minusj2
minusj5
infin0 02
05
1 2 5
(b)
Figure 19 (a) Contours of maximal efficiency of PA and (b) contours of corresponding PUF associated with the waveform pair (187)-(188)as functions of normalized second harmonic impedance 119911
2119899= 1199112Re119911
1
study the procedure for calculation of maximal efficiency ofPA with waveform pair (187)-(188) is the same as presentedin Algorithm 32 In this case study the coefficients of funda-mental and second harmonic of current waveform are 119886
1119894=
(1 + radic5)2 and 1198862119894= 2radic55 respectively
Maximal efficiency of PA associated with the waveformpair (187)-(188) as a function of normalized secondharmonicimpedance 119911
2119899= 1199112Re119911
1 is presented in Figure 19(a)
Efficiency of 08 is achieved at the edge of Smith chart wheresecond harmonic impedance has small resistive part Thetheoretical upper bound 120578 = (1 + radic5)4 asymp 08090 isattained when second harmonic is short-circuitedWhen thisupper bound is reached both second and third harmonic areshort-circuited which implies that we are dealing with finiteharmonic class-C [6 8] or dually when current and voltageinterchange their roles with finite harmonic inverse class-C[6 9] Corresponding PUF calculated according to (179) ispresented in Figure 19(b) Peak value of PUF asymp 08541 isattained when second harmonic is short-circuited
For example for second harmonic impedance 1199112
=
007 minus 11989504 and current waveform (188) from Algorithm 32 itfollows that 2120582
2V le 1 minus 1198862V Furthermore according to step
(iii) of Algorithm 32 maximal efficiency of PA is attainedwith voltage waveform of type (187) with coefficients 119886
2V =
minus00626 1198872V = minus03578 119886
1V = minus09374 and 1198871V = 07155 (see
Figure 20) Corresponding efficiency PUF and normalizedsecond harmonic impedance are 120578 = 07584 PUF = 06337and 1199112119899= 01208 minus 11989506904 respectively
On the other hand for 1199112= 005 minus 11989507 and current
waveform (187) it follows that 21205822V gt 1minus119886
2VThen accordingto step (iii) of Algorithm 32 themaximal efficiency is attainedwith voltage waveform of type (187) with coefficients 119886
2V =
minus00447 1198872V = minus06261 119886
1V = minus09318 and 1198871V = 10007 (see
Figure 21) Efficiency PUF and normalized second harmonic
2
3
4
1
0
2 3 410
Wav
efor
ms
Angle 120579120587
i(120579)
(120579)
Figure 20 Waveform pair (187)-(188) that provides maximal effi-ciency for 119911
2= 007 minus 11989504
impedance are 120578 = 07538 PUF = 05314 and 1199112119899= 00868minus
11989512156 respectively
72 Nonnegative Waveforms for 119896 = 3 in PA EfficiencyAnalysis In this subsection we provide another two casestudies of usage of description of nonnegative waveforms inPA efficiency analysis this time with fundamental and thirdharmonic (119896 = 3)
Case Study 73 Let us consider current-voltage pair such thatvoltage is nonnegative waveform with fundamental and thirdharmonic
V (120579) = 1 + 1198861V cos 120579 + 1198871V sin 120579 + 1198863V cos 3120579 + 1198873V sin 3120579
(189)
Mathematical Problems in Engineering 25
2
3
4
1
0
2 3 410
Wav
efor
ms
Angle 120579120587
i(120579)
(120579)
Figure 21 Waveform pair (187)-(188) that provides maximal effi-ciency for 119911
2= 005 minus 11989507
and current is nonnegative cosine waveform given by (188)Load impedances at fundamental second and third har-monic are 119911
1= minus(119886
1V minus 1198951198871V)1198861119894 1199112 = 0 and 119911
3= minus(119886
3V minus
1198951198873V)1198863119894 respectively According to our assumption the load
is passive and therefore Re1199111 gt 0 and Re119911
3 ge 0 which
further imply 11988611198941198861V lt 0 and 119886
31198941198863V le 0
In this subsection we consider the problem of findingmaximal efficiency of PA with waveform pair (188)-(189)for given third harmonic impedance As we mentionedearlier problem of finding maximal efficiency of PA withcurrent-voltage pair (188)-(189) for prescribed third har-monic impedance can be reduced to the problem of findingvoltage waveform of type (189) withmaximal coefficient |119886
1V|for prescribed coefficients of third harmonic (see Section 52)
The following algorithm provides the procedure forcalculation of maximal efficiency with current-voltage pair(188)-(189) The definition of function atan 2(119910 119909) whichappears in step (iii) of the following algorithm is given by(105)
Algorithm 33 (i) Choose 1199113= 1199033+1198951199093such that |119911
3| le 1|119886
3119894|
(ii) calculate 1198863V minus 1198951198873V = minus119911
31198863119894and 120582
3V = radic11988623V + 1198872
3V(iii) if 271205822
3V le (1 minus 21198863V)3 then calculate 119886
1V = minus1 minus 1198863V
and 1198871V = minus3119887
3V else calculate 1205821V = 3(3radic1205823V minus 120582
3V) 1205790V minus120585V3 = (13)atan 2(119887
3V 1198863V) 1198861V = minus1205821V cos(1205790V minus 120585V3) and
1198871V = minus120582
1V sin(1205790V minus 120585V3)(iv) calculate efficiency 120578 = minus119886
11198941198861V2
(v) calculate 1199111= minus(119886
1V minus 1198951198871V)1198861119894 and 1199113119899 = 1199113Re119911
1
In this case study coefficients of fundamental and thirdharmonic of current waveform are 119886
1119894= (1 + radic5)2 and
1198863119894= (5 minus radic5)10 respectively For the waveform pair (188)-
(189) maximal efficiency of PA as a function of normalizedthird harmonic impedance 119911
3119899= 1199113Re119911
1 is presented in
Figure 22 Efficiency of 08 is reached when third harmonicimpedance has small resistive part Peak efficiency 120578 = (1 +
radic5)4 asymp 08090 is achieved when third harmonic is short-circuited
+j2
+j5
+j1
+j2
+j5
minusj2
minusj5
minusj1
minusj2
minusj5
infin0
08
08
075
075
07
07
06
05
02
05
1 2
Figure 22 Contours of maximal efficiency of PA associated withthe waveform pair (188)-(189) as a function of normalized thirdharmonic impedance 119911
3119899= 1199113Re119911
1
For the present case study in what follows we show thatpower utilization factor is proportional to efficiency For volt-age waveform of type (189) it is easy to see that V(120579 + 120587) =
2minusV(120579) holdsThis relation along with the fact that waveformV(120579) that provides maximal efficiency has at least one zeroimplies that max[V(120579)] = 2 On the other hand currentwaveform (188) is cosine waveform with positive coefficientsand therefore max[119894(120579)] = 119894(0) = 2 + 4radic5 Consequentlyaccording to (179) the following relation holds
PUFCase study 73 = 2 (5 minus 2radic5) 120578 = 10557120578 (190)
Clearly the ratio PUF120578 is constant and therefore in this casestudy PUF can be easily calculated from the correspondingefficiency Accordingly peak efficiency and peak value ofPUFCase study 73 = 3radic52 minus 52 = 08541 are attained forthe same voltage waveform (when third harmonic is short-circuited)
In the first example current waveform (188) and 1199113=
02 minus 11989505 imply that 2712058223V le (1 minus 2119886
3V)3 Then according
to Algorithm 33 the voltage waveform of type (189) thatprovides maximal efficiency has the following coefficients1198863V = minus00553 119887
3V = minus01382 1198861V = minus09447 and 119887
1V =
04146 (see Figure 23) Efficiency PUF and normalized thirdharmonic impedance are 120578 = 07643 PUF = 08069 and1199113119899= 03425 minus 11989508564 respectivelyIn the second example current waveform (188) and 119911
3=
01 minus 11989511 imply that 2712058223V gt (1 minus 2119886
3V)3 Then according
to Algorithm 33 the voltage waveform of type (189) thatprovides maximal efficiency has the following coefficients1198863V = minus00276 119887
3V = minus03040 1198861V = minus09391 and 119887
1V =
05807 (see Figure 24) Efficiency PUF and normalized thirdharmonic impedance are 120578 = 07598 PUF = 08021 and1199113119899= 01723 minus 11989518952 respectively
26 Mathematical Problems in Engineering
2
3
4
1
0
2 3 410
Wav
efor
ms
Angle 120579120587
i(120579)
(120579)
Figure 23 Waveform pair (188)-(189) that provides maximal effi-ciency for 119911
3= 02 minus 11989505
2
3
4
1
0
2 3 410
Wav
efor
ms
Angle 120579120587
i(120579)
(120579)
Figure 24 Waveform pair (188)-(189) that provides maximal effi-ciency for 119911
3= 01 minus 11989511
Case Study 74 In this case study let us consider current-voltage pair where current is normalized waveform of type(180) with conduction angle 120572 = 115120587 (207∘) and voltageis nonnegative waveform of type (189) Substitution of 120572 =
115120587 and 119868dc = 1 into (182) leads to 119868119863
= 22535Furthermore substitution of 120572 = 115120587 and 119868
119863= 22535 into
(180) leads to
119894 (120579)
=
22535 [cos 120579 minus cos(1151205872
)] |120579| le115120587
2
0115120587
2le |120579| le 120587
(191)
Similarly substitution of 120572 = 115120587 and 119868119863
= 22535 into(183) and (184) for 119899 = 3 yields coefficients of fundamentaland third harmonic of waveform (191)
1198861119894= 14586 119886
3119894= minus01026 (192)
Because current waveform (191) contains only cosineterms and voltage waveform is the same as in previous case
+j2
+j5
+j1
+j2
+j5
minusj2
minusj5
minusj1
minusj2
minusj5
infin0 02
05
1 2
074
076
08
084
082
078
076
074
Figure 25 Contours of maximal efficiency of PA associated withthe waveform pair (189) and (191) as a function of normalized thirdharmonic impedance 119911
3119899= 1199113Re119911
1
study the procedure for calculation of maximal efficiencyof PA with waveform pair (189)ndash(191) is the same as thatpresented in Algorithm 33 In this case study the coefficientsof fundamental and third harmonic of current waveform aregiven by (192)
For the waveform pair (189) and (191) maximal efficiencyof PA as a function of normalized third harmonic impedance1199113119899= 1199113Re119911
1 is presented in Figure 25 Efficiency of 084
is obtained in vicinity of 1199113119899= 23685 (corresponding to 119911
3=
18750) Peak efficiency 120578 asymp 08421 is achieved for voltagewaveform of type (189) with coefficients 119886
1V = minus2radic3 1198863V =
radic39 and 1198871V = 1198873V = 0
In the course of finding power utilization factor noticethat currentwaveformof type (191) attains itsmaximumvaluefor 120579 = 0 Insertion of max[119894(120579)] = 119894(0) = 278 andmax[V(120579)] = 2 for voltage waveform of type (189) into (179)leads to
PUFCase study 74 = 1439120578 (193)
Again the ratio PUF120578 is constant and PUF can be easilycalculated from the corresponding efficiency Accordinglypeak value of PUFCase study 74 asymp 12118 and peak efficiencyare attained for the same voltage waveform
In the first example current waveform (191) and 1199113=
1 minus 11989502 imply that 2712058223V le (1 minus 2119886
3V)3 Then according to
Algorithm 33 voltage waveform of type (189) which providesmaximal efficiency has coefficients 119886
3V = 01026 1198873V =
00205 1198861V = minus11026 and 119887
1V = minus00616 (see Figure 26)Efficiency PUF and normalized third harmonic impedanceare 120578 = 08042 PUF = 11572 and 119911
3119899= 13228 minus 11989502646
respectivelyIn second example current waveform (191) and 119911
3=
15 minus 11989512 imply that 2712058223V gt (1 minus 2119886
3V)3 Then according to
Algorithm 33 voltage waveform of type (189) which providesmaximal efficiency has coefficients 119886
3V = 01540 1198873V =
Mathematical Problems in Engineering 27
2
3
4
1
0
2 3 410
Wav
efor
ms
Angle 120579120587
i(120579)
(120579)
Figure 26 Waveform pair (189) and (191) that provides maximalefficiency for 119911
3= 1 minus 11989502
2
3
4
1
0
2 3 410
Wav
efor
ms
Angle 120579120587
i(120579)
(120579)
Figure 27 Waveform pair (189) and (191) that provides maximalefficiency for 119911
3= 15 minus 11989512
01232 1198861V = minus11255 and 119887
1V = minus02575 (see Figure 27)Efficiency PUF and normalized third harmonic impedanceare 120578 = 08208 PUF = 11812 and 119911
3119899= 19439 minus 11989515552
respectively
8 Conclusion
In this paper we consider a problem of finding general de-scriptions of various classes of nonnegative waveforms withfundamental and 119896th harmonic These classes include non-negative waveforms with at least one zero nonnegative wave-forms with maximal amplitude of fundamental harmonic forprescribed amplitude of 119896th harmonic nonnegative wave-forms withmaximal coefficient of cosine part of fundamentalharmonic for prescribed coefficients of 119896th harmonic andnonnegative cosine waveforms with at least one zero Mainresults are stated in six propositions (Propositions 1 6 918 22 and 26) four corollaries (Corollaries 2ndash5) twentyremarks and three algorithms Four case studies of usageof closed form descriptions of nonnegative waveforms in PAefficiency analysis are considered in detail in Section 7
Appendices
Here we provide a list of finite sums of trigonometric func-tions used in this paper (Appendix A) and brief account ofthe Chebyshev polynomials (Appendix B)
A List of Some Finite Sums ofTrigonometric Functions
Dirichlet kernel (eg see [27]) is as follows
119863119896minus1
(120591) = 1 + 2
119896minus1
sum
119899=1
cos 119899120591 = sin ((2119896 minus 1) 1205912)sin (1205912)
(A1)
Fejer kernel (eg see [27]) can be expressed in the fol-lowing equivalent forms
119865119896minus1
(120591) =1
119896
119896minus1
sum
119899=0
119863119899(120591) = 1 +
2
119896
119896minus1
sum
119899=1
(119896 minus 119899) cos 119899120591
=(1 minus cos 119896120591)119896 (1 minus cos 120591)
(A2)
Lagrangersquos trigonometric identity (eg see [28]) is as fol-lows
1198781(120591) =
119896minus1
sum
119899=1
sin 119899120591 = sin (1198961205912) sin ((119896 minus 1) 1205912)sin (1205912)
(A3)
In what follows we show that the following three trigono-metric identities also hold
2
119896minus1
sum
119899=1
(119896 minus 119899) sin 119899120591 = 119896 sin 120591 minus sin 1198961205911 minus cos 120591
(A4)
119896minus1
sum
119899=1
cos (119896 minus 2119899) 120591 = sin (119896 minus 1) 120591sin 120591
(A5)
119896minus1
sum
119899=1
119899 (119896 minus 119899) cos (119896 minus 2119899) 120591
=sin (119896120591) cos 120591 minus 119896 cos (119896120591) sin 120591
2sin3120591
(A6)
Denote 1198782(120591) = 2sum
119896minus1
119899=1(119896 minus 119899) sin 119899120591 119878
3(120591) = sum
119896minus1
119899=1cos(119896 minus
2119899)120591 and 1198784(120591) = sum
119896minus1
119899=1119899(119896 minus 119899) cos(119896 minus 2119899)120591
Notice that 1198782(120591) = 2119896119878
1(120591) + 119889119863
119896minus1(120591)119889120591 which
immediately leads to (A4)
28 Mathematical Problems in Engineering
Identity (A5) can be obtained as follows
sin (119896 minus 1) 120591sin 120591
=119890119895(119896minus1)120591
minus 119890minus119895(119896minus1)120591
119890119895120591 minus 119890minus119895120591
= 119890119895119896120591
119890minus2119895120591
minus 119890minus2119895(119896minus1)120591
1 minus 119890minus2119895120591
= 119890119895119896120591
119896minus1
sum
119899=1
119890minus2119895119899120591
=
119896minus1
sum
119899=1
119890119895(119896minus2119899)120591
=
119896minus1
sum
119899=1
cos (119896 minus 2119899) 120591
(A7)
From 4119899(119896 minus 119899) = 1198962
minus (119896 minus 2119899)2 it follows that 4119878
4(120591) =
1198962
1198783(120591) + 119889
2
1198783(120591)119889120591
2 which leads to (A6)
B The Chebyshev Polynomials
The Chebyshev polynomials of the first kind 119881119899(119909) can be
defined by the following relation (eg see [29])
119881119899(119909) = cos 119899120591 when 119909 = cos 120591 (B1)
The Chebyshev polynomials of the second kind 119880119899(119909) can be
defined by the following relation (eg see [29])
119880119899(119909) =
sin (119899 + 1) 120591sin 120591
when 119909 = cos 120591 (B2)
The Chebyshev polynomials satisfy the following recur-rence relations (eg see [29])
1198810(119909) = 1 119881
1(119909) = 119909
119881119899+1
(119909) = 2119909119881119899(119909) minus 119881
119899minus1(119909)
1198800(119909) = 1 119880
1(119909) = 2119909
119880119899+1
(119909) = 2119909119880119899(119909) minus 119880
119899minus1(119909)
(B3)
The first few Chebyshev polynomials of the first and secondkind are 119881
2(119909) = 2119909
2
minus 1 1198813(119909) = 4119909
3
minus 3119909 1198814(119909) = 8119909
4
minus
81199092
+ 1 1198802(119909) = 4119909
2
minus 1 1198803(119909) = 8119909
3
minus 4119909 and 1198804(119909) =
161199094
minus 121199092
+ 1
Conflict of Interests
The authors declare that there is no conflict of interestsregarding the publication of this paper
Acknowledgment
This work is supported by the SerbianMinistry of EducationScience and Technology Development as a part of ProjectTP32016
References
[1] V I Arnolrsquod V S Afrajmovich Y S Ilrsquoyashenko and L PShilrsquonikov Dynamical Systems V Bifurcation Theory and Catas-trophe Theory Springer Berlin Germany 1994
[2] E Polak ldquoOn the mathematical foundations of nondifferen-tiable optimization in engineering designrdquo SIAM Review vol29 no 1 pp 21ndash89 1987
[3] N S Fuzik ldquoBiharmonic modes of a tuned RF power amplifierrdquoRadiotehnika vol 25 no 7 pp 62ndash71 1970 (Russian)
[4] P Colantonio F Giannini G Leuzzi and E Limiti ldquoClass Gapproach for low-voltage high-efficiency PA designrdquo Interna-tional Journal of RF and Microwave Computer-Aided Engineer-ing vol 10 no 6 pp 366ndash378 2000
[5] F H Raab ldquoMaximum efficiency and output of class-F poweramplifiersrdquo IEEE Transactions on Microwave Theory and Tech-niques vol 49 no 6 pp 1162ndash1166 2001
[6] FH Raab ldquoClass-E class-C and class-F power amplifiers basedupon a finite number of harmonicsrdquo IEEE Transactions onMicrowaveTheory and Techniques vol 49 no 8 pp 1462ndash14682001
[7] J D Rhodes ldquoOutput universality inmaximum efficiency linearpower amplifiersrdquo International Journal of Circuit Theory andApplications vol 31 no 4 pp 385ndash405 2003
[8] A Juhas and L A Novak ldquoComments on lsquoClass-E class-Cand classF power amplifier based upon a finite number ofharmonicsrsquordquo IEEE Transactions on Microwave Theory and Tech-niques vol 57 no 6 pp 1623ndash1625 2009
[9] M Roberg and Z Popovic ldquoAnalysis of high-efficiency poweramplifiers with arbitrary output harmonic terminationsrdquo IEEETransactions on Microwave Theory and Techniques vol 59 no8 pp 2037ndash2048 2011
[10] A Grebennikov N O Sokal and M J Franco Switchmode RFPower Amplifiers ElsevierAcademic Press San Diego CalifUSA 2nd edition 2012
[11] T Canning P J Tasker and S C Cripps ldquoContinuous modepower amplifier design using harmonic clipping contourstheory and practicerdquo IEEE Transactions on Microwave Theoryand Techniques vol 62 no 1 pp 100ndash110 2014
[12] A Juhas and L A Novak ldquoGeneral description of nonnegativewaveforms up to second harmonic for power amplifier mod-ellingrdquoMathematical Problems in Engineering vol 2014 ArticleID 709762 18 pages 2014
[13] V I Arnolrsquod V V Goryunov O V Lyashko and V A VasilrsquoevDynamical Systems VIIImdashSingularity Theory II ApplicationsSpringer Berlin Germany 1993
[14] D Siersma ldquoProperties of conflict sets in the planerdquo BanachCenter Publications Polish Academy of Sciences vol 50 no 1 pp267ndash276 1999 Proceedings of the Banach Center Symposiumon Geometry and Topology of Caustics (Caustics rsquo98) WarsawPoland
[15] M van Manen The geometry of conflict sets [Dissertation]Universiteit Utrecht UtrechtThe Netherlands 2003 httpigi-tur-archivelibraryuunldissertations2003-0912-123058c4pdf
[16] Y L Sachkov ldquoMaxwell strata and symmetries in the problemofoptimal rolling of a sphere over a planerdquo Sbornik Mathematicsvol 201 no 7-8 pp 1029ndash1051 2010
[17] I A Bogaevsky ldquoPerestroikas of shocks and singularities ofminimum functionsrdquoPhysicaDNonlinear Phenomena vol 173no 1-2 pp 1ndash28 2002
[18] Y L Sachkov ldquoMaxwell strata in the Euler elastic problemrdquoJournal of Dynamical andControl Systems vol 14 no 2 pp 169ndash234 2008
[19] M Siino and T Koike ldquoTopological classification of black holesgeneric Maxwell set and crease set of a horizonrdquo InternationalJournal ofModern Physics D Gravitation Astrophysics Cosmol-ogy vol 20 no 6 pp 1095ndash1122 2011
Mathematical Problems in Engineering 29
[20] F H Raab ldquoClass-F power amplifiers with maximally flat wave-formsrdquo IEEETransactions onMicrowaveTheory and Techniquesvol 45 no 11 pp 2007ndash2012 1997
[21] A Juhas and L A Novak ldquoMaximally flat waveforms with finitenumber of harmonics in class-F power amplifiersrdquo Mathemat-ical Problems in Engineering vol 2013 Article ID 169590 9pages 2013
[22] L Fejer ldquoUber trigonometrische polynomerdquo Journal fur dieReine und Angewandte Mathematik vol 1916 no 146 pp 53ndash82 1916 (German)
[23] S C Cripps ldquoBessel Waives [microwave bytes]rdquo IEEE Micro-wave Magazine vol 10 no 7 pp 30ndash36 117 2009
[24] L N Bryzgalova ldquoSingularities of the maximum of para-metrically dependent functionrdquo Functional Analysis and ItsApplications vol 11 no 1 pp 49ndash51 1977
[25] V I Arnold A A Davydov V A Vassiliev and V MZakalyukin Mathematical Models of Catastrophes Control ofCatastrophic Process Encyclopedia of Life Support Systems(EOLSS) EOLSS Publishers Oxford UK 2006
[26] S C Cripps RF Power Amplifiers for Wireless CommunicationsArtech House Norwood Mass USA 2nd edition 2006
[27] A Zygmund Trigonometric Series vol 1 Cambridge UniversityPress Cambridge UK 2nd edition 1959
[28] A Jeffrey and H DaiHandbook of Mathematical Formulas andIntegrals ElsevierAcademic Press San Diego Calif USA 4thedition 2008
[29] J C Mason and D C Handscomb Chebyshev PolynomialsChapman amp Hall CRC Press Boca Raton Fla USA 2003
Submit your manuscripts athttpwwwhindawicom
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical Problems in Engineering
Hindawi Publishing Corporationhttpwwwhindawicom
Differential EquationsInternational Journal of
Volume 2014
Applied MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Probability and StatisticsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical PhysicsAdvances in
Complex AnalysisJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
OptimizationJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
CombinatoricsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Operations ResearchAdvances in
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Function Spaces
Abstract and Applied AnalysisHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of Mathematics and Mathematical Sciences
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Algebra
Discrete Dynamics in Nature and Society
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Decision SciencesAdvances in
Discrete MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom
Volume 2014 Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Stochastic AnalysisInternational Journal of
16 Mathematical Problems in Engineering
1
05
0
minus05
minus1
minus1 minus05 0 05 1
Coefficient ak
Coe
ffici
entb
k
radica2k+ b2
kle 1
k = 2k = 3k = 4
Figure 11 Plot of (119886119896 119887119896) satisfying 1 + 119886
119896= 119896120582
119896[sin 120575 sin(120575
119896)] cos(120575119896) for 119896 le 4
By solving (129) and (130) for 119896 le 4 we obtain
119910 = radic1 + 1198862
2 (1 minus 1198862) minus1 le 119886
2le1
3
119910 = radic3
4 (1 minus 21198863) minus1 le 119886
3le1
8
119910 =radicradic2 minus 4119886
4+ 1011988624minus 2 (1 minus 119886
4)
4 (1 minus 31198864)
minus1 le 1198864le
1
15
(132)
Insertion of (132) into (128) leads to the following explicitexpressions for the amplitude 120582
119896 119896 le 4
1205822=1
2(1 minus 119886
2) minus1 le 119886
2le1
3 (133)
1205822
3= [
1
3(1 minus 2119886
3)]
3
minus1 le 1198863le1
8 (134)
1205824=1
4(minus1 minus 119886
4+ radic2 minus 4119886
4+ 1011988624) minus1 le 119886
4le
1
15
(135)
Relations (133)ndash(135) define closed lines (see Figure 11) whichseparate points representing waveforms of type (121) frompoints representing waveforms of type (122) For given 119896points inside the corresponding curve refer to nonnegativewaveforms of type (121) whereas points outside curve (andradic1198862119896+ 1198872119896le 1) correspond to nonnegative waveforms of type
(122) Points on the respective curve correspond to the wave-forms which can be expressed in both forms (121) and (122)
Remark 25 Themaximum absolute value of coefficient 1198861of
nonnegative waveform of type (35) is
100381610038161003816100381611988611003816100381610038161003816max =
1
cos (120587 (2119896)) (136)
This maximum value is attained for |120585| = 1205872 and 120575 = 0
(see (124)) Notice that |1198861|max is equal to the maximum value
1205821max of amplitude of fundamental harmonic (see (113))
Coefficients of waveform with maximum absolute value ofcoefficient 119886
1 1198861lt 0 are
1198861= minus
1
cos (120587 (2119896)) 119886
119896=1
119896tan( 120587
(2119896))
1198871= 119887119896= 0
(137)
Waveformdescribed by (137) is cosinewaveformhaving zerosat 120587(2119896) and minus120587(2119896)
In the course of proving (136) notice first that |1198861|max le
1205821max holds According to (123) and (124) maximum of |119886
1|
occurs for 119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) ge 1 + 119886
119896 From (124)
it immediately follows that maximum value of |1198861| is attained
if and only if 1205821= 1205821max and 120575 = 0 which because of
120575119896 = 1205910minus120585119896 further implies 120591
0= 120585119896 Sincemaximumvalue
of 1205821is attained for |120585| = 1205872 it follows that corresponding
waveform has zeros at 120587(2119896) and minus120587(2119896)
Proof of Proposition 22 As it was mentioned earlier in thissection we can assume without loss of generality that 119886
1le 0
We consider waveforms119879119896(120591) of type (35) such that119879
119896(120591) ge 0
and119879119896(120591) = 0 for some 120591
0 Fromassumption that nonnegative
waveform 119879119896(120591) of type (35) has at least one zero it follows
that it can be expressed in form (38)Let us also assume that 120591
0is position of nondegenerate
critical point Therefore 119879119896(1205910) = 0 implies 1198791015840
119896(1205910) = 0 and
11987910158401015840
119896(1205910) gt 0 According to (55) second derivative of 119879
119896(120591) at
1205910can be expressed as 11987910158401015840
119896(1205910) = 1 minus 120582
119896(1198962
minus 1) cos 120585 Since11987910158401015840
119896(1205910) gt 0 it follows immediately that
1 minus 120582119896(1198962
minus 1) cos 120585 gt 0 (138)
Let us further assume that 119879119896(120591) has exactly one zeroThe
problem of finding maximum absolute value of 1198861is con-
nected to the problem of finding maximum of the minimumfunction (see Section 21) If waveforms possess unique globalminimum at nondegenerate critical point then correspond-ing minimum function is a smooth function of parameters[13] Consequently assumption that 119879
119896(120591) has exactly one
zero at nondegenerate critical point leads to the conclusionthat coefficient 119886
1is differentiable function of 120591
0 First
derivative of 1198861(see (43)) with respect to 120591
0 taking into
account that 1205971205851205971205910= 119896 (see (50)) can be expressed in the
following factorized form
1205971198861
1205971205910
= sin 1205910[1 minus 120582
119896(1198962
minus 1) cos 120585] (139)
Mathematical Problems in Engineering 17
From (138) and (139) it is clear that 12059711988611205971205910= 0 if and only if
sin 1205910= 0 According toRemark 12 assumption that119879
119896(120591)has
exactly one zero implies 120582119896lt 1 From (51) (48) and 120582
119896lt 1
it follows that 1198861cos 1205910+ 1198871sin 1205910lt 0 which together with
sin 1205910= 0 implies that 119886
1cos 1205910lt 0 Assumption 119886
1le 0
together with relations 1198861cos 1205910lt 0 and sin 120591
0= 0 further
implies 1198861
= 0 and
1205910= 0 (140)
Insertion of 1205910= 0 into (38) leads to
119879119896(120591)
= [1 minus cos 120591]
sdot [1 minus (119896 minus 1) 120582119896cos 120585 minus 2120582
119896
119896minus1
sum
119899=1
(119896 minus 119899) cos (119899120591 + 120585)]
(141)
Substitution of 1205910= 0 into (45) and (46) yields 119886
119896= 120582119896cos 120585
and 119887119896
= minus120582119896sin 120585 respectively Replacing 120582
119896cos 120585 with
119886119896and 120582
119896cos(119899120591 + 120585) with (119886
119896cos 119899120591 + 119887
119896sin 119899120591) in (141)
immediately leads to (121)Furthermore 119886
119896= 120582119896cos 120585 119887
119896= minus120582
119896sin 120585 and (118)
imply that
120575 = minus120585 (142)
According to (38)ndash(40) and (142) it follows that (141) is non-negative if and only if
120582119896[(119896 minus 1) cos 120575 + 119896 sin (120575 minus 120575119896)
sin (120575119896)] lt 1 (143)
Notice that 119886119896= 120582119896cos 120575 implies that the following relation
holds
120582119896[(119896 minus 1) cos 120575 + 119896 sin (120575 minus 120575119896)
sin (120575119896)]
= minus119886119896+ 119896120582119896
sin 120575sin (120575119896)
cos(120575119896)
(144)
Finally substitution of (144) into (143) leads to 119896120582119896[sin 120575
sin(120575119896)] cos(120575119896) lt 1 + 119886119896 which proves that (121) holds
when 119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) lt 1 + 119886
119896
Apart from nonnegative waveforms with exactly one zeroat nondegenerate critical point in what follows we will alsoconsider other types of nonnegative waveforms with at leastone zero According to Proposition 9 and Remark 11 thesewaveforms can be described by (66)ndash(68) providing that 0 le|120585| le 120587
According to (35) 119879119896(0) ge 0 implies 1 + 119886
1+ 119886119896ge 0
Consequently 1198861le 0 implies that |119886
1| le 1 + 119886
119896 On the other
hand according to (123) |1198861| = 1 + 119886
119896holds for waveforms
of type (121) The converse is also true 1198861le 0 and |119886
1| =
1 + 119886119896imply 119886
1= minus1 minus 119886
119896 which further from (35) implies
119879119896(0) = 0 Therefore in what follows it is enough to consider
only nonnegativewaveformswhich can be described by (66)ndash(68) and 0 le |120585| le 120587 with coefficients 119886
119896and 119887119896satisfying
119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) ge 1 + 119886
119896
For prescribed coefficients 119886119896and 119887119896 the amplitude 120582
119896=
radic1198862119896+ 1198872119896of 119896th harmonic is also prescribed According to
Remark 15 (see also Remark 16) 120582119896is monotonically
decreasing function of 119909 = cos(120585119896) The value of 119909 can beobtained by solving (90) subject to the constraint cos(120587119896) le119909 le 1 Then 120582
1can be determined from (88) From (106) it
immediately follows that maximal absolute value of 1198861le 0
corresponds to 119902 = 0 which from (104) and (120) furtherimplies that
120575 = 1198961205910minus 120585 (145)
Furthermore 119902 = 0 according to (107) implies that waveformzeros are
1205910=(120575 + 120585)
119896 120591
1015840
0= 1205910minus2120585
119896=(120575 minus 120585)
119896 (146)
Substitution of 1205910= (120575 + 120585)119896 into (66) yields (122) which
proves that (122) holds when 119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) ge
1 + 119886119896
In what follows we prove that (121) also holds when119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) = 1 + 119886
119896 Substitution of 119886
119896=
120582119896cos 120575 into 119896120582
119896[sin 120575 sin(120575119896)] cos(120575119896) = 1 + 119886
119896leads to
120582119896[(119896 minus 1) cos 120575 + 119896 sin (120575 minus 120575119896)
sin (120575119896)] = 1 (147)
As we mentioned earlier relation (142) holds for all wave-forms of type (121) Substituting (142) into (147) we obtain
120582119896[(119896 minus 1) cos 120585 + 119896 sin (120585 minus 120585119896)
sin (120585119896)] = 1 (148)
This expression can be rearranged as
120582119896
119896 sin ((119896 minus 1) 120585119896)sin 120585119896
= 1 minus (119896 minus 1) 120582119896cos 120585 (149)
On the other hand for waveforms of type (122) according to(68) relations (148) and (149) also hold Substitution of 120591
0=
(120575 + 120585)119896 (see (145)) and (67) into (122) leads to
119879119896(120591)
= 120582119896[1 minus cos (120591 minus 120591
0)]
sdot [119896 sin ((119896 minus 1) 120585119896)
sin 120585119896minus 2
119896minus1
sum
119899=1
(119896 minus 119899) cos (119899 (120591 minus 1205910) + 120585)]
(150)
Furthermore substitution of (142) into (145) implies that1205910
= 0 Finally substitution of 1205910
= 0 and (149) into(150) leads to (141) Therefore (141) holds when 119896120582
119896[sin 120575
sin(120575119896)] cos(120575119896) = 1 + 119886119896 which in turn shows that (121)
holds when 119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) = 1 + 119886
119896 This
completes the proof
18 Mathematical Problems in Engineering
52 Nonnegative Waveforms with Maximal Absolute Valueof Coefficient 119886
1for 119896 = 3 Nonnegative waveform of type
(35) for 119896 = 3 is widely used in PA design (eg see [10])In this subsection we illustrate results of Section 51 for thisparticular case The case 119896 = 2 is presented in detail in [12]
Coefficients of fundamental harmonic of nonnegativewaveform of type (35) with 119896 = 3 and maximal absolutevalue of coefficient 119886
1le 0 for prescribed coefficients 119886
3and
1198873(1205823= radic11988623+ 11988723) according to (123) (124) (134) (94) and
(120) are equal to
1198861= minus1 minus 119886
3 119887
1= minus3119887
3 (151)
if 12058223le [(1 minus 2119886
3)3]3
1198861= minus1205821cos(120575
3) 119887
1= minus1205821sin(120575
3) (152)
where 1205821= 3(
3radic1205823minus 1205823) and 120575 = atan 2(119887
3 1198863) if [(1 minus
21198863)3]3
le 1205822
3le 1The line 1205822
3= [(1minus2119886
3)3]3 (see case 119896 = 3
in Figure 11) separates points representing waveforms withcoefficients satisfying (151) from points representing wave-forms with coefficients satisfying (152) Waveforms describedby (151) for 1205822
3lt [(1 minus 2119886
3)3]3 have exactly one zero at
1205910= 0 Waveforms described by (151) and (152) for 1205822
3= [(1 minus
21198863)3]3 also have zero at 120591
0= 0 These waveforms as a rule
have exactly two zeros However there are two exceptionsone related to the maximally flat nonnegative waveform withcoefficients 119886
1= minus98 119886
3= 18 and 119887
1= 1198873= 0 which
has only one zero and the other related to the waveform withcoefficients 119886
1= 0 119886
3= minus1 and 119887
1= 1198873= 0 which has three
zerosWaveforms described by (152) for [(1minus21198863)3]3
lt 1205822
3lt
1 have two zeros Waveforms with 1205823= 1 have only third
harmonic (fundamental harmonic is zero)Plot of contours of maximal absolute value of coefficient
1198861 1198861le 0 for prescribed coefficients 119886
3and 1198873is presented
in Figure 12 According to Remark 25 the waveform withmaximum absolute value of 119886
1le 0 is fully described with
the following coefficients 1198861
= minus2radic3 1198863
= radic39 and1198871= 1198873= 0 This waveform has two zeros at plusmn1205876
Two examples of nonnegative waveforms for 119896 = 3
and maximal absolute value of coefficient 1198861 1198861le 0 with
prescribed coefficients 1198863and 1198873are presented in Figure 13
One waveform corresponds to the case 12058223lt [(1 minus 2119886
3)3]3
(solid line) and the other to the case 12058223gt [(1 minus 2119886
3)3]3
(dashed line)Thewaveform represented by solid line has onezero and its coefficients are 119886
3= minus01 119887
3= 01 119886
1= minus09
and 1198871= minus03 Dashed line corresponds to the waveform
having two zeros with coefficients 1198863= minus01 119887
3= 03 119886
1=
minus08844 and 1198871= minus06460 (case 1205822
3gt [(1 minus 2119886
3)3]3)
6 Nonnegative Cosine Waveforms withat Least One Zero
Nonnegative cosine waveforms have proved to be of impor-tance for waveform modelling in PA design (eg see [10])In this section we consider nonnegative cosine waveforms
1
05
0
minus05
minus1
minus1 minus05 0 05 1
Coefficient a3
Coe
ffici
entb
3
02
04
06
08
10
11
Figure 12 Contours ofmaximal absolute value of coefficient 1198861 1198861le
0 as a function of 1198863and 1198873
2
1
0
0 1 2 3 4
Angle 120591120587
Wav
efor
ms
a3 = minus01 b3 = 01
a3 = minus01 b3 = 03
Figure 13 Nonnegative waveforms for 119896 = 3 and maximal absolutevalue of 119886
1 1198861le 0 with prescribed coefficients 119886
3and 1198873
containing fundamental and 119896th harmonic with at least onezero
Cosine waveform with dc component fundamental and119896th harmonic can be obtained from (35) by setting 119887
1= 119887119896=
0 that is
119879119896(120591) = 1 + 119886
1cos 120591 + 119886
119896cos 119896120591 (153)
In Section 61 we provide general description of non-negative cosine waveforms of type (153) with at least onezero We show that nonnegative cosine waveforms with atleast one zero coincide with nonnegative cosine waveformswith maximal absolute value of coefficient 119886
1for prescribed
coefficient 119886119896 In Section 62 we illustrate results of Section 61
for particular case 119896 = 3
61 Nonnegative Cosine Waveforms with at Least One Zerofor 119896 ge 2 Amplitudes of fundamental and 119896th harmonic
Mathematical Problems in Engineering 19
of cosine waveform of type (153) are 1205821= |1198861| and 120582
119896=
|119886119896| respectively According to (42) for nonnegative cosine
waveforms of type (153) the following relation holds
minus1 le 119886119896le 1 (154)
This explains why 119896th harmonic coefficient 119886119896in Proposi-
tion 26 goes through interval [minus1 1]Waveforms (153) with 119886
1ge 0 can be obtained from
waveforms with 1198861le 0 by shifting by 120587 and therefore with-
out loss of generality we can assume that 1198861le 0
Proposition 26 Each nonnegative cosine waveform of type(153) with 119886
1le 0 and at least one zero can be represented as
119879119896(120591) = [1 minus cos 120591] [1 minus (119896 minus 1) 119886
119896minus 2119886119896
119896minus1
sum
119899=1
(119896 minus 119899) cos 119899120591]
(155)
if minus1 le 119886119896le 1(119896
2
minus 1) or
119879119896(120591) = 119886
119896[1 minus cos (120591 minus 120591
0)] [1 minus cos (120591 + 120591
0)]
sdot [1198880+ 2
119896minus2
sum
119899=1
119888119899cos 119899120591]
(156)
where
119888119899=sin ((119896 minus 119899) 120591
0) cos 120591
0minus (119896 minus 119899) cos ((119896 minus 119899) 120591
0) sin 120591
0
sin31205910
(157)
119886119896=
sin 1205910
119896 sin (1198961205910) cos 120591
0minus cos (119896120591
0) sin 120591
0
(158)
100381610038161003816100381612059101003816100381610038161003816 le
120587
119896 (159)
if 1(1198962 minus 1) le 119886119896le 1
Remark 27 Identity [1minuscos(120591minus1205910)][1minuscos(120591+120591
0)] = [cos 120591
0minus
cos 120591]2 implies that (156) can be rewritten as
119879119896(120591) = 119886
119896[cos 120591
0minus cos 120591]2 [119888
0+ 2
119896minus2
sum
119899=1
119888119899cos 119899120591] (160)
Furthermore substitution of (157) into (160) leads to
119879119896(120591) = 119886
119896[cos 120591
0minus cos 120591]
sdot [(119896 minus 1) sin 119896120591
0
sin 1205910
minus 2
119896minus1
sum
119899=1
sin ((119896 minus 119899) 1205910)
sin 1205910
cos 119899120591]
(161)
Remark 28 All nonnegative cosine waveforms of type (153)with at least one zero and 119886
1le 0 except one of them can be
represented either in form (155) or form (156)This exceptionis maximally flat cosine waveform with 119886
1lt 0 which can be
obtained from (155) for 119886119896= 1(119896
2
minus 1) or from (156) for 1205910=
0 Maximally flat cosine waveform with 1198861lt 0 can also be
obtained from (70) by setting 1205910= 0 Furthermore setting
1205910= 0 in (71) leads to maximally flat cosine waveforms for
119896 le 4 and 1198861lt 0
Remark 29 Nonnegative cosine waveform of type (155) with1198861lt 0 and minus1 lt 119886
119896le 1(119896
2
minus 1) has exactly one zero at120591 = 0 Nonnegative cosine waveform described by (156) with1198861lt 0 and 1(1198962 minus 1) lt 119886
119896lt 1 has two zeros at plusmn120591
0 where
0 lt |1205910| lt 120587119896 For 119886
119896= ∓1 nonnegative cosine waveform
of type (153) reduces to 119879119896(120591) = 1 ∓ cos 119896120591 (clearly these two
waveforms both have 119896 zeros)
Remark 30 Transformation of (155) into an additive formleads to the following relation
1198861= minus1 minus 119886
119896 (162)
where minus1 le 119886119896le 1(119896
2
minus1) Similarly transformation of (156)leads to the following relation
1198861= minus119886119896
119896 sin 1198961205910
sin 1205910
(163)
where 119886119896is given by (158) 1(1198962minus1) le 119886
119896le 1 and |120591
0| le 120587119896
Notice that coefficients of maximally flat cosine waveformnamely 119886
119896= 1(119896
2
minus1) and 1198861= minus1198962
(1198962
minus1) satisfy relation(162) They also satisfy relation (163) for 120591
0= 0
Remark 31 Nonnegative cosine waveforms of type (153) withat least one zero coincide with nonnegative cosine waveformswith maximal absolute value of coefficient 119886
1for prescribed
coefficient 119886119896
In proving that Remark 31 holds notice that expression(155) can be obtained from (121) by setting 119887
119896= 0 Fur-
thermore if 119886119896
ge 0 then 120582119896
= 119886119896 which together
with 119887119896
= 0 and (118) implies 120575 = 0 In this case119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) le 1 + 119886
119896becomes 1198962119886
119896le 1 +
119886119896 On the other hand if 119886
119896lt 0 then 120582
119896= minus119886
119896 which
together with 119887119896= 0 and (118) implies |120575| = 120587 In this case
119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) le 1 + 119886
119896becomes 0 le 1 + 119886
119896
Therefore every nonnegative cosine waveform of type (155)has maximal absolute value of coefficient 119886
1for prescribed
coefficient 119886119896 when minus1 le 119886
119896le 1(119896
2
minus 1)Let us now show that expression (156) can be obtained
from (122) by setting 119887119896= 0 and 119886
119896gt 0 For waveforms
of type (122) according to (118) 119887119896= 0 and 119886
119896gt 0 imply
120575 = 0 and 120582119896= 119886119896 Substitution of 120582
119896= 119886119896and 120575 = 0 into
119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) ge 1+119886
119896leads to 119886
119896ge 1(119896
2
minus1)Furthermore substitution of 120575 = 0 into (145) yields 120591
0= 120585119896
Insertion of 120582119896= 119886119896 120575 = 0 and 120591
0= 120585119896 into (122) leads
to (156) Therefore every nonnegative cosine waveform oftype (156) has maximal absolute value of coefficient 119886
1for
prescribed coefficient 119886119896 when 1(1198962 minus 1) le 119886
119896le 1
Proof of Proposition 26 Let us start with nonnegative cosinewaveform of type (153) with 120582
119896= |119886119896| = 1 According to
Remark 7 120582119896= |119886119896| = 1 implies that 120582
1= |1198861| = 0
Substitution of 119886119896
= minus1 into (155) and using (A2) (seeAppendices) lead to 119879
119896(120591) = 1 minus cos 119896120591 Consequently (155)
holds for 119886119896= minus1 On the other hand substitution of 119886
119896= 1
into (158) yields |1205910| = 120587119896 Furthermore substitution of
119886119896= 1 and 120591
0= 120587119896 (or 120591
0= minus120587119896) into (156) along
with performing all multiplications and using (A2) leads to
20 Mathematical Problems in Engineering
119879119896(120591) = 1 + cos 119896120591 Consequently (156)ndash(158) hold for 119886
119896= 1
and |1205910| = 120587119896
It is easy to see that 120582119896= |119886119896| lt 1 and 119879
119896(1205910) = 0 for some
1205910imply 120582
1= |1198861| = 0 Therefore in what follows we assume
that |119886119896| = 1 and 119886
1lt 0
Cosine waveforms are even functions of 120591 Therefore ifnonnegative cosine waveform has exactly one zero it has to beeither at 0 or at 120587 On the other hand if nonnegative cosinewaveform with 119886
1= 0 has exactly two zeros then these zeros
are placed at plusmn1205910 such that 120591
0is neither 0 nor 120587
In order to prove that (155) holds for minus1 lt 119886119896le 1(1minus119896
2
)let us start by referring to the description (38) of nonnegativewaveformswith at least one zero As wementioned earlier fornonnegative cosine waveformwith exactly one zero (denotedby 1205910) it is either 120591
0= 0 or 120591
0= 120587 Therefore in both cases
sin 1205910= 0 Substitution of sin 120591
0= 0 into (43) together with
1198861
= 0 and 120582119896= |119886119896| lt 1 leads to
1205910= 0 (164)
Clearly 1205910= 0 119887
1= 0 and 119887
119896= 0 according to (44) and (46)
imply 120582119896sin 120585 = 0 Since 120582
119896= |119886119896| it follows that |119886
119896| sin 120585 = 0
also holds which further implies 120582119896= 119886119896= 0 or sin 120585 =
0 In the case when 120582119896= 119886119896= 0 from (164) and (43) we
obtain 1198861= minus1 which further implies that 119879
119896(120591) = 1 minus cos 120591
Consequently (155) holds for 119886119896= 0 In the case when sin 120585 =
0 from (164) and (45) we obtain 119886119896= 120582119896if 120585 = 0 or 119886
119896= minus120582119896
if 120585 = 120587 Relations 119886119896= 120582119896and 120585 = 0 according to (40) imply
that 0 le 119886119896le 1(1 minus 119896
2
) Substitution of 120585 = 0 120582119896= 119886119896 and
(164) into (38) leads to (155) which proves that (155) holdsfor 0 le 119886
119896le 1(1 minus 119896
2
) On the other hand relations 119886119896=
minus120582119896and 120585 = 120587 according to (40) imply that minus1 lt 119886
119896le 0
Substitution of 120585 = 120587 120582119896= minus119886119896 and (164) into (38) also leads
to (155) which proves that (155) also holds for minus1 lt 119886119896le 0
Consequently (155) holds for minus1 lt 119886119896le 1(1 minus 119896
2
)In what follows we first prove that (156)-(157) hold for
1(1 minus 1198962
) lt 119886119896lt 1 For this purpose let us start with non-
negative waveforms with two zeros described by (66) As wementioned before nonnegative cosine waveforms with twozeros have zeros at 120591
0and minus120591
0 such that 120591
0= 0 and 120591
0= 120587
Relations 1198861
lt 0 and 1198871
= 0 according to (84) implycos(1205910minus 120585119896) = 1 and therefore
120585
119896= 1205910 (165)
From 120585119896 = 1205910and 0 lt |120585| lt 120587 it follows that 0 lt |120591
0| lt
120587119896 Insertion of 120585119896 = 1205910into (45) yields 119886
119896= 120582119896 Relations
119886119896= 120582119896and (82) imply that 1(1 minus 1198962) lt 119886
119896lt 1 Substitution
of 120582119896= 119886119896and 120585119896 = 120591
0into (66)ndash(68) leads to (156)ndash(158)
which proves that (156)ndash(158) hold for 1(1 minus 1198962
) lt 119886119896lt 1
and 0 lt |1205910| lt 120587119896
Finally substitution of 119886119896= 1(1 minus 119896
2
) and 1205910= 0 into
(161) leads to
119879119896(120591) =
[1 minus cos 120591](1 minus 1198962)
[119896 (119896 minus 1) minus 2
119896minus1
sum
119899=1
(119896 minus 119899) cos 119899120591]
(166)
Waveform (166) coincides with waveform (155) for 119886119896
=
1(1 minus 1198962
) which in turn proves that (156) holds for 119886119896=
1(1 minus 1198962
) and 1205910= 0 This completes the proof
62 Nonnegative CosineWaveforms with at Least One Zero for119896 = 3 In this subsection we consider nonnegative cosinewaveforms with at least one zero for 119896 = 3 (for case 119896 = 2
see [12])Cosine waveform with fundamental and third harmonic
reads
1198793(120591) = 1 + 119886
1cos 120591 + 119886
3cos 3120591 (167)
For 1198861le 0 and minus1 le 119886
3le 18 according to (155) non-
negative cosine waveform of type (167) with at least one zerocan be expressed as
1198793(120591) = (1 minus cos 120591) [1 minus 2119886
3(1 + 2 cos 120591 + cos 2120591)] (168)
From 1198793(120591 + 120587) = 2 minus 119879
3(120591) it immediately follows that for
1198861ge 0 and minus18 le 119886
3le 1 119879
3(120591) can be expressed as
1198793(120591) = (1 + cos 120591) [1 + 2119886
3(1 minus 2 cos 120591 + cos 2120591)] (169)
For 1198861le 0 and 18 le 119886
3le 1 from (158) it follows that 119886
3=
[8cos31205910]minus1 This relation along with (160) and (157) further
implies that 1198793(120591) can be expressed as
1198793(120591) =
[cos 1205910minus cos 120591]2 [2 cos 120591
0+ cos 120591]
2cos31205910
(170)
providing that |1205910| le 1205873 From 119879
3(120591 + 120587) = 2 minus 119879
3(120591) it
follows that (170) also holds for 1198861ge 0 and minus1 le 119886
3le minus18
providing that 1205910isin [21205873 41205873]
Maximally flat nonnegative cosinewaveformof type (167)with 119886
1lt 0 (minimum at 120591
0= 0) reads 119879
3(120591) = [1 minus
cos 120591]2[1 + (12) cos 120591] Dually maximally flat nonnegativecosine waveform with 119886
1gt 0 (minimum at 120591
0= 120587) reads
1198793(120591) = [1 + cos 120591]2[1 minus (12) cos 120591]In what follows we provide relations between coefficients
1198861and 1198863of nonnegative cosine waveforms of type (167) with
at least one zeroFor 1198861le 0 conversion of (168) into an additive form
immediately leads to the following relation
1198861= minus1 minus 119886
3for minus 1 le 119886
3le1
8 (171)
Conversion of (170) into an additive form leads to 1198861
=
minus31198863(2 cos 2120591
0+ 1) which can be also expressed as 119886
1=
minus31198863(4cos2120591
0minus 1) For 119886
1le 0 relations |120591
0| le 1205873 119886
1=
minus31198863(4cos2120591
0minus 1) and 119886
3= [8cos3120591
0]minus1 lead to
1198861= minus3 [ 3radic119886
3minus 1198863] for 1
8le 1198863le 1 (172)
Similarly for 1198861ge 0 conversion of (169) into an additive form
leads to the following relation
1198861= 1 minus 119886
3for minus
1
8le 1198863le 1 (173)
Mathematical Problems in Engineering 21
1
15
15
05
0
minus05
minus15minus15
minus1
minus1 minus05 0 05 1
Coefficient a1
Coe
ffici
enta
3 a1= minus
3(3radica3minus a3)
a1 =
minus1 minus a3
a1 =
1 minusa3
a1 =3(3radic|a3
| +a3)
Figure 14 Parameter space of cosine waveforms for 119896 = 3
For waveform of type (170) with 1198861ge 0 relations 120591
0isin [21205873
41205873] 1198861= minus3119886
3(4cos2120591
0minus 1) and 119886
3= [8cos3120591
0]minus1 lead to
1198861= 3 [
3radic10038161003816100381610038161198863
1003816100381610038161003816 + 1198863] for minus 1 le 1198863le minus
1
8 (174)
Every cosine waveform of type (167) corresponds to apair of real numbers (119886
1 1198863) and vice versa Points (119886
1 1198863)
in grey area in Figure 14 correspond to nonnegative cosinewaveforms for 119896 = 3 The points at the boundary of grey areacorrespond to nonnegative cosinewaveformswith at least onezero A number of shapes of nonnegative cosine waveformswith 119896 = 3 and at least one zero plotted on interval [minus120587 120587]are also presented in Figure 14 The boundary of grey area inFigure 14 consists of four line segments described by relations(171)ndash(174) The common point of line segments (172) and(173) is cusp point with coordinates 119886
1= 0 and 119886
3= 1
Another cusp point with coordinates 1198861= 0 and 119886
3= minus1
is the common point of line segments (171) and (174) Thecommon point of line segments (171)-(172) has coordinates(minus98 18) and common point of line segments (173)-(174)has coordinates (98 minus18) These points are representedby white circle dots and they correspond to maximallyflat cosine waveforms (eg see [21]) White triangle dotswith coordinates (2radic3 minusradic39) and (minus2radic3radic39) refer tothe nonnegative cosine waveforms with maximum value ofamplitude of fundamental harmonic
7 Four Case Studies of Usage of NonnegativeWaveforms in PA Efficiency Analysis
In this section we provide four case studies of usage ofdescription of nonnegative waveforms with fundamental and119896th harmonic in PA efficiency analysis In first two casestudies to be presented in Section 71 voltage is nonnegativewaveform with fundamental and second harmonic with atleast one zero In remaining two case studies to be consideredin Section 72 voltage waveform contains fundamental andthird harmonic
i
v Load
Vdc
Idc
Lch
vL
iL
Cb
+ +in
Figure 15 Generic PA circuit diagram
Let us consider generic PA circuit diagram as shown inFigure 15 We assume here that voltage and current wave-forms at the transistor output are
V (120579) = 1 + 1198861V cos 120579 + 1198871V sin 120579 + 119886119896V cos 119896120579 + 119887119896V sin 119896120579
119894 (120579) = 1 + 1198861119894cos 120579 +
infin
sum
119899=2
119886119899119894cos 119899120579
(175)
where 120579 stands for 120596119905 Both waveforms are normalized in thesense that dc components of voltage and current are 119881dc =
1 and 119868dc = 1 respectively Under assumption that blockingcapacitor 119862
119887behaves as short-circuit at the fundamental and
higher harmonics current and voltage waveforms at the loadare
V119871(120579) = 119886
1V cos 120579 + 1198871V sin 120579 + 119886119896V cos 119896120579 + 119887119896V sin 119896120579
119894119871(120579) = minus119886
1119894cos 120579 minus
infin
sum
119899=2
119886119899119894cos 119899120579
(176)
In terms of coefficients of voltage and current waveforms theload impedance at fundamental harmonic is 119911
1= minus(119886
1V minus
1198951198871V)1198861119894 whereas load impedance at 119896th harmonic is 119911
119896=
minus(119886119896Vminus119895119887119896V)119886119896119894 All other harmonics are short-circuited (119911
119899=
0 for 119899 = 1 and 119899 = 119896) Time average output power of PA (egsee [10]) with waveform pair (175) at fundamental frequencycan be expressed as
1198751= minus
11988611198941198861V
2 (177)
For normalized waveforms (175) with 119881dc = 1 and 119868dc = 1dc power is 119875dc = 1 Consequently PA efficiency 120578 = 119875
1119875dc
(eg see [10 26]) is equal to
120578 = minus11988611198941198861V
2 (178)
Thus time average output power 1198751of PA with pair of nor-
malized waveform (175) is equal to efficiency (178)Power utilization factor (PUF) is defined [26] as ldquothe
ratio of power delivered in a given situation to the power
22 Mathematical Problems in Engineering
delivered by the same device with the same supply voltagein Class A moderdquo Since the output power in class-A modeis 1198751class-A = max[V(120579)] sdotmax[119894(120579)]8 (eg see [9]) it follows
that power utilization factor PUF = 11987511198751class-A for PA with
pair of normalized waveforms (175) can be expressed as
PUF =8120578
max [V (120579)] sdotmax [119894 (120579)] (179)
71 NonnegativeWaveforms for 119896 = 2 in PAEfficiencyAnalysisIn this subsection we provide two case studies of usage ofdescription of nonnegative waveforms with fundamental andsecond harmonic (119896 = 2) in PA efficiency analysis For moreexamples of usage of descriptions of nonnegative waveformswith fundamental and second harmonic in PA efficiencyanalysis see [12]
Case Study 71 In this case study we consider efficiency of PAfor given second harmonic impedance providing that voltageis nonnegative waveform with fundamental and second har-monic and current is ldquohalf-sinerdquo waveform frequently used inefficiency analysis of classical PA operation (eg see [10])
Standard model of current waveform for classical PAoperation has the form (eg see [10 26])
119894119863(120579) =
119868119863[cos 120579 minus cos(120572
2)] |120579| le
120572
2
0120572
2le |120579| le 120587
(180)
where 120572 is conduction angle and 119868119863
gt 0 Since 119894119863(120579) is
even function it immediately follows that its Fourier seriescontains only dc component and cosine terms
119894119863(120579) = 119868dc +
infin
sum
119899=1
119868119899cos 119899120579 (181)
The dc component of the waveform (180) is
119868dc =119868119863120572
2120587[sinc(120572
2) minus cos(120572
2)] (182)
where sinc119909 = (sin119909)119909 The coefficient of the fundamentalharmonic component reads
1198681=119868119863120572
2120587(1 minus sinc120572) (183)
and the coefficient of 119899th harmonic component can bewrittenin the form
119868119899=119868119863
119899120587[sin ((119899 minus 1) 1205722)
(119899 minus 1)minussin ((119899 + 1) 1205722)
(119899 + 1)] 119899 ge 2
(184)
For ldquohalf-sinerdquo current waveform conduction angle is equalto 120587 (class-B conduction angle) According to (182) thisfurther implies that 119868dc = 119868
119863120587 To obtain normalized form of
waveform (180) we set 119868dc = 1which implies that 119868119863= 120587 Fur-
thermore substitution of 120572 = 120587 and 119868119863= 120587 in (180) leads to
119894 (120579) =
120587 cos 120579 |120579| lt120587
2
0120587
2lt |120579| le 120587
(185)
Similarly substitution of 119868119863= 120587 and 120572 = 120587 into (183) and
(184) leads to the coefficients of waveform (185) Coefficientsof fundamental and second harmonic respectively are
1198861119894=120587
2 119886
2119894=2
3 (186)
On the other hand voltage waveform of type (35) for 119896 =2 reads
V (120579) = 1 + 1198861V cos 120579 + 1198871V sin 120579 + 1198862V cos 2120579 + 1198872V sin 2120579
(187)
This waveform contains only fundamental and second har-monic and therefore all harmonics of order higher thantwo are short-circuited (119911
119899= 0 for 119899 gt 2) For current
voltage pair (185) and (187) load impedance at fundamentalharmonic is 119911
1= minus(119886
1V minus 1198951198871V)1198861119894 whereas load impedance
at second harmonic is 1199112= minus(119886
2Vminus1198951198872V)1198862119894 According to ourassumption the load is passive and therefore Re119911
1 gt 0 and
Re1199112 ge 0 which further imply 119886
11198941198861V lt 0 and 119886
21198941198862V le 0
respectivelyIt is easy to see that problem of findingmaximal efficiency
of PA with current-voltage pair (185) and (187) for prescribedsecond harmonic impedance can be reduced to the problemof finding voltage waveform of type (187) with maximal coef-ficient |119886
1V| for prescribed coefficients of second harmonic(see Section 5)
The following algorithm (analogous to Algorithm 22presented in [12]) provides the procedure for calculation ofmaximal efficiency with current-voltage pair (185) and (187)for prescribed second harmonic impedance The definitionof function atan 2(119910 119909) which appears in the step (iii) of thefollowing algorithm is given by (105)
Algorithm 32 (i) Choose 1199112= 1199032+1198951199092such that |119911
2| le 1|119886
2119894|
(ii) calculate 1198862V minus 1198951198872V = minus119911
21198862119894and 120582
2V = radic11988622V + 1198872
2V(iii) if 2120582
2V le 1 minus 1198862V then calculate 119886
1V = minus1 minus 1198862V and
1198871V = minus2119887
2V else calculate 1205821V = radic81205822V(1 minus 1205822V) 1205790V minus 120585V2 =
(12)atan2(1198872V 1198862V) 1198861V = minus120582
1V cos(1205790V minus 120585V2) and 1198871V =
minus1205821V sin(1205790V minus 120585V2)(iv) calculate efficiency 120578 = minus119886
11198941198861V2
(v) calculate 1199111= minus(119886
1V minus 1198951198871V)1198861119894 and 1199112119899 = 1199112Re119911
1
In this case study coefficients of fundamental and secondharmonic of current waveform are given by (186) Maximalefficiency of PA associated with the waveform pair (185)and (187) as a function of normalized second harmonicimpedance 119911
2119899= 1199112Re119911
1 is presented in Figure 16(a)
As can be seen from Figure 16(a) efficiency of 078 isachieved at the edge of Smith chart where second harmonicimpedance has small resistive part Corresponding PUFcalculated according to (179) is presented in Figure 16(b)Peak efficiency 120578 = 1205874 = 07854 and peak value of PUF = 1
are attained when second harmonic is short-circuited (whichcorresponds to ideal class-B operation [10 26])
For example for second harmonic impedance 1199112= 01 minus
11989505 and currentwaveform (185) fromAlgorithm32 it followsthat 2120582
2V le 1 minus 1198862V Furthermore according to step (iii)
of above algorithm maximal efficiency of PA is attained
Mathematical Problems in Engineering 23
+j2
+j5
+j1
+j2
+j5
minusj2
minusj5
minusj1
minusj2
minusj5
infin0
078
02
05
1 2 5
075 07 065
06 05
075
07
065
06
05
120578 lt 05
(a)
099
095
091
083 075
067
0 59 051
051
099
095
091
083
075 0
67 059
+j5
+j1
+j2
+j5
minusj5
minusj1
minusj2
minusj5
infin
+j2
minusj2
0 02
05
1 2 5
120578 lt 05
(b)
Figure 16 (a) Contours of maximal efficiency of PA and (b) contours of corresponding PUF associated with the waveform pair (185) and(187) as functions of normalized second harmonic impedance 119911
2119899= 1199112Re119911
1
2
3
4
1
0
2 3 410
Wav
efor
ms
Angle 120579120587
i(120579)
(120579)
Figure 17 Waveform pair (185) and (187) that provides maximalefficiency for 119911
2= 01 minus 11989505
with voltage waveform of type (187) with coefficients 1198862V =
minus00667 1198872V = minus03333 119886
1V = minus09333 and 1198871V = 06667 (see
Figure 17) Corresponding efficiency PUF and normalizedsecond harmonic impedance are 120578 = 07330 PUF = 07572and 1199112119899= 01683 minus 11989508415 respectively
On the other hand for second harmonic impedance 1199112=
01 minus 11989508 and current waveform (185) from Algorithm 32 itfollows that 2120582
2V gt 1 minus 1198862V Then according to step (iii) of
above algorithm maximal efficiency is attained with voltagewaveform of type (187) with coefficients 119886
2V = minus00667 1198872V =
minus05333 1198861V = minus09333 and 119887
1V = 10572 (see Figure 18)Efficiency PUF andnormalized secondharmonic impedanceare 120578 = 07330 PUF = 06332 and 119911
2119899= 01683 minus 11989513465
respectively
2
3
4
1
0
2 3 410
Wav
efor
ms
Angle 120579120587
i(120579)
(120579)
Figure 18 Waveform pair (185) and (187) that provides maximalefficiency for 119911
2= 01 minus 11989508
Case Study 72 As another case study let us consider the effi-ciency of PA providing that current waveform is nonnegativecosine waveform up to third harmonic with maximum valueof amplitude of fundamental harmonic [22] (see also [8])
119894 (120579) = 1 +1 + radic5
2cos 120579 + 2radic5
5cos 2120579 + 5 minus radic5
10cos 3120579
(188)
and voltage waveform is nonnegative waveform of type(187) Load impedances at fundamental second and thirdharmonic are 119911
1= minus(119886
1Vminus1198951198871V)1198861119894 1199112 = minus(1198862Vminus1198951198872V)1198862119894 and
1199113= 0 respectively According to our assumption the load
is passive and therefore Re1199111 gt 0 and Re119911
2 ge 0 which
further imply 11988611198941198861V lt 0 and 119886
21198941198862V le 0 respectively
Because current waveform (188) contains only cosineterms and voltage waveform is the same as in previous case
24 Mathematical Problems in Engineering
08
075
07
06
06
065
065
05
05
075 0
7
120578 lt 05
+j2
+j5
+j1
+j2
+j5
minusj2
minusj5
minusj1
minusj2
minusj5
infin0
02
05
1 2 5
(a)
085
08 0
75 07 0
6
05 04
08
075
0405
060
7
120578 lt 05
+j2
+j5
+j1
+j2
+j5
minusj2
minusj5
minusj1
minusj2
minusj5
infin0 02
05
1 2 5
(b)
Figure 19 (a) Contours of maximal efficiency of PA and (b) contours of corresponding PUF associated with the waveform pair (187)-(188)as functions of normalized second harmonic impedance 119911
2119899= 1199112Re119911
1
study the procedure for calculation of maximal efficiency ofPA with waveform pair (187)-(188) is the same as presentedin Algorithm 32 In this case study the coefficients of funda-mental and second harmonic of current waveform are 119886
1119894=
(1 + radic5)2 and 1198862119894= 2radic55 respectively
Maximal efficiency of PA associated with the waveformpair (187)-(188) as a function of normalized secondharmonicimpedance 119911
2119899= 1199112Re119911
1 is presented in Figure 19(a)
Efficiency of 08 is achieved at the edge of Smith chart wheresecond harmonic impedance has small resistive part Thetheoretical upper bound 120578 = (1 + radic5)4 asymp 08090 isattained when second harmonic is short-circuitedWhen thisupper bound is reached both second and third harmonic areshort-circuited which implies that we are dealing with finiteharmonic class-C [6 8] or dually when current and voltageinterchange their roles with finite harmonic inverse class-C[6 9] Corresponding PUF calculated according to (179) ispresented in Figure 19(b) Peak value of PUF asymp 08541 isattained when second harmonic is short-circuited
For example for second harmonic impedance 1199112
=
007 minus 11989504 and current waveform (188) from Algorithm 32 itfollows that 2120582
2V le 1 minus 1198862V Furthermore according to step
(iii) of Algorithm 32 maximal efficiency of PA is attainedwith voltage waveform of type (187) with coefficients 119886
2V =
minus00626 1198872V = minus03578 119886
1V = minus09374 and 1198871V = 07155 (see
Figure 20) Corresponding efficiency PUF and normalizedsecond harmonic impedance are 120578 = 07584 PUF = 06337and 1199112119899= 01208 minus 11989506904 respectively
On the other hand for 1199112= 005 minus 11989507 and current
waveform (187) it follows that 21205822V gt 1minus119886
2VThen accordingto step (iii) of Algorithm 32 themaximal efficiency is attainedwith voltage waveform of type (187) with coefficients 119886
2V =
minus00447 1198872V = minus06261 119886
1V = minus09318 and 1198871V = 10007 (see
Figure 21) Efficiency PUF and normalized second harmonic
2
3
4
1
0
2 3 410
Wav
efor
ms
Angle 120579120587
i(120579)
(120579)
Figure 20 Waveform pair (187)-(188) that provides maximal effi-ciency for 119911
2= 007 minus 11989504
impedance are 120578 = 07538 PUF = 05314 and 1199112119899= 00868minus
11989512156 respectively
72 Nonnegative Waveforms for 119896 = 3 in PA EfficiencyAnalysis In this subsection we provide another two casestudies of usage of description of nonnegative waveforms inPA efficiency analysis this time with fundamental and thirdharmonic (119896 = 3)
Case Study 73 Let us consider current-voltage pair such thatvoltage is nonnegative waveform with fundamental and thirdharmonic
V (120579) = 1 + 1198861V cos 120579 + 1198871V sin 120579 + 1198863V cos 3120579 + 1198873V sin 3120579
(189)
Mathematical Problems in Engineering 25
2
3
4
1
0
2 3 410
Wav
efor
ms
Angle 120579120587
i(120579)
(120579)
Figure 21 Waveform pair (187)-(188) that provides maximal effi-ciency for 119911
2= 005 minus 11989507
and current is nonnegative cosine waveform given by (188)Load impedances at fundamental second and third har-monic are 119911
1= minus(119886
1V minus 1198951198871V)1198861119894 1199112 = 0 and 119911
3= minus(119886
3V minus
1198951198873V)1198863119894 respectively According to our assumption the load
is passive and therefore Re1199111 gt 0 and Re119911
3 ge 0 which
further imply 11988611198941198861V lt 0 and 119886
31198941198863V le 0
In this subsection we consider the problem of findingmaximal efficiency of PA with waveform pair (188)-(189)for given third harmonic impedance As we mentionedearlier problem of finding maximal efficiency of PA withcurrent-voltage pair (188)-(189) for prescribed third har-monic impedance can be reduced to the problem of findingvoltage waveform of type (189) withmaximal coefficient |119886
1V|for prescribed coefficients of third harmonic (see Section 52)
The following algorithm provides the procedure forcalculation of maximal efficiency with current-voltage pair(188)-(189) The definition of function atan 2(119910 119909) whichappears in step (iii) of the following algorithm is given by(105)
Algorithm 33 (i) Choose 1199113= 1199033+1198951199093such that |119911
3| le 1|119886
3119894|
(ii) calculate 1198863V minus 1198951198873V = minus119911
31198863119894and 120582
3V = radic11988623V + 1198872
3V(iii) if 271205822
3V le (1 minus 21198863V)3 then calculate 119886
1V = minus1 minus 1198863V
and 1198871V = minus3119887
3V else calculate 1205821V = 3(3radic1205823V minus 120582
3V) 1205790V minus120585V3 = (13)atan 2(119887
3V 1198863V) 1198861V = minus1205821V cos(1205790V minus 120585V3) and
1198871V = minus120582
1V sin(1205790V minus 120585V3)(iv) calculate efficiency 120578 = minus119886
11198941198861V2
(v) calculate 1199111= minus(119886
1V minus 1198951198871V)1198861119894 and 1199113119899 = 1199113Re119911
1
In this case study coefficients of fundamental and thirdharmonic of current waveform are 119886
1119894= (1 + radic5)2 and
1198863119894= (5 minus radic5)10 respectively For the waveform pair (188)-
(189) maximal efficiency of PA as a function of normalizedthird harmonic impedance 119911
3119899= 1199113Re119911
1 is presented in
Figure 22 Efficiency of 08 is reached when third harmonicimpedance has small resistive part Peak efficiency 120578 = (1 +
radic5)4 asymp 08090 is achieved when third harmonic is short-circuited
+j2
+j5
+j1
+j2
+j5
minusj2
minusj5
minusj1
minusj2
minusj5
infin0
08
08
075
075
07
07
06
05
02
05
1 2
Figure 22 Contours of maximal efficiency of PA associated withthe waveform pair (188)-(189) as a function of normalized thirdharmonic impedance 119911
3119899= 1199113Re119911
1
For the present case study in what follows we show thatpower utilization factor is proportional to efficiency For volt-age waveform of type (189) it is easy to see that V(120579 + 120587) =
2minusV(120579) holdsThis relation along with the fact that waveformV(120579) that provides maximal efficiency has at least one zeroimplies that max[V(120579)] = 2 On the other hand currentwaveform (188) is cosine waveform with positive coefficientsand therefore max[119894(120579)] = 119894(0) = 2 + 4radic5 Consequentlyaccording to (179) the following relation holds
PUFCase study 73 = 2 (5 minus 2radic5) 120578 = 10557120578 (190)
Clearly the ratio PUF120578 is constant and therefore in this casestudy PUF can be easily calculated from the correspondingefficiency Accordingly peak efficiency and peak value ofPUFCase study 73 = 3radic52 minus 52 = 08541 are attained forthe same voltage waveform (when third harmonic is short-circuited)
In the first example current waveform (188) and 1199113=
02 minus 11989505 imply that 2712058223V le (1 minus 2119886
3V)3 Then according
to Algorithm 33 the voltage waveform of type (189) thatprovides maximal efficiency has the following coefficients1198863V = minus00553 119887
3V = minus01382 1198861V = minus09447 and 119887
1V =
04146 (see Figure 23) Efficiency PUF and normalized thirdharmonic impedance are 120578 = 07643 PUF = 08069 and1199113119899= 03425 minus 11989508564 respectivelyIn the second example current waveform (188) and 119911
3=
01 minus 11989511 imply that 2712058223V gt (1 minus 2119886
3V)3 Then according
to Algorithm 33 the voltage waveform of type (189) thatprovides maximal efficiency has the following coefficients1198863V = minus00276 119887
3V = minus03040 1198861V = minus09391 and 119887
1V =
05807 (see Figure 24) Efficiency PUF and normalized thirdharmonic impedance are 120578 = 07598 PUF = 08021 and1199113119899= 01723 minus 11989518952 respectively
26 Mathematical Problems in Engineering
2
3
4
1
0
2 3 410
Wav
efor
ms
Angle 120579120587
i(120579)
(120579)
Figure 23 Waveform pair (188)-(189) that provides maximal effi-ciency for 119911
3= 02 minus 11989505
2
3
4
1
0
2 3 410
Wav
efor
ms
Angle 120579120587
i(120579)
(120579)
Figure 24 Waveform pair (188)-(189) that provides maximal effi-ciency for 119911
3= 01 minus 11989511
Case Study 74 In this case study let us consider current-voltage pair where current is normalized waveform of type(180) with conduction angle 120572 = 115120587 (207∘) and voltageis nonnegative waveform of type (189) Substitution of 120572 =
115120587 and 119868dc = 1 into (182) leads to 119868119863
= 22535Furthermore substitution of 120572 = 115120587 and 119868
119863= 22535 into
(180) leads to
119894 (120579)
=
22535 [cos 120579 minus cos(1151205872
)] |120579| le115120587
2
0115120587
2le |120579| le 120587
(191)
Similarly substitution of 120572 = 115120587 and 119868119863
= 22535 into(183) and (184) for 119899 = 3 yields coefficients of fundamentaland third harmonic of waveform (191)
1198861119894= 14586 119886
3119894= minus01026 (192)
Because current waveform (191) contains only cosineterms and voltage waveform is the same as in previous case
+j2
+j5
+j1
+j2
+j5
minusj2
minusj5
minusj1
minusj2
minusj5
infin0 02
05
1 2
074
076
08
084
082
078
076
074
Figure 25 Contours of maximal efficiency of PA associated withthe waveform pair (189) and (191) as a function of normalized thirdharmonic impedance 119911
3119899= 1199113Re119911
1
study the procedure for calculation of maximal efficiencyof PA with waveform pair (189)ndash(191) is the same as thatpresented in Algorithm 33 In this case study the coefficientsof fundamental and third harmonic of current waveform aregiven by (192)
For the waveform pair (189) and (191) maximal efficiencyof PA as a function of normalized third harmonic impedance1199113119899= 1199113Re119911
1 is presented in Figure 25 Efficiency of 084
is obtained in vicinity of 1199113119899= 23685 (corresponding to 119911
3=
18750) Peak efficiency 120578 asymp 08421 is achieved for voltagewaveform of type (189) with coefficients 119886
1V = minus2radic3 1198863V =
radic39 and 1198871V = 1198873V = 0
In the course of finding power utilization factor noticethat currentwaveformof type (191) attains itsmaximumvaluefor 120579 = 0 Insertion of max[119894(120579)] = 119894(0) = 278 andmax[V(120579)] = 2 for voltage waveform of type (189) into (179)leads to
PUFCase study 74 = 1439120578 (193)
Again the ratio PUF120578 is constant and PUF can be easilycalculated from the corresponding efficiency Accordinglypeak value of PUFCase study 74 asymp 12118 and peak efficiencyare attained for the same voltage waveform
In the first example current waveform (191) and 1199113=
1 minus 11989502 imply that 2712058223V le (1 minus 2119886
3V)3 Then according to
Algorithm 33 voltage waveform of type (189) which providesmaximal efficiency has coefficients 119886
3V = 01026 1198873V =
00205 1198861V = minus11026 and 119887
1V = minus00616 (see Figure 26)Efficiency PUF and normalized third harmonic impedanceare 120578 = 08042 PUF = 11572 and 119911
3119899= 13228 minus 11989502646
respectivelyIn second example current waveform (191) and 119911
3=
15 minus 11989512 imply that 2712058223V gt (1 minus 2119886
3V)3 Then according to
Algorithm 33 voltage waveform of type (189) which providesmaximal efficiency has coefficients 119886
3V = 01540 1198873V =
Mathematical Problems in Engineering 27
2
3
4
1
0
2 3 410
Wav
efor
ms
Angle 120579120587
i(120579)
(120579)
Figure 26 Waveform pair (189) and (191) that provides maximalefficiency for 119911
3= 1 minus 11989502
2
3
4
1
0
2 3 410
Wav
efor
ms
Angle 120579120587
i(120579)
(120579)
Figure 27 Waveform pair (189) and (191) that provides maximalefficiency for 119911
3= 15 minus 11989512
01232 1198861V = minus11255 and 119887
1V = minus02575 (see Figure 27)Efficiency PUF and normalized third harmonic impedanceare 120578 = 08208 PUF = 11812 and 119911
3119899= 19439 minus 11989515552
respectively
8 Conclusion
In this paper we consider a problem of finding general de-scriptions of various classes of nonnegative waveforms withfundamental and 119896th harmonic These classes include non-negative waveforms with at least one zero nonnegative wave-forms with maximal amplitude of fundamental harmonic forprescribed amplitude of 119896th harmonic nonnegative wave-forms withmaximal coefficient of cosine part of fundamentalharmonic for prescribed coefficients of 119896th harmonic andnonnegative cosine waveforms with at least one zero Mainresults are stated in six propositions (Propositions 1 6 918 22 and 26) four corollaries (Corollaries 2ndash5) twentyremarks and three algorithms Four case studies of usageof closed form descriptions of nonnegative waveforms in PAefficiency analysis are considered in detail in Section 7
Appendices
Here we provide a list of finite sums of trigonometric func-tions used in this paper (Appendix A) and brief account ofthe Chebyshev polynomials (Appendix B)
A List of Some Finite Sums ofTrigonometric Functions
Dirichlet kernel (eg see [27]) is as follows
119863119896minus1
(120591) = 1 + 2
119896minus1
sum
119899=1
cos 119899120591 = sin ((2119896 minus 1) 1205912)sin (1205912)
(A1)
Fejer kernel (eg see [27]) can be expressed in the fol-lowing equivalent forms
119865119896minus1
(120591) =1
119896
119896minus1
sum
119899=0
119863119899(120591) = 1 +
2
119896
119896minus1
sum
119899=1
(119896 minus 119899) cos 119899120591
=(1 minus cos 119896120591)119896 (1 minus cos 120591)
(A2)
Lagrangersquos trigonometric identity (eg see [28]) is as fol-lows
1198781(120591) =
119896minus1
sum
119899=1
sin 119899120591 = sin (1198961205912) sin ((119896 minus 1) 1205912)sin (1205912)
(A3)
In what follows we show that the following three trigono-metric identities also hold
2
119896minus1
sum
119899=1
(119896 minus 119899) sin 119899120591 = 119896 sin 120591 minus sin 1198961205911 minus cos 120591
(A4)
119896minus1
sum
119899=1
cos (119896 minus 2119899) 120591 = sin (119896 minus 1) 120591sin 120591
(A5)
119896minus1
sum
119899=1
119899 (119896 minus 119899) cos (119896 minus 2119899) 120591
=sin (119896120591) cos 120591 minus 119896 cos (119896120591) sin 120591
2sin3120591
(A6)
Denote 1198782(120591) = 2sum
119896minus1
119899=1(119896 minus 119899) sin 119899120591 119878
3(120591) = sum
119896minus1
119899=1cos(119896 minus
2119899)120591 and 1198784(120591) = sum
119896minus1
119899=1119899(119896 minus 119899) cos(119896 minus 2119899)120591
Notice that 1198782(120591) = 2119896119878
1(120591) + 119889119863
119896minus1(120591)119889120591 which
immediately leads to (A4)
28 Mathematical Problems in Engineering
Identity (A5) can be obtained as follows
sin (119896 minus 1) 120591sin 120591
=119890119895(119896minus1)120591
minus 119890minus119895(119896minus1)120591
119890119895120591 minus 119890minus119895120591
= 119890119895119896120591
119890minus2119895120591
minus 119890minus2119895(119896minus1)120591
1 minus 119890minus2119895120591
= 119890119895119896120591
119896minus1
sum
119899=1
119890minus2119895119899120591
=
119896minus1
sum
119899=1
119890119895(119896minus2119899)120591
=
119896minus1
sum
119899=1
cos (119896 minus 2119899) 120591
(A7)
From 4119899(119896 minus 119899) = 1198962
minus (119896 minus 2119899)2 it follows that 4119878
4(120591) =
1198962
1198783(120591) + 119889
2
1198783(120591)119889120591
2 which leads to (A6)
B The Chebyshev Polynomials
The Chebyshev polynomials of the first kind 119881119899(119909) can be
defined by the following relation (eg see [29])
119881119899(119909) = cos 119899120591 when 119909 = cos 120591 (B1)
The Chebyshev polynomials of the second kind 119880119899(119909) can be
defined by the following relation (eg see [29])
119880119899(119909) =
sin (119899 + 1) 120591sin 120591
when 119909 = cos 120591 (B2)
The Chebyshev polynomials satisfy the following recur-rence relations (eg see [29])
1198810(119909) = 1 119881
1(119909) = 119909
119881119899+1
(119909) = 2119909119881119899(119909) minus 119881
119899minus1(119909)
1198800(119909) = 1 119880
1(119909) = 2119909
119880119899+1
(119909) = 2119909119880119899(119909) minus 119880
119899minus1(119909)
(B3)
The first few Chebyshev polynomials of the first and secondkind are 119881
2(119909) = 2119909
2
minus 1 1198813(119909) = 4119909
3
minus 3119909 1198814(119909) = 8119909
4
minus
81199092
+ 1 1198802(119909) = 4119909
2
minus 1 1198803(119909) = 8119909
3
minus 4119909 and 1198804(119909) =
161199094
minus 121199092
+ 1
Conflict of Interests
The authors declare that there is no conflict of interestsregarding the publication of this paper
Acknowledgment
This work is supported by the SerbianMinistry of EducationScience and Technology Development as a part of ProjectTP32016
References
[1] V I Arnolrsquod V S Afrajmovich Y S Ilrsquoyashenko and L PShilrsquonikov Dynamical Systems V Bifurcation Theory and Catas-trophe Theory Springer Berlin Germany 1994
[2] E Polak ldquoOn the mathematical foundations of nondifferen-tiable optimization in engineering designrdquo SIAM Review vol29 no 1 pp 21ndash89 1987
[3] N S Fuzik ldquoBiharmonic modes of a tuned RF power amplifierrdquoRadiotehnika vol 25 no 7 pp 62ndash71 1970 (Russian)
[4] P Colantonio F Giannini G Leuzzi and E Limiti ldquoClass Gapproach for low-voltage high-efficiency PA designrdquo Interna-tional Journal of RF and Microwave Computer-Aided Engineer-ing vol 10 no 6 pp 366ndash378 2000
[5] F H Raab ldquoMaximum efficiency and output of class-F poweramplifiersrdquo IEEE Transactions on Microwave Theory and Tech-niques vol 49 no 6 pp 1162ndash1166 2001
[6] FH Raab ldquoClass-E class-C and class-F power amplifiers basedupon a finite number of harmonicsrdquo IEEE Transactions onMicrowaveTheory and Techniques vol 49 no 8 pp 1462ndash14682001
[7] J D Rhodes ldquoOutput universality inmaximum efficiency linearpower amplifiersrdquo International Journal of Circuit Theory andApplications vol 31 no 4 pp 385ndash405 2003
[8] A Juhas and L A Novak ldquoComments on lsquoClass-E class-Cand classF power amplifier based upon a finite number ofharmonicsrsquordquo IEEE Transactions on Microwave Theory and Tech-niques vol 57 no 6 pp 1623ndash1625 2009
[9] M Roberg and Z Popovic ldquoAnalysis of high-efficiency poweramplifiers with arbitrary output harmonic terminationsrdquo IEEETransactions on Microwave Theory and Techniques vol 59 no8 pp 2037ndash2048 2011
[10] A Grebennikov N O Sokal and M J Franco Switchmode RFPower Amplifiers ElsevierAcademic Press San Diego CalifUSA 2nd edition 2012
[11] T Canning P J Tasker and S C Cripps ldquoContinuous modepower amplifier design using harmonic clipping contourstheory and practicerdquo IEEE Transactions on Microwave Theoryand Techniques vol 62 no 1 pp 100ndash110 2014
[12] A Juhas and L A Novak ldquoGeneral description of nonnegativewaveforms up to second harmonic for power amplifier mod-ellingrdquoMathematical Problems in Engineering vol 2014 ArticleID 709762 18 pages 2014
[13] V I Arnolrsquod V V Goryunov O V Lyashko and V A VasilrsquoevDynamical Systems VIIImdashSingularity Theory II ApplicationsSpringer Berlin Germany 1993
[14] D Siersma ldquoProperties of conflict sets in the planerdquo BanachCenter Publications Polish Academy of Sciences vol 50 no 1 pp267ndash276 1999 Proceedings of the Banach Center Symposiumon Geometry and Topology of Caustics (Caustics rsquo98) WarsawPoland
[15] M van Manen The geometry of conflict sets [Dissertation]Universiteit Utrecht UtrechtThe Netherlands 2003 httpigi-tur-archivelibraryuunldissertations2003-0912-123058c4pdf
[16] Y L Sachkov ldquoMaxwell strata and symmetries in the problemofoptimal rolling of a sphere over a planerdquo Sbornik Mathematicsvol 201 no 7-8 pp 1029ndash1051 2010
[17] I A Bogaevsky ldquoPerestroikas of shocks and singularities ofminimum functionsrdquoPhysicaDNonlinear Phenomena vol 173no 1-2 pp 1ndash28 2002
[18] Y L Sachkov ldquoMaxwell strata in the Euler elastic problemrdquoJournal of Dynamical andControl Systems vol 14 no 2 pp 169ndash234 2008
[19] M Siino and T Koike ldquoTopological classification of black holesgeneric Maxwell set and crease set of a horizonrdquo InternationalJournal ofModern Physics D Gravitation Astrophysics Cosmol-ogy vol 20 no 6 pp 1095ndash1122 2011
Mathematical Problems in Engineering 29
[20] F H Raab ldquoClass-F power amplifiers with maximally flat wave-formsrdquo IEEETransactions onMicrowaveTheory and Techniquesvol 45 no 11 pp 2007ndash2012 1997
[21] A Juhas and L A Novak ldquoMaximally flat waveforms with finitenumber of harmonics in class-F power amplifiersrdquo Mathemat-ical Problems in Engineering vol 2013 Article ID 169590 9pages 2013
[22] L Fejer ldquoUber trigonometrische polynomerdquo Journal fur dieReine und Angewandte Mathematik vol 1916 no 146 pp 53ndash82 1916 (German)
[23] S C Cripps ldquoBessel Waives [microwave bytes]rdquo IEEE Micro-wave Magazine vol 10 no 7 pp 30ndash36 117 2009
[24] L N Bryzgalova ldquoSingularities of the maximum of para-metrically dependent functionrdquo Functional Analysis and ItsApplications vol 11 no 1 pp 49ndash51 1977
[25] V I Arnold A A Davydov V A Vassiliev and V MZakalyukin Mathematical Models of Catastrophes Control ofCatastrophic Process Encyclopedia of Life Support Systems(EOLSS) EOLSS Publishers Oxford UK 2006
[26] S C Cripps RF Power Amplifiers for Wireless CommunicationsArtech House Norwood Mass USA 2nd edition 2006
[27] A Zygmund Trigonometric Series vol 1 Cambridge UniversityPress Cambridge UK 2nd edition 1959
[28] A Jeffrey and H DaiHandbook of Mathematical Formulas andIntegrals ElsevierAcademic Press San Diego Calif USA 4thedition 2008
[29] J C Mason and D C Handscomb Chebyshev PolynomialsChapman amp Hall CRC Press Boca Raton Fla USA 2003
Submit your manuscripts athttpwwwhindawicom
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical Problems in Engineering
Hindawi Publishing Corporationhttpwwwhindawicom
Differential EquationsInternational Journal of
Volume 2014
Applied MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Probability and StatisticsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical PhysicsAdvances in
Complex AnalysisJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
OptimizationJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
CombinatoricsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Operations ResearchAdvances in
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Function Spaces
Abstract and Applied AnalysisHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of Mathematics and Mathematical Sciences
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Algebra
Discrete Dynamics in Nature and Society
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Decision SciencesAdvances in
Discrete MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom
Volume 2014 Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Stochastic AnalysisInternational Journal of
Mathematical Problems in Engineering 17
From (138) and (139) it is clear that 12059711988611205971205910= 0 if and only if
sin 1205910= 0 According toRemark 12 assumption that119879
119896(120591)has
exactly one zero implies 120582119896lt 1 From (51) (48) and 120582
119896lt 1
it follows that 1198861cos 1205910+ 1198871sin 1205910lt 0 which together with
sin 1205910= 0 implies that 119886
1cos 1205910lt 0 Assumption 119886
1le 0
together with relations 1198861cos 1205910lt 0 and sin 120591
0= 0 further
implies 1198861
= 0 and
1205910= 0 (140)
Insertion of 1205910= 0 into (38) leads to
119879119896(120591)
= [1 minus cos 120591]
sdot [1 minus (119896 minus 1) 120582119896cos 120585 minus 2120582
119896
119896minus1
sum
119899=1
(119896 minus 119899) cos (119899120591 + 120585)]
(141)
Substitution of 1205910= 0 into (45) and (46) yields 119886
119896= 120582119896cos 120585
and 119887119896
= minus120582119896sin 120585 respectively Replacing 120582
119896cos 120585 with
119886119896and 120582
119896cos(119899120591 + 120585) with (119886
119896cos 119899120591 + 119887
119896sin 119899120591) in (141)
immediately leads to (121)Furthermore 119886
119896= 120582119896cos 120585 119887
119896= minus120582
119896sin 120585 and (118)
imply that
120575 = minus120585 (142)
According to (38)ndash(40) and (142) it follows that (141) is non-negative if and only if
120582119896[(119896 minus 1) cos 120575 + 119896 sin (120575 minus 120575119896)
sin (120575119896)] lt 1 (143)
Notice that 119886119896= 120582119896cos 120575 implies that the following relation
holds
120582119896[(119896 minus 1) cos 120575 + 119896 sin (120575 minus 120575119896)
sin (120575119896)]
= minus119886119896+ 119896120582119896
sin 120575sin (120575119896)
cos(120575119896)
(144)
Finally substitution of (144) into (143) leads to 119896120582119896[sin 120575
sin(120575119896)] cos(120575119896) lt 1 + 119886119896 which proves that (121) holds
when 119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) lt 1 + 119886
119896
Apart from nonnegative waveforms with exactly one zeroat nondegenerate critical point in what follows we will alsoconsider other types of nonnegative waveforms with at leastone zero According to Proposition 9 and Remark 11 thesewaveforms can be described by (66)ndash(68) providing that 0 le|120585| le 120587
According to (35) 119879119896(0) ge 0 implies 1 + 119886
1+ 119886119896ge 0
Consequently 1198861le 0 implies that |119886
1| le 1 + 119886
119896 On the other
hand according to (123) |1198861| = 1 + 119886
119896holds for waveforms
of type (121) The converse is also true 1198861le 0 and |119886
1| =
1 + 119886119896imply 119886
1= minus1 minus 119886
119896 which further from (35) implies
119879119896(0) = 0 Therefore in what follows it is enough to consider
only nonnegativewaveformswhich can be described by (66)ndash(68) and 0 le |120585| le 120587 with coefficients 119886
119896and 119887119896satisfying
119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) ge 1 + 119886
119896
For prescribed coefficients 119886119896and 119887119896 the amplitude 120582
119896=
radic1198862119896+ 1198872119896of 119896th harmonic is also prescribed According to
Remark 15 (see also Remark 16) 120582119896is monotonically
decreasing function of 119909 = cos(120585119896) The value of 119909 can beobtained by solving (90) subject to the constraint cos(120587119896) le119909 le 1 Then 120582
1can be determined from (88) From (106) it
immediately follows that maximal absolute value of 1198861le 0
corresponds to 119902 = 0 which from (104) and (120) furtherimplies that
120575 = 1198961205910minus 120585 (145)
Furthermore 119902 = 0 according to (107) implies that waveformzeros are
1205910=(120575 + 120585)
119896 120591
1015840
0= 1205910minus2120585
119896=(120575 minus 120585)
119896 (146)
Substitution of 1205910= (120575 + 120585)119896 into (66) yields (122) which
proves that (122) holds when 119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) ge
1 + 119886119896
In what follows we prove that (121) also holds when119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) = 1 + 119886
119896 Substitution of 119886
119896=
120582119896cos 120575 into 119896120582
119896[sin 120575 sin(120575119896)] cos(120575119896) = 1 + 119886
119896leads to
120582119896[(119896 minus 1) cos 120575 + 119896 sin (120575 minus 120575119896)
sin (120575119896)] = 1 (147)
As we mentioned earlier relation (142) holds for all wave-forms of type (121) Substituting (142) into (147) we obtain
120582119896[(119896 minus 1) cos 120585 + 119896 sin (120585 minus 120585119896)
sin (120585119896)] = 1 (148)
This expression can be rearranged as
120582119896
119896 sin ((119896 minus 1) 120585119896)sin 120585119896
= 1 minus (119896 minus 1) 120582119896cos 120585 (149)
On the other hand for waveforms of type (122) according to(68) relations (148) and (149) also hold Substitution of 120591
0=
(120575 + 120585)119896 (see (145)) and (67) into (122) leads to
119879119896(120591)
= 120582119896[1 minus cos (120591 minus 120591
0)]
sdot [119896 sin ((119896 minus 1) 120585119896)
sin 120585119896minus 2
119896minus1
sum
119899=1
(119896 minus 119899) cos (119899 (120591 minus 1205910) + 120585)]
(150)
Furthermore substitution of (142) into (145) implies that1205910
= 0 Finally substitution of 1205910
= 0 and (149) into(150) leads to (141) Therefore (141) holds when 119896120582
119896[sin 120575
sin(120575119896)] cos(120575119896) = 1 + 119886119896 which in turn shows that (121)
holds when 119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) = 1 + 119886
119896 This
completes the proof
18 Mathematical Problems in Engineering
52 Nonnegative Waveforms with Maximal Absolute Valueof Coefficient 119886
1for 119896 = 3 Nonnegative waveform of type
(35) for 119896 = 3 is widely used in PA design (eg see [10])In this subsection we illustrate results of Section 51 for thisparticular case The case 119896 = 2 is presented in detail in [12]
Coefficients of fundamental harmonic of nonnegativewaveform of type (35) with 119896 = 3 and maximal absolutevalue of coefficient 119886
1le 0 for prescribed coefficients 119886
3and
1198873(1205823= radic11988623+ 11988723) according to (123) (124) (134) (94) and
(120) are equal to
1198861= minus1 minus 119886
3 119887
1= minus3119887
3 (151)
if 12058223le [(1 minus 2119886
3)3]3
1198861= minus1205821cos(120575
3) 119887
1= minus1205821sin(120575
3) (152)
where 1205821= 3(
3radic1205823minus 1205823) and 120575 = atan 2(119887
3 1198863) if [(1 minus
21198863)3]3
le 1205822
3le 1The line 1205822
3= [(1minus2119886
3)3]3 (see case 119896 = 3
in Figure 11) separates points representing waveforms withcoefficients satisfying (151) from points representing wave-forms with coefficients satisfying (152) Waveforms describedby (151) for 1205822
3lt [(1 minus 2119886
3)3]3 have exactly one zero at
1205910= 0 Waveforms described by (151) and (152) for 1205822
3= [(1 minus
21198863)3]3 also have zero at 120591
0= 0 These waveforms as a rule
have exactly two zeros However there are two exceptionsone related to the maximally flat nonnegative waveform withcoefficients 119886
1= minus98 119886
3= 18 and 119887
1= 1198873= 0 which
has only one zero and the other related to the waveform withcoefficients 119886
1= 0 119886
3= minus1 and 119887
1= 1198873= 0 which has three
zerosWaveforms described by (152) for [(1minus21198863)3]3
lt 1205822
3lt
1 have two zeros Waveforms with 1205823= 1 have only third
harmonic (fundamental harmonic is zero)Plot of contours of maximal absolute value of coefficient
1198861 1198861le 0 for prescribed coefficients 119886
3and 1198873is presented
in Figure 12 According to Remark 25 the waveform withmaximum absolute value of 119886
1le 0 is fully described with
the following coefficients 1198861
= minus2radic3 1198863
= radic39 and1198871= 1198873= 0 This waveform has two zeros at plusmn1205876
Two examples of nonnegative waveforms for 119896 = 3
and maximal absolute value of coefficient 1198861 1198861le 0 with
prescribed coefficients 1198863and 1198873are presented in Figure 13
One waveform corresponds to the case 12058223lt [(1 minus 2119886
3)3]3
(solid line) and the other to the case 12058223gt [(1 minus 2119886
3)3]3
(dashed line)Thewaveform represented by solid line has onezero and its coefficients are 119886
3= minus01 119887
3= 01 119886
1= minus09
and 1198871= minus03 Dashed line corresponds to the waveform
having two zeros with coefficients 1198863= minus01 119887
3= 03 119886
1=
minus08844 and 1198871= minus06460 (case 1205822
3gt [(1 minus 2119886
3)3]3)
6 Nonnegative Cosine Waveforms withat Least One Zero
Nonnegative cosine waveforms have proved to be of impor-tance for waveform modelling in PA design (eg see [10])In this section we consider nonnegative cosine waveforms
1
05
0
minus05
minus1
minus1 minus05 0 05 1
Coefficient a3
Coe
ffici
entb
3
02
04
06
08
10
11
Figure 12 Contours ofmaximal absolute value of coefficient 1198861 1198861le
0 as a function of 1198863and 1198873
2
1
0
0 1 2 3 4
Angle 120591120587
Wav
efor
ms
a3 = minus01 b3 = 01
a3 = minus01 b3 = 03
Figure 13 Nonnegative waveforms for 119896 = 3 and maximal absolutevalue of 119886
1 1198861le 0 with prescribed coefficients 119886
3and 1198873
containing fundamental and 119896th harmonic with at least onezero
Cosine waveform with dc component fundamental and119896th harmonic can be obtained from (35) by setting 119887
1= 119887119896=
0 that is
119879119896(120591) = 1 + 119886
1cos 120591 + 119886
119896cos 119896120591 (153)
In Section 61 we provide general description of non-negative cosine waveforms of type (153) with at least onezero We show that nonnegative cosine waveforms with atleast one zero coincide with nonnegative cosine waveformswith maximal absolute value of coefficient 119886
1for prescribed
coefficient 119886119896 In Section 62 we illustrate results of Section 61
for particular case 119896 = 3
61 Nonnegative Cosine Waveforms with at Least One Zerofor 119896 ge 2 Amplitudes of fundamental and 119896th harmonic
Mathematical Problems in Engineering 19
of cosine waveform of type (153) are 1205821= |1198861| and 120582
119896=
|119886119896| respectively According to (42) for nonnegative cosine
waveforms of type (153) the following relation holds
minus1 le 119886119896le 1 (154)
This explains why 119896th harmonic coefficient 119886119896in Proposi-
tion 26 goes through interval [minus1 1]Waveforms (153) with 119886
1ge 0 can be obtained from
waveforms with 1198861le 0 by shifting by 120587 and therefore with-
out loss of generality we can assume that 1198861le 0
Proposition 26 Each nonnegative cosine waveform of type(153) with 119886
1le 0 and at least one zero can be represented as
119879119896(120591) = [1 minus cos 120591] [1 minus (119896 minus 1) 119886
119896minus 2119886119896
119896minus1
sum
119899=1
(119896 minus 119899) cos 119899120591]
(155)
if minus1 le 119886119896le 1(119896
2
minus 1) or
119879119896(120591) = 119886
119896[1 minus cos (120591 minus 120591
0)] [1 minus cos (120591 + 120591
0)]
sdot [1198880+ 2
119896minus2
sum
119899=1
119888119899cos 119899120591]
(156)
where
119888119899=sin ((119896 minus 119899) 120591
0) cos 120591
0minus (119896 minus 119899) cos ((119896 minus 119899) 120591
0) sin 120591
0
sin31205910
(157)
119886119896=
sin 1205910
119896 sin (1198961205910) cos 120591
0minus cos (119896120591
0) sin 120591
0
(158)
100381610038161003816100381612059101003816100381610038161003816 le
120587
119896 (159)
if 1(1198962 minus 1) le 119886119896le 1
Remark 27 Identity [1minuscos(120591minus1205910)][1minuscos(120591+120591
0)] = [cos 120591
0minus
cos 120591]2 implies that (156) can be rewritten as
119879119896(120591) = 119886
119896[cos 120591
0minus cos 120591]2 [119888
0+ 2
119896minus2
sum
119899=1
119888119899cos 119899120591] (160)
Furthermore substitution of (157) into (160) leads to
119879119896(120591) = 119886
119896[cos 120591
0minus cos 120591]
sdot [(119896 minus 1) sin 119896120591
0
sin 1205910
minus 2
119896minus1
sum
119899=1
sin ((119896 minus 119899) 1205910)
sin 1205910
cos 119899120591]
(161)
Remark 28 All nonnegative cosine waveforms of type (153)with at least one zero and 119886
1le 0 except one of them can be
represented either in form (155) or form (156)This exceptionis maximally flat cosine waveform with 119886
1lt 0 which can be
obtained from (155) for 119886119896= 1(119896
2
minus 1) or from (156) for 1205910=
0 Maximally flat cosine waveform with 1198861lt 0 can also be
obtained from (70) by setting 1205910= 0 Furthermore setting
1205910= 0 in (71) leads to maximally flat cosine waveforms for
119896 le 4 and 1198861lt 0
Remark 29 Nonnegative cosine waveform of type (155) with1198861lt 0 and minus1 lt 119886
119896le 1(119896
2
minus 1) has exactly one zero at120591 = 0 Nonnegative cosine waveform described by (156) with1198861lt 0 and 1(1198962 minus 1) lt 119886
119896lt 1 has two zeros at plusmn120591
0 where
0 lt |1205910| lt 120587119896 For 119886
119896= ∓1 nonnegative cosine waveform
of type (153) reduces to 119879119896(120591) = 1 ∓ cos 119896120591 (clearly these two
waveforms both have 119896 zeros)
Remark 30 Transformation of (155) into an additive formleads to the following relation
1198861= minus1 minus 119886
119896 (162)
where minus1 le 119886119896le 1(119896
2
minus1) Similarly transformation of (156)leads to the following relation
1198861= minus119886119896
119896 sin 1198961205910
sin 1205910
(163)
where 119886119896is given by (158) 1(1198962minus1) le 119886
119896le 1 and |120591
0| le 120587119896
Notice that coefficients of maximally flat cosine waveformnamely 119886
119896= 1(119896
2
minus1) and 1198861= minus1198962
(1198962
minus1) satisfy relation(162) They also satisfy relation (163) for 120591
0= 0
Remark 31 Nonnegative cosine waveforms of type (153) withat least one zero coincide with nonnegative cosine waveformswith maximal absolute value of coefficient 119886
1for prescribed
coefficient 119886119896
In proving that Remark 31 holds notice that expression(155) can be obtained from (121) by setting 119887
119896= 0 Fur-
thermore if 119886119896
ge 0 then 120582119896
= 119886119896 which together
with 119887119896
= 0 and (118) implies 120575 = 0 In this case119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) le 1 + 119886
119896becomes 1198962119886
119896le 1 +
119886119896 On the other hand if 119886
119896lt 0 then 120582
119896= minus119886
119896 which
together with 119887119896= 0 and (118) implies |120575| = 120587 In this case
119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) le 1 + 119886
119896becomes 0 le 1 + 119886
119896
Therefore every nonnegative cosine waveform of type (155)has maximal absolute value of coefficient 119886
1for prescribed
coefficient 119886119896 when minus1 le 119886
119896le 1(119896
2
minus 1)Let us now show that expression (156) can be obtained
from (122) by setting 119887119896= 0 and 119886
119896gt 0 For waveforms
of type (122) according to (118) 119887119896= 0 and 119886
119896gt 0 imply
120575 = 0 and 120582119896= 119886119896 Substitution of 120582
119896= 119886119896and 120575 = 0 into
119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) ge 1+119886
119896leads to 119886
119896ge 1(119896
2
minus1)Furthermore substitution of 120575 = 0 into (145) yields 120591
0= 120585119896
Insertion of 120582119896= 119886119896 120575 = 0 and 120591
0= 120585119896 into (122) leads
to (156) Therefore every nonnegative cosine waveform oftype (156) has maximal absolute value of coefficient 119886
1for
prescribed coefficient 119886119896 when 1(1198962 minus 1) le 119886
119896le 1
Proof of Proposition 26 Let us start with nonnegative cosinewaveform of type (153) with 120582
119896= |119886119896| = 1 According to
Remark 7 120582119896= |119886119896| = 1 implies that 120582
1= |1198861| = 0
Substitution of 119886119896
= minus1 into (155) and using (A2) (seeAppendices) lead to 119879
119896(120591) = 1 minus cos 119896120591 Consequently (155)
holds for 119886119896= minus1 On the other hand substitution of 119886
119896= 1
into (158) yields |1205910| = 120587119896 Furthermore substitution of
119886119896= 1 and 120591
0= 120587119896 (or 120591
0= minus120587119896) into (156) along
with performing all multiplications and using (A2) leads to
20 Mathematical Problems in Engineering
119879119896(120591) = 1 + cos 119896120591 Consequently (156)ndash(158) hold for 119886
119896= 1
and |1205910| = 120587119896
It is easy to see that 120582119896= |119886119896| lt 1 and 119879
119896(1205910) = 0 for some
1205910imply 120582
1= |1198861| = 0 Therefore in what follows we assume
that |119886119896| = 1 and 119886
1lt 0
Cosine waveforms are even functions of 120591 Therefore ifnonnegative cosine waveform has exactly one zero it has to beeither at 0 or at 120587 On the other hand if nonnegative cosinewaveform with 119886
1= 0 has exactly two zeros then these zeros
are placed at plusmn1205910 such that 120591
0is neither 0 nor 120587
In order to prove that (155) holds for minus1 lt 119886119896le 1(1minus119896
2
)let us start by referring to the description (38) of nonnegativewaveformswith at least one zero As wementioned earlier fornonnegative cosine waveformwith exactly one zero (denotedby 1205910) it is either 120591
0= 0 or 120591
0= 120587 Therefore in both cases
sin 1205910= 0 Substitution of sin 120591
0= 0 into (43) together with
1198861
= 0 and 120582119896= |119886119896| lt 1 leads to
1205910= 0 (164)
Clearly 1205910= 0 119887
1= 0 and 119887
119896= 0 according to (44) and (46)
imply 120582119896sin 120585 = 0 Since 120582
119896= |119886119896| it follows that |119886
119896| sin 120585 = 0
also holds which further implies 120582119896= 119886119896= 0 or sin 120585 =
0 In the case when 120582119896= 119886119896= 0 from (164) and (43) we
obtain 1198861= minus1 which further implies that 119879
119896(120591) = 1 minus cos 120591
Consequently (155) holds for 119886119896= 0 In the case when sin 120585 =
0 from (164) and (45) we obtain 119886119896= 120582119896if 120585 = 0 or 119886
119896= minus120582119896
if 120585 = 120587 Relations 119886119896= 120582119896and 120585 = 0 according to (40) imply
that 0 le 119886119896le 1(1 minus 119896
2
) Substitution of 120585 = 0 120582119896= 119886119896 and
(164) into (38) leads to (155) which proves that (155) holdsfor 0 le 119886
119896le 1(1 minus 119896
2
) On the other hand relations 119886119896=
minus120582119896and 120585 = 120587 according to (40) imply that minus1 lt 119886
119896le 0
Substitution of 120585 = 120587 120582119896= minus119886119896 and (164) into (38) also leads
to (155) which proves that (155) also holds for minus1 lt 119886119896le 0
Consequently (155) holds for minus1 lt 119886119896le 1(1 minus 119896
2
)In what follows we first prove that (156)-(157) hold for
1(1 minus 1198962
) lt 119886119896lt 1 For this purpose let us start with non-
negative waveforms with two zeros described by (66) As wementioned before nonnegative cosine waveforms with twozeros have zeros at 120591
0and minus120591
0 such that 120591
0= 0 and 120591
0= 120587
Relations 1198861
lt 0 and 1198871
= 0 according to (84) implycos(1205910minus 120585119896) = 1 and therefore
120585
119896= 1205910 (165)
From 120585119896 = 1205910and 0 lt |120585| lt 120587 it follows that 0 lt |120591
0| lt
120587119896 Insertion of 120585119896 = 1205910into (45) yields 119886
119896= 120582119896 Relations
119886119896= 120582119896and (82) imply that 1(1 minus 1198962) lt 119886
119896lt 1 Substitution
of 120582119896= 119886119896and 120585119896 = 120591
0into (66)ndash(68) leads to (156)ndash(158)
which proves that (156)ndash(158) hold for 1(1 minus 1198962
) lt 119886119896lt 1
and 0 lt |1205910| lt 120587119896
Finally substitution of 119886119896= 1(1 minus 119896
2
) and 1205910= 0 into
(161) leads to
119879119896(120591) =
[1 minus cos 120591](1 minus 1198962)
[119896 (119896 minus 1) minus 2
119896minus1
sum
119899=1
(119896 minus 119899) cos 119899120591]
(166)
Waveform (166) coincides with waveform (155) for 119886119896
=
1(1 minus 1198962
) which in turn proves that (156) holds for 119886119896=
1(1 minus 1198962
) and 1205910= 0 This completes the proof
62 Nonnegative CosineWaveforms with at Least One Zero for119896 = 3 In this subsection we consider nonnegative cosinewaveforms with at least one zero for 119896 = 3 (for case 119896 = 2
see [12])Cosine waveform with fundamental and third harmonic
reads
1198793(120591) = 1 + 119886
1cos 120591 + 119886
3cos 3120591 (167)
For 1198861le 0 and minus1 le 119886
3le 18 according to (155) non-
negative cosine waveform of type (167) with at least one zerocan be expressed as
1198793(120591) = (1 minus cos 120591) [1 minus 2119886
3(1 + 2 cos 120591 + cos 2120591)] (168)
From 1198793(120591 + 120587) = 2 minus 119879
3(120591) it immediately follows that for
1198861ge 0 and minus18 le 119886
3le 1 119879
3(120591) can be expressed as
1198793(120591) = (1 + cos 120591) [1 + 2119886
3(1 minus 2 cos 120591 + cos 2120591)] (169)
For 1198861le 0 and 18 le 119886
3le 1 from (158) it follows that 119886
3=
[8cos31205910]minus1 This relation along with (160) and (157) further
implies that 1198793(120591) can be expressed as
1198793(120591) =
[cos 1205910minus cos 120591]2 [2 cos 120591
0+ cos 120591]
2cos31205910
(170)
providing that |1205910| le 1205873 From 119879
3(120591 + 120587) = 2 minus 119879
3(120591) it
follows that (170) also holds for 1198861ge 0 and minus1 le 119886
3le minus18
providing that 1205910isin [21205873 41205873]
Maximally flat nonnegative cosinewaveformof type (167)with 119886
1lt 0 (minimum at 120591
0= 0) reads 119879
3(120591) = [1 minus
cos 120591]2[1 + (12) cos 120591] Dually maximally flat nonnegativecosine waveform with 119886
1gt 0 (minimum at 120591
0= 120587) reads
1198793(120591) = [1 + cos 120591]2[1 minus (12) cos 120591]In what follows we provide relations between coefficients
1198861and 1198863of nonnegative cosine waveforms of type (167) with
at least one zeroFor 1198861le 0 conversion of (168) into an additive form
immediately leads to the following relation
1198861= minus1 minus 119886
3for minus 1 le 119886
3le1
8 (171)
Conversion of (170) into an additive form leads to 1198861
=
minus31198863(2 cos 2120591
0+ 1) which can be also expressed as 119886
1=
minus31198863(4cos2120591
0minus 1) For 119886
1le 0 relations |120591
0| le 1205873 119886
1=
minus31198863(4cos2120591
0minus 1) and 119886
3= [8cos3120591
0]minus1 lead to
1198861= minus3 [ 3radic119886
3minus 1198863] for 1
8le 1198863le 1 (172)
Similarly for 1198861ge 0 conversion of (169) into an additive form
leads to the following relation
1198861= 1 minus 119886
3for minus
1
8le 1198863le 1 (173)
Mathematical Problems in Engineering 21
1
15
15
05
0
minus05
minus15minus15
minus1
minus1 minus05 0 05 1
Coefficient a1
Coe
ffici
enta
3 a1= minus
3(3radica3minus a3)
a1 =
minus1 minus a3
a1 =
1 minusa3
a1 =3(3radic|a3
| +a3)
Figure 14 Parameter space of cosine waveforms for 119896 = 3
For waveform of type (170) with 1198861ge 0 relations 120591
0isin [21205873
41205873] 1198861= minus3119886
3(4cos2120591
0minus 1) and 119886
3= [8cos3120591
0]minus1 lead to
1198861= 3 [
3radic10038161003816100381610038161198863
1003816100381610038161003816 + 1198863] for minus 1 le 1198863le minus
1
8 (174)
Every cosine waveform of type (167) corresponds to apair of real numbers (119886
1 1198863) and vice versa Points (119886
1 1198863)
in grey area in Figure 14 correspond to nonnegative cosinewaveforms for 119896 = 3 The points at the boundary of grey areacorrespond to nonnegative cosinewaveformswith at least onezero A number of shapes of nonnegative cosine waveformswith 119896 = 3 and at least one zero plotted on interval [minus120587 120587]are also presented in Figure 14 The boundary of grey area inFigure 14 consists of four line segments described by relations(171)ndash(174) The common point of line segments (172) and(173) is cusp point with coordinates 119886
1= 0 and 119886
3= 1
Another cusp point with coordinates 1198861= 0 and 119886
3= minus1
is the common point of line segments (171) and (174) Thecommon point of line segments (171)-(172) has coordinates(minus98 18) and common point of line segments (173)-(174)has coordinates (98 minus18) These points are representedby white circle dots and they correspond to maximallyflat cosine waveforms (eg see [21]) White triangle dotswith coordinates (2radic3 minusradic39) and (minus2radic3radic39) refer tothe nonnegative cosine waveforms with maximum value ofamplitude of fundamental harmonic
7 Four Case Studies of Usage of NonnegativeWaveforms in PA Efficiency Analysis
In this section we provide four case studies of usage ofdescription of nonnegative waveforms with fundamental and119896th harmonic in PA efficiency analysis In first two casestudies to be presented in Section 71 voltage is nonnegativewaveform with fundamental and second harmonic with atleast one zero In remaining two case studies to be consideredin Section 72 voltage waveform contains fundamental andthird harmonic
i
v Load
Vdc
Idc
Lch
vL
iL
Cb
+ +in
Figure 15 Generic PA circuit diagram
Let us consider generic PA circuit diagram as shown inFigure 15 We assume here that voltage and current wave-forms at the transistor output are
V (120579) = 1 + 1198861V cos 120579 + 1198871V sin 120579 + 119886119896V cos 119896120579 + 119887119896V sin 119896120579
119894 (120579) = 1 + 1198861119894cos 120579 +
infin
sum
119899=2
119886119899119894cos 119899120579
(175)
where 120579 stands for 120596119905 Both waveforms are normalized in thesense that dc components of voltage and current are 119881dc =
1 and 119868dc = 1 respectively Under assumption that blockingcapacitor 119862
119887behaves as short-circuit at the fundamental and
higher harmonics current and voltage waveforms at the loadare
V119871(120579) = 119886
1V cos 120579 + 1198871V sin 120579 + 119886119896V cos 119896120579 + 119887119896V sin 119896120579
119894119871(120579) = minus119886
1119894cos 120579 minus
infin
sum
119899=2
119886119899119894cos 119899120579
(176)
In terms of coefficients of voltage and current waveforms theload impedance at fundamental harmonic is 119911
1= minus(119886
1V minus
1198951198871V)1198861119894 whereas load impedance at 119896th harmonic is 119911
119896=
minus(119886119896Vminus119895119887119896V)119886119896119894 All other harmonics are short-circuited (119911
119899=
0 for 119899 = 1 and 119899 = 119896) Time average output power of PA (egsee [10]) with waveform pair (175) at fundamental frequencycan be expressed as
1198751= minus
11988611198941198861V
2 (177)
For normalized waveforms (175) with 119881dc = 1 and 119868dc = 1dc power is 119875dc = 1 Consequently PA efficiency 120578 = 119875
1119875dc
(eg see [10 26]) is equal to
120578 = minus11988611198941198861V
2 (178)
Thus time average output power 1198751of PA with pair of nor-
malized waveform (175) is equal to efficiency (178)Power utilization factor (PUF) is defined [26] as ldquothe
ratio of power delivered in a given situation to the power
22 Mathematical Problems in Engineering
delivered by the same device with the same supply voltagein Class A moderdquo Since the output power in class-A modeis 1198751class-A = max[V(120579)] sdotmax[119894(120579)]8 (eg see [9]) it follows
that power utilization factor PUF = 11987511198751class-A for PA with
pair of normalized waveforms (175) can be expressed as
PUF =8120578
max [V (120579)] sdotmax [119894 (120579)] (179)
71 NonnegativeWaveforms for 119896 = 2 in PAEfficiencyAnalysisIn this subsection we provide two case studies of usage ofdescription of nonnegative waveforms with fundamental andsecond harmonic (119896 = 2) in PA efficiency analysis For moreexamples of usage of descriptions of nonnegative waveformswith fundamental and second harmonic in PA efficiencyanalysis see [12]
Case Study 71 In this case study we consider efficiency of PAfor given second harmonic impedance providing that voltageis nonnegative waveform with fundamental and second har-monic and current is ldquohalf-sinerdquo waveform frequently used inefficiency analysis of classical PA operation (eg see [10])
Standard model of current waveform for classical PAoperation has the form (eg see [10 26])
119894119863(120579) =
119868119863[cos 120579 minus cos(120572
2)] |120579| le
120572
2
0120572
2le |120579| le 120587
(180)
where 120572 is conduction angle and 119868119863
gt 0 Since 119894119863(120579) is
even function it immediately follows that its Fourier seriescontains only dc component and cosine terms
119894119863(120579) = 119868dc +
infin
sum
119899=1
119868119899cos 119899120579 (181)
The dc component of the waveform (180) is
119868dc =119868119863120572
2120587[sinc(120572
2) minus cos(120572
2)] (182)
where sinc119909 = (sin119909)119909 The coefficient of the fundamentalharmonic component reads
1198681=119868119863120572
2120587(1 minus sinc120572) (183)
and the coefficient of 119899th harmonic component can bewrittenin the form
119868119899=119868119863
119899120587[sin ((119899 minus 1) 1205722)
(119899 minus 1)minussin ((119899 + 1) 1205722)
(119899 + 1)] 119899 ge 2
(184)
For ldquohalf-sinerdquo current waveform conduction angle is equalto 120587 (class-B conduction angle) According to (182) thisfurther implies that 119868dc = 119868
119863120587 To obtain normalized form of
waveform (180) we set 119868dc = 1which implies that 119868119863= 120587 Fur-
thermore substitution of 120572 = 120587 and 119868119863= 120587 in (180) leads to
119894 (120579) =
120587 cos 120579 |120579| lt120587
2
0120587
2lt |120579| le 120587
(185)
Similarly substitution of 119868119863= 120587 and 120572 = 120587 into (183) and
(184) leads to the coefficients of waveform (185) Coefficientsof fundamental and second harmonic respectively are
1198861119894=120587
2 119886
2119894=2
3 (186)
On the other hand voltage waveform of type (35) for 119896 =2 reads
V (120579) = 1 + 1198861V cos 120579 + 1198871V sin 120579 + 1198862V cos 2120579 + 1198872V sin 2120579
(187)
This waveform contains only fundamental and second har-monic and therefore all harmonics of order higher thantwo are short-circuited (119911
119899= 0 for 119899 gt 2) For current
voltage pair (185) and (187) load impedance at fundamentalharmonic is 119911
1= minus(119886
1V minus 1198951198871V)1198861119894 whereas load impedance
at second harmonic is 1199112= minus(119886
2Vminus1198951198872V)1198862119894 According to ourassumption the load is passive and therefore Re119911
1 gt 0 and
Re1199112 ge 0 which further imply 119886
11198941198861V lt 0 and 119886
21198941198862V le 0
respectivelyIt is easy to see that problem of findingmaximal efficiency
of PA with current-voltage pair (185) and (187) for prescribedsecond harmonic impedance can be reduced to the problemof finding voltage waveform of type (187) with maximal coef-ficient |119886
1V| for prescribed coefficients of second harmonic(see Section 5)
The following algorithm (analogous to Algorithm 22presented in [12]) provides the procedure for calculation ofmaximal efficiency with current-voltage pair (185) and (187)for prescribed second harmonic impedance The definitionof function atan 2(119910 119909) which appears in the step (iii) of thefollowing algorithm is given by (105)
Algorithm 32 (i) Choose 1199112= 1199032+1198951199092such that |119911
2| le 1|119886
2119894|
(ii) calculate 1198862V minus 1198951198872V = minus119911
21198862119894and 120582
2V = radic11988622V + 1198872
2V(iii) if 2120582
2V le 1 minus 1198862V then calculate 119886
1V = minus1 minus 1198862V and
1198871V = minus2119887
2V else calculate 1205821V = radic81205822V(1 minus 1205822V) 1205790V minus 120585V2 =
(12)atan2(1198872V 1198862V) 1198861V = minus120582
1V cos(1205790V minus 120585V2) and 1198871V =
minus1205821V sin(1205790V minus 120585V2)(iv) calculate efficiency 120578 = minus119886
11198941198861V2
(v) calculate 1199111= minus(119886
1V minus 1198951198871V)1198861119894 and 1199112119899 = 1199112Re119911
1
In this case study coefficients of fundamental and secondharmonic of current waveform are given by (186) Maximalefficiency of PA associated with the waveform pair (185)and (187) as a function of normalized second harmonicimpedance 119911
2119899= 1199112Re119911
1 is presented in Figure 16(a)
As can be seen from Figure 16(a) efficiency of 078 isachieved at the edge of Smith chart where second harmonicimpedance has small resistive part Corresponding PUFcalculated according to (179) is presented in Figure 16(b)Peak efficiency 120578 = 1205874 = 07854 and peak value of PUF = 1
are attained when second harmonic is short-circuited (whichcorresponds to ideal class-B operation [10 26])
For example for second harmonic impedance 1199112= 01 minus
11989505 and currentwaveform (185) fromAlgorithm32 it followsthat 2120582
2V le 1 minus 1198862V Furthermore according to step (iii)
of above algorithm maximal efficiency of PA is attained
Mathematical Problems in Engineering 23
+j2
+j5
+j1
+j2
+j5
minusj2
minusj5
minusj1
minusj2
minusj5
infin0
078
02
05
1 2 5
075 07 065
06 05
075
07
065
06
05
120578 lt 05
(a)
099
095
091
083 075
067
0 59 051
051
099
095
091
083
075 0
67 059
+j5
+j1
+j2
+j5
minusj5
minusj1
minusj2
minusj5
infin
+j2
minusj2
0 02
05
1 2 5
120578 lt 05
(b)
Figure 16 (a) Contours of maximal efficiency of PA and (b) contours of corresponding PUF associated with the waveform pair (185) and(187) as functions of normalized second harmonic impedance 119911
2119899= 1199112Re119911
1
2
3
4
1
0
2 3 410
Wav
efor
ms
Angle 120579120587
i(120579)
(120579)
Figure 17 Waveform pair (185) and (187) that provides maximalefficiency for 119911
2= 01 minus 11989505
with voltage waveform of type (187) with coefficients 1198862V =
minus00667 1198872V = minus03333 119886
1V = minus09333 and 1198871V = 06667 (see
Figure 17) Corresponding efficiency PUF and normalizedsecond harmonic impedance are 120578 = 07330 PUF = 07572and 1199112119899= 01683 minus 11989508415 respectively
On the other hand for second harmonic impedance 1199112=
01 minus 11989508 and current waveform (185) from Algorithm 32 itfollows that 2120582
2V gt 1 minus 1198862V Then according to step (iii) of
above algorithm maximal efficiency is attained with voltagewaveform of type (187) with coefficients 119886
2V = minus00667 1198872V =
minus05333 1198861V = minus09333 and 119887
1V = 10572 (see Figure 18)Efficiency PUF andnormalized secondharmonic impedanceare 120578 = 07330 PUF = 06332 and 119911
2119899= 01683 minus 11989513465
respectively
2
3
4
1
0
2 3 410
Wav
efor
ms
Angle 120579120587
i(120579)
(120579)
Figure 18 Waveform pair (185) and (187) that provides maximalefficiency for 119911
2= 01 minus 11989508
Case Study 72 As another case study let us consider the effi-ciency of PA providing that current waveform is nonnegativecosine waveform up to third harmonic with maximum valueof amplitude of fundamental harmonic [22] (see also [8])
119894 (120579) = 1 +1 + radic5
2cos 120579 + 2radic5
5cos 2120579 + 5 minus radic5
10cos 3120579
(188)
and voltage waveform is nonnegative waveform of type(187) Load impedances at fundamental second and thirdharmonic are 119911
1= minus(119886
1Vminus1198951198871V)1198861119894 1199112 = minus(1198862Vminus1198951198872V)1198862119894 and
1199113= 0 respectively According to our assumption the load
is passive and therefore Re1199111 gt 0 and Re119911
2 ge 0 which
further imply 11988611198941198861V lt 0 and 119886
21198941198862V le 0 respectively
Because current waveform (188) contains only cosineterms and voltage waveform is the same as in previous case
24 Mathematical Problems in Engineering
08
075
07
06
06
065
065
05
05
075 0
7
120578 lt 05
+j2
+j5
+j1
+j2
+j5
minusj2
minusj5
minusj1
minusj2
minusj5
infin0
02
05
1 2 5
(a)
085
08 0
75 07 0
6
05 04
08
075
0405
060
7
120578 lt 05
+j2
+j5
+j1
+j2
+j5
minusj2
minusj5
minusj1
minusj2
minusj5
infin0 02
05
1 2 5
(b)
Figure 19 (a) Contours of maximal efficiency of PA and (b) contours of corresponding PUF associated with the waveform pair (187)-(188)as functions of normalized second harmonic impedance 119911
2119899= 1199112Re119911
1
study the procedure for calculation of maximal efficiency ofPA with waveform pair (187)-(188) is the same as presentedin Algorithm 32 In this case study the coefficients of funda-mental and second harmonic of current waveform are 119886
1119894=
(1 + radic5)2 and 1198862119894= 2radic55 respectively
Maximal efficiency of PA associated with the waveformpair (187)-(188) as a function of normalized secondharmonicimpedance 119911
2119899= 1199112Re119911
1 is presented in Figure 19(a)
Efficiency of 08 is achieved at the edge of Smith chart wheresecond harmonic impedance has small resistive part Thetheoretical upper bound 120578 = (1 + radic5)4 asymp 08090 isattained when second harmonic is short-circuitedWhen thisupper bound is reached both second and third harmonic areshort-circuited which implies that we are dealing with finiteharmonic class-C [6 8] or dually when current and voltageinterchange their roles with finite harmonic inverse class-C[6 9] Corresponding PUF calculated according to (179) ispresented in Figure 19(b) Peak value of PUF asymp 08541 isattained when second harmonic is short-circuited
For example for second harmonic impedance 1199112
=
007 minus 11989504 and current waveform (188) from Algorithm 32 itfollows that 2120582
2V le 1 minus 1198862V Furthermore according to step
(iii) of Algorithm 32 maximal efficiency of PA is attainedwith voltage waveform of type (187) with coefficients 119886
2V =
minus00626 1198872V = minus03578 119886
1V = minus09374 and 1198871V = 07155 (see
Figure 20) Corresponding efficiency PUF and normalizedsecond harmonic impedance are 120578 = 07584 PUF = 06337and 1199112119899= 01208 minus 11989506904 respectively
On the other hand for 1199112= 005 minus 11989507 and current
waveform (187) it follows that 21205822V gt 1minus119886
2VThen accordingto step (iii) of Algorithm 32 themaximal efficiency is attainedwith voltage waveform of type (187) with coefficients 119886
2V =
minus00447 1198872V = minus06261 119886
1V = minus09318 and 1198871V = 10007 (see
Figure 21) Efficiency PUF and normalized second harmonic
2
3
4
1
0
2 3 410
Wav
efor
ms
Angle 120579120587
i(120579)
(120579)
Figure 20 Waveform pair (187)-(188) that provides maximal effi-ciency for 119911
2= 007 minus 11989504
impedance are 120578 = 07538 PUF = 05314 and 1199112119899= 00868minus
11989512156 respectively
72 Nonnegative Waveforms for 119896 = 3 in PA EfficiencyAnalysis In this subsection we provide another two casestudies of usage of description of nonnegative waveforms inPA efficiency analysis this time with fundamental and thirdharmonic (119896 = 3)
Case Study 73 Let us consider current-voltage pair such thatvoltage is nonnegative waveform with fundamental and thirdharmonic
V (120579) = 1 + 1198861V cos 120579 + 1198871V sin 120579 + 1198863V cos 3120579 + 1198873V sin 3120579
(189)
Mathematical Problems in Engineering 25
2
3
4
1
0
2 3 410
Wav
efor
ms
Angle 120579120587
i(120579)
(120579)
Figure 21 Waveform pair (187)-(188) that provides maximal effi-ciency for 119911
2= 005 minus 11989507
and current is nonnegative cosine waveform given by (188)Load impedances at fundamental second and third har-monic are 119911
1= minus(119886
1V minus 1198951198871V)1198861119894 1199112 = 0 and 119911
3= minus(119886
3V minus
1198951198873V)1198863119894 respectively According to our assumption the load
is passive and therefore Re1199111 gt 0 and Re119911
3 ge 0 which
further imply 11988611198941198861V lt 0 and 119886
31198941198863V le 0
In this subsection we consider the problem of findingmaximal efficiency of PA with waveform pair (188)-(189)for given third harmonic impedance As we mentionedearlier problem of finding maximal efficiency of PA withcurrent-voltage pair (188)-(189) for prescribed third har-monic impedance can be reduced to the problem of findingvoltage waveform of type (189) withmaximal coefficient |119886
1V|for prescribed coefficients of third harmonic (see Section 52)
The following algorithm provides the procedure forcalculation of maximal efficiency with current-voltage pair(188)-(189) The definition of function atan 2(119910 119909) whichappears in step (iii) of the following algorithm is given by(105)
Algorithm 33 (i) Choose 1199113= 1199033+1198951199093such that |119911
3| le 1|119886
3119894|
(ii) calculate 1198863V minus 1198951198873V = minus119911
31198863119894and 120582
3V = radic11988623V + 1198872
3V(iii) if 271205822
3V le (1 minus 21198863V)3 then calculate 119886
1V = minus1 minus 1198863V
and 1198871V = minus3119887
3V else calculate 1205821V = 3(3radic1205823V minus 120582
3V) 1205790V minus120585V3 = (13)atan 2(119887
3V 1198863V) 1198861V = minus1205821V cos(1205790V minus 120585V3) and
1198871V = minus120582
1V sin(1205790V minus 120585V3)(iv) calculate efficiency 120578 = minus119886
11198941198861V2
(v) calculate 1199111= minus(119886
1V minus 1198951198871V)1198861119894 and 1199113119899 = 1199113Re119911
1
In this case study coefficients of fundamental and thirdharmonic of current waveform are 119886
1119894= (1 + radic5)2 and
1198863119894= (5 minus radic5)10 respectively For the waveform pair (188)-
(189) maximal efficiency of PA as a function of normalizedthird harmonic impedance 119911
3119899= 1199113Re119911
1 is presented in
Figure 22 Efficiency of 08 is reached when third harmonicimpedance has small resistive part Peak efficiency 120578 = (1 +
radic5)4 asymp 08090 is achieved when third harmonic is short-circuited
+j2
+j5
+j1
+j2
+j5
minusj2
minusj5
minusj1
minusj2
minusj5
infin0
08
08
075
075
07
07
06
05
02
05
1 2
Figure 22 Contours of maximal efficiency of PA associated withthe waveform pair (188)-(189) as a function of normalized thirdharmonic impedance 119911
3119899= 1199113Re119911
1
For the present case study in what follows we show thatpower utilization factor is proportional to efficiency For volt-age waveform of type (189) it is easy to see that V(120579 + 120587) =
2minusV(120579) holdsThis relation along with the fact that waveformV(120579) that provides maximal efficiency has at least one zeroimplies that max[V(120579)] = 2 On the other hand currentwaveform (188) is cosine waveform with positive coefficientsand therefore max[119894(120579)] = 119894(0) = 2 + 4radic5 Consequentlyaccording to (179) the following relation holds
PUFCase study 73 = 2 (5 minus 2radic5) 120578 = 10557120578 (190)
Clearly the ratio PUF120578 is constant and therefore in this casestudy PUF can be easily calculated from the correspondingefficiency Accordingly peak efficiency and peak value ofPUFCase study 73 = 3radic52 minus 52 = 08541 are attained forthe same voltage waveform (when third harmonic is short-circuited)
In the first example current waveform (188) and 1199113=
02 minus 11989505 imply that 2712058223V le (1 minus 2119886
3V)3 Then according
to Algorithm 33 the voltage waveform of type (189) thatprovides maximal efficiency has the following coefficients1198863V = minus00553 119887
3V = minus01382 1198861V = minus09447 and 119887
1V =
04146 (see Figure 23) Efficiency PUF and normalized thirdharmonic impedance are 120578 = 07643 PUF = 08069 and1199113119899= 03425 minus 11989508564 respectivelyIn the second example current waveform (188) and 119911
3=
01 minus 11989511 imply that 2712058223V gt (1 minus 2119886
3V)3 Then according
to Algorithm 33 the voltage waveform of type (189) thatprovides maximal efficiency has the following coefficients1198863V = minus00276 119887
3V = minus03040 1198861V = minus09391 and 119887
1V =
05807 (see Figure 24) Efficiency PUF and normalized thirdharmonic impedance are 120578 = 07598 PUF = 08021 and1199113119899= 01723 minus 11989518952 respectively
26 Mathematical Problems in Engineering
2
3
4
1
0
2 3 410
Wav
efor
ms
Angle 120579120587
i(120579)
(120579)
Figure 23 Waveform pair (188)-(189) that provides maximal effi-ciency for 119911
3= 02 minus 11989505
2
3
4
1
0
2 3 410
Wav
efor
ms
Angle 120579120587
i(120579)
(120579)
Figure 24 Waveform pair (188)-(189) that provides maximal effi-ciency for 119911
3= 01 minus 11989511
Case Study 74 In this case study let us consider current-voltage pair where current is normalized waveform of type(180) with conduction angle 120572 = 115120587 (207∘) and voltageis nonnegative waveform of type (189) Substitution of 120572 =
115120587 and 119868dc = 1 into (182) leads to 119868119863
= 22535Furthermore substitution of 120572 = 115120587 and 119868
119863= 22535 into
(180) leads to
119894 (120579)
=
22535 [cos 120579 minus cos(1151205872
)] |120579| le115120587
2
0115120587
2le |120579| le 120587
(191)
Similarly substitution of 120572 = 115120587 and 119868119863
= 22535 into(183) and (184) for 119899 = 3 yields coefficients of fundamentaland third harmonic of waveform (191)
1198861119894= 14586 119886
3119894= minus01026 (192)
Because current waveform (191) contains only cosineterms and voltage waveform is the same as in previous case
+j2
+j5
+j1
+j2
+j5
minusj2
minusj5
minusj1
minusj2
minusj5
infin0 02
05
1 2
074
076
08
084
082
078
076
074
Figure 25 Contours of maximal efficiency of PA associated withthe waveform pair (189) and (191) as a function of normalized thirdharmonic impedance 119911
3119899= 1199113Re119911
1
study the procedure for calculation of maximal efficiencyof PA with waveform pair (189)ndash(191) is the same as thatpresented in Algorithm 33 In this case study the coefficientsof fundamental and third harmonic of current waveform aregiven by (192)
For the waveform pair (189) and (191) maximal efficiencyof PA as a function of normalized third harmonic impedance1199113119899= 1199113Re119911
1 is presented in Figure 25 Efficiency of 084
is obtained in vicinity of 1199113119899= 23685 (corresponding to 119911
3=
18750) Peak efficiency 120578 asymp 08421 is achieved for voltagewaveform of type (189) with coefficients 119886
1V = minus2radic3 1198863V =
radic39 and 1198871V = 1198873V = 0
In the course of finding power utilization factor noticethat currentwaveformof type (191) attains itsmaximumvaluefor 120579 = 0 Insertion of max[119894(120579)] = 119894(0) = 278 andmax[V(120579)] = 2 for voltage waveform of type (189) into (179)leads to
PUFCase study 74 = 1439120578 (193)
Again the ratio PUF120578 is constant and PUF can be easilycalculated from the corresponding efficiency Accordinglypeak value of PUFCase study 74 asymp 12118 and peak efficiencyare attained for the same voltage waveform
In the first example current waveform (191) and 1199113=
1 minus 11989502 imply that 2712058223V le (1 minus 2119886
3V)3 Then according to
Algorithm 33 voltage waveform of type (189) which providesmaximal efficiency has coefficients 119886
3V = 01026 1198873V =
00205 1198861V = minus11026 and 119887
1V = minus00616 (see Figure 26)Efficiency PUF and normalized third harmonic impedanceare 120578 = 08042 PUF = 11572 and 119911
3119899= 13228 minus 11989502646
respectivelyIn second example current waveform (191) and 119911
3=
15 minus 11989512 imply that 2712058223V gt (1 minus 2119886
3V)3 Then according to
Algorithm 33 voltage waveform of type (189) which providesmaximal efficiency has coefficients 119886
3V = 01540 1198873V =
Mathematical Problems in Engineering 27
2
3
4
1
0
2 3 410
Wav
efor
ms
Angle 120579120587
i(120579)
(120579)
Figure 26 Waveform pair (189) and (191) that provides maximalefficiency for 119911
3= 1 minus 11989502
2
3
4
1
0
2 3 410
Wav
efor
ms
Angle 120579120587
i(120579)
(120579)
Figure 27 Waveform pair (189) and (191) that provides maximalefficiency for 119911
3= 15 minus 11989512
01232 1198861V = minus11255 and 119887
1V = minus02575 (see Figure 27)Efficiency PUF and normalized third harmonic impedanceare 120578 = 08208 PUF = 11812 and 119911
3119899= 19439 minus 11989515552
respectively
8 Conclusion
In this paper we consider a problem of finding general de-scriptions of various classes of nonnegative waveforms withfundamental and 119896th harmonic These classes include non-negative waveforms with at least one zero nonnegative wave-forms with maximal amplitude of fundamental harmonic forprescribed amplitude of 119896th harmonic nonnegative wave-forms withmaximal coefficient of cosine part of fundamentalharmonic for prescribed coefficients of 119896th harmonic andnonnegative cosine waveforms with at least one zero Mainresults are stated in six propositions (Propositions 1 6 918 22 and 26) four corollaries (Corollaries 2ndash5) twentyremarks and three algorithms Four case studies of usageof closed form descriptions of nonnegative waveforms in PAefficiency analysis are considered in detail in Section 7
Appendices
Here we provide a list of finite sums of trigonometric func-tions used in this paper (Appendix A) and brief account ofthe Chebyshev polynomials (Appendix B)
A List of Some Finite Sums ofTrigonometric Functions
Dirichlet kernel (eg see [27]) is as follows
119863119896minus1
(120591) = 1 + 2
119896minus1
sum
119899=1
cos 119899120591 = sin ((2119896 minus 1) 1205912)sin (1205912)
(A1)
Fejer kernel (eg see [27]) can be expressed in the fol-lowing equivalent forms
119865119896minus1
(120591) =1
119896
119896minus1
sum
119899=0
119863119899(120591) = 1 +
2
119896
119896minus1
sum
119899=1
(119896 minus 119899) cos 119899120591
=(1 minus cos 119896120591)119896 (1 minus cos 120591)
(A2)
Lagrangersquos trigonometric identity (eg see [28]) is as fol-lows
1198781(120591) =
119896minus1
sum
119899=1
sin 119899120591 = sin (1198961205912) sin ((119896 minus 1) 1205912)sin (1205912)
(A3)
In what follows we show that the following three trigono-metric identities also hold
2
119896minus1
sum
119899=1
(119896 minus 119899) sin 119899120591 = 119896 sin 120591 minus sin 1198961205911 minus cos 120591
(A4)
119896minus1
sum
119899=1
cos (119896 minus 2119899) 120591 = sin (119896 minus 1) 120591sin 120591
(A5)
119896minus1
sum
119899=1
119899 (119896 minus 119899) cos (119896 minus 2119899) 120591
=sin (119896120591) cos 120591 minus 119896 cos (119896120591) sin 120591
2sin3120591
(A6)
Denote 1198782(120591) = 2sum
119896minus1
119899=1(119896 minus 119899) sin 119899120591 119878
3(120591) = sum
119896minus1
119899=1cos(119896 minus
2119899)120591 and 1198784(120591) = sum
119896minus1
119899=1119899(119896 minus 119899) cos(119896 minus 2119899)120591
Notice that 1198782(120591) = 2119896119878
1(120591) + 119889119863
119896minus1(120591)119889120591 which
immediately leads to (A4)
28 Mathematical Problems in Engineering
Identity (A5) can be obtained as follows
sin (119896 minus 1) 120591sin 120591
=119890119895(119896minus1)120591
minus 119890minus119895(119896minus1)120591
119890119895120591 minus 119890minus119895120591
= 119890119895119896120591
119890minus2119895120591
minus 119890minus2119895(119896minus1)120591
1 minus 119890minus2119895120591
= 119890119895119896120591
119896minus1
sum
119899=1
119890minus2119895119899120591
=
119896minus1
sum
119899=1
119890119895(119896minus2119899)120591
=
119896minus1
sum
119899=1
cos (119896 minus 2119899) 120591
(A7)
From 4119899(119896 minus 119899) = 1198962
minus (119896 minus 2119899)2 it follows that 4119878
4(120591) =
1198962
1198783(120591) + 119889
2
1198783(120591)119889120591
2 which leads to (A6)
B The Chebyshev Polynomials
The Chebyshev polynomials of the first kind 119881119899(119909) can be
defined by the following relation (eg see [29])
119881119899(119909) = cos 119899120591 when 119909 = cos 120591 (B1)
The Chebyshev polynomials of the second kind 119880119899(119909) can be
defined by the following relation (eg see [29])
119880119899(119909) =
sin (119899 + 1) 120591sin 120591
when 119909 = cos 120591 (B2)
The Chebyshev polynomials satisfy the following recur-rence relations (eg see [29])
1198810(119909) = 1 119881
1(119909) = 119909
119881119899+1
(119909) = 2119909119881119899(119909) minus 119881
119899minus1(119909)
1198800(119909) = 1 119880
1(119909) = 2119909
119880119899+1
(119909) = 2119909119880119899(119909) minus 119880
119899minus1(119909)
(B3)
The first few Chebyshev polynomials of the first and secondkind are 119881
2(119909) = 2119909
2
minus 1 1198813(119909) = 4119909
3
minus 3119909 1198814(119909) = 8119909
4
minus
81199092
+ 1 1198802(119909) = 4119909
2
minus 1 1198803(119909) = 8119909
3
minus 4119909 and 1198804(119909) =
161199094
minus 121199092
+ 1
Conflict of Interests
The authors declare that there is no conflict of interestsregarding the publication of this paper
Acknowledgment
This work is supported by the SerbianMinistry of EducationScience and Technology Development as a part of ProjectTP32016
References
[1] V I Arnolrsquod V S Afrajmovich Y S Ilrsquoyashenko and L PShilrsquonikov Dynamical Systems V Bifurcation Theory and Catas-trophe Theory Springer Berlin Germany 1994
[2] E Polak ldquoOn the mathematical foundations of nondifferen-tiable optimization in engineering designrdquo SIAM Review vol29 no 1 pp 21ndash89 1987
[3] N S Fuzik ldquoBiharmonic modes of a tuned RF power amplifierrdquoRadiotehnika vol 25 no 7 pp 62ndash71 1970 (Russian)
[4] P Colantonio F Giannini G Leuzzi and E Limiti ldquoClass Gapproach for low-voltage high-efficiency PA designrdquo Interna-tional Journal of RF and Microwave Computer-Aided Engineer-ing vol 10 no 6 pp 366ndash378 2000
[5] F H Raab ldquoMaximum efficiency and output of class-F poweramplifiersrdquo IEEE Transactions on Microwave Theory and Tech-niques vol 49 no 6 pp 1162ndash1166 2001
[6] FH Raab ldquoClass-E class-C and class-F power amplifiers basedupon a finite number of harmonicsrdquo IEEE Transactions onMicrowaveTheory and Techniques vol 49 no 8 pp 1462ndash14682001
[7] J D Rhodes ldquoOutput universality inmaximum efficiency linearpower amplifiersrdquo International Journal of Circuit Theory andApplications vol 31 no 4 pp 385ndash405 2003
[8] A Juhas and L A Novak ldquoComments on lsquoClass-E class-Cand classF power amplifier based upon a finite number ofharmonicsrsquordquo IEEE Transactions on Microwave Theory and Tech-niques vol 57 no 6 pp 1623ndash1625 2009
[9] M Roberg and Z Popovic ldquoAnalysis of high-efficiency poweramplifiers with arbitrary output harmonic terminationsrdquo IEEETransactions on Microwave Theory and Techniques vol 59 no8 pp 2037ndash2048 2011
[10] A Grebennikov N O Sokal and M J Franco Switchmode RFPower Amplifiers ElsevierAcademic Press San Diego CalifUSA 2nd edition 2012
[11] T Canning P J Tasker and S C Cripps ldquoContinuous modepower amplifier design using harmonic clipping contourstheory and practicerdquo IEEE Transactions on Microwave Theoryand Techniques vol 62 no 1 pp 100ndash110 2014
[12] A Juhas and L A Novak ldquoGeneral description of nonnegativewaveforms up to second harmonic for power amplifier mod-ellingrdquoMathematical Problems in Engineering vol 2014 ArticleID 709762 18 pages 2014
[13] V I Arnolrsquod V V Goryunov O V Lyashko and V A VasilrsquoevDynamical Systems VIIImdashSingularity Theory II ApplicationsSpringer Berlin Germany 1993
[14] D Siersma ldquoProperties of conflict sets in the planerdquo BanachCenter Publications Polish Academy of Sciences vol 50 no 1 pp267ndash276 1999 Proceedings of the Banach Center Symposiumon Geometry and Topology of Caustics (Caustics rsquo98) WarsawPoland
[15] M van Manen The geometry of conflict sets [Dissertation]Universiteit Utrecht UtrechtThe Netherlands 2003 httpigi-tur-archivelibraryuunldissertations2003-0912-123058c4pdf
[16] Y L Sachkov ldquoMaxwell strata and symmetries in the problemofoptimal rolling of a sphere over a planerdquo Sbornik Mathematicsvol 201 no 7-8 pp 1029ndash1051 2010
[17] I A Bogaevsky ldquoPerestroikas of shocks and singularities ofminimum functionsrdquoPhysicaDNonlinear Phenomena vol 173no 1-2 pp 1ndash28 2002
[18] Y L Sachkov ldquoMaxwell strata in the Euler elastic problemrdquoJournal of Dynamical andControl Systems vol 14 no 2 pp 169ndash234 2008
[19] M Siino and T Koike ldquoTopological classification of black holesgeneric Maxwell set and crease set of a horizonrdquo InternationalJournal ofModern Physics D Gravitation Astrophysics Cosmol-ogy vol 20 no 6 pp 1095ndash1122 2011
Mathematical Problems in Engineering 29
[20] F H Raab ldquoClass-F power amplifiers with maximally flat wave-formsrdquo IEEETransactions onMicrowaveTheory and Techniquesvol 45 no 11 pp 2007ndash2012 1997
[21] A Juhas and L A Novak ldquoMaximally flat waveforms with finitenumber of harmonics in class-F power amplifiersrdquo Mathemat-ical Problems in Engineering vol 2013 Article ID 169590 9pages 2013
[22] L Fejer ldquoUber trigonometrische polynomerdquo Journal fur dieReine und Angewandte Mathematik vol 1916 no 146 pp 53ndash82 1916 (German)
[23] S C Cripps ldquoBessel Waives [microwave bytes]rdquo IEEE Micro-wave Magazine vol 10 no 7 pp 30ndash36 117 2009
[24] L N Bryzgalova ldquoSingularities of the maximum of para-metrically dependent functionrdquo Functional Analysis and ItsApplications vol 11 no 1 pp 49ndash51 1977
[25] V I Arnold A A Davydov V A Vassiliev and V MZakalyukin Mathematical Models of Catastrophes Control ofCatastrophic Process Encyclopedia of Life Support Systems(EOLSS) EOLSS Publishers Oxford UK 2006
[26] S C Cripps RF Power Amplifiers for Wireless CommunicationsArtech House Norwood Mass USA 2nd edition 2006
[27] A Zygmund Trigonometric Series vol 1 Cambridge UniversityPress Cambridge UK 2nd edition 1959
[28] A Jeffrey and H DaiHandbook of Mathematical Formulas andIntegrals ElsevierAcademic Press San Diego Calif USA 4thedition 2008
[29] J C Mason and D C Handscomb Chebyshev PolynomialsChapman amp Hall CRC Press Boca Raton Fla USA 2003
Submit your manuscripts athttpwwwhindawicom
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical Problems in Engineering
Hindawi Publishing Corporationhttpwwwhindawicom
Differential EquationsInternational Journal of
Volume 2014
Applied MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Probability and StatisticsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical PhysicsAdvances in
Complex AnalysisJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
OptimizationJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
CombinatoricsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Operations ResearchAdvances in
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Function Spaces
Abstract and Applied AnalysisHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of Mathematics and Mathematical Sciences
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Algebra
Discrete Dynamics in Nature and Society
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Decision SciencesAdvances in
Discrete MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom
Volume 2014 Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Stochastic AnalysisInternational Journal of
18 Mathematical Problems in Engineering
52 Nonnegative Waveforms with Maximal Absolute Valueof Coefficient 119886
1for 119896 = 3 Nonnegative waveform of type
(35) for 119896 = 3 is widely used in PA design (eg see [10])In this subsection we illustrate results of Section 51 for thisparticular case The case 119896 = 2 is presented in detail in [12]
Coefficients of fundamental harmonic of nonnegativewaveform of type (35) with 119896 = 3 and maximal absolutevalue of coefficient 119886
1le 0 for prescribed coefficients 119886
3and
1198873(1205823= radic11988623+ 11988723) according to (123) (124) (134) (94) and
(120) are equal to
1198861= minus1 minus 119886
3 119887
1= minus3119887
3 (151)
if 12058223le [(1 minus 2119886
3)3]3
1198861= minus1205821cos(120575
3) 119887
1= minus1205821sin(120575
3) (152)
where 1205821= 3(
3radic1205823minus 1205823) and 120575 = atan 2(119887
3 1198863) if [(1 minus
21198863)3]3
le 1205822
3le 1The line 1205822
3= [(1minus2119886
3)3]3 (see case 119896 = 3
in Figure 11) separates points representing waveforms withcoefficients satisfying (151) from points representing wave-forms with coefficients satisfying (152) Waveforms describedby (151) for 1205822
3lt [(1 minus 2119886
3)3]3 have exactly one zero at
1205910= 0 Waveforms described by (151) and (152) for 1205822
3= [(1 minus
21198863)3]3 also have zero at 120591
0= 0 These waveforms as a rule
have exactly two zeros However there are two exceptionsone related to the maximally flat nonnegative waveform withcoefficients 119886
1= minus98 119886
3= 18 and 119887
1= 1198873= 0 which
has only one zero and the other related to the waveform withcoefficients 119886
1= 0 119886
3= minus1 and 119887
1= 1198873= 0 which has three
zerosWaveforms described by (152) for [(1minus21198863)3]3
lt 1205822
3lt
1 have two zeros Waveforms with 1205823= 1 have only third
harmonic (fundamental harmonic is zero)Plot of contours of maximal absolute value of coefficient
1198861 1198861le 0 for prescribed coefficients 119886
3and 1198873is presented
in Figure 12 According to Remark 25 the waveform withmaximum absolute value of 119886
1le 0 is fully described with
the following coefficients 1198861
= minus2radic3 1198863
= radic39 and1198871= 1198873= 0 This waveform has two zeros at plusmn1205876
Two examples of nonnegative waveforms for 119896 = 3
and maximal absolute value of coefficient 1198861 1198861le 0 with
prescribed coefficients 1198863and 1198873are presented in Figure 13
One waveform corresponds to the case 12058223lt [(1 minus 2119886
3)3]3
(solid line) and the other to the case 12058223gt [(1 minus 2119886
3)3]3
(dashed line)Thewaveform represented by solid line has onezero and its coefficients are 119886
3= minus01 119887
3= 01 119886
1= minus09
and 1198871= minus03 Dashed line corresponds to the waveform
having two zeros with coefficients 1198863= minus01 119887
3= 03 119886
1=
minus08844 and 1198871= minus06460 (case 1205822
3gt [(1 minus 2119886
3)3]3)
6 Nonnegative Cosine Waveforms withat Least One Zero
Nonnegative cosine waveforms have proved to be of impor-tance for waveform modelling in PA design (eg see [10])In this section we consider nonnegative cosine waveforms
1
05
0
minus05
minus1
minus1 minus05 0 05 1
Coefficient a3
Coe
ffici
entb
3
02
04
06
08
10
11
Figure 12 Contours ofmaximal absolute value of coefficient 1198861 1198861le
0 as a function of 1198863and 1198873
2
1
0
0 1 2 3 4
Angle 120591120587
Wav
efor
ms
a3 = minus01 b3 = 01
a3 = minus01 b3 = 03
Figure 13 Nonnegative waveforms for 119896 = 3 and maximal absolutevalue of 119886
1 1198861le 0 with prescribed coefficients 119886
3and 1198873
containing fundamental and 119896th harmonic with at least onezero
Cosine waveform with dc component fundamental and119896th harmonic can be obtained from (35) by setting 119887
1= 119887119896=
0 that is
119879119896(120591) = 1 + 119886
1cos 120591 + 119886
119896cos 119896120591 (153)
In Section 61 we provide general description of non-negative cosine waveforms of type (153) with at least onezero We show that nonnegative cosine waveforms with atleast one zero coincide with nonnegative cosine waveformswith maximal absolute value of coefficient 119886
1for prescribed
coefficient 119886119896 In Section 62 we illustrate results of Section 61
for particular case 119896 = 3
61 Nonnegative Cosine Waveforms with at Least One Zerofor 119896 ge 2 Amplitudes of fundamental and 119896th harmonic
Mathematical Problems in Engineering 19
of cosine waveform of type (153) are 1205821= |1198861| and 120582
119896=
|119886119896| respectively According to (42) for nonnegative cosine
waveforms of type (153) the following relation holds
minus1 le 119886119896le 1 (154)
This explains why 119896th harmonic coefficient 119886119896in Proposi-
tion 26 goes through interval [minus1 1]Waveforms (153) with 119886
1ge 0 can be obtained from
waveforms with 1198861le 0 by shifting by 120587 and therefore with-
out loss of generality we can assume that 1198861le 0
Proposition 26 Each nonnegative cosine waveform of type(153) with 119886
1le 0 and at least one zero can be represented as
119879119896(120591) = [1 minus cos 120591] [1 minus (119896 minus 1) 119886
119896minus 2119886119896
119896minus1
sum
119899=1
(119896 minus 119899) cos 119899120591]
(155)
if minus1 le 119886119896le 1(119896
2
minus 1) or
119879119896(120591) = 119886
119896[1 minus cos (120591 minus 120591
0)] [1 minus cos (120591 + 120591
0)]
sdot [1198880+ 2
119896minus2
sum
119899=1
119888119899cos 119899120591]
(156)
where
119888119899=sin ((119896 minus 119899) 120591
0) cos 120591
0minus (119896 minus 119899) cos ((119896 minus 119899) 120591
0) sin 120591
0
sin31205910
(157)
119886119896=
sin 1205910
119896 sin (1198961205910) cos 120591
0minus cos (119896120591
0) sin 120591
0
(158)
100381610038161003816100381612059101003816100381610038161003816 le
120587
119896 (159)
if 1(1198962 minus 1) le 119886119896le 1
Remark 27 Identity [1minuscos(120591minus1205910)][1minuscos(120591+120591
0)] = [cos 120591
0minus
cos 120591]2 implies that (156) can be rewritten as
119879119896(120591) = 119886
119896[cos 120591
0minus cos 120591]2 [119888
0+ 2
119896minus2
sum
119899=1
119888119899cos 119899120591] (160)
Furthermore substitution of (157) into (160) leads to
119879119896(120591) = 119886
119896[cos 120591
0minus cos 120591]
sdot [(119896 minus 1) sin 119896120591
0
sin 1205910
minus 2
119896minus1
sum
119899=1
sin ((119896 minus 119899) 1205910)
sin 1205910
cos 119899120591]
(161)
Remark 28 All nonnegative cosine waveforms of type (153)with at least one zero and 119886
1le 0 except one of them can be
represented either in form (155) or form (156)This exceptionis maximally flat cosine waveform with 119886
1lt 0 which can be
obtained from (155) for 119886119896= 1(119896
2
minus 1) or from (156) for 1205910=
0 Maximally flat cosine waveform with 1198861lt 0 can also be
obtained from (70) by setting 1205910= 0 Furthermore setting
1205910= 0 in (71) leads to maximally flat cosine waveforms for
119896 le 4 and 1198861lt 0
Remark 29 Nonnegative cosine waveform of type (155) with1198861lt 0 and minus1 lt 119886
119896le 1(119896
2
minus 1) has exactly one zero at120591 = 0 Nonnegative cosine waveform described by (156) with1198861lt 0 and 1(1198962 minus 1) lt 119886
119896lt 1 has two zeros at plusmn120591
0 where
0 lt |1205910| lt 120587119896 For 119886
119896= ∓1 nonnegative cosine waveform
of type (153) reduces to 119879119896(120591) = 1 ∓ cos 119896120591 (clearly these two
waveforms both have 119896 zeros)
Remark 30 Transformation of (155) into an additive formleads to the following relation
1198861= minus1 minus 119886
119896 (162)
where minus1 le 119886119896le 1(119896
2
minus1) Similarly transformation of (156)leads to the following relation
1198861= minus119886119896
119896 sin 1198961205910
sin 1205910
(163)
where 119886119896is given by (158) 1(1198962minus1) le 119886
119896le 1 and |120591
0| le 120587119896
Notice that coefficients of maximally flat cosine waveformnamely 119886
119896= 1(119896
2
minus1) and 1198861= minus1198962
(1198962
minus1) satisfy relation(162) They also satisfy relation (163) for 120591
0= 0
Remark 31 Nonnegative cosine waveforms of type (153) withat least one zero coincide with nonnegative cosine waveformswith maximal absolute value of coefficient 119886
1for prescribed
coefficient 119886119896
In proving that Remark 31 holds notice that expression(155) can be obtained from (121) by setting 119887
119896= 0 Fur-
thermore if 119886119896
ge 0 then 120582119896
= 119886119896 which together
with 119887119896
= 0 and (118) implies 120575 = 0 In this case119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) le 1 + 119886
119896becomes 1198962119886
119896le 1 +
119886119896 On the other hand if 119886
119896lt 0 then 120582
119896= minus119886
119896 which
together with 119887119896= 0 and (118) implies |120575| = 120587 In this case
119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) le 1 + 119886
119896becomes 0 le 1 + 119886
119896
Therefore every nonnegative cosine waveform of type (155)has maximal absolute value of coefficient 119886
1for prescribed
coefficient 119886119896 when minus1 le 119886
119896le 1(119896
2
minus 1)Let us now show that expression (156) can be obtained
from (122) by setting 119887119896= 0 and 119886
119896gt 0 For waveforms
of type (122) according to (118) 119887119896= 0 and 119886
119896gt 0 imply
120575 = 0 and 120582119896= 119886119896 Substitution of 120582
119896= 119886119896and 120575 = 0 into
119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) ge 1+119886
119896leads to 119886
119896ge 1(119896
2
minus1)Furthermore substitution of 120575 = 0 into (145) yields 120591
0= 120585119896
Insertion of 120582119896= 119886119896 120575 = 0 and 120591
0= 120585119896 into (122) leads
to (156) Therefore every nonnegative cosine waveform oftype (156) has maximal absolute value of coefficient 119886
1for
prescribed coefficient 119886119896 when 1(1198962 minus 1) le 119886
119896le 1
Proof of Proposition 26 Let us start with nonnegative cosinewaveform of type (153) with 120582
119896= |119886119896| = 1 According to
Remark 7 120582119896= |119886119896| = 1 implies that 120582
1= |1198861| = 0
Substitution of 119886119896
= minus1 into (155) and using (A2) (seeAppendices) lead to 119879
119896(120591) = 1 minus cos 119896120591 Consequently (155)
holds for 119886119896= minus1 On the other hand substitution of 119886
119896= 1
into (158) yields |1205910| = 120587119896 Furthermore substitution of
119886119896= 1 and 120591
0= 120587119896 (or 120591
0= minus120587119896) into (156) along
with performing all multiplications and using (A2) leads to
20 Mathematical Problems in Engineering
119879119896(120591) = 1 + cos 119896120591 Consequently (156)ndash(158) hold for 119886
119896= 1
and |1205910| = 120587119896
It is easy to see that 120582119896= |119886119896| lt 1 and 119879
119896(1205910) = 0 for some
1205910imply 120582
1= |1198861| = 0 Therefore in what follows we assume
that |119886119896| = 1 and 119886
1lt 0
Cosine waveforms are even functions of 120591 Therefore ifnonnegative cosine waveform has exactly one zero it has to beeither at 0 or at 120587 On the other hand if nonnegative cosinewaveform with 119886
1= 0 has exactly two zeros then these zeros
are placed at plusmn1205910 such that 120591
0is neither 0 nor 120587
In order to prove that (155) holds for minus1 lt 119886119896le 1(1minus119896
2
)let us start by referring to the description (38) of nonnegativewaveformswith at least one zero As wementioned earlier fornonnegative cosine waveformwith exactly one zero (denotedby 1205910) it is either 120591
0= 0 or 120591
0= 120587 Therefore in both cases
sin 1205910= 0 Substitution of sin 120591
0= 0 into (43) together with
1198861
= 0 and 120582119896= |119886119896| lt 1 leads to
1205910= 0 (164)
Clearly 1205910= 0 119887
1= 0 and 119887
119896= 0 according to (44) and (46)
imply 120582119896sin 120585 = 0 Since 120582
119896= |119886119896| it follows that |119886
119896| sin 120585 = 0
also holds which further implies 120582119896= 119886119896= 0 or sin 120585 =
0 In the case when 120582119896= 119886119896= 0 from (164) and (43) we
obtain 1198861= minus1 which further implies that 119879
119896(120591) = 1 minus cos 120591
Consequently (155) holds for 119886119896= 0 In the case when sin 120585 =
0 from (164) and (45) we obtain 119886119896= 120582119896if 120585 = 0 or 119886
119896= minus120582119896
if 120585 = 120587 Relations 119886119896= 120582119896and 120585 = 0 according to (40) imply
that 0 le 119886119896le 1(1 minus 119896
2
) Substitution of 120585 = 0 120582119896= 119886119896 and
(164) into (38) leads to (155) which proves that (155) holdsfor 0 le 119886
119896le 1(1 minus 119896
2
) On the other hand relations 119886119896=
minus120582119896and 120585 = 120587 according to (40) imply that minus1 lt 119886
119896le 0
Substitution of 120585 = 120587 120582119896= minus119886119896 and (164) into (38) also leads
to (155) which proves that (155) also holds for minus1 lt 119886119896le 0
Consequently (155) holds for minus1 lt 119886119896le 1(1 minus 119896
2
)In what follows we first prove that (156)-(157) hold for
1(1 minus 1198962
) lt 119886119896lt 1 For this purpose let us start with non-
negative waveforms with two zeros described by (66) As wementioned before nonnegative cosine waveforms with twozeros have zeros at 120591
0and minus120591
0 such that 120591
0= 0 and 120591
0= 120587
Relations 1198861
lt 0 and 1198871
= 0 according to (84) implycos(1205910minus 120585119896) = 1 and therefore
120585
119896= 1205910 (165)
From 120585119896 = 1205910and 0 lt |120585| lt 120587 it follows that 0 lt |120591
0| lt
120587119896 Insertion of 120585119896 = 1205910into (45) yields 119886
119896= 120582119896 Relations
119886119896= 120582119896and (82) imply that 1(1 minus 1198962) lt 119886
119896lt 1 Substitution
of 120582119896= 119886119896and 120585119896 = 120591
0into (66)ndash(68) leads to (156)ndash(158)
which proves that (156)ndash(158) hold for 1(1 minus 1198962
) lt 119886119896lt 1
and 0 lt |1205910| lt 120587119896
Finally substitution of 119886119896= 1(1 minus 119896
2
) and 1205910= 0 into
(161) leads to
119879119896(120591) =
[1 minus cos 120591](1 minus 1198962)
[119896 (119896 minus 1) minus 2
119896minus1
sum
119899=1
(119896 minus 119899) cos 119899120591]
(166)
Waveform (166) coincides with waveform (155) for 119886119896
=
1(1 minus 1198962
) which in turn proves that (156) holds for 119886119896=
1(1 minus 1198962
) and 1205910= 0 This completes the proof
62 Nonnegative CosineWaveforms with at Least One Zero for119896 = 3 In this subsection we consider nonnegative cosinewaveforms with at least one zero for 119896 = 3 (for case 119896 = 2
see [12])Cosine waveform with fundamental and third harmonic
reads
1198793(120591) = 1 + 119886
1cos 120591 + 119886
3cos 3120591 (167)
For 1198861le 0 and minus1 le 119886
3le 18 according to (155) non-
negative cosine waveform of type (167) with at least one zerocan be expressed as
1198793(120591) = (1 minus cos 120591) [1 minus 2119886
3(1 + 2 cos 120591 + cos 2120591)] (168)
From 1198793(120591 + 120587) = 2 minus 119879
3(120591) it immediately follows that for
1198861ge 0 and minus18 le 119886
3le 1 119879
3(120591) can be expressed as
1198793(120591) = (1 + cos 120591) [1 + 2119886
3(1 minus 2 cos 120591 + cos 2120591)] (169)
For 1198861le 0 and 18 le 119886
3le 1 from (158) it follows that 119886
3=
[8cos31205910]minus1 This relation along with (160) and (157) further
implies that 1198793(120591) can be expressed as
1198793(120591) =
[cos 1205910minus cos 120591]2 [2 cos 120591
0+ cos 120591]
2cos31205910
(170)
providing that |1205910| le 1205873 From 119879
3(120591 + 120587) = 2 minus 119879
3(120591) it
follows that (170) also holds for 1198861ge 0 and minus1 le 119886
3le minus18
providing that 1205910isin [21205873 41205873]
Maximally flat nonnegative cosinewaveformof type (167)with 119886
1lt 0 (minimum at 120591
0= 0) reads 119879
3(120591) = [1 minus
cos 120591]2[1 + (12) cos 120591] Dually maximally flat nonnegativecosine waveform with 119886
1gt 0 (minimum at 120591
0= 120587) reads
1198793(120591) = [1 + cos 120591]2[1 minus (12) cos 120591]In what follows we provide relations between coefficients
1198861and 1198863of nonnegative cosine waveforms of type (167) with
at least one zeroFor 1198861le 0 conversion of (168) into an additive form
immediately leads to the following relation
1198861= minus1 minus 119886
3for minus 1 le 119886
3le1
8 (171)
Conversion of (170) into an additive form leads to 1198861
=
minus31198863(2 cos 2120591
0+ 1) which can be also expressed as 119886
1=
minus31198863(4cos2120591
0minus 1) For 119886
1le 0 relations |120591
0| le 1205873 119886
1=
minus31198863(4cos2120591
0minus 1) and 119886
3= [8cos3120591
0]minus1 lead to
1198861= minus3 [ 3radic119886
3minus 1198863] for 1
8le 1198863le 1 (172)
Similarly for 1198861ge 0 conversion of (169) into an additive form
leads to the following relation
1198861= 1 minus 119886
3for minus
1
8le 1198863le 1 (173)
Mathematical Problems in Engineering 21
1
15
15
05
0
minus05
minus15minus15
minus1
minus1 minus05 0 05 1
Coefficient a1
Coe
ffici
enta
3 a1= minus
3(3radica3minus a3)
a1 =
minus1 minus a3
a1 =
1 minusa3
a1 =3(3radic|a3
| +a3)
Figure 14 Parameter space of cosine waveforms for 119896 = 3
For waveform of type (170) with 1198861ge 0 relations 120591
0isin [21205873
41205873] 1198861= minus3119886
3(4cos2120591
0minus 1) and 119886
3= [8cos3120591
0]minus1 lead to
1198861= 3 [
3radic10038161003816100381610038161198863
1003816100381610038161003816 + 1198863] for minus 1 le 1198863le minus
1
8 (174)
Every cosine waveform of type (167) corresponds to apair of real numbers (119886
1 1198863) and vice versa Points (119886
1 1198863)
in grey area in Figure 14 correspond to nonnegative cosinewaveforms for 119896 = 3 The points at the boundary of grey areacorrespond to nonnegative cosinewaveformswith at least onezero A number of shapes of nonnegative cosine waveformswith 119896 = 3 and at least one zero plotted on interval [minus120587 120587]are also presented in Figure 14 The boundary of grey area inFigure 14 consists of four line segments described by relations(171)ndash(174) The common point of line segments (172) and(173) is cusp point with coordinates 119886
1= 0 and 119886
3= 1
Another cusp point with coordinates 1198861= 0 and 119886
3= minus1
is the common point of line segments (171) and (174) Thecommon point of line segments (171)-(172) has coordinates(minus98 18) and common point of line segments (173)-(174)has coordinates (98 minus18) These points are representedby white circle dots and they correspond to maximallyflat cosine waveforms (eg see [21]) White triangle dotswith coordinates (2radic3 minusradic39) and (minus2radic3radic39) refer tothe nonnegative cosine waveforms with maximum value ofamplitude of fundamental harmonic
7 Four Case Studies of Usage of NonnegativeWaveforms in PA Efficiency Analysis
In this section we provide four case studies of usage ofdescription of nonnegative waveforms with fundamental and119896th harmonic in PA efficiency analysis In first two casestudies to be presented in Section 71 voltage is nonnegativewaveform with fundamental and second harmonic with atleast one zero In remaining two case studies to be consideredin Section 72 voltage waveform contains fundamental andthird harmonic
i
v Load
Vdc
Idc
Lch
vL
iL
Cb
+ +in
Figure 15 Generic PA circuit diagram
Let us consider generic PA circuit diagram as shown inFigure 15 We assume here that voltage and current wave-forms at the transistor output are
V (120579) = 1 + 1198861V cos 120579 + 1198871V sin 120579 + 119886119896V cos 119896120579 + 119887119896V sin 119896120579
119894 (120579) = 1 + 1198861119894cos 120579 +
infin
sum
119899=2
119886119899119894cos 119899120579
(175)
where 120579 stands for 120596119905 Both waveforms are normalized in thesense that dc components of voltage and current are 119881dc =
1 and 119868dc = 1 respectively Under assumption that blockingcapacitor 119862
119887behaves as short-circuit at the fundamental and
higher harmonics current and voltage waveforms at the loadare
V119871(120579) = 119886
1V cos 120579 + 1198871V sin 120579 + 119886119896V cos 119896120579 + 119887119896V sin 119896120579
119894119871(120579) = minus119886
1119894cos 120579 minus
infin
sum
119899=2
119886119899119894cos 119899120579
(176)
In terms of coefficients of voltage and current waveforms theload impedance at fundamental harmonic is 119911
1= minus(119886
1V minus
1198951198871V)1198861119894 whereas load impedance at 119896th harmonic is 119911
119896=
minus(119886119896Vminus119895119887119896V)119886119896119894 All other harmonics are short-circuited (119911
119899=
0 for 119899 = 1 and 119899 = 119896) Time average output power of PA (egsee [10]) with waveform pair (175) at fundamental frequencycan be expressed as
1198751= minus
11988611198941198861V
2 (177)
For normalized waveforms (175) with 119881dc = 1 and 119868dc = 1dc power is 119875dc = 1 Consequently PA efficiency 120578 = 119875
1119875dc
(eg see [10 26]) is equal to
120578 = minus11988611198941198861V
2 (178)
Thus time average output power 1198751of PA with pair of nor-
malized waveform (175) is equal to efficiency (178)Power utilization factor (PUF) is defined [26] as ldquothe
ratio of power delivered in a given situation to the power
22 Mathematical Problems in Engineering
delivered by the same device with the same supply voltagein Class A moderdquo Since the output power in class-A modeis 1198751class-A = max[V(120579)] sdotmax[119894(120579)]8 (eg see [9]) it follows
that power utilization factor PUF = 11987511198751class-A for PA with
pair of normalized waveforms (175) can be expressed as
PUF =8120578
max [V (120579)] sdotmax [119894 (120579)] (179)
71 NonnegativeWaveforms for 119896 = 2 in PAEfficiencyAnalysisIn this subsection we provide two case studies of usage ofdescription of nonnegative waveforms with fundamental andsecond harmonic (119896 = 2) in PA efficiency analysis For moreexamples of usage of descriptions of nonnegative waveformswith fundamental and second harmonic in PA efficiencyanalysis see [12]
Case Study 71 In this case study we consider efficiency of PAfor given second harmonic impedance providing that voltageis nonnegative waveform with fundamental and second har-monic and current is ldquohalf-sinerdquo waveform frequently used inefficiency analysis of classical PA operation (eg see [10])
Standard model of current waveform for classical PAoperation has the form (eg see [10 26])
119894119863(120579) =
119868119863[cos 120579 minus cos(120572
2)] |120579| le
120572
2
0120572
2le |120579| le 120587
(180)
where 120572 is conduction angle and 119868119863
gt 0 Since 119894119863(120579) is
even function it immediately follows that its Fourier seriescontains only dc component and cosine terms
119894119863(120579) = 119868dc +
infin
sum
119899=1
119868119899cos 119899120579 (181)
The dc component of the waveform (180) is
119868dc =119868119863120572
2120587[sinc(120572
2) minus cos(120572
2)] (182)
where sinc119909 = (sin119909)119909 The coefficient of the fundamentalharmonic component reads
1198681=119868119863120572
2120587(1 minus sinc120572) (183)
and the coefficient of 119899th harmonic component can bewrittenin the form
119868119899=119868119863
119899120587[sin ((119899 minus 1) 1205722)
(119899 minus 1)minussin ((119899 + 1) 1205722)
(119899 + 1)] 119899 ge 2
(184)
For ldquohalf-sinerdquo current waveform conduction angle is equalto 120587 (class-B conduction angle) According to (182) thisfurther implies that 119868dc = 119868
119863120587 To obtain normalized form of
waveform (180) we set 119868dc = 1which implies that 119868119863= 120587 Fur-
thermore substitution of 120572 = 120587 and 119868119863= 120587 in (180) leads to
119894 (120579) =
120587 cos 120579 |120579| lt120587
2
0120587
2lt |120579| le 120587
(185)
Similarly substitution of 119868119863= 120587 and 120572 = 120587 into (183) and
(184) leads to the coefficients of waveform (185) Coefficientsof fundamental and second harmonic respectively are
1198861119894=120587
2 119886
2119894=2
3 (186)
On the other hand voltage waveform of type (35) for 119896 =2 reads
V (120579) = 1 + 1198861V cos 120579 + 1198871V sin 120579 + 1198862V cos 2120579 + 1198872V sin 2120579
(187)
This waveform contains only fundamental and second har-monic and therefore all harmonics of order higher thantwo are short-circuited (119911
119899= 0 for 119899 gt 2) For current
voltage pair (185) and (187) load impedance at fundamentalharmonic is 119911
1= minus(119886
1V minus 1198951198871V)1198861119894 whereas load impedance
at second harmonic is 1199112= minus(119886
2Vminus1198951198872V)1198862119894 According to ourassumption the load is passive and therefore Re119911
1 gt 0 and
Re1199112 ge 0 which further imply 119886
11198941198861V lt 0 and 119886
21198941198862V le 0
respectivelyIt is easy to see that problem of findingmaximal efficiency
of PA with current-voltage pair (185) and (187) for prescribedsecond harmonic impedance can be reduced to the problemof finding voltage waveform of type (187) with maximal coef-ficient |119886
1V| for prescribed coefficients of second harmonic(see Section 5)
The following algorithm (analogous to Algorithm 22presented in [12]) provides the procedure for calculation ofmaximal efficiency with current-voltage pair (185) and (187)for prescribed second harmonic impedance The definitionof function atan 2(119910 119909) which appears in the step (iii) of thefollowing algorithm is given by (105)
Algorithm 32 (i) Choose 1199112= 1199032+1198951199092such that |119911
2| le 1|119886
2119894|
(ii) calculate 1198862V minus 1198951198872V = minus119911
21198862119894and 120582
2V = radic11988622V + 1198872
2V(iii) if 2120582
2V le 1 minus 1198862V then calculate 119886
1V = minus1 minus 1198862V and
1198871V = minus2119887
2V else calculate 1205821V = radic81205822V(1 minus 1205822V) 1205790V minus 120585V2 =
(12)atan2(1198872V 1198862V) 1198861V = minus120582
1V cos(1205790V minus 120585V2) and 1198871V =
minus1205821V sin(1205790V minus 120585V2)(iv) calculate efficiency 120578 = minus119886
11198941198861V2
(v) calculate 1199111= minus(119886
1V minus 1198951198871V)1198861119894 and 1199112119899 = 1199112Re119911
1
In this case study coefficients of fundamental and secondharmonic of current waveform are given by (186) Maximalefficiency of PA associated with the waveform pair (185)and (187) as a function of normalized second harmonicimpedance 119911
2119899= 1199112Re119911
1 is presented in Figure 16(a)
As can be seen from Figure 16(a) efficiency of 078 isachieved at the edge of Smith chart where second harmonicimpedance has small resistive part Corresponding PUFcalculated according to (179) is presented in Figure 16(b)Peak efficiency 120578 = 1205874 = 07854 and peak value of PUF = 1
are attained when second harmonic is short-circuited (whichcorresponds to ideal class-B operation [10 26])
For example for second harmonic impedance 1199112= 01 minus
11989505 and currentwaveform (185) fromAlgorithm32 it followsthat 2120582
2V le 1 minus 1198862V Furthermore according to step (iii)
of above algorithm maximal efficiency of PA is attained
Mathematical Problems in Engineering 23
+j2
+j5
+j1
+j2
+j5
minusj2
minusj5
minusj1
minusj2
minusj5
infin0
078
02
05
1 2 5
075 07 065
06 05
075
07
065
06
05
120578 lt 05
(a)
099
095
091
083 075
067
0 59 051
051
099
095
091
083
075 0
67 059
+j5
+j1
+j2
+j5
minusj5
minusj1
minusj2
minusj5
infin
+j2
minusj2
0 02
05
1 2 5
120578 lt 05
(b)
Figure 16 (a) Contours of maximal efficiency of PA and (b) contours of corresponding PUF associated with the waveform pair (185) and(187) as functions of normalized second harmonic impedance 119911
2119899= 1199112Re119911
1
2
3
4
1
0
2 3 410
Wav
efor
ms
Angle 120579120587
i(120579)
(120579)
Figure 17 Waveform pair (185) and (187) that provides maximalefficiency for 119911
2= 01 minus 11989505
with voltage waveform of type (187) with coefficients 1198862V =
minus00667 1198872V = minus03333 119886
1V = minus09333 and 1198871V = 06667 (see
Figure 17) Corresponding efficiency PUF and normalizedsecond harmonic impedance are 120578 = 07330 PUF = 07572and 1199112119899= 01683 minus 11989508415 respectively
On the other hand for second harmonic impedance 1199112=
01 minus 11989508 and current waveform (185) from Algorithm 32 itfollows that 2120582
2V gt 1 minus 1198862V Then according to step (iii) of
above algorithm maximal efficiency is attained with voltagewaveform of type (187) with coefficients 119886
2V = minus00667 1198872V =
minus05333 1198861V = minus09333 and 119887
1V = 10572 (see Figure 18)Efficiency PUF andnormalized secondharmonic impedanceare 120578 = 07330 PUF = 06332 and 119911
2119899= 01683 minus 11989513465
respectively
2
3
4
1
0
2 3 410
Wav
efor
ms
Angle 120579120587
i(120579)
(120579)
Figure 18 Waveform pair (185) and (187) that provides maximalefficiency for 119911
2= 01 minus 11989508
Case Study 72 As another case study let us consider the effi-ciency of PA providing that current waveform is nonnegativecosine waveform up to third harmonic with maximum valueof amplitude of fundamental harmonic [22] (see also [8])
119894 (120579) = 1 +1 + radic5
2cos 120579 + 2radic5
5cos 2120579 + 5 minus radic5
10cos 3120579
(188)
and voltage waveform is nonnegative waveform of type(187) Load impedances at fundamental second and thirdharmonic are 119911
1= minus(119886
1Vminus1198951198871V)1198861119894 1199112 = minus(1198862Vminus1198951198872V)1198862119894 and
1199113= 0 respectively According to our assumption the load
is passive and therefore Re1199111 gt 0 and Re119911
2 ge 0 which
further imply 11988611198941198861V lt 0 and 119886
21198941198862V le 0 respectively
Because current waveform (188) contains only cosineterms and voltage waveform is the same as in previous case
24 Mathematical Problems in Engineering
08
075
07
06
06
065
065
05
05
075 0
7
120578 lt 05
+j2
+j5
+j1
+j2
+j5
minusj2
minusj5
minusj1
minusj2
minusj5
infin0
02
05
1 2 5
(a)
085
08 0
75 07 0
6
05 04
08
075
0405
060
7
120578 lt 05
+j2
+j5
+j1
+j2
+j5
minusj2
minusj5
minusj1
minusj2
minusj5
infin0 02
05
1 2 5
(b)
Figure 19 (a) Contours of maximal efficiency of PA and (b) contours of corresponding PUF associated with the waveform pair (187)-(188)as functions of normalized second harmonic impedance 119911
2119899= 1199112Re119911
1
study the procedure for calculation of maximal efficiency ofPA with waveform pair (187)-(188) is the same as presentedin Algorithm 32 In this case study the coefficients of funda-mental and second harmonic of current waveform are 119886
1119894=
(1 + radic5)2 and 1198862119894= 2radic55 respectively
Maximal efficiency of PA associated with the waveformpair (187)-(188) as a function of normalized secondharmonicimpedance 119911
2119899= 1199112Re119911
1 is presented in Figure 19(a)
Efficiency of 08 is achieved at the edge of Smith chart wheresecond harmonic impedance has small resistive part Thetheoretical upper bound 120578 = (1 + radic5)4 asymp 08090 isattained when second harmonic is short-circuitedWhen thisupper bound is reached both second and third harmonic areshort-circuited which implies that we are dealing with finiteharmonic class-C [6 8] or dually when current and voltageinterchange their roles with finite harmonic inverse class-C[6 9] Corresponding PUF calculated according to (179) ispresented in Figure 19(b) Peak value of PUF asymp 08541 isattained when second harmonic is short-circuited
For example for second harmonic impedance 1199112
=
007 minus 11989504 and current waveform (188) from Algorithm 32 itfollows that 2120582
2V le 1 minus 1198862V Furthermore according to step
(iii) of Algorithm 32 maximal efficiency of PA is attainedwith voltage waveform of type (187) with coefficients 119886
2V =
minus00626 1198872V = minus03578 119886
1V = minus09374 and 1198871V = 07155 (see
Figure 20) Corresponding efficiency PUF and normalizedsecond harmonic impedance are 120578 = 07584 PUF = 06337and 1199112119899= 01208 minus 11989506904 respectively
On the other hand for 1199112= 005 minus 11989507 and current
waveform (187) it follows that 21205822V gt 1minus119886
2VThen accordingto step (iii) of Algorithm 32 themaximal efficiency is attainedwith voltage waveform of type (187) with coefficients 119886
2V =
minus00447 1198872V = minus06261 119886
1V = minus09318 and 1198871V = 10007 (see
Figure 21) Efficiency PUF and normalized second harmonic
2
3
4
1
0
2 3 410
Wav
efor
ms
Angle 120579120587
i(120579)
(120579)
Figure 20 Waveform pair (187)-(188) that provides maximal effi-ciency for 119911
2= 007 minus 11989504
impedance are 120578 = 07538 PUF = 05314 and 1199112119899= 00868minus
11989512156 respectively
72 Nonnegative Waveforms for 119896 = 3 in PA EfficiencyAnalysis In this subsection we provide another two casestudies of usage of description of nonnegative waveforms inPA efficiency analysis this time with fundamental and thirdharmonic (119896 = 3)
Case Study 73 Let us consider current-voltage pair such thatvoltage is nonnegative waveform with fundamental and thirdharmonic
V (120579) = 1 + 1198861V cos 120579 + 1198871V sin 120579 + 1198863V cos 3120579 + 1198873V sin 3120579
(189)
Mathematical Problems in Engineering 25
2
3
4
1
0
2 3 410
Wav
efor
ms
Angle 120579120587
i(120579)
(120579)
Figure 21 Waveform pair (187)-(188) that provides maximal effi-ciency for 119911
2= 005 minus 11989507
and current is nonnegative cosine waveform given by (188)Load impedances at fundamental second and third har-monic are 119911
1= minus(119886
1V minus 1198951198871V)1198861119894 1199112 = 0 and 119911
3= minus(119886
3V minus
1198951198873V)1198863119894 respectively According to our assumption the load
is passive and therefore Re1199111 gt 0 and Re119911
3 ge 0 which
further imply 11988611198941198861V lt 0 and 119886
31198941198863V le 0
In this subsection we consider the problem of findingmaximal efficiency of PA with waveform pair (188)-(189)for given third harmonic impedance As we mentionedearlier problem of finding maximal efficiency of PA withcurrent-voltage pair (188)-(189) for prescribed third har-monic impedance can be reduced to the problem of findingvoltage waveform of type (189) withmaximal coefficient |119886
1V|for prescribed coefficients of third harmonic (see Section 52)
The following algorithm provides the procedure forcalculation of maximal efficiency with current-voltage pair(188)-(189) The definition of function atan 2(119910 119909) whichappears in step (iii) of the following algorithm is given by(105)
Algorithm 33 (i) Choose 1199113= 1199033+1198951199093such that |119911
3| le 1|119886
3119894|
(ii) calculate 1198863V minus 1198951198873V = minus119911
31198863119894and 120582
3V = radic11988623V + 1198872
3V(iii) if 271205822
3V le (1 minus 21198863V)3 then calculate 119886
1V = minus1 minus 1198863V
and 1198871V = minus3119887
3V else calculate 1205821V = 3(3radic1205823V minus 120582
3V) 1205790V minus120585V3 = (13)atan 2(119887
3V 1198863V) 1198861V = minus1205821V cos(1205790V minus 120585V3) and
1198871V = minus120582
1V sin(1205790V minus 120585V3)(iv) calculate efficiency 120578 = minus119886
11198941198861V2
(v) calculate 1199111= minus(119886
1V minus 1198951198871V)1198861119894 and 1199113119899 = 1199113Re119911
1
In this case study coefficients of fundamental and thirdharmonic of current waveform are 119886
1119894= (1 + radic5)2 and
1198863119894= (5 minus radic5)10 respectively For the waveform pair (188)-
(189) maximal efficiency of PA as a function of normalizedthird harmonic impedance 119911
3119899= 1199113Re119911
1 is presented in
Figure 22 Efficiency of 08 is reached when third harmonicimpedance has small resistive part Peak efficiency 120578 = (1 +
radic5)4 asymp 08090 is achieved when third harmonic is short-circuited
+j2
+j5
+j1
+j2
+j5
minusj2
minusj5
minusj1
minusj2
minusj5
infin0
08
08
075
075
07
07
06
05
02
05
1 2
Figure 22 Contours of maximal efficiency of PA associated withthe waveform pair (188)-(189) as a function of normalized thirdharmonic impedance 119911
3119899= 1199113Re119911
1
For the present case study in what follows we show thatpower utilization factor is proportional to efficiency For volt-age waveform of type (189) it is easy to see that V(120579 + 120587) =
2minusV(120579) holdsThis relation along with the fact that waveformV(120579) that provides maximal efficiency has at least one zeroimplies that max[V(120579)] = 2 On the other hand currentwaveform (188) is cosine waveform with positive coefficientsand therefore max[119894(120579)] = 119894(0) = 2 + 4radic5 Consequentlyaccording to (179) the following relation holds
PUFCase study 73 = 2 (5 minus 2radic5) 120578 = 10557120578 (190)
Clearly the ratio PUF120578 is constant and therefore in this casestudy PUF can be easily calculated from the correspondingefficiency Accordingly peak efficiency and peak value ofPUFCase study 73 = 3radic52 minus 52 = 08541 are attained forthe same voltage waveform (when third harmonic is short-circuited)
In the first example current waveform (188) and 1199113=
02 minus 11989505 imply that 2712058223V le (1 minus 2119886
3V)3 Then according
to Algorithm 33 the voltage waveform of type (189) thatprovides maximal efficiency has the following coefficients1198863V = minus00553 119887
3V = minus01382 1198861V = minus09447 and 119887
1V =
04146 (see Figure 23) Efficiency PUF and normalized thirdharmonic impedance are 120578 = 07643 PUF = 08069 and1199113119899= 03425 minus 11989508564 respectivelyIn the second example current waveform (188) and 119911
3=
01 minus 11989511 imply that 2712058223V gt (1 minus 2119886
3V)3 Then according
to Algorithm 33 the voltage waveform of type (189) thatprovides maximal efficiency has the following coefficients1198863V = minus00276 119887
3V = minus03040 1198861V = minus09391 and 119887
1V =
05807 (see Figure 24) Efficiency PUF and normalized thirdharmonic impedance are 120578 = 07598 PUF = 08021 and1199113119899= 01723 minus 11989518952 respectively
26 Mathematical Problems in Engineering
2
3
4
1
0
2 3 410
Wav
efor
ms
Angle 120579120587
i(120579)
(120579)
Figure 23 Waveform pair (188)-(189) that provides maximal effi-ciency for 119911
3= 02 minus 11989505
2
3
4
1
0
2 3 410
Wav
efor
ms
Angle 120579120587
i(120579)
(120579)
Figure 24 Waveform pair (188)-(189) that provides maximal effi-ciency for 119911
3= 01 minus 11989511
Case Study 74 In this case study let us consider current-voltage pair where current is normalized waveform of type(180) with conduction angle 120572 = 115120587 (207∘) and voltageis nonnegative waveform of type (189) Substitution of 120572 =
115120587 and 119868dc = 1 into (182) leads to 119868119863
= 22535Furthermore substitution of 120572 = 115120587 and 119868
119863= 22535 into
(180) leads to
119894 (120579)
=
22535 [cos 120579 minus cos(1151205872
)] |120579| le115120587
2
0115120587
2le |120579| le 120587
(191)
Similarly substitution of 120572 = 115120587 and 119868119863
= 22535 into(183) and (184) for 119899 = 3 yields coefficients of fundamentaland third harmonic of waveform (191)
1198861119894= 14586 119886
3119894= minus01026 (192)
Because current waveform (191) contains only cosineterms and voltage waveform is the same as in previous case
+j2
+j5
+j1
+j2
+j5
minusj2
minusj5
minusj1
minusj2
minusj5
infin0 02
05
1 2
074
076
08
084
082
078
076
074
Figure 25 Contours of maximal efficiency of PA associated withthe waveform pair (189) and (191) as a function of normalized thirdharmonic impedance 119911
3119899= 1199113Re119911
1
study the procedure for calculation of maximal efficiencyof PA with waveform pair (189)ndash(191) is the same as thatpresented in Algorithm 33 In this case study the coefficientsof fundamental and third harmonic of current waveform aregiven by (192)
For the waveform pair (189) and (191) maximal efficiencyof PA as a function of normalized third harmonic impedance1199113119899= 1199113Re119911
1 is presented in Figure 25 Efficiency of 084
is obtained in vicinity of 1199113119899= 23685 (corresponding to 119911
3=
18750) Peak efficiency 120578 asymp 08421 is achieved for voltagewaveform of type (189) with coefficients 119886
1V = minus2radic3 1198863V =
radic39 and 1198871V = 1198873V = 0
In the course of finding power utilization factor noticethat currentwaveformof type (191) attains itsmaximumvaluefor 120579 = 0 Insertion of max[119894(120579)] = 119894(0) = 278 andmax[V(120579)] = 2 for voltage waveform of type (189) into (179)leads to
PUFCase study 74 = 1439120578 (193)
Again the ratio PUF120578 is constant and PUF can be easilycalculated from the corresponding efficiency Accordinglypeak value of PUFCase study 74 asymp 12118 and peak efficiencyare attained for the same voltage waveform
In the first example current waveform (191) and 1199113=
1 minus 11989502 imply that 2712058223V le (1 minus 2119886
3V)3 Then according to
Algorithm 33 voltage waveform of type (189) which providesmaximal efficiency has coefficients 119886
3V = 01026 1198873V =
00205 1198861V = minus11026 and 119887
1V = minus00616 (see Figure 26)Efficiency PUF and normalized third harmonic impedanceare 120578 = 08042 PUF = 11572 and 119911
3119899= 13228 minus 11989502646
respectivelyIn second example current waveform (191) and 119911
3=
15 minus 11989512 imply that 2712058223V gt (1 minus 2119886
3V)3 Then according to
Algorithm 33 voltage waveform of type (189) which providesmaximal efficiency has coefficients 119886
3V = 01540 1198873V =
Mathematical Problems in Engineering 27
2
3
4
1
0
2 3 410
Wav
efor
ms
Angle 120579120587
i(120579)
(120579)
Figure 26 Waveform pair (189) and (191) that provides maximalefficiency for 119911
3= 1 minus 11989502
2
3
4
1
0
2 3 410
Wav
efor
ms
Angle 120579120587
i(120579)
(120579)
Figure 27 Waveform pair (189) and (191) that provides maximalefficiency for 119911
3= 15 minus 11989512
01232 1198861V = minus11255 and 119887
1V = minus02575 (see Figure 27)Efficiency PUF and normalized third harmonic impedanceare 120578 = 08208 PUF = 11812 and 119911
3119899= 19439 minus 11989515552
respectively
8 Conclusion
In this paper we consider a problem of finding general de-scriptions of various classes of nonnegative waveforms withfundamental and 119896th harmonic These classes include non-negative waveforms with at least one zero nonnegative wave-forms with maximal amplitude of fundamental harmonic forprescribed amplitude of 119896th harmonic nonnegative wave-forms withmaximal coefficient of cosine part of fundamentalharmonic for prescribed coefficients of 119896th harmonic andnonnegative cosine waveforms with at least one zero Mainresults are stated in six propositions (Propositions 1 6 918 22 and 26) four corollaries (Corollaries 2ndash5) twentyremarks and three algorithms Four case studies of usageof closed form descriptions of nonnegative waveforms in PAefficiency analysis are considered in detail in Section 7
Appendices
Here we provide a list of finite sums of trigonometric func-tions used in this paper (Appendix A) and brief account ofthe Chebyshev polynomials (Appendix B)
A List of Some Finite Sums ofTrigonometric Functions
Dirichlet kernel (eg see [27]) is as follows
119863119896minus1
(120591) = 1 + 2
119896minus1
sum
119899=1
cos 119899120591 = sin ((2119896 minus 1) 1205912)sin (1205912)
(A1)
Fejer kernel (eg see [27]) can be expressed in the fol-lowing equivalent forms
119865119896minus1
(120591) =1
119896
119896minus1
sum
119899=0
119863119899(120591) = 1 +
2
119896
119896minus1
sum
119899=1
(119896 minus 119899) cos 119899120591
=(1 minus cos 119896120591)119896 (1 minus cos 120591)
(A2)
Lagrangersquos trigonometric identity (eg see [28]) is as fol-lows
1198781(120591) =
119896minus1
sum
119899=1
sin 119899120591 = sin (1198961205912) sin ((119896 minus 1) 1205912)sin (1205912)
(A3)
In what follows we show that the following three trigono-metric identities also hold
2
119896minus1
sum
119899=1
(119896 minus 119899) sin 119899120591 = 119896 sin 120591 minus sin 1198961205911 minus cos 120591
(A4)
119896minus1
sum
119899=1
cos (119896 minus 2119899) 120591 = sin (119896 minus 1) 120591sin 120591
(A5)
119896minus1
sum
119899=1
119899 (119896 minus 119899) cos (119896 minus 2119899) 120591
=sin (119896120591) cos 120591 minus 119896 cos (119896120591) sin 120591
2sin3120591
(A6)
Denote 1198782(120591) = 2sum
119896minus1
119899=1(119896 minus 119899) sin 119899120591 119878
3(120591) = sum
119896minus1
119899=1cos(119896 minus
2119899)120591 and 1198784(120591) = sum
119896minus1
119899=1119899(119896 minus 119899) cos(119896 minus 2119899)120591
Notice that 1198782(120591) = 2119896119878
1(120591) + 119889119863
119896minus1(120591)119889120591 which
immediately leads to (A4)
28 Mathematical Problems in Engineering
Identity (A5) can be obtained as follows
sin (119896 minus 1) 120591sin 120591
=119890119895(119896minus1)120591
minus 119890minus119895(119896minus1)120591
119890119895120591 minus 119890minus119895120591
= 119890119895119896120591
119890minus2119895120591
minus 119890minus2119895(119896minus1)120591
1 minus 119890minus2119895120591
= 119890119895119896120591
119896minus1
sum
119899=1
119890minus2119895119899120591
=
119896minus1
sum
119899=1
119890119895(119896minus2119899)120591
=
119896minus1
sum
119899=1
cos (119896 minus 2119899) 120591
(A7)
From 4119899(119896 minus 119899) = 1198962
minus (119896 minus 2119899)2 it follows that 4119878
4(120591) =
1198962
1198783(120591) + 119889
2
1198783(120591)119889120591
2 which leads to (A6)
B The Chebyshev Polynomials
The Chebyshev polynomials of the first kind 119881119899(119909) can be
defined by the following relation (eg see [29])
119881119899(119909) = cos 119899120591 when 119909 = cos 120591 (B1)
The Chebyshev polynomials of the second kind 119880119899(119909) can be
defined by the following relation (eg see [29])
119880119899(119909) =
sin (119899 + 1) 120591sin 120591
when 119909 = cos 120591 (B2)
The Chebyshev polynomials satisfy the following recur-rence relations (eg see [29])
1198810(119909) = 1 119881
1(119909) = 119909
119881119899+1
(119909) = 2119909119881119899(119909) minus 119881
119899minus1(119909)
1198800(119909) = 1 119880
1(119909) = 2119909
119880119899+1
(119909) = 2119909119880119899(119909) minus 119880
119899minus1(119909)
(B3)
The first few Chebyshev polynomials of the first and secondkind are 119881
2(119909) = 2119909
2
minus 1 1198813(119909) = 4119909
3
minus 3119909 1198814(119909) = 8119909
4
minus
81199092
+ 1 1198802(119909) = 4119909
2
minus 1 1198803(119909) = 8119909
3
minus 4119909 and 1198804(119909) =
161199094
minus 121199092
+ 1
Conflict of Interests
The authors declare that there is no conflict of interestsregarding the publication of this paper
Acknowledgment
This work is supported by the SerbianMinistry of EducationScience and Technology Development as a part of ProjectTP32016
References
[1] V I Arnolrsquod V S Afrajmovich Y S Ilrsquoyashenko and L PShilrsquonikov Dynamical Systems V Bifurcation Theory and Catas-trophe Theory Springer Berlin Germany 1994
[2] E Polak ldquoOn the mathematical foundations of nondifferen-tiable optimization in engineering designrdquo SIAM Review vol29 no 1 pp 21ndash89 1987
[3] N S Fuzik ldquoBiharmonic modes of a tuned RF power amplifierrdquoRadiotehnika vol 25 no 7 pp 62ndash71 1970 (Russian)
[4] P Colantonio F Giannini G Leuzzi and E Limiti ldquoClass Gapproach for low-voltage high-efficiency PA designrdquo Interna-tional Journal of RF and Microwave Computer-Aided Engineer-ing vol 10 no 6 pp 366ndash378 2000
[5] F H Raab ldquoMaximum efficiency and output of class-F poweramplifiersrdquo IEEE Transactions on Microwave Theory and Tech-niques vol 49 no 6 pp 1162ndash1166 2001
[6] FH Raab ldquoClass-E class-C and class-F power amplifiers basedupon a finite number of harmonicsrdquo IEEE Transactions onMicrowaveTheory and Techniques vol 49 no 8 pp 1462ndash14682001
[7] J D Rhodes ldquoOutput universality inmaximum efficiency linearpower amplifiersrdquo International Journal of Circuit Theory andApplications vol 31 no 4 pp 385ndash405 2003
[8] A Juhas and L A Novak ldquoComments on lsquoClass-E class-Cand classF power amplifier based upon a finite number ofharmonicsrsquordquo IEEE Transactions on Microwave Theory and Tech-niques vol 57 no 6 pp 1623ndash1625 2009
[9] M Roberg and Z Popovic ldquoAnalysis of high-efficiency poweramplifiers with arbitrary output harmonic terminationsrdquo IEEETransactions on Microwave Theory and Techniques vol 59 no8 pp 2037ndash2048 2011
[10] A Grebennikov N O Sokal and M J Franco Switchmode RFPower Amplifiers ElsevierAcademic Press San Diego CalifUSA 2nd edition 2012
[11] T Canning P J Tasker and S C Cripps ldquoContinuous modepower amplifier design using harmonic clipping contourstheory and practicerdquo IEEE Transactions on Microwave Theoryand Techniques vol 62 no 1 pp 100ndash110 2014
[12] A Juhas and L A Novak ldquoGeneral description of nonnegativewaveforms up to second harmonic for power amplifier mod-ellingrdquoMathematical Problems in Engineering vol 2014 ArticleID 709762 18 pages 2014
[13] V I Arnolrsquod V V Goryunov O V Lyashko and V A VasilrsquoevDynamical Systems VIIImdashSingularity Theory II ApplicationsSpringer Berlin Germany 1993
[14] D Siersma ldquoProperties of conflict sets in the planerdquo BanachCenter Publications Polish Academy of Sciences vol 50 no 1 pp267ndash276 1999 Proceedings of the Banach Center Symposiumon Geometry and Topology of Caustics (Caustics rsquo98) WarsawPoland
[15] M van Manen The geometry of conflict sets [Dissertation]Universiteit Utrecht UtrechtThe Netherlands 2003 httpigi-tur-archivelibraryuunldissertations2003-0912-123058c4pdf
[16] Y L Sachkov ldquoMaxwell strata and symmetries in the problemofoptimal rolling of a sphere over a planerdquo Sbornik Mathematicsvol 201 no 7-8 pp 1029ndash1051 2010
[17] I A Bogaevsky ldquoPerestroikas of shocks and singularities ofminimum functionsrdquoPhysicaDNonlinear Phenomena vol 173no 1-2 pp 1ndash28 2002
[18] Y L Sachkov ldquoMaxwell strata in the Euler elastic problemrdquoJournal of Dynamical andControl Systems vol 14 no 2 pp 169ndash234 2008
[19] M Siino and T Koike ldquoTopological classification of black holesgeneric Maxwell set and crease set of a horizonrdquo InternationalJournal ofModern Physics D Gravitation Astrophysics Cosmol-ogy vol 20 no 6 pp 1095ndash1122 2011
Mathematical Problems in Engineering 29
[20] F H Raab ldquoClass-F power amplifiers with maximally flat wave-formsrdquo IEEETransactions onMicrowaveTheory and Techniquesvol 45 no 11 pp 2007ndash2012 1997
[21] A Juhas and L A Novak ldquoMaximally flat waveforms with finitenumber of harmonics in class-F power amplifiersrdquo Mathemat-ical Problems in Engineering vol 2013 Article ID 169590 9pages 2013
[22] L Fejer ldquoUber trigonometrische polynomerdquo Journal fur dieReine und Angewandte Mathematik vol 1916 no 146 pp 53ndash82 1916 (German)
[23] S C Cripps ldquoBessel Waives [microwave bytes]rdquo IEEE Micro-wave Magazine vol 10 no 7 pp 30ndash36 117 2009
[24] L N Bryzgalova ldquoSingularities of the maximum of para-metrically dependent functionrdquo Functional Analysis and ItsApplications vol 11 no 1 pp 49ndash51 1977
[25] V I Arnold A A Davydov V A Vassiliev and V MZakalyukin Mathematical Models of Catastrophes Control ofCatastrophic Process Encyclopedia of Life Support Systems(EOLSS) EOLSS Publishers Oxford UK 2006
[26] S C Cripps RF Power Amplifiers for Wireless CommunicationsArtech House Norwood Mass USA 2nd edition 2006
[27] A Zygmund Trigonometric Series vol 1 Cambridge UniversityPress Cambridge UK 2nd edition 1959
[28] A Jeffrey and H DaiHandbook of Mathematical Formulas andIntegrals ElsevierAcademic Press San Diego Calif USA 4thedition 2008
[29] J C Mason and D C Handscomb Chebyshev PolynomialsChapman amp Hall CRC Press Boca Raton Fla USA 2003
Submit your manuscripts athttpwwwhindawicom
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical Problems in Engineering
Hindawi Publishing Corporationhttpwwwhindawicom
Differential EquationsInternational Journal of
Volume 2014
Applied MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Probability and StatisticsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical PhysicsAdvances in
Complex AnalysisJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
OptimizationJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
CombinatoricsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Operations ResearchAdvances in
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Function Spaces
Abstract and Applied AnalysisHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of Mathematics and Mathematical Sciences
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Algebra
Discrete Dynamics in Nature and Society
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Decision SciencesAdvances in
Discrete MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom
Volume 2014 Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Stochastic AnalysisInternational Journal of
Mathematical Problems in Engineering 19
of cosine waveform of type (153) are 1205821= |1198861| and 120582
119896=
|119886119896| respectively According to (42) for nonnegative cosine
waveforms of type (153) the following relation holds
minus1 le 119886119896le 1 (154)
This explains why 119896th harmonic coefficient 119886119896in Proposi-
tion 26 goes through interval [minus1 1]Waveforms (153) with 119886
1ge 0 can be obtained from
waveforms with 1198861le 0 by shifting by 120587 and therefore with-
out loss of generality we can assume that 1198861le 0
Proposition 26 Each nonnegative cosine waveform of type(153) with 119886
1le 0 and at least one zero can be represented as
119879119896(120591) = [1 minus cos 120591] [1 minus (119896 minus 1) 119886
119896minus 2119886119896
119896minus1
sum
119899=1
(119896 minus 119899) cos 119899120591]
(155)
if minus1 le 119886119896le 1(119896
2
minus 1) or
119879119896(120591) = 119886
119896[1 minus cos (120591 minus 120591
0)] [1 minus cos (120591 + 120591
0)]
sdot [1198880+ 2
119896minus2
sum
119899=1
119888119899cos 119899120591]
(156)
where
119888119899=sin ((119896 minus 119899) 120591
0) cos 120591
0minus (119896 minus 119899) cos ((119896 minus 119899) 120591
0) sin 120591
0
sin31205910
(157)
119886119896=
sin 1205910
119896 sin (1198961205910) cos 120591
0minus cos (119896120591
0) sin 120591
0
(158)
100381610038161003816100381612059101003816100381610038161003816 le
120587
119896 (159)
if 1(1198962 minus 1) le 119886119896le 1
Remark 27 Identity [1minuscos(120591minus1205910)][1minuscos(120591+120591
0)] = [cos 120591
0minus
cos 120591]2 implies that (156) can be rewritten as
119879119896(120591) = 119886
119896[cos 120591
0minus cos 120591]2 [119888
0+ 2
119896minus2
sum
119899=1
119888119899cos 119899120591] (160)
Furthermore substitution of (157) into (160) leads to
119879119896(120591) = 119886
119896[cos 120591
0minus cos 120591]
sdot [(119896 minus 1) sin 119896120591
0
sin 1205910
minus 2
119896minus1
sum
119899=1
sin ((119896 minus 119899) 1205910)
sin 1205910
cos 119899120591]
(161)
Remark 28 All nonnegative cosine waveforms of type (153)with at least one zero and 119886
1le 0 except one of them can be
represented either in form (155) or form (156)This exceptionis maximally flat cosine waveform with 119886
1lt 0 which can be
obtained from (155) for 119886119896= 1(119896
2
minus 1) or from (156) for 1205910=
0 Maximally flat cosine waveform with 1198861lt 0 can also be
obtained from (70) by setting 1205910= 0 Furthermore setting
1205910= 0 in (71) leads to maximally flat cosine waveforms for
119896 le 4 and 1198861lt 0
Remark 29 Nonnegative cosine waveform of type (155) with1198861lt 0 and minus1 lt 119886
119896le 1(119896
2
minus 1) has exactly one zero at120591 = 0 Nonnegative cosine waveform described by (156) with1198861lt 0 and 1(1198962 minus 1) lt 119886
119896lt 1 has two zeros at plusmn120591
0 where
0 lt |1205910| lt 120587119896 For 119886
119896= ∓1 nonnegative cosine waveform
of type (153) reduces to 119879119896(120591) = 1 ∓ cos 119896120591 (clearly these two
waveforms both have 119896 zeros)
Remark 30 Transformation of (155) into an additive formleads to the following relation
1198861= minus1 minus 119886
119896 (162)
where minus1 le 119886119896le 1(119896
2
minus1) Similarly transformation of (156)leads to the following relation
1198861= minus119886119896
119896 sin 1198961205910
sin 1205910
(163)
where 119886119896is given by (158) 1(1198962minus1) le 119886
119896le 1 and |120591
0| le 120587119896
Notice that coefficients of maximally flat cosine waveformnamely 119886
119896= 1(119896
2
minus1) and 1198861= minus1198962
(1198962
minus1) satisfy relation(162) They also satisfy relation (163) for 120591
0= 0
Remark 31 Nonnegative cosine waveforms of type (153) withat least one zero coincide with nonnegative cosine waveformswith maximal absolute value of coefficient 119886
1for prescribed
coefficient 119886119896
In proving that Remark 31 holds notice that expression(155) can be obtained from (121) by setting 119887
119896= 0 Fur-
thermore if 119886119896
ge 0 then 120582119896
= 119886119896 which together
with 119887119896
= 0 and (118) implies 120575 = 0 In this case119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) le 1 + 119886
119896becomes 1198962119886
119896le 1 +
119886119896 On the other hand if 119886
119896lt 0 then 120582
119896= minus119886
119896 which
together with 119887119896= 0 and (118) implies |120575| = 120587 In this case
119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) le 1 + 119886
119896becomes 0 le 1 + 119886
119896
Therefore every nonnegative cosine waveform of type (155)has maximal absolute value of coefficient 119886
1for prescribed
coefficient 119886119896 when minus1 le 119886
119896le 1(119896
2
minus 1)Let us now show that expression (156) can be obtained
from (122) by setting 119887119896= 0 and 119886
119896gt 0 For waveforms
of type (122) according to (118) 119887119896= 0 and 119886
119896gt 0 imply
120575 = 0 and 120582119896= 119886119896 Substitution of 120582
119896= 119886119896and 120575 = 0 into
119896120582119896[sin 120575 sin(120575119896)] cos(120575119896) ge 1+119886
119896leads to 119886
119896ge 1(119896
2
minus1)Furthermore substitution of 120575 = 0 into (145) yields 120591
0= 120585119896
Insertion of 120582119896= 119886119896 120575 = 0 and 120591
0= 120585119896 into (122) leads
to (156) Therefore every nonnegative cosine waveform oftype (156) has maximal absolute value of coefficient 119886
1for
prescribed coefficient 119886119896 when 1(1198962 minus 1) le 119886
119896le 1
Proof of Proposition 26 Let us start with nonnegative cosinewaveform of type (153) with 120582
119896= |119886119896| = 1 According to
Remark 7 120582119896= |119886119896| = 1 implies that 120582
1= |1198861| = 0
Substitution of 119886119896
= minus1 into (155) and using (A2) (seeAppendices) lead to 119879
119896(120591) = 1 minus cos 119896120591 Consequently (155)
holds for 119886119896= minus1 On the other hand substitution of 119886
119896= 1
into (158) yields |1205910| = 120587119896 Furthermore substitution of
119886119896= 1 and 120591
0= 120587119896 (or 120591
0= minus120587119896) into (156) along
with performing all multiplications and using (A2) leads to
20 Mathematical Problems in Engineering
119879119896(120591) = 1 + cos 119896120591 Consequently (156)ndash(158) hold for 119886
119896= 1
and |1205910| = 120587119896
It is easy to see that 120582119896= |119886119896| lt 1 and 119879
119896(1205910) = 0 for some
1205910imply 120582
1= |1198861| = 0 Therefore in what follows we assume
that |119886119896| = 1 and 119886
1lt 0
Cosine waveforms are even functions of 120591 Therefore ifnonnegative cosine waveform has exactly one zero it has to beeither at 0 or at 120587 On the other hand if nonnegative cosinewaveform with 119886
1= 0 has exactly two zeros then these zeros
are placed at plusmn1205910 such that 120591
0is neither 0 nor 120587
In order to prove that (155) holds for minus1 lt 119886119896le 1(1minus119896
2
)let us start by referring to the description (38) of nonnegativewaveformswith at least one zero As wementioned earlier fornonnegative cosine waveformwith exactly one zero (denotedby 1205910) it is either 120591
0= 0 or 120591
0= 120587 Therefore in both cases
sin 1205910= 0 Substitution of sin 120591
0= 0 into (43) together with
1198861
= 0 and 120582119896= |119886119896| lt 1 leads to
1205910= 0 (164)
Clearly 1205910= 0 119887
1= 0 and 119887
119896= 0 according to (44) and (46)
imply 120582119896sin 120585 = 0 Since 120582
119896= |119886119896| it follows that |119886
119896| sin 120585 = 0
also holds which further implies 120582119896= 119886119896= 0 or sin 120585 =
0 In the case when 120582119896= 119886119896= 0 from (164) and (43) we
obtain 1198861= minus1 which further implies that 119879
119896(120591) = 1 minus cos 120591
Consequently (155) holds for 119886119896= 0 In the case when sin 120585 =
0 from (164) and (45) we obtain 119886119896= 120582119896if 120585 = 0 or 119886
119896= minus120582119896
if 120585 = 120587 Relations 119886119896= 120582119896and 120585 = 0 according to (40) imply
that 0 le 119886119896le 1(1 minus 119896
2
) Substitution of 120585 = 0 120582119896= 119886119896 and
(164) into (38) leads to (155) which proves that (155) holdsfor 0 le 119886
119896le 1(1 minus 119896
2
) On the other hand relations 119886119896=
minus120582119896and 120585 = 120587 according to (40) imply that minus1 lt 119886
119896le 0
Substitution of 120585 = 120587 120582119896= minus119886119896 and (164) into (38) also leads
to (155) which proves that (155) also holds for minus1 lt 119886119896le 0
Consequently (155) holds for minus1 lt 119886119896le 1(1 minus 119896
2
)In what follows we first prove that (156)-(157) hold for
1(1 minus 1198962
) lt 119886119896lt 1 For this purpose let us start with non-
negative waveforms with two zeros described by (66) As wementioned before nonnegative cosine waveforms with twozeros have zeros at 120591
0and minus120591
0 such that 120591
0= 0 and 120591
0= 120587
Relations 1198861
lt 0 and 1198871
= 0 according to (84) implycos(1205910minus 120585119896) = 1 and therefore
120585
119896= 1205910 (165)
From 120585119896 = 1205910and 0 lt |120585| lt 120587 it follows that 0 lt |120591
0| lt
120587119896 Insertion of 120585119896 = 1205910into (45) yields 119886
119896= 120582119896 Relations
119886119896= 120582119896and (82) imply that 1(1 minus 1198962) lt 119886
119896lt 1 Substitution
of 120582119896= 119886119896and 120585119896 = 120591
0into (66)ndash(68) leads to (156)ndash(158)
which proves that (156)ndash(158) hold for 1(1 minus 1198962
) lt 119886119896lt 1
and 0 lt |1205910| lt 120587119896
Finally substitution of 119886119896= 1(1 minus 119896
2
) and 1205910= 0 into
(161) leads to
119879119896(120591) =
[1 minus cos 120591](1 minus 1198962)
[119896 (119896 minus 1) minus 2
119896minus1
sum
119899=1
(119896 minus 119899) cos 119899120591]
(166)
Waveform (166) coincides with waveform (155) for 119886119896
=
1(1 minus 1198962
) which in turn proves that (156) holds for 119886119896=
1(1 minus 1198962
) and 1205910= 0 This completes the proof
62 Nonnegative CosineWaveforms with at Least One Zero for119896 = 3 In this subsection we consider nonnegative cosinewaveforms with at least one zero for 119896 = 3 (for case 119896 = 2
see [12])Cosine waveform with fundamental and third harmonic
reads
1198793(120591) = 1 + 119886
1cos 120591 + 119886
3cos 3120591 (167)
For 1198861le 0 and minus1 le 119886
3le 18 according to (155) non-
negative cosine waveform of type (167) with at least one zerocan be expressed as
1198793(120591) = (1 minus cos 120591) [1 minus 2119886
3(1 + 2 cos 120591 + cos 2120591)] (168)
From 1198793(120591 + 120587) = 2 minus 119879
3(120591) it immediately follows that for
1198861ge 0 and minus18 le 119886
3le 1 119879
3(120591) can be expressed as
1198793(120591) = (1 + cos 120591) [1 + 2119886
3(1 minus 2 cos 120591 + cos 2120591)] (169)
For 1198861le 0 and 18 le 119886
3le 1 from (158) it follows that 119886
3=
[8cos31205910]minus1 This relation along with (160) and (157) further
implies that 1198793(120591) can be expressed as
1198793(120591) =
[cos 1205910minus cos 120591]2 [2 cos 120591
0+ cos 120591]
2cos31205910
(170)
providing that |1205910| le 1205873 From 119879
3(120591 + 120587) = 2 minus 119879
3(120591) it
follows that (170) also holds for 1198861ge 0 and minus1 le 119886
3le minus18
providing that 1205910isin [21205873 41205873]
Maximally flat nonnegative cosinewaveformof type (167)with 119886
1lt 0 (minimum at 120591
0= 0) reads 119879
3(120591) = [1 minus
cos 120591]2[1 + (12) cos 120591] Dually maximally flat nonnegativecosine waveform with 119886
1gt 0 (minimum at 120591
0= 120587) reads
1198793(120591) = [1 + cos 120591]2[1 minus (12) cos 120591]In what follows we provide relations between coefficients
1198861and 1198863of nonnegative cosine waveforms of type (167) with
at least one zeroFor 1198861le 0 conversion of (168) into an additive form
immediately leads to the following relation
1198861= minus1 minus 119886
3for minus 1 le 119886
3le1
8 (171)
Conversion of (170) into an additive form leads to 1198861
=
minus31198863(2 cos 2120591
0+ 1) which can be also expressed as 119886
1=
minus31198863(4cos2120591
0minus 1) For 119886
1le 0 relations |120591
0| le 1205873 119886
1=
minus31198863(4cos2120591
0minus 1) and 119886
3= [8cos3120591
0]minus1 lead to
1198861= minus3 [ 3radic119886
3minus 1198863] for 1
8le 1198863le 1 (172)
Similarly for 1198861ge 0 conversion of (169) into an additive form
leads to the following relation
1198861= 1 minus 119886
3for minus
1
8le 1198863le 1 (173)
Mathematical Problems in Engineering 21
1
15
15
05
0
minus05
minus15minus15
minus1
minus1 minus05 0 05 1
Coefficient a1
Coe
ffici
enta
3 a1= minus
3(3radica3minus a3)
a1 =
minus1 minus a3
a1 =
1 minusa3
a1 =3(3radic|a3
| +a3)
Figure 14 Parameter space of cosine waveforms for 119896 = 3
For waveform of type (170) with 1198861ge 0 relations 120591
0isin [21205873
41205873] 1198861= minus3119886
3(4cos2120591
0minus 1) and 119886
3= [8cos3120591
0]minus1 lead to
1198861= 3 [
3radic10038161003816100381610038161198863
1003816100381610038161003816 + 1198863] for minus 1 le 1198863le minus
1
8 (174)
Every cosine waveform of type (167) corresponds to apair of real numbers (119886
1 1198863) and vice versa Points (119886
1 1198863)
in grey area in Figure 14 correspond to nonnegative cosinewaveforms for 119896 = 3 The points at the boundary of grey areacorrespond to nonnegative cosinewaveformswith at least onezero A number of shapes of nonnegative cosine waveformswith 119896 = 3 and at least one zero plotted on interval [minus120587 120587]are also presented in Figure 14 The boundary of grey area inFigure 14 consists of four line segments described by relations(171)ndash(174) The common point of line segments (172) and(173) is cusp point with coordinates 119886
1= 0 and 119886
3= 1
Another cusp point with coordinates 1198861= 0 and 119886
3= minus1
is the common point of line segments (171) and (174) Thecommon point of line segments (171)-(172) has coordinates(minus98 18) and common point of line segments (173)-(174)has coordinates (98 minus18) These points are representedby white circle dots and they correspond to maximallyflat cosine waveforms (eg see [21]) White triangle dotswith coordinates (2radic3 minusradic39) and (minus2radic3radic39) refer tothe nonnegative cosine waveforms with maximum value ofamplitude of fundamental harmonic
7 Four Case Studies of Usage of NonnegativeWaveforms in PA Efficiency Analysis
In this section we provide four case studies of usage ofdescription of nonnegative waveforms with fundamental and119896th harmonic in PA efficiency analysis In first two casestudies to be presented in Section 71 voltage is nonnegativewaveform with fundamental and second harmonic with atleast one zero In remaining two case studies to be consideredin Section 72 voltage waveform contains fundamental andthird harmonic
i
v Load
Vdc
Idc
Lch
vL
iL
Cb
+ +in
Figure 15 Generic PA circuit diagram
Let us consider generic PA circuit diagram as shown inFigure 15 We assume here that voltage and current wave-forms at the transistor output are
V (120579) = 1 + 1198861V cos 120579 + 1198871V sin 120579 + 119886119896V cos 119896120579 + 119887119896V sin 119896120579
119894 (120579) = 1 + 1198861119894cos 120579 +
infin
sum
119899=2
119886119899119894cos 119899120579
(175)
where 120579 stands for 120596119905 Both waveforms are normalized in thesense that dc components of voltage and current are 119881dc =
1 and 119868dc = 1 respectively Under assumption that blockingcapacitor 119862
119887behaves as short-circuit at the fundamental and
higher harmonics current and voltage waveforms at the loadare
V119871(120579) = 119886
1V cos 120579 + 1198871V sin 120579 + 119886119896V cos 119896120579 + 119887119896V sin 119896120579
119894119871(120579) = minus119886
1119894cos 120579 minus
infin
sum
119899=2
119886119899119894cos 119899120579
(176)
In terms of coefficients of voltage and current waveforms theload impedance at fundamental harmonic is 119911
1= minus(119886
1V minus
1198951198871V)1198861119894 whereas load impedance at 119896th harmonic is 119911
119896=
minus(119886119896Vminus119895119887119896V)119886119896119894 All other harmonics are short-circuited (119911
119899=
0 for 119899 = 1 and 119899 = 119896) Time average output power of PA (egsee [10]) with waveform pair (175) at fundamental frequencycan be expressed as
1198751= minus
11988611198941198861V
2 (177)
For normalized waveforms (175) with 119881dc = 1 and 119868dc = 1dc power is 119875dc = 1 Consequently PA efficiency 120578 = 119875
1119875dc
(eg see [10 26]) is equal to
120578 = minus11988611198941198861V
2 (178)
Thus time average output power 1198751of PA with pair of nor-
malized waveform (175) is equal to efficiency (178)Power utilization factor (PUF) is defined [26] as ldquothe
ratio of power delivered in a given situation to the power
22 Mathematical Problems in Engineering
delivered by the same device with the same supply voltagein Class A moderdquo Since the output power in class-A modeis 1198751class-A = max[V(120579)] sdotmax[119894(120579)]8 (eg see [9]) it follows
that power utilization factor PUF = 11987511198751class-A for PA with
pair of normalized waveforms (175) can be expressed as
PUF =8120578
max [V (120579)] sdotmax [119894 (120579)] (179)
71 NonnegativeWaveforms for 119896 = 2 in PAEfficiencyAnalysisIn this subsection we provide two case studies of usage ofdescription of nonnegative waveforms with fundamental andsecond harmonic (119896 = 2) in PA efficiency analysis For moreexamples of usage of descriptions of nonnegative waveformswith fundamental and second harmonic in PA efficiencyanalysis see [12]
Case Study 71 In this case study we consider efficiency of PAfor given second harmonic impedance providing that voltageis nonnegative waveform with fundamental and second har-monic and current is ldquohalf-sinerdquo waveform frequently used inefficiency analysis of classical PA operation (eg see [10])
Standard model of current waveform for classical PAoperation has the form (eg see [10 26])
119894119863(120579) =
119868119863[cos 120579 minus cos(120572
2)] |120579| le
120572
2
0120572
2le |120579| le 120587
(180)
where 120572 is conduction angle and 119868119863
gt 0 Since 119894119863(120579) is
even function it immediately follows that its Fourier seriescontains only dc component and cosine terms
119894119863(120579) = 119868dc +
infin
sum
119899=1
119868119899cos 119899120579 (181)
The dc component of the waveform (180) is
119868dc =119868119863120572
2120587[sinc(120572
2) minus cos(120572
2)] (182)
where sinc119909 = (sin119909)119909 The coefficient of the fundamentalharmonic component reads
1198681=119868119863120572
2120587(1 minus sinc120572) (183)
and the coefficient of 119899th harmonic component can bewrittenin the form
119868119899=119868119863
119899120587[sin ((119899 minus 1) 1205722)
(119899 minus 1)minussin ((119899 + 1) 1205722)
(119899 + 1)] 119899 ge 2
(184)
For ldquohalf-sinerdquo current waveform conduction angle is equalto 120587 (class-B conduction angle) According to (182) thisfurther implies that 119868dc = 119868
119863120587 To obtain normalized form of
waveform (180) we set 119868dc = 1which implies that 119868119863= 120587 Fur-
thermore substitution of 120572 = 120587 and 119868119863= 120587 in (180) leads to
119894 (120579) =
120587 cos 120579 |120579| lt120587
2
0120587
2lt |120579| le 120587
(185)
Similarly substitution of 119868119863= 120587 and 120572 = 120587 into (183) and
(184) leads to the coefficients of waveform (185) Coefficientsof fundamental and second harmonic respectively are
1198861119894=120587
2 119886
2119894=2
3 (186)
On the other hand voltage waveform of type (35) for 119896 =2 reads
V (120579) = 1 + 1198861V cos 120579 + 1198871V sin 120579 + 1198862V cos 2120579 + 1198872V sin 2120579
(187)
This waveform contains only fundamental and second har-monic and therefore all harmonics of order higher thantwo are short-circuited (119911
119899= 0 for 119899 gt 2) For current
voltage pair (185) and (187) load impedance at fundamentalharmonic is 119911
1= minus(119886
1V minus 1198951198871V)1198861119894 whereas load impedance
at second harmonic is 1199112= minus(119886
2Vminus1198951198872V)1198862119894 According to ourassumption the load is passive and therefore Re119911
1 gt 0 and
Re1199112 ge 0 which further imply 119886
11198941198861V lt 0 and 119886
21198941198862V le 0
respectivelyIt is easy to see that problem of findingmaximal efficiency
of PA with current-voltage pair (185) and (187) for prescribedsecond harmonic impedance can be reduced to the problemof finding voltage waveform of type (187) with maximal coef-ficient |119886
1V| for prescribed coefficients of second harmonic(see Section 5)
The following algorithm (analogous to Algorithm 22presented in [12]) provides the procedure for calculation ofmaximal efficiency with current-voltage pair (185) and (187)for prescribed second harmonic impedance The definitionof function atan 2(119910 119909) which appears in the step (iii) of thefollowing algorithm is given by (105)
Algorithm 32 (i) Choose 1199112= 1199032+1198951199092such that |119911
2| le 1|119886
2119894|
(ii) calculate 1198862V minus 1198951198872V = minus119911
21198862119894and 120582
2V = radic11988622V + 1198872
2V(iii) if 2120582
2V le 1 minus 1198862V then calculate 119886
1V = minus1 minus 1198862V and
1198871V = minus2119887
2V else calculate 1205821V = radic81205822V(1 minus 1205822V) 1205790V minus 120585V2 =
(12)atan2(1198872V 1198862V) 1198861V = minus120582
1V cos(1205790V minus 120585V2) and 1198871V =
minus1205821V sin(1205790V minus 120585V2)(iv) calculate efficiency 120578 = minus119886
11198941198861V2
(v) calculate 1199111= minus(119886
1V minus 1198951198871V)1198861119894 and 1199112119899 = 1199112Re119911
1
In this case study coefficients of fundamental and secondharmonic of current waveform are given by (186) Maximalefficiency of PA associated with the waveform pair (185)and (187) as a function of normalized second harmonicimpedance 119911
2119899= 1199112Re119911
1 is presented in Figure 16(a)
As can be seen from Figure 16(a) efficiency of 078 isachieved at the edge of Smith chart where second harmonicimpedance has small resistive part Corresponding PUFcalculated according to (179) is presented in Figure 16(b)Peak efficiency 120578 = 1205874 = 07854 and peak value of PUF = 1
are attained when second harmonic is short-circuited (whichcorresponds to ideal class-B operation [10 26])
For example for second harmonic impedance 1199112= 01 minus
11989505 and currentwaveform (185) fromAlgorithm32 it followsthat 2120582
2V le 1 minus 1198862V Furthermore according to step (iii)
of above algorithm maximal efficiency of PA is attained
Mathematical Problems in Engineering 23
+j2
+j5
+j1
+j2
+j5
minusj2
minusj5
minusj1
minusj2
minusj5
infin0
078
02
05
1 2 5
075 07 065
06 05
075
07
065
06
05
120578 lt 05
(a)
099
095
091
083 075
067
0 59 051
051
099
095
091
083
075 0
67 059
+j5
+j1
+j2
+j5
minusj5
minusj1
minusj2
minusj5
infin
+j2
minusj2
0 02
05
1 2 5
120578 lt 05
(b)
Figure 16 (a) Contours of maximal efficiency of PA and (b) contours of corresponding PUF associated with the waveform pair (185) and(187) as functions of normalized second harmonic impedance 119911
2119899= 1199112Re119911
1
2
3
4
1
0
2 3 410
Wav
efor
ms
Angle 120579120587
i(120579)
(120579)
Figure 17 Waveform pair (185) and (187) that provides maximalefficiency for 119911
2= 01 minus 11989505
with voltage waveform of type (187) with coefficients 1198862V =
minus00667 1198872V = minus03333 119886
1V = minus09333 and 1198871V = 06667 (see
Figure 17) Corresponding efficiency PUF and normalizedsecond harmonic impedance are 120578 = 07330 PUF = 07572and 1199112119899= 01683 minus 11989508415 respectively
On the other hand for second harmonic impedance 1199112=
01 minus 11989508 and current waveform (185) from Algorithm 32 itfollows that 2120582
2V gt 1 minus 1198862V Then according to step (iii) of
above algorithm maximal efficiency is attained with voltagewaveform of type (187) with coefficients 119886
2V = minus00667 1198872V =
minus05333 1198861V = minus09333 and 119887
1V = 10572 (see Figure 18)Efficiency PUF andnormalized secondharmonic impedanceare 120578 = 07330 PUF = 06332 and 119911
2119899= 01683 minus 11989513465
respectively
2
3
4
1
0
2 3 410
Wav
efor
ms
Angle 120579120587
i(120579)
(120579)
Figure 18 Waveform pair (185) and (187) that provides maximalefficiency for 119911
2= 01 minus 11989508
Case Study 72 As another case study let us consider the effi-ciency of PA providing that current waveform is nonnegativecosine waveform up to third harmonic with maximum valueof amplitude of fundamental harmonic [22] (see also [8])
119894 (120579) = 1 +1 + radic5
2cos 120579 + 2radic5
5cos 2120579 + 5 minus radic5
10cos 3120579
(188)
and voltage waveform is nonnegative waveform of type(187) Load impedances at fundamental second and thirdharmonic are 119911
1= minus(119886
1Vminus1198951198871V)1198861119894 1199112 = minus(1198862Vminus1198951198872V)1198862119894 and
1199113= 0 respectively According to our assumption the load
is passive and therefore Re1199111 gt 0 and Re119911
2 ge 0 which
further imply 11988611198941198861V lt 0 and 119886
21198941198862V le 0 respectively
Because current waveform (188) contains only cosineterms and voltage waveform is the same as in previous case
24 Mathematical Problems in Engineering
08
075
07
06
06
065
065
05
05
075 0
7
120578 lt 05
+j2
+j5
+j1
+j2
+j5
minusj2
minusj5
minusj1
minusj2
minusj5
infin0
02
05
1 2 5
(a)
085
08 0
75 07 0
6
05 04
08
075
0405
060
7
120578 lt 05
+j2
+j5
+j1
+j2
+j5
minusj2
minusj5
minusj1
minusj2
minusj5
infin0 02
05
1 2 5
(b)
Figure 19 (a) Contours of maximal efficiency of PA and (b) contours of corresponding PUF associated with the waveform pair (187)-(188)as functions of normalized second harmonic impedance 119911
2119899= 1199112Re119911
1
study the procedure for calculation of maximal efficiency ofPA with waveform pair (187)-(188) is the same as presentedin Algorithm 32 In this case study the coefficients of funda-mental and second harmonic of current waveform are 119886
1119894=
(1 + radic5)2 and 1198862119894= 2radic55 respectively
Maximal efficiency of PA associated with the waveformpair (187)-(188) as a function of normalized secondharmonicimpedance 119911
2119899= 1199112Re119911
1 is presented in Figure 19(a)
Efficiency of 08 is achieved at the edge of Smith chart wheresecond harmonic impedance has small resistive part Thetheoretical upper bound 120578 = (1 + radic5)4 asymp 08090 isattained when second harmonic is short-circuitedWhen thisupper bound is reached both second and third harmonic areshort-circuited which implies that we are dealing with finiteharmonic class-C [6 8] or dually when current and voltageinterchange their roles with finite harmonic inverse class-C[6 9] Corresponding PUF calculated according to (179) ispresented in Figure 19(b) Peak value of PUF asymp 08541 isattained when second harmonic is short-circuited
For example for second harmonic impedance 1199112
=
007 minus 11989504 and current waveform (188) from Algorithm 32 itfollows that 2120582
2V le 1 minus 1198862V Furthermore according to step
(iii) of Algorithm 32 maximal efficiency of PA is attainedwith voltage waveform of type (187) with coefficients 119886
2V =
minus00626 1198872V = minus03578 119886
1V = minus09374 and 1198871V = 07155 (see
Figure 20) Corresponding efficiency PUF and normalizedsecond harmonic impedance are 120578 = 07584 PUF = 06337and 1199112119899= 01208 minus 11989506904 respectively
On the other hand for 1199112= 005 minus 11989507 and current
waveform (187) it follows that 21205822V gt 1minus119886
2VThen accordingto step (iii) of Algorithm 32 themaximal efficiency is attainedwith voltage waveform of type (187) with coefficients 119886
2V =
minus00447 1198872V = minus06261 119886
1V = minus09318 and 1198871V = 10007 (see
Figure 21) Efficiency PUF and normalized second harmonic
2
3
4
1
0
2 3 410
Wav
efor
ms
Angle 120579120587
i(120579)
(120579)
Figure 20 Waveform pair (187)-(188) that provides maximal effi-ciency for 119911
2= 007 minus 11989504
impedance are 120578 = 07538 PUF = 05314 and 1199112119899= 00868minus
11989512156 respectively
72 Nonnegative Waveforms for 119896 = 3 in PA EfficiencyAnalysis In this subsection we provide another two casestudies of usage of description of nonnegative waveforms inPA efficiency analysis this time with fundamental and thirdharmonic (119896 = 3)
Case Study 73 Let us consider current-voltage pair such thatvoltage is nonnegative waveform with fundamental and thirdharmonic
V (120579) = 1 + 1198861V cos 120579 + 1198871V sin 120579 + 1198863V cos 3120579 + 1198873V sin 3120579
(189)
Mathematical Problems in Engineering 25
2
3
4
1
0
2 3 410
Wav
efor
ms
Angle 120579120587
i(120579)
(120579)
Figure 21 Waveform pair (187)-(188) that provides maximal effi-ciency for 119911
2= 005 minus 11989507
and current is nonnegative cosine waveform given by (188)Load impedances at fundamental second and third har-monic are 119911
1= minus(119886
1V minus 1198951198871V)1198861119894 1199112 = 0 and 119911
3= minus(119886
3V minus
1198951198873V)1198863119894 respectively According to our assumption the load
is passive and therefore Re1199111 gt 0 and Re119911
3 ge 0 which
further imply 11988611198941198861V lt 0 and 119886
31198941198863V le 0
In this subsection we consider the problem of findingmaximal efficiency of PA with waveform pair (188)-(189)for given third harmonic impedance As we mentionedearlier problem of finding maximal efficiency of PA withcurrent-voltage pair (188)-(189) for prescribed third har-monic impedance can be reduced to the problem of findingvoltage waveform of type (189) withmaximal coefficient |119886
1V|for prescribed coefficients of third harmonic (see Section 52)
The following algorithm provides the procedure forcalculation of maximal efficiency with current-voltage pair(188)-(189) The definition of function atan 2(119910 119909) whichappears in step (iii) of the following algorithm is given by(105)
Algorithm 33 (i) Choose 1199113= 1199033+1198951199093such that |119911
3| le 1|119886
3119894|
(ii) calculate 1198863V minus 1198951198873V = minus119911
31198863119894and 120582
3V = radic11988623V + 1198872
3V(iii) if 271205822
3V le (1 minus 21198863V)3 then calculate 119886
1V = minus1 minus 1198863V
and 1198871V = minus3119887
3V else calculate 1205821V = 3(3radic1205823V minus 120582
3V) 1205790V minus120585V3 = (13)atan 2(119887
3V 1198863V) 1198861V = minus1205821V cos(1205790V minus 120585V3) and
1198871V = minus120582
1V sin(1205790V minus 120585V3)(iv) calculate efficiency 120578 = minus119886
11198941198861V2
(v) calculate 1199111= minus(119886
1V minus 1198951198871V)1198861119894 and 1199113119899 = 1199113Re119911
1
In this case study coefficients of fundamental and thirdharmonic of current waveform are 119886
1119894= (1 + radic5)2 and
1198863119894= (5 minus radic5)10 respectively For the waveform pair (188)-
(189) maximal efficiency of PA as a function of normalizedthird harmonic impedance 119911
3119899= 1199113Re119911
1 is presented in
Figure 22 Efficiency of 08 is reached when third harmonicimpedance has small resistive part Peak efficiency 120578 = (1 +
radic5)4 asymp 08090 is achieved when third harmonic is short-circuited
+j2
+j5
+j1
+j2
+j5
minusj2
minusj5
minusj1
minusj2
minusj5
infin0
08
08
075
075
07
07
06
05
02
05
1 2
Figure 22 Contours of maximal efficiency of PA associated withthe waveform pair (188)-(189) as a function of normalized thirdharmonic impedance 119911
3119899= 1199113Re119911
1
For the present case study in what follows we show thatpower utilization factor is proportional to efficiency For volt-age waveform of type (189) it is easy to see that V(120579 + 120587) =
2minusV(120579) holdsThis relation along with the fact that waveformV(120579) that provides maximal efficiency has at least one zeroimplies that max[V(120579)] = 2 On the other hand currentwaveform (188) is cosine waveform with positive coefficientsand therefore max[119894(120579)] = 119894(0) = 2 + 4radic5 Consequentlyaccording to (179) the following relation holds
PUFCase study 73 = 2 (5 minus 2radic5) 120578 = 10557120578 (190)
Clearly the ratio PUF120578 is constant and therefore in this casestudy PUF can be easily calculated from the correspondingefficiency Accordingly peak efficiency and peak value ofPUFCase study 73 = 3radic52 minus 52 = 08541 are attained forthe same voltage waveform (when third harmonic is short-circuited)
In the first example current waveform (188) and 1199113=
02 minus 11989505 imply that 2712058223V le (1 minus 2119886
3V)3 Then according
to Algorithm 33 the voltage waveform of type (189) thatprovides maximal efficiency has the following coefficients1198863V = minus00553 119887
3V = minus01382 1198861V = minus09447 and 119887
1V =
04146 (see Figure 23) Efficiency PUF and normalized thirdharmonic impedance are 120578 = 07643 PUF = 08069 and1199113119899= 03425 minus 11989508564 respectivelyIn the second example current waveform (188) and 119911
3=
01 minus 11989511 imply that 2712058223V gt (1 minus 2119886
3V)3 Then according
to Algorithm 33 the voltage waveform of type (189) thatprovides maximal efficiency has the following coefficients1198863V = minus00276 119887
3V = minus03040 1198861V = minus09391 and 119887
1V =
05807 (see Figure 24) Efficiency PUF and normalized thirdharmonic impedance are 120578 = 07598 PUF = 08021 and1199113119899= 01723 minus 11989518952 respectively
26 Mathematical Problems in Engineering
2
3
4
1
0
2 3 410
Wav
efor
ms
Angle 120579120587
i(120579)
(120579)
Figure 23 Waveform pair (188)-(189) that provides maximal effi-ciency for 119911
3= 02 minus 11989505
2
3
4
1
0
2 3 410
Wav
efor
ms
Angle 120579120587
i(120579)
(120579)
Figure 24 Waveform pair (188)-(189) that provides maximal effi-ciency for 119911
3= 01 minus 11989511
Case Study 74 In this case study let us consider current-voltage pair where current is normalized waveform of type(180) with conduction angle 120572 = 115120587 (207∘) and voltageis nonnegative waveform of type (189) Substitution of 120572 =
115120587 and 119868dc = 1 into (182) leads to 119868119863
= 22535Furthermore substitution of 120572 = 115120587 and 119868
119863= 22535 into
(180) leads to
119894 (120579)
=
22535 [cos 120579 minus cos(1151205872
)] |120579| le115120587
2
0115120587
2le |120579| le 120587
(191)
Similarly substitution of 120572 = 115120587 and 119868119863
= 22535 into(183) and (184) for 119899 = 3 yields coefficients of fundamentaland third harmonic of waveform (191)
1198861119894= 14586 119886
3119894= minus01026 (192)
Because current waveform (191) contains only cosineterms and voltage waveform is the same as in previous case
+j2
+j5
+j1
+j2
+j5
minusj2
minusj5
minusj1
minusj2
minusj5
infin0 02
05
1 2
074
076
08
084
082
078
076
074
Figure 25 Contours of maximal efficiency of PA associated withthe waveform pair (189) and (191) as a function of normalized thirdharmonic impedance 119911
3119899= 1199113Re119911
1
study the procedure for calculation of maximal efficiencyof PA with waveform pair (189)ndash(191) is the same as thatpresented in Algorithm 33 In this case study the coefficientsof fundamental and third harmonic of current waveform aregiven by (192)
For the waveform pair (189) and (191) maximal efficiencyof PA as a function of normalized third harmonic impedance1199113119899= 1199113Re119911
1 is presented in Figure 25 Efficiency of 084
is obtained in vicinity of 1199113119899= 23685 (corresponding to 119911
3=
18750) Peak efficiency 120578 asymp 08421 is achieved for voltagewaveform of type (189) with coefficients 119886
1V = minus2radic3 1198863V =
radic39 and 1198871V = 1198873V = 0
In the course of finding power utilization factor noticethat currentwaveformof type (191) attains itsmaximumvaluefor 120579 = 0 Insertion of max[119894(120579)] = 119894(0) = 278 andmax[V(120579)] = 2 for voltage waveform of type (189) into (179)leads to
PUFCase study 74 = 1439120578 (193)
Again the ratio PUF120578 is constant and PUF can be easilycalculated from the corresponding efficiency Accordinglypeak value of PUFCase study 74 asymp 12118 and peak efficiencyare attained for the same voltage waveform
In the first example current waveform (191) and 1199113=
1 minus 11989502 imply that 2712058223V le (1 minus 2119886
3V)3 Then according to
Algorithm 33 voltage waveform of type (189) which providesmaximal efficiency has coefficients 119886
3V = 01026 1198873V =
00205 1198861V = minus11026 and 119887
1V = minus00616 (see Figure 26)Efficiency PUF and normalized third harmonic impedanceare 120578 = 08042 PUF = 11572 and 119911
3119899= 13228 minus 11989502646
respectivelyIn second example current waveform (191) and 119911
3=
15 minus 11989512 imply that 2712058223V gt (1 minus 2119886
3V)3 Then according to
Algorithm 33 voltage waveform of type (189) which providesmaximal efficiency has coefficients 119886
3V = 01540 1198873V =
Mathematical Problems in Engineering 27
2
3
4
1
0
2 3 410
Wav
efor
ms
Angle 120579120587
i(120579)
(120579)
Figure 26 Waveform pair (189) and (191) that provides maximalefficiency for 119911
3= 1 minus 11989502
2
3
4
1
0
2 3 410
Wav
efor
ms
Angle 120579120587
i(120579)
(120579)
Figure 27 Waveform pair (189) and (191) that provides maximalefficiency for 119911
3= 15 minus 11989512
01232 1198861V = minus11255 and 119887
1V = minus02575 (see Figure 27)Efficiency PUF and normalized third harmonic impedanceare 120578 = 08208 PUF = 11812 and 119911
3119899= 19439 minus 11989515552
respectively
8 Conclusion
In this paper we consider a problem of finding general de-scriptions of various classes of nonnegative waveforms withfundamental and 119896th harmonic These classes include non-negative waveforms with at least one zero nonnegative wave-forms with maximal amplitude of fundamental harmonic forprescribed amplitude of 119896th harmonic nonnegative wave-forms withmaximal coefficient of cosine part of fundamentalharmonic for prescribed coefficients of 119896th harmonic andnonnegative cosine waveforms with at least one zero Mainresults are stated in six propositions (Propositions 1 6 918 22 and 26) four corollaries (Corollaries 2ndash5) twentyremarks and three algorithms Four case studies of usageof closed form descriptions of nonnegative waveforms in PAefficiency analysis are considered in detail in Section 7
Appendices
Here we provide a list of finite sums of trigonometric func-tions used in this paper (Appendix A) and brief account ofthe Chebyshev polynomials (Appendix B)
A List of Some Finite Sums ofTrigonometric Functions
Dirichlet kernel (eg see [27]) is as follows
119863119896minus1
(120591) = 1 + 2
119896minus1
sum
119899=1
cos 119899120591 = sin ((2119896 minus 1) 1205912)sin (1205912)
(A1)
Fejer kernel (eg see [27]) can be expressed in the fol-lowing equivalent forms
119865119896minus1
(120591) =1
119896
119896minus1
sum
119899=0
119863119899(120591) = 1 +
2
119896
119896minus1
sum
119899=1
(119896 minus 119899) cos 119899120591
=(1 minus cos 119896120591)119896 (1 minus cos 120591)
(A2)
Lagrangersquos trigonometric identity (eg see [28]) is as fol-lows
1198781(120591) =
119896minus1
sum
119899=1
sin 119899120591 = sin (1198961205912) sin ((119896 minus 1) 1205912)sin (1205912)
(A3)
In what follows we show that the following three trigono-metric identities also hold
2
119896minus1
sum
119899=1
(119896 minus 119899) sin 119899120591 = 119896 sin 120591 minus sin 1198961205911 minus cos 120591
(A4)
119896minus1
sum
119899=1
cos (119896 minus 2119899) 120591 = sin (119896 minus 1) 120591sin 120591
(A5)
119896minus1
sum
119899=1
119899 (119896 minus 119899) cos (119896 minus 2119899) 120591
=sin (119896120591) cos 120591 minus 119896 cos (119896120591) sin 120591
2sin3120591
(A6)
Denote 1198782(120591) = 2sum
119896minus1
119899=1(119896 minus 119899) sin 119899120591 119878
3(120591) = sum
119896minus1
119899=1cos(119896 minus
2119899)120591 and 1198784(120591) = sum
119896minus1
119899=1119899(119896 minus 119899) cos(119896 minus 2119899)120591
Notice that 1198782(120591) = 2119896119878
1(120591) + 119889119863
119896minus1(120591)119889120591 which
immediately leads to (A4)
28 Mathematical Problems in Engineering
Identity (A5) can be obtained as follows
sin (119896 minus 1) 120591sin 120591
=119890119895(119896minus1)120591
minus 119890minus119895(119896minus1)120591
119890119895120591 minus 119890minus119895120591
= 119890119895119896120591
119890minus2119895120591
minus 119890minus2119895(119896minus1)120591
1 minus 119890minus2119895120591
= 119890119895119896120591
119896minus1
sum
119899=1
119890minus2119895119899120591
=
119896minus1
sum
119899=1
119890119895(119896minus2119899)120591
=
119896minus1
sum
119899=1
cos (119896 minus 2119899) 120591
(A7)
From 4119899(119896 minus 119899) = 1198962
minus (119896 minus 2119899)2 it follows that 4119878
4(120591) =
1198962
1198783(120591) + 119889
2
1198783(120591)119889120591
2 which leads to (A6)
B The Chebyshev Polynomials
The Chebyshev polynomials of the first kind 119881119899(119909) can be
defined by the following relation (eg see [29])
119881119899(119909) = cos 119899120591 when 119909 = cos 120591 (B1)
The Chebyshev polynomials of the second kind 119880119899(119909) can be
defined by the following relation (eg see [29])
119880119899(119909) =
sin (119899 + 1) 120591sin 120591
when 119909 = cos 120591 (B2)
The Chebyshev polynomials satisfy the following recur-rence relations (eg see [29])
1198810(119909) = 1 119881
1(119909) = 119909
119881119899+1
(119909) = 2119909119881119899(119909) minus 119881
119899minus1(119909)
1198800(119909) = 1 119880
1(119909) = 2119909
119880119899+1
(119909) = 2119909119880119899(119909) minus 119880
119899minus1(119909)
(B3)
The first few Chebyshev polynomials of the first and secondkind are 119881
2(119909) = 2119909
2
minus 1 1198813(119909) = 4119909
3
minus 3119909 1198814(119909) = 8119909
4
minus
81199092
+ 1 1198802(119909) = 4119909
2
minus 1 1198803(119909) = 8119909
3
minus 4119909 and 1198804(119909) =
161199094
minus 121199092
+ 1
Conflict of Interests
The authors declare that there is no conflict of interestsregarding the publication of this paper
Acknowledgment
This work is supported by the SerbianMinistry of EducationScience and Technology Development as a part of ProjectTP32016
References
[1] V I Arnolrsquod V S Afrajmovich Y S Ilrsquoyashenko and L PShilrsquonikov Dynamical Systems V Bifurcation Theory and Catas-trophe Theory Springer Berlin Germany 1994
[2] E Polak ldquoOn the mathematical foundations of nondifferen-tiable optimization in engineering designrdquo SIAM Review vol29 no 1 pp 21ndash89 1987
[3] N S Fuzik ldquoBiharmonic modes of a tuned RF power amplifierrdquoRadiotehnika vol 25 no 7 pp 62ndash71 1970 (Russian)
[4] P Colantonio F Giannini G Leuzzi and E Limiti ldquoClass Gapproach for low-voltage high-efficiency PA designrdquo Interna-tional Journal of RF and Microwave Computer-Aided Engineer-ing vol 10 no 6 pp 366ndash378 2000
[5] F H Raab ldquoMaximum efficiency and output of class-F poweramplifiersrdquo IEEE Transactions on Microwave Theory and Tech-niques vol 49 no 6 pp 1162ndash1166 2001
[6] FH Raab ldquoClass-E class-C and class-F power amplifiers basedupon a finite number of harmonicsrdquo IEEE Transactions onMicrowaveTheory and Techniques vol 49 no 8 pp 1462ndash14682001
[7] J D Rhodes ldquoOutput universality inmaximum efficiency linearpower amplifiersrdquo International Journal of Circuit Theory andApplications vol 31 no 4 pp 385ndash405 2003
[8] A Juhas and L A Novak ldquoComments on lsquoClass-E class-Cand classF power amplifier based upon a finite number ofharmonicsrsquordquo IEEE Transactions on Microwave Theory and Tech-niques vol 57 no 6 pp 1623ndash1625 2009
[9] M Roberg and Z Popovic ldquoAnalysis of high-efficiency poweramplifiers with arbitrary output harmonic terminationsrdquo IEEETransactions on Microwave Theory and Techniques vol 59 no8 pp 2037ndash2048 2011
[10] A Grebennikov N O Sokal and M J Franco Switchmode RFPower Amplifiers ElsevierAcademic Press San Diego CalifUSA 2nd edition 2012
[11] T Canning P J Tasker and S C Cripps ldquoContinuous modepower amplifier design using harmonic clipping contourstheory and practicerdquo IEEE Transactions on Microwave Theoryand Techniques vol 62 no 1 pp 100ndash110 2014
[12] A Juhas and L A Novak ldquoGeneral description of nonnegativewaveforms up to second harmonic for power amplifier mod-ellingrdquoMathematical Problems in Engineering vol 2014 ArticleID 709762 18 pages 2014
[13] V I Arnolrsquod V V Goryunov O V Lyashko and V A VasilrsquoevDynamical Systems VIIImdashSingularity Theory II ApplicationsSpringer Berlin Germany 1993
[14] D Siersma ldquoProperties of conflict sets in the planerdquo BanachCenter Publications Polish Academy of Sciences vol 50 no 1 pp267ndash276 1999 Proceedings of the Banach Center Symposiumon Geometry and Topology of Caustics (Caustics rsquo98) WarsawPoland
[15] M van Manen The geometry of conflict sets [Dissertation]Universiteit Utrecht UtrechtThe Netherlands 2003 httpigi-tur-archivelibraryuunldissertations2003-0912-123058c4pdf
[16] Y L Sachkov ldquoMaxwell strata and symmetries in the problemofoptimal rolling of a sphere over a planerdquo Sbornik Mathematicsvol 201 no 7-8 pp 1029ndash1051 2010
[17] I A Bogaevsky ldquoPerestroikas of shocks and singularities ofminimum functionsrdquoPhysicaDNonlinear Phenomena vol 173no 1-2 pp 1ndash28 2002
[18] Y L Sachkov ldquoMaxwell strata in the Euler elastic problemrdquoJournal of Dynamical andControl Systems vol 14 no 2 pp 169ndash234 2008
[19] M Siino and T Koike ldquoTopological classification of black holesgeneric Maxwell set and crease set of a horizonrdquo InternationalJournal ofModern Physics D Gravitation Astrophysics Cosmol-ogy vol 20 no 6 pp 1095ndash1122 2011
Mathematical Problems in Engineering 29
[20] F H Raab ldquoClass-F power amplifiers with maximally flat wave-formsrdquo IEEETransactions onMicrowaveTheory and Techniquesvol 45 no 11 pp 2007ndash2012 1997
[21] A Juhas and L A Novak ldquoMaximally flat waveforms with finitenumber of harmonics in class-F power amplifiersrdquo Mathemat-ical Problems in Engineering vol 2013 Article ID 169590 9pages 2013
[22] L Fejer ldquoUber trigonometrische polynomerdquo Journal fur dieReine und Angewandte Mathematik vol 1916 no 146 pp 53ndash82 1916 (German)
[23] S C Cripps ldquoBessel Waives [microwave bytes]rdquo IEEE Micro-wave Magazine vol 10 no 7 pp 30ndash36 117 2009
[24] L N Bryzgalova ldquoSingularities of the maximum of para-metrically dependent functionrdquo Functional Analysis and ItsApplications vol 11 no 1 pp 49ndash51 1977
[25] V I Arnold A A Davydov V A Vassiliev and V MZakalyukin Mathematical Models of Catastrophes Control ofCatastrophic Process Encyclopedia of Life Support Systems(EOLSS) EOLSS Publishers Oxford UK 2006
[26] S C Cripps RF Power Amplifiers for Wireless CommunicationsArtech House Norwood Mass USA 2nd edition 2006
[27] A Zygmund Trigonometric Series vol 1 Cambridge UniversityPress Cambridge UK 2nd edition 1959
[28] A Jeffrey and H DaiHandbook of Mathematical Formulas andIntegrals ElsevierAcademic Press San Diego Calif USA 4thedition 2008
[29] J C Mason and D C Handscomb Chebyshev PolynomialsChapman amp Hall CRC Press Boca Raton Fla USA 2003
Submit your manuscripts athttpwwwhindawicom
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical Problems in Engineering
Hindawi Publishing Corporationhttpwwwhindawicom
Differential EquationsInternational Journal of
Volume 2014
Applied MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Probability and StatisticsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical PhysicsAdvances in
Complex AnalysisJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
OptimizationJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
CombinatoricsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Operations ResearchAdvances in
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Function Spaces
Abstract and Applied AnalysisHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of Mathematics and Mathematical Sciences
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Algebra
Discrete Dynamics in Nature and Society
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Decision SciencesAdvances in
Discrete MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom
Volume 2014 Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Stochastic AnalysisInternational Journal of
20 Mathematical Problems in Engineering
119879119896(120591) = 1 + cos 119896120591 Consequently (156)ndash(158) hold for 119886
119896= 1
and |1205910| = 120587119896
It is easy to see that 120582119896= |119886119896| lt 1 and 119879
119896(1205910) = 0 for some
1205910imply 120582
1= |1198861| = 0 Therefore in what follows we assume
that |119886119896| = 1 and 119886
1lt 0
Cosine waveforms are even functions of 120591 Therefore ifnonnegative cosine waveform has exactly one zero it has to beeither at 0 or at 120587 On the other hand if nonnegative cosinewaveform with 119886
1= 0 has exactly two zeros then these zeros
are placed at plusmn1205910 such that 120591
0is neither 0 nor 120587
In order to prove that (155) holds for minus1 lt 119886119896le 1(1minus119896
2
)let us start by referring to the description (38) of nonnegativewaveformswith at least one zero As wementioned earlier fornonnegative cosine waveformwith exactly one zero (denotedby 1205910) it is either 120591
0= 0 or 120591
0= 120587 Therefore in both cases
sin 1205910= 0 Substitution of sin 120591
0= 0 into (43) together with
1198861
= 0 and 120582119896= |119886119896| lt 1 leads to
1205910= 0 (164)
Clearly 1205910= 0 119887
1= 0 and 119887
119896= 0 according to (44) and (46)
imply 120582119896sin 120585 = 0 Since 120582
119896= |119886119896| it follows that |119886
119896| sin 120585 = 0
also holds which further implies 120582119896= 119886119896= 0 or sin 120585 =
0 In the case when 120582119896= 119886119896= 0 from (164) and (43) we
obtain 1198861= minus1 which further implies that 119879
119896(120591) = 1 minus cos 120591
Consequently (155) holds for 119886119896= 0 In the case when sin 120585 =
0 from (164) and (45) we obtain 119886119896= 120582119896if 120585 = 0 or 119886
119896= minus120582119896
if 120585 = 120587 Relations 119886119896= 120582119896and 120585 = 0 according to (40) imply
that 0 le 119886119896le 1(1 minus 119896
2
) Substitution of 120585 = 0 120582119896= 119886119896 and
(164) into (38) leads to (155) which proves that (155) holdsfor 0 le 119886
119896le 1(1 minus 119896
2
) On the other hand relations 119886119896=
minus120582119896and 120585 = 120587 according to (40) imply that minus1 lt 119886
119896le 0
Substitution of 120585 = 120587 120582119896= minus119886119896 and (164) into (38) also leads
to (155) which proves that (155) also holds for minus1 lt 119886119896le 0
Consequently (155) holds for minus1 lt 119886119896le 1(1 minus 119896
2
)In what follows we first prove that (156)-(157) hold for
1(1 minus 1198962
) lt 119886119896lt 1 For this purpose let us start with non-
negative waveforms with two zeros described by (66) As wementioned before nonnegative cosine waveforms with twozeros have zeros at 120591
0and minus120591
0 such that 120591
0= 0 and 120591
0= 120587
Relations 1198861
lt 0 and 1198871
= 0 according to (84) implycos(1205910minus 120585119896) = 1 and therefore
120585
119896= 1205910 (165)
From 120585119896 = 1205910and 0 lt |120585| lt 120587 it follows that 0 lt |120591
0| lt
120587119896 Insertion of 120585119896 = 1205910into (45) yields 119886
119896= 120582119896 Relations
119886119896= 120582119896and (82) imply that 1(1 minus 1198962) lt 119886
119896lt 1 Substitution
of 120582119896= 119886119896and 120585119896 = 120591
0into (66)ndash(68) leads to (156)ndash(158)
which proves that (156)ndash(158) hold for 1(1 minus 1198962
) lt 119886119896lt 1
and 0 lt |1205910| lt 120587119896
Finally substitution of 119886119896= 1(1 minus 119896
2
) and 1205910= 0 into
(161) leads to
119879119896(120591) =
[1 minus cos 120591](1 minus 1198962)
[119896 (119896 minus 1) minus 2
119896minus1
sum
119899=1
(119896 minus 119899) cos 119899120591]
(166)
Waveform (166) coincides with waveform (155) for 119886119896
=
1(1 minus 1198962
) which in turn proves that (156) holds for 119886119896=
1(1 minus 1198962
) and 1205910= 0 This completes the proof
62 Nonnegative CosineWaveforms with at Least One Zero for119896 = 3 In this subsection we consider nonnegative cosinewaveforms with at least one zero for 119896 = 3 (for case 119896 = 2
see [12])Cosine waveform with fundamental and third harmonic
reads
1198793(120591) = 1 + 119886
1cos 120591 + 119886
3cos 3120591 (167)
For 1198861le 0 and minus1 le 119886
3le 18 according to (155) non-
negative cosine waveform of type (167) with at least one zerocan be expressed as
1198793(120591) = (1 minus cos 120591) [1 minus 2119886
3(1 + 2 cos 120591 + cos 2120591)] (168)
From 1198793(120591 + 120587) = 2 minus 119879
3(120591) it immediately follows that for
1198861ge 0 and minus18 le 119886
3le 1 119879
3(120591) can be expressed as
1198793(120591) = (1 + cos 120591) [1 + 2119886
3(1 minus 2 cos 120591 + cos 2120591)] (169)
For 1198861le 0 and 18 le 119886
3le 1 from (158) it follows that 119886
3=
[8cos31205910]minus1 This relation along with (160) and (157) further
implies that 1198793(120591) can be expressed as
1198793(120591) =
[cos 1205910minus cos 120591]2 [2 cos 120591
0+ cos 120591]
2cos31205910
(170)
providing that |1205910| le 1205873 From 119879
3(120591 + 120587) = 2 minus 119879
3(120591) it
follows that (170) also holds for 1198861ge 0 and minus1 le 119886
3le minus18
providing that 1205910isin [21205873 41205873]
Maximally flat nonnegative cosinewaveformof type (167)with 119886
1lt 0 (minimum at 120591
0= 0) reads 119879
3(120591) = [1 minus
cos 120591]2[1 + (12) cos 120591] Dually maximally flat nonnegativecosine waveform with 119886
1gt 0 (minimum at 120591
0= 120587) reads
1198793(120591) = [1 + cos 120591]2[1 minus (12) cos 120591]In what follows we provide relations between coefficients
1198861and 1198863of nonnegative cosine waveforms of type (167) with
at least one zeroFor 1198861le 0 conversion of (168) into an additive form
immediately leads to the following relation
1198861= minus1 minus 119886
3for minus 1 le 119886
3le1
8 (171)
Conversion of (170) into an additive form leads to 1198861
=
minus31198863(2 cos 2120591
0+ 1) which can be also expressed as 119886
1=
minus31198863(4cos2120591
0minus 1) For 119886
1le 0 relations |120591
0| le 1205873 119886
1=
minus31198863(4cos2120591
0minus 1) and 119886
3= [8cos3120591
0]minus1 lead to
1198861= minus3 [ 3radic119886
3minus 1198863] for 1
8le 1198863le 1 (172)
Similarly for 1198861ge 0 conversion of (169) into an additive form
leads to the following relation
1198861= 1 minus 119886
3for minus
1
8le 1198863le 1 (173)
Mathematical Problems in Engineering 21
1
15
15
05
0
minus05
minus15minus15
minus1
minus1 minus05 0 05 1
Coefficient a1
Coe
ffici
enta
3 a1= minus
3(3radica3minus a3)
a1 =
minus1 minus a3
a1 =
1 minusa3
a1 =3(3radic|a3
| +a3)
Figure 14 Parameter space of cosine waveforms for 119896 = 3
For waveform of type (170) with 1198861ge 0 relations 120591
0isin [21205873
41205873] 1198861= minus3119886
3(4cos2120591
0minus 1) and 119886
3= [8cos3120591
0]minus1 lead to
1198861= 3 [
3radic10038161003816100381610038161198863
1003816100381610038161003816 + 1198863] for minus 1 le 1198863le minus
1
8 (174)
Every cosine waveform of type (167) corresponds to apair of real numbers (119886
1 1198863) and vice versa Points (119886
1 1198863)
in grey area in Figure 14 correspond to nonnegative cosinewaveforms for 119896 = 3 The points at the boundary of grey areacorrespond to nonnegative cosinewaveformswith at least onezero A number of shapes of nonnegative cosine waveformswith 119896 = 3 and at least one zero plotted on interval [minus120587 120587]are also presented in Figure 14 The boundary of grey area inFigure 14 consists of four line segments described by relations(171)ndash(174) The common point of line segments (172) and(173) is cusp point with coordinates 119886
1= 0 and 119886
3= 1
Another cusp point with coordinates 1198861= 0 and 119886
3= minus1
is the common point of line segments (171) and (174) Thecommon point of line segments (171)-(172) has coordinates(minus98 18) and common point of line segments (173)-(174)has coordinates (98 minus18) These points are representedby white circle dots and they correspond to maximallyflat cosine waveforms (eg see [21]) White triangle dotswith coordinates (2radic3 minusradic39) and (minus2radic3radic39) refer tothe nonnegative cosine waveforms with maximum value ofamplitude of fundamental harmonic
7 Four Case Studies of Usage of NonnegativeWaveforms in PA Efficiency Analysis
In this section we provide four case studies of usage ofdescription of nonnegative waveforms with fundamental and119896th harmonic in PA efficiency analysis In first two casestudies to be presented in Section 71 voltage is nonnegativewaveform with fundamental and second harmonic with atleast one zero In remaining two case studies to be consideredin Section 72 voltage waveform contains fundamental andthird harmonic
i
v Load
Vdc
Idc
Lch
vL
iL
Cb
+ +in
Figure 15 Generic PA circuit diagram
Let us consider generic PA circuit diagram as shown inFigure 15 We assume here that voltage and current wave-forms at the transistor output are
V (120579) = 1 + 1198861V cos 120579 + 1198871V sin 120579 + 119886119896V cos 119896120579 + 119887119896V sin 119896120579
119894 (120579) = 1 + 1198861119894cos 120579 +
infin
sum
119899=2
119886119899119894cos 119899120579
(175)
where 120579 stands for 120596119905 Both waveforms are normalized in thesense that dc components of voltage and current are 119881dc =
1 and 119868dc = 1 respectively Under assumption that blockingcapacitor 119862
119887behaves as short-circuit at the fundamental and
higher harmonics current and voltage waveforms at the loadare
V119871(120579) = 119886
1V cos 120579 + 1198871V sin 120579 + 119886119896V cos 119896120579 + 119887119896V sin 119896120579
119894119871(120579) = minus119886
1119894cos 120579 minus
infin
sum
119899=2
119886119899119894cos 119899120579
(176)
In terms of coefficients of voltage and current waveforms theload impedance at fundamental harmonic is 119911
1= minus(119886
1V minus
1198951198871V)1198861119894 whereas load impedance at 119896th harmonic is 119911
119896=
minus(119886119896Vminus119895119887119896V)119886119896119894 All other harmonics are short-circuited (119911
119899=
0 for 119899 = 1 and 119899 = 119896) Time average output power of PA (egsee [10]) with waveform pair (175) at fundamental frequencycan be expressed as
1198751= minus
11988611198941198861V
2 (177)
For normalized waveforms (175) with 119881dc = 1 and 119868dc = 1dc power is 119875dc = 1 Consequently PA efficiency 120578 = 119875
1119875dc
(eg see [10 26]) is equal to
120578 = minus11988611198941198861V
2 (178)
Thus time average output power 1198751of PA with pair of nor-
malized waveform (175) is equal to efficiency (178)Power utilization factor (PUF) is defined [26] as ldquothe
ratio of power delivered in a given situation to the power
22 Mathematical Problems in Engineering
delivered by the same device with the same supply voltagein Class A moderdquo Since the output power in class-A modeis 1198751class-A = max[V(120579)] sdotmax[119894(120579)]8 (eg see [9]) it follows
that power utilization factor PUF = 11987511198751class-A for PA with
pair of normalized waveforms (175) can be expressed as
PUF =8120578
max [V (120579)] sdotmax [119894 (120579)] (179)
71 NonnegativeWaveforms for 119896 = 2 in PAEfficiencyAnalysisIn this subsection we provide two case studies of usage ofdescription of nonnegative waveforms with fundamental andsecond harmonic (119896 = 2) in PA efficiency analysis For moreexamples of usage of descriptions of nonnegative waveformswith fundamental and second harmonic in PA efficiencyanalysis see [12]
Case Study 71 In this case study we consider efficiency of PAfor given second harmonic impedance providing that voltageis nonnegative waveform with fundamental and second har-monic and current is ldquohalf-sinerdquo waveform frequently used inefficiency analysis of classical PA operation (eg see [10])
Standard model of current waveform for classical PAoperation has the form (eg see [10 26])
119894119863(120579) =
119868119863[cos 120579 minus cos(120572
2)] |120579| le
120572
2
0120572
2le |120579| le 120587
(180)
where 120572 is conduction angle and 119868119863
gt 0 Since 119894119863(120579) is
even function it immediately follows that its Fourier seriescontains only dc component and cosine terms
119894119863(120579) = 119868dc +
infin
sum
119899=1
119868119899cos 119899120579 (181)
The dc component of the waveform (180) is
119868dc =119868119863120572
2120587[sinc(120572
2) minus cos(120572
2)] (182)
where sinc119909 = (sin119909)119909 The coefficient of the fundamentalharmonic component reads
1198681=119868119863120572
2120587(1 minus sinc120572) (183)
and the coefficient of 119899th harmonic component can bewrittenin the form
119868119899=119868119863
119899120587[sin ((119899 minus 1) 1205722)
(119899 minus 1)minussin ((119899 + 1) 1205722)
(119899 + 1)] 119899 ge 2
(184)
For ldquohalf-sinerdquo current waveform conduction angle is equalto 120587 (class-B conduction angle) According to (182) thisfurther implies that 119868dc = 119868
119863120587 To obtain normalized form of
waveform (180) we set 119868dc = 1which implies that 119868119863= 120587 Fur-
thermore substitution of 120572 = 120587 and 119868119863= 120587 in (180) leads to
119894 (120579) =
120587 cos 120579 |120579| lt120587
2
0120587
2lt |120579| le 120587
(185)
Similarly substitution of 119868119863= 120587 and 120572 = 120587 into (183) and
(184) leads to the coefficients of waveform (185) Coefficientsof fundamental and second harmonic respectively are
1198861119894=120587
2 119886
2119894=2
3 (186)
On the other hand voltage waveform of type (35) for 119896 =2 reads
V (120579) = 1 + 1198861V cos 120579 + 1198871V sin 120579 + 1198862V cos 2120579 + 1198872V sin 2120579
(187)
This waveform contains only fundamental and second har-monic and therefore all harmonics of order higher thantwo are short-circuited (119911
119899= 0 for 119899 gt 2) For current
voltage pair (185) and (187) load impedance at fundamentalharmonic is 119911
1= minus(119886
1V minus 1198951198871V)1198861119894 whereas load impedance
at second harmonic is 1199112= minus(119886
2Vminus1198951198872V)1198862119894 According to ourassumption the load is passive and therefore Re119911
1 gt 0 and
Re1199112 ge 0 which further imply 119886
11198941198861V lt 0 and 119886
21198941198862V le 0
respectivelyIt is easy to see that problem of findingmaximal efficiency
of PA with current-voltage pair (185) and (187) for prescribedsecond harmonic impedance can be reduced to the problemof finding voltage waveform of type (187) with maximal coef-ficient |119886
1V| for prescribed coefficients of second harmonic(see Section 5)
The following algorithm (analogous to Algorithm 22presented in [12]) provides the procedure for calculation ofmaximal efficiency with current-voltage pair (185) and (187)for prescribed second harmonic impedance The definitionof function atan 2(119910 119909) which appears in the step (iii) of thefollowing algorithm is given by (105)
Algorithm 32 (i) Choose 1199112= 1199032+1198951199092such that |119911
2| le 1|119886
2119894|
(ii) calculate 1198862V minus 1198951198872V = minus119911
21198862119894and 120582
2V = radic11988622V + 1198872
2V(iii) if 2120582
2V le 1 minus 1198862V then calculate 119886
1V = minus1 minus 1198862V and
1198871V = minus2119887
2V else calculate 1205821V = radic81205822V(1 minus 1205822V) 1205790V minus 120585V2 =
(12)atan2(1198872V 1198862V) 1198861V = minus120582
1V cos(1205790V minus 120585V2) and 1198871V =
minus1205821V sin(1205790V minus 120585V2)(iv) calculate efficiency 120578 = minus119886
11198941198861V2
(v) calculate 1199111= minus(119886
1V minus 1198951198871V)1198861119894 and 1199112119899 = 1199112Re119911
1
In this case study coefficients of fundamental and secondharmonic of current waveform are given by (186) Maximalefficiency of PA associated with the waveform pair (185)and (187) as a function of normalized second harmonicimpedance 119911
2119899= 1199112Re119911
1 is presented in Figure 16(a)
As can be seen from Figure 16(a) efficiency of 078 isachieved at the edge of Smith chart where second harmonicimpedance has small resistive part Corresponding PUFcalculated according to (179) is presented in Figure 16(b)Peak efficiency 120578 = 1205874 = 07854 and peak value of PUF = 1
are attained when second harmonic is short-circuited (whichcorresponds to ideal class-B operation [10 26])
For example for second harmonic impedance 1199112= 01 minus
11989505 and currentwaveform (185) fromAlgorithm32 it followsthat 2120582
2V le 1 minus 1198862V Furthermore according to step (iii)
of above algorithm maximal efficiency of PA is attained
Mathematical Problems in Engineering 23
+j2
+j5
+j1
+j2
+j5
minusj2
minusj5
minusj1
minusj2
minusj5
infin0
078
02
05
1 2 5
075 07 065
06 05
075
07
065
06
05
120578 lt 05
(a)
099
095
091
083 075
067
0 59 051
051
099
095
091
083
075 0
67 059
+j5
+j1
+j2
+j5
minusj5
minusj1
minusj2
minusj5
infin
+j2
minusj2
0 02
05
1 2 5
120578 lt 05
(b)
Figure 16 (a) Contours of maximal efficiency of PA and (b) contours of corresponding PUF associated with the waveform pair (185) and(187) as functions of normalized second harmonic impedance 119911
2119899= 1199112Re119911
1
2
3
4
1
0
2 3 410
Wav
efor
ms
Angle 120579120587
i(120579)
(120579)
Figure 17 Waveform pair (185) and (187) that provides maximalefficiency for 119911
2= 01 minus 11989505
with voltage waveform of type (187) with coefficients 1198862V =
minus00667 1198872V = minus03333 119886
1V = minus09333 and 1198871V = 06667 (see
Figure 17) Corresponding efficiency PUF and normalizedsecond harmonic impedance are 120578 = 07330 PUF = 07572and 1199112119899= 01683 minus 11989508415 respectively
On the other hand for second harmonic impedance 1199112=
01 minus 11989508 and current waveform (185) from Algorithm 32 itfollows that 2120582
2V gt 1 minus 1198862V Then according to step (iii) of
above algorithm maximal efficiency is attained with voltagewaveform of type (187) with coefficients 119886
2V = minus00667 1198872V =
minus05333 1198861V = minus09333 and 119887
1V = 10572 (see Figure 18)Efficiency PUF andnormalized secondharmonic impedanceare 120578 = 07330 PUF = 06332 and 119911
2119899= 01683 minus 11989513465
respectively
2
3
4
1
0
2 3 410
Wav
efor
ms
Angle 120579120587
i(120579)
(120579)
Figure 18 Waveform pair (185) and (187) that provides maximalefficiency for 119911
2= 01 minus 11989508
Case Study 72 As another case study let us consider the effi-ciency of PA providing that current waveform is nonnegativecosine waveform up to third harmonic with maximum valueof amplitude of fundamental harmonic [22] (see also [8])
119894 (120579) = 1 +1 + radic5
2cos 120579 + 2radic5
5cos 2120579 + 5 minus radic5
10cos 3120579
(188)
and voltage waveform is nonnegative waveform of type(187) Load impedances at fundamental second and thirdharmonic are 119911
1= minus(119886
1Vminus1198951198871V)1198861119894 1199112 = minus(1198862Vminus1198951198872V)1198862119894 and
1199113= 0 respectively According to our assumption the load
is passive and therefore Re1199111 gt 0 and Re119911
2 ge 0 which
further imply 11988611198941198861V lt 0 and 119886
21198941198862V le 0 respectively
Because current waveform (188) contains only cosineterms and voltage waveform is the same as in previous case
24 Mathematical Problems in Engineering
08
075
07
06
06
065
065
05
05
075 0
7
120578 lt 05
+j2
+j5
+j1
+j2
+j5
minusj2
minusj5
minusj1
minusj2
minusj5
infin0
02
05
1 2 5
(a)
085
08 0
75 07 0
6
05 04
08
075
0405
060
7
120578 lt 05
+j2
+j5
+j1
+j2
+j5
minusj2
minusj5
minusj1
minusj2
minusj5
infin0 02
05
1 2 5
(b)
Figure 19 (a) Contours of maximal efficiency of PA and (b) contours of corresponding PUF associated with the waveform pair (187)-(188)as functions of normalized second harmonic impedance 119911
2119899= 1199112Re119911
1
study the procedure for calculation of maximal efficiency ofPA with waveform pair (187)-(188) is the same as presentedin Algorithm 32 In this case study the coefficients of funda-mental and second harmonic of current waveform are 119886
1119894=
(1 + radic5)2 and 1198862119894= 2radic55 respectively
Maximal efficiency of PA associated with the waveformpair (187)-(188) as a function of normalized secondharmonicimpedance 119911
2119899= 1199112Re119911
1 is presented in Figure 19(a)
Efficiency of 08 is achieved at the edge of Smith chart wheresecond harmonic impedance has small resistive part Thetheoretical upper bound 120578 = (1 + radic5)4 asymp 08090 isattained when second harmonic is short-circuitedWhen thisupper bound is reached both second and third harmonic areshort-circuited which implies that we are dealing with finiteharmonic class-C [6 8] or dually when current and voltageinterchange their roles with finite harmonic inverse class-C[6 9] Corresponding PUF calculated according to (179) ispresented in Figure 19(b) Peak value of PUF asymp 08541 isattained when second harmonic is short-circuited
For example for second harmonic impedance 1199112
=
007 minus 11989504 and current waveform (188) from Algorithm 32 itfollows that 2120582
2V le 1 minus 1198862V Furthermore according to step
(iii) of Algorithm 32 maximal efficiency of PA is attainedwith voltage waveform of type (187) with coefficients 119886
2V =
minus00626 1198872V = minus03578 119886
1V = minus09374 and 1198871V = 07155 (see
Figure 20) Corresponding efficiency PUF and normalizedsecond harmonic impedance are 120578 = 07584 PUF = 06337and 1199112119899= 01208 minus 11989506904 respectively
On the other hand for 1199112= 005 minus 11989507 and current
waveform (187) it follows that 21205822V gt 1minus119886
2VThen accordingto step (iii) of Algorithm 32 themaximal efficiency is attainedwith voltage waveform of type (187) with coefficients 119886
2V =
minus00447 1198872V = minus06261 119886
1V = minus09318 and 1198871V = 10007 (see
Figure 21) Efficiency PUF and normalized second harmonic
2
3
4
1
0
2 3 410
Wav
efor
ms
Angle 120579120587
i(120579)
(120579)
Figure 20 Waveform pair (187)-(188) that provides maximal effi-ciency for 119911
2= 007 minus 11989504
impedance are 120578 = 07538 PUF = 05314 and 1199112119899= 00868minus
11989512156 respectively
72 Nonnegative Waveforms for 119896 = 3 in PA EfficiencyAnalysis In this subsection we provide another two casestudies of usage of description of nonnegative waveforms inPA efficiency analysis this time with fundamental and thirdharmonic (119896 = 3)
Case Study 73 Let us consider current-voltage pair such thatvoltage is nonnegative waveform with fundamental and thirdharmonic
V (120579) = 1 + 1198861V cos 120579 + 1198871V sin 120579 + 1198863V cos 3120579 + 1198873V sin 3120579
(189)
Mathematical Problems in Engineering 25
2
3
4
1
0
2 3 410
Wav
efor
ms
Angle 120579120587
i(120579)
(120579)
Figure 21 Waveform pair (187)-(188) that provides maximal effi-ciency for 119911
2= 005 minus 11989507
and current is nonnegative cosine waveform given by (188)Load impedances at fundamental second and third har-monic are 119911
1= minus(119886
1V minus 1198951198871V)1198861119894 1199112 = 0 and 119911
3= minus(119886
3V minus
1198951198873V)1198863119894 respectively According to our assumption the load
is passive and therefore Re1199111 gt 0 and Re119911
3 ge 0 which
further imply 11988611198941198861V lt 0 and 119886
31198941198863V le 0
In this subsection we consider the problem of findingmaximal efficiency of PA with waveform pair (188)-(189)for given third harmonic impedance As we mentionedearlier problem of finding maximal efficiency of PA withcurrent-voltage pair (188)-(189) for prescribed third har-monic impedance can be reduced to the problem of findingvoltage waveform of type (189) withmaximal coefficient |119886
1V|for prescribed coefficients of third harmonic (see Section 52)
The following algorithm provides the procedure forcalculation of maximal efficiency with current-voltage pair(188)-(189) The definition of function atan 2(119910 119909) whichappears in step (iii) of the following algorithm is given by(105)
Algorithm 33 (i) Choose 1199113= 1199033+1198951199093such that |119911
3| le 1|119886
3119894|
(ii) calculate 1198863V minus 1198951198873V = minus119911
31198863119894and 120582
3V = radic11988623V + 1198872
3V(iii) if 271205822
3V le (1 minus 21198863V)3 then calculate 119886
1V = minus1 minus 1198863V
and 1198871V = minus3119887
3V else calculate 1205821V = 3(3radic1205823V minus 120582
3V) 1205790V minus120585V3 = (13)atan 2(119887
3V 1198863V) 1198861V = minus1205821V cos(1205790V minus 120585V3) and
1198871V = minus120582
1V sin(1205790V minus 120585V3)(iv) calculate efficiency 120578 = minus119886
11198941198861V2
(v) calculate 1199111= minus(119886
1V minus 1198951198871V)1198861119894 and 1199113119899 = 1199113Re119911
1
In this case study coefficients of fundamental and thirdharmonic of current waveform are 119886
1119894= (1 + radic5)2 and
1198863119894= (5 minus radic5)10 respectively For the waveform pair (188)-
(189) maximal efficiency of PA as a function of normalizedthird harmonic impedance 119911
3119899= 1199113Re119911
1 is presented in
Figure 22 Efficiency of 08 is reached when third harmonicimpedance has small resistive part Peak efficiency 120578 = (1 +
radic5)4 asymp 08090 is achieved when third harmonic is short-circuited
+j2
+j5
+j1
+j2
+j5
minusj2
minusj5
minusj1
minusj2
minusj5
infin0
08
08
075
075
07
07
06
05
02
05
1 2
Figure 22 Contours of maximal efficiency of PA associated withthe waveform pair (188)-(189) as a function of normalized thirdharmonic impedance 119911
3119899= 1199113Re119911
1
For the present case study in what follows we show thatpower utilization factor is proportional to efficiency For volt-age waveform of type (189) it is easy to see that V(120579 + 120587) =
2minusV(120579) holdsThis relation along with the fact that waveformV(120579) that provides maximal efficiency has at least one zeroimplies that max[V(120579)] = 2 On the other hand currentwaveform (188) is cosine waveform with positive coefficientsand therefore max[119894(120579)] = 119894(0) = 2 + 4radic5 Consequentlyaccording to (179) the following relation holds
PUFCase study 73 = 2 (5 minus 2radic5) 120578 = 10557120578 (190)
Clearly the ratio PUF120578 is constant and therefore in this casestudy PUF can be easily calculated from the correspondingefficiency Accordingly peak efficiency and peak value ofPUFCase study 73 = 3radic52 minus 52 = 08541 are attained forthe same voltage waveform (when third harmonic is short-circuited)
In the first example current waveform (188) and 1199113=
02 minus 11989505 imply that 2712058223V le (1 minus 2119886
3V)3 Then according
to Algorithm 33 the voltage waveform of type (189) thatprovides maximal efficiency has the following coefficients1198863V = minus00553 119887
3V = minus01382 1198861V = minus09447 and 119887
1V =
04146 (see Figure 23) Efficiency PUF and normalized thirdharmonic impedance are 120578 = 07643 PUF = 08069 and1199113119899= 03425 minus 11989508564 respectivelyIn the second example current waveform (188) and 119911
3=
01 minus 11989511 imply that 2712058223V gt (1 minus 2119886
3V)3 Then according
to Algorithm 33 the voltage waveform of type (189) thatprovides maximal efficiency has the following coefficients1198863V = minus00276 119887
3V = minus03040 1198861V = minus09391 and 119887
1V =
05807 (see Figure 24) Efficiency PUF and normalized thirdharmonic impedance are 120578 = 07598 PUF = 08021 and1199113119899= 01723 minus 11989518952 respectively
26 Mathematical Problems in Engineering
2
3
4
1
0
2 3 410
Wav
efor
ms
Angle 120579120587
i(120579)
(120579)
Figure 23 Waveform pair (188)-(189) that provides maximal effi-ciency for 119911
3= 02 minus 11989505
2
3
4
1
0
2 3 410
Wav
efor
ms
Angle 120579120587
i(120579)
(120579)
Figure 24 Waveform pair (188)-(189) that provides maximal effi-ciency for 119911
3= 01 minus 11989511
Case Study 74 In this case study let us consider current-voltage pair where current is normalized waveform of type(180) with conduction angle 120572 = 115120587 (207∘) and voltageis nonnegative waveform of type (189) Substitution of 120572 =
115120587 and 119868dc = 1 into (182) leads to 119868119863
= 22535Furthermore substitution of 120572 = 115120587 and 119868
119863= 22535 into
(180) leads to
119894 (120579)
=
22535 [cos 120579 minus cos(1151205872
)] |120579| le115120587
2
0115120587
2le |120579| le 120587
(191)
Similarly substitution of 120572 = 115120587 and 119868119863
= 22535 into(183) and (184) for 119899 = 3 yields coefficients of fundamentaland third harmonic of waveform (191)
1198861119894= 14586 119886
3119894= minus01026 (192)
Because current waveform (191) contains only cosineterms and voltage waveform is the same as in previous case
+j2
+j5
+j1
+j2
+j5
minusj2
minusj5
minusj1
minusj2
minusj5
infin0 02
05
1 2
074
076
08
084
082
078
076
074
Figure 25 Contours of maximal efficiency of PA associated withthe waveform pair (189) and (191) as a function of normalized thirdharmonic impedance 119911
3119899= 1199113Re119911
1
study the procedure for calculation of maximal efficiencyof PA with waveform pair (189)ndash(191) is the same as thatpresented in Algorithm 33 In this case study the coefficientsof fundamental and third harmonic of current waveform aregiven by (192)
For the waveform pair (189) and (191) maximal efficiencyof PA as a function of normalized third harmonic impedance1199113119899= 1199113Re119911
1 is presented in Figure 25 Efficiency of 084
is obtained in vicinity of 1199113119899= 23685 (corresponding to 119911
3=
18750) Peak efficiency 120578 asymp 08421 is achieved for voltagewaveform of type (189) with coefficients 119886
1V = minus2radic3 1198863V =
radic39 and 1198871V = 1198873V = 0
In the course of finding power utilization factor noticethat currentwaveformof type (191) attains itsmaximumvaluefor 120579 = 0 Insertion of max[119894(120579)] = 119894(0) = 278 andmax[V(120579)] = 2 for voltage waveform of type (189) into (179)leads to
PUFCase study 74 = 1439120578 (193)
Again the ratio PUF120578 is constant and PUF can be easilycalculated from the corresponding efficiency Accordinglypeak value of PUFCase study 74 asymp 12118 and peak efficiencyare attained for the same voltage waveform
In the first example current waveform (191) and 1199113=
1 minus 11989502 imply that 2712058223V le (1 minus 2119886
3V)3 Then according to
Algorithm 33 voltage waveform of type (189) which providesmaximal efficiency has coefficients 119886
3V = 01026 1198873V =
00205 1198861V = minus11026 and 119887
1V = minus00616 (see Figure 26)Efficiency PUF and normalized third harmonic impedanceare 120578 = 08042 PUF = 11572 and 119911
3119899= 13228 minus 11989502646
respectivelyIn second example current waveform (191) and 119911
3=
15 minus 11989512 imply that 2712058223V gt (1 minus 2119886
3V)3 Then according to
Algorithm 33 voltage waveform of type (189) which providesmaximal efficiency has coefficients 119886
3V = 01540 1198873V =
Mathematical Problems in Engineering 27
2
3
4
1
0
2 3 410
Wav
efor
ms
Angle 120579120587
i(120579)
(120579)
Figure 26 Waveform pair (189) and (191) that provides maximalefficiency for 119911
3= 1 minus 11989502
2
3
4
1
0
2 3 410
Wav
efor
ms
Angle 120579120587
i(120579)
(120579)
Figure 27 Waveform pair (189) and (191) that provides maximalefficiency for 119911
3= 15 minus 11989512
01232 1198861V = minus11255 and 119887
1V = minus02575 (see Figure 27)Efficiency PUF and normalized third harmonic impedanceare 120578 = 08208 PUF = 11812 and 119911
3119899= 19439 minus 11989515552
respectively
8 Conclusion
In this paper we consider a problem of finding general de-scriptions of various classes of nonnegative waveforms withfundamental and 119896th harmonic These classes include non-negative waveforms with at least one zero nonnegative wave-forms with maximal amplitude of fundamental harmonic forprescribed amplitude of 119896th harmonic nonnegative wave-forms withmaximal coefficient of cosine part of fundamentalharmonic for prescribed coefficients of 119896th harmonic andnonnegative cosine waveforms with at least one zero Mainresults are stated in six propositions (Propositions 1 6 918 22 and 26) four corollaries (Corollaries 2ndash5) twentyremarks and three algorithms Four case studies of usageof closed form descriptions of nonnegative waveforms in PAefficiency analysis are considered in detail in Section 7
Appendices
Here we provide a list of finite sums of trigonometric func-tions used in this paper (Appendix A) and brief account ofthe Chebyshev polynomials (Appendix B)
A List of Some Finite Sums ofTrigonometric Functions
Dirichlet kernel (eg see [27]) is as follows
119863119896minus1
(120591) = 1 + 2
119896minus1
sum
119899=1
cos 119899120591 = sin ((2119896 minus 1) 1205912)sin (1205912)
(A1)
Fejer kernel (eg see [27]) can be expressed in the fol-lowing equivalent forms
119865119896minus1
(120591) =1
119896
119896minus1
sum
119899=0
119863119899(120591) = 1 +
2
119896
119896minus1
sum
119899=1
(119896 minus 119899) cos 119899120591
=(1 minus cos 119896120591)119896 (1 minus cos 120591)
(A2)
Lagrangersquos trigonometric identity (eg see [28]) is as fol-lows
1198781(120591) =
119896minus1
sum
119899=1
sin 119899120591 = sin (1198961205912) sin ((119896 minus 1) 1205912)sin (1205912)
(A3)
In what follows we show that the following three trigono-metric identities also hold
2
119896minus1
sum
119899=1
(119896 minus 119899) sin 119899120591 = 119896 sin 120591 minus sin 1198961205911 minus cos 120591
(A4)
119896minus1
sum
119899=1
cos (119896 minus 2119899) 120591 = sin (119896 minus 1) 120591sin 120591
(A5)
119896minus1
sum
119899=1
119899 (119896 minus 119899) cos (119896 minus 2119899) 120591
=sin (119896120591) cos 120591 minus 119896 cos (119896120591) sin 120591
2sin3120591
(A6)
Denote 1198782(120591) = 2sum
119896minus1
119899=1(119896 minus 119899) sin 119899120591 119878
3(120591) = sum
119896minus1
119899=1cos(119896 minus
2119899)120591 and 1198784(120591) = sum
119896minus1
119899=1119899(119896 minus 119899) cos(119896 minus 2119899)120591
Notice that 1198782(120591) = 2119896119878
1(120591) + 119889119863
119896minus1(120591)119889120591 which
immediately leads to (A4)
28 Mathematical Problems in Engineering
Identity (A5) can be obtained as follows
sin (119896 minus 1) 120591sin 120591
=119890119895(119896minus1)120591
minus 119890minus119895(119896minus1)120591
119890119895120591 minus 119890minus119895120591
= 119890119895119896120591
119890minus2119895120591
minus 119890minus2119895(119896minus1)120591
1 minus 119890minus2119895120591
= 119890119895119896120591
119896minus1
sum
119899=1
119890minus2119895119899120591
=
119896minus1
sum
119899=1
119890119895(119896minus2119899)120591
=
119896minus1
sum
119899=1
cos (119896 minus 2119899) 120591
(A7)
From 4119899(119896 minus 119899) = 1198962
minus (119896 minus 2119899)2 it follows that 4119878
4(120591) =
1198962
1198783(120591) + 119889
2
1198783(120591)119889120591
2 which leads to (A6)
B The Chebyshev Polynomials
The Chebyshev polynomials of the first kind 119881119899(119909) can be
defined by the following relation (eg see [29])
119881119899(119909) = cos 119899120591 when 119909 = cos 120591 (B1)
The Chebyshev polynomials of the second kind 119880119899(119909) can be
defined by the following relation (eg see [29])
119880119899(119909) =
sin (119899 + 1) 120591sin 120591
when 119909 = cos 120591 (B2)
The Chebyshev polynomials satisfy the following recur-rence relations (eg see [29])
1198810(119909) = 1 119881
1(119909) = 119909
119881119899+1
(119909) = 2119909119881119899(119909) minus 119881
119899minus1(119909)
1198800(119909) = 1 119880
1(119909) = 2119909
119880119899+1
(119909) = 2119909119880119899(119909) minus 119880
119899minus1(119909)
(B3)
The first few Chebyshev polynomials of the first and secondkind are 119881
2(119909) = 2119909
2
minus 1 1198813(119909) = 4119909
3
minus 3119909 1198814(119909) = 8119909
4
minus
81199092
+ 1 1198802(119909) = 4119909
2
minus 1 1198803(119909) = 8119909
3
minus 4119909 and 1198804(119909) =
161199094
minus 121199092
+ 1
Conflict of Interests
The authors declare that there is no conflict of interestsregarding the publication of this paper
Acknowledgment
This work is supported by the SerbianMinistry of EducationScience and Technology Development as a part of ProjectTP32016
References
[1] V I Arnolrsquod V S Afrajmovich Y S Ilrsquoyashenko and L PShilrsquonikov Dynamical Systems V Bifurcation Theory and Catas-trophe Theory Springer Berlin Germany 1994
[2] E Polak ldquoOn the mathematical foundations of nondifferen-tiable optimization in engineering designrdquo SIAM Review vol29 no 1 pp 21ndash89 1987
[3] N S Fuzik ldquoBiharmonic modes of a tuned RF power amplifierrdquoRadiotehnika vol 25 no 7 pp 62ndash71 1970 (Russian)
[4] P Colantonio F Giannini G Leuzzi and E Limiti ldquoClass Gapproach for low-voltage high-efficiency PA designrdquo Interna-tional Journal of RF and Microwave Computer-Aided Engineer-ing vol 10 no 6 pp 366ndash378 2000
[5] F H Raab ldquoMaximum efficiency and output of class-F poweramplifiersrdquo IEEE Transactions on Microwave Theory and Tech-niques vol 49 no 6 pp 1162ndash1166 2001
[6] FH Raab ldquoClass-E class-C and class-F power amplifiers basedupon a finite number of harmonicsrdquo IEEE Transactions onMicrowaveTheory and Techniques vol 49 no 8 pp 1462ndash14682001
[7] J D Rhodes ldquoOutput universality inmaximum efficiency linearpower amplifiersrdquo International Journal of Circuit Theory andApplications vol 31 no 4 pp 385ndash405 2003
[8] A Juhas and L A Novak ldquoComments on lsquoClass-E class-Cand classF power amplifier based upon a finite number ofharmonicsrsquordquo IEEE Transactions on Microwave Theory and Tech-niques vol 57 no 6 pp 1623ndash1625 2009
[9] M Roberg and Z Popovic ldquoAnalysis of high-efficiency poweramplifiers with arbitrary output harmonic terminationsrdquo IEEETransactions on Microwave Theory and Techniques vol 59 no8 pp 2037ndash2048 2011
[10] A Grebennikov N O Sokal and M J Franco Switchmode RFPower Amplifiers ElsevierAcademic Press San Diego CalifUSA 2nd edition 2012
[11] T Canning P J Tasker and S C Cripps ldquoContinuous modepower amplifier design using harmonic clipping contourstheory and practicerdquo IEEE Transactions on Microwave Theoryand Techniques vol 62 no 1 pp 100ndash110 2014
[12] A Juhas and L A Novak ldquoGeneral description of nonnegativewaveforms up to second harmonic for power amplifier mod-ellingrdquoMathematical Problems in Engineering vol 2014 ArticleID 709762 18 pages 2014
[13] V I Arnolrsquod V V Goryunov O V Lyashko and V A VasilrsquoevDynamical Systems VIIImdashSingularity Theory II ApplicationsSpringer Berlin Germany 1993
[14] D Siersma ldquoProperties of conflict sets in the planerdquo BanachCenter Publications Polish Academy of Sciences vol 50 no 1 pp267ndash276 1999 Proceedings of the Banach Center Symposiumon Geometry and Topology of Caustics (Caustics rsquo98) WarsawPoland
[15] M van Manen The geometry of conflict sets [Dissertation]Universiteit Utrecht UtrechtThe Netherlands 2003 httpigi-tur-archivelibraryuunldissertations2003-0912-123058c4pdf
[16] Y L Sachkov ldquoMaxwell strata and symmetries in the problemofoptimal rolling of a sphere over a planerdquo Sbornik Mathematicsvol 201 no 7-8 pp 1029ndash1051 2010
[17] I A Bogaevsky ldquoPerestroikas of shocks and singularities ofminimum functionsrdquoPhysicaDNonlinear Phenomena vol 173no 1-2 pp 1ndash28 2002
[18] Y L Sachkov ldquoMaxwell strata in the Euler elastic problemrdquoJournal of Dynamical andControl Systems vol 14 no 2 pp 169ndash234 2008
[19] M Siino and T Koike ldquoTopological classification of black holesgeneric Maxwell set and crease set of a horizonrdquo InternationalJournal ofModern Physics D Gravitation Astrophysics Cosmol-ogy vol 20 no 6 pp 1095ndash1122 2011
Mathematical Problems in Engineering 29
[20] F H Raab ldquoClass-F power amplifiers with maximally flat wave-formsrdquo IEEETransactions onMicrowaveTheory and Techniquesvol 45 no 11 pp 2007ndash2012 1997
[21] A Juhas and L A Novak ldquoMaximally flat waveforms with finitenumber of harmonics in class-F power amplifiersrdquo Mathemat-ical Problems in Engineering vol 2013 Article ID 169590 9pages 2013
[22] L Fejer ldquoUber trigonometrische polynomerdquo Journal fur dieReine und Angewandte Mathematik vol 1916 no 146 pp 53ndash82 1916 (German)
[23] S C Cripps ldquoBessel Waives [microwave bytes]rdquo IEEE Micro-wave Magazine vol 10 no 7 pp 30ndash36 117 2009
[24] L N Bryzgalova ldquoSingularities of the maximum of para-metrically dependent functionrdquo Functional Analysis and ItsApplications vol 11 no 1 pp 49ndash51 1977
[25] V I Arnold A A Davydov V A Vassiliev and V MZakalyukin Mathematical Models of Catastrophes Control ofCatastrophic Process Encyclopedia of Life Support Systems(EOLSS) EOLSS Publishers Oxford UK 2006
[26] S C Cripps RF Power Amplifiers for Wireless CommunicationsArtech House Norwood Mass USA 2nd edition 2006
[27] A Zygmund Trigonometric Series vol 1 Cambridge UniversityPress Cambridge UK 2nd edition 1959
[28] A Jeffrey and H DaiHandbook of Mathematical Formulas andIntegrals ElsevierAcademic Press San Diego Calif USA 4thedition 2008
[29] J C Mason and D C Handscomb Chebyshev PolynomialsChapman amp Hall CRC Press Boca Raton Fla USA 2003
Submit your manuscripts athttpwwwhindawicom
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical Problems in Engineering
Hindawi Publishing Corporationhttpwwwhindawicom
Differential EquationsInternational Journal of
Volume 2014
Applied MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Probability and StatisticsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical PhysicsAdvances in
Complex AnalysisJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
OptimizationJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
CombinatoricsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Operations ResearchAdvances in
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Function Spaces
Abstract and Applied AnalysisHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of Mathematics and Mathematical Sciences
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Algebra
Discrete Dynamics in Nature and Society
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Decision SciencesAdvances in
Discrete MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom
Volume 2014 Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Stochastic AnalysisInternational Journal of
Mathematical Problems in Engineering 21
1
15
15
05
0
minus05
minus15minus15
minus1
minus1 minus05 0 05 1
Coefficient a1
Coe
ffici
enta
3 a1= minus
3(3radica3minus a3)
a1 =
minus1 minus a3
a1 =
1 minusa3
a1 =3(3radic|a3
| +a3)
Figure 14 Parameter space of cosine waveforms for 119896 = 3
For waveform of type (170) with 1198861ge 0 relations 120591
0isin [21205873
41205873] 1198861= minus3119886
3(4cos2120591
0minus 1) and 119886
3= [8cos3120591
0]minus1 lead to
1198861= 3 [
3radic10038161003816100381610038161198863
1003816100381610038161003816 + 1198863] for minus 1 le 1198863le minus
1
8 (174)
Every cosine waveform of type (167) corresponds to apair of real numbers (119886
1 1198863) and vice versa Points (119886
1 1198863)
in grey area in Figure 14 correspond to nonnegative cosinewaveforms for 119896 = 3 The points at the boundary of grey areacorrespond to nonnegative cosinewaveformswith at least onezero A number of shapes of nonnegative cosine waveformswith 119896 = 3 and at least one zero plotted on interval [minus120587 120587]are also presented in Figure 14 The boundary of grey area inFigure 14 consists of four line segments described by relations(171)ndash(174) The common point of line segments (172) and(173) is cusp point with coordinates 119886
1= 0 and 119886
3= 1
Another cusp point with coordinates 1198861= 0 and 119886
3= minus1
is the common point of line segments (171) and (174) Thecommon point of line segments (171)-(172) has coordinates(minus98 18) and common point of line segments (173)-(174)has coordinates (98 minus18) These points are representedby white circle dots and they correspond to maximallyflat cosine waveforms (eg see [21]) White triangle dotswith coordinates (2radic3 minusradic39) and (minus2radic3radic39) refer tothe nonnegative cosine waveforms with maximum value ofamplitude of fundamental harmonic
7 Four Case Studies of Usage of NonnegativeWaveforms in PA Efficiency Analysis
In this section we provide four case studies of usage ofdescription of nonnegative waveforms with fundamental and119896th harmonic in PA efficiency analysis In first two casestudies to be presented in Section 71 voltage is nonnegativewaveform with fundamental and second harmonic with atleast one zero In remaining two case studies to be consideredin Section 72 voltage waveform contains fundamental andthird harmonic
i
v Load
Vdc
Idc
Lch
vL
iL
Cb
+ +in
Figure 15 Generic PA circuit diagram
Let us consider generic PA circuit diagram as shown inFigure 15 We assume here that voltage and current wave-forms at the transistor output are
V (120579) = 1 + 1198861V cos 120579 + 1198871V sin 120579 + 119886119896V cos 119896120579 + 119887119896V sin 119896120579
119894 (120579) = 1 + 1198861119894cos 120579 +
infin
sum
119899=2
119886119899119894cos 119899120579
(175)
where 120579 stands for 120596119905 Both waveforms are normalized in thesense that dc components of voltage and current are 119881dc =
1 and 119868dc = 1 respectively Under assumption that blockingcapacitor 119862
119887behaves as short-circuit at the fundamental and
higher harmonics current and voltage waveforms at the loadare
V119871(120579) = 119886
1V cos 120579 + 1198871V sin 120579 + 119886119896V cos 119896120579 + 119887119896V sin 119896120579
119894119871(120579) = minus119886
1119894cos 120579 minus
infin
sum
119899=2
119886119899119894cos 119899120579
(176)
In terms of coefficients of voltage and current waveforms theload impedance at fundamental harmonic is 119911
1= minus(119886
1V minus
1198951198871V)1198861119894 whereas load impedance at 119896th harmonic is 119911
119896=
minus(119886119896Vminus119895119887119896V)119886119896119894 All other harmonics are short-circuited (119911
119899=
0 for 119899 = 1 and 119899 = 119896) Time average output power of PA (egsee [10]) with waveform pair (175) at fundamental frequencycan be expressed as
1198751= minus
11988611198941198861V
2 (177)
For normalized waveforms (175) with 119881dc = 1 and 119868dc = 1dc power is 119875dc = 1 Consequently PA efficiency 120578 = 119875
1119875dc
(eg see [10 26]) is equal to
120578 = minus11988611198941198861V
2 (178)
Thus time average output power 1198751of PA with pair of nor-
malized waveform (175) is equal to efficiency (178)Power utilization factor (PUF) is defined [26] as ldquothe
ratio of power delivered in a given situation to the power
22 Mathematical Problems in Engineering
delivered by the same device with the same supply voltagein Class A moderdquo Since the output power in class-A modeis 1198751class-A = max[V(120579)] sdotmax[119894(120579)]8 (eg see [9]) it follows
that power utilization factor PUF = 11987511198751class-A for PA with
pair of normalized waveforms (175) can be expressed as
PUF =8120578
max [V (120579)] sdotmax [119894 (120579)] (179)
71 NonnegativeWaveforms for 119896 = 2 in PAEfficiencyAnalysisIn this subsection we provide two case studies of usage ofdescription of nonnegative waveforms with fundamental andsecond harmonic (119896 = 2) in PA efficiency analysis For moreexamples of usage of descriptions of nonnegative waveformswith fundamental and second harmonic in PA efficiencyanalysis see [12]
Case Study 71 In this case study we consider efficiency of PAfor given second harmonic impedance providing that voltageis nonnegative waveform with fundamental and second har-monic and current is ldquohalf-sinerdquo waveform frequently used inefficiency analysis of classical PA operation (eg see [10])
Standard model of current waveform for classical PAoperation has the form (eg see [10 26])
119894119863(120579) =
119868119863[cos 120579 minus cos(120572
2)] |120579| le
120572
2
0120572
2le |120579| le 120587
(180)
where 120572 is conduction angle and 119868119863
gt 0 Since 119894119863(120579) is
even function it immediately follows that its Fourier seriescontains only dc component and cosine terms
119894119863(120579) = 119868dc +
infin
sum
119899=1
119868119899cos 119899120579 (181)
The dc component of the waveform (180) is
119868dc =119868119863120572
2120587[sinc(120572
2) minus cos(120572
2)] (182)
where sinc119909 = (sin119909)119909 The coefficient of the fundamentalharmonic component reads
1198681=119868119863120572
2120587(1 minus sinc120572) (183)
and the coefficient of 119899th harmonic component can bewrittenin the form
119868119899=119868119863
119899120587[sin ((119899 minus 1) 1205722)
(119899 minus 1)minussin ((119899 + 1) 1205722)
(119899 + 1)] 119899 ge 2
(184)
For ldquohalf-sinerdquo current waveform conduction angle is equalto 120587 (class-B conduction angle) According to (182) thisfurther implies that 119868dc = 119868
119863120587 To obtain normalized form of
waveform (180) we set 119868dc = 1which implies that 119868119863= 120587 Fur-
thermore substitution of 120572 = 120587 and 119868119863= 120587 in (180) leads to
119894 (120579) =
120587 cos 120579 |120579| lt120587
2
0120587
2lt |120579| le 120587
(185)
Similarly substitution of 119868119863= 120587 and 120572 = 120587 into (183) and
(184) leads to the coefficients of waveform (185) Coefficientsof fundamental and second harmonic respectively are
1198861119894=120587
2 119886
2119894=2
3 (186)
On the other hand voltage waveform of type (35) for 119896 =2 reads
V (120579) = 1 + 1198861V cos 120579 + 1198871V sin 120579 + 1198862V cos 2120579 + 1198872V sin 2120579
(187)
This waveform contains only fundamental and second har-monic and therefore all harmonics of order higher thantwo are short-circuited (119911
119899= 0 for 119899 gt 2) For current
voltage pair (185) and (187) load impedance at fundamentalharmonic is 119911
1= minus(119886
1V minus 1198951198871V)1198861119894 whereas load impedance
at second harmonic is 1199112= minus(119886
2Vminus1198951198872V)1198862119894 According to ourassumption the load is passive and therefore Re119911
1 gt 0 and
Re1199112 ge 0 which further imply 119886
11198941198861V lt 0 and 119886
21198941198862V le 0
respectivelyIt is easy to see that problem of findingmaximal efficiency
of PA with current-voltage pair (185) and (187) for prescribedsecond harmonic impedance can be reduced to the problemof finding voltage waveform of type (187) with maximal coef-ficient |119886
1V| for prescribed coefficients of second harmonic(see Section 5)
The following algorithm (analogous to Algorithm 22presented in [12]) provides the procedure for calculation ofmaximal efficiency with current-voltage pair (185) and (187)for prescribed second harmonic impedance The definitionof function atan 2(119910 119909) which appears in the step (iii) of thefollowing algorithm is given by (105)
Algorithm 32 (i) Choose 1199112= 1199032+1198951199092such that |119911
2| le 1|119886
2119894|
(ii) calculate 1198862V minus 1198951198872V = minus119911
21198862119894and 120582
2V = radic11988622V + 1198872
2V(iii) if 2120582
2V le 1 minus 1198862V then calculate 119886
1V = minus1 minus 1198862V and
1198871V = minus2119887
2V else calculate 1205821V = radic81205822V(1 minus 1205822V) 1205790V minus 120585V2 =
(12)atan2(1198872V 1198862V) 1198861V = minus120582
1V cos(1205790V minus 120585V2) and 1198871V =
minus1205821V sin(1205790V minus 120585V2)(iv) calculate efficiency 120578 = minus119886
11198941198861V2
(v) calculate 1199111= minus(119886
1V minus 1198951198871V)1198861119894 and 1199112119899 = 1199112Re119911
1
In this case study coefficients of fundamental and secondharmonic of current waveform are given by (186) Maximalefficiency of PA associated with the waveform pair (185)and (187) as a function of normalized second harmonicimpedance 119911
2119899= 1199112Re119911
1 is presented in Figure 16(a)
As can be seen from Figure 16(a) efficiency of 078 isachieved at the edge of Smith chart where second harmonicimpedance has small resistive part Corresponding PUFcalculated according to (179) is presented in Figure 16(b)Peak efficiency 120578 = 1205874 = 07854 and peak value of PUF = 1
are attained when second harmonic is short-circuited (whichcorresponds to ideal class-B operation [10 26])
For example for second harmonic impedance 1199112= 01 minus
11989505 and currentwaveform (185) fromAlgorithm32 it followsthat 2120582
2V le 1 minus 1198862V Furthermore according to step (iii)
of above algorithm maximal efficiency of PA is attained
Mathematical Problems in Engineering 23
+j2
+j5
+j1
+j2
+j5
minusj2
minusj5
minusj1
minusj2
minusj5
infin0
078
02
05
1 2 5
075 07 065
06 05
075
07
065
06
05
120578 lt 05
(a)
099
095
091
083 075
067
0 59 051
051
099
095
091
083
075 0
67 059
+j5
+j1
+j2
+j5
minusj5
minusj1
minusj2
minusj5
infin
+j2
minusj2
0 02
05
1 2 5
120578 lt 05
(b)
Figure 16 (a) Contours of maximal efficiency of PA and (b) contours of corresponding PUF associated with the waveform pair (185) and(187) as functions of normalized second harmonic impedance 119911
2119899= 1199112Re119911
1
2
3
4
1
0
2 3 410
Wav
efor
ms
Angle 120579120587
i(120579)
(120579)
Figure 17 Waveform pair (185) and (187) that provides maximalefficiency for 119911
2= 01 minus 11989505
with voltage waveform of type (187) with coefficients 1198862V =
minus00667 1198872V = minus03333 119886
1V = minus09333 and 1198871V = 06667 (see
Figure 17) Corresponding efficiency PUF and normalizedsecond harmonic impedance are 120578 = 07330 PUF = 07572and 1199112119899= 01683 minus 11989508415 respectively
On the other hand for second harmonic impedance 1199112=
01 minus 11989508 and current waveform (185) from Algorithm 32 itfollows that 2120582
2V gt 1 minus 1198862V Then according to step (iii) of
above algorithm maximal efficiency is attained with voltagewaveform of type (187) with coefficients 119886
2V = minus00667 1198872V =
minus05333 1198861V = minus09333 and 119887
1V = 10572 (see Figure 18)Efficiency PUF andnormalized secondharmonic impedanceare 120578 = 07330 PUF = 06332 and 119911
2119899= 01683 minus 11989513465
respectively
2
3
4
1
0
2 3 410
Wav
efor
ms
Angle 120579120587
i(120579)
(120579)
Figure 18 Waveform pair (185) and (187) that provides maximalefficiency for 119911
2= 01 minus 11989508
Case Study 72 As another case study let us consider the effi-ciency of PA providing that current waveform is nonnegativecosine waveform up to third harmonic with maximum valueof amplitude of fundamental harmonic [22] (see also [8])
119894 (120579) = 1 +1 + radic5
2cos 120579 + 2radic5
5cos 2120579 + 5 minus radic5
10cos 3120579
(188)
and voltage waveform is nonnegative waveform of type(187) Load impedances at fundamental second and thirdharmonic are 119911
1= minus(119886
1Vminus1198951198871V)1198861119894 1199112 = minus(1198862Vminus1198951198872V)1198862119894 and
1199113= 0 respectively According to our assumption the load
is passive and therefore Re1199111 gt 0 and Re119911
2 ge 0 which
further imply 11988611198941198861V lt 0 and 119886
21198941198862V le 0 respectively
Because current waveform (188) contains only cosineterms and voltage waveform is the same as in previous case
24 Mathematical Problems in Engineering
08
075
07
06
06
065
065
05
05
075 0
7
120578 lt 05
+j2
+j5
+j1
+j2
+j5
minusj2
minusj5
minusj1
minusj2
minusj5
infin0
02
05
1 2 5
(a)
085
08 0
75 07 0
6
05 04
08
075
0405
060
7
120578 lt 05
+j2
+j5
+j1
+j2
+j5
minusj2
minusj5
minusj1
minusj2
minusj5
infin0 02
05
1 2 5
(b)
Figure 19 (a) Contours of maximal efficiency of PA and (b) contours of corresponding PUF associated with the waveform pair (187)-(188)as functions of normalized second harmonic impedance 119911
2119899= 1199112Re119911
1
study the procedure for calculation of maximal efficiency ofPA with waveform pair (187)-(188) is the same as presentedin Algorithm 32 In this case study the coefficients of funda-mental and second harmonic of current waveform are 119886
1119894=
(1 + radic5)2 and 1198862119894= 2radic55 respectively
Maximal efficiency of PA associated with the waveformpair (187)-(188) as a function of normalized secondharmonicimpedance 119911
2119899= 1199112Re119911
1 is presented in Figure 19(a)
Efficiency of 08 is achieved at the edge of Smith chart wheresecond harmonic impedance has small resistive part Thetheoretical upper bound 120578 = (1 + radic5)4 asymp 08090 isattained when second harmonic is short-circuitedWhen thisupper bound is reached both second and third harmonic areshort-circuited which implies that we are dealing with finiteharmonic class-C [6 8] or dually when current and voltageinterchange their roles with finite harmonic inverse class-C[6 9] Corresponding PUF calculated according to (179) ispresented in Figure 19(b) Peak value of PUF asymp 08541 isattained when second harmonic is short-circuited
For example for second harmonic impedance 1199112
=
007 minus 11989504 and current waveform (188) from Algorithm 32 itfollows that 2120582
2V le 1 minus 1198862V Furthermore according to step
(iii) of Algorithm 32 maximal efficiency of PA is attainedwith voltage waveform of type (187) with coefficients 119886
2V =
minus00626 1198872V = minus03578 119886
1V = minus09374 and 1198871V = 07155 (see
Figure 20) Corresponding efficiency PUF and normalizedsecond harmonic impedance are 120578 = 07584 PUF = 06337and 1199112119899= 01208 minus 11989506904 respectively
On the other hand for 1199112= 005 minus 11989507 and current
waveform (187) it follows that 21205822V gt 1minus119886
2VThen accordingto step (iii) of Algorithm 32 themaximal efficiency is attainedwith voltage waveform of type (187) with coefficients 119886
2V =
minus00447 1198872V = minus06261 119886
1V = minus09318 and 1198871V = 10007 (see
Figure 21) Efficiency PUF and normalized second harmonic
2
3
4
1
0
2 3 410
Wav
efor
ms
Angle 120579120587
i(120579)
(120579)
Figure 20 Waveform pair (187)-(188) that provides maximal effi-ciency for 119911
2= 007 minus 11989504
impedance are 120578 = 07538 PUF = 05314 and 1199112119899= 00868minus
11989512156 respectively
72 Nonnegative Waveforms for 119896 = 3 in PA EfficiencyAnalysis In this subsection we provide another two casestudies of usage of description of nonnegative waveforms inPA efficiency analysis this time with fundamental and thirdharmonic (119896 = 3)
Case Study 73 Let us consider current-voltage pair such thatvoltage is nonnegative waveform with fundamental and thirdharmonic
V (120579) = 1 + 1198861V cos 120579 + 1198871V sin 120579 + 1198863V cos 3120579 + 1198873V sin 3120579
(189)
Mathematical Problems in Engineering 25
2
3
4
1
0
2 3 410
Wav
efor
ms
Angle 120579120587
i(120579)
(120579)
Figure 21 Waveform pair (187)-(188) that provides maximal effi-ciency for 119911
2= 005 minus 11989507
and current is nonnegative cosine waveform given by (188)Load impedances at fundamental second and third har-monic are 119911
1= minus(119886
1V minus 1198951198871V)1198861119894 1199112 = 0 and 119911
3= minus(119886
3V minus
1198951198873V)1198863119894 respectively According to our assumption the load
is passive and therefore Re1199111 gt 0 and Re119911
3 ge 0 which
further imply 11988611198941198861V lt 0 and 119886
31198941198863V le 0
In this subsection we consider the problem of findingmaximal efficiency of PA with waveform pair (188)-(189)for given third harmonic impedance As we mentionedearlier problem of finding maximal efficiency of PA withcurrent-voltage pair (188)-(189) for prescribed third har-monic impedance can be reduced to the problem of findingvoltage waveform of type (189) withmaximal coefficient |119886
1V|for prescribed coefficients of third harmonic (see Section 52)
The following algorithm provides the procedure forcalculation of maximal efficiency with current-voltage pair(188)-(189) The definition of function atan 2(119910 119909) whichappears in step (iii) of the following algorithm is given by(105)
Algorithm 33 (i) Choose 1199113= 1199033+1198951199093such that |119911
3| le 1|119886
3119894|
(ii) calculate 1198863V minus 1198951198873V = minus119911
31198863119894and 120582
3V = radic11988623V + 1198872
3V(iii) if 271205822
3V le (1 minus 21198863V)3 then calculate 119886
1V = minus1 minus 1198863V
and 1198871V = minus3119887
3V else calculate 1205821V = 3(3radic1205823V minus 120582
3V) 1205790V minus120585V3 = (13)atan 2(119887
3V 1198863V) 1198861V = minus1205821V cos(1205790V minus 120585V3) and
1198871V = minus120582
1V sin(1205790V minus 120585V3)(iv) calculate efficiency 120578 = minus119886
11198941198861V2
(v) calculate 1199111= minus(119886
1V minus 1198951198871V)1198861119894 and 1199113119899 = 1199113Re119911
1
In this case study coefficients of fundamental and thirdharmonic of current waveform are 119886
1119894= (1 + radic5)2 and
1198863119894= (5 minus radic5)10 respectively For the waveform pair (188)-
(189) maximal efficiency of PA as a function of normalizedthird harmonic impedance 119911
3119899= 1199113Re119911
1 is presented in
Figure 22 Efficiency of 08 is reached when third harmonicimpedance has small resistive part Peak efficiency 120578 = (1 +
radic5)4 asymp 08090 is achieved when third harmonic is short-circuited
+j2
+j5
+j1
+j2
+j5
minusj2
minusj5
minusj1
minusj2
minusj5
infin0
08
08
075
075
07
07
06
05
02
05
1 2
Figure 22 Contours of maximal efficiency of PA associated withthe waveform pair (188)-(189) as a function of normalized thirdharmonic impedance 119911
3119899= 1199113Re119911
1
For the present case study in what follows we show thatpower utilization factor is proportional to efficiency For volt-age waveform of type (189) it is easy to see that V(120579 + 120587) =
2minusV(120579) holdsThis relation along with the fact that waveformV(120579) that provides maximal efficiency has at least one zeroimplies that max[V(120579)] = 2 On the other hand currentwaveform (188) is cosine waveform with positive coefficientsand therefore max[119894(120579)] = 119894(0) = 2 + 4radic5 Consequentlyaccording to (179) the following relation holds
PUFCase study 73 = 2 (5 minus 2radic5) 120578 = 10557120578 (190)
Clearly the ratio PUF120578 is constant and therefore in this casestudy PUF can be easily calculated from the correspondingefficiency Accordingly peak efficiency and peak value ofPUFCase study 73 = 3radic52 minus 52 = 08541 are attained forthe same voltage waveform (when third harmonic is short-circuited)
In the first example current waveform (188) and 1199113=
02 minus 11989505 imply that 2712058223V le (1 minus 2119886
3V)3 Then according
to Algorithm 33 the voltage waveform of type (189) thatprovides maximal efficiency has the following coefficients1198863V = minus00553 119887
3V = minus01382 1198861V = minus09447 and 119887
1V =
04146 (see Figure 23) Efficiency PUF and normalized thirdharmonic impedance are 120578 = 07643 PUF = 08069 and1199113119899= 03425 minus 11989508564 respectivelyIn the second example current waveform (188) and 119911
3=
01 minus 11989511 imply that 2712058223V gt (1 minus 2119886
3V)3 Then according
to Algorithm 33 the voltage waveform of type (189) thatprovides maximal efficiency has the following coefficients1198863V = minus00276 119887
3V = minus03040 1198861V = minus09391 and 119887
1V =
05807 (see Figure 24) Efficiency PUF and normalized thirdharmonic impedance are 120578 = 07598 PUF = 08021 and1199113119899= 01723 minus 11989518952 respectively
26 Mathematical Problems in Engineering
2
3
4
1
0
2 3 410
Wav
efor
ms
Angle 120579120587
i(120579)
(120579)
Figure 23 Waveform pair (188)-(189) that provides maximal effi-ciency for 119911
3= 02 minus 11989505
2
3
4
1
0
2 3 410
Wav
efor
ms
Angle 120579120587
i(120579)
(120579)
Figure 24 Waveform pair (188)-(189) that provides maximal effi-ciency for 119911
3= 01 minus 11989511
Case Study 74 In this case study let us consider current-voltage pair where current is normalized waveform of type(180) with conduction angle 120572 = 115120587 (207∘) and voltageis nonnegative waveform of type (189) Substitution of 120572 =
115120587 and 119868dc = 1 into (182) leads to 119868119863
= 22535Furthermore substitution of 120572 = 115120587 and 119868
119863= 22535 into
(180) leads to
119894 (120579)
=
22535 [cos 120579 minus cos(1151205872
)] |120579| le115120587
2
0115120587
2le |120579| le 120587
(191)
Similarly substitution of 120572 = 115120587 and 119868119863
= 22535 into(183) and (184) for 119899 = 3 yields coefficients of fundamentaland third harmonic of waveform (191)
1198861119894= 14586 119886
3119894= minus01026 (192)
Because current waveform (191) contains only cosineterms and voltage waveform is the same as in previous case
+j2
+j5
+j1
+j2
+j5
minusj2
minusj5
minusj1
minusj2
minusj5
infin0 02
05
1 2
074
076
08
084
082
078
076
074
Figure 25 Contours of maximal efficiency of PA associated withthe waveform pair (189) and (191) as a function of normalized thirdharmonic impedance 119911
3119899= 1199113Re119911
1
study the procedure for calculation of maximal efficiencyof PA with waveform pair (189)ndash(191) is the same as thatpresented in Algorithm 33 In this case study the coefficientsof fundamental and third harmonic of current waveform aregiven by (192)
For the waveform pair (189) and (191) maximal efficiencyof PA as a function of normalized third harmonic impedance1199113119899= 1199113Re119911
1 is presented in Figure 25 Efficiency of 084
is obtained in vicinity of 1199113119899= 23685 (corresponding to 119911
3=
18750) Peak efficiency 120578 asymp 08421 is achieved for voltagewaveform of type (189) with coefficients 119886
1V = minus2radic3 1198863V =
radic39 and 1198871V = 1198873V = 0
In the course of finding power utilization factor noticethat currentwaveformof type (191) attains itsmaximumvaluefor 120579 = 0 Insertion of max[119894(120579)] = 119894(0) = 278 andmax[V(120579)] = 2 for voltage waveform of type (189) into (179)leads to
PUFCase study 74 = 1439120578 (193)
Again the ratio PUF120578 is constant and PUF can be easilycalculated from the corresponding efficiency Accordinglypeak value of PUFCase study 74 asymp 12118 and peak efficiencyare attained for the same voltage waveform
In the first example current waveform (191) and 1199113=
1 minus 11989502 imply that 2712058223V le (1 minus 2119886
3V)3 Then according to
Algorithm 33 voltage waveform of type (189) which providesmaximal efficiency has coefficients 119886
3V = 01026 1198873V =
00205 1198861V = minus11026 and 119887
1V = minus00616 (see Figure 26)Efficiency PUF and normalized third harmonic impedanceare 120578 = 08042 PUF = 11572 and 119911
3119899= 13228 minus 11989502646
respectivelyIn second example current waveform (191) and 119911
3=
15 minus 11989512 imply that 2712058223V gt (1 minus 2119886
3V)3 Then according to
Algorithm 33 voltage waveform of type (189) which providesmaximal efficiency has coefficients 119886
3V = 01540 1198873V =
Mathematical Problems in Engineering 27
2
3
4
1
0
2 3 410
Wav
efor
ms
Angle 120579120587
i(120579)
(120579)
Figure 26 Waveform pair (189) and (191) that provides maximalefficiency for 119911
3= 1 minus 11989502
2
3
4
1
0
2 3 410
Wav
efor
ms
Angle 120579120587
i(120579)
(120579)
Figure 27 Waveform pair (189) and (191) that provides maximalefficiency for 119911
3= 15 minus 11989512
01232 1198861V = minus11255 and 119887
1V = minus02575 (see Figure 27)Efficiency PUF and normalized third harmonic impedanceare 120578 = 08208 PUF = 11812 and 119911
3119899= 19439 minus 11989515552
respectively
8 Conclusion
In this paper we consider a problem of finding general de-scriptions of various classes of nonnegative waveforms withfundamental and 119896th harmonic These classes include non-negative waveforms with at least one zero nonnegative wave-forms with maximal amplitude of fundamental harmonic forprescribed amplitude of 119896th harmonic nonnegative wave-forms withmaximal coefficient of cosine part of fundamentalharmonic for prescribed coefficients of 119896th harmonic andnonnegative cosine waveforms with at least one zero Mainresults are stated in six propositions (Propositions 1 6 918 22 and 26) four corollaries (Corollaries 2ndash5) twentyremarks and three algorithms Four case studies of usageof closed form descriptions of nonnegative waveforms in PAefficiency analysis are considered in detail in Section 7
Appendices
Here we provide a list of finite sums of trigonometric func-tions used in this paper (Appendix A) and brief account ofthe Chebyshev polynomials (Appendix B)
A List of Some Finite Sums ofTrigonometric Functions
Dirichlet kernel (eg see [27]) is as follows
119863119896minus1
(120591) = 1 + 2
119896minus1
sum
119899=1
cos 119899120591 = sin ((2119896 minus 1) 1205912)sin (1205912)
(A1)
Fejer kernel (eg see [27]) can be expressed in the fol-lowing equivalent forms
119865119896minus1
(120591) =1
119896
119896minus1
sum
119899=0
119863119899(120591) = 1 +
2
119896
119896minus1
sum
119899=1
(119896 minus 119899) cos 119899120591
=(1 minus cos 119896120591)119896 (1 minus cos 120591)
(A2)
Lagrangersquos trigonometric identity (eg see [28]) is as fol-lows
1198781(120591) =
119896minus1
sum
119899=1
sin 119899120591 = sin (1198961205912) sin ((119896 minus 1) 1205912)sin (1205912)
(A3)
In what follows we show that the following three trigono-metric identities also hold
2
119896minus1
sum
119899=1
(119896 minus 119899) sin 119899120591 = 119896 sin 120591 minus sin 1198961205911 minus cos 120591
(A4)
119896minus1
sum
119899=1
cos (119896 minus 2119899) 120591 = sin (119896 minus 1) 120591sin 120591
(A5)
119896minus1
sum
119899=1
119899 (119896 minus 119899) cos (119896 minus 2119899) 120591
=sin (119896120591) cos 120591 minus 119896 cos (119896120591) sin 120591
2sin3120591
(A6)
Denote 1198782(120591) = 2sum
119896minus1
119899=1(119896 minus 119899) sin 119899120591 119878
3(120591) = sum
119896minus1
119899=1cos(119896 minus
2119899)120591 and 1198784(120591) = sum
119896minus1
119899=1119899(119896 minus 119899) cos(119896 minus 2119899)120591
Notice that 1198782(120591) = 2119896119878
1(120591) + 119889119863
119896minus1(120591)119889120591 which
immediately leads to (A4)
28 Mathematical Problems in Engineering
Identity (A5) can be obtained as follows
sin (119896 minus 1) 120591sin 120591
=119890119895(119896minus1)120591
minus 119890minus119895(119896minus1)120591
119890119895120591 minus 119890minus119895120591
= 119890119895119896120591
119890minus2119895120591
minus 119890minus2119895(119896minus1)120591
1 minus 119890minus2119895120591
= 119890119895119896120591
119896minus1
sum
119899=1
119890minus2119895119899120591
=
119896minus1
sum
119899=1
119890119895(119896minus2119899)120591
=
119896minus1
sum
119899=1
cos (119896 minus 2119899) 120591
(A7)
From 4119899(119896 minus 119899) = 1198962
minus (119896 minus 2119899)2 it follows that 4119878
4(120591) =
1198962
1198783(120591) + 119889
2
1198783(120591)119889120591
2 which leads to (A6)
B The Chebyshev Polynomials
The Chebyshev polynomials of the first kind 119881119899(119909) can be
defined by the following relation (eg see [29])
119881119899(119909) = cos 119899120591 when 119909 = cos 120591 (B1)
The Chebyshev polynomials of the second kind 119880119899(119909) can be
defined by the following relation (eg see [29])
119880119899(119909) =
sin (119899 + 1) 120591sin 120591
when 119909 = cos 120591 (B2)
The Chebyshev polynomials satisfy the following recur-rence relations (eg see [29])
1198810(119909) = 1 119881
1(119909) = 119909
119881119899+1
(119909) = 2119909119881119899(119909) minus 119881
119899minus1(119909)
1198800(119909) = 1 119880
1(119909) = 2119909
119880119899+1
(119909) = 2119909119880119899(119909) minus 119880
119899minus1(119909)
(B3)
The first few Chebyshev polynomials of the first and secondkind are 119881
2(119909) = 2119909
2
minus 1 1198813(119909) = 4119909
3
minus 3119909 1198814(119909) = 8119909
4
minus
81199092
+ 1 1198802(119909) = 4119909
2
minus 1 1198803(119909) = 8119909
3
minus 4119909 and 1198804(119909) =
161199094
minus 121199092
+ 1
Conflict of Interests
The authors declare that there is no conflict of interestsregarding the publication of this paper
Acknowledgment
This work is supported by the SerbianMinistry of EducationScience and Technology Development as a part of ProjectTP32016
References
[1] V I Arnolrsquod V S Afrajmovich Y S Ilrsquoyashenko and L PShilrsquonikov Dynamical Systems V Bifurcation Theory and Catas-trophe Theory Springer Berlin Germany 1994
[2] E Polak ldquoOn the mathematical foundations of nondifferen-tiable optimization in engineering designrdquo SIAM Review vol29 no 1 pp 21ndash89 1987
[3] N S Fuzik ldquoBiharmonic modes of a tuned RF power amplifierrdquoRadiotehnika vol 25 no 7 pp 62ndash71 1970 (Russian)
[4] P Colantonio F Giannini G Leuzzi and E Limiti ldquoClass Gapproach for low-voltage high-efficiency PA designrdquo Interna-tional Journal of RF and Microwave Computer-Aided Engineer-ing vol 10 no 6 pp 366ndash378 2000
[5] F H Raab ldquoMaximum efficiency and output of class-F poweramplifiersrdquo IEEE Transactions on Microwave Theory and Tech-niques vol 49 no 6 pp 1162ndash1166 2001
[6] FH Raab ldquoClass-E class-C and class-F power amplifiers basedupon a finite number of harmonicsrdquo IEEE Transactions onMicrowaveTheory and Techniques vol 49 no 8 pp 1462ndash14682001
[7] J D Rhodes ldquoOutput universality inmaximum efficiency linearpower amplifiersrdquo International Journal of Circuit Theory andApplications vol 31 no 4 pp 385ndash405 2003
[8] A Juhas and L A Novak ldquoComments on lsquoClass-E class-Cand classF power amplifier based upon a finite number ofharmonicsrsquordquo IEEE Transactions on Microwave Theory and Tech-niques vol 57 no 6 pp 1623ndash1625 2009
[9] M Roberg and Z Popovic ldquoAnalysis of high-efficiency poweramplifiers with arbitrary output harmonic terminationsrdquo IEEETransactions on Microwave Theory and Techniques vol 59 no8 pp 2037ndash2048 2011
[10] A Grebennikov N O Sokal and M J Franco Switchmode RFPower Amplifiers ElsevierAcademic Press San Diego CalifUSA 2nd edition 2012
[11] T Canning P J Tasker and S C Cripps ldquoContinuous modepower amplifier design using harmonic clipping contourstheory and practicerdquo IEEE Transactions on Microwave Theoryand Techniques vol 62 no 1 pp 100ndash110 2014
[12] A Juhas and L A Novak ldquoGeneral description of nonnegativewaveforms up to second harmonic for power amplifier mod-ellingrdquoMathematical Problems in Engineering vol 2014 ArticleID 709762 18 pages 2014
[13] V I Arnolrsquod V V Goryunov O V Lyashko and V A VasilrsquoevDynamical Systems VIIImdashSingularity Theory II ApplicationsSpringer Berlin Germany 1993
[14] D Siersma ldquoProperties of conflict sets in the planerdquo BanachCenter Publications Polish Academy of Sciences vol 50 no 1 pp267ndash276 1999 Proceedings of the Banach Center Symposiumon Geometry and Topology of Caustics (Caustics rsquo98) WarsawPoland
[15] M van Manen The geometry of conflict sets [Dissertation]Universiteit Utrecht UtrechtThe Netherlands 2003 httpigi-tur-archivelibraryuunldissertations2003-0912-123058c4pdf
[16] Y L Sachkov ldquoMaxwell strata and symmetries in the problemofoptimal rolling of a sphere over a planerdquo Sbornik Mathematicsvol 201 no 7-8 pp 1029ndash1051 2010
[17] I A Bogaevsky ldquoPerestroikas of shocks and singularities ofminimum functionsrdquoPhysicaDNonlinear Phenomena vol 173no 1-2 pp 1ndash28 2002
[18] Y L Sachkov ldquoMaxwell strata in the Euler elastic problemrdquoJournal of Dynamical andControl Systems vol 14 no 2 pp 169ndash234 2008
[19] M Siino and T Koike ldquoTopological classification of black holesgeneric Maxwell set and crease set of a horizonrdquo InternationalJournal ofModern Physics D Gravitation Astrophysics Cosmol-ogy vol 20 no 6 pp 1095ndash1122 2011
Mathematical Problems in Engineering 29
[20] F H Raab ldquoClass-F power amplifiers with maximally flat wave-formsrdquo IEEETransactions onMicrowaveTheory and Techniquesvol 45 no 11 pp 2007ndash2012 1997
[21] A Juhas and L A Novak ldquoMaximally flat waveforms with finitenumber of harmonics in class-F power amplifiersrdquo Mathemat-ical Problems in Engineering vol 2013 Article ID 169590 9pages 2013
[22] L Fejer ldquoUber trigonometrische polynomerdquo Journal fur dieReine und Angewandte Mathematik vol 1916 no 146 pp 53ndash82 1916 (German)
[23] S C Cripps ldquoBessel Waives [microwave bytes]rdquo IEEE Micro-wave Magazine vol 10 no 7 pp 30ndash36 117 2009
[24] L N Bryzgalova ldquoSingularities of the maximum of para-metrically dependent functionrdquo Functional Analysis and ItsApplications vol 11 no 1 pp 49ndash51 1977
[25] V I Arnold A A Davydov V A Vassiliev and V MZakalyukin Mathematical Models of Catastrophes Control ofCatastrophic Process Encyclopedia of Life Support Systems(EOLSS) EOLSS Publishers Oxford UK 2006
[26] S C Cripps RF Power Amplifiers for Wireless CommunicationsArtech House Norwood Mass USA 2nd edition 2006
[27] A Zygmund Trigonometric Series vol 1 Cambridge UniversityPress Cambridge UK 2nd edition 1959
[28] A Jeffrey and H DaiHandbook of Mathematical Formulas andIntegrals ElsevierAcademic Press San Diego Calif USA 4thedition 2008
[29] J C Mason and D C Handscomb Chebyshev PolynomialsChapman amp Hall CRC Press Boca Raton Fla USA 2003
Submit your manuscripts athttpwwwhindawicom
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical Problems in Engineering
Hindawi Publishing Corporationhttpwwwhindawicom
Differential EquationsInternational Journal of
Volume 2014
Applied MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Probability and StatisticsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical PhysicsAdvances in
Complex AnalysisJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
OptimizationJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
CombinatoricsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Operations ResearchAdvances in
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Function Spaces
Abstract and Applied AnalysisHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of Mathematics and Mathematical Sciences
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Algebra
Discrete Dynamics in Nature and Society
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Decision SciencesAdvances in
Discrete MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom
Volume 2014 Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Stochastic AnalysisInternational Journal of
22 Mathematical Problems in Engineering
delivered by the same device with the same supply voltagein Class A moderdquo Since the output power in class-A modeis 1198751class-A = max[V(120579)] sdotmax[119894(120579)]8 (eg see [9]) it follows
that power utilization factor PUF = 11987511198751class-A for PA with
pair of normalized waveforms (175) can be expressed as
PUF =8120578
max [V (120579)] sdotmax [119894 (120579)] (179)
71 NonnegativeWaveforms for 119896 = 2 in PAEfficiencyAnalysisIn this subsection we provide two case studies of usage ofdescription of nonnegative waveforms with fundamental andsecond harmonic (119896 = 2) in PA efficiency analysis For moreexamples of usage of descriptions of nonnegative waveformswith fundamental and second harmonic in PA efficiencyanalysis see [12]
Case Study 71 In this case study we consider efficiency of PAfor given second harmonic impedance providing that voltageis nonnegative waveform with fundamental and second har-monic and current is ldquohalf-sinerdquo waveform frequently used inefficiency analysis of classical PA operation (eg see [10])
Standard model of current waveform for classical PAoperation has the form (eg see [10 26])
119894119863(120579) =
119868119863[cos 120579 minus cos(120572
2)] |120579| le
120572
2
0120572
2le |120579| le 120587
(180)
where 120572 is conduction angle and 119868119863
gt 0 Since 119894119863(120579) is
even function it immediately follows that its Fourier seriescontains only dc component and cosine terms
119894119863(120579) = 119868dc +
infin
sum
119899=1
119868119899cos 119899120579 (181)
The dc component of the waveform (180) is
119868dc =119868119863120572
2120587[sinc(120572
2) minus cos(120572
2)] (182)
where sinc119909 = (sin119909)119909 The coefficient of the fundamentalharmonic component reads
1198681=119868119863120572
2120587(1 minus sinc120572) (183)
and the coefficient of 119899th harmonic component can bewrittenin the form
119868119899=119868119863
119899120587[sin ((119899 minus 1) 1205722)
(119899 minus 1)minussin ((119899 + 1) 1205722)
(119899 + 1)] 119899 ge 2
(184)
For ldquohalf-sinerdquo current waveform conduction angle is equalto 120587 (class-B conduction angle) According to (182) thisfurther implies that 119868dc = 119868
119863120587 To obtain normalized form of
waveform (180) we set 119868dc = 1which implies that 119868119863= 120587 Fur-
thermore substitution of 120572 = 120587 and 119868119863= 120587 in (180) leads to
119894 (120579) =
120587 cos 120579 |120579| lt120587
2
0120587
2lt |120579| le 120587
(185)
Similarly substitution of 119868119863= 120587 and 120572 = 120587 into (183) and
(184) leads to the coefficients of waveform (185) Coefficientsof fundamental and second harmonic respectively are
1198861119894=120587
2 119886
2119894=2
3 (186)
On the other hand voltage waveform of type (35) for 119896 =2 reads
V (120579) = 1 + 1198861V cos 120579 + 1198871V sin 120579 + 1198862V cos 2120579 + 1198872V sin 2120579
(187)
This waveform contains only fundamental and second har-monic and therefore all harmonics of order higher thantwo are short-circuited (119911
119899= 0 for 119899 gt 2) For current
voltage pair (185) and (187) load impedance at fundamentalharmonic is 119911
1= minus(119886
1V minus 1198951198871V)1198861119894 whereas load impedance
at second harmonic is 1199112= minus(119886
2Vminus1198951198872V)1198862119894 According to ourassumption the load is passive and therefore Re119911
1 gt 0 and
Re1199112 ge 0 which further imply 119886
11198941198861V lt 0 and 119886
21198941198862V le 0
respectivelyIt is easy to see that problem of findingmaximal efficiency
of PA with current-voltage pair (185) and (187) for prescribedsecond harmonic impedance can be reduced to the problemof finding voltage waveform of type (187) with maximal coef-ficient |119886
1V| for prescribed coefficients of second harmonic(see Section 5)
The following algorithm (analogous to Algorithm 22presented in [12]) provides the procedure for calculation ofmaximal efficiency with current-voltage pair (185) and (187)for prescribed second harmonic impedance The definitionof function atan 2(119910 119909) which appears in the step (iii) of thefollowing algorithm is given by (105)
Algorithm 32 (i) Choose 1199112= 1199032+1198951199092such that |119911
2| le 1|119886
2119894|
(ii) calculate 1198862V minus 1198951198872V = minus119911
21198862119894and 120582
2V = radic11988622V + 1198872
2V(iii) if 2120582
2V le 1 minus 1198862V then calculate 119886
1V = minus1 minus 1198862V and
1198871V = minus2119887
2V else calculate 1205821V = radic81205822V(1 minus 1205822V) 1205790V minus 120585V2 =
(12)atan2(1198872V 1198862V) 1198861V = minus120582
1V cos(1205790V minus 120585V2) and 1198871V =
minus1205821V sin(1205790V minus 120585V2)(iv) calculate efficiency 120578 = minus119886
11198941198861V2
(v) calculate 1199111= minus(119886
1V minus 1198951198871V)1198861119894 and 1199112119899 = 1199112Re119911
1
In this case study coefficients of fundamental and secondharmonic of current waveform are given by (186) Maximalefficiency of PA associated with the waveform pair (185)and (187) as a function of normalized second harmonicimpedance 119911
2119899= 1199112Re119911
1 is presented in Figure 16(a)
As can be seen from Figure 16(a) efficiency of 078 isachieved at the edge of Smith chart where second harmonicimpedance has small resistive part Corresponding PUFcalculated according to (179) is presented in Figure 16(b)Peak efficiency 120578 = 1205874 = 07854 and peak value of PUF = 1
are attained when second harmonic is short-circuited (whichcorresponds to ideal class-B operation [10 26])
For example for second harmonic impedance 1199112= 01 minus
11989505 and currentwaveform (185) fromAlgorithm32 it followsthat 2120582
2V le 1 minus 1198862V Furthermore according to step (iii)
of above algorithm maximal efficiency of PA is attained
Mathematical Problems in Engineering 23
+j2
+j5
+j1
+j2
+j5
minusj2
minusj5
minusj1
minusj2
minusj5
infin0
078
02
05
1 2 5
075 07 065
06 05
075
07
065
06
05
120578 lt 05
(a)
099
095
091
083 075
067
0 59 051
051
099
095
091
083
075 0
67 059
+j5
+j1
+j2
+j5
minusj5
minusj1
minusj2
minusj5
infin
+j2
minusj2
0 02
05
1 2 5
120578 lt 05
(b)
Figure 16 (a) Contours of maximal efficiency of PA and (b) contours of corresponding PUF associated with the waveform pair (185) and(187) as functions of normalized second harmonic impedance 119911
2119899= 1199112Re119911
1
2
3
4
1
0
2 3 410
Wav
efor
ms
Angle 120579120587
i(120579)
(120579)
Figure 17 Waveform pair (185) and (187) that provides maximalefficiency for 119911
2= 01 minus 11989505
with voltage waveform of type (187) with coefficients 1198862V =
minus00667 1198872V = minus03333 119886
1V = minus09333 and 1198871V = 06667 (see
Figure 17) Corresponding efficiency PUF and normalizedsecond harmonic impedance are 120578 = 07330 PUF = 07572and 1199112119899= 01683 minus 11989508415 respectively
On the other hand for second harmonic impedance 1199112=
01 minus 11989508 and current waveform (185) from Algorithm 32 itfollows that 2120582
2V gt 1 minus 1198862V Then according to step (iii) of
above algorithm maximal efficiency is attained with voltagewaveform of type (187) with coefficients 119886
2V = minus00667 1198872V =
minus05333 1198861V = minus09333 and 119887
1V = 10572 (see Figure 18)Efficiency PUF andnormalized secondharmonic impedanceare 120578 = 07330 PUF = 06332 and 119911
2119899= 01683 minus 11989513465
respectively
2
3
4
1
0
2 3 410
Wav
efor
ms
Angle 120579120587
i(120579)
(120579)
Figure 18 Waveform pair (185) and (187) that provides maximalefficiency for 119911
2= 01 minus 11989508
Case Study 72 As another case study let us consider the effi-ciency of PA providing that current waveform is nonnegativecosine waveform up to third harmonic with maximum valueof amplitude of fundamental harmonic [22] (see also [8])
119894 (120579) = 1 +1 + radic5
2cos 120579 + 2radic5
5cos 2120579 + 5 minus radic5
10cos 3120579
(188)
and voltage waveform is nonnegative waveform of type(187) Load impedances at fundamental second and thirdharmonic are 119911
1= minus(119886
1Vminus1198951198871V)1198861119894 1199112 = minus(1198862Vminus1198951198872V)1198862119894 and
1199113= 0 respectively According to our assumption the load
is passive and therefore Re1199111 gt 0 and Re119911
2 ge 0 which
further imply 11988611198941198861V lt 0 and 119886
21198941198862V le 0 respectively
Because current waveform (188) contains only cosineterms and voltage waveform is the same as in previous case
24 Mathematical Problems in Engineering
08
075
07
06
06
065
065
05
05
075 0
7
120578 lt 05
+j2
+j5
+j1
+j2
+j5
minusj2
minusj5
minusj1
minusj2
minusj5
infin0
02
05
1 2 5
(a)
085
08 0
75 07 0
6
05 04
08
075
0405
060
7
120578 lt 05
+j2
+j5
+j1
+j2
+j5
minusj2
minusj5
minusj1
minusj2
minusj5
infin0 02
05
1 2 5
(b)
Figure 19 (a) Contours of maximal efficiency of PA and (b) contours of corresponding PUF associated with the waveform pair (187)-(188)as functions of normalized second harmonic impedance 119911
2119899= 1199112Re119911
1
study the procedure for calculation of maximal efficiency ofPA with waveform pair (187)-(188) is the same as presentedin Algorithm 32 In this case study the coefficients of funda-mental and second harmonic of current waveform are 119886
1119894=
(1 + radic5)2 and 1198862119894= 2radic55 respectively
Maximal efficiency of PA associated with the waveformpair (187)-(188) as a function of normalized secondharmonicimpedance 119911
2119899= 1199112Re119911
1 is presented in Figure 19(a)
Efficiency of 08 is achieved at the edge of Smith chart wheresecond harmonic impedance has small resistive part Thetheoretical upper bound 120578 = (1 + radic5)4 asymp 08090 isattained when second harmonic is short-circuitedWhen thisupper bound is reached both second and third harmonic areshort-circuited which implies that we are dealing with finiteharmonic class-C [6 8] or dually when current and voltageinterchange their roles with finite harmonic inverse class-C[6 9] Corresponding PUF calculated according to (179) ispresented in Figure 19(b) Peak value of PUF asymp 08541 isattained when second harmonic is short-circuited
For example for second harmonic impedance 1199112
=
007 minus 11989504 and current waveform (188) from Algorithm 32 itfollows that 2120582
2V le 1 minus 1198862V Furthermore according to step
(iii) of Algorithm 32 maximal efficiency of PA is attainedwith voltage waveform of type (187) with coefficients 119886
2V =
minus00626 1198872V = minus03578 119886
1V = minus09374 and 1198871V = 07155 (see
Figure 20) Corresponding efficiency PUF and normalizedsecond harmonic impedance are 120578 = 07584 PUF = 06337and 1199112119899= 01208 minus 11989506904 respectively
On the other hand for 1199112= 005 minus 11989507 and current
waveform (187) it follows that 21205822V gt 1minus119886
2VThen accordingto step (iii) of Algorithm 32 themaximal efficiency is attainedwith voltage waveform of type (187) with coefficients 119886
2V =
minus00447 1198872V = minus06261 119886
1V = minus09318 and 1198871V = 10007 (see
Figure 21) Efficiency PUF and normalized second harmonic
2
3
4
1
0
2 3 410
Wav
efor
ms
Angle 120579120587
i(120579)
(120579)
Figure 20 Waveform pair (187)-(188) that provides maximal effi-ciency for 119911
2= 007 minus 11989504
impedance are 120578 = 07538 PUF = 05314 and 1199112119899= 00868minus
11989512156 respectively
72 Nonnegative Waveforms for 119896 = 3 in PA EfficiencyAnalysis In this subsection we provide another two casestudies of usage of description of nonnegative waveforms inPA efficiency analysis this time with fundamental and thirdharmonic (119896 = 3)
Case Study 73 Let us consider current-voltage pair such thatvoltage is nonnegative waveform with fundamental and thirdharmonic
V (120579) = 1 + 1198861V cos 120579 + 1198871V sin 120579 + 1198863V cos 3120579 + 1198873V sin 3120579
(189)
Mathematical Problems in Engineering 25
2
3
4
1
0
2 3 410
Wav
efor
ms
Angle 120579120587
i(120579)
(120579)
Figure 21 Waveform pair (187)-(188) that provides maximal effi-ciency for 119911
2= 005 minus 11989507
and current is nonnegative cosine waveform given by (188)Load impedances at fundamental second and third har-monic are 119911
1= minus(119886
1V minus 1198951198871V)1198861119894 1199112 = 0 and 119911
3= minus(119886
3V minus
1198951198873V)1198863119894 respectively According to our assumption the load
is passive and therefore Re1199111 gt 0 and Re119911
3 ge 0 which
further imply 11988611198941198861V lt 0 and 119886
31198941198863V le 0
In this subsection we consider the problem of findingmaximal efficiency of PA with waveform pair (188)-(189)for given third harmonic impedance As we mentionedearlier problem of finding maximal efficiency of PA withcurrent-voltage pair (188)-(189) for prescribed third har-monic impedance can be reduced to the problem of findingvoltage waveform of type (189) withmaximal coefficient |119886
1V|for prescribed coefficients of third harmonic (see Section 52)
The following algorithm provides the procedure forcalculation of maximal efficiency with current-voltage pair(188)-(189) The definition of function atan 2(119910 119909) whichappears in step (iii) of the following algorithm is given by(105)
Algorithm 33 (i) Choose 1199113= 1199033+1198951199093such that |119911
3| le 1|119886
3119894|
(ii) calculate 1198863V minus 1198951198873V = minus119911
31198863119894and 120582
3V = radic11988623V + 1198872
3V(iii) if 271205822
3V le (1 minus 21198863V)3 then calculate 119886
1V = minus1 minus 1198863V
and 1198871V = minus3119887
3V else calculate 1205821V = 3(3radic1205823V minus 120582
3V) 1205790V minus120585V3 = (13)atan 2(119887
3V 1198863V) 1198861V = minus1205821V cos(1205790V minus 120585V3) and
1198871V = minus120582
1V sin(1205790V minus 120585V3)(iv) calculate efficiency 120578 = minus119886
11198941198861V2
(v) calculate 1199111= minus(119886
1V minus 1198951198871V)1198861119894 and 1199113119899 = 1199113Re119911
1
In this case study coefficients of fundamental and thirdharmonic of current waveform are 119886
1119894= (1 + radic5)2 and
1198863119894= (5 minus radic5)10 respectively For the waveform pair (188)-
(189) maximal efficiency of PA as a function of normalizedthird harmonic impedance 119911
3119899= 1199113Re119911
1 is presented in
Figure 22 Efficiency of 08 is reached when third harmonicimpedance has small resistive part Peak efficiency 120578 = (1 +
radic5)4 asymp 08090 is achieved when third harmonic is short-circuited
+j2
+j5
+j1
+j2
+j5
minusj2
minusj5
minusj1
minusj2
minusj5
infin0
08
08
075
075
07
07
06
05
02
05
1 2
Figure 22 Contours of maximal efficiency of PA associated withthe waveform pair (188)-(189) as a function of normalized thirdharmonic impedance 119911
3119899= 1199113Re119911
1
For the present case study in what follows we show thatpower utilization factor is proportional to efficiency For volt-age waveform of type (189) it is easy to see that V(120579 + 120587) =
2minusV(120579) holdsThis relation along with the fact that waveformV(120579) that provides maximal efficiency has at least one zeroimplies that max[V(120579)] = 2 On the other hand currentwaveform (188) is cosine waveform with positive coefficientsand therefore max[119894(120579)] = 119894(0) = 2 + 4radic5 Consequentlyaccording to (179) the following relation holds
PUFCase study 73 = 2 (5 minus 2radic5) 120578 = 10557120578 (190)
Clearly the ratio PUF120578 is constant and therefore in this casestudy PUF can be easily calculated from the correspondingefficiency Accordingly peak efficiency and peak value ofPUFCase study 73 = 3radic52 minus 52 = 08541 are attained forthe same voltage waveform (when third harmonic is short-circuited)
In the first example current waveform (188) and 1199113=
02 minus 11989505 imply that 2712058223V le (1 minus 2119886
3V)3 Then according
to Algorithm 33 the voltage waveform of type (189) thatprovides maximal efficiency has the following coefficients1198863V = minus00553 119887
3V = minus01382 1198861V = minus09447 and 119887
1V =
04146 (see Figure 23) Efficiency PUF and normalized thirdharmonic impedance are 120578 = 07643 PUF = 08069 and1199113119899= 03425 minus 11989508564 respectivelyIn the second example current waveform (188) and 119911
3=
01 minus 11989511 imply that 2712058223V gt (1 minus 2119886
3V)3 Then according
to Algorithm 33 the voltage waveform of type (189) thatprovides maximal efficiency has the following coefficients1198863V = minus00276 119887
3V = minus03040 1198861V = minus09391 and 119887
1V =
05807 (see Figure 24) Efficiency PUF and normalized thirdharmonic impedance are 120578 = 07598 PUF = 08021 and1199113119899= 01723 minus 11989518952 respectively
26 Mathematical Problems in Engineering
2
3
4
1
0
2 3 410
Wav
efor
ms
Angle 120579120587
i(120579)
(120579)
Figure 23 Waveform pair (188)-(189) that provides maximal effi-ciency for 119911
3= 02 minus 11989505
2
3
4
1
0
2 3 410
Wav
efor
ms
Angle 120579120587
i(120579)
(120579)
Figure 24 Waveform pair (188)-(189) that provides maximal effi-ciency for 119911
3= 01 minus 11989511
Case Study 74 In this case study let us consider current-voltage pair where current is normalized waveform of type(180) with conduction angle 120572 = 115120587 (207∘) and voltageis nonnegative waveform of type (189) Substitution of 120572 =
115120587 and 119868dc = 1 into (182) leads to 119868119863
= 22535Furthermore substitution of 120572 = 115120587 and 119868
119863= 22535 into
(180) leads to
119894 (120579)
=
22535 [cos 120579 minus cos(1151205872
)] |120579| le115120587
2
0115120587
2le |120579| le 120587
(191)
Similarly substitution of 120572 = 115120587 and 119868119863
= 22535 into(183) and (184) for 119899 = 3 yields coefficients of fundamentaland third harmonic of waveform (191)
1198861119894= 14586 119886
3119894= minus01026 (192)
Because current waveform (191) contains only cosineterms and voltage waveform is the same as in previous case
+j2
+j5
+j1
+j2
+j5
minusj2
minusj5
minusj1
minusj2
minusj5
infin0 02
05
1 2
074
076
08
084
082
078
076
074
Figure 25 Contours of maximal efficiency of PA associated withthe waveform pair (189) and (191) as a function of normalized thirdharmonic impedance 119911
3119899= 1199113Re119911
1
study the procedure for calculation of maximal efficiencyof PA with waveform pair (189)ndash(191) is the same as thatpresented in Algorithm 33 In this case study the coefficientsof fundamental and third harmonic of current waveform aregiven by (192)
For the waveform pair (189) and (191) maximal efficiencyof PA as a function of normalized third harmonic impedance1199113119899= 1199113Re119911
1 is presented in Figure 25 Efficiency of 084
is obtained in vicinity of 1199113119899= 23685 (corresponding to 119911
3=
18750) Peak efficiency 120578 asymp 08421 is achieved for voltagewaveform of type (189) with coefficients 119886
1V = minus2radic3 1198863V =
radic39 and 1198871V = 1198873V = 0
In the course of finding power utilization factor noticethat currentwaveformof type (191) attains itsmaximumvaluefor 120579 = 0 Insertion of max[119894(120579)] = 119894(0) = 278 andmax[V(120579)] = 2 for voltage waveform of type (189) into (179)leads to
PUFCase study 74 = 1439120578 (193)
Again the ratio PUF120578 is constant and PUF can be easilycalculated from the corresponding efficiency Accordinglypeak value of PUFCase study 74 asymp 12118 and peak efficiencyare attained for the same voltage waveform
In the first example current waveform (191) and 1199113=
1 minus 11989502 imply that 2712058223V le (1 minus 2119886
3V)3 Then according to
Algorithm 33 voltage waveform of type (189) which providesmaximal efficiency has coefficients 119886
3V = 01026 1198873V =
00205 1198861V = minus11026 and 119887
1V = minus00616 (see Figure 26)Efficiency PUF and normalized third harmonic impedanceare 120578 = 08042 PUF = 11572 and 119911
3119899= 13228 minus 11989502646
respectivelyIn second example current waveform (191) and 119911
3=
15 minus 11989512 imply that 2712058223V gt (1 minus 2119886
3V)3 Then according to
Algorithm 33 voltage waveform of type (189) which providesmaximal efficiency has coefficients 119886
3V = 01540 1198873V =
Mathematical Problems in Engineering 27
2
3
4
1
0
2 3 410
Wav
efor
ms
Angle 120579120587
i(120579)
(120579)
Figure 26 Waveform pair (189) and (191) that provides maximalefficiency for 119911
3= 1 minus 11989502
2
3
4
1
0
2 3 410
Wav
efor
ms
Angle 120579120587
i(120579)
(120579)
Figure 27 Waveform pair (189) and (191) that provides maximalefficiency for 119911
3= 15 minus 11989512
01232 1198861V = minus11255 and 119887
1V = minus02575 (see Figure 27)Efficiency PUF and normalized third harmonic impedanceare 120578 = 08208 PUF = 11812 and 119911
3119899= 19439 minus 11989515552
respectively
8 Conclusion
In this paper we consider a problem of finding general de-scriptions of various classes of nonnegative waveforms withfundamental and 119896th harmonic These classes include non-negative waveforms with at least one zero nonnegative wave-forms with maximal amplitude of fundamental harmonic forprescribed amplitude of 119896th harmonic nonnegative wave-forms withmaximal coefficient of cosine part of fundamentalharmonic for prescribed coefficients of 119896th harmonic andnonnegative cosine waveforms with at least one zero Mainresults are stated in six propositions (Propositions 1 6 918 22 and 26) four corollaries (Corollaries 2ndash5) twentyremarks and three algorithms Four case studies of usageof closed form descriptions of nonnegative waveforms in PAefficiency analysis are considered in detail in Section 7
Appendices
Here we provide a list of finite sums of trigonometric func-tions used in this paper (Appendix A) and brief account ofthe Chebyshev polynomials (Appendix B)
A List of Some Finite Sums ofTrigonometric Functions
Dirichlet kernel (eg see [27]) is as follows
119863119896minus1
(120591) = 1 + 2
119896minus1
sum
119899=1
cos 119899120591 = sin ((2119896 minus 1) 1205912)sin (1205912)
(A1)
Fejer kernel (eg see [27]) can be expressed in the fol-lowing equivalent forms
119865119896minus1
(120591) =1
119896
119896minus1
sum
119899=0
119863119899(120591) = 1 +
2
119896
119896minus1
sum
119899=1
(119896 minus 119899) cos 119899120591
=(1 minus cos 119896120591)119896 (1 minus cos 120591)
(A2)
Lagrangersquos trigonometric identity (eg see [28]) is as fol-lows
1198781(120591) =
119896minus1
sum
119899=1
sin 119899120591 = sin (1198961205912) sin ((119896 minus 1) 1205912)sin (1205912)
(A3)
In what follows we show that the following three trigono-metric identities also hold
2
119896minus1
sum
119899=1
(119896 minus 119899) sin 119899120591 = 119896 sin 120591 minus sin 1198961205911 minus cos 120591
(A4)
119896minus1
sum
119899=1
cos (119896 minus 2119899) 120591 = sin (119896 minus 1) 120591sin 120591
(A5)
119896minus1
sum
119899=1
119899 (119896 minus 119899) cos (119896 minus 2119899) 120591
=sin (119896120591) cos 120591 minus 119896 cos (119896120591) sin 120591
2sin3120591
(A6)
Denote 1198782(120591) = 2sum
119896minus1
119899=1(119896 minus 119899) sin 119899120591 119878
3(120591) = sum
119896minus1
119899=1cos(119896 minus
2119899)120591 and 1198784(120591) = sum
119896minus1
119899=1119899(119896 minus 119899) cos(119896 minus 2119899)120591
Notice that 1198782(120591) = 2119896119878
1(120591) + 119889119863
119896minus1(120591)119889120591 which
immediately leads to (A4)
28 Mathematical Problems in Engineering
Identity (A5) can be obtained as follows
sin (119896 minus 1) 120591sin 120591
=119890119895(119896minus1)120591
minus 119890minus119895(119896minus1)120591
119890119895120591 minus 119890minus119895120591
= 119890119895119896120591
119890minus2119895120591
minus 119890minus2119895(119896minus1)120591
1 minus 119890minus2119895120591
= 119890119895119896120591
119896minus1
sum
119899=1
119890minus2119895119899120591
=
119896minus1
sum
119899=1
119890119895(119896minus2119899)120591
=
119896minus1
sum
119899=1
cos (119896 minus 2119899) 120591
(A7)
From 4119899(119896 minus 119899) = 1198962
minus (119896 minus 2119899)2 it follows that 4119878
4(120591) =
1198962
1198783(120591) + 119889
2
1198783(120591)119889120591
2 which leads to (A6)
B The Chebyshev Polynomials
The Chebyshev polynomials of the first kind 119881119899(119909) can be
defined by the following relation (eg see [29])
119881119899(119909) = cos 119899120591 when 119909 = cos 120591 (B1)
The Chebyshev polynomials of the second kind 119880119899(119909) can be
defined by the following relation (eg see [29])
119880119899(119909) =
sin (119899 + 1) 120591sin 120591
when 119909 = cos 120591 (B2)
The Chebyshev polynomials satisfy the following recur-rence relations (eg see [29])
1198810(119909) = 1 119881
1(119909) = 119909
119881119899+1
(119909) = 2119909119881119899(119909) minus 119881
119899minus1(119909)
1198800(119909) = 1 119880
1(119909) = 2119909
119880119899+1
(119909) = 2119909119880119899(119909) minus 119880
119899minus1(119909)
(B3)
The first few Chebyshev polynomials of the first and secondkind are 119881
2(119909) = 2119909
2
minus 1 1198813(119909) = 4119909
3
minus 3119909 1198814(119909) = 8119909
4
minus
81199092
+ 1 1198802(119909) = 4119909
2
minus 1 1198803(119909) = 8119909
3
minus 4119909 and 1198804(119909) =
161199094
minus 121199092
+ 1
Conflict of Interests
The authors declare that there is no conflict of interestsregarding the publication of this paper
Acknowledgment
This work is supported by the SerbianMinistry of EducationScience and Technology Development as a part of ProjectTP32016
References
[1] V I Arnolrsquod V S Afrajmovich Y S Ilrsquoyashenko and L PShilrsquonikov Dynamical Systems V Bifurcation Theory and Catas-trophe Theory Springer Berlin Germany 1994
[2] E Polak ldquoOn the mathematical foundations of nondifferen-tiable optimization in engineering designrdquo SIAM Review vol29 no 1 pp 21ndash89 1987
[3] N S Fuzik ldquoBiharmonic modes of a tuned RF power amplifierrdquoRadiotehnika vol 25 no 7 pp 62ndash71 1970 (Russian)
[4] P Colantonio F Giannini G Leuzzi and E Limiti ldquoClass Gapproach for low-voltage high-efficiency PA designrdquo Interna-tional Journal of RF and Microwave Computer-Aided Engineer-ing vol 10 no 6 pp 366ndash378 2000
[5] F H Raab ldquoMaximum efficiency and output of class-F poweramplifiersrdquo IEEE Transactions on Microwave Theory and Tech-niques vol 49 no 6 pp 1162ndash1166 2001
[6] FH Raab ldquoClass-E class-C and class-F power amplifiers basedupon a finite number of harmonicsrdquo IEEE Transactions onMicrowaveTheory and Techniques vol 49 no 8 pp 1462ndash14682001
[7] J D Rhodes ldquoOutput universality inmaximum efficiency linearpower amplifiersrdquo International Journal of Circuit Theory andApplications vol 31 no 4 pp 385ndash405 2003
[8] A Juhas and L A Novak ldquoComments on lsquoClass-E class-Cand classF power amplifier based upon a finite number ofharmonicsrsquordquo IEEE Transactions on Microwave Theory and Tech-niques vol 57 no 6 pp 1623ndash1625 2009
[9] M Roberg and Z Popovic ldquoAnalysis of high-efficiency poweramplifiers with arbitrary output harmonic terminationsrdquo IEEETransactions on Microwave Theory and Techniques vol 59 no8 pp 2037ndash2048 2011
[10] A Grebennikov N O Sokal and M J Franco Switchmode RFPower Amplifiers ElsevierAcademic Press San Diego CalifUSA 2nd edition 2012
[11] T Canning P J Tasker and S C Cripps ldquoContinuous modepower amplifier design using harmonic clipping contourstheory and practicerdquo IEEE Transactions on Microwave Theoryand Techniques vol 62 no 1 pp 100ndash110 2014
[12] A Juhas and L A Novak ldquoGeneral description of nonnegativewaveforms up to second harmonic for power amplifier mod-ellingrdquoMathematical Problems in Engineering vol 2014 ArticleID 709762 18 pages 2014
[13] V I Arnolrsquod V V Goryunov O V Lyashko and V A VasilrsquoevDynamical Systems VIIImdashSingularity Theory II ApplicationsSpringer Berlin Germany 1993
[14] D Siersma ldquoProperties of conflict sets in the planerdquo BanachCenter Publications Polish Academy of Sciences vol 50 no 1 pp267ndash276 1999 Proceedings of the Banach Center Symposiumon Geometry and Topology of Caustics (Caustics rsquo98) WarsawPoland
[15] M van Manen The geometry of conflict sets [Dissertation]Universiteit Utrecht UtrechtThe Netherlands 2003 httpigi-tur-archivelibraryuunldissertations2003-0912-123058c4pdf
[16] Y L Sachkov ldquoMaxwell strata and symmetries in the problemofoptimal rolling of a sphere over a planerdquo Sbornik Mathematicsvol 201 no 7-8 pp 1029ndash1051 2010
[17] I A Bogaevsky ldquoPerestroikas of shocks and singularities ofminimum functionsrdquoPhysicaDNonlinear Phenomena vol 173no 1-2 pp 1ndash28 2002
[18] Y L Sachkov ldquoMaxwell strata in the Euler elastic problemrdquoJournal of Dynamical andControl Systems vol 14 no 2 pp 169ndash234 2008
[19] M Siino and T Koike ldquoTopological classification of black holesgeneric Maxwell set and crease set of a horizonrdquo InternationalJournal ofModern Physics D Gravitation Astrophysics Cosmol-ogy vol 20 no 6 pp 1095ndash1122 2011
Mathematical Problems in Engineering 29
[20] F H Raab ldquoClass-F power amplifiers with maximally flat wave-formsrdquo IEEETransactions onMicrowaveTheory and Techniquesvol 45 no 11 pp 2007ndash2012 1997
[21] A Juhas and L A Novak ldquoMaximally flat waveforms with finitenumber of harmonics in class-F power amplifiersrdquo Mathemat-ical Problems in Engineering vol 2013 Article ID 169590 9pages 2013
[22] L Fejer ldquoUber trigonometrische polynomerdquo Journal fur dieReine und Angewandte Mathematik vol 1916 no 146 pp 53ndash82 1916 (German)
[23] S C Cripps ldquoBessel Waives [microwave bytes]rdquo IEEE Micro-wave Magazine vol 10 no 7 pp 30ndash36 117 2009
[24] L N Bryzgalova ldquoSingularities of the maximum of para-metrically dependent functionrdquo Functional Analysis and ItsApplications vol 11 no 1 pp 49ndash51 1977
[25] V I Arnold A A Davydov V A Vassiliev and V MZakalyukin Mathematical Models of Catastrophes Control ofCatastrophic Process Encyclopedia of Life Support Systems(EOLSS) EOLSS Publishers Oxford UK 2006
[26] S C Cripps RF Power Amplifiers for Wireless CommunicationsArtech House Norwood Mass USA 2nd edition 2006
[27] A Zygmund Trigonometric Series vol 1 Cambridge UniversityPress Cambridge UK 2nd edition 1959
[28] A Jeffrey and H DaiHandbook of Mathematical Formulas andIntegrals ElsevierAcademic Press San Diego Calif USA 4thedition 2008
[29] J C Mason and D C Handscomb Chebyshev PolynomialsChapman amp Hall CRC Press Boca Raton Fla USA 2003
Submit your manuscripts athttpwwwhindawicom
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical Problems in Engineering
Hindawi Publishing Corporationhttpwwwhindawicom
Differential EquationsInternational Journal of
Volume 2014
Applied MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Probability and StatisticsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical PhysicsAdvances in
Complex AnalysisJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
OptimizationJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
CombinatoricsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Operations ResearchAdvances in
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Function Spaces
Abstract and Applied AnalysisHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of Mathematics and Mathematical Sciences
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Algebra
Discrete Dynamics in Nature and Society
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Decision SciencesAdvances in
Discrete MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom
Volume 2014 Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Stochastic AnalysisInternational Journal of
Mathematical Problems in Engineering 23
+j2
+j5
+j1
+j2
+j5
minusj2
minusj5
minusj1
minusj2
minusj5
infin0
078
02
05
1 2 5
075 07 065
06 05
075
07
065
06
05
120578 lt 05
(a)
099
095
091
083 075
067
0 59 051
051
099
095
091
083
075 0
67 059
+j5
+j1
+j2
+j5
minusj5
minusj1
minusj2
minusj5
infin
+j2
minusj2
0 02
05
1 2 5
120578 lt 05
(b)
Figure 16 (a) Contours of maximal efficiency of PA and (b) contours of corresponding PUF associated with the waveform pair (185) and(187) as functions of normalized second harmonic impedance 119911
2119899= 1199112Re119911
1
2
3
4
1
0
2 3 410
Wav
efor
ms
Angle 120579120587
i(120579)
(120579)
Figure 17 Waveform pair (185) and (187) that provides maximalefficiency for 119911
2= 01 minus 11989505
with voltage waveform of type (187) with coefficients 1198862V =
minus00667 1198872V = minus03333 119886
1V = minus09333 and 1198871V = 06667 (see
Figure 17) Corresponding efficiency PUF and normalizedsecond harmonic impedance are 120578 = 07330 PUF = 07572and 1199112119899= 01683 minus 11989508415 respectively
On the other hand for second harmonic impedance 1199112=
01 minus 11989508 and current waveform (185) from Algorithm 32 itfollows that 2120582
2V gt 1 minus 1198862V Then according to step (iii) of
above algorithm maximal efficiency is attained with voltagewaveform of type (187) with coefficients 119886
2V = minus00667 1198872V =
minus05333 1198861V = minus09333 and 119887
1V = 10572 (see Figure 18)Efficiency PUF andnormalized secondharmonic impedanceare 120578 = 07330 PUF = 06332 and 119911
2119899= 01683 minus 11989513465
respectively
2
3
4
1
0
2 3 410
Wav
efor
ms
Angle 120579120587
i(120579)
(120579)
Figure 18 Waveform pair (185) and (187) that provides maximalefficiency for 119911
2= 01 minus 11989508
Case Study 72 As another case study let us consider the effi-ciency of PA providing that current waveform is nonnegativecosine waveform up to third harmonic with maximum valueof amplitude of fundamental harmonic [22] (see also [8])
119894 (120579) = 1 +1 + radic5
2cos 120579 + 2radic5
5cos 2120579 + 5 minus radic5
10cos 3120579
(188)
and voltage waveform is nonnegative waveform of type(187) Load impedances at fundamental second and thirdharmonic are 119911
1= minus(119886
1Vminus1198951198871V)1198861119894 1199112 = minus(1198862Vminus1198951198872V)1198862119894 and
1199113= 0 respectively According to our assumption the load
is passive and therefore Re1199111 gt 0 and Re119911
2 ge 0 which
further imply 11988611198941198861V lt 0 and 119886
21198941198862V le 0 respectively
Because current waveform (188) contains only cosineterms and voltage waveform is the same as in previous case
24 Mathematical Problems in Engineering
08
075
07
06
06
065
065
05
05
075 0
7
120578 lt 05
+j2
+j5
+j1
+j2
+j5
minusj2
minusj5
minusj1
minusj2
minusj5
infin0
02
05
1 2 5
(a)
085
08 0
75 07 0
6
05 04
08
075
0405
060
7
120578 lt 05
+j2
+j5
+j1
+j2
+j5
minusj2
minusj5
minusj1
minusj2
minusj5
infin0 02
05
1 2 5
(b)
Figure 19 (a) Contours of maximal efficiency of PA and (b) contours of corresponding PUF associated with the waveform pair (187)-(188)as functions of normalized second harmonic impedance 119911
2119899= 1199112Re119911
1
study the procedure for calculation of maximal efficiency ofPA with waveform pair (187)-(188) is the same as presentedin Algorithm 32 In this case study the coefficients of funda-mental and second harmonic of current waveform are 119886
1119894=
(1 + radic5)2 and 1198862119894= 2radic55 respectively
Maximal efficiency of PA associated with the waveformpair (187)-(188) as a function of normalized secondharmonicimpedance 119911
2119899= 1199112Re119911
1 is presented in Figure 19(a)
Efficiency of 08 is achieved at the edge of Smith chart wheresecond harmonic impedance has small resistive part Thetheoretical upper bound 120578 = (1 + radic5)4 asymp 08090 isattained when second harmonic is short-circuitedWhen thisupper bound is reached both second and third harmonic areshort-circuited which implies that we are dealing with finiteharmonic class-C [6 8] or dually when current and voltageinterchange their roles with finite harmonic inverse class-C[6 9] Corresponding PUF calculated according to (179) ispresented in Figure 19(b) Peak value of PUF asymp 08541 isattained when second harmonic is short-circuited
For example for second harmonic impedance 1199112
=
007 minus 11989504 and current waveform (188) from Algorithm 32 itfollows that 2120582
2V le 1 minus 1198862V Furthermore according to step
(iii) of Algorithm 32 maximal efficiency of PA is attainedwith voltage waveform of type (187) with coefficients 119886
2V =
minus00626 1198872V = minus03578 119886
1V = minus09374 and 1198871V = 07155 (see
Figure 20) Corresponding efficiency PUF and normalizedsecond harmonic impedance are 120578 = 07584 PUF = 06337and 1199112119899= 01208 minus 11989506904 respectively
On the other hand for 1199112= 005 minus 11989507 and current
waveform (187) it follows that 21205822V gt 1minus119886
2VThen accordingto step (iii) of Algorithm 32 themaximal efficiency is attainedwith voltage waveform of type (187) with coefficients 119886
2V =
minus00447 1198872V = minus06261 119886
1V = minus09318 and 1198871V = 10007 (see
Figure 21) Efficiency PUF and normalized second harmonic
2
3
4
1
0
2 3 410
Wav
efor
ms
Angle 120579120587
i(120579)
(120579)
Figure 20 Waveform pair (187)-(188) that provides maximal effi-ciency for 119911
2= 007 minus 11989504
impedance are 120578 = 07538 PUF = 05314 and 1199112119899= 00868minus
11989512156 respectively
72 Nonnegative Waveforms for 119896 = 3 in PA EfficiencyAnalysis In this subsection we provide another two casestudies of usage of description of nonnegative waveforms inPA efficiency analysis this time with fundamental and thirdharmonic (119896 = 3)
Case Study 73 Let us consider current-voltage pair such thatvoltage is nonnegative waveform with fundamental and thirdharmonic
V (120579) = 1 + 1198861V cos 120579 + 1198871V sin 120579 + 1198863V cos 3120579 + 1198873V sin 3120579
(189)
Mathematical Problems in Engineering 25
2
3
4
1
0
2 3 410
Wav
efor
ms
Angle 120579120587
i(120579)
(120579)
Figure 21 Waveform pair (187)-(188) that provides maximal effi-ciency for 119911
2= 005 minus 11989507
and current is nonnegative cosine waveform given by (188)Load impedances at fundamental second and third har-monic are 119911
1= minus(119886
1V minus 1198951198871V)1198861119894 1199112 = 0 and 119911
3= minus(119886
3V minus
1198951198873V)1198863119894 respectively According to our assumption the load
is passive and therefore Re1199111 gt 0 and Re119911
3 ge 0 which
further imply 11988611198941198861V lt 0 and 119886
31198941198863V le 0
In this subsection we consider the problem of findingmaximal efficiency of PA with waveform pair (188)-(189)for given third harmonic impedance As we mentionedearlier problem of finding maximal efficiency of PA withcurrent-voltage pair (188)-(189) for prescribed third har-monic impedance can be reduced to the problem of findingvoltage waveform of type (189) withmaximal coefficient |119886
1V|for prescribed coefficients of third harmonic (see Section 52)
The following algorithm provides the procedure forcalculation of maximal efficiency with current-voltage pair(188)-(189) The definition of function atan 2(119910 119909) whichappears in step (iii) of the following algorithm is given by(105)
Algorithm 33 (i) Choose 1199113= 1199033+1198951199093such that |119911
3| le 1|119886
3119894|
(ii) calculate 1198863V minus 1198951198873V = minus119911
31198863119894and 120582
3V = radic11988623V + 1198872
3V(iii) if 271205822
3V le (1 minus 21198863V)3 then calculate 119886
1V = minus1 minus 1198863V
and 1198871V = minus3119887
3V else calculate 1205821V = 3(3radic1205823V minus 120582
3V) 1205790V minus120585V3 = (13)atan 2(119887
3V 1198863V) 1198861V = minus1205821V cos(1205790V minus 120585V3) and
1198871V = minus120582
1V sin(1205790V minus 120585V3)(iv) calculate efficiency 120578 = minus119886
11198941198861V2
(v) calculate 1199111= minus(119886
1V minus 1198951198871V)1198861119894 and 1199113119899 = 1199113Re119911
1
In this case study coefficients of fundamental and thirdharmonic of current waveform are 119886
1119894= (1 + radic5)2 and
1198863119894= (5 minus radic5)10 respectively For the waveform pair (188)-
(189) maximal efficiency of PA as a function of normalizedthird harmonic impedance 119911
3119899= 1199113Re119911
1 is presented in
Figure 22 Efficiency of 08 is reached when third harmonicimpedance has small resistive part Peak efficiency 120578 = (1 +
radic5)4 asymp 08090 is achieved when third harmonic is short-circuited
+j2
+j5
+j1
+j2
+j5
minusj2
minusj5
minusj1
minusj2
minusj5
infin0
08
08
075
075
07
07
06
05
02
05
1 2
Figure 22 Contours of maximal efficiency of PA associated withthe waveform pair (188)-(189) as a function of normalized thirdharmonic impedance 119911
3119899= 1199113Re119911
1
For the present case study in what follows we show thatpower utilization factor is proportional to efficiency For volt-age waveform of type (189) it is easy to see that V(120579 + 120587) =
2minusV(120579) holdsThis relation along with the fact that waveformV(120579) that provides maximal efficiency has at least one zeroimplies that max[V(120579)] = 2 On the other hand currentwaveform (188) is cosine waveform with positive coefficientsand therefore max[119894(120579)] = 119894(0) = 2 + 4radic5 Consequentlyaccording to (179) the following relation holds
PUFCase study 73 = 2 (5 minus 2radic5) 120578 = 10557120578 (190)
Clearly the ratio PUF120578 is constant and therefore in this casestudy PUF can be easily calculated from the correspondingefficiency Accordingly peak efficiency and peak value ofPUFCase study 73 = 3radic52 minus 52 = 08541 are attained forthe same voltage waveform (when third harmonic is short-circuited)
In the first example current waveform (188) and 1199113=
02 minus 11989505 imply that 2712058223V le (1 minus 2119886
3V)3 Then according
to Algorithm 33 the voltage waveform of type (189) thatprovides maximal efficiency has the following coefficients1198863V = minus00553 119887
3V = minus01382 1198861V = minus09447 and 119887
1V =
04146 (see Figure 23) Efficiency PUF and normalized thirdharmonic impedance are 120578 = 07643 PUF = 08069 and1199113119899= 03425 minus 11989508564 respectivelyIn the second example current waveform (188) and 119911
3=
01 minus 11989511 imply that 2712058223V gt (1 minus 2119886
3V)3 Then according
to Algorithm 33 the voltage waveform of type (189) thatprovides maximal efficiency has the following coefficients1198863V = minus00276 119887
3V = minus03040 1198861V = minus09391 and 119887
1V =
05807 (see Figure 24) Efficiency PUF and normalized thirdharmonic impedance are 120578 = 07598 PUF = 08021 and1199113119899= 01723 minus 11989518952 respectively
26 Mathematical Problems in Engineering
2
3
4
1
0
2 3 410
Wav
efor
ms
Angle 120579120587
i(120579)
(120579)
Figure 23 Waveform pair (188)-(189) that provides maximal effi-ciency for 119911
3= 02 minus 11989505
2
3
4
1
0
2 3 410
Wav
efor
ms
Angle 120579120587
i(120579)
(120579)
Figure 24 Waveform pair (188)-(189) that provides maximal effi-ciency for 119911
3= 01 minus 11989511
Case Study 74 In this case study let us consider current-voltage pair where current is normalized waveform of type(180) with conduction angle 120572 = 115120587 (207∘) and voltageis nonnegative waveform of type (189) Substitution of 120572 =
115120587 and 119868dc = 1 into (182) leads to 119868119863
= 22535Furthermore substitution of 120572 = 115120587 and 119868
119863= 22535 into
(180) leads to
119894 (120579)
=
22535 [cos 120579 minus cos(1151205872
)] |120579| le115120587
2
0115120587
2le |120579| le 120587
(191)
Similarly substitution of 120572 = 115120587 and 119868119863
= 22535 into(183) and (184) for 119899 = 3 yields coefficients of fundamentaland third harmonic of waveform (191)
1198861119894= 14586 119886
3119894= minus01026 (192)
Because current waveform (191) contains only cosineterms and voltage waveform is the same as in previous case
+j2
+j5
+j1
+j2
+j5
minusj2
minusj5
minusj1
minusj2
minusj5
infin0 02
05
1 2
074
076
08
084
082
078
076
074
Figure 25 Contours of maximal efficiency of PA associated withthe waveform pair (189) and (191) as a function of normalized thirdharmonic impedance 119911
3119899= 1199113Re119911
1
study the procedure for calculation of maximal efficiencyof PA with waveform pair (189)ndash(191) is the same as thatpresented in Algorithm 33 In this case study the coefficientsof fundamental and third harmonic of current waveform aregiven by (192)
For the waveform pair (189) and (191) maximal efficiencyof PA as a function of normalized third harmonic impedance1199113119899= 1199113Re119911
1 is presented in Figure 25 Efficiency of 084
is obtained in vicinity of 1199113119899= 23685 (corresponding to 119911
3=
18750) Peak efficiency 120578 asymp 08421 is achieved for voltagewaveform of type (189) with coefficients 119886
1V = minus2radic3 1198863V =
radic39 and 1198871V = 1198873V = 0
In the course of finding power utilization factor noticethat currentwaveformof type (191) attains itsmaximumvaluefor 120579 = 0 Insertion of max[119894(120579)] = 119894(0) = 278 andmax[V(120579)] = 2 for voltage waveform of type (189) into (179)leads to
PUFCase study 74 = 1439120578 (193)
Again the ratio PUF120578 is constant and PUF can be easilycalculated from the corresponding efficiency Accordinglypeak value of PUFCase study 74 asymp 12118 and peak efficiencyare attained for the same voltage waveform
In the first example current waveform (191) and 1199113=
1 minus 11989502 imply that 2712058223V le (1 minus 2119886
3V)3 Then according to
Algorithm 33 voltage waveform of type (189) which providesmaximal efficiency has coefficients 119886
3V = 01026 1198873V =
00205 1198861V = minus11026 and 119887
1V = minus00616 (see Figure 26)Efficiency PUF and normalized third harmonic impedanceare 120578 = 08042 PUF = 11572 and 119911
3119899= 13228 minus 11989502646
respectivelyIn second example current waveform (191) and 119911
3=
15 minus 11989512 imply that 2712058223V gt (1 minus 2119886
3V)3 Then according to
Algorithm 33 voltage waveform of type (189) which providesmaximal efficiency has coefficients 119886
3V = 01540 1198873V =
Mathematical Problems in Engineering 27
2
3
4
1
0
2 3 410
Wav
efor
ms
Angle 120579120587
i(120579)
(120579)
Figure 26 Waveform pair (189) and (191) that provides maximalefficiency for 119911
3= 1 minus 11989502
2
3
4
1
0
2 3 410
Wav
efor
ms
Angle 120579120587
i(120579)
(120579)
Figure 27 Waveform pair (189) and (191) that provides maximalefficiency for 119911
3= 15 minus 11989512
01232 1198861V = minus11255 and 119887
1V = minus02575 (see Figure 27)Efficiency PUF and normalized third harmonic impedanceare 120578 = 08208 PUF = 11812 and 119911
3119899= 19439 minus 11989515552
respectively
8 Conclusion
In this paper we consider a problem of finding general de-scriptions of various classes of nonnegative waveforms withfundamental and 119896th harmonic These classes include non-negative waveforms with at least one zero nonnegative wave-forms with maximal amplitude of fundamental harmonic forprescribed amplitude of 119896th harmonic nonnegative wave-forms withmaximal coefficient of cosine part of fundamentalharmonic for prescribed coefficients of 119896th harmonic andnonnegative cosine waveforms with at least one zero Mainresults are stated in six propositions (Propositions 1 6 918 22 and 26) four corollaries (Corollaries 2ndash5) twentyremarks and three algorithms Four case studies of usageof closed form descriptions of nonnegative waveforms in PAefficiency analysis are considered in detail in Section 7
Appendices
Here we provide a list of finite sums of trigonometric func-tions used in this paper (Appendix A) and brief account ofthe Chebyshev polynomials (Appendix B)
A List of Some Finite Sums ofTrigonometric Functions
Dirichlet kernel (eg see [27]) is as follows
119863119896minus1
(120591) = 1 + 2
119896minus1
sum
119899=1
cos 119899120591 = sin ((2119896 minus 1) 1205912)sin (1205912)
(A1)
Fejer kernel (eg see [27]) can be expressed in the fol-lowing equivalent forms
119865119896minus1
(120591) =1
119896
119896minus1
sum
119899=0
119863119899(120591) = 1 +
2
119896
119896minus1
sum
119899=1
(119896 minus 119899) cos 119899120591
=(1 minus cos 119896120591)119896 (1 minus cos 120591)
(A2)
Lagrangersquos trigonometric identity (eg see [28]) is as fol-lows
1198781(120591) =
119896minus1
sum
119899=1
sin 119899120591 = sin (1198961205912) sin ((119896 minus 1) 1205912)sin (1205912)
(A3)
In what follows we show that the following three trigono-metric identities also hold
2
119896minus1
sum
119899=1
(119896 minus 119899) sin 119899120591 = 119896 sin 120591 minus sin 1198961205911 minus cos 120591
(A4)
119896minus1
sum
119899=1
cos (119896 minus 2119899) 120591 = sin (119896 minus 1) 120591sin 120591
(A5)
119896minus1
sum
119899=1
119899 (119896 minus 119899) cos (119896 minus 2119899) 120591
=sin (119896120591) cos 120591 minus 119896 cos (119896120591) sin 120591
2sin3120591
(A6)
Denote 1198782(120591) = 2sum
119896minus1
119899=1(119896 minus 119899) sin 119899120591 119878
3(120591) = sum
119896minus1
119899=1cos(119896 minus
2119899)120591 and 1198784(120591) = sum
119896minus1
119899=1119899(119896 minus 119899) cos(119896 minus 2119899)120591
Notice that 1198782(120591) = 2119896119878
1(120591) + 119889119863
119896minus1(120591)119889120591 which
immediately leads to (A4)
28 Mathematical Problems in Engineering
Identity (A5) can be obtained as follows
sin (119896 minus 1) 120591sin 120591
=119890119895(119896minus1)120591
minus 119890minus119895(119896minus1)120591
119890119895120591 minus 119890minus119895120591
= 119890119895119896120591
119890minus2119895120591
minus 119890minus2119895(119896minus1)120591
1 minus 119890minus2119895120591
= 119890119895119896120591
119896minus1
sum
119899=1
119890minus2119895119899120591
=
119896minus1
sum
119899=1
119890119895(119896minus2119899)120591
=
119896minus1
sum
119899=1
cos (119896 minus 2119899) 120591
(A7)
From 4119899(119896 minus 119899) = 1198962
minus (119896 minus 2119899)2 it follows that 4119878
4(120591) =
1198962
1198783(120591) + 119889
2
1198783(120591)119889120591
2 which leads to (A6)
B The Chebyshev Polynomials
The Chebyshev polynomials of the first kind 119881119899(119909) can be
defined by the following relation (eg see [29])
119881119899(119909) = cos 119899120591 when 119909 = cos 120591 (B1)
The Chebyshev polynomials of the second kind 119880119899(119909) can be
defined by the following relation (eg see [29])
119880119899(119909) =
sin (119899 + 1) 120591sin 120591
when 119909 = cos 120591 (B2)
The Chebyshev polynomials satisfy the following recur-rence relations (eg see [29])
1198810(119909) = 1 119881
1(119909) = 119909
119881119899+1
(119909) = 2119909119881119899(119909) minus 119881
119899minus1(119909)
1198800(119909) = 1 119880
1(119909) = 2119909
119880119899+1
(119909) = 2119909119880119899(119909) minus 119880
119899minus1(119909)
(B3)
The first few Chebyshev polynomials of the first and secondkind are 119881
2(119909) = 2119909
2
minus 1 1198813(119909) = 4119909
3
minus 3119909 1198814(119909) = 8119909
4
minus
81199092
+ 1 1198802(119909) = 4119909
2
minus 1 1198803(119909) = 8119909
3
minus 4119909 and 1198804(119909) =
161199094
minus 121199092
+ 1
Conflict of Interests
The authors declare that there is no conflict of interestsregarding the publication of this paper
Acknowledgment
This work is supported by the SerbianMinistry of EducationScience and Technology Development as a part of ProjectTP32016
References
[1] V I Arnolrsquod V S Afrajmovich Y S Ilrsquoyashenko and L PShilrsquonikov Dynamical Systems V Bifurcation Theory and Catas-trophe Theory Springer Berlin Germany 1994
[2] E Polak ldquoOn the mathematical foundations of nondifferen-tiable optimization in engineering designrdquo SIAM Review vol29 no 1 pp 21ndash89 1987
[3] N S Fuzik ldquoBiharmonic modes of a tuned RF power amplifierrdquoRadiotehnika vol 25 no 7 pp 62ndash71 1970 (Russian)
[4] P Colantonio F Giannini G Leuzzi and E Limiti ldquoClass Gapproach for low-voltage high-efficiency PA designrdquo Interna-tional Journal of RF and Microwave Computer-Aided Engineer-ing vol 10 no 6 pp 366ndash378 2000
[5] F H Raab ldquoMaximum efficiency and output of class-F poweramplifiersrdquo IEEE Transactions on Microwave Theory and Tech-niques vol 49 no 6 pp 1162ndash1166 2001
[6] FH Raab ldquoClass-E class-C and class-F power amplifiers basedupon a finite number of harmonicsrdquo IEEE Transactions onMicrowaveTheory and Techniques vol 49 no 8 pp 1462ndash14682001
[7] J D Rhodes ldquoOutput universality inmaximum efficiency linearpower amplifiersrdquo International Journal of Circuit Theory andApplications vol 31 no 4 pp 385ndash405 2003
[8] A Juhas and L A Novak ldquoComments on lsquoClass-E class-Cand classF power amplifier based upon a finite number ofharmonicsrsquordquo IEEE Transactions on Microwave Theory and Tech-niques vol 57 no 6 pp 1623ndash1625 2009
[9] M Roberg and Z Popovic ldquoAnalysis of high-efficiency poweramplifiers with arbitrary output harmonic terminationsrdquo IEEETransactions on Microwave Theory and Techniques vol 59 no8 pp 2037ndash2048 2011
[10] A Grebennikov N O Sokal and M J Franco Switchmode RFPower Amplifiers ElsevierAcademic Press San Diego CalifUSA 2nd edition 2012
[11] T Canning P J Tasker and S C Cripps ldquoContinuous modepower amplifier design using harmonic clipping contourstheory and practicerdquo IEEE Transactions on Microwave Theoryand Techniques vol 62 no 1 pp 100ndash110 2014
[12] A Juhas and L A Novak ldquoGeneral description of nonnegativewaveforms up to second harmonic for power amplifier mod-ellingrdquoMathematical Problems in Engineering vol 2014 ArticleID 709762 18 pages 2014
[13] V I Arnolrsquod V V Goryunov O V Lyashko and V A VasilrsquoevDynamical Systems VIIImdashSingularity Theory II ApplicationsSpringer Berlin Germany 1993
[14] D Siersma ldquoProperties of conflict sets in the planerdquo BanachCenter Publications Polish Academy of Sciences vol 50 no 1 pp267ndash276 1999 Proceedings of the Banach Center Symposiumon Geometry and Topology of Caustics (Caustics rsquo98) WarsawPoland
[15] M van Manen The geometry of conflict sets [Dissertation]Universiteit Utrecht UtrechtThe Netherlands 2003 httpigi-tur-archivelibraryuunldissertations2003-0912-123058c4pdf
[16] Y L Sachkov ldquoMaxwell strata and symmetries in the problemofoptimal rolling of a sphere over a planerdquo Sbornik Mathematicsvol 201 no 7-8 pp 1029ndash1051 2010
[17] I A Bogaevsky ldquoPerestroikas of shocks and singularities ofminimum functionsrdquoPhysicaDNonlinear Phenomena vol 173no 1-2 pp 1ndash28 2002
[18] Y L Sachkov ldquoMaxwell strata in the Euler elastic problemrdquoJournal of Dynamical andControl Systems vol 14 no 2 pp 169ndash234 2008
[19] M Siino and T Koike ldquoTopological classification of black holesgeneric Maxwell set and crease set of a horizonrdquo InternationalJournal ofModern Physics D Gravitation Astrophysics Cosmol-ogy vol 20 no 6 pp 1095ndash1122 2011
Mathematical Problems in Engineering 29
[20] F H Raab ldquoClass-F power amplifiers with maximally flat wave-formsrdquo IEEETransactions onMicrowaveTheory and Techniquesvol 45 no 11 pp 2007ndash2012 1997
[21] A Juhas and L A Novak ldquoMaximally flat waveforms with finitenumber of harmonics in class-F power amplifiersrdquo Mathemat-ical Problems in Engineering vol 2013 Article ID 169590 9pages 2013
[22] L Fejer ldquoUber trigonometrische polynomerdquo Journal fur dieReine und Angewandte Mathematik vol 1916 no 146 pp 53ndash82 1916 (German)
[23] S C Cripps ldquoBessel Waives [microwave bytes]rdquo IEEE Micro-wave Magazine vol 10 no 7 pp 30ndash36 117 2009
[24] L N Bryzgalova ldquoSingularities of the maximum of para-metrically dependent functionrdquo Functional Analysis and ItsApplications vol 11 no 1 pp 49ndash51 1977
[25] V I Arnold A A Davydov V A Vassiliev and V MZakalyukin Mathematical Models of Catastrophes Control ofCatastrophic Process Encyclopedia of Life Support Systems(EOLSS) EOLSS Publishers Oxford UK 2006
[26] S C Cripps RF Power Amplifiers for Wireless CommunicationsArtech House Norwood Mass USA 2nd edition 2006
[27] A Zygmund Trigonometric Series vol 1 Cambridge UniversityPress Cambridge UK 2nd edition 1959
[28] A Jeffrey and H DaiHandbook of Mathematical Formulas andIntegrals ElsevierAcademic Press San Diego Calif USA 4thedition 2008
[29] J C Mason and D C Handscomb Chebyshev PolynomialsChapman amp Hall CRC Press Boca Raton Fla USA 2003
Submit your manuscripts athttpwwwhindawicom
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical Problems in Engineering
Hindawi Publishing Corporationhttpwwwhindawicom
Differential EquationsInternational Journal of
Volume 2014
Applied MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Probability and StatisticsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical PhysicsAdvances in
Complex AnalysisJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
OptimizationJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
CombinatoricsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Operations ResearchAdvances in
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Function Spaces
Abstract and Applied AnalysisHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of Mathematics and Mathematical Sciences
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Algebra
Discrete Dynamics in Nature and Society
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Decision SciencesAdvances in
Discrete MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom
Volume 2014 Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Stochastic AnalysisInternational Journal of
24 Mathematical Problems in Engineering
08
075
07
06
06
065
065
05
05
075 0
7
120578 lt 05
+j2
+j5
+j1
+j2
+j5
minusj2
minusj5
minusj1
minusj2
minusj5
infin0
02
05
1 2 5
(a)
085
08 0
75 07 0
6
05 04
08
075
0405
060
7
120578 lt 05
+j2
+j5
+j1
+j2
+j5
minusj2
minusj5
minusj1
minusj2
minusj5
infin0 02
05
1 2 5
(b)
Figure 19 (a) Contours of maximal efficiency of PA and (b) contours of corresponding PUF associated with the waveform pair (187)-(188)as functions of normalized second harmonic impedance 119911
2119899= 1199112Re119911
1
study the procedure for calculation of maximal efficiency ofPA with waveform pair (187)-(188) is the same as presentedin Algorithm 32 In this case study the coefficients of funda-mental and second harmonic of current waveform are 119886
1119894=
(1 + radic5)2 and 1198862119894= 2radic55 respectively
Maximal efficiency of PA associated with the waveformpair (187)-(188) as a function of normalized secondharmonicimpedance 119911
2119899= 1199112Re119911
1 is presented in Figure 19(a)
Efficiency of 08 is achieved at the edge of Smith chart wheresecond harmonic impedance has small resistive part Thetheoretical upper bound 120578 = (1 + radic5)4 asymp 08090 isattained when second harmonic is short-circuitedWhen thisupper bound is reached both second and third harmonic areshort-circuited which implies that we are dealing with finiteharmonic class-C [6 8] or dually when current and voltageinterchange their roles with finite harmonic inverse class-C[6 9] Corresponding PUF calculated according to (179) ispresented in Figure 19(b) Peak value of PUF asymp 08541 isattained when second harmonic is short-circuited
For example for second harmonic impedance 1199112
=
007 minus 11989504 and current waveform (188) from Algorithm 32 itfollows that 2120582
2V le 1 minus 1198862V Furthermore according to step
(iii) of Algorithm 32 maximal efficiency of PA is attainedwith voltage waveform of type (187) with coefficients 119886
2V =
minus00626 1198872V = minus03578 119886
1V = minus09374 and 1198871V = 07155 (see
Figure 20) Corresponding efficiency PUF and normalizedsecond harmonic impedance are 120578 = 07584 PUF = 06337and 1199112119899= 01208 minus 11989506904 respectively
On the other hand for 1199112= 005 minus 11989507 and current
waveform (187) it follows that 21205822V gt 1minus119886
2VThen accordingto step (iii) of Algorithm 32 themaximal efficiency is attainedwith voltage waveform of type (187) with coefficients 119886
2V =
minus00447 1198872V = minus06261 119886
1V = minus09318 and 1198871V = 10007 (see
Figure 21) Efficiency PUF and normalized second harmonic
2
3
4
1
0
2 3 410
Wav
efor
ms
Angle 120579120587
i(120579)
(120579)
Figure 20 Waveform pair (187)-(188) that provides maximal effi-ciency for 119911
2= 007 minus 11989504
impedance are 120578 = 07538 PUF = 05314 and 1199112119899= 00868minus
11989512156 respectively
72 Nonnegative Waveforms for 119896 = 3 in PA EfficiencyAnalysis In this subsection we provide another two casestudies of usage of description of nonnegative waveforms inPA efficiency analysis this time with fundamental and thirdharmonic (119896 = 3)
Case Study 73 Let us consider current-voltage pair such thatvoltage is nonnegative waveform with fundamental and thirdharmonic
V (120579) = 1 + 1198861V cos 120579 + 1198871V sin 120579 + 1198863V cos 3120579 + 1198873V sin 3120579
(189)
Mathematical Problems in Engineering 25
2
3
4
1
0
2 3 410
Wav
efor
ms
Angle 120579120587
i(120579)
(120579)
Figure 21 Waveform pair (187)-(188) that provides maximal effi-ciency for 119911
2= 005 minus 11989507
and current is nonnegative cosine waveform given by (188)Load impedances at fundamental second and third har-monic are 119911
1= minus(119886
1V minus 1198951198871V)1198861119894 1199112 = 0 and 119911
3= minus(119886
3V minus
1198951198873V)1198863119894 respectively According to our assumption the load
is passive and therefore Re1199111 gt 0 and Re119911
3 ge 0 which
further imply 11988611198941198861V lt 0 and 119886
31198941198863V le 0
In this subsection we consider the problem of findingmaximal efficiency of PA with waveform pair (188)-(189)for given third harmonic impedance As we mentionedearlier problem of finding maximal efficiency of PA withcurrent-voltage pair (188)-(189) for prescribed third har-monic impedance can be reduced to the problem of findingvoltage waveform of type (189) withmaximal coefficient |119886
1V|for prescribed coefficients of third harmonic (see Section 52)
The following algorithm provides the procedure forcalculation of maximal efficiency with current-voltage pair(188)-(189) The definition of function atan 2(119910 119909) whichappears in step (iii) of the following algorithm is given by(105)
Algorithm 33 (i) Choose 1199113= 1199033+1198951199093such that |119911
3| le 1|119886
3119894|
(ii) calculate 1198863V minus 1198951198873V = minus119911
31198863119894and 120582
3V = radic11988623V + 1198872
3V(iii) if 271205822
3V le (1 minus 21198863V)3 then calculate 119886
1V = minus1 minus 1198863V
and 1198871V = minus3119887
3V else calculate 1205821V = 3(3radic1205823V minus 120582
3V) 1205790V minus120585V3 = (13)atan 2(119887
3V 1198863V) 1198861V = minus1205821V cos(1205790V minus 120585V3) and
1198871V = minus120582
1V sin(1205790V minus 120585V3)(iv) calculate efficiency 120578 = minus119886
11198941198861V2
(v) calculate 1199111= minus(119886
1V minus 1198951198871V)1198861119894 and 1199113119899 = 1199113Re119911
1
In this case study coefficients of fundamental and thirdharmonic of current waveform are 119886
1119894= (1 + radic5)2 and
1198863119894= (5 minus radic5)10 respectively For the waveform pair (188)-
(189) maximal efficiency of PA as a function of normalizedthird harmonic impedance 119911
3119899= 1199113Re119911
1 is presented in
Figure 22 Efficiency of 08 is reached when third harmonicimpedance has small resistive part Peak efficiency 120578 = (1 +
radic5)4 asymp 08090 is achieved when third harmonic is short-circuited
+j2
+j5
+j1
+j2
+j5
minusj2
minusj5
minusj1
minusj2
minusj5
infin0
08
08
075
075
07
07
06
05
02
05
1 2
Figure 22 Contours of maximal efficiency of PA associated withthe waveform pair (188)-(189) as a function of normalized thirdharmonic impedance 119911
3119899= 1199113Re119911
1
For the present case study in what follows we show thatpower utilization factor is proportional to efficiency For volt-age waveform of type (189) it is easy to see that V(120579 + 120587) =
2minusV(120579) holdsThis relation along with the fact that waveformV(120579) that provides maximal efficiency has at least one zeroimplies that max[V(120579)] = 2 On the other hand currentwaveform (188) is cosine waveform with positive coefficientsand therefore max[119894(120579)] = 119894(0) = 2 + 4radic5 Consequentlyaccording to (179) the following relation holds
PUFCase study 73 = 2 (5 minus 2radic5) 120578 = 10557120578 (190)
Clearly the ratio PUF120578 is constant and therefore in this casestudy PUF can be easily calculated from the correspondingefficiency Accordingly peak efficiency and peak value ofPUFCase study 73 = 3radic52 minus 52 = 08541 are attained forthe same voltage waveform (when third harmonic is short-circuited)
In the first example current waveform (188) and 1199113=
02 minus 11989505 imply that 2712058223V le (1 minus 2119886
3V)3 Then according
to Algorithm 33 the voltage waveform of type (189) thatprovides maximal efficiency has the following coefficients1198863V = minus00553 119887
3V = minus01382 1198861V = minus09447 and 119887
1V =
04146 (see Figure 23) Efficiency PUF and normalized thirdharmonic impedance are 120578 = 07643 PUF = 08069 and1199113119899= 03425 minus 11989508564 respectivelyIn the second example current waveform (188) and 119911
3=
01 minus 11989511 imply that 2712058223V gt (1 minus 2119886
3V)3 Then according
to Algorithm 33 the voltage waveform of type (189) thatprovides maximal efficiency has the following coefficients1198863V = minus00276 119887
3V = minus03040 1198861V = minus09391 and 119887
1V =
05807 (see Figure 24) Efficiency PUF and normalized thirdharmonic impedance are 120578 = 07598 PUF = 08021 and1199113119899= 01723 minus 11989518952 respectively
26 Mathematical Problems in Engineering
2
3
4
1
0
2 3 410
Wav
efor
ms
Angle 120579120587
i(120579)
(120579)
Figure 23 Waveform pair (188)-(189) that provides maximal effi-ciency for 119911
3= 02 minus 11989505
2
3
4
1
0
2 3 410
Wav
efor
ms
Angle 120579120587
i(120579)
(120579)
Figure 24 Waveform pair (188)-(189) that provides maximal effi-ciency for 119911
3= 01 minus 11989511
Case Study 74 In this case study let us consider current-voltage pair where current is normalized waveform of type(180) with conduction angle 120572 = 115120587 (207∘) and voltageis nonnegative waveform of type (189) Substitution of 120572 =
115120587 and 119868dc = 1 into (182) leads to 119868119863
= 22535Furthermore substitution of 120572 = 115120587 and 119868
119863= 22535 into
(180) leads to
119894 (120579)
=
22535 [cos 120579 minus cos(1151205872
)] |120579| le115120587
2
0115120587
2le |120579| le 120587
(191)
Similarly substitution of 120572 = 115120587 and 119868119863
= 22535 into(183) and (184) for 119899 = 3 yields coefficients of fundamentaland third harmonic of waveform (191)
1198861119894= 14586 119886
3119894= minus01026 (192)
Because current waveform (191) contains only cosineterms and voltage waveform is the same as in previous case
+j2
+j5
+j1
+j2
+j5
minusj2
minusj5
minusj1
minusj2
minusj5
infin0 02
05
1 2
074
076
08
084
082
078
076
074
Figure 25 Contours of maximal efficiency of PA associated withthe waveform pair (189) and (191) as a function of normalized thirdharmonic impedance 119911
3119899= 1199113Re119911
1
study the procedure for calculation of maximal efficiencyof PA with waveform pair (189)ndash(191) is the same as thatpresented in Algorithm 33 In this case study the coefficientsof fundamental and third harmonic of current waveform aregiven by (192)
For the waveform pair (189) and (191) maximal efficiencyof PA as a function of normalized third harmonic impedance1199113119899= 1199113Re119911
1 is presented in Figure 25 Efficiency of 084
is obtained in vicinity of 1199113119899= 23685 (corresponding to 119911
3=
18750) Peak efficiency 120578 asymp 08421 is achieved for voltagewaveform of type (189) with coefficients 119886
1V = minus2radic3 1198863V =
radic39 and 1198871V = 1198873V = 0
In the course of finding power utilization factor noticethat currentwaveformof type (191) attains itsmaximumvaluefor 120579 = 0 Insertion of max[119894(120579)] = 119894(0) = 278 andmax[V(120579)] = 2 for voltage waveform of type (189) into (179)leads to
PUFCase study 74 = 1439120578 (193)
Again the ratio PUF120578 is constant and PUF can be easilycalculated from the corresponding efficiency Accordinglypeak value of PUFCase study 74 asymp 12118 and peak efficiencyare attained for the same voltage waveform
In the first example current waveform (191) and 1199113=
1 minus 11989502 imply that 2712058223V le (1 minus 2119886
3V)3 Then according to
Algorithm 33 voltage waveform of type (189) which providesmaximal efficiency has coefficients 119886
3V = 01026 1198873V =
00205 1198861V = minus11026 and 119887
1V = minus00616 (see Figure 26)Efficiency PUF and normalized third harmonic impedanceare 120578 = 08042 PUF = 11572 and 119911
3119899= 13228 minus 11989502646
respectivelyIn second example current waveform (191) and 119911
3=
15 minus 11989512 imply that 2712058223V gt (1 minus 2119886
3V)3 Then according to
Algorithm 33 voltage waveform of type (189) which providesmaximal efficiency has coefficients 119886
3V = 01540 1198873V =
Mathematical Problems in Engineering 27
2
3
4
1
0
2 3 410
Wav
efor
ms
Angle 120579120587
i(120579)
(120579)
Figure 26 Waveform pair (189) and (191) that provides maximalefficiency for 119911
3= 1 minus 11989502
2
3
4
1
0
2 3 410
Wav
efor
ms
Angle 120579120587
i(120579)
(120579)
Figure 27 Waveform pair (189) and (191) that provides maximalefficiency for 119911
3= 15 minus 11989512
01232 1198861V = minus11255 and 119887
1V = minus02575 (see Figure 27)Efficiency PUF and normalized third harmonic impedanceare 120578 = 08208 PUF = 11812 and 119911
3119899= 19439 minus 11989515552
respectively
8 Conclusion
In this paper we consider a problem of finding general de-scriptions of various classes of nonnegative waveforms withfundamental and 119896th harmonic These classes include non-negative waveforms with at least one zero nonnegative wave-forms with maximal amplitude of fundamental harmonic forprescribed amplitude of 119896th harmonic nonnegative wave-forms withmaximal coefficient of cosine part of fundamentalharmonic for prescribed coefficients of 119896th harmonic andnonnegative cosine waveforms with at least one zero Mainresults are stated in six propositions (Propositions 1 6 918 22 and 26) four corollaries (Corollaries 2ndash5) twentyremarks and three algorithms Four case studies of usageof closed form descriptions of nonnegative waveforms in PAefficiency analysis are considered in detail in Section 7
Appendices
Here we provide a list of finite sums of trigonometric func-tions used in this paper (Appendix A) and brief account ofthe Chebyshev polynomials (Appendix B)
A List of Some Finite Sums ofTrigonometric Functions
Dirichlet kernel (eg see [27]) is as follows
119863119896minus1
(120591) = 1 + 2
119896minus1
sum
119899=1
cos 119899120591 = sin ((2119896 minus 1) 1205912)sin (1205912)
(A1)
Fejer kernel (eg see [27]) can be expressed in the fol-lowing equivalent forms
119865119896minus1
(120591) =1
119896
119896minus1
sum
119899=0
119863119899(120591) = 1 +
2
119896
119896minus1
sum
119899=1
(119896 minus 119899) cos 119899120591
=(1 minus cos 119896120591)119896 (1 minus cos 120591)
(A2)
Lagrangersquos trigonometric identity (eg see [28]) is as fol-lows
1198781(120591) =
119896minus1
sum
119899=1
sin 119899120591 = sin (1198961205912) sin ((119896 minus 1) 1205912)sin (1205912)
(A3)
In what follows we show that the following three trigono-metric identities also hold
2
119896minus1
sum
119899=1
(119896 minus 119899) sin 119899120591 = 119896 sin 120591 minus sin 1198961205911 minus cos 120591
(A4)
119896minus1
sum
119899=1
cos (119896 minus 2119899) 120591 = sin (119896 minus 1) 120591sin 120591
(A5)
119896minus1
sum
119899=1
119899 (119896 minus 119899) cos (119896 minus 2119899) 120591
=sin (119896120591) cos 120591 minus 119896 cos (119896120591) sin 120591
2sin3120591
(A6)
Denote 1198782(120591) = 2sum
119896minus1
119899=1(119896 minus 119899) sin 119899120591 119878
3(120591) = sum
119896minus1
119899=1cos(119896 minus
2119899)120591 and 1198784(120591) = sum
119896minus1
119899=1119899(119896 minus 119899) cos(119896 minus 2119899)120591
Notice that 1198782(120591) = 2119896119878
1(120591) + 119889119863
119896minus1(120591)119889120591 which
immediately leads to (A4)
28 Mathematical Problems in Engineering
Identity (A5) can be obtained as follows
sin (119896 minus 1) 120591sin 120591
=119890119895(119896minus1)120591
minus 119890minus119895(119896minus1)120591
119890119895120591 minus 119890minus119895120591
= 119890119895119896120591
119890minus2119895120591
minus 119890minus2119895(119896minus1)120591
1 minus 119890minus2119895120591
= 119890119895119896120591
119896minus1
sum
119899=1
119890minus2119895119899120591
=
119896minus1
sum
119899=1
119890119895(119896minus2119899)120591
=
119896minus1
sum
119899=1
cos (119896 minus 2119899) 120591
(A7)
From 4119899(119896 minus 119899) = 1198962
minus (119896 minus 2119899)2 it follows that 4119878
4(120591) =
1198962
1198783(120591) + 119889
2
1198783(120591)119889120591
2 which leads to (A6)
B The Chebyshev Polynomials
The Chebyshev polynomials of the first kind 119881119899(119909) can be
defined by the following relation (eg see [29])
119881119899(119909) = cos 119899120591 when 119909 = cos 120591 (B1)
The Chebyshev polynomials of the second kind 119880119899(119909) can be
defined by the following relation (eg see [29])
119880119899(119909) =
sin (119899 + 1) 120591sin 120591
when 119909 = cos 120591 (B2)
The Chebyshev polynomials satisfy the following recur-rence relations (eg see [29])
1198810(119909) = 1 119881
1(119909) = 119909
119881119899+1
(119909) = 2119909119881119899(119909) minus 119881
119899minus1(119909)
1198800(119909) = 1 119880
1(119909) = 2119909
119880119899+1
(119909) = 2119909119880119899(119909) minus 119880
119899minus1(119909)
(B3)
The first few Chebyshev polynomials of the first and secondkind are 119881
2(119909) = 2119909
2
minus 1 1198813(119909) = 4119909
3
minus 3119909 1198814(119909) = 8119909
4
minus
81199092
+ 1 1198802(119909) = 4119909
2
minus 1 1198803(119909) = 8119909
3
minus 4119909 and 1198804(119909) =
161199094
minus 121199092
+ 1
Conflict of Interests
The authors declare that there is no conflict of interestsregarding the publication of this paper
Acknowledgment
This work is supported by the SerbianMinistry of EducationScience and Technology Development as a part of ProjectTP32016
References
[1] V I Arnolrsquod V S Afrajmovich Y S Ilrsquoyashenko and L PShilrsquonikov Dynamical Systems V Bifurcation Theory and Catas-trophe Theory Springer Berlin Germany 1994
[2] E Polak ldquoOn the mathematical foundations of nondifferen-tiable optimization in engineering designrdquo SIAM Review vol29 no 1 pp 21ndash89 1987
[3] N S Fuzik ldquoBiharmonic modes of a tuned RF power amplifierrdquoRadiotehnika vol 25 no 7 pp 62ndash71 1970 (Russian)
[4] P Colantonio F Giannini G Leuzzi and E Limiti ldquoClass Gapproach for low-voltage high-efficiency PA designrdquo Interna-tional Journal of RF and Microwave Computer-Aided Engineer-ing vol 10 no 6 pp 366ndash378 2000
[5] F H Raab ldquoMaximum efficiency and output of class-F poweramplifiersrdquo IEEE Transactions on Microwave Theory and Tech-niques vol 49 no 6 pp 1162ndash1166 2001
[6] FH Raab ldquoClass-E class-C and class-F power amplifiers basedupon a finite number of harmonicsrdquo IEEE Transactions onMicrowaveTheory and Techniques vol 49 no 8 pp 1462ndash14682001
[7] J D Rhodes ldquoOutput universality inmaximum efficiency linearpower amplifiersrdquo International Journal of Circuit Theory andApplications vol 31 no 4 pp 385ndash405 2003
[8] A Juhas and L A Novak ldquoComments on lsquoClass-E class-Cand classF power amplifier based upon a finite number ofharmonicsrsquordquo IEEE Transactions on Microwave Theory and Tech-niques vol 57 no 6 pp 1623ndash1625 2009
[9] M Roberg and Z Popovic ldquoAnalysis of high-efficiency poweramplifiers with arbitrary output harmonic terminationsrdquo IEEETransactions on Microwave Theory and Techniques vol 59 no8 pp 2037ndash2048 2011
[10] A Grebennikov N O Sokal and M J Franco Switchmode RFPower Amplifiers ElsevierAcademic Press San Diego CalifUSA 2nd edition 2012
[11] T Canning P J Tasker and S C Cripps ldquoContinuous modepower amplifier design using harmonic clipping contourstheory and practicerdquo IEEE Transactions on Microwave Theoryand Techniques vol 62 no 1 pp 100ndash110 2014
[12] A Juhas and L A Novak ldquoGeneral description of nonnegativewaveforms up to second harmonic for power amplifier mod-ellingrdquoMathematical Problems in Engineering vol 2014 ArticleID 709762 18 pages 2014
[13] V I Arnolrsquod V V Goryunov O V Lyashko and V A VasilrsquoevDynamical Systems VIIImdashSingularity Theory II ApplicationsSpringer Berlin Germany 1993
[14] D Siersma ldquoProperties of conflict sets in the planerdquo BanachCenter Publications Polish Academy of Sciences vol 50 no 1 pp267ndash276 1999 Proceedings of the Banach Center Symposiumon Geometry and Topology of Caustics (Caustics rsquo98) WarsawPoland
[15] M van Manen The geometry of conflict sets [Dissertation]Universiteit Utrecht UtrechtThe Netherlands 2003 httpigi-tur-archivelibraryuunldissertations2003-0912-123058c4pdf
[16] Y L Sachkov ldquoMaxwell strata and symmetries in the problemofoptimal rolling of a sphere over a planerdquo Sbornik Mathematicsvol 201 no 7-8 pp 1029ndash1051 2010
[17] I A Bogaevsky ldquoPerestroikas of shocks and singularities ofminimum functionsrdquoPhysicaDNonlinear Phenomena vol 173no 1-2 pp 1ndash28 2002
[18] Y L Sachkov ldquoMaxwell strata in the Euler elastic problemrdquoJournal of Dynamical andControl Systems vol 14 no 2 pp 169ndash234 2008
[19] M Siino and T Koike ldquoTopological classification of black holesgeneric Maxwell set and crease set of a horizonrdquo InternationalJournal ofModern Physics D Gravitation Astrophysics Cosmol-ogy vol 20 no 6 pp 1095ndash1122 2011
Mathematical Problems in Engineering 29
[20] F H Raab ldquoClass-F power amplifiers with maximally flat wave-formsrdquo IEEETransactions onMicrowaveTheory and Techniquesvol 45 no 11 pp 2007ndash2012 1997
[21] A Juhas and L A Novak ldquoMaximally flat waveforms with finitenumber of harmonics in class-F power amplifiersrdquo Mathemat-ical Problems in Engineering vol 2013 Article ID 169590 9pages 2013
[22] L Fejer ldquoUber trigonometrische polynomerdquo Journal fur dieReine und Angewandte Mathematik vol 1916 no 146 pp 53ndash82 1916 (German)
[23] S C Cripps ldquoBessel Waives [microwave bytes]rdquo IEEE Micro-wave Magazine vol 10 no 7 pp 30ndash36 117 2009
[24] L N Bryzgalova ldquoSingularities of the maximum of para-metrically dependent functionrdquo Functional Analysis and ItsApplications vol 11 no 1 pp 49ndash51 1977
[25] V I Arnold A A Davydov V A Vassiliev and V MZakalyukin Mathematical Models of Catastrophes Control ofCatastrophic Process Encyclopedia of Life Support Systems(EOLSS) EOLSS Publishers Oxford UK 2006
[26] S C Cripps RF Power Amplifiers for Wireless CommunicationsArtech House Norwood Mass USA 2nd edition 2006
[27] A Zygmund Trigonometric Series vol 1 Cambridge UniversityPress Cambridge UK 2nd edition 1959
[28] A Jeffrey and H DaiHandbook of Mathematical Formulas andIntegrals ElsevierAcademic Press San Diego Calif USA 4thedition 2008
[29] J C Mason and D C Handscomb Chebyshev PolynomialsChapman amp Hall CRC Press Boca Raton Fla USA 2003
Submit your manuscripts athttpwwwhindawicom
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical Problems in Engineering
Hindawi Publishing Corporationhttpwwwhindawicom
Differential EquationsInternational Journal of
Volume 2014
Applied MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Probability and StatisticsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical PhysicsAdvances in
Complex AnalysisJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
OptimizationJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
CombinatoricsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Operations ResearchAdvances in
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Function Spaces
Abstract and Applied AnalysisHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of Mathematics and Mathematical Sciences
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Algebra
Discrete Dynamics in Nature and Society
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Decision SciencesAdvances in
Discrete MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom
Volume 2014 Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Stochastic AnalysisInternational Journal of
Mathematical Problems in Engineering 25
2
3
4
1
0
2 3 410
Wav
efor
ms
Angle 120579120587
i(120579)
(120579)
Figure 21 Waveform pair (187)-(188) that provides maximal effi-ciency for 119911
2= 005 minus 11989507
and current is nonnegative cosine waveform given by (188)Load impedances at fundamental second and third har-monic are 119911
1= minus(119886
1V minus 1198951198871V)1198861119894 1199112 = 0 and 119911
3= minus(119886
3V minus
1198951198873V)1198863119894 respectively According to our assumption the load
is passive and therefore Re1199111 gt 0 and Re119911
3 ge 0 which
further imply 11988611198941198861V lt 0 and 119886
31198941198863V le 0
In this subsection we consider the problem of findingmaximal efficiency of PA with waveform pair (188)-(189)for given third harmonic impedance As we mentionedearlier problem of finding maximal efficiency of PA withcurrent-voltage pair (188)-(189) for prescribed third har-monic impedance can be reduced to the problem of findingvoltage waveform of type (189) withmaximal coefficient |119886
1V|for prescribed coefficients of third harmonic (see Section 52)
The following algorithm provides the procedure forcalculation of maximal efficiency with current-voltage pair(188)-(189) The definition of function atan 2(119910 119909) whichappears in step (iii) of the following algorithm is given by(105)
Algorithm 33 (i) Choose 1199113= 1199033+1198951199093such that |119911
3| le 1|119886
3119894|
(ii) calculate 1198863V minus 1198951198873V = minus119911
31198863119894and 120582
3V = radic11988623V + 1198872
3V(iii) if 271205822
3V le (1 minus 21198863V)3 then calculate 119886
1V = minus1 minus 1198863V
and 1198871V = minus3119887
3V else calculate 1205821V = 3(3radic1205823V minus 120582
3V) 1205790V minus120585V3 = (13)atan 2(119887
3V 1198863V) 1198861V = minus1205821V cos(1205790V minus 120585V3) and
1198871V = minus120582
1V sin(1205790V minus 120585V3)(iv) calculate efficiency 120578 = minus119886
11198941198861V2
(v) calculate 1199111= minus(119886
1V minus 1198951198871V)1198861119894 and 1199113119899 = 1199113Re119911
1
In this case study coefficients of fundamental and thirdharmonic of current waveform are 119886
1119894= (1 + radic5)2 and
1198863119894= (5 minus radic5)10 respectively For the waveform pair (188)-
(189) maximal efficiency of PA as a function of normalizedthird harmonic impedance 119911
3119899= 1199113Re119911
1 is presented in
Figure 22 Efficiency of 08 is reached when third harmonicimpedance has small resistive part Peak efficiency 120578 = (1 +
radic5)4 asymp 08090 is achieved when third harmonic is short-circuited
+j2
+j5
+j1
+j2
+j5
minusj2
minusj5
minusj1
minusj2
minusj5
infin0
08
08
075
075
07
07
06
05
02
05
1 2
Figure 22 Contours of maximal efficiency of PA associated withthe waveform pair (188)-(189) as a function of normalized thirdharmonic impedance 119911
3119899= 1199113Re119911
1
For the present case study in what follows we show thatpower utilization factor is proportional to efficiency For volt-age waveform of type (189) it is easy to see that V(120579 + 120587) =
2minusV(120579) holdsThis relation along with the fact that waveformV(120579) that provides maximal efficiency has at least one zeroimplies that max[V(120579)] = 2 On the other hand currentwaveform (188) is cosine waveform with positive coefficientsand therefore max[119894(120579)] = 119894(0) = 2 + 4radic5 Consequentlyaccording to (179) the following relation holds
PUFCase study 73 = 2 (5 minus 2radic5) 120578 = 10557120578 (190)
Clearly the ratio PUF120578 is constant and therefore in this casestudy PUF can be easily calculated from the correspondingefficiency Accordingly peak efficiency and peak value ofPUFCase study 73 = 3radic52 minus 52 = 08541 are attained forthe same voltage waveform (when third harmonic is short-circuited)
In the first example current waveform (188) and 1199113=
02 minus 11989505 imply that 2712058223V le (1 minus 2119886
3V)3 Then according
to Algorithm 33 the voltage waveform of type (189) thatprovides maximal efficiency has the following coefficients1198863V = minus00553 119887
3V = minus01382 1198861V = minus09447 and 119887
1V =
04146 (see Figure 23) Efficiency PUF and normalized thirdharmonic impedance are 120578 = 07643 PUF = 08069 and1199113119899= 03425 minus 11989508564 respectivelyIn the second example current waveform (188) and 119911
3=
01 minus 11989511 imply that 2712058223V gt (1 minus 2119886
3V)3 Then according
to Algorithm 33 the voltage waveform of type (189) thatprovides maximal efficiency has the following coefficients1198863V = minus00276 119887
3V = minus03040 1198861V = minus09391 and 119887
1V =
05807 (see Figure 24) Efficiency PUF and normalized thirdharmonic impedance are 120578 = 07598 PUF = 08021 and1199113119899= 01723 minus 11989518952 respectively
26 Mathematical Problems in Engineering
2
3
4
1
0
2 3 410
Wav
efor
ms
Angle 120579120587
i(120579)
(120579)
Figure 23 Waveform pair (188)-(189) that provides maximal effi-ciency for 119911
3= 02 minus 11989505
2
3
4
1
0
2 3 410
Wav
efor
ms
Angle 120579120587
i(120579)
(120579)
Figure 24 Waveform pair (188)-(189) that provides maximal effi-ciency for 119911
3= 01 minus 11989511
Case Study 74 In this case study let us consider current-voltage pair where current is normalized waveform of type(180) with conduction angle 120572 = 115120587 (207∘) and voltageis nonnegative waveform of type (189) Substitution of 120572 =
115120587 and 119868dc = 1 into (182) leads to 119868119863
= 22535Furthermore substitution of 120572 = 115120587 and 119868
119863= 22535 into
(180) leads to
119894 (120579)
=
22535 [cos 120579 minus cos(1151205872
)] |120579| le115120587
2
0115120587
2le |120579| le 120587
(191)
Similarly substitution of 120572 = 115120587 and 119868119863
= 22535 into(183) and (184) for 119899 = 3 yields coefficients of fundamentaland third harmonic of waveform (191)
1198861119894= 14586 119886
3119894= minus01026 (192)
Because current waveform (191) contains only cosineterms and voltage waveform is the same as in previous case
+j2
+j5
+j1
+j2
+j5
minusj2
minusj5
minusj1
minusj2
minusj5
infin0 02
05
1 2
074
076
08
084
082
078
076
074
Figure 25 Contours of maximal efficiency of PA associated withthe waveform pair (189) and (191) as a function of normalized thirdharmonic impedance 119911
3119899= 1199113Re119911
1
study the procedure for calculation of maximal efficiencyof PA with waveform pair (189)ndash(191) is the same as thatpresented in Algorithm 33 In this case study the coefficientsof fundamental and third harmonic of current waveform aregiven by (192)
For the waveform pair (189) and (191) maximal efficiencyof PA as a function of normalized third harmonic impedance1199113119899= 1199113Re119911
1 is presented in Figure 25 Efficiency of 084
is obtained in vicinity of 1199113119899= 23685 (corresponding to 119911
3=
18750) Peak efficiency 120578 asymp 08421 is achieved for voltagewaveform of type (189) with coefficients 119886
1V = minus2radic3 1198863V =
radic39 and 1198871V = 1198873V = 0
In the course of finding power utilization factor noticethat currentwaveformof type (191) attains itsmaximumvaluefor 120579 = 0 Insertion of max[119894(120579)] = 119894(0) = 278 andmax[V(120579)] = 2 for voltage waveform of type (189) into (179)leads to
PUFCase study 74 = 1439120578 (193)
Again the ratio PUF120578 is constant and PUF can be easilycalculated from the corresponding efficiency Accordinglypeak value of PUFCase study 74 asymp 12118 and peak efficiencyare attained for the same voltage waveform
In the first example current waveform (191) and 1199113=
1 minus 11989502 imply that 2712058223V le (1 minus 2119886
3V)3 Then according to
Algorithm 33 voltage waveform of type (189) which providesmaximal efficiency has coefficients 119886
3V = 01026 1198873V =
00205 1198861V = minus11026 and 119887
1V = minus00616 (see Figure 26)Efficiency PUF and normalized third harmonic impedanceare 120578 = 08042 PUF = 11572 and 119911
3119899= 13228 minus 11989502646
respectivelyIn second example current waveform (191) and 119911
3=
15 minus 11989512 imply that 2712058223V gt (1 minus 2119886
3V)3 Then according to
Algorithm 33 voltage waveform of type (189) which providesmaximal efficiency has coefficients 119886
3V = 01540 1198873V =
Mathematical Problems in Engineering 27
2
3
4
1
0
2 3 410
Wav
efor
ms
Angle 120579120587
i(120579)
(120579)
Figure 26 Waveform pair (189) and (191) that provides maximalefficiency for 119911
3= 1 minus 11989502
2
3
4
1
0
2 3 410
Wav
efor
ms
Angle 120579120587
i(120579)
(120579)
Figure 27 Waveform pair (189) and (191) that provides maximalefficiency for 119911
3= 15 minus 11989512
01232 1198861V = minus11255 and 119887
1V = minus02575 (see Figure 27)Efficiency PUF and normalized third harmonic impedanceare 120578 = 08208 PUF = 11812 and 119911
3119899= 19439 minus 11989515552
respectively
8 Conclusion
In this paper we consider a problem of finding general de-scriptions of various classes of nonnegative waveforms withfundamental and 119896th harmonic These classes include non-negative waveforms with at least one zero nonnegative wave-forms with maximal amplitude of fundamental harmonic forprescribed amplitude of 119896th harmonic nonnegative wave-forms withmaximal coefficient of cosine part of fundamentalharmonic for prescribed coefficients of 119896th harmonic andnonnegative cosine waveforms with at least one zero Mainresults are stated in six propositions (Propositions 1 6 918 22 and 26) four corollaries (Corollaries 2ndash5) twentyremarks and three algorithms Four case studies of usageof closed form descriptions of nonnegative waveforms in PAefficiency analysis are considered in detail in Section 7
Appendices
Here we provide a list of finite sums of trigonometric func-tions used in this paper (Appendix A) and brief account ofthe Chebyshev polynomials (Appendix B)
A List of Some Finite Sums ofTrigonometric Functions
Dirichlet kernel (eg see [27]) is as follows
119863119896minus1
(120591) = 1 + 2
119896minus1
sum
119899=1
cos 119899120591 = sin ((2119896 minus 1) 1205912)sin (1205912)
(A1)
Fejer kernel (eg see [27]) can be expressed in the fol-lowing equivalent forms
119865119896minus1
(120591) =1
119896
119896minus1
sum
119899=0
119863119899(120591) = 1 +
2
119896
119896minus1
sum
119899=1
(119896 minus 119899) cos 119899120591
=(1 minus cos 119896120591)119896 (1 minus cos 120591)
(A2)
Lagrangersquos trigonometric identity (eg see [28]) is as fol-lows
1198781(120591) =
119896minus1
sum
119899=1
sin 119899120591 = sin (1198961205912) sin ((119896 minus 1) 1205912)sin (1205912)
(A3)
In what follows we show that the following three trigono-metric identities also hold
2
119896minus1
sum
119899=1
(119896 minus 119899) sin 119899120591 = 119896 sin 120591 minus sin 1198961205911 minus cos 120591
(A4)
119896minus1
sum
119899=1
cos (119896 minus 2119899) 120591 = sin (119896 minus 1) 120591sin 120591
(A5)
119896minus1
sum
119899=1
119899 (119896 minus 119899) cos (119896 minus 2119899) 120591
=sin (119896120591) cos 120591 minus 119896 cos (119896120591) sin 120591
2sin3120591
(A6)
Denote 1198782(120591) = 2sum
119896minus1
119899=1(119896 minus 119899) sin 119899120591 119878
3(120591) = sum
119896minus1
119899=1cos(119896 minus
2119899)120591 and 1198784(120591) = sum
119896minus1
119899=1119899(119896 minus 119899) cos(119896 minus 2119899)120591
Notice that 1198782(120591) = 2119896119878
1(120591) + 119889119863
119896minus1(120591)119889120591 which
immediately leads to (A4)
28 Mathematical Problems in Engineering
Identity (A5) can be obtained as follows
sin (119896 minus 1) 120591sin 120591
=119890119895(119896minus1)120591
minus 119890minus119895(119896minus1)120591
119890119895120591 minus 119890minus119895120591
= 119890119895119896120591
119890minus2119895120591
minus 119890minus2119895(119896minus1)120591
1 minus 119890minus2119895120591
= 119890119895119896120591
119896minus1
sum
119899=1
119890minus2119895119899120591
=
119896minus1
sum
119899=1
119890119895(119896minus2119899)120591
=
119896minus1
sum
119899=1
cos (119896 minus 2119899) 120591
(A7)
From 4119899(119896 minus 119899) = 1198962
minus (119896 minus 2119899)2 it follows that 4119878
4(120591) =
1198962
1198783(120591) + 119889
2
1198783(120591)119889120591
2 which leads to (A6)
B The Chebyshev Polynomials
The Chebyshev polynomials of the first kind 119881119899(119909) can be
defined by the following relation (eg see [29])
119881119899(119909) = cos 119899120591 when 119909 = cos 120591 (B1)
The Chebyshev polynomials of the second kind 119880119899(119909) can be
defined by the following relation (eg see [29])
119880119899(119909) =
sin (119899 + 1) 120591sin 120591
when 119909 = cos 120591 (B2)
The Chebyshev polynomials satisfy the following recur-rence relations (eg see [29])
1198810(119909) = 1 119881
1(119909) = 119909
119881119899+1
(119909) = 2119909119881119899(119909) minus 119881
119899minus1(119909)
1198800(119909) = 1 119880
1(119909) = 2119909
119880119899+1
(119909) = 2119909119880119899(119909) minus 119880
119899minus1(119909)
(B3)
The first few Chebyshev polynomials of the first and secondkind are 119881
2(119909) = 2119909
2
minus 1 1198813(119909) = 4119909
3
minus 3119909 1198814(119909) = 8119909
4
minus
81199092
+ 1 1198802(119909) = 4119909
2
minus 1 1198803(119909) = 8119909
3
minus 4119909 and 1198804(119909) =
161199094
minus 121199092
+ 1
Conflict of Interests
The authors declare that there is no conflict of interestsregarding the publication of this paper
Acknowledgment
This work is supported by the SerbianMinistry of EducationScience and Technology Development as a part of ProjectTP32016
References
[1] V I Arnolrsquod V S Afrajmovich Y S Ilrsquoyashenko and L PShilrsquonikov Dynamical Systems V Bifurcation Theory and Catas-trophe Theory Springer Berlin Germany 1994
[2] E Polak ldquoOn the mathematical foundations of nondifferen-tiable optimization in engineering designrdquo SIAM Review vol29 no 1 pp 21ndash89 1987
[3] N S Fuzik ldquoBiharmonic modes of a tuned RF power amplifierrdquoRadiotehnika vol 25 no 7 pp 62ndash71 1970 (Russian)
[4] P Colantonio F Giannini G Leuzzi and E Limiti ldquoClass Gapproach for low-voltage high-efficiency PA designrdquo Interna-tional Journal of RF and Microwave Computer-Aided Engineer-ing vol 10 no 6 pp 366ndash378 2000
[5] F H Raab ldquoMaximum efficiency and output of class-F poweramplifiersrdquo IEEE Transactions on Microwave Theory and Tech-niques vol 49 no 6 pp 1162ndash1166 2001
[6] FH Raab ldquoClass-E class-C and class-F power amplifiers basedupon a finite number of harmonicsrdquo IEEE Transactions onMicrowaveTheory and Techniques vol 49 no 8 pp 1462ndash14682001
[7] J D Rhodes ldquoOutput universality inmaximum efficiency linearpower amplifiersrdquo International Journal of Circuit Theory andApplications vol 31 no 4 pp 385ndash405 2003
[8] A Juhas and L A Novak ldquoComments on lsquoClass-E class-Cand classF power amplifier based upon a finite number ofharmonicsrsquordquo IEEE Transactions on Microwave Theory and Tech-niques vol 57 no 6 pp 1623ndash1625 2009
[9] M Roberg and Z Popovic ldquoAnalysis of high-efficiency poweramplifiers with arbitrary output harmonic terminationsrdquo IEEETransactions on Microwave Theory and Techniques vol 59 no8 pp 2037ndash2048 2011
[10] A Grebennikov N O Sokal and M J Franco Switchmode RFPower Amplifiers ElsevierAcademic Press San Diego CalifUSA 2nd edition 2012
[11] T Canning P J Tasker and S C Cripps ldquoContinuous modepower amplifier design using harmonic clipping contourstheory and practicerdquo IEEE Transactions on Microwave Theoryand Techniques vol 62 no 1 pp 100ndash110 2014
[12] A Juhas and L A Novak ldquoGeneral description of nonnegativewaveforms up to second harmonic for power amplifier mod-ellingrdquoMathematical Problems in Engineering vol 2014 ArticleID 709762 18 pages 2014
[13] V I Arnolrsquod V V Goryunov O V Lyashko and V A VasilrsquoevDynamical Systems VIIImdashSingularity Theory II ApplicationsSpringer Berlin Germany 1993
[14] D Siersma ldquoProperties of conflict sets in the planerdquo BanachCenter Publications Polish Academy of Sciences vol 50 no 1 pp267ndash276 1999 Proceedings of the Banach Center Symposiumon Geometry and Topology of Caustics (Caustics rsquo98) WarsawPoland
[15] M van Manen The geometry of conflict sets [Dissertation]Universiteit Utrecht UtrechtThe Netherlands 2003 httpigi-tur-archivelibraryuunldissertations2003-0912-123058c4pdf
[16] Y L Sachkov ldquoMaxwell strata and symmetries in the problemofoptimal rolling of a sphere over a planerdquo Sbornik Mathematicsvol 201 no 7-8 pp 1029ndash1051 2010
[17] I A Bogaevsky ldquoPerestroikas of shocks and singularities ofminimum functionsrdquoPhysicaDNonlinear Phenomena vol 173no 1-2 pp 1ndash28 2002
[18] Y L Sachkov ldquoMaxwell strata in the Euler elastic problemrdquoJournal of Dynamical andControl Systems vol 14 no 2 pp 169ndash234 2008
[19] M Siino and T Koike ldquoTopological classification of black holesgeneric Maxwell set and crease set of a horizonrdquo InternationalJournal ofModern Physics D Gravitation Astrophysics Cosmol-ogy vol 20 no 6 pp 1095ndash1122 2011
Mathematical Problems in Engineering 29
[20] F H Raab ldquoClass-F power amplifiers with maximally flat wave-formsrdquo IEEETransactions onMicrowaveTheory and Techniquesvol 45 no 11 pp 2007ndash2012 1997
[21] A Juhas and L A Novak ldquoMaximally flat waveforms with finitenumber of harmonics in class-F power amplifiersrdquo Mathemat-ical Problems in Engineering vol 2013 Article ID 169590 9pages 2013
[22] L Fejer ldquoUber trigonometrische polynomerdquo Journal fur dieReine und Angewandte Mathematik vol 1916 no 146 pp 53ndash82 1916 (German)
[23] S C Cripps ldquoBessel Waives [microwave bytes]rdquo IEEE Micro-wave Magazine vol 10 no 7 pp 30ndash36 117 2009
[24] L N Bryzgalova ldquoSingularities of the maximum of para-metrically dependent functionrdquo Functional Analysis and ItsApplications vol 11 no 1 pp 49ndash51 1977
[25] V I Arnold A A Davydov V A Vassiliev and V MZakalyukin Mathematical Models of Catastrophes Control ofCatastrophic Process Encyclopedia of Life Support Systems(EOLSS) EOLSS Publishers Oxford UK 2006
[26] S C Cripps RF Power Amplifiers for Wireless CommunicationsArtech House Norwood Mass USA 2nd edition 2006
[27] A Zygmund Trigonometric Series vol 1 Cambridge UniversityPress Cambridge UK 2nd edition 1959
[28] A Jeffrey and H DaiHandbook of Mathematical Formulas andIntegrals ElsevierAcademic Press San Diego Calif USA 4thedition 2008
[29] J C Mason and D C Handscomb Chebyshev PolynomialsChapman amp Hall CRC Press Boca Raton Fla USA 2003
Submit your manuscripts athttpwwwhindawicom
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical Problems in Engineering
Hindawi Publishing Corporationhttpwwwhindawicom
Differential EquationsInternational Journal of
Volume 2014
Applied MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Probability and StatisticsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical PhysicsAdvances in
Complex AnalysisJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
OptimizationJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
CombinatoricsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Operations ResearchAdvances in
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Function Spaces
Abstract and Applied AnalysisHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of Mathematics and Mathematical Sciences
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Algebra
Discrete Dynamics in Nature and Society
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Decision SciencesAdvances in
Discrete MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom
Volume 2014 Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Stochastic AnalysisInternational Journal of
26 Mathematical Problems in Engineering
2
3
4
1
0
2 3 410
Wav
efor
ms
Angle 120579120587
i(120579)
(120579)
Figure 23 Waveform pair (188)-(189) that provides maximal effi-ciency for 119911
3= 02 minus 11989505
2
3
4
1
0
2 3 410
Wav
efor
ms
Angle 120579120587
i(120579)
(120579)
Figure 24 Waveform pair (188)-(189) that provides maximal effi-ciency for 119911
3= 01 minus 11989511
Case Study 74 In this case study let us consider current-voltage pair where current is normalized waveform of type(180) with conduction angle 120572 = 115120587 (207∘) and voltageis nonnegative waveform of type (189) Substitution of 120572 =
115120587 and 119868dc = 1 into (182) leads to 119868119863
= 22535Furthermore substitution of 120572 = 115120587 and 119868
119863= 22535 into
(180) leads to
119894 (120579)
=
22535 [cos 120579 minus cos(1151205872
)] |120579| le115120587
2
0115120587
2le |120579| le 120587
(191)
Similarly substitution of 120572 = 115120587 and 119868119863
= 22535 into(183) and (184) for 119899 = 3 yields coefficients of fundamentaland third harmonic of waveform (191)
1198861119894= 14586 119886
3119894= minus01026 (192)
Because current waveform (191) contains only cosineterms and voltage waveform is the same as in previous case
+j2
+j5
+j1
+j2
+j5
minusj2
minusj5
minusj1
minusj2
minusj5
infin0 02
05
1 2
074
076
08
084
082
078
076
074
Figure 25 Contours of maximal efficiency of PA associated withthe waveform pair (189) and (191) as a function of normalized thirdharmonic impedance 119911
3119899= 1199113Re119911
1
study the procedure for calculation of maximal efficiencyof PA with waveform pair (189)ndash(191) is the same as thatpresented in Algorithm 33 In this case study the coefficientsof fundamental and third harmonic of current waveform aregiven by (192)
For the waveform pair (189) and (191) maximal efficiencyof PA as a function of normalized third harmonic impedance1199113119899= 1199113Re119911
1 is presented in Figure 25 Efficiency of 084
is obtained in vicinity of 1199113119899= 23685 (corresponding to 119911
3=
18750) Peak efficiency 120578 asymp 08421 is achieved for voltagewaveform of type (189) with coefficients 119886
1V = minus2radic3 1198863V =
radic39 and 1198871V = 1198873V = 0
In the course of finding power utilization factor noticethat currentwaveformof type (191) attains itsmaximumvaluefor 120579 = 0 Insertion of max[119894(120579)] = 119894(0) = 278 andmax[V(120579)] = 2 for voltage waveform of type (189) into (179)leads to
PUFCase study 74 = 1439120578 (193)
Again the ratio PUF120578 is constant and PUF can be easilycalculated from the corresponding efficiency Accordinglypeak value of PUFCase study 74 asymp 12118 and peak efficiencyare attained for the same voltage waveform
In the first example current waveform (191) and 1199113=
1 minus 11989502 imply that 2712058223V le (1 minus 2119886
3V)3 Then according to
Algorithm 33 voltage waveform of type (189) which providesmaximal efficiency has coefficients 119886
3V = 01026 1198873V =
00205 1198861V = minus11026 and 119887
1V = minus00616 (see Figure 26)Efficiency PUF and normalized third harmonic impedanceare 120578 = 08042 PUF = 11572 and 119911
3119899= 13228 minus 11989502646
respectivelyIn second example current waveform (191) and 119911
3=
15 minus 11989512 imply that 2712058223V gt (1 minus 2119886
3V)3 Then according to
Algorithm 33 voltage waveform of type (189) which providesmaximal efficiency has coefficients 119886
3V = 01540 1198873V =
Mathematical Problems in Engineering 27
2
3
4
1
0
2 3 410
Wav
efor
ms
Angle 120579120587
i(120579)
(120579)
Figure 26 Waveform pair (189) and (191) that provides maximalefficiency for 119911
3= 1 minus 11989502
2
3
4
1
0
2 3 410
Wav
efor
ms
Angle 120579120587
i(120579)
(120579)
Figure 27 Waveform pair (189) and (191) that provides maximalefficiency for 119911
3= 15 minus 11989512
01232 1198861V = minus11255 and 119887
1V = minus02575 (see Figure 27)Efficiency PUF and normalized third harmonic impedanceare 120578 = 08208 PUF = 11812 and 119911
3119899= 19439 minus 11989515552
respectively
8 Conclusion
In this paper we consider a problem of finding general de-scriptions of various classes of nonnegative waveforms withfundamental and 119896th harmonic These classes include non-negative waveforms with at least one zero nonnegative wave-forms with maximal amplitude of fundamental harmonic forprescribed amplitude of 119896th harmonic nonnegative wave-forms withmaximal coefficient of cosine part of fundamentalharmonic for prescribed coefficients of 119896th harmonic andnonnegative cosine waveforms with at least one zero Mainresults are stated in six propositions (Propositions 1 6 918 22 and 26) four corollaries (Corollaries 2ndash5) twentyremarks and three algorithms Four case studies of usageof closed form descriptions of nonnegative waveforms in PAefficiency analysis are considered in detail in Section 7
Appendices
Here we provide a list of finite sums of trigonometric func-tions used in this paper (Appendix A) and brief account ofthe Chebyshev polynomials (Appendix B)
A List of Some Finite Sums ofTrigonometric Functions
Dirichlet kernel (eg see [27]) is as follows
119863119896minus1
(120591) = 1 + 2
119896minus1
sum
119899=1
cos 119899120591 = sin ((2119896 minus 1) 1205912)sin (1205912)
(A1)
Fejer kernel (eg see [27]) can be expressed in the fol-lowing equivalent forms
119865119896minus1
(120591) =1
119896
119896minus1
sum
119899=0
119863119899(120591) = 1 +
2
119896
119896minus1
sum
119899=1
(119896 minus 119899) cos 119899120591
=(1 minus cos 119896120591)119896 (1 minus cos 120591)
(A2)
Lagrangersquos trigonometric identity (eg see [28]) is as fol-lows
1198781(120591) =
119896minus1
sum
119899=1
sin 119899120591 = sin (1198961205912) sin ((119896 minus 1) 1205912)sin (1205912)
(A3)
In what follows we show that the following three trigono-metric identities also hold
2
119896minus1
sum
119899=1
(119896 minus 119899) sin 119899120591 = 119896 sin 120591 minus sin 1198961205911 minus cos 120591
(A4)
119896minus1
sum
119899=1
cos (119896 minus 2119899) 120591 = sin (119896 minus 1) 120591sin 120591
(A5)
119896minus1
sum
119899=1
119899 (119896 minus 119899) cos (119896 minus 2119899) 120591
=sin (119896120591) cos 120591 minus 119896 cos (119896120591) sin 120591
2sin3120591
(A6)
Denote 1198782(120591) = 2sum
119896minus1
119899=1(119896 minus 119899) sin 119899120591 119878
3(120591) = sum
119896minus1
119899=1cos(119896 minus
2119899)120591 and 1198784(120591) = sum
119896minus1
119899=1119899(119896 minus 119899) cos(119896 minus 2119899)120591
Notice that 1198782(120591) = 2119896119878
1(120591) + 119889119863
119896minus1(120591)119889120591 which
immediately leads to (A4)
28 Mathematical Problems in Engineering
Identity (A5) can be obtained as follows
sin (119896 minus 1) 120591sin 120591
=119890119895(119896minus1)120591
minus 119890minus119895(119896minus1)120591
119890119895120591 minus 119890minus119895120591
= 119890119895119896120591
119890minus2119895120591
minus 119890minus2119895(119896minus1)120591
1 minus 119890minus2119895120591
= 119890119895119896120591
119896minus1
sum
119899=1
119890minus2119895119899120591
=
119896minus1
sum
119899=1
119890119895(119896minus2119899)120591
=
119896minus1
sum
119899=1
cos (119896 minus 2119899) 120591
(A7)
From 4119899(119896 minus 119899) = 1198962
minus (119896 minus 2119899)2 it follows that 4119878
4(120591) =
1198962
1198783(120591) + 119889
2
1198783(120591)119889120591
2 which leads to (A6)
B The Chebyshev Polynomials
The Chebyshev polynomials of the first kind 119881119899(119909) can be
defined by the following relation (eg see [29])
119881119899(119909) = cos 119899120591 when 119909 = cos 120591 (B1)
The Chebyshev polynomials of the second kind 119880119899(119909) can be
defined by the following relation (eg see [29])
119880119899(119909) =
sin (119899 + 1) 120591sin 120591
when 119909 = cos 120591 (B2)
The Chebyshev polynomials satisfy the following recur-rence relations (eg see [29])
1198810(119909) = 1 119881
1(119909) = 119909
119881119899+1
(119909) = 2119909119881119899(119909) minus 119881
119899minus1(119909)
1198800(119909) = 1 119880
1(119909) = 2119909
119880119899+1
(119909) = 2119909119880119899(119909) minus 119880
119899minus1(119909)
(B3)
The first few Chebyshev polynomials of the first and secondkind are 119881
2(119909) = 2119909
2
minus 1 1198813(119909) = 4119909
3
minus 3119909 1198814(119909) = 8119909
4
minus
81199092
+ 1 1198802(119909) = 4119909
2
minus 1 1198803(119909) = 8119909
3
minus 4119909 and 1198804(119909) =
161199094
minus 121199092
+ 1
Conflict of Interests
The authors declare that there is no conflict of interestsregarding the publication of this paper
Acknowledgment
This work is supported by the SerbianMinistry of EducationScience and Technology Development as a part of ProjectTP32016
References
[1] V I Arnolrsquod V S Afrajmovich Y S Ilrsquoyashenko and L PShilrsquonikov Dynamical Systems V Bifurcation Theory and Catas-trophe Theory Springer Berlin Germany 1994
[2] E Polak ldquoOn the mathematical foundations of nondifferen-tiable optimization in engineering designrdquo SIAM Review vol29 no 1 pp 21ndash89 1987
[3] N S Fuzik ldquoBiharmonic modes of a tuned RF power amplifierrdquoRadiotehnika vol 25 no 7 pp 62ndash71 1970 (Russian)
[4] P Colantonio F Giannini G Leuzzi and E Limiti ldquoClass Gapproach for low-voltage high-efficiency PA designrdquo Interna-tional Journal of RF and Microwave Computer-Aided Engineer-ing vol 10 no 6 pp 366ndash378 2000
[5] F H Raab ldquoMaximum efficiency and output of class-F poweramplifiersrdquo IEEE Transactions on Microwave Theory and Tech-niques vol 49 no 6 pp 1162ndash1166 2001
[6] FH Raab ldquoClass-E class-C and class-F power amplifiers basedupon a finite number of harmonicsrdquo IEEE Transactions onMicrowaveTheory and Techniques vol 49 no 8 pp 1462ndash14682001
[7] J D Rhodes ldquoOutput universality inmaximum efficiency linearpower amplifiersrdquo International Journal of Circuit Theory andApplications vol 31 no 4 pp 385ndash405 2003
[8] A Juhas and L A Novak ldquoComments on lsquoClass-E class-Cand classF power amplifier based upon a finite number ofharmonicsrsquordquo IEEE Transactions on Microwave Theory and Tech-niques vol 57 no 6 pp 1623ndash1625 2009
[9] M Roberg and Z Popovic ldquoAnalysis of high-efficiency poweramplifiers with arbitrary output harmonic terminationsrdquo IEEETransactions on Microwave Theory and Techniques vol 59 no8 pp 2037ndash2048 2011
[10] A Grebennikov N O Sokal and M J Franco Switchmode RFPower Amplifiers ElsevierAcademic Press San Diego CalifUSA 2nd edition 2012
[11] T Canning P J Tasker and S C Cripps ldquoContinuous modepower amplifier design using harmonic clipping contourstheory and practicerdquo IEEE Transactions on Microwave Theoryand Techniques vol 62 no 1 pp 100ndash110 2014
[12] A Juhas and L A Novak ldquoGeneral description of nonnegativewaveforms up to second harmonic for power amplifier mod-ellingrdquoMathematical Problems in Engineering vol 2014 ArticleID 709762 18 pages 2014
[13] V I Arnolrsquod V V Goryunov O V Lyashko and V A VasilrsquoevDynamical Systems VIIImdashSingularity Theory II ApplicationsSpringer Berlin Germany 1993
[14] D Siersma ldquoProperties of conflict sets in the planerdquo BanachCenter Publications Polish Academy of Sciences vol 50 no 1 pp267ndash276 1999 Proceedings of the Banach Center Symposiumon Geometry and Topology of Caustics (Caustics rsquo98) WarsawPoland
[15] M van Manen The geometry of conflict sets [Dissertation]Universiteit Utrecht UtrechtThe Netherlands 2003 httpigi-tur-archivelibraryuunldissertations2003-0912-123058c4pdf
[16] Y L Sachkov ldquoMaxwell strata and symmetries in the problemofoptimal rolling of a sphere over a planerdquo Sbornik Mathematicsvol 201 no 7-8 pp 1029ndash1051 2010
[17] I A Bogaevsky ldquoPerestroikas of shocks and singularities ofminimum functionsrdquoPhysicaDNonlinear Phenomena vol 173no 1-2 pp 1ndash28 2002
[18] Y L Sachkov ldquoMaxwell strata in the Euler elastic problemrdquoJournal of Dynamical andControl Systems vol 14 no 2 pp 169ndash234 2008
[19] M Siino and T Koike ldquoTopological classification of black holesgeneric Maxwell set and crease set of a horizonrdquo InternationalJournal ofModern Physics D Gravitation Astrophysics Cosmol-ogy vol 20 no 6 pp 1095ndash1122 2011
Mathematical Problems in Engineering 29
[20] F H Raab ldquoClass-F power amplifiers with maximally flat wave-formsrdquo IEEETransactions onMicrowaveTheory and Techniquesvol 45 no 11 pp 2007ndash2012 1997
[21] A Juhas and L A Novak ldquoMaximally flat waveforms with finitenumber of harmonics in class-F power amplifiersrdquo Mathemat-ical Problems in Engineering vol 2013 Article ID 169590 9pages 2013
[22] L Fejer ldquoUber trigonometrische polynomerdquo Journal fur dieReine und Angewandte Mathematik vol 1916 no 146 pp 53ndash82 1916 (German)
[23] S C Cripps ldquoBessel Waives [microwave bytes]rdquo IEEE Micro-wave Magazine vol 10 no 7 pp 30ndash36 117 2009
[24] L N Bryzgalova ldquoSingularities of the maximum of para-metrically dependent functionrdquo Functional Analysis and ItsApplications vol 11 no 1 pp 49ndash51 1977
[25] V I Arnold A A Davydov V A Vassiliev and V MZakalyukin Mathematical Models of Catastrophes Control ofCatastrophic Process Encyclopedia of Life Support Systems(EOLSS) EOLSS Publishers Oxford UK 2006
[26] S C Cripps RF Power Amplifiers for Wireless CommunicationsArtech House Norwood Mass USA 2nd edition 2006
[27] A Zygmund Trigonometric Series vol 1 Cambridge UniversityPress Cambridge UK 2nd edition 1959
[28] A Jeffrey and H DaiHandbook of Mathematical Formulas andIntegrals ElsevierAcademic Press San Diego Calif USA 4thedition 2008
[29] J C Mason and D C Handscomb Chebyshev PolynomialsChapman amp Hall CRC Press Boca Raton Fla USA 2003
Submit your manuscripts athttpwwwhindawicom
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical Problems in Engineering
Hindawi Publishing Corporationhttpwwwhindawicom
Differential EquationsInternational Journal of
Volume 2014
Applied MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Probability and StatisticsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical PhysicsAdvances in
Complex AnalysisJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
OptimizationJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
CombinatoricsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Operations ResearchAdvances in
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Function Spaces
Abstract and Applied AnalysisHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of Mathematics and Mathematical Sciences
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Algebra
Discrete Dynamics in Nature and Society
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Decision SciencesAdvances in
Discrete MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom
Volume 2014 Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Stochastic AnalysisInternational Journal of
Mathematical Problems in Engineering 27
2
3
4
1
0
2 3 410
Wav
efor
ms
Angle 120579120587
i(120579)
(120579)
Figure 26 Waveform pair (189) and (191) that provides maximalefficiency for 119911
3= 1 minus 11989502
2
3
4
1
0
2 3 410
Wav
efor
ms
Angle 120579120587
i(120579)
(120579)
Figure 27 Waveform pair (189) and (191) that provides maximalefficiency for 119911
3= 15 minus 11989512
01232 1198861V = minus11255 and 119887
1V = minus02575 (see Figure 27)Efficiency PUF and normalized third harmonic impedanceare 120578 = 08208 PUF = 11812 and 119911
3119899= 19439 minus 11989515552
respectively
8 Conclusion
In this paper we consider a problem of finding general de-scriptions of various classes of nonnegative waveforms withfundamental and 119896th harmonic These classes include non-negative waveforms with at least one zero nonnegative wave-forms with maximal amplitude of fundamental harmonic forprescribed amplitude of 119896th harmonic nonnegative wave-forms withmaximal coefficient of cosine part of fundamentalharmonic for prescribed coefficients of 119896th harmonic andnonnegative cosine waveforms with at least one zero Mainresults are stated in six propositions (Propositions 1 6 918 22 and 26) four corollaries (Corollaries 2ndash5) twentyremarks and three algorithms Four case studies of usageof closed form descriptions of nonnegative waveforms in PAefficiency analysis are considered in detail in Section 7
Appendices
Here we provide a list of finite sums of trigonometric func-tions used in this paper (Appendix A) and brief account ofthe Chebyshev polynomials (Appendix B)
A List of Some Finite Sums ofTrigonometric Functions
Dirichlet kernel (eg see [27]) is as follows
119863119896minus1
(120591) = 1 + 2
119896minus1
sum
119899=1
cos 119899120591 = sin ((2119896 minus 1) 1205912)sin (1205912)
(A1)
Fejer kernel (eg see [27]) can be expressed in the fol-lowing equivalent forms
119865119896minus1
(120591) =1
119896
119896minus1
sum
119899=0
119863119899(120591) = 1 +
2
119896
119896minus1
sum
119899=1
(119896 minus 119899) cos 119899120591
=(1 minus cos 119896120591)119896 (1 minus cos 120591)
(A2)
Lagrangersquos trigonometric identity (eg see [28]) is as fol-lows
1198781(120591) =
119896minus1
sum
119899=1
sin 119899120591 = sin (1198961205912) sin ((119896 minus 1) 1205912)sin (1205912)
(A3)
In what follows we show that the following three trigono-metric identities also hold
2
119896minus1
sum
119899=1
(119896 minus 119899) sin 119899120591 = 119896 sin 120591 minus sin 1198961205911 minus cos 120591
(A4)
119896minus1
sum
119899=1
cos (119896 minus 2119899) 120591 = sin (119896 minus 1) 120591sin 120591
(A5)
119896minus1
sum
119899=1
119899 (119896 minus 119899) cos (119896 minus 2119899) 120591
=sin (119896120591) cos 120591 minus 119896 cos (119896120591) sin 120591
2sin3120591
(A6)
Denote 1198782(120591) = 2sum
119896minus1
119899=1(119896 minus 119899) sin 119899120591 119878
3(120591) = sum
119896minus1
119899=1cos(119896 minus
2119899)120591 and 1198784(120591) = sum
119896minus1
119899=1119899(119896 minus 119899) cos(119896 minus 2119899)120591
Notice that 1198782(120591) = 2119896119878
1(120591) + 119889119863
119896minus1(120591)119889120591 which
immediately leads to (A4)
28 Mathematical Problems in Engineering
Identity (A5) can be obtained as follows
sin (119896 minus 1) 120591sin 120591
=119890119895(119896minus1)120591
minus 119890minus119895(119896minus1)120591
119890119895120591 minus 119890minus119895120591
= 119890119895119896120591
119890minus2119895120591
minus 119890minus2119895(119896minus1)120591
1 minus 119890minus2119895120591
= 119890119895119896120591
119896minus1
sum
119899=1
119890minus2119895119899120591
=
119896minus1
sum
119899=1
119890119895(119896minus2119899)120591
=
119896minus1
sum
119899=1
cos (119896 minus 2119899) 120591
(A7)
From 4119899(119896 minus 119899) = 1198962
minus (119896 minus 2119899)2 it follows that 4119878
4(120591) =
1198962
1198783(120591) + 119889
2
1198783(120591)119889120591
2 which leads to (A6)
B The Chebyshev Polynomials
The Chebyshev polynomials of the first kind 119881119899(119909) can be
defined by the following relation (eg see [29])
119881119899(119909) = cos 119899120591 when 119909 = cos 120591 (B1)
The Chebyshev polynomials of the second kind 119880119899(119909) can be
defined by the following relation (eg see [29])
119880119899(119909) =
sin (119899 + 1) 120591sin 120591
when 119909 = cos 120591 (B2)
The Chebyshev polynomials satisfy the following recur-rence relations (eg see [29])
1198810(119909) = 1 119881
1(119909) = 119909
119881119899+1
(119909) = 2119909119881119899(119909) minus 119881
119899minus1(119909)
1198800(119909) = 1 119880
1(119909) = 2119909
119880119899+1
(119909) = 2119909119880119899(119909) minus 119880
119899minus1(119909)
(B3)
The first few Chebyshev polynomials of the first and secondkind are 119881
2(119909) = 2119909
2
minus 1 1198813(119909) = 4119909
3
minus 3119909 1198814(119909) = 8119909
4
minus
81199092
+ 1 1198802(119909) = 4119909
2
minus 1 1198803(119909) = 8119909
3
minus 4119909 and 1198804(119909) =
161199094
minus 121199092
+ 1
Conflict of Interests
The authors declare that there is no conflict of interestsregarding the publication of this paper
Acknowledgment
This work is supported by the SerbianMinistry of EducationScience and Technology Development as a part of ProjectTP32016
References
[1] V I Arnolrsquod V S Afrajmovich Y S Ilrsquoyashenko and L PShilrsquonikov Dynamical Systems V Bifurcation Theory and Catas-trophe Theory Springer Berlin Germany 1994
[2] E Polak ldquoOn the mathematical foundations of nondifferen-tiable optimization in engineering designrdquo SIAM Review vol29 no 1 pp 21ndash89 1987
[3] N S Fuzik ldquoBiharmonic modes of a tuned RF power amplifierrdquoRadiotehnika vol 25 no 7 pp 62ndash71 1970 (Russian)
[4] P Colantonio F Giannini G Leuzzi and E Limiti ldquoClass Gapproach for low-voltage high-efficiency PA designrdquo Interna-tional Journal of RF and Microwave Computer-Aided Engineer-ing vol 10 no 6 pp 366ndash378 2000
[5] F H Raab ldquoMaximum efficiency and output of class-F poweramplifiersrdquo IEEE Transactions on Microwave Theory and Tech-niques vol 49 no 6 pp 1162ndash1166 2001
[6] FH Raab ldquoClass-E class-C and class-F power amplifiers basedupon a finite number of harmonicsrdquo IEEE Transactions onMicrowaveTheory and Techniques vol 49 no 8 pp 1462ndash14682001
[7] J D Rhodes ldquoOutput universality inmaximum efficiency linearpower amplifiersrdquo International Journal of Circuit Theory andApplications vol 31 no 4 pp 385ndash405 2003
[8] A Juhas and L A Novak ldquoComments on lsquoClass-E class-Cand classF power amplifier based upon a finite number ofharmonicsrsquordquo IEEE Transactions on Microwave Theory and Tech-niques vol 57 no 6 pp 1623ndash1625 2009
[9] M Roberg and Z Popovic ldquoAnalysis of high-efficiency poweramplifiers with arbitrary output harmonic terminationsrdquo IEEETransactions on Microwave Theory and Techniques vol 59 no8 pp 2037ndash2048 2011
[10] A Grebennikov N O Sokal and M J Franco Switchmode RFPower Amplifiers ElsevierAcademic Press San Diego CalifUSA 2nd edition 2012
[11] T Canning P J Tasker and S C Cripps ldquoContinuous modepower amplifier design using harmonic clipping contourstheory and practicerdquo IEEE Transactions on Microwave Theoryand Techniques vol 62 no 1 pp 100ndash110 2014
[12] A Juhas and L A Novak ldquoGeneral description of nonnegativewaveforms up to second harmonic for power amplifier mod-ellingrdquoMathematical Problems in Engineering vol 2014 ArticleID 709762 18 pages 2014
[13] V I Arnolrsquod V V Goryunov O V Lyashko and V A VasilrsquoevDynamical Systems VIIImdashSingularity Theory II ApplicationsSpringer Berlin Germany 1993
[14] D Siersma ldquoProperties of conflict sets in the planerdquo BanachCenter Publications Polish Academy of Sciences vol 50 no 1 pp267ndash276 1999 Proceedings of the Banach Center Symposiumon Geometry and Topology of Caustics (Caustics rsquo98) WarsawPoland
[15] M van Manen The geometry of conflict sets [Dissertation]Universiteit Utrecht UtrechtThe Netherlands 2003 httpigi-tur-archivelibraryuunldissertations2003-0912-123058c4pdf
[16] Y L Sachkov ldquoMaxwell strata and symmetries in the problemofoptimal rolling of a sphere over a planerdquo Sbornik Mathematicsvol 201 no 7-8 pp 1029ndash1051 2010
[17] I A Bogaevsky ldquoPerestroikas of shocks and singularities ofminimum functionsrdquoPhysicaDNonlinear Phenomena vol 173no 1-2 pp 1ndash28 2002
[18] Y L Sachkov ldquoMaxwell strata in the Euler elastic problemrdquoJournal of Dynamical andControl Systems vol 14 no 2 pp 169ndash234 2008
[19] M Siino and T Koike ldquoTopological classification of black holesgeneric Maxwell set and crease set of a horizonrdquo InternationalJournal ofModern Physics D Gravitation Astrophysics Cosmol-ogy vol 20 no 6 pp 1095ndash1122 2011
Mathematical Problems in Engineering 29
[20] F H Raab ldquoClass-F power amplifiers with maximally flat wave-formsrdquo IEEETransactions onMicrowaveTheory and Techniquesvol 45 no 11 pp 2007ndash2012 1997
[21] A Juhas and L A Novak ldquoMaximally flat waveforms with finitenumber of harmonics in class-F power amplifiersrdquo Mathemat-ical Problems in Engineering vol 2013 Article ID 169590 9pages 2013
[22] L Fejer ldquoUber trigonometrische polynomerdquo Journal fur dieReine und Angewandte Mathematik vol 1916 no 146 pp 53ndash82 1916 (German)
[23] S C Cripps ldquoBessel Waives [microwave bytes]rdquo IEEE Micro-wave Magazine vol 10 no 7 pp 30ndash36 117 2009
[24] L N Bryzgalova ldquoSingularities of the maximum of para-metrically dependent functionrdquo Functional Analysis and ItsApplications vol 11 no 1 pp 49ndash51 1977
[25] V I Arnold A A Davydov V A Vassiliev and V MZakalyukin Mathematical Models of Catastrophes Control ofCatastrophic Process Encyclopedia of Life Support Systems(EOLSS) EOLSS Publishers Oxford UK 2006
[26] S C Cripps RF Power Amplifiers for Wireless CommunicationsArtech House Norwood Mass USA 2nd edition 2006
[27] A Zygmund Trigonometric Series vol 1 Cambridge UniversityPress Cambridge UK 2nd edition 1959
[28] A Jeffrey and H DaiHandbook of Mathematical Formulas andIntegrals ElsevierAcademic Press San Diego Calif USA 4thedition 2008
[29] J C Mason and D C Handscomb Chebyshev PolynomialsChapman amp Hall CRC Press Boca Raton Fla USA 2003
Submit your manuscripts athttpwwwhindawicom
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical Problems in Engineering
Hindawi Publishing Corporationhttpwwwhindawicom
Differential EquationsInternational Journal of
Volume 2014
Applied MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Probability and StatisticsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical PhysicsAdvances in
Complex AnalysisJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
OptimizationJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
CombinatoricsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Operations ResearchAdvances in
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Function Spaces
Abstract and Applied AnalysisHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of Mathematics and Mathematical Sciences
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Algebra
Discrete Dynamics in Nature and Society
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Decision SciencesAdvances in
Discrete MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom
Volume 2014 Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Stochastic AnalysisInternational Journal of
28 Mathematical Problems in Engineering
Identity (A5) can be obtained as follows
sin (119896 minus 1) 120591sin 120591
=119890119895(119896minus1)120591
minus 119890minus119895(119896minus1)120591
119890119895120591 minus 119890minus119895120591
= 119890119895119896120591
119890minus2119895120591
minus 119890minus2119895(119896minus1)120591
1 minus 119890minus2119895120591
= 119890119895119896120591
119896minus1
sum
119899=1
119890minus2119895119899120591
=
119896minus1
sum
119899=1
119890119895(119896minus2119899)120591
=
119896minus1
sum
119899=1
cos (119896 minus 2119899) 120591
(A7)
From 4119899(119896 minus 119899) = 1198962
minus (119896 minus 2119899)2 it follows that 4119878
4(120591) =
1198962
1198783(120591) + 119889
2
1198783(120591)119889120591
2 which leads to (A6)
B The Chebyshev Polynomials
The Chebyshev polynomials of the first kind 119881119899(119909) can be
defined by the following relation (eg see [29])
119881119899(119909) = cos 119899120591 when 119909 = cos 120591 (B1)
The Chebyshev polynomials of the second kind 119880119899(119909) can be
defined by the following relation (eg see [29])
119880119899(119909) =
sin (119899 + 1) 120591sin 120591
when 119909 = cos 120591 (B2)
The Chebyshev polynomials satisfy the following recur-rence relations (eg see [29])
1198810(119909) = 1 119881
1(119909) = 119909
119881119899+1
(119909) = 2119909119881119899(119909) minus 119881
119899minus1(119909)
1198800(119909) = 1 119880
1(119909) = 2119909
119880119899+1
(119909) = 2119909119880119899(119909) minus 119880
119899minus1(119909)
(B3)
The first few Chebyshev polynomials of the first and secondkind are 119881
2(119909) = 2119909
2
minus 1 1198813(119909) = 4119909
3
minus 3119909 1198814(119909) = 8119909
4
minus
81199092
+ 1 1198802(119909) = 4119909
2
minus 1 1198803(119909) = 8119909
3
minus 4119909 and 1198804(119909) =
161199094
minus 121199092
+ 1
Conflict of Interests
The authors declare that there is no conflict of interestsregarding the publication of this paper
Acknowledgment
This work is supported by the SerbianMinistry of EducationScience and Technology Development as a part of ProjectTP32016
References
[1] V I Arnolrsquod V S Afrajmovich Y S Ilrsquoyashenko and L PShilrsquonikov Dynamical Systems V Bifurcation Theory and Catas-trophe Theory Springer Berlin Germany 1994
[2] E Polak ldquoOn the mathematical foundations of nondifferen-tiable optimization in engineering designrdquo SIAM Review vol29 no 1 pp 21ndash89 1987
[3] N S Fuzik ldquoBiharmonic modes of a tuned RF power amplifierrdquoRadiotehnika vol 25 no 7 pp 62ndash71 1970 (Russian)
[4] P Colantonio F Giannini G Leuzzi and E Limiti ldquoClass Gapproach for low-voltage high-efficiency PA designrdquo Interna-tional Journal of RF and Microwave Computer-Aided Engineer-ing vol 10 no 6 pp 366ndash378 2000
[5] F H Raab ldquoMaximum efficiency and output of class-F poweramplifiersrdquo IEEE Transactions on Microwave Theory and Tech-niques vol 49 no 6 pp 1162ndash1166 2001
[6] FH Raab ldquoClass-E class-C and class-F power amplifiers basedupon a finite number of harmonicsrdquo IEEE Transactions onMicrowaveTheory and Techniques vol 49 no 8 pp 1462ndash14682001
[7] J D Rhodes ldquoOutput universality inmaximum efficiency linearpower amplifiersrdquo International Journal of Circuit Theory andApplications vol 31 no 4 pp 385ndash405 2003
[8] A Juhas and L A Novak ldquoComments on lsquoClass-E class-Cand classF power amplifier based upon a finite number ofharmonicsrsquordquo IEEE Transactions on Microwave Theory and Tech-niques vol 57 no 6 pp 1623ndash1625 2009
[9] M Roberg and Z Popovic ldquoAnalysis of high-efficiency poweramplifiers with arbitrary output harmonic terminationsrdquo IEEETransactions on Microwave Theory and Techniques vol 59 no8 pp 2037ndash2048 2011
[10] A Grebennikov N O Sokal and M J Franco Switchmode RFPower Amplifiers ElsevierAcademic Press San Diego CalifUSA 2nd edition 2012
[11] T Canning P J Tasker and S C Cripps ldquoContinuous modepower amplifier design using harmonic clipping contourstheory and practicerdquo IEEE Transactions on Microwave Theoryand Techniques vol 62 no 1 pp 100ndash110 2014
[12] A Juhas and L A Novak ldquoGeneral description of nonnegativewaveforms up to second harmonic for power amplifier mod-ellingrdquoMathematical Problems in Engineering vol 2014 ArticleID 709762 18 pages 2014
[13] V I Arnolrsquod V V Goryunov O V Lyashko and V A VasilrsquoevDynamical Systems VIIImdashSingularity Theory II ApplicationsSpringer Berlin Germany 1993
[14] D Siersma ldquoProperties of conflict sets in the planerdquo BanachCenter Publications Polish Academy of Sciences vol 50 no 1 pp267ndash276 1999 Proceedings of the Banach Center Symposiumon Geometry and Topology of Caustics (Caustics rsquo98) WarsawPoland
[15] M van Manen The geometry of conflict sets [Dissertation]Universiteit Utrecht UtrechtThe Netherlands 2003 httpigi-tur-archivelibraryuunldissertations2003-0912-123058c4pdf
[16] Y L Sachkov ldquoMaxwell strata and symmetries in the problemofoptimal rolling of a sphere over a planerdquo Sbornik Mathematicsvol 201 no 7-8 pp 1029ndash1051 2010
[17] I A Bogaevsky ldquoPerestroikas of shocks and singularities ofminimum functionsrdquoPhysicaDNonlinear Phenomena vol 173no 1-2 pp 1ndash28 2002
[18] Y L Sachkov ldquoMaxwell strata in the Euler elastic problemrdquoJournal of Dynamical andControl Systems vol 14 no 2 pp 169ndash234 2008
[19] M Siino and T Koike ldquoTopological classification of black holesgeneric Maxwell set and crease set of a horizonrdquo InternationalJournal ofModern Physics D Gravitation Astrophysics Cosmol-ogy vol 20 no 6 pp 1095ndash1122 2011
Mathematical Problems in Engineering 29
[20] F H Raab ldquoClass-F power amplifiers with maximally flat wave-formsrdquo IEEETransactions onMicrowaveTheory and Techniquesvol 45 no 11 pp 2007ndash2012 1997
[21] A Juhas and L A Novak ldquoMaximally flat waveforms with finitenumber of harmonics in class-F power amplifiersrdquo Mathemat-ical Problems in Engineering vol 2013 Article ID 169590 9pages 2013
[22] L Fejer ldquoUber trigonometrische polynomerdquo Journal fur dieReine und Angewandte Mathematik vol 1916 no 146 pp 53ndash82 1916 (German)
[23] S C Cripps ldquoBessel Waives [microwave bytes]rdquo IEEE Micro-wave Magazine vol 10 no 7 pp 30ndash36 117 2009
[24] L N Bryzgalova ldquoSingularities of the maximum of para-metrically dependent functionrdquo Functional Analysis and ItsApplications vol 11 no 1 pp 49ndash51 1977
[25] V I Arnold A A Davydov V A Vassiliev and V MZakalyukin Mathematical Models of Catastrophes Control ofCatastrophic Process Encyclopedia of Life Support Systems(EOLSS) EOLSS Publishers Oxford UK 2006
[26] S C Cripps RF Power Amplifiers for Wireless CommunicationsArtech House Norwood Mass USA 2nd edition 2006
[27] A Zygmund Trigonometric Series vol 1 Cambridge UniversityPress Cambridge UK 2nd edition 1959
[28] A Jeffrey and H DaiHandbook of Mathematical Formulas andIntegrals ElsevierAcademic Press San Diego Calif USA 4thedition 2008
[29] J C Mason and D C Handscomb Chebyshev PolynomialsChapman amp Hall CRC Press Boca Raton Fla USA 2003
Submit your manuscripts athttpwwwhindawicom
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical Problems in Engineering
Hindawi Publishing Corporationhttpwwwhindawicom
Differential EquationsInternational Journal of
Volume 2014
Applied MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Probability and StatisticsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical PhysicsAdvances in
Complex AnalysisJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
OptimizationJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
CombinatoricsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Operations ResearchAdvances in
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Function Spaces
Abstract and Applied AnalysisHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of Mathematics and Mathematical Sciences
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Algebra
Discrete Dynamics in Nature and Society
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Decision SciencesAdvances in
Discrete MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom
Volume 2014 Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Stochastic AnalysisInternational Journal of
Mathematical Problems in Engineering 29
[20] F H Raab ldquoClass-F power amplifiers with maximally flat wave-formsrdquo IEEETransactions onMicrowaveTheory and Techniquesvol 45 no 11 pp 2007ndash2012 1997
[21] A Juhas and L A Novak ldquoMaximally flat waveforms with finitenumber of harmonics in class-F power amplifiersrdquo Mathemat-ical Problems in Engineering vol 2013 Article ID 169590 9pages 2013
[22] L Fejer ldquoUber trigonometrische polynomerdquo Journal fur dieReine und Angewandte Mathematik vol 1916 no 146 pp 53ndash82 1916 (German)
[23] S C Cripps ldquoBessel Waives [microwave bytes]rdquo IEEE Micro-wave Magazine vol 10 no 7 pp 30ndash36 117 2009
[24] L N Bryzgalova ldquoSingularities of the maximum of para-metrically dependent functionrdquo Functional Analysis and ItsApplications vol 11 no 1 pp 49ndash51 1977
[25] V I Arnold A A Davydov V A Vassiliev and V MZakalyukin Mathematical Models of Catastrophes Control ofCatastrophic Process Encyclopedia of Life Support Systems(EOLSS) EOLSS Publishers Oxford UK 2006
[26] S C Cripps RF Power Amplifiers for Wireless CommunicationsArtech House Norwood Mass USA 2nd edition 2006
[27] A Zygmund Trigonometric Series vol 1 Cambridge UniversityPress Cambridge UK 2nd edition 1959
[28] A Jeffrey and H DaiHandbook of Mathematical Formulas andIntegrals ElsevierAcademic Press San Diego Calif USA 4thedition 2008
[29] J C Mason and D C Handscomb Chebyshev PolynomialsChapman amp Hall CRC Press Boca Raton Fla USA 2003
Submit your manuscripts athttpwwwhindawicom
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical Problems in Engineering
Hindawi Publishing Corporationhttpwwwhindawicom
Differential EquationsInternational Journal of
Volume 2014
Applied MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Probability and StatisticsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical PhysicsAdvances in
Complex AnalysisJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
OptimizationJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
CombinatoricsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Operations ResearchAdvances in
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Function Spaces
Abstract and Applied AnalysisHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of Mathematics and Mathematical Sciences
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Algebra
Discrete Dynamics in Nature and Society
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Decision SciencesAdvances in
Discrete MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom
Volume 2014 Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Stochastic AnalysisInternational Journal of
Submit your manuscripts athttpwwwhindawicom
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical Problems in Engineering
Hindawi Publishing Corporationhttpwwwhindawicom
Differential EquationsInternational Journal of
Volume 2014
Applied MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Probability and StatisticsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical PhysicsAdvances in
Complex AnalysisJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
OptimizationJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
CombinatoricsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Operations ResearchAdvances in
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Function Spaces
Abstract and Applied AnalysisHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of Mathematics and Mathematical Sciences
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Algebra
Discrete Dynamics in Nature and Society
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Decision SciencesAdvances in
Discrete MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom
Volume 2014 Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Stochastic AnalysisInternational Journal of