Log-Periodic Dipole Antennas in Printed Circuit Technology

Guilherme Conde Vieiraguilhermeconde.v@gmail.com

Instituto Superior Tecnico, Lisboa, Portugal

June 2018

Abstract

The traffic of information nowadays is becoming larger than ever before. Devices and connections aregrowing faster than both the population and Internet users. By the year of 2021, it is expected for theSmartphone traffic to exceed the PC traffic. With new communication systems being developed everyyear, it is essential to have good measurement antennas to test and validate them. The Log PeriodicDipole Antenna (LPDA) is one of the most used frequency independent antennas, with applicationsranging from HF (or even MF) to microwave. This is a broadband, multi-element and directionalantenna. Combining this antenna with Printed Circuit Technology, which is a low cost, easy fabricationand installation technology that stands for low profile, compact, light weight and robust antennas, theapplication possibilities are tremendously extended. This thesis proposes the study, design, optimizationand fabrication of a pair of printed LPDA antennas, designed to work in the frequency range 500-1750MHz, with an input impedance of 50 Ω, a reflection coefficient below -10 dB and a gain as high aspossible, for application as standard antenna for low frequency far-field anechoic chamber measurements.The design procedure has to deal with size restriction imposed by the PCB technology fabricationlaboratory. Another LPDA to be fabricated using not PCB technology, and therefore free from sizelimitations is also designed. The experimental input reflection coefficient and radiation pattern resultsshow a good agreement with numerical simulation (CST) results providing validation of the designprocedure and proof of the proposed antenna concept.Keywords: Log-periodic dipole antenna, Printed antennas, Miniaturization techniques, Standardantennas, Antenna measurements, Broadband antenna.

1. Introduction

Today, sharing information is easy, fast and cheap.This ability to share information has made it pos-sible for people and industries to collaborate more,develop new solutions and combine different areas ofexpertise, overturning the old and traditional busi-ness models. The traffic of information nowadays isbecoming larger than ever before. Devices and con-nections are growing faster than both the popula-tion and Internet users. In 2016, computers were re-sponsible for almost 46 percent of the total IP traf-fic, with Smartphones only taking a smaller part, 13percent. By the year of 2021, it is expected for theSmartphone traffic to exceed the PC traffic, withthe first one having 33 percent of the total IP traf-fic, where Computers will decrease to 25 percent.Also, the majority of IP traffic will be Wireless,with 63 percent.

With new Wireless communication systems beingdeveloped and with Wireless IP traffic increasingsubstantially every year, it is essential to have goodmeasurement systems and techniques to validatethe requirements. To test those systems, the an-

tenna research and development (RD) communityhas a lot of computational electromagnetic (CEM)tools at their disposal. Tools that minimize the fi-nancial resources for the design of new antennas byallowing to simulate a virtual environment but inorder to validate the simulated results there is astrong need for new measuring antennas. Despitethe fact that the software available is now accu-rate enough to achieve a point of convergence withthe measured results, only by putting the antennaunder test (AUT) to a real environment can pro-vide enough confidence for a total validation andfinal characterization of the antenna. To do that,an anechoic chamber seems the ideal place. It is aroom designed to suppress contributions from thesurrounding environment during the experimentaltests, by providing protection from weather andby absorbing the reflections of the electromagneticwaves. Ideally, since the chamber can isolate elec-tromagnetic (EM) waves from the exterior, the re-ceiving antenna would only receive the EM wavesdirectly from the emitting antenna, since the cham-ber walls (ideally) absorb the waves inciding on

1

them. The measuring process of either radiationpattern or antenna gain in the chamber is quitesimple. The antenna in the receiving position isplaced in the far field of the source antenna andby measuring the received power and comparing itwith the gain of the standard antenna one can ob-tain the gain of the AUT. That is why standardgain antennas are very important, as they are com-monly used for gain reference and antenna measure-ments. These are antennas that are able to oper-ate in a specific range of frequencies with a reason-able gain and with their characteristics well-known.The LPDA is one of the most used frequency inde-pendent antennas, ranging from HF (or even MF)to microwave applications. This antenna is knownfor its wide bandwidth and its simple design. Itis a broadband, multi-element, directional antennawith radiation and impedance characteristics thatideally repeat themselves regularly as a logarithmicfunction of the excitation frequency. This charac-teristics and the fact that at Instituto Superior Tc-nico and at Instituto de Telecomunicaes there areno standard antennas able to cover frequencies be-low 1.72 GHz (IT/IST has antennas that cover afrequency spectrum from 1.72 GHz to 75.8 GHz),makes the LPDA antenna a perfect candidate toprovide a solution to fill this gap. With the devel-opment of large scale integrated circuit and printedcircuit board (PCB) technology, printed antennashave drawn a lot of attention in the antenna com-munity. Their light weight, low cost, easy fabri-cation and installation and good performance hasattracted a lot of interest. These antennas are lowprofile, robust and not expensive to manufacture us-ing modern printed-circuit technology, and they arevery versatile in terms of resonant frequency, radi-ation pattern, polarization and impedance. More-over, they are very easy to integrate with electroniccomponents and arrays and they offer the 2 possi-bility of being printed on curved surfaces to accom-plish the creation of conformal antennas. With thistechnology being cheaper than the traditional an-tennas and since IT/IST has requested an antennaable to work at low frequencies for application asstandard antenna for low frequency far-field ane-choic chamber measurements.

This paper proposes to contribute with a step-by-step design procedure and simulated performanceanalysis of a printed log-periodic dipole antenna(LPDA) that is able to operate over a wide fre-quency range 500-2400 MHz. It contributes withthe development of an expertise in the antennafield. The work carried out goes through all theimportant stages of an EM engineering process,namely, analysis, design (with optimization), fab-rication and test. Initially it has been necessaryto obtain know how on the LPDA antenna work-

ing principles and get acquainted with the softwaresimulation tool (CST). It was then possible to de-velop a well sustained activity that has allowed thefollowing original contributions: detailed paramet-ric analysis of LPDA printed antenna. All the rel-evant parameters such as: dipole‘s size, number ofdipoles, feeding line, feeding coaxial cable, termina-tion, dipole width and substrate shape, have beenstudied; design (with optimization), fabrication andtest of a printed LPDA antenna with strict max-imum size restrictions, large bandwidth and gaincontrol; design and optimization of a non-printedLPDA antenna supported in foam and providingnot only large bandwidth but also gain improve-ment; enhancement of the IT/IST set of standardgain antenna with a pair of printed LPDA antennasthat can be used in the frequency range [500-2400]MHz and therefore extending the measurement ca-pabilities lower to 500 MHz.

2. Antenna Characteristics2.1. Geometry Description

With a rather simple design, the antenna is com-posed by a sequence of linear dipoles, displacedside-by-side, forming a coplanar array. The dipolesdiffer from sizes and are fed alternately by a com-mon transmission line. The geometrical dimensionsof the elements in the LPDA follow a very specificpattern. For the following analysis, the notationbellow will be used:

Figure 1: LPDA configuration [1].

dn or Dn - distance between elements n and n+1;

an - radius of the element n;

ln - half the length of the element n (n = 1,2,...N);

LN - length of the element n (n = 1,2,...N);

Xn - distance of element n to the antenna (vir-tual) apex;

La - Total length of the antenna;

LN - Length of the shortest dipole;

L1 - Length of the longest dipole;

2 α - Apex angle;

B - Usable band;

K1 and K2 - Upper and lower truncation coeffi-cients.

2

These parameters are related through a spacingfactor (σ) and a scalling factor (τ) and the antennaarray increases logarithmically as defined by thescaling factor presented in equation 1.

τ =ln+1

ln=Xn+1

Xn=dn+1

dn=an+1

an, 1 ≤ n ≤ N

(1)

Frequently the radius of the dipoles do not re-spect entirely the geometric progression. Usuallydipoles have the same radius.

Also, the spacing factor (σ) is defined by:

σ =dn

2Ln(2)

The angle 2α is a characteristic of frequency in-dependent structures. It is called the apex angleand it relates with τ and σ.

tanα =Ln

2Xn=Ln − Ln+1

2dn=

1− τ4σ

(3)

Constant dimensions are used because it is veryhard to obtain wires of different diameters and tomaintain tolerances of very little gap spacings. Con-trary to the Yagi-Uda antenna, where only one el-ement of the array is directly empowered by thefeed line, while all the others remain operating asparasitics, in the log-periodic array all the elementswork connected to each other [pp.619-635]Balanis.

To achieve a true log-periodic configuration an in-finite structure would be needed. Yet, to use the an-tenna as a broadband radiator, the structure needsto be truncated at both edges, which limits the op-erational frequency to a given bandwith.

The electrical lengths of the longest and short-est dipoles determine the cutoff frequencies and thehigh cutoff frequency appears when the shortestdipole is approximately λ

2 and only when the ac-tive region is very narrow, while the low cutoff fre-quency appears when the longest dipole is about λ

2[pp.619-635]Balanis.

In terms of functionality, the antenna can be di-vided in tree regions: the transmission region, theactive region and the reflexion region.

Polarization: the wave that is radiated of a sin-gle LPDA is linearly polarized, having horizontalpolarization when the antenna plane is parallel tothe ground. When one wants to obtain bidirectionalpatterns or circular polarization, one has to com-bine multiple LPDA, and in that case, the totaleffective phase center can be sustained at the feed[pp.619-635]Balanis.

2.2. Design rulesCarrel, in 1961, introduced a mathematical modelfor the LPDA analysis, whose results agreed reason-ably with the experimental values Carrele. That al-lowed him to define a step-by-step design procedurethat will be described below. To help in the designprocess, he defined a set of curves and monographs[pp.619-635]Balanis. The antenna configuration isdescribed mostly in terms of the design parametersσ, τ and α and they relate to each other as pre-sented in equation 2.3.

There is an important relation between the valuesof α and τ . If the value of α increases then the valueof τ decreases and vice-versa. Furthermore, alongwith the decrease of α or the increase of τ , the num-ber of dipoles that are closer to each other increases,increasing the number of elements in the active re-gion which are approximatly λ

2 . Nonetheless, thevariations of impedance and other characteristicsas a function of frequency are smaller, due to thesmoother transition between the elements. Also, ifα decreases then the gain increases and vice-verse[pp.619-635]Balanis.

The original study of Carrel allowed to obtainthe dependence of the LPDA directivity with theparameters σ and τ , for antennas with Z0 igual to100 Ohm and L

D = 177. It shows that dependence,along with the relative space between the dipole el-ements (σ), called Optimum σ, which maximizesthe directivity for a specific value of τ sebentacom-plementosantenas. However, the original directivitycurves are incorrect due to an error in the expres-sion for the E-plane field pattern, so the values arevery optimistic, being 1-2 dB higher than what it issupposed.

Figure 2: Carrel curves and corrections [2].

According to Peixeiro1988, the corrected direc-tivity contour curves are represented in figure 2 be-low. The corrected curves (not the dashed ones)indicate that in order to achieve only 1dB addi-tional gain, a larger spacing factor σ and a biggerscaling factor τ are needed, which require the an-

3

tenna to extend its length by having more radiatingelements.

The first step regarding the LPDA design is tochose the desired directivity, D0. That will de-termine the scale (τ) and space factors (σ) usingCarrel‘s design contours presented in the figuresabove, D(τ ,σ). The operation frequency range andthe truncation coefficients (K1, K2) allow to obtainthe length of the longest (L1) and shortest (LN )elements Peixeiro1988.

Figure 3 shows a corrected diagram of the trunca-tion coefficients Peixeiro1988, corresponding to thecase Z0 = 100 Ω and Ln

Dn= 177 .

Figure 3: Truncattion coefficients for optimum (τ ,σ) pairs [2].

In figures 2 and 3 the solid curves correspond toCarrel results Carrele while the deshed curved wereproposed in Peixeiro1988.

The following expressions are used:

L1 ≥ K1λmax (4)

LN ≤ K2λmin (5)

The number of dipoles is given by:

N = 1 +log(L1/LN )

log(1/τ)(6)

Where the operational band ratio is representedby:

B =fmaxfmin

=λmaxλmin

(7)

The active region bandwidth is :

Bar =K1

K2(8)

After obtaining Bar and B it is possible to obtainthe lengths of the shortest and longest dipoles.

Length of the longest dipole:

L1 = k1λmax = LNBBar (9)

Length of the shortest dipole:

LN = k2λmin = τN−1L1 (10)

Xn represents the spacing between elements andit can be obtained by:

Xn =2σLn

1− τ(11)

The total length of the antenna (LA) correspondsto the distance between the shortest and the longestdipoles:

LA = X1 −XN =L1LN2tanα

=2σ(L1LN )

1− τ(12)

Also, the dipoles thickness is represented by theratio:

Ln2an

(13)

Having in mind the desired input impedance, thevoltage standing wave ratio (VSWR) allowed, thepower-handling capability and some practical con-struction limitations, the characteristic impedanceof the feeder transmission line can be obtained bythe following expression:

Z0 =Z

πarc cos h

e

2a≈ Z

πLn

e

a(e a) (14)

Where the transmission line conductors radius isrepresented by ”a” and spacing between wires by”e” (for further analysis sebentacomplementosante-nas).For high power emission antennas it is neces-sary to have transmission lines with high character-istic impedances, which means more space betweenthe conductors to hold the voltages without disrup-tion sebentacomplementosantenas.

2.3. Feeding Technique

Figure 4: Balun implementation [3].

The antenna feeding is made using a copper semi-rigid coaxial cable directly soldered to the LPDA

4

feeding point and by having a SMA connector onthe other end, the balun is implemented by solder-ing the outer conductor of the coaxial cable alongthe feeding line of the LPDA and the inner con-ductor soldered directly to the other feeding linethrough a via, as it can be seen in figure 4. Strictlyspeaking, this configuration is in essence composedby two antenna feeder lines each on the oppositeside of the PCB that provide a 180 degree phaseshift and an array of dipole elements connecting al-ternately to the one on the upper and to the otheron the bottom side of the PCB [4].

2.4. Gain Measurement

There are two simple methods that can be used tomeasure the antenna gain: the absolute gain andthe gain comparison techniques (or gain-transfermethod) gainmeasur. The absolute gain methoddoes not require previous knowledge of the trans-mitting (T) or receiving (R) antenna gain. If theantennas (T and R) are identical, one measurementand use of the Friis transmission formula is suffi-cient to determine the gain Balanis.

On the other hand, the gain-transfer techniqueneeds to be used in parallel with standard gain an-tennas to obtain the absolute gain of the AUT. Ev-ery technique has its own advantages and disadvan-tages, in terms of accuracy, time consumption andcost.

For example, the gain transfer method producesa simple and accurate solution for measuring theantenna gain. It uses a standard antenna with itscharacteristics well-known, such as the reflection co-efficient and gain to determine the gain of an AUTin conjunction with equation 15 (below).

GAUT =PrAUTGSGH

PrSGH(1− | τt |2AUT )(dAUTdSGH

)2 (15)

GAUT , dAUT , PrAUT , and | τt |2AUT representthe unknown gain, the distance between the AUTand the transmitting antenna, the received powerand the reflection coefficient related with the AUT,respectively. Further, dSGH , PrSGH and GSGH arethe distance between the standard gain antenna andthe transmitting antenna, the received power asso-ciated with the standard gain antenna and standardantenna gain, respectively. It is also important tosay that pyramidal horn antennas are universallyaccepted and used as standard gain antennas Sim-ulatingmeasurments.

In the anechoic chamber of IT/IST, the two an-tennas (AUT and standard antenna) are 5 m apart.This means that the far-field distance needs to beless than 5 m.

3. Antenna Design and Optimization3.1. Design Considerations

When designing an antenna, certain limitations aretaken under consideration. Every antenna is differ-ent and for every purpose, different requirementsare set.For the simulation procedure, the authorused the CST MWS software, which allowed thesimulation of the antenna under a controlled envi-ronment, in order to understand how the antennaperforms by studying its radiation characteristicsand each parameter influence on the antenna‘s be-haviour.

During the simulation process, some importantcharacteristics were monitored and taken underconsideration, such as the antenna realized gain,that takes into account all the losses (like theloss of mismatching), the bandwidth, the cross-polarization, the reflection coefficient (S11), the effi-ciency, the directivity, the antenna‘s radiation pat-terns and some other important characteristics. Itis imperative to say that during the optimizationprocess the antenna was simulated using 1 coaxialcable since it was known from the start that the pro-totypes would be fed that way. The feeding methoddepends on the type of device and the required char-acteristics one intends to use the antennas for.

In order for the prototype antenna to meet itsspecific requirements, the gain should be the highestpossible, since the antenna will be used as a stan-dard antenna for gain measurements. As for thereflection coefficient, this is one of the parametersthat better describes how well matched to the feed-ing coaxial cable is the antenna for each working fre-quency, by measuring and comparing the power ofthe reflected electromagnetic wave versus the powerof the incident wave. The lower the power of the re-flected wave, the better matched the antenna is.The input reflection coefficient, or S11, is a veryimportant parameter since it defines the frequencyband in which the antenna operates. In otherwords, for this specific project one important re-quirement is for the antenna to cover frequenciesranging from 500 MHz to 1750 MHz, so this condi-tion must be satisfied | S11| ≤ -10 dB within thatfrequency range.

This criteria is used as a rigorous reference andis suitable for this specific project, not always con-sidered as such for every antenna (different valuessuch as -6 dB i.e. for mobile communications, or-15 dB are used, depending on the antenna applica-tion). Moreover, despite CST being a tremendoustool in the simulation of a variety of antennas, notalways the simulation and the experimental resultsare exactly the same.

The choice of FR4 was simple. It is a relativelycheaper material that is easy to get access to andthat has high fabrication tolerances. Sure it is not

5

the most suitable solution if one wants a low lossboard, but since this project requires for the an-tenna to operate mostly at lower frequencies, thislimitation is certainly not a drawback, since this is alossy substrate that exhibits poor performance withthe increase in frequency. Moreover, the choice ofthis substrate was not only to provide robustnessbut also to help decrease the antenna overall size.

For this project, the maximum production sizefor the PCB board is 420x350mm and it is imposedby the UV machine, used in the production process.This represents a huge problem, since the lower thefrequency of operation the bigger the antenna needsto be. Moreover, the antenna will be used has astandard antenna for low frequency far-field mea-surements in the anechoic chamber and so in orderto minimize the reflections from the walls of thechamber, the antenna has to have a gain as high aspossible.

To reach the best compromise possible, a lot ofsimulations were performed. In terms of polariza-tion, it will have linear polarization, which is com-mon in antennas with dipoles. It will also have aninput impedance of 50 Ω and a reflection coefficientbelow -10 dB for the required bandwidth.

4. Optimized antenna in FR4

Figure 5: Top view geometry.

4.1. Configuration

From figure 5 it is noticeable that the designed an-tenna has a trapezoidal design that meets the sizelimit requirement of 420 x 350 mm, with a size of417.07 x 325 mm. This includes a 1 mm marginbetween the largest dipole (L1) and the limit of theboard to prevent future production problems likedamage L1 when cutting the dielectric. The dielec-tric (represented in yellow) used is a FR-4 lossy ma-terial with a relative permitivity (εr) of 4.3, loss tan-gent (tanδ) of 0.025 @ 10GHz, thermal conductivity(k) of 0.3 W/k/m and a thickness of 1.6 mm. Re-garding the feeding technique, conclusions from sec-tion ?? and also tests realized on this configurationdictated that the antenna would be fed thought asemi-rigid UT-086 coaxial cable positioned straightalong the antenna feed line. The cable has a length

of 419 mm and was excited with a wave guide portat first and then tested with an SMA connector.

The antenna has the following configuration (τ0.89 and σ 0.11):

Table 1: Prototype length, spacing and width ofdipole elements.

The dipoles were designed and simulated using anannealed cooper material (lossy) of thickness 0.035mm since, like all materials described before, thematerials used in the simulation will match the onesused in the laboratory. The feeding lines width andsize are a very important part of the LPDA antenna,since they determine the antenna impedance. Oneof the feed lines have a width of 2.2 mm (the onewith the coaxial on top) while the other has a widthof 3.7 mm. This size difference was reached after anintensive optimization study and is used to compen-sate the effect of the external shield coaxial cablesoldered to obtain an equivalent electroconductibil-ity on both sides of the feed lines. Also, one of thefeed lines has an extension of 1 mm for the innerconductor to be soldered to.

4.2. Mounting structure interface and fabricatedprototype

After the antenna was fabricated, an interface toconnect it to the positioner of the anehoic chamberhad to be made specific for this LPDA antenna.The structure is made of acrylic material to not in-terfere with the antenna‘s radiation characteristicsand had to be strong enough to hold the antennain vertical and horizontal alignments.

The disc has an external diameter of 190 mm withfour holes of 8.1 mm of diameter, distributed uni-formly over a circumference of 170 mm of diameter.

Figure 6 below shows the mounting structuremade in the IT/IST workshop with the antenna at-tached. Figure 7 shows the fabricated prototype

6

(with a coin of 1 euro on its side for reference).

Figure 6: Mountingstructure with theLPDA on it.

Figure 7: Fabricatedprototype.

4.3. Input Reflection Coefficient ResultsFigure 8 contains the S11 simulation (CST) andthe two measurement results. Measurement 1 (inblue) correspond to the first measurement wherethe antenna was placed on top of a foam, whereasmeasurement 2 (represented by red) correspond tothe second measurement where the antenna wasmounted on the structure.

Figure 8: Simulated vs measured S11 results.

The simulated and measured results have verygood agreement between each other and match therequirement of being below -10 dB for the entirebandwidth of interest. Furthermore, the antenna isable to operate in a wider band than the requiredone, being able to operate from 500 MHz up to al-most 2.5 GHz, which is 650 MHz more than theinitial required band.

Aside from the average magnitude of | S11| be-ing 6 dB difference between the simulated and mea-sured results, which can be caused by the simula-tor itself (accuracy and mesh used) and the non-anechoic laboratory environment where | S11| wasmeasured, there are always other factors that hap-pen during the fabrication process that can causethis little discrepancies such as fabrication toler-ances, the way the antenna was cut, the soldering

and connection of the cable that introduce lossesand even the materials used, specially FR4 which isnot a high quality substrate.

4.4. Simulated Radiation Pattern Results

Figure 9: Antenna coordinates: front view.

From figure 9 it is possible to identify that theX axis positive is set toward the apex of the an-tenna (after a coordinate transformation), wherethe shortest dipoles are. Below are represented theradiation pattern cuts for both E and H planes forsome specific frequencies of f = (500, 600, 800, 950,1100, 1400, 1750, 2000, 2300, 2400, 2500) MHz,which were carefully chosen to show how the radi-ated fields change with frequency. As expected, thelower frequency (500 and 600 MHz) provide muchhigher back radiation. The front-to-back ratio isabout 10 dB.• E-plane: Constant Theta, Theta = 90

Figure 10: E-plane radiation pattern simulation re-sults.

• H-plane: Constant Phi, Phi = 0

7

Figure 11: H-plane radiation pattern simulation re-sults.

4.5. Measured Radiation Patterns

This section presents the radiation patterns mea-sured in the two anechoic chambers previously indi-cated. As explained before, the chamber of IT/ISTis certified for frequencies above 2 GHz. To over-come this, and because it was imperative to measureand test the antenna performance in the lower fre-quencies (500 MHz), the author went to IT at theDepartment of Electronics, Telecommunication andInformatics (DETI) of University of Aveiro (UA) totest the prototypes in their chamber. The main dif-ference between their chamber and the one in theIT/IST is that they have standard gain horn anten-nas down to 700 MHz.

The results presented below show both patterns,E-plane and H-plane, with their correspondent ver-tical and horizontal polarizations. Each figure cor-responds to a certain frequency and includes thesimulated patterns, the patterns measured in theIT/IST anechoic chamber and the ones measuredat IT/UA.

The measured frequencies in Aveiro were: 500MHz, 700 MHz, 900 MHz, 1100 MHz, 1300 MHz,1500 MHz and 1750 MHz. For each frequency, 4different angular sweeps were made. Measurementsat 500 MHz were made using the standard antennaIMC: IWH-560 and for every other frequency theantenna A-INFO LB-7180 was used. Also, for afuture comparison between more than one trans-mitting antenna, measurements at 1750 MHz werealso made using the horn SGH1726. In sum, the to-tal azimuthal sweeps were 32 using three antennasin total (4 sweeps with IMC: IWH-560, 24 sweepswith A-INFO LB-7180 and 4 sweeps with the hornSGH1726).

Regarding the anechoic chamber of IT/IST, themeasurements were: radiation pattern for the fre-quencies 1100 MHz, 1300 MHz, 1500 MHz and 1700MHz using the horn FMI 06240-10 and the radia-tion pattern and gain for frequencies 1750 MHz,

2000 MHz and 2300 MHz using a horn FMI 08240-10. (Note: the Horn FMI 08240-10 plus the transi-tion 08093-NF10 plus the transition N-SMA wigth6.1 KG.)

Figure 12: E-Plane and H-plane measured radiationpatterns at 500 MHz: measured in IT/UA.

Figure 13: E-Plane and H-plane measured radiationpatterns at 1100 MHz: measured in IT/UA andIT/IST.

Figure 14: E-Plane and H-plane measured radiationpatterns at 1750 MHz: measured in IT/UA andIT/IST.

For more radiation pattern representation see thisThesis.

8

4.6. Gain

Figure 15: Measured Gain at 1750 MHz in IT/ISTand in IT/UA.

The summary of the gain results of both FR4 andfoam antennas are included in tables 2 and 3. Thereis a good agreement between CST simulation andthe experimental IT/IST results. One can see fromtable 2 that for example for the frequency 1750MHz the difference is negligible. However there is asignificant difference concerning the IT/UA exper-imental results. At the moment there is not a con-solidated explanation for this difference, despite theamount of reflections from the walls of the chamberwhen operating at lower frequencies.

Table 2: Comparison of gain results for the FR4antenna.

Table 3: Simulation gain results for the foam an-tenna.

5. ConclusionsThis paper describes the study, design, optimiza-tion and fabrication of a LPDA antenna in printedcircuit technology. The design and optimizationof a non-printed LPDA antenna is also presented.

It contributes with a printed LPDA antenna us-ing FR4 dielectric material for low frequency far-field anechoic chamber measurements. The de-signed configuration provides the best combinationbetween size and performance for the project re-quirements and it not only contributes to the col-lection of standard gain antennas of IT/IST but alsoto the scientific antenna community, by presentingsome interesting new features and studies related tothe specific LPDA configuration. The antenna ini-tial requirement was to cover a frequency spectrumstarting from 500 MHz up to 1750 MHz, which ex-tends the low frequency limit of the standard gainhorn antenna that IT/IST has. In fact they do actu-ally have a horn antenna that can start to operate ata frequency of 1.14 GHz but they do not have a pairof that specific antenna, so no direct gain measure-ments can be made. Having the desired bandwidthas a project requirement in mind, the author suc-ceeded in surpassing that requirement by designingan antenna with a relatively small size that is verywell adapted to work in a frequency band of 500MHz to 2.4 GHz, with an average realized gain of6.5 dB. To make that possible, a detailed paramet-ric analysis of the LPDA printed antenna was per-formed, with all relevant parameters studied, suchas: dipole‘s size, number of dipoles, feeding line,feeding coaxial cable, termination, dipole width andsubstrate shape, have been studied. Furthermore,the antenna is fed through a coaxial cable in astraight forward position and so different feed linewidths were used to help reduce the coaxial shieldsoldering effect, which was proved to have a hugeimpact in the antenna reflection coefficient, lead-ing to a more balanced current distribution. Also,in the anechoic chamber, the transmitting antennaand the AUT have to be in the far-field region ofeach other in order to achieve good and accurate re-sults. Considering that the distance in the chamberbetween each positioner is approximately 5 meters,the antenna cannot be very large and also difficultto wiggle around, or it will be very hard to attachto the positioner. On the other hand, the cham-ber of IT/IST is only certified to operate above 2GHz, and so at lower frequencies the walls startto reflect the EM radiation which degradates themeasurements. That is why the author tried to de-sign an antenna with the gain as high as possibleand with as much dipoles as possible so that theantenna can have a very large active region. Thehigher the gain the more directive is the beamwidthof the antenna, minimizing the reflections from thewalls. Additionally, a new LPDA configuration infoam substrate was study and designed. This wasmade to demonstrate that the gain difference is verynoticeable, which turned out to be true due to anaverage simulated gain in foam of about 9.5 dB for

9

the same bandwidth as the FR4 antenna. This wasalso true because since it did not require any useof the printed technology, there was no size restric-tion and so the antenna has almost 1 meter of totalboom length, compared to the 43 cm of total lengthof the FR4 LPDA. The dimensions and characteris-tics of the two optimized antennas are listed below.Both antennas provide a 50 Ω input impedance andlinear polarization.

Figure 16: LPDA in FR4 prototype dimensions.

• Printed LPDA

Size [cm]: ZZ = 32.5, YY = 41.70, XX = 11.79;Thickness [mm]: 1.6; Weight (without sma): 273.8g; Bandwidth (| S11| ≤ -10 dB): 4.8-1 (500-2400MHz); Average gain: 6.5 dB;

• LPDA in foam

Size [cm]: ZZ = 38, YY = 98.99, XX = 13.55;Thickness [mm]: 5.35; Bandwidth (| S11| ≤ -10 dB):4.2-1 (500-2100 MHz); Average gain: 9.5 dB;

5.1. Future Work

First of all, it is necessary to find a good explana-tion for the significant difference obtained in gainbetween experimental IT/UA and IT/IST results.Secondly, he optimized LPDA antenna in foamshould be produced, tested and its results com-pared to the simulated ones, with the same thingfor the second LPDA antenna printed in FR4. Fur-ther, every optimization process is very unique andhas specific requirements that need to be addressed.Despite the fact that the proposed objectives forthis project have been achieved, the author con-siders there is always room for improvement. Oneinteresting suggestion for future work would be tocontinue the optimization process of the LPDA an-tenna in foam, which was not concluded in this the-sis due to the unavailable materials and the shorttime frame of this project. Also, since the an-tenna in FR4 was tested in two anechoic cham-bers, one in the IT/IST and other in IT/UA, itwould be interesting to realize new measurementsin a low frequency certified chamber such as theone in the Polytechnic University of Madrid. As fu-ture work, it would also be interesting to refine the

mesh structure in the CST simulator and to usethe maximum number of accuracy and cells to ob-tain more accurate simulation results. This wouldbe particularly relevant for gain results at low fre-quency. Finally, one interesting project would beto design and simulate a log-yagi antenna with thesame project requirements, which is a combinationof a Log-Periodic antenna with a Yagi-Uda antennalogyagi, that provides an excellent wideband pat-tern with a very high gain, specially when bendingsome of its elements. This would be very interestingto compare with the printed LPDA antenna, sincethey are very similar in terms of configuration andoperation.

AcknowledgementsFirstly, the author would like to express his sinceregratitude to Professor Custdio Peixeiro for the con-tinuous support during this thesis, for his patience,dedication, knowledge and guidance.To Mr. CarlosBrito and Eng. Antnio Almeida, for their knowl-edge, experience and participation in the fabrica-tion and test of the prototypes. Also, to Mr. Far-inha, for his time and effort helping manufacturethe mounting structure to hold the antenna ontothe positioner of the anechoic chamber. Further,a special thanks to the author′s parents, brotherand friends, for their unconditional encouragementand support during this journey. To Instituto Su-perior Tcnico (IST), a school filled with science andknowledge, where the author completed his studies.To IT/UA (DETI), Prof. Nuno Borges de Carvalhoand Eng. Hugo Mostardinha, for their help, knowl-edge, sympathy and availability in the execution ofthe prototype anechoic chamber measurements. Fi-nally, a very special thanks to Instituto de Teleco-municaes (IT), the institution that sponsored andhosted this thesis.

References[1] A. Brinca, “Advanced Antennas (in Por-

tuguese).” Instituto Superior Tcnico, 1980.

[2] C. Peixeiro, “Design of Log-Periodic Dipole An-tennas,” IEE Proceedings part H, Microwaves,Optics and Antennas, vol. 135, no. 2, pp. 98–102, 1988.

[3] K. Britain, “Printed circuit board antennas - logperiodic.” [online; accessed February 10, 2018],http://www.wa5vjb.com/.

[4] T. Limpiti and A. Chantaveerod, “Design of aPrinted Log-Periodic Dipole Antenna (LPDA)for 0.8-2.5 GHz Band Applications, electricalengineering/electronics, computer, telecommu-nications and information technology (ecti-con),2016 13th international conference,” September2016.

10