analysis, simulation and experimental study of chaotic behaviour in parallel-connected dc–dc boost...
TRANSCRIPT
Chaos, Solitons and Fractals 39 (2009) 2465–2476
www.elsevier.com/locate/chaos
Analysis, simulation and experimental study of chaoticbehaviour in parallel-connected DC–DC boost converters
Ammar N. Natsheh a, J. Gordon Kettleborough a, Jamal M. Nazzal b,*
a Department of Electronic and Electrical Engineering, Loughborough University, Loughborough, Leicestershire LE11 3TU, UKb Faculty of Engineering, Al-Ahliyya Amman University, Post Code 19328 Amman, Jordan
Accepted 5 July 2007
Communicated by Professor G. Iovane
Abstract
The paper describes an experimental study of the bifurcation behaviour of a modular peak current-mode controlledDC–DC boost converter. The parallel-input/parallel-output converter comprises two identical boost circuits and oper-ates in the continuous-current conduction mode. A comparison is made between the results obtained from an experi-mental converter with those obtained from bifurcation diagrams generated from previous work and waveforms from anew MATLAB/SIMULINK simulation presented in this paper. Another comparison is made between the modularconverter diagrams with those of the single boost converter.� 2007 Elsevier Ltd. All rights reserved.
1. Introduction
DC–DC switch-mode converters are non-linear devices and recently it has been observed that they may behave in achaotic manner. This has generated much interest, particularly in single-stage topologies such as the buck, boost, andbuck-boost converters [1–7].
The connection of several DC–DC converters in parallel, with the load shared between modules, reduces currentstress on the switching devices and increases system reliability. However, despite the growing popularity of these mod-ular converters, their bifurcation phenomena have rarely been studied. The work presented in [8] describes the bifur-cation phenomena in a parallel system of two boost converters under a master-slave current-sharing control scheme.One converter has voltage feedback control, while the other has an additional inner current loop which adjusts the volt-age feedback loop, to ensure equal sharing of the load current.
This paper investigates chaotic behaviour in a two-module parallel input/parallel-output boost converter operatingunder a peak current-mode control scheme, where each module has its own current feedback loop. The converterconsists of two identical boost circuits and operates in the continuous-current conduction mode. A comparison ismade between waveforms obtained from an experimental converter and both a MATLAB/SIMULINK model
0960-0779/$ - see front matter � 2007 Elsevier Ltd. All rights reserved.doi:10.1016/j.chaos.2007.07.020
* Corresponding author. Tel.: +962 79 5261 490; fax: +962 65 3351 69.E-mail address: [email protected] (J.M. Nazzal).
2466 A.N. Natsheh et al. / Chaos, Solitons and Fractals 39 (2009) 2465–2476
and results obtained from bifurcation diagrams generated from a nonlinear mapping in closed form usingMATLAB [9].
2. Bifurcation and chaos theory
Bifurcation theory, originally developed by Poincare, is used to indicate the qualitative change in behavior of a sys-tem in terms of the number and the type of solutions, under the variation of one or more parameters on which the sys-tem depends [10].
In bifurcation problems, in addition to state variables, there are control parameters. The relationship between any ofthese controls parameters and any state variable is called the state-control space. In this space, locations at which bifur-cations occur are called bifurcation points. Bifurcations of an equilibrium or fixed-point solution are classified as either
Fig. 1. Simplified circuit diagram for the two-module converter.
Fig. 2. Module current voltage waveforms for the circuit of Fig. 1.
A.N. Natsheh et al. / Chaos, Solitons and Fractals 39 (2009) 2465–2476 2467
static bifurcations, such as saddle-node, pitch fork, or transcritical bifurcations; or as dynamic bifurcations which arealso known as the Hopf bifurcation that exhibits periodic solutions. For the fixed-point solutions, the local stabilityof the system is determined from the eigenvalues of the Jacobian matrix of the linearized system. On the other hand,with periodic-solutions, the system stability depends on what is known as the Floquet theory and the eigenvalues of theMonodromy matrix that are known in the literature as Floquet or characteristic multipliers. The types of bifurcationare determined from the manner in which the Floquet multipliers leave the unit circle. There are three possible ways forthis to happen [10]:
(i) If the Floquet multiplier leaves the unit circle through +1, then three possible bifurcations may occur: transcrit-ical, symmetry-breaking, or cyclic-fold bifurcation.
(ii) If the Floquet multiplier leaves through �1, the period-doubling (Flip bifurcation) occurs.(iii) If the Floquet multipliers are complex conjugate and leave the unit circle from the real axis, the system exhibits
secondary Hopf bifurcation.
To observe the system dynamics under all the above possible bifurcations, a complete diagram known as a bifurca-
tion diagram is constructed. A bifurcation diagram shows the variation of one of the state variables with one of thesystem parameters, otherwise known as a control parameter.
3. Mathematical modelling and numerical analysis
A simplified diagram for the proposed converter is shown in Fig. 1. It consists of two peak current-mode controlledDC–DC boost converters whose outputs are connected in parallel to feed a common resistive load. Each converter hasits own current feedback loop comprising a comparator and a flip-flop. Each comparator compares its respective peakinductor current with a reference value, to determine the on-time of the switch.
The current and voltage waveforms for the circuit of Fig. 1 are shown in Fig. 2. Switches S1 and S2 are controlledsuch that they close at the same time. When they are closed, diodes D1 and D2 are non-conducting, and inductor
Fig. 3. Circuit diagram for the two-module converter with S1, S2 closed and D1, D2 open.
Fig. 4. Circuit diagram of the two-module converter with S1, S2 open and D1, D2 conducting.
2468 A.N. Natsheh et al. / Chaos, Solitons and Fractals 39 (2009) 2465–2476
currents i1 and i2 increase linearly. S1 and S2 open respectively when i1 = Iref and i2 = Iref, whereupon D1 and D2 con-duct and the inductor currents decay linearly. S1 and S2 close again when the next clock pulse arrives.
The iterative map describing the system takes the form:
TableThe se
Circui
SwitchInputInductInductCapacCapacLoad
xnþ1 ¼ f ðxn; I refÞ ð1Þ
where subscript n denotes the value at the beginning of the nth cycle and x is the state vector
1t of circuit parameter
t components Values
ing period T 100 lsvoltage VI 10 Vance L1 1 mHance L2 1 mHitance C1 10 lFitance C2 10 lFresistance R 20 X
Fig. 5. Bifurcation diagram of i1 for the two-module converter.
Fig. 6. Bifurcation diagram of i2 for the two-module converter.
A.N. Natsheh et al. / Chaos, Solitons and Fractals 39 (2009) 2465–2476 2469
x ¼
i1
vc1
i2
vc2
26664
37775 ð2Þ
with i1, i2, the currents through inductors L1 and L2, respectively, vc1, vc2, the voltages across capacitors C1 and C2,respectively, and vc1, vc2, considered as separate states for clarity.
vC2
vC1
iL2
iL1
10
Vin
Cl
Error
Memory
U
Truth Table2
Cl
Error
Memory
U
Truth Table1
PulseGenerator
Memory2
Memory1
5
Iref
1s
Integrator4
1s
Integrator3
1s
Integrator2
1s
Integrator1
1
Constant2
1
Constant1
Add8
Add7
Add6
Add5
Add4
Add3
Add2
Add1
Add
-K-
1/R2
-K-
1/R1
-K-
1/L2
-K-
1/L1
-K-
1/2*C2
-K-
1/2*C1
(1-u)*vC2
(1-u)*vC1
(1-u)*(iL2+iL1)
(1-u)*(iL1+iL2)
Fig. 8. MATLAB/SIMULINK module for the two-module converter.
Fig. 7. Bifurcation diagram for a single boost converter with Iref as the bifurcation parameter.
2470 A.N. Natsheh et al. / Chaos, Solitons and Fractals 39 (2009) 2465–2476
3.1. S1 and S2 closed
Diodes D1 and D2 block at t = 0 and the circuit is as shown in Fig. 3. Assuming that L1 = L2 = L and C1 = C2 = C,the circuit can be described by the following set of first-order differential equations:
di1
dt¼ di2
dt¼ V I
Lð3Þ
dvc1
dt¼ � 1
2RCvc1 ð4Þ
dvc2
dt¼ � 1
2RCvc2 ð5Þ
Assuming that initially the module inductor current is in and the capacitor voltage is vn, and that the inductor current isIref at time tn, the solution of these equations gives
tn ¼L I ref � inð Þ
V Ið6Þ
vc1ðtÞ ¼ vc2ðtÞ ¼ vn e�2kt ð7Þ
where
k ¼ 1
4RCð8Þ
and tn is the time at which switches S1 and S2 open.
8.5 9 9.5 10
x 10-3
-2
-1.5
-1
-0.5
0
0.5
1
1.5
2
time (sec)
iL (
A)
15.8 16 16.2 16.4 16.6 16.8 17 17.2 17.4 17.60.45
0.5
0.55
0.6
0.65
0.7
0.75
0.8
0.85
0.9
0.95
Vc (V)
iL (
A)
Fig. 9. Period-1 (experimental waveforms 1A/V) at Iref = 09 A {left: practical, right: SIMULINK.}.
A.N. Natsheh et al. / Chaos, Solitons and Fractals 39 (2009) 2465–2476 2471
3.2. S1 and S2 open
When S1 and S2 open, D1 and D2 conduct. The circuit is as shown in Fig. 4, and is described by the following dif-ferential equations:
di1
dt¼ � vc1
Lþ V I
Lð9Þ
di2
dt¼ � vc2
Lþ V I
Lð10Þ
dvc1
dt¼ � vc1
2RCþ i1
2Cþ i2
2Cð11Þ
dvc2
dt¼ � vc2
2RCþ i1
2Cþ i2
2Cð12Þ
whose general solution is
i1ðtÞ ¼ i2ðtÞ ¼ e�ktðA1 sin xt þ A2 cos xtÞ þ V I
2Rð13Þ
where
x ¼ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi1
LC� k2
rð14Þ
and k is given in Eq. (8).
8.4 8.6 8.8 9 9.2 9.4 9.6 9.8
x 10-3
-2
-1.5
-1
-0.5
0
0.5
1
1.5
2
Vc (V)
iL (
A)
17 17.5 18 18.5 19 19.5 20 20.50.4
0.5
0.6
0.7
0.8
0.9
1
1.1
1.2
1.3
iL (
A)
Vc (V)
Fig. 10. Period-2 (experimental waveforms 1A/V) at Iref = 1.2 A {left: practical, right: SIMULINK.}.
2472 A.N. Natsheh et al. / Chaos, Solitons and Fractals 39 (2009) 2465–2476
The constants A1 and A2 are set by the boundary conditions.The circuit is defined by these equations until the next set of clock pulses arrives, causing S1 and S2 to close. This
happens at a time t0n, where
t0n ¼ T ½1� ðtn=T Þ� ð15Þ
Setting t = 0 immediately after S1 and S2 have opened gives
i1ð0Þ ¼ i2ð0Þ ¼ A2 þV I
2R¼ I ref ð16Þ
and
vc1ð0Þ ¼ vc2ð0Þ ¼ vn e�2ktn ¼ V I � Ldi1
dtt¼0
¼ V I � Ldi2
dtt¼0
ð17Þ
Using Eq. (13), and solving gives
A1 ¼V I � vne�2ktn þ LI 0ref k
xLð18Þ
A2 ¼ I ref �V I
2Rð19Þ
where
I 0ref ¼ I ref �V I
2Rð20Þ
5 5.5 6 6.5 7 7.5 8 8.5 9 9.5 10x 10-3
-4
-3
-2
-1
0
1
2
3
4
time (sec)
iL (
A)
20 22 24 26 28 30 32 34
0.8
1
1.2
1.4
1.6
1.8
2
2.2
2.4
2.6
Vc (V)
iL (
A)
Fig. 11. Chaotic (experimental waveforms 1A/V) at Iref = 2.5A {left: practical, right: SIMULINK.}.
A.N. Natsheh et al. / Chaos, Solitons and Fractals 39 (2009) 2465–2476 2473
Since i1ðnþ1Þ ¼ i1ðt0nÞ, and i2ðnþ1Þ ¼ i2ðt0nÞ, using Eq. 13 gives
i1ðnþ1Þ ¼ i2ðnþ1Þ ¼ e�kt0nkLI 0ref V I � vn e�2ktn
xLsin xt0n þ I 0ref cos xt0n
� �þ V I
2Rð21Þ
Similarly,
vc1ðnþ1Þ ¼ vc2ðnþ1Þ ¼ V I e�kt0nkvn e�2ktn � kV I � I 0ref=C
xsin xt0n þ V I � vne�2ktn
� �cos xt0n
� �ð22Þ
Eqs. (21) and (22) were programmed into MATLAB.The mapping (refer to Eqs. 21 and 22 above) is a function that relates the voltage and current vector (vn+1, in+1)
sampled at one instant, to the vector (vn, in) at a previous instant, with numerical analysis carried out using MAT-LAB. To obtain the bifurcation diagrams, initial conditions (i1(0), vc1(0), i2(0), vc2(0)) are specified, together with agiven Iref. The equations are iterated 750 times, with the first 500 values discarded. The last 250 values are plottedtaking Iref as the bifurcation parameter (Iref was swept from 0.5 to 5.5) with the set of circuit parameters listed inTable 1.
Figs. 5 and 6 show the inductor current bifurcation diagrams for the two parallel converters. It is evident that per-iod-1 is stable until a critical bifurcation point Irefo = 1.05A. A further increase in Irefo results in one of the Floquetmultipliers leaving the unit circle through �1, which causes a period-doubling bifurcation. A further increase in Iref
beyond Iref results in the system undergoing stable period-2, and a route to chaos. For Iref > Irefo, the period-1 solutionwill continue to be an unstable period-1 solution.
8.5 9 9.5 10x 10-3
-2
-1.5
-1
-0.5
0
0.5
1
1.5
2
time (sec)
iL (
A)
12.4 12.6 12.8 13 13.2 13.4 13.6 13.8 140.75
0.8
0.85
0.9
0.95
1
1.05
1.1
iL (
A)
Vc (V)
Fig. 12. Period-1 (experimental waveforms 1A/V) at Iref = 1A {left: practical, right: SIMULINK.}.
2474 A.N. Natsheh et al. / Chaos, Solitons and Fractals 39 (2009) 2465–2476
To compare the inductor current bifurcation behaviour in the two-module system with that of a single-unit boostconverter, which has the same parameter values of Table (1), the bifurcation diagram of the latter is plotted inFig. 7. It is clear that the critical bifurcation point of the modular converter takes place at a lower value of referencecurrent compared to that of the single-boost converter.
To check the validity of these findings, experimental results and a MATLAB/SIMULINK model were employedand the results are discussed in the following section.
4. Experimental implementation and MATLAB/SIMULINK simulation
To verify the theoretical model, an experimental two-module boost converter was built, according to the block dia-gram of Fig. 1. Also a MATLAB/SIMULINK model was developed and this is shown in Fig. 8.
Figs. 9–11 show both experimental and MATLAB/SIMULINK inductor current waveforms, phase portraits show-ing the relationship between inductor current and capacitor voltage, and power spectral densities for different values ofthe bifurcation parameter Iref. The figures are in good agreement and clearly demonstrate period-1, period-2, and cha-otic behavior. The power spectrum in the chaotic region has a wide band, unlike that for period-1, which contains afundamental switching frequency component, together with higher order harmonics at multiples of the switching fre-quency. Comparing the results obtained from the bifurcation diagrams of Figs. 5 and 6 with those given in Figs. 9–11 show good agreements in terms of operating region.
Figs. 12–14 show the steady-state waveforms for the single converter inductor current, the phase portraits, andpower spectral densities for different values of the bifurcation parameter Iref. Again, the figures are in good agreementand clearly demonstrate period-1, period-2, and chaotic behavior. The bifurcation diagram of Fig. 7 also agrees with theexperimental results and those obtained from the MATLAB/SIMULINK model.
8 8.2 8.4 8.6 8.8 9 9.2 9.4 9.6 9.8 10x 10-3
-2
-1.5
-1
-0.5
0
0.5
1
1.5
2
time (sec)
iL (
A)
15 15.5 16 16.5 17 17.5 18 18.5 19 19.51.1
1.2
1.3
1.4
1.5
1.6
1.7
1.8
1.9
Vc (V)
iL (
A)
Fig. 13. Period-2 (experimental waveforms 1A/V) at Iref = 2.5A {left: practical, right: SIMULINK.}.
6 6.5 7 7.5 8 8.5 9 9.5 10x 10-3
-4
-3
-2
-1
0
1
2
3
4
time (sec)
iL (
A)
15 20 25 301.4
1.6
1.8
2
2.2
2.4
2.6
2.8
3
3.2
Vc (V)
iL (
A)
Fig. 14. Chaotic (experimental waveforms 1A/V) at Iref = 3A {left: practical, right: SIMULINK.}.
A.N. Natsheh et al. / Chaos, Solitons and Fractals 39 (2009) 2465–2476 2475
5. Conclusion
The waveforms produced by the MATLAB/SIMULINK models for both the two-module and the single convertersshow excellent agreement with experimental waveforms. They also agree with the transition points predicted by thebifurcation diagrams. These models will be used in an on-going research programme which will investigate methodsof controlling and ultimately eliminating chaotic behaviour.
References
[1] Deane JHB, Hamill DC. Instability, subharmonic and chaos in power electronic circuits. IEEE Trans Power Electron1990;5:260–8.
[2] Deane JHB. Chaos in a current-mode controlled boost DC–DC converter. IEEE Trans Circ Syst1: Fundam Theory Appl1992;39(8):680–3.
[3] Hamill DC, Deane JHB, Jefferies DJ. Modeling of chaotic DC–DC converters by iterated nonlinear mapping. IEEE Trans PowerElectron 1992;7(1):25–36.
[4] Tse CK. Flip bifurcation and chaos in three-state boost switching regulators. IEEE Trans Circ Syst1: Fundam Theory Appl1994;41(1):16–21.
[5] Tse CK, Chan WCY. Chaos from a current programmed Cuk converter. Int J Circ Theory Appl 1995;23:217–25.[6] Marrero JLR, Font JM, Verghese GC. Analysis of the chaotic regime for DC–DC converters under current-mode control. In:
Power Electronics Specialists Conference, PESC; 1996.p. 1477–83.[7] Banerjee S. Nonlinear modeling and bifurcation in boost converter. IEEE Trans Power Electron 1998;13(2):25–260.
2476 A.N. Natsheh et al. / Chaos, Solitons and Fractals 39 (2009) 2465–2476
[8] Iu HHC, Tse CK, Lai YM. Bifurcation in current-sharing parallel-connected boost converters. In: Power electronics and motioncontrol conference, PIEMC; 2000. p. 921–24.
[9] Natsheh A, Nazzal J. Application of bifurcation theory to current mode controlled parallel-connected DC–DC boost converterswith multi bifurcation parameters. Chaos, Solitons & Fractals 2007;33(4):1135–56.
[10] Nayfeh AH, Balachandran B. Applied nonlinear dynamics. New York: John Wiley; 1995.