design of a wearable active cospas-sarsat beacon …

Erica Debels

antenna.Design of a wearable active COSPAS-SARSAT beacon

Academic year 2013-2014Faculty of Engineering and Architecture

Chairman: Prof. dr. ir. Jan Van CampenhoutDepartment of Electronics and Information Systems

Chairman: Prof. dr. ir. Daniël De ZutterDepartment of Information Technology

Master of Science in Electrical EngineeringMaster's dissertation submitted in order to obtain the academic degree of

Vervust, Ir. Sam AgneessensCounsellors: Sam Lemey, Patrick Van Torre, Ir. Arnaut Dierck, Dr. ir. Thomas

Supervisors: Prof. dr. ir. Hendrik Rogier, Prof. dr. ir. Jan Vanfleteren

Preface

After five year of this wonderful study, I can now proudly present my master thesis. It was

a very challenging experience, that I could not have accomplished without the help of many

people.

First of all, I would like to thank my promotor and co-promotor, Professor Hendrik Rogier

and Professor Jan Vanfleteren, for presenting me this interesting subject. I want to thank

Professor Hendrik Rogier for his guidance troughout the year, which helped me to successfully

complete this thesis. I also want to thank Professor Johan Bauwelinck for giving me advice,

and letting me use the equipment of his group.

Furthermore, I would like to thank my counselors, Sam Lemey, Patrick Van Torre, Arnaut

Dierck, Thomas Vervurst and Sam Agneessens, for helping me whenever I needed it. In par-

ticular, I want to thank Sam Agneessens, Arnaut Dierck and Frederick Declercq, for stopping

by and giving useful counsel during the year. I would like to thank Sam Lemey for his ex-

cellent feedback and guidance. Despite his own busy schedule, he always helped me out. I

would not have been able the finish my thesis successfully without him.

I also want to thank my family, in particular my parents and sister, for supporting me over

these five years. Although I loved all five years of my study, there have been some tough

periods, during which they supported me unconditionally.

Finally I would like to thank all my friends. During my study of electrical engineering, I

met some wonderful people who made this five year lasting journey an amazing experience.

I want to thank Anne Roets for her friendship, and giving me so much fun during my life

as a student. Furthermore, I want to thank my fellow students who did their thesis at the

same research group. They helped me out and encouraged me when things went wrong, and

created a pleasant working atmosphere.

Erica Debels, June 2014

Admission to Loan

The authors give permission to make this master dissertation available for consultation and

to copy parts of this master dissertation for personal use. In the case of any other use, the

limitations of the copyright have to be respected, in particular with regard to the obligation

to state expressly the source when quoting results from this master dissertation.

De auteurs geven de toelating deze masterproef voor consultatie beschikbaar te stellen en

delen van de masterproef te kopieren voor persoonlijk gebruik. Elk ander gebruik valt onder

de beperkingen van het auteursrecht, in het bijzonder met betrekking tot de verplichting de

bron uitdrukkelijk te vermelden bij het aanhalen van resultaten uit deze masterproef.

Erica Debels, June 2014 , Ghent

Design of a Wearable Active

COSPAS-SARSAT Beacon Antennaby

Erica Debels

Master’s Thesis submitted to obtain the academic degree of

Master of Science in Electrical Engineering

Academic 2013-2014

Promoters: Prof. Dr. Ir. Hendrik ROGIER, Prof. Dr. Ir. Jan VANFLETEREN

Supervisors: SAM LEMEY, Patrick VAN TORRE, Ir. Arnaut DIERCK,

Dr. ir. Thomas VERVURST, Ir. Sam AGNEESSENS

Faculty of Engineering and Architecture

Ghent University

Departement of Information Technology

President: Prof. Dr. Ir. Daniel DE ZUTTER

Departement of Electronics and Information Systems

President: Prof. Dr. Ir. Jan VAN CAMPENHOUT

Summary

This master thesis handles the design of an active antenna, that is compatible with Cospas-

Sarsat distress beacons. Cospas-Sarsat is a search and rescue system that is based on beacons

that send out a signal that will be received by satellites, who will, in turn, inform Local User

Terminals on the ground. The active antenna will be designed such that it can be integrated

into a life jacket. Integrating the beacon into the life jacket increases the comfort of the user,

and will result in a more robust design since the system will be protected by the life jacket’s

cover. For this purpose, an aperture coupled shorted patch antenna will be designed that

has a resonance frequency of 406 MHz, and a bandwidth larger than 6 MHz. Measurements

will verify the proper operation and will be used to test the influence of the human body and

bending of the antenna. Furthermore, a class-E power amplifier will be designed and tested

that can deliver the required power of 5 W to the antenna input terminals.

Keywords

Cospas-Sarsat; life jacket; wearable electronics; active antenna; shorted patch antenna; aper-

ture coupling, class-E amplifiers.

Design of a Wearable Active COSPAS-SARSATBeacon Antenna

Erica Debels

Supervisors: Prof. dr. ir. H. Rogier, Prof. dr. ir. J. Vanfleteren, S. Lemey, P. Van Torre, Ir. A. Dierck,Dr. ir. T. Vervurst and Ir. S. Agneessens

Abstract— In this master thesis, the design of an active antenna for aCOSPAS-SARSAT distress beacon, which is intended to be integrated intoa life jacket, is presented. Integrating the beacon into the life jacket in-creases the comfort of the user, and will result in a more robust design sincethe system will be protected by the life jacket’s cover. A small, planar an-tenna, that resonates at a frequency of 406 MHz, and solely exists of flexibleand light weighted materials, will be designed and verified. The antennatopology used is an aperture coupled shorted patch antenna, and the ob-tained bandwidth is 12.8 MHz. Furthermore, a Class-E power amplifierhas been designed that delivers the required output power of 5 W to theantenna terminals, and can be integrated onto the antenna.

Keywords— Cospas-Sarsat; life jacket; wearable electronics; active an-tenna; shorted patch antenna; aperture coupling, class-E amplifiers.

I. INTRODUCTION

THE Cospas-Sarsat (C/S) search and rescue system is re-sponsible for rescuing over 27,000 people since its intro-

duction in 1982. It is a satellite aided tracking system where, incase of distress, a beacon sends out an emergency signal at 406MHz, that can be received by satellites. These satellites will ontheir turn inform Local User Terminals (LUT) on earth.

Initially, the system was designed for use in aircraft and ships,but recently, more and more Personal Locater Beacons (PLB)are being used. During the past few years, making electronicswearable to increase the comfort of the user, has received a lot ofattention. Since one of the target user groups of PLBs are peo-ple on sea, a system that can be integrated into a life jacket willbe presented. Unlike the message generator and signal modu-lator, which can be made very small these days, the design ofa small flexible antenna, resonating at 406 MHz, with an inte-grated power amplifier that can deliver 5 Watt output power, isnot trivial. Therefore an active antenna will be designed in thismaster thesis [1].

II. DESIGN SPECIFICATIONS

All C/S frequency channels lay in the [406-406.1] MHzfrequency band. This means that the antenna must have animpedance bandwidth of at least 100 kHz. The impedance band-width will further on be expressed in terms of the input reflectioncoefficient, |S11|, which has to be less than -10 dB in the desiredfrequency band. However, as the antenna will be flexible, it canbe bent while it is used. Furthermore, the proximity of the hu-man body will influence the antenna performance. To take intoaccount these effects, a safety margin on the antenna bandwidthwill be maintained, and the target is to obtain a bandwidth ofat least 6 MHz. Finally, a hemispherical radiation pattern is re-quired, with radiation pointed away from the human body.

It will be assumed that the signal generator will produce aphase modulated output signal at a carrier frequency of 406

MHz, with a power of 18 dBm. Since C/S requires an outputpower between 35 an 39 dBm, an amplifier that has a gain be-tween 17 and 21 dB for an input power of 18 dBm will be de-signed. Because the amplifier will be used for wearable appli-cations, it will dependent on a battery, that is preferably not toolarge, and so the goal is to design a high efficiency amplifier.High efficiency amplifiers are non-linear devices, but since themessage will be phase modulated, this will not be of concern.

Fig. 1. Structure of the aperture coupled patch antenna.

III. ACTIVE ANTENNA DESIGN

AntennaA small, planar structure for the antenna topology needs cho-

sen, as well as lightweight and flexible materials in order toobtain unobtrusive integration into the life jacket. An antennatopology that is promising to meet the imposed requirements isan aperture coupled shorted patch antenna, of which the topol-ogy is shown in Figure 1. A microstrip patch antenna consists ofa rectangular metallic patch, and a groundplane, that are placedat the opposite sides of the antenna substrate. This structure willbehave as a resonant cavity, inducing electrical fields in the sub-strate. These fields will extend further than the length of thepatch, bringing the fields outside of the cavity, resulting in radi-ation. This topology has a planar structure, and a hemispheri-cal radiation pattern due to shielding of the groundplane, whichmakes it very well suited for wearable applications [2].

Because normal patch antennas have a length in the order

TABLE IOVERVIEW OF THE USED MATERIALS.

Material εr tan δAntenna polyethurane foam 1.58 0.038substrate h = 11 mm

Feed aramid 1.84 0.015structure h = 0.8 mm

Material Rs [Ω/sq] σ [S/m]patch & copper polyester 3 x 10−3 4.8 x 104

groundplane taffeta fabricFeedline Copper foil 59.6 x 106

of λ/2, a shorting plane will be used, cutting the length of thepatch in half by approximation. The chosen feeding techniqueis aperture coupling, as it results in the largest bandwidth [3].The substrates that were chosen are a 11 mm thick black foamfor the patch antenna, and two layers of 0.4 mm thick aramidtextile for the feed structure. For the patch antenna conductivesheets, a Pure Copper Polyester Taffeta fabric was used. Thefeed line was made in copper foil, to obtain a higher efficiency.An overview of the materials used is given in Table I

TABLE IIDIMENSIONS (MM) OF THE ANTENNA.

Patch Wp 48Lp 140

groundplane Wg 80Lg 190

Aperture Ha 72S1 25.7S2 35.1Wa 40Th 1.8Tv 3.6

feed line t 29Wf 2.7

A first design was made in ADS momentum, because this pla-nar 3D solver is much faster than the full wave 3D solver of CSTmicrowave studio. Afterwards, the design was optimised usingthe time domain solver of CST. The optimisation was done inCST because it includes the influence of an infinite groundplaneand substrate to obtain more accurate results. The optimisedantenna dimensions can be seen in Table II. According to thesimulations, a 6.6 MHz bandwidth was achieved, with a |S11|below -20 dB for the [406 - 406.1] MHz band. This result canbe seen in Figure 3. The simulated efficiency is 62.2 %, and themaximum gain along broadside is 0.27 dBi at 406 MHz.

Power AmplifierBesides the antenna, a Class-E power amplifier was designed

to deliver the required output power to the antenna terminals.The principle of high efficiency amplifiers is that the active de-vice is used as a switch. In the ideal case, there will never bea high current flowing trough the transistor, and a high voltage

across the device at the same time. Hence, the power dissipationin the transistor will be very low.

The working principle of the class-E amplifier is that of a sin-gle active device, driven as a switch, followed by a series LCfilter, to block all higher harmonics. An inductor will be used toseparate the RF signal from the DC current. Design equationsare available that make it possible to calculate the componentvalues to obtain Class-E operations for a certain frequency anddesired output power [4] . These component values can than beused as initial values to further optimise the power amplifier. Aninput and output matching network was added to the circuit tohave ideal operation when the input and output are terminatedto a 50 Ω impedance [5]. Decoupling capacitors are added, aswell as low capacitance values at the gate and drain, to short cir-cuit noise over a large frequency band, to avoid interference andto limit unwanted voltage or current spikes. The entire circuitis depicted in Figure 2, and the final component are listed Ta-ble III. The design is made on a 1mm FR4 substrate with a 0.35µm copper layer.

TABLE IIICOMPONENT VALUES OF THE AMPLIFIER, WITH A AFT05MS031N RF

POWER TRANSISTOR.

Component Value Component ValueLRF,choke2 100 nH CG1 180 pFLRF,choke1 100 nH CG2 200 pFCDc,block 22 nF CG3 0.01 µ F

L0 5.1 nH CG4 0.1 µFC0 18 pF CG5 22 µFLo1 5.1 nH CG6 10 pFCo1 39 pF CG7 27 pFLi1 3.3 nH CD1 180 pfCi1 39 pF CD2 0.01 µFLi2 5.1 nH CD3 0.1 µFCi2 22 pF CD4 470 µFCi3 18 pF CD5 27 pF

350 360 370 380 390 400 410 420 430 440 450−40

−35

−30

−25

−20

−15

−10

−5

0

−13.76

396.1 408.9401.7 408.3

−20.9

frequency [MHz]

|S11

| (dB

)

Simulated |S11

|

Measured |S11

|

Fig. 3. Measured and simulated |S11| of the antenna.

VGS

LRF,choke,2

CG5CG4CG3CG2CG2

CG7CG6

LRF,choke,1

VDS

CD1 CD2 CD3 CD4

CD5

L0C0 Co1

Lo1 50Ω

+

−

CDC,blockLi1

Ci1Ci2Li2

Ci3

50Ω

+

−

Fig. 2. Final amplifier topology.

At 406 MHz, a good input matching was achieved, havingan |S11| of -18 dB. The transmission coefficient, |S21| is 18 dBat 406 MHz. A good output matching could not be obtainedwhile keeping a sufficient, amplification and the final |S22| was-5.2 dB at 406 MHz. The S-parameter curves are depicted inFigure 4. Stability was achieved for all possible input and outputterminations. The obtained power gain at 18 dBm input poweris 17.6 dB, and the power added efficiency (PAE) is 38.12 %.

The efficiency is much lower than expected from a class-Eamplifier. Initially, with ideal components, an efficiency of 75% was obtained, but efficiency was decreased when replacingthem by real component models, and adding the PCB layout.Moreover, when the first circuit was built, the amplifier turnedout to be instable. To solve this stability problem, some compo-nents were changed, which resulted in a much lower efficiencyfor the same gain. A possible explanation for this can be thatchanging the components at the output network, resulted in anoutput transfer function that was not ideal for class-E operation.

IV. MEASUREMENTS

AntennaThe antenna input reflection coefficient was measured using a

N5242A PNA-X by Agilent Technologies. The measured |S11|is plotted together with the simulated values in Figure 3. It canbe seen that the measured results corresponds very well with thesimulated ones. However, it should be remarked that the patchlength needed to be reduced from 140 mm to 134 mm to obtaina resonance frequency of 406 MHz, and that the length of thefeedline was increased from t= 29 mm to t= 36 mm, to obtaingood matching. The resonance frequency is still slightly lowerthan the simulated one, but the obtained bandwidth is larger.The antenna was also tested under different bending angles andin the presence of a human body, with different spacings. The|S11| was smaller than -10 dB for all the conditions measured.

To obtain the gain of the antenna, a three antenna measure-ment was done in an anechoic chamber. The resulting maxi-mum gain was -1.8 dBi, which is almost 2 dB lower than thesimulated value. However, the anechoic chamber that was used,is designed for higher frequencies, and doesn’t have optimal ab-

300 320 340 360 380 400 420 440 460 480 500

−40

−30

−20

−10

0

10

20

frequency [MHz]

S−

para

met

ers

(dB

)

Measured |S11

|

Simulated|S11

|

Measured |S21

|

Simulated|S21

|

Measured |S22

|

Simulated|S22

|

Fig. 4. Measured versus simulated S-parameters of the power amplifier.

sorption at a frequency of 406 MHz, making the measurementsinaccurate.

Power AmplifierThe S-parameters of the power amplifier were measured with

the same PNA-X as the antenna. The simulated and measuredS-parameters can be seen in Figure 4. A few adjustments hadto be made to have a good input matching. Li2 and Ci3 werechanged to h nH and 15 pF respectively. Since the changes thathad to be made, are only small, they are probably due to vari-ations in the component values. The measured amplification atan input power of 18 dBm is 18.1 dB, which is higher than thesimulated 17.6 dB. This can be caused by an over estimation ofthe losses by the simulation program, or also due to variationsin component values. The measured PAE was 37.12 %, whichis almost the same as the 38.12 % obtained in simulations.

V. CONCLUSIONS AND FUTURE WORK

An antenna was designed that is flexible, lightweight and hasdimensions of 80x190x12 mm, and is thus suitable for integra-

tion into a life jacket. The impedance bandwidth obtained is12.8 MHz, and the |S11| at 406 MHz was less than -10 dB fordifferent bending angles and when tested on-body. However,C/S also imposes requirements on the radiation pattern and gain,which could not be accurately tested because of an anechoicchamber that was not suited for this frequency. Further testsare thus required to investigate whether these requirements aremet.

A power amplifier was designed that is able to deliver 36.1dBm output power at an input power of 18 dBm, and thus ful-fills the C/S requirements. However, the PAE is rather low, witha value of 37 %, which will result in short battery lifetime, and alot of heat generation. Besides, the input matching is not ideal,with a reflection coefficient of -9.3 dB. The current power am-plifier is designed on a rigid FR4 substrate, and should be re-designed for the same substrate as the antenna feed structure, sothat it can be integrated onto the antenna.

REFERENCES

[1] Cospas-Sarsat Secretariat, “C/S T.001: Specifications for COSPAS-SARSAT 406 MHz distress beacons”, Issue 3 - Revision 13 , October 2012.

[2] Kin-Lu Wong, “Compact and Broadband Microstrip Antennas”, pg 48-49, Wiley, 2002.

[3] David M. Poza, “A Review of Aperture Coupled Microstrip Antennas: His-tory, Operation, Development, and Applications”, May 1996

[4] Nathan O. Sokal, Alan D. Sokal, “Class E- A New Class of High-EfficiencyTuned Single-Ended Switching Power Amplifiers”, IEEE Journal of Solid-State Circuits, Vol. SC-10, No. 3, June 1975

[5] Firas Mohammed ,Ali Al-Raiet, “Design of Input Matching Networks forClass-E RF Power Amplifiers”, High frequency electronics, pg. 40-48, Jan-uary 2011

Ontwerp van een actieve COSPAS-SARSAT bakenantenne voor integratie in reddingspakken.

Erica Debels

Begeleiders: Prof. dr. ir. H. Rogier, Prof. dr. ir. J. Vanfleteren, S. Lemey, P. Van Torre, Ir. A. Dierck,Dr. ir. T. Vervurst and Ir. S. Agneessens

Abstract—In deze master thesis, zal een actieve antenne ontworpen wor-den voor een Cospas-Sarsat (C/S) baken, die kan geıntegreerd worden inreddingsvesten. Door integratie in de reddingsvest zal het baken beschermdworden door de behuizing, en indien de antenne flexibel, klein en licht is, zalhet comfort van de gebruiker verhoogd worden. Een antenne die voldoendeklein is, en enkel bestaat uit flexibele materialen en een resonantie frequen-tie heeft van 406 MHz, is ontworpen en getest. De gebruikte topologie is eenapertuur gekoppelde kortgesloten patch antenne. Daarnaast is er ook eenklasse-E vermogensversterker ontworpen die het gewenste vermogen levertaan de ingangspoorten van de antenne.

Trefwoorden— Cospas-Sarsat; reddingspakken; draagbare elektronica;active antenne; shorted patch antenne; apertuur koppeling, klasse-E ver-sterkers .

I. INLEIDING

HET Cospas-Sarsat (C/S) systeem voor het opsporen en red-den van personen, heeft sinds haar aanvang in 1982, al

meer dan 27.000 personen gered. Het systeem berust op hetvolgende principe: wanneer een persoon in nood is, dan zal eennoodbaken een noodsignaal uitzenden op 406 MHz. Dit sig-naal zal dan opgevangen worden door satellieten, die op hunbeurt een signaal terug naar de aarde zullen sturen met informa-tie waarmee een reddingsmissie kan gestart worden.

Aanvankelijk was het systeem bedoeld voor gebruik in vlieg-tuigen en schepen, maar het wordt nu meer en meer toegepastvoor individueel gebruik. Deze bakens worden Personal Loca-tor Beacons (PLB) genoemd. Indien een baken gebruikt wordtvoor persoonlijk gebruik, moet het voldoen aan enkele extra ver-eisten. Het systeem moet kleiner en lichter zijn, zodat het mak-kelijk te dragen is. De laaste jaren is het onderzoek naar hetintegreren van elektronica in kledij zeer groot geworden. In ditkader wordt er nu ook onderzoek gedaan naar het integreren vanPLBs in kleren, en hier meer specifiek in reddingsvesten, zodathet comfort van de gebruiker zo hoog mogelijk is.

In tegenstelling tot de signaalgenerator en modulator, die de-zer dagen zeer klein zijn, zal het ontwerp van een antenne meteen resonantie frequentie op 406 MHz, die klein en tegelijk ef-ficient is, niet triviaal zijn. Daarboven moet er 5 Watt vermogengeleverd worden aan deze antenna, waardoor er een versterkerontworpen moet worden die voor 406 MHz een aanzienlijk ver-mogen moet kunnen leveren. [1].

II. ONTWERPSDOELSTELLINGEN

C/S heeft verschillende kanalen beschikbaar, die allen in de[406-406.1] MHz frequentieband liggen. Voor deze band moetde ingangsreflectiecoefficient zeer laag zijn, zodat het vermo-gen afgeleverd wordt aan de antenne, en niet terug gereflec-teerd wordt naar de bron. Er zal vanaf nu naar de ingangsre-flectiecoefficient verwezen worden aan de hand van de |S11|.

Hoewel de vereiste bandbreedte maar 100 kHz is, werd er ge-streefd naar een bandbreedte (BB) van ten minste 6 MHz, omeen zekere marge te hebben. Deze marge is nodig, omdat deantenne ontworpen zal worden in flexibele materialen, en zalgeıntegreerd worden in een reddingsvest. Bij het gebruik vande reddingsvest, kan de antenne mogelijk gebogen worden, ener zal een menselijk lichaam in de buurt zijn op een ongekendeafstand. Zowel buiging als het menselijk lichaam kunnen eeninvloed hebben op de resonantiefrequentie, en als er van in hetbegin een te kleine BB is, is het mogelijk dat 406 MHz uit deresonantieband van de antenne zal komen te liggen. Daarnaastvereist C/S een hemisferisch stralingspatroon, waar er in een ze-ker gebied een gain tussen -3 en 4 dBi moet behaald worden.

Zoals vermeld, moet er een vermogen van ± 5 W, of dus eenvermogen tussen 35 en 39 dBm aan de antenne geleverd worden.In deze thesis werd er vanuit gegaan dat er een signaalgeneratorgebruikt wordt die een 18 dBm uitgangsvermogen levert. Ditwil dus zeggen dat er een versterking nodig is tussen 17 en 21dB. Aangezien de versterker in een draagbare toepassing zal ge-bruikt worden, zal het geleverd vermogen van een batterij ko-men. Omdat alles zo klein mogelijk gehouden moet worden, iseen grote batterij niet aan te raden, en langs de andere kant moetdeze batterij lang genoeg meegaan. Daarom werd er geopteerdvoor een hoog efficiente versterker. Hoog efficiente versterkerszijn echter niet linear, maar dit is geen probleem gezien het sig-naal fase gemoduleerd is.

Fig. 1. Topologie van de apertuur gekoppelde kortgesloten patch antenne.

III. ONTWERP VAN EEN ACTIEVE ANTENNE

AntenneOm een makkelijke integratie in een reddingsvest mogelijk

te maken, zijn alle gebruikte materialen flexibel en zeer licht.Daarnaast is er gekozen voor een kleine, planaire antenne to-pologie. Een veel belovende technologie voor een antenne diezowel klein is als een degelijke BB heeft, is de apertuur gekop-pelde kortgesloten patch antenne. Een patch antenne bestaat uiteen rechthoekig, geleidend vlak, en een groter grondvlak, dieaangebracht zijn op de tegenovergestelde zijden van een sub-straat. De structuur zal zich gedragen als een resonante caviteit,waardoor er velden zullen ontstaan in het substraat. Deze vel-den zullen echter niet stoppen aan de rand van de patch, maarzullen verder doordringen in het substraat. Deze velden die be-staan buiten de grenzen van de patch, zullen zich gedeeltelijkin de vrije ruimte begeven, en zo het veld naar buiten brengenen straling veroorzaken. Deze topologie is planair, en bestaatuit een grondvlak dat straling in deze richting zal tegenhouden,waardoor het stralingspatroon hemisferisch wordt [2].

TABLE IOVERZICHT VAN DE GEBRUIKTE MATERIALEN VOOR DE ACTIEVE

ANTENNE.

Materiaal εr tan δAntenne polyethureen 1.58 0.038

schuimsubstraat h = 11 mmvoedings aramide 1.84 0.015structuur h = 0.8 mm

Materiaal Rs [Ω/sq] σ [S/m]patch & koper polyester 3 x 10−3 4.8 x 104

grondvlak taffeta textielVoedingslijn Koperfolie 59.6 x 106

Omdat de lengte van een patch antenne normaal in de ordevan λ/2 is, is er een kortsluitingsvlak geplaats aan een zijde vande patch, waardoor de lengte van de patch kan gereduceerd wor-den tot de helft van een normale patch antenne. Apertuur kop-peling is gekozen als voedingstechniek, omdat deze leidt tot debeste impedantie aanpassing [3]. De topologie en parameterszijn te zien in Figuur 1.

De gekozen substraten zijn een 11 mm dik zwart schuim voorde patch, en 2 lagen aramide die samen 0.8 mm dik zijn. Voorhet grondvlak en de patch is er gekozen voor een koper polyestertaffetta textiel. Voor de voedingslijn werd er koperfolie gebruikt,omdat dit minder verliezen heeft, en zo een hogere efficientiewordt bekomen. Een overzicht van alle gebruikte materialen isgegeven in Tabel I

Voor het ontwerpen en optimaliseren van de antenne para-meters, werd er gebruik gemaakt van ADS momentum en CSTmicrowave studio. Het eerste ontwerp is gemaakt in ADS mo-mentum omdat deze beschikt over een snelle planaire 3D sol-ver. Daarna werd het ontwerp verder geoptimaliseerd in CST,die over een volledige 3D-golf oplosser beschikt, die accuraterstralingspatronen kan berekenen. De dimensies na optimalisatiezijn te zien in Tabel II. In de simulaties werd er een BB van 6.6MHz bekomen, en de |S11| was onder -20 dB voor de gehele

TABLE IIDIMENSIES (MM) VAN DE GEOPTIMALISEERDE ANTENNE.

Patch Wp 48Lp 140

grondvlak Wg 80Lg 190

Apertuur Ha 72S1 25.7S2 35.1Wa 40Th 1.8Tv 3.6

Voedingslijn t 29Wf 2.7

[406 - 406.1] MHz band. Het resultaat kan gezien worden in Fi-guur 3. De efficientie is 62.2 %, en de maximale antenne winstis 0.27 dBi.

VermogensversterkerNaast een antenne werd er ook een Klasse-E versterker ont-

worpen om het vereiste vermogen te leveren aan de antenne.Het principe van alle hoog efficiente versterkers is dat de ac-tieve component wordt gebruikt als een schakelaar. In het idealegeval zal er nooit tegelijk stroom lopen door de transistor eneen spanning over staan, waardoor de vermogensdissipatie in detransistor zeer laag zal zijn.

Klasse-E versterkers bestaan uit 1 enkele transistor die aan-gestuurd wordt als een schakelaar, gevolgd door een LC filterom de hogere harmonischen er uit te filteren. Er zal een spoelgeplaatst worden tussen de voedingsbron en de drain om lekvan het RF signaal naar de voedingsbron te voorkomen. Voorklasse-E versterkers zijn er reeds formules beschikbaar die decomponentwaarden uitrekenen voor een gegeven frequentie engewenst uitgangsvermogen. [4]

Daarnaast is er een ingangs- en uitgangsaanpassings netwerkvoorzien zodat beide aangepast zijn aan 50 Ω. Evenals zijner een reeks van ontkoppelingscapaciteiten toegevoegd, alsookkleine capaciteiten aan de gate en drain. Deze dienen omruis over een grote frequentie band kort te sluiten en eventu-ele stroom- of spanningspieken te beperken die de transisitorzouden kunnen beschadigen. Het volledige circuit is te zien inFiguur 1, en de uiteindelijke waarden in Tabel III. Het ontwerpis gemaakt op een 1mm dik FR4 substraat, en de dikte van dekoperlagen is 0.35µm.

Voor 406 MHz is er een goede input matching, met een |S11|gelijk aan -18 dB. Daarintegen lukte het niet een goede uitgangsaanpassing te verkrijgen en tegelijk de gain te behouden. De ge-simuleerde |S22| is -5.2 dB voor 406 MHz. De ingangs-uitgangstransmissie coefficient is 18 dB. De S-parameter curves zijn tezien in figuur 4. De versterker is stabiel voor alle mogelijke in-gangs en uitgangs impedanties. De gesimuleerde versterking is17.6 dB met een ingangsvermogen van 18 dBm, en de efficientieis 39.6%.

De efficientie is veel lager als wat verwacht zou worden vanKlasse-E versterkers. Initieel, met dezelfde transistor, en ide-

VGS

LRF,choke,2

CG5CG4CG3CG2CG2

CG7CG6

LRF,choke,1

VDS

CD1 CD2 CD3 CD4

CD5

L0C0 Co1

Lo1 50Ω

+

−

CDC,blockLi1

Ci1Ci2Li2

Ci3

50Ω

+

−

Fig. 2. Schema van de uiteindelijke versterker.

TABLE IIICOMPONENTWAARDEN BIJ HET GEBRUIK VAN DE AFT05MS031N RF

VERMOGENS TRANISTOR..

Component Waarde Component WaardeLRF,choke2 100 nH CG1 180 pFLRF,choke1 100 nH CG2 200 pFLDc,block 22 nF CG3 0.01 µ F

L0 5.1 nH CG4 0.1 µFC0 18 pF CG5 22 µFLo1 39 pF CG6 10 pFCo1 5.1 nH CG7 27 pFLi1 3.3 nH CD1 180 pfCi1 39 pF CD2 0.01 µFLi2 5.1 nH CD3 0.1 µFCi2 33 pF CD4 470 µFCi3 16 pF CD5 27 pF

ale passieve componenten, was de efficientie 75% bij hetzelfdeingangsvermogen. De efficientie was al veel lager wanneer deideale componenten vervangen werden door realistische compo-nent modellen, en de PCB layout was toegevoegd in de simula-ties. Daarnaast werden enkele componenten vervangen om deversterker stabiel te maken, waardoor de efficientie nog zakte,tot 39.6%.

IV. METINGEN

AntenneDe antenne ingangsreflectiecoefficient is opgemeten met be-

hulp van een N5242A PNA-X geleverd door Agilent Technolo-gies. De gemeten |S11| is geplot samen met de gesimuleerdewaarden in Figuur 3. De gemeten resultaten komen goed over-een met de simulaties. Er moet wel opgemerkt worden dat hetnodig was de lengte van de patch te verminderen met 6 mm,tot een lengte van 134 mm, om de gewenste resonantiefrequen-tie te verkrijgen. Daarnaast moest de lengte van de voedings-lijn verhoogd worden met 7 mm, tot een lengte t = 36 mm, omgoede aanpassing te verkrijgen. De opgemeten bandbreedte is

350 360 370 380 390 400 410 420 430 440 450−40

−35

−30

−25

−20

−15

−10

−5

0

−13.76

396.1 408.9401.7 408.3

−20.9

frequentie [MHz]

|S11

| (dB

)

Gesimuleerde |S11

|

Gemeten |S11

|

Fig. 3. Gemeten en gesimuleerde |S11| van de antenne.

12.8 MHz, wat een pak groter is dan de gesimuleerde BB. An-derzijds is de |S11| hoger bij de gemeten antenne op 406 MHZ,doordat de resonantiefrequentie iets lager ligt.

De |S11| was daarnaast ook opgemeten wanneer de antennegebogen was en wanneer er een menselijk lichaam in de nabij-heid was. De |S11| bleef onder de -10 dB voor alle testen.

Om de antennewinst op te meten is er een drie antenne metinguitgevoerd in een anechoische kamer. De bekomen winst is -1.8dBi, wat bijna 2 dB lager is dan de gesimuleerde winst. Dit kante verklaren zijn doordat de gebruikte anechoische kamer geenideale absorptie heeft voor 406 MHz, waardoor het mogelijk isdat er reflecties zijn opgetreden..

VermogensversterkerDe S-parameters van de vermogensversterker zijn opgemeten

met dezelfde netwerk analyser als de antenne. De gesimuleerdeen gemeten waarden van de S-parameters zijn afgebeeld in Fi-guur 4. Om een goede ingangsaanpassing te bekomen, werdenLi2 and Ci3 vervangen door de waarden 5 nH and 15 pF respec-tievelijk. Aangezien deze waarden maar zeer licht afwijken van

Fig. 4. Gemeten versus gesimuleerde S-parameters van de vermogensversterker.

300 320 340 360 380 400 420 440 460 480 500

−40

−30

−20

−10

0

10

20

frequentie [MHz]

S−

para

met

ers

(dB

)

Gemeten|S11

|

Gesimuleerde|S11

|

Gemeten |S21

|

Gesimuleerde|S21

|

Gemeten|S22

|

Gesimuleerd|S22

|

de oorspronkelijk waarden, zijn deze verschillen met de simula-tie waarschijnlijk te wijten aan variaties in de individuele com-ponentwaarden. De gemeten versterking bij een ingangsvermo-gen van 18 dBm is 18.1 dB, en dus hoger dan in simulaties.De oorzaak hiervan kan zijn dat de verliezen in simulaties over-schat worden of ook weer door variaties in componentwaarde.De opgemeten effcientie is 38.6 %, en is ongeveer gelijk aan degesimuleerde 39.6 %.

V. CONCLUSIES EN VERDER ONDERZOEK

Een antenne werd ontworpen die licht, flexibel en klein is,met dimensies van 80x190x12 mm, en die dus makkelijk kangeıntegreerd worden in een reddingsvest zonder het comfort teverminderen. De verkregen BB is 12.7 MHz, en de |S11| op 406MHz was altijd kleiner dan -10 dB voor verschillende buigingenen wanneer aangebracht op een menselijk lichaam. Een goede|S11| is echter niet genoeg, aangezien C/S ook eigenschappenoplegt voor het stralingspatroon. Accurate metingen voor hetstralingspatroon konden echter niet verkregen worden door hetniet geschikt zijn van de anechoische kamer op deze frequentie.Verdere testen zijn dus noodzakelijk om te verifieren of er aandeze voorwaarden is voldaan.

Een vermogensversterker werd ontworpen die in staat is hetgewenste vermogen te leveren wanneer het ingangsvermogen 18dBm is, en dus voldoet aan de C/S specificaties. Hoewel eenhoge efficientie geen vereiste is opgelegd door C/S, is het tochwenselijk om een lange levensduur te voorzien met een com-pacte batterij. De ontworpen versterker heeft een efficientie van38.6%, wat betrekkelijk laag is voor Klasse-E. Daarnaast is deversterker nu ontworpen voor een rigide FR4 substraat. De ver-sterker zou opnieuw ontworpen moeten worden op het zelfdesubstraat als de antenne voedingsstructuur, om hierop te wordengeıntegreerd.

REFERENTIES

[1] Cospas-Sarsat Secretariat, “C/S T.001: Specifications for COSPAS-SARSAT 406 MHz distress beacons”, Issue 3 - Revision 13 , October 2012.

[2] Kin-Lu Wong, “Compact and Broadband Microstrip Antennas”, pg 48-49, Wiley, 2002.

[3] David M. Poza, “A Review of Aperture Coupled Microstrip Antennas: His-tory, Operation, Development, and Applications”, May 1996

[4] Nathan O. Sokal, Alan D. Sokal, “Class E- A New Class of High-EfficiencyTuned Single-Ended Switching Power Amplifiers”, IEEE Journal of Solid-State Circuits, Vol. SC-10, No. 3, June 1975

[5] Firas Mohammed ,Ali Al-Raiet, “Design of Input Matching Networks forClass-E RF Power Amplifiers”, High frequency electronics, pg. 40-48, Ja-nuary 2011

Contents

List of Abbreviations iii

1 Introduction 1

1.1 Situation of the Thesis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1

1.2 Design Challenges . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2

1.3 Applied Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2

1.4 Chapter Outline . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3

2 Literature Study and Background Research 4

2.1 Cospas-Sarsat . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4

2.1.1 History . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4

2.1.2 Principle of Operation . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5

2.1.3 Technical Requirements . . . . . . . . . . . . . . . . . . . . . . . . . . . 7

2.2 Wearable Active Antennas . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13

2.2.1 Wearable Antennas . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14

2.2.2 Power Amplifier Design . . . . . . . . . . . . . . . . . . . . . . . . . . . 19

3 Active Antenna Design 23

3.1 Design Goals . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23

3.2 Active Antenna Topology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26

i

3.2.1 Antenna Topology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26

3.2.2 Power Amplifier Topology . . . . . . . . . . . . . . . . . . . . . . . . . . 30

3.3 Bill of Materials . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33

3.4 Computer Aided Design of the Active Antenna . . . . . . . . . . . . . . . . . . 35

3.4.1 Design of the Flexible Antenna . . . . . . . . . . . . . . . . . . . . . . . 36

3.4.2 Design of the Power Amplifier . . . . . . . . . . . . . . . . . . . . . . . . 43

4 Measurements and Results 61

4.1 Antenna Measurements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61

4.1.1 Input Reflection Coefficient . . . . . . . . . . . . . . . . . . . . . . . . . 61

4.1.2 Radiation Pattern . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66

4.1.3 Influence of the Human Body and Bending of the Antenna. . . . . . . . 68

4.2 Power Amplifier Measurements . . . . . . . . . . . . . . . . . . . . . . . . . . . 69

5 Conclusions and Future Work 78

A AFT05MS031N RF power LDMOS, Freescale Semiconducter 81

B List of measurements on the ouput power of the amplifier 88

References 90

List of Figures 98

List of Tables 99

ii

List of Abbreviations

ADS Advanced Design Systems

COSPAS-SARSAT Cosmicheskaya Sistema Poiska Avariynyh Sudov

(Space System for the Search of Vessels in Distress)

- Search And Rescue Satellite-Aided Tracking

C/S COSPAS-SARSAT

CST Computer Simulation Technology

DC Direct Current

ELT Emergency Locater Transmitters

EPIRB Emergency Position-Indication Radio Beacons

GEO Geosynchronous

LDMOS laterally diffused metal oxide semiconductor

LEO Low-Earth Orbit

LUT Local User Terminals

MCC Mission Control Centres

MEO Medium Earth Orbit

PCB Printed circuit Board

PLB Personal Locater Beacon

PAE Power Added efficiency

PNA Programmmable Network Analyser

RF Radio frequency

SAR Search and Rescue

iii

Chapter 1

Introduction

In this introductory chapter, the context of this master thesis and its relevance for research

will be outlined. Afterwards, the design goals of this thesis are presented. The chapter will

be concluded with an overview of what will be discussed in the next chapters.

This thesis is executed at the electromagnetism research group, which is part of the INTEC

department of Ghent University, in cooperation with ELIS. They will provide the materials,

simulation tools and measurement equipment to realise this master thesis.

1.1 Situation of the Thesis

The subject of this masterthesis is the design of an active antenna that can be used in Cospas-

Sarsat distress beacons, and which can be integrated into a life jacket. Cospas-Sarsat is an

international satellite based Search and Rescue system, that helped saving 21,000 lives in over

5,700 emergency missions, since the first satellite was launched in 1982.

When a person or vehicle is in distress, the emergency beacon transmits a signal that will be

received by satellites, who will, in turn, send information to ground stations. This information

is then used to coordinate rescue operations. Many types of emergency beacons already exist.

Initially the system was designed to use in aircraft and ships. For these applications, larger

and heavier systems could be used. However, recently, Personal Locator Beacons (PLB)

were introduced, which are intended for individuals who are far away from normal emergency

services, like hikers or people on sea. Because these PLBs are made for personal use, they

need to be smaller and lighter than the beacons used for aircraft and ships. In this thesis an

active antenna for a PLB, that can be integrated into a life jacket will be designed.

1

1.2 Design Challenges

Over the past years, more and more research has been done on the integration of electronics

into clothes. Making electronics wearable results in some design challenges. It requires them to

be lightweight, flexible, small and to have a planar structure, in order to realise an unobtrusive

integration. In this master thesis, an active antenna is designed that can be integrated into

a life jacket. By integrating the beacon inside the life jacket, it will be protected by the life

jacket’s cover, which will extend the system’s lifetime. Unlike the message generator and signal

modulator, which can be made very small these days, the design of a power amplifier and

antenna for the Cospas-Sarsat distress beacon, that can be integrated into the life jacket, will

be the major concern. The resonance wavelength, which is 74 cm for Cospas-Sarsat, will result

in rather large antennas, which makes it necessary to use miniaturising techniques. However,

to do so, greatest care must be taken when designing the active antenna. Furthermore, the

antenna materials must be chosen in such a way that the antenna is light-weighted and flexible

such that it does not obstruct the person wearing the life jacket. The power amplifier must

be capable of delivering 5 Watts to the antenna, without needing excessively large batteries.

Besides, it should be small, but, at the same time, have a large efficiency to increase the

batteries lifetime.

1.3 Applied Methods

After choosing an antenna and power amplifier topology, they will be simulated and optimised

using simulation software. For the antenna design, two full-wave electromagnetic solvers were

used, namely ADS Momentum and CST Microwave Studio. For the power amplifier, the

circuit simulator of ADS was used, as well as the ADS Momentum solver to simulate the

effect of losses and coupling in the printed circuit board (PCB) structure.

After achieving the design goals in simulation, the power amplifier and antenna will be as-

sembled, and tested to see if the measured results correspond with the simulated ones. To

accomplish these measurements, the electromagnetism research group had a PNA-X N5242A

Network Analyzer from Agilent technologies, that can provide accurate S-parameter measure-

ments, as well as a U2000B USB power sensor from Agilent, and a signal analyser (Rhode

& Schwarz FSV40). There was also an anechoic chamber available to measure the efficiency

and radiation pattern of the antenna.

2

1.4 Chapter Outline

In Chapter 2, the operation principle of Cospas-Sarsat will be explained. Afterwards, the

possible choices for antenna and power amplifier topologies are discussed, giving their pro’s

and con’s that led to the choice of topology.

In Chapter 3, the design steps are formulated and explained. The simulation results are

given and discussed as well. In Chapter 4, the measurements done on the antenna and power

amplifier will be discussed, and compared with the simulation results.

Chapter 5 will consist of the conclusions that were made, as well as further possible improve-

ments.

3

Chapter 2

Literature Study and Background

Research

2.1 Cospas-Sarsat

2.1.1 History

Cospas-Sarsat is an international satellite based search and rescue system, established by

Canada, France, the United States, and the former Soviet Union and was founded in 1979.

Cospas-Sarsat is a Russian-American acronym standing for “Cosmicheskaya Sistema Poiska

Avariynyh Sudov” (Space System for the Search of Vessels in Distress) - “Search And Res-

cue Satellite-Aided Tracking”. It is a system that detects and locates emergency beacons,

activated by aircraft, ships or individuals in distress [1] [2].

The first satellite (Cospas-1) was launched on the thirtieth of June, 1982. The first signal of

an emergency beacon was detected not long after the activation of the first satellite. At that

time, analog transmission at 121.5 or 243 MHz was used, without any message. However,

soon after the launch of the system, the switch from analog to digital transmission was made

to enhance the performance and to enable the transmission of information along with the

signal. The carrier frequency also changed from 121.5/243 MHz to 406 MHz. Since 2009,

support for the analog 121.5/243 MHz system stopped because too many Search and Rescue

(SAR) resources were wasted on false alerts. [3].

Over the years, more and more countries joined the project. Currently twenty-six more

countries are providers of ground segments, eleven countries are user states, and there are

two participating organisations. Since the establishing of the Cospas-Sarsat sytem in 1982, it

helped saving 21,000 lives in over 5,700 emergency missions.

4

2.1.2 Principle of Operation

The Cospas-Sarsat system consists of three main parts. The first part consists of emergency

beacons, sending out a signal when a person or vehicle is in distress. The second part is

the space segment that consists of geosynchronous satellites called GEOSARs and low-earth

polar orbit satellites LEOSARs. Finally, ground segments consisting of Local User Terminals

(LUTs) and Mission Control Centres (MCCs) were developed, to process the signals received

by the satellites. The LUTs are divided into GEOLUTs and LEOLUTs, depending on the

type of satellites they communicate with.

When in distress, the beacons will send out an emergency signal at 406 MHz, which will be

received by both types of satellites. These satellites will process and relay the information

from the beacons to the ground segment. The information from the ground segments is then

used to coordinate the rescue mission. An overview of this principle is shown in Figure 2.1.

Figure 2.1: Cospas-Sarsat (C/S) principle of operation.

There are three types of emergency beacons available at this moment:

5

• ”EPIRBs” (Emergency Position-Indication Radio Beacons), are used to signal maritime

distress;

• ”ELTs” (Emergency Locator Transmitters), for the use in aircraft;

• ”PLBs” (Personal Locator Beacons), which are intended for personal use, and indicate

a person that is in distress and away from normal emergency services, like backcountry

hikers.

When the system was launched, only EPIRBs and ELTs existed. It is only recently that

beacons for personal use were brought on the market. The difference between the three

beacons, lies in the appliciation in which they are used, and in the size and features, which

are adapted to the requirements of its application area, mentioned above. Initially, the ELTs

for general aviation were constructed to transmit at 121.5/243 MHz, whereas the EPIRBs

were designed to transmit at 406 MHz. More recently, all beacons are designed and optimised

to transmit at 406 MHz. In addition, all beacons communicate in a similar way with the

satellites.

The space segment consists of satellites that are equipped with a SARR/SARP (search and

rescue repeater/processor) system. These SARR/SARP devices are small packages that are

most of the time added to satellites that were already launched for other applications then

Cospas-Sarsat (C/S). These satellites will process and relay the signal back to LUTs on earth

at a frequency of 1544.5 MHz. There are two different satellite systems in use at this moment,

the initial space segment that consists of Low-Earth-Orbit (LEO) satellites and Geostationary

(GEO) satellites that were added later on.

LUTs will receive the signals sent by the satellites. There are also two different kinds of

LUTs. The LEOLUTs and GEOLUTs process the data sent by the LEOSARs and GEOSARs

respectively. From this data, the beacons are localised, and the position is sent to the MCCs

(Mission Control Centres). All MCCs are interconnected, and share the information provided

by the LUTs, to ensure reliable information distribution. [4]

The Cospas-Sarsat Low-altitude Earth Orbit (LEOSAR) System for Search and Rescue is

based on LEO satellites that detect 406 MHz distress signals and communicate with LEO-

LUTs. At this moment, six LEOSAR satellites are monitored by 44 LEOLUTs. The satellites

have a complete, but not continuous coverage of the earth. The satellite can only produce a

distress alert when it “sees” the distress beacon. Therefore, a memory module is included in

the SARP to store distress beacon information that is rebroadcast when the satellite comes

into the view of a LUT.

The LEOSAR system calculates the location of distress beacons, using Doppler processing

techniques. The position is calculated based on the frequency shift that is induced by the

6

Doppler effect. By monitoring the change of the frequency and knowing the position of the

satellite, the distance between the satellite and beacon can be calculated. The position of the

beacon can be triangulated by the LUTs if the distance to three or more satellites is known.

[5].

The GEOSAR system consists of 406 MHz repeaters, carried on board of eight GEO satellites.

The signals sent by the GEO satellites are processed at the sixteen associated ground facilities

called GEOLUTs. These GEOSAR satellites are fixed relative to the earth, which makes it

impossible to use the Doppler shift technique to calculate the position of the beacon, since

there will be none. The location must be provided by the beacon itself or must be derived

from the LEOSAR system. However, this system provides continuous coverage of the earth

up to 70 latitude [6].

In the future, a MEOSAR system will be added, and will consist of MEO satellites (Medium

Earth Orbit search and rescue satellites). The possibility to expand the system and make it

compatible with the Gallileo system is also discussed in literature. [7].

2.1.3 Technical Requirements

Since the moment PLBs came into use, and became more popular, integration of these PLBs

into clothing has become an important topic in several research areas. The goal of this master

thesis is to design an active antenna that is compatible with the Cospas-Sarsat system and

that can be integrated in a life jacket. In the past, two types of signals were used in the

Cospas-Sarsat system, one at 121.5/243 MHz and one in the 406 MHz band. In the next

part, the type of signal send by the emergency beacons and the message contained in the

signal, will be explained in more detail, as well as the requirements imposed by C/S on the

transmitted power and antenna properties. Afterwards, integration of an PLB in a life jacket

will be discussed, and the additional requirements on the system to make it wearable.

Signal type, and message sent by Cospas-Sarsat emergency beacons

The first signal send by C/S beacons was at 121.5/243 MHz. It was a continuous signal without

any identification, and was initially not designed to operate trough the satellite system. The

signal at 406 MHz is a phase modulated burst signal. A short message, with a duration of

± 0.5 seconds, modulated on a carrier in the 406 MHz band, will be transmitted every 50

seconds, with a power of 5 Watt. [2]

At the moment, only alerts transmitted in the 406 MHz band are still detected, so all beacons

are optimised for this frequency. The decision to stop supporting the processing of 121.5

MHz signals was due to the less accurate information it provided, which led to numerous false

7

alerts, adversely impacting the effectiveness of search and rescue operations. The 406 MHz

system is more expensive, but gives more reliable and complete information, leading to better

results.

To be compatible with the Cospas-Sarsat system, the active antenna must be able to send

out a message, with information specified by the Cospas-Sarsat technical documents. The

duration of the message is 440 ms for a short message and 520ms for a long message. The

first 160ms of the transmitted signal consists of an unmodulated carrier, after which 15 bits

of 1’s are transmitted for bit synchronisation. The next 9 bits are always 000101111 and

are used for frame synchronisation. Then there is 1 bit to indicate if it is a long or short

message, followed by the actual message which is 87 (short) or 119 (long) bits. Both messages

contain a protocol flag, country code, beacon identification, and sometimes the position of

the beacon. This information is encoded by a (82,61) BCH code for error detection. The

difference between a short and long message lies in the last bits. The short message has

another 6 uncoded bits, that can be used for additional information, where the long message

adds 26 bits of supplementary information, that is encoded by a (38,26) BCH code. The

repetition period of the bursts is randomly chosen between 47.5 and 52.5 seconds, to make

sure that no two beacons have all of their burst at the same time and interfere with each

other.

Figure 2.2: Message format.

The message is biphase L modulated on a carrier within the 406 MHz band, which is from 406

to 406.1 MHz. The principle of biphase L modulation can be seen in Figure 2.3. The channels

in the 406 MHz band are defined by their centre frequency, and are assigned as provided in

[8]. The frequency variation in time must be limited as described in [9].

Requirements on the transmitted power and antenna characteristics

The power of the signal delivered to the antenna terminals must be between the limits of

5W ± 2 dB, or thus between 35 and 39 dBm. Four C/S signal generators were found that

could be used. Two of them are from Atlantic RF, and two from Syrlinks. The first one from

Atlantic RF is the ”MOSAR − EMODULE, 406MHzRFmodule”. This signal generator

already generates the required power of 37 dBm, but is quite large and thick, with dimenions

55x35x14 mm (length, width, height). The second from Atlantic RF is the “ARF406MHz”

module, and is smaller, with dimensions 20x20x9 mm. However, this module generates a very

8

11 10 0 0 0Data

Phase of the

signal+φ

φ -

Ref 0° rad

Figure 2.3: Biphase L encoding principle.

low output power, 4 dBm. An amplifier with a power gain of at least 31dB will be necessary.

To acquire this power gain, more than one amplifier stage will be required. The third module is

the “MRB1005−0001” module from Syrlinks, and already delivers 37 dBm output power. It

has dimensions of 55x35x7 mm. The last one is the “MRB1002” module from Syrlinks. This

module generates 18 dBm output power and has dimensions of 32x20x4 mm. It is the thinnest

of the four, and thus easiest to integrate in a life jacket. The “MRB1005−0001” is the signal

generator that will be used to base the design of the power amplifier on. The architecture can

be found in Figure 2.4. Since the output power is 18 dBm, a power amplification between 17

and 21 dB will be required.

Figure 2.4: Architecuture of the “MRB1002” signal generator of Syrlinks. [10]

The power available from the source should bet delivered to the antenna, instead of being

reflected back to the source. The most common way to express this requirements, is by means

of the reflection coefficient at the antenna input port, the S11. The standard is to make sure

the |S11| is below -10 dB, in the frequency band specified by the application. A low S11

is obtained when the antenna input impedance is matched to the source impedance. For a

|S11| < - 10dB, less then 10% of the power is reflected. The C/S system however, specifies

9

that the VSWR should be smaller than 1.5:1, which corresponds with a |S11| smaller than

-14 dB, and is thus a stricter condition. The reflection coefficient will be dependent on the

characteristic impedance of the source as well as the input impedance of the antenna. Optimal

power transfer will occur when the source impedance is conjugate matched to the antenna

impedance. This means the S11 needs to be defined, compared to a reference impedance.

The part of the power that is not reflected back to the source, will either be dissipated in the

antenna due to losses, or will be radiated. The percentage of power delivered to the antenna

terminals that is radiated is called the antenna efficiency. The total antenna efficiency also

includes the losses due to impedance mismatch. The efficiency should always made as high

as possible.

The radiation efficiency, defined as the amount of power that is radiated, divided trough the

power delivered to the antenna terminals,

Another important property of antennas is the radiation pattern. An antenna will not radiate

the same amount of power towards all directions. The radiation pattern will express the

directivity of the antenna by giving the power gain of the antenna for every polar angle θ and

every azimuthal angle φ. It is not only a measure for how well power is radiated in a certain

direction, but will also describe how good radiated power is received. This assumption is valid

because of reciprocity. The standard is to give the gain in dBi, which means it is expressed

compared to the gain of an isotropic antenna. The antenna radiation pattern is an important

parameter, because certain applications require an omnidirectional radiation pattern, while

other applications will require a high gain in only one direction, to send a strong signal in

that direction only.

The C/S specifications for the antenna radiation pattern are that it must have a hemispherical

radiation pattern. For 90% of the elevation angles between 5 and 60 in this hemispherical

pattern, the gain must be between -3 dBi and 4 dBi, as shown in Figure 2.5. On this plot, an

angle of 0 corresponds with an elevation angle of 90 , and is the direction perpendicular to

the sea. This gain must be achieved over 90% of this area, and must be true for all vertical

cross sections of the radiation pattern.

A last parameter that will influence the amount of power received is the polarisation of the

waves in the far field of the antenna. To have maximum power transfer, the polarisation axis

of the transmit and receive antenna should coincide. For the design of an antenna for a C/S

beacon, this will not be of great importance, since the specifications are that the polarisation

can be both linear and circular.

Integration of the active antenna into a life jacket

Because the design should be comfortable for the person wearing the life jacket, the goal is to

10

-2

0

2

4

30

210

60

240

90

270

120

300

150

330

180 0

Figure 2.5: Vertical cross section of the radiation pattern with areas where the gain must be

between -3 and 4 dBi.

integrate the active antenna into the life jacket. Large rigid parts coming out of the life jacket

would obstruct the person wearing it. Another benefit of integration into the life jacket, is

that the system will be protected by the cover of the jacket, which will extend the beacons

lifetime. However, integration into the life jacket imposes a few more constraints on the design

and will be discussed in this subsection. The system will more specifically be designed to fit

into the HI-RISE 275N Solas Life Jacket, made by Mullion and provided by Sioen N.V.

The HI-RISE 275N Solas Life Jacket is designed to make sure that the wearer is always faced

upwards, laying flat on the water. Hence, the antenna will always be faced upwards as well,

assuming that the position of the antenna in the life jacket is well chosen. This means that

the antenna only needs to meet the requirements for the radiation pattern, depicted in Figure

2.5, when lying down horizontally.

First of all, a suitable position for the active antenna in the life jacket needs to be found.

The rather low operation frequency (406 MHz) of the active antenna corresponds with a

wavelength of 74 cm. Hence, a quite large antenna is expected, requiring a large area in the

life jacket. The largest areas of the life jacket are depicted in Figure 2.6.

The first area (1) is situated behind the head. However, this position is not ideal, since the

antenna will radiate towards the head. Because of health issues, and given the large amount

of radiated power, it should be avoided to have radiation towards the body, and certainly not

towards the head. In addition, the absorption and reflection caused by the head, will influence

the antenna performance.

11

The next region (3) is positioned on the chest. Here, already a few items are attached to

the jacket, who will reduce the actual available space, and can also influence the radiation

pattern.

The last region (2) is located on the shoulder, and is also the smallest one. Nonetheless,

this area is preferred because there are no objects attached yet, so the system can be easily

integrated. The antenna will only see the human body on one side, which is the same direction

as the sea. No objects will be seen by the system in the upper hemisphere, which is also the

direction in which the antenna must radiate according to the C/S specifications. The antenna

itself can be chosen in such a way that the radiation pattern is mainly in one hemisphere, by

e.g. choosing an antenna with a groundplane, which will shield the radiation in the direction

of the groundplane.

Figure 2.6: Possible places in the life jacket to integrate the beacon.

Since comfort is the reason to integrate the active antenna into the life jacket, the integration

of the active antenna may not reduce the wearability and the performance of the life jacket.

This implies that the system should be light-weighted, in order to maintain the life jacket’s

floating performance and its ability to maintain the wearer in its flat position. Another way

to keep the life jacket comfortable is making an active antenna that has a low profile, is flat

and flexible, so it doesn’t limit the movements of the person carrying it, with no rigid parts

coming out of the antenna. Another reason to avoid rigid parts is because they are more

susceptible for cracking under tension.

The active antenna must be robust and maintain its performance, even when deployed in the

harsh environments on sea, which are humidity and salt. Although the PLB will be designed

so it can be fully integrated in the life jacket, and will be isolated, these conditions should

still be taken into account. These conditions can change the permittivity of the substrate

12

and thus the resonance frequency. They could also limit the systems lifetime by damaging

components.

Besides the environmental conditions, it must maintain its performance when being pushed

together and being bent in deflated state and still work when the life jacket is inflated. The

presence of the human body needs to be considered in the design procedure of the active

antenna and it needs to be tested under these conditions. Bending the antenna, and the

presence of a human body are very likely to change the resonance frequency of the antenna.

Because the exact conditions in which the system will be used are not exactly known, it is

best to design the antenna with a certain safety margin on the bandwidth, so that it will still

work under the all the conditions mentioned above.

2.2 Wearable Active Antennas

An active antenna is an antenna that contains integrated active electronic components. In

this master thesis, the active electronics consist of a power amplifier that delivers the required

power to the input terminals of the antenna. The power amplifier can also be used to com-

pensate for losses in the antenna if necessary. The integration of the power amplifier onto

the antenna has multiple advantages. First of all, the systems dimension can be reduced by

matching the power amplifier’s output impedance directly to the antenna input impedane,

instead of matching both to 50 Ω. This can reduce the amount of components by making

the use of two matching networks unnecessary. By a proper choice of antenna topology, the

antenna can be reused as heat sink as explained in[11]. Antennas will in general have a high

conductivity and will consist of a large surface exposed to the surrounding media, which can

remove a large amount of heat. Reusing the antenna as a heatsink will result in a more

compact, low-profile and flexible overall system. A circuit and full-wave cosimulation of both

structures can also be useful to see the influence of the electronics on the antenna and vice

versa. [12]

Making the system smaller by optimising the power amplifier and antenna together is an

important benefit, since the goal is to make the system small and light-weighted. The choice

for the amplifier and antenna topology, will mainly be based on the specifications mentioned

in the previous section. Different type of antennas and amplifiers found in literature will be

discussed hereunder.

13

2.2.1 Wearable Antennas

As discussed in the previous section, for wearable applications, the antenna must meet certain

requirements that will limit the amount of antennas that can be used. Because the antenna

needs to be thin and integrable into clothing, it needs to have a planar structure. For this

purpose, a lot of classical thin wire antenna topologies are made planar to make them wearable.

Therefore, a thin, flexible conductor sheet is used instead of a conductive wire. The conductor

sheet can be placed on a textile or flexible substrate, or the conductive material can directly be

integrated into textile. Examples are planar dipoles and monopoles, of which some examples

are shown in Figures 2.7 to 2.10b.

Figure 2.7: Planar dipole antenna.

Figure 2.8: Planar meandered dipole

antenna

Figure 2.9: Planar bowtie antenna

Figure 2.7 depicts a simple planar dipole antenna consisting of two thin rectangular conductive

sheets that are fed differentially. The meandered dipole is a variation of the simple planar

dipole, and results from meandering the arms of a conventional planar dipole. Hence, a more

compact design is achieved, but the electrical length and resonance frequency, are still the

same. [13][14]

A bowtie antenna, seen in Figure 2.9, has a topology similar to the planar dipole antenna.

However, the width of the sheet enlarges towards the ends. Like this, the electrical length of

the antenna will be larger than the physical length of the antenna. The bowtie antenna is

often used for UWB (Ultra Wide Band) applications. To understand why a bowtie antenna

has a very large bandwidth, it should first be considered infinitely long. If the length is

infinite, the antenna itself is only described by the angle between the metal sheets. Normally,

if the dimensions of an antenna are scaled by a certain factor, the resonance wavelength

of the antenna will scale with the same factor. If the shape of the antenna is invariant to

physical scaling, like an infinite bowtie antenna, the antenna must be frequency independent.

14

Of course, the length can not be infinite, but the current will decrease towards the ends

of the antenna because of radiation. If, for a certain wavelength, the current is sufficiently

low at a certain length, cutting it of at this position will not influence the radiation pattern

significantly. [15]

All three types of planar dipole antennas are fed differentially, and no groundplane is used. The

radiation pattern will be omnidirectional like a standard dipole antenna, and the polarisation

will be vertical.

(a) Microstrip feed. (b) Coplanar waveguid feed.

Figure 2.10: Two examples of planar monopole antennas.

In Figure 2.10a and 2.10b, two types of planar monopoles can be seen. The first one is fed by

a microstrip line, and the second one with a coplanar waveguide. A lot of shapes for the flat

monopole can be found in literature. This kind op antenna is also used for UWB applications

by making it electrically small. Both antennas will have an omnidirectional radiation pattern.

[15]

Another common type of planar antennas are microstrip patch antennas. A microstrip patch

antenna is an antenna that consists of a flat rectangular metallic sheet, the patch, and a

larger groundplane. The patch and groundplane are placed on opposite sides of a substrate,

as shown in Figure 2.11. The field distributions of a patch can be seen in Figure 2.12. It

can be seen that the length of the patch needs to be in the order of half a wavelength. The

electrical field is zero in the middle of the patch, and at a maximum or minimum at the sides.

These electric fields do not stop at the edges, but continue to the sides, and will extend above

Groundplane

Patch

Substrate

Figure 2.11: Topology of a microstrip patch antenna.

15

Figure 2.12: Electrical field distribution in a patch antenna.

the substrate. These are called the fringing fields, who cause the patch to radiate, by bringing

the electric field outside the cavity. The actual resonant length of the patch is:

L =1

2

λ0√εr,eff

=λd2, (2.1)

with εr,eff the effective permittivity experienced by the fringing fields, which will be slightly

lower than the permittivity of the substrate because in realistic patch antennas, field lines

and in particular, those of the fringing fields will partly exist in free space. This can also be

seen in Figure 2.12. The actual resonant length will differ slightly from this formula, since

fringing fields will extend the electrical length of the patch compared to the physical length.

A correction factor slightly smaller than one needs to be used, depending on the substrate

thickness and its dielectric constant. [16] [17]

The matching of the antenna to a certain impedance depends on the feeding technique used

to excite the patch. The first manner to excite the patch consists of a microstrip feedline at

the edge of the patch, as illustrated in Figure 2.13a. The drawback of this method, is that the

impedance at the edge is very high. The impedance can be decreased by making the patch

wider, but this makes the antenna a lot bigger, which takes a lot of valuable space. A λ/4

transmission line can be used (fig. 2.13b), to transform the impedance at the edge of patch,

but for low frequencies this also takes a lot of space. This is why most of the time other

feeding techniques are used, where the patch is fed closer to the centre. The impedance of the

patch antenna decreases towards the centre because the current increases, and the electrical

field decreases (Z = VI ). One way to excite the patch near its centre, is by still feeding it with

a microstrip feedline, but now with an inset (fig. 2.13c). Another common method to feed

the patch closer to its centre, is by using a coaxial feed (fig. 2.13d), with the inner conductor

connected with the patch and the outer conductor with the groundplane. The patch can also

be fed by means of aperture coupling (fig. 2.13e). The feedline is located on a substrate on the

16

(a) Microstrip edge feed.

λ/4

(b) Edge feed with λ/4

impedance transformer.

(c) Microtrip edge feed

with inset.

(d) Probe feed. (e) Aperture coupled

feed.

Figure 2.13: Different feeding techniques for microstrip patch antennas.

back of the groundplane. The current on the feedline will excite electrical fields in an aperture

in the groundplane. These fields will then generate currents on the patch. Aperture coupling

is the easiest way to get a good impedance matching, because there are more parameters that

can be chosen freely. This is one of the reasons why it is used a lot for wearable applications,

because they require a larger bandwidth to cope with the conditions in which it will be used,

as described above. [18]

Microstrip antennas are used a lot these days in a large variety of applications. Their stan-

dard low-profile planar structure allows them to be easily integrated into clothing and portable

devices. The substrate for a microstrip patch antenna can be chosen freely to meet the re-

quirements of the applications. For wearable applications, a thin, light and flexible substrate

can be chosen. For portable applications, they are mostly made on the same substrate as the

Printed Circuit Board (PCB), which makes them also very popular for all portable applica-

tions, like mobile phones. The structure of the microstrip patch antenna with a groundplane

makes it very easy to integrate electronics on them. When using aperture coupling, the elec-

tronics can be integrated on the same substrate as the feedline, and be directly coupled to

the feedline, gaining space. [19]

Another property of microstrip antennas that makes them popular for wearable applications,

is that the groundplane minimizes radiation towards the human body , whereas the beamwidth

in the direction away from the human body is generally sufficiently large to have good coverage

17

in that hemisphere. The only drawback of microstrip patch antennas for wearable applications

is their poor efficiency due to losses in the dielectric substrate.

For the thickness of the substrate used for microstrip patch antennas, a trade-off has to be

made. A thin substrate is easier to integrate and will be more comfortable, but will result in

a lower bandwidth. Using a very thick substrate, can on the other hand induce higher order

modes, which needs to be avoided. The permittivity of the substrate is usually chosen very

low since a low εr results in a higher bandwidth, gain and efficiency. [20]

The normal length of microstrip patch antennas is in the order of λ/2, which will make them

vary large for applications with frequencies below 1 GHz. This is why a lot of shortening

techniques are found in literature to make the antenna smaller. The first technique is to use

a shorting plane as illustrated in Figure 2.14. The length of the plane can be varied from a

thin shortening pin or a plane that has the same width as the patch. The shorting plane will

force the fields at that point to zero, whereas the current can be at a maximum because of the

shorting plane. This results in a similar field-current distribution as in one half of a normal

patch antenna. The current and field distribution can be seen in Figure 2.15. [21]

Figure 2.14: Shorted patch antenna.

| |E

| |I

Figure 2.15: Electrical field and

current magnitude in

patch antenna.

Using this shortening technique, the length of the patch can be reduced by half of its original

length, if the shorting plane is extended over the whole width of the antenna, and can be even

shorter if a shorting pin is used. The drawback of this technique, is that there will now be

only fringing fields at one side op the patch. This means that electrical fields can escape the

cavity at only one side, reducing the gain and efficiency of the antenna.

Another technique to reduce the dimensions of a microstrip patch antenna consists of me-

andering the patch or the ground plane of the antenna. Figure 2.16 demonstrates that the

current is forced to follow a longer path because of the meanders, resulting in a larger electrical

length and hence a more compact design.

After choosing an antenna topology that can be integrated into clothing, the materials need

to be chosen. For wearable applications, the chosen materials should be flexible and light

18

Figure 2.16: Current flowing in a meandered patch antenna.

weighted to keep the comfort of the clothes. Flexible materials need to be used for the

substrate as well as for the conductor sheets. Very flexible materials increase the comfort of

the user, but will bent easier, and the antenna must thus work under more extreme bending

conditions.

For some applications the substrate must also withstand extreme conditions, without signif-

icant changes of the material characteristics. For example, antennas for in fire jackets, must

be able to withstand high temperatures, and substrates used in humid environments should

not absorb moisture, since this influences the permittivity substantially.

To integrate an antenna operating at the Cospas-Sarsat frequency (406 MHz), corresponding

with a wavelenght of ± 74 cm, miniaturising techniques will be necessary to enable unobtrusive

integration into a life jacket. All the design specifications mentioned in section 2.1.3 will be

applicable. Previous research has already been done to integrate a C/S-antenna in a life

jacket. The design of a PIFA antenna resonating at 406 MHz, for integration in a life jacket

is described in [22] this antenna has a very large bandwidth, but the total dimensions are

283x200x17.5 mm, which will make full integration into the life jacket impossible. Another,

more compact design of a PIFA antenna for C/S can be found in [23], but the obtained

bandwidth is very small. In [24] a bowtie and meandered dipole are designed for 406 MHz,

and integration into a life jacket. However, these are also large, having dimensions of 270x60

mm and 300x20 mm respectively.

2.2.2 Power Amplifier Design

The term power amplifier is mostly used to refer to the last stage in a circuit, that is used to

amplify a signal to the required output power. There are two main groups of amplifiers, the

linear amplifiers, and the non-linear high-efficiency amplifiers. The most important types of

each group will be discussed briefly hereunder.

The first group of amplifiers are the linear amplifiers. In this category, the active device

operates in its linear region to amplify the signal. When the amplifier is driven correctly,

19

the power gain stays the same, regardless of the input power. The most basic type of linear

amplifier are the Class-A power amplifiers. The transistor is biased in such a way that it

conducts over the entire range of the input cycle. The principle can be seen in seen Figure

2.17. The advantage of a class A amplifier is the easy operation principle, making the circuit

very small. Because the device is always conducting, there is no off-state, and there will

be no problems with charge storage. For this reason, the same active device can operate at

higher frequencies then if it had an off phase. The continuous conducting state also results

in very linear amplification. This is why class A amplifiers are mostly used in applications

where linearity is very important. The drawback of Class-A amplifiers, is that they are very

inefficient. The DC bias point is chosen in such a way that the current and voltage over the

transistor are always bigger than zero. This can be seen on the waveforms in Figure 2.17.

This results in a lot of power dissipation in the transistor. This not only wastes power and

limits the battery lifetime, but the power dissipation in the transistor can lead to enormous

heat generation which can be hard to evacuate.

V I

VD D

I D C

gmVi n

LRf ,choke

CD C,bl ock

VD D

I D C

Vi n

Vout

Vi n Vout

Figure 2.17: Principle of Class-A power amplifier.

A second type of linear amplifier, Class B, uses two active devices. Only one of both active

devices conducts at a time, each during half of a period. This will substantially increase the

efficiency of the power amplifier. Class-B amplifiers are also linear amplifiers in the sense

that variations in amplitude of the input signal will still be the same at the output. However,

using two active devices, will induce distortion. For example, when there is a certain amount

of time where none or both of the transistors are conducting, or if the two transistors don’t

behave identical. The principle of Class-B amplifiers can be seen in Figure 2.18.

The second group of amplifiers are the high-efficiency power amplifiers. The active device is

driven as a switch instead of being used as a voltage controlled current source. In the ideal

case, either the current going trough the transistor, or the voltage across the transistor, is zero.

20

Vi n Vout

VD D

i 1

i 2

i 1 − i 2

i 1

i 2

i 1 − i 2

Figure 2.18: Principle of Class-B power amplifier.

vi n

v

C

vDL 0

C0

vout

LRF ,choke

VD D

vi n

iD

D Di

vout

Figure 2.19: Principle of Class-E power amplifier.

Hence, in the ideal case, as the power dissipation equals the product of the voltage and current

in the transistor, the power dissipation in the active device will be zero. However, non-ideal

switching, parasitics, finite conductivity, leakage current and other non-ideal characteristics

of the active device, will reduce the actual efficiency. But despite these effects, the efficiency

will still exceed those of linear amplifiers.

Again, many different types of high-efficiency amplifiers, using the switching principle, exist.

The type that is most often used in wearable applications is the Class-E amplifier. This

amplifier is characterised by a very high efficiency, with losses a factor 2.3 less then for class B

amplifiers. In addition, class E amplifiers require only one active device for proper operation

and have the lowest amount of components in their circuit compared to other high-efficiency

amplifiers. The principle of class-E operation can be seen in Figure 2.19. [25] [26]

The transistor is driven as a switch, and an output bandpass filter will filter out all higher

harmonics, except for the base tone. Variations of class E-amplifiers add more filters to short

harmonics to ground and ameliorate the harmonic suppression. This can be very useful, but

21

is less used in wearable applications since it increases the amount of required components.

Class-E amplifiers are by far the most common type of amplifier used in portable applica-

tions. Having a a very high efficiency is the most reason for this. In portable applications,

the electronics will almost always depend on a battery, which has a limited lifetime, and is

preferable not too big. Hence, a high efficiency is desired to enhance battery life time and/or

reduce battery size. In addition, a higher efficiency reduces the heat generation, resulting

in less problems with heat evacuation. Another reason is the small topology of this type of

amplifiers.

22

Chapter 3

Active Antenna Design

3.1 Design Goals

A C/S compatible emergency beacon consists of two functional elements, a digital message

generator and a modulator with 406 MHz transmitter. Since this master thesis only describes

the design of a C/S compatible active antenna, only the modulator and transmitter will be

discussed hereunder. The active antenna will consist of a power amplifier that amplifies the

modulated signal from the signal generator to the required power of 5 Watt, and an antenna,

capable of transmitting signal at 406 MHz.

Transmitted frequency

A number of channels have been defined for Cospas-Sarsat, all in the [406-406.1] MHz band.

The active antenna should fulfill all specifications for this frequency band. Each channels

is defined by its centre frequency, and shall be set in accordance to the channel assignment

table of [8]. This carrier frequency should not deviate more than 5 kHz from its nominal

value over a period of five year. Other limitations on the short- and medium-term variations

can be found in [9]. Frequency stability is of major importance to acquire accurate Doppler

measurements, which will determine the position of the person in distress.

Given a certain carrier frequency, the in-band spurious emissions can not exceed the levels

specified by Figure 3.1. When designing both antenna and power amplifier, the spurious

emissions should be controlled to keep them under these levels.

Data encoding and modulation

23

Figure 3.1: Spurious emission mask for the [406.0-406.1] MHz band that is required for each

individual channel [9]

The data is biphase-L encoded. The phase will be positive or negative 1.1 ± 0.1 radians peak,

referenced to the unmodulated carrier. Since phase modulation is used, linearity of the power

amplifier will not be of concern. The rise and fall times of the phas must be between 150 ±100 µs. Since the modulation is already done by the modulator, the implication for the design

of the active antenna, is that it cannot increase the rise and fall time in such a way that the

C/S requirements aren’t met anymore.

Transmitter output power

The transmitted output power needs to be within the limits of 5W ± 2 dB, referenced to a

50 Ω load impedance. This means that the power delivered to the antenna terminals must

be between 35 and 39 dBm. A 0.5 seconds lasting message will be send every 50 seconds.

The output rise time measured between the 10% and 90% power points needs to be lower

then 5 ms. In addition, the power must rise linear, which means that the power must be

zero (or below -10 dB), 0.6 ms before the rise time measurements. This burst power must be

maintained for at least twenty-four hours.

The transmission duration of the signal is 440 ms for a short message, and 520 ms for a long

message. The repetition period is between 47.5 and 52.5 seconds, and is chosen randomly.

The implications for the design are that the amplifier must be able to work 0.5 seconds every

50 seconds, without heating up too much. This equals an average duty cycle of ± 1%. Because

of the low duty cycle, it can be assumed that the transistor will be back on the environmental

temperature when the next burst is sent. Hence, the amplifier needs to work during 0.5 ms.

24

Gain Between 17 and 21 dB

PAE > 50 %

Linearity Not required

Spurious emissions According to Figure 3.1

Input reflection coefficient |S11| <-10 dB

Table 3.1: Design goals of the power amplifier, given an input power of 18 dBm.

The signal generator used is the “MRB1002” module provided by Syrlinks. This signal

generator delivers a 18 dBm output power to a 50 Ω load. This means an amplification of at

least 17 dB is necessary, given this input power. A small circuit is required such that it can

easily be integrated into the life jacket. Besides, the amplifier must be able to work during

twenty-four hours. Because the system depends on a battery, and the required output power

is large, high efficiency will be favorable. The power efficiency will further on be defined as

the Power Added Efficiency (PAE) which is defined as:

PAE[%] = 100Pout,RF − Pin,RF

PDC,total(3.1)

In this master thesis, the goals will be set to obtain an efficiency of at least 50 %. An overview

of the design goals for the amplifier are listed in Table 3.1

Antenna characteristics

As already described in Chapter 2, the antenna needs to fulfill specific requirement imposed

by [9], in order to be compatible with the Cospas-Sarsat standard. For all azimuth angles and

for elevation angles between 5 and 60 , the antenna must have a hemispherical radiation

pattern, with a gain between -3 dBi and 4 dBi in 90% of this region. The polarisation can be

either circular or linear. The VSWR needs to be smaller than 1.5:1 for the [406-406.1] MHz

band. This VSWR corresponds with a reflection coefficient lower than -14 dB. This condition

is stricter then the general accepted standard, which is obtaining a reflection coefficient, or

|S11|, below -10 dB. In this master thesis, obtainin an |S11| < -10 dB will be set as the design

goal.

According to the C/S specifications, only a bandwidth of 100 kHz is required. Given the

application in which the active antenna will be used, the possibility exists that the antenna

will be bent, or is affected by the environmental conditions, which can cause a shift in the

resonance frequency of the antenna. This is why the goal in this master thesis will be to

acquire a bandwidth of at least 4 MHz, in order to have a safety margin. For integration

into a life jacket, the antenna sizes should be kept minimal, and the materials used should be

25

|S11| <-10 dB

Bandwidth 4 MHz

Centre frequency 406 MHz

Radiation pattern According to 2.5

dimensions < 100x200x15 mm

materials Flexible

Table 3.2: Design goals for the antenna.

flexible. An overview of the design goals for the antenna can be found in Table 3.2 [9]

3.2 Active Antenna Topology

Because the antenna needs to be integrated into a life jacket, and electronics will be integrated

onto the antenna, the most logical choice was to use a microstrip patch antenna. As described

in the previous chapter, microstrip patch antennas are preferred for wearable applications

because of their planar structure, hemispherical radiation pattern, and because the topology

enables easy integration of electronic circuitry. When aperture coupling is used, the substrate

and groundplane of the antenna feed structure can be reused for the power amplifier.

A class E power amplifier was chosen because of its high efficiency and because its topology

only requires few components. Class-E amplifiers are non-linear devices., but since the signal

is phase modulated, the non-linear nature of class E power amplifiers is allowed. On the

other hand, the spectrum will need to be checked if harmonics of the signal frequency are

sufficiently suppressed, and it needs to be verified that the spurious emissions don’t exceed

the levels specified by C/S.

3.2.1 Antenna Topology

The antenna topology chosen for this master thesis is an aperture coupled shorted patch

antenna. Aperture coupling is chosen as feed technique because it results in a larger impedance

bandwidth then other feeding techniques. Furthermore, it doesn’t require a pin to be inserted

from the groundplane to the patch, which would be a weak point in the antenna that is

susceptible for cracking, and it would decrease the antenna’s flexibility. Aperture coupling

also makes it easier to integrate a power amplifier onto the antenna, since it could be placed

on the same substrate as the feedline.

The length of a normal patch antenna is in the order of λ/2, which corresponds with a length

of 37 cm for the C/S frequency, which is too long for wearable applications. This is why a

26

Figure 3.2: Topology of a planar inverted-F antenna (PIFA) with its electrical field distribution.

shorting plane will be used to reduce the dimensions of the antenna.

If a shorting plane is be placed at the middle of the patch antenna, it would create a mirror-

effect in the patch. The shorting plane will force the electrical fields to zero, while the currents

will be at their maximum. At the edge of the antenna, the electrical field will be maximal,

and there will be no current flowing. Without a shorting plane, the electrical field will also be

at a minimum in the middle of the patch, and maximal at the edges, and the current maximal

in the middle, and minimal at the edges. The shorting plane will thus not alter the current

and electrical field distribution substantially.

If now one half of the patch is removed, and the shorting plane is kept, the electrical field

and current distribution would still be the same as before this half was removed. This is

due to the mirror effect of the groundplane. The field distributions in a normal patch and

shortened patch can be seen in Figures 2.12 and 2.15 respectively. Since the field and current

distribution are the same, the resonance frequency will stay the same too, but having only

half of the original length. Hence, a more unobtrusive integration can be achieved.

A coaxial fed patch antenna with shorting plane was already designed in [23]. This design is

also called a planar inverted F antenna or PIFA antenna because of its structure, which can

be seen in Figure 3.2. The measured |S11| of this antenna can be found in Figure 3.3.

On this graph, it can be seen that the bandwidth is lower then 4 MHz. Although only a very

small bandwidth of 100 kHz is required for the Cospas-Sarsat system, using an antenna with a

small bandwidth can give certain problems when making it by hand and using it for wearable

applications, where it will be bend, and a human body will be present. Small variations in

the fabrication procedure of the antenna, bending of the antenna, or deploying the antenna in

the vicinity of the human body, will cause a shift of the resonance frequency of the antenna.

If the antenna exhibits a very small bandwidth, using the antenna under these conditions will

almost certainly mean that the Cospas-Sarsat frequency will fall out of the resonance band of

the antenna. It is thus better to have a larger bandwidth, as a safety margin for design and

environmental variations. Therefore, the PIFA design of [23] will be redesigned in this master

thesis to obtain a larger bandwidth by using aperture coupling instead of coaxial feeding.

Aperture coupling, as was illustrated in Figure 2.13e, is based on a feedline that generates

27

Figure 3.3: Measured |S11| of a coaxial fed PIFA antenna.

electrical fields in an aperture by magnetic polarisation. This aperture is present in the

groundplane. The electrical fields in the aperture will generate currents on the patch. Another

advantage of using aperture coupling is that two substrates can be used, where the substrate

for the patch can be made thick and with low permittivity to have a large bandwidth, and

the substrate for the feedline can be made thinner to have a tightly coupled field without

introducing spurious emission, or exciting higher order modes.

When using aperture coupling, impedance matching is easily controlled by optimising the

parameters of the antenna feed structure, and good impedance matching can be obtained

for a larger frequency band compared with other feeding techniques. This is because there

are more parameters that can be optimised for ideal impedance matching, being the shape,

size and position of the aperture, the characteristics of both substrates, and the width and

height of the feedline. Another reason why aperture coupling is preferred over coaxial feeding,

is because probe feeding means that a coaxial feed needs to be inserted into the substrate,

and soldered. This will form a weak point in the antenna that is susceptible for cracking.

When using aperture coupling, the power amplifier can directly be connected to the feedline,

without using cables. The only drawback of aperture coupling is the increased complexity of

the design, and the need for the second layer, which makes the design and production process

more difficult. [27] [28]

Because reusing the PIFA design of [23] and replacing the coaxial feed structure by an aperture

coupled feed structure, led to unsatisfactory results, it was decided to start from a new design,

based on the design of an aperture coupled shorted patch antenna, described in [29]. This

28

Figure 3.4: Structure of an aperture coupled shorted patch antenna.

book states that an aperture coupled PIFA antenna is a very promising structure to become

a small, but broadband antenna.

Figure 3.4 shows the geometry of the patch antenna. The size and shape of the slot influences

the coupling level and the back radiation. In general, smaller aperture areas results in lower

back radiation, and thus in a higher efficiency. On the other hand, coupling is increased by

using longer or wider rectangular slots. For simple rectangular slots, the transverse electric

fields must vanish at the end of the aperture. However, a longer or wider slot would increase

the input impedance. It can be noticed that the aperture of Figure 3.4 has an H-shape.

The advantage over a simple rectangular aperture is that by adding slots at the end of the

rectangular slot, the fields in the slot become more uniform, hence the coupling increases. By

using this principle, thinner slots can be used, to obtain better impedance matching. Other

shapes of apertures, to increase the coupling, are the dog bone and bow tie slot, which can be

seen in Figure 3.5. Best coupling is reached when the slot is placed at the centre of the patch,

where the patch has its maximum magnetic field, and maximum magnetic coupling will be

obtained. [30]

(a) H slot. (b) Dog bone.

(c) Bow tie slot.

Figure 3.5: Different shapes of apertures that can be used to increase coupling.

29

The width of the feedline will set the characteristic impedance of the feedline, and will thus

influence the input impedance of the antenna. Furthermore, it will also have an influence on

the coupling level. Thinner feed lines will, to a certain degree, result in a better coupling. The

length of the stub, which is the part of the feedline that extends beyond the aperture, will

influence the input impedance and the impedance bandwidth. Making the length shorter, will

move the impedance to capacitive side on the Smith chart. Therefore, the length should be

optimised for a given structure. Thicker substrates will normally result in a more inductive

impedance, and this can be compensated with the stub length. The coupling will be best

if the feed line is positioned perpendicular to the slot, and at the centre of the patch, since

the fields in the slot will then be symmetrical. Making the angle between the aperture and

feedline less then 90 will decrease coupling, as will positioning the feed toward one of the

edges of the slot.

The substrate for the patch and feed structure can be chosen differently. For the antenna,

a substrate with a low permittivity will result in a larger bandwidth and higher efficiency.

Making the substrate thicker will also increase the bandwidth and the efficiency because there

is a larger aperture through which power can be radiated. On the other hand, using a too

high substrate will result in exciting surface waves, and will induce spurious emissions, and

will reduce the coupling with the feedline as well as the efficiency. A thicker substrate will

also result in a more inductive impedance, but, as mentioned before, this can be compensated

by adjusting the length of the feedline. For the feed structure, a thinner substrate will result

in less spurious emissions, but in higher losses. The height and dielectric constant of this

substrate will define the width of the feedline, if it is designed for 50 Ω. The height and

permittivity should be chosen to have good microstrip characteristics.[31] When choosing the

substrates, it also has to be taken into account that the system will be used in a life jacket,

and the materials should be chosen accordingly.

The computer aided design of an aperture coupled patch antenna with a shortening plane for

a frequency of 406 MHz will be explained further on, where the principles of this section will

be applied to design a small antenna that has a large bandwidth. The available materials and

those chosen will be explained in the section about the bill of materials.

3.2.2 Power Amplifier Topology

Out of the different types of amplifiers, the class-E amplifier was chosen because of its high

efficiency and small topology. Class-E amplifiers are non-linear amplifiers, which means that

variations in signal amplitude at the input will not be entirely reproduced at the output, since

the gain will decrease for increasing input power. However, this is not a problem since the

signal sent out by the Cospas-Sarsat system is phase modulated, and variations in amplitude

30

don’t need to be reproduced at the output. A high-efficiency, non-linear amplifier was chosen

because there is less heat dissipation in the transistor, which could damage the system if the

heat is not properly evacuated by means of a heat sink or air circulation. Large heat sinks

make the system a lot bigger which is a large drawback in portable or wearable applications.

In addition, choosing for a highly efficient class E power amplifier will extend battery life or

will reduce the battery size. Both results are beneficial for wearable applications.

Class E topology is preferred over other non-linear amplifiers because of it small structure,

and higher efficiency. The basic circuit consists of only one transistor that will be used as a

switch, a shunt capacitor that represents the transistors output capacitance, as well as other

parasitic shunt capacitances of the circuit, a high reactance RF choke, and an LC tank to

filter out higher harmonics of the resonance frequency. Another advantage is that there are

design equations available that calculate the components values for a certain frequency and

output power. These equations will also show the effect of variations in component values.

Figure 3.6: Ideal waveforms for obtaining highly efficient class-E operation

The principle of high efficiency amplifiers is based on limiting the power dissipation in the

active device by minimising the voltage across the device (vds in Figure 3.7) when current flows,

and reducing the current (ids) as when a voltage exists across the transistor. Furthermore

the duration of the transition period, in which the voltage as well as well as the current are

both non-zero, should be kept as short as possible. By doing this, the average voltage-current

product will be small, and, hence, also the power dissipation. This principle will be applied

by driving the active device as a switch, and keeping the transition periods short by choosing

a good transistor, and driving it correctly. Optimising the load networks transient response,

can also reduce the switching time. Besides reducing the transition periods between off and on

state, power dissipation can further be reduced by forcing the voltage to zero before current

starts to flow at switch turn on, and delaying the voltage at turn off till current has reduced

31

almost to zero. If the voltage slope is also zero at turn on, the voltage will remain zero during

the short interval after turn on, which permits slight delay on turn off time. These last two

conditions can be expressed by the following equations:

dvs(t)

dt

∣∣∣∣t=Ts

= 0, (3.2)

vs(t) = 0|t=Ts , (3.3)

where Ts is the time at which the switch is turned on. The ideal waveforms for the current

and voltage depicted on Figure 3.6. This principle was first presented by Sokal in 1975 [32].

Figure 3.7: Schematic representation of the operation principle of class E amplifiers.

The basic topology of a class E amplifier, representing the operation principle, can be seen

in Figure 3.7. L1 is the RF choke to separate the AC-current from the bias current. Capac-

itor C represents the transistor’s output capacitance, as well as other parasitic capacitances.

Capacitor C0 and inductor L0 form a bandpass filter to suppress harmonics of the switching

frequency, and ZL represents the actual load. When the transistor is switched off, C will need

time to charge, keeping the voltage low some time after turn off. The complete load network

will have the response of a damped second order system. When the Q-factor of this system

is chosen correctly, there will be no overshoot, bringing the voltage under zero, and the the

voltage slope will become zero as well.

The conditions above are obtained by choosing the correct values for C, ZL, C0 and L0. These

component values will depend on the choice of the Q-factor, QL, of the load network. For

the choice of QL, a trade-off has to be made between high suppression of the harmonics (high

QL), and high efficiency to get ideal damping (low QL).

From the principles above, formulas can be found that calculate the best values for C, C0 and

32

L0, given the required output power, switching frequency, the chosen bias voltage, QL and

L1.[25] These equations are:[33] [34].

ZL = 0.5768

(V 2DD

Pout

).

(1− 0.451759

QL+

0.402444

Q2L

)(3.4)

C =1

5.44658ωZL

(1 +

0.91424

QL− 1.03175

Q2L

)+

0.6

ω2L1(3.5)

C0 =1

ωR

(1

QL0.104823

)(1 +

1

QL − 1.7879

)− 0.2

ω2L1(3.6)

L0 =QLR

ω(3.7)

By solving these equations, ideal class-E operation with a high efficiency can be achieved.

Off course, when working with higher frequencies, a lot of parasitic effects, interference, and

non-ideal components will impact the performance of the circuit. This is why the circuit will

be simulated and optimised, taking into account these effects.

3.3 Bill of Materials

When choosing the materials which will be used to assemble the active antenna, great care

must be taken to obtain a flexible, light-weight and robust design that maintains its perfor-

mance when it is slightly bent, or when it is applied on-body.

Choice of antenna materials

First of all, substrates have to be chosen for the microstrip patch antenna. Since aperture cou-

pling was chosen to feed the patch, two different substrates need to be chosen. The importance

of the substrate materials, and their effect on antenna characteristics was already explained

in the previous section (3.2.1). Both substrates need to be flexible and light weighted, and

breathable, so they can easily be integrated, without obstructing the person wearing it. Using

flexible materials also reduces the chances of cracking when the antenna is bent. Four different

flexible substrates were available to use as substrate for the antenna, and will be discussed

hereunder.

The first material is a 3.94 mm thick black foam with a relative dielectric constant, εr =

1.58, and a loss tangent, tan(δ), of 0.038. The second possible substrate is a thicker black

foam, having a height of 11 mm. This foam has a permittivity of 1.15, and a loss tangent

33

Material name thickness (mm) εr tan δ

polyurethane 3.94 1.58 0.038

black foam 11 1.15 0.004

3D substrate 10 To be specified To be specified

Aramid 0.4 1.84 0.015

Table 3.3: Overview of available materials.

of 0.004. Both materials where characterised at 406 MHz, in [23]. The characterisation was

performed by making a patch antenna, and then fitting the simulated |S11| on the measured

one by adjusting the substrate parameters in the simulation. The black foams are made out

of protective, fire retardant polyurethane, and are manufactured by Javaux. This material is

used for firefighter clothes. This closed-cell expanded rubber possesses good chemical stability,

low moisture regain and good resistance against oil and solvents [35].

As a alternative for the 11 mm black foam substrate, a 3D-textile with a thickness of 10

mm could be used as substrate. 3D-textiles mainly consist of air, resulting in a relative

permittivity close to one, and a very low loss tangent. Hence, it is very suitable as antenna

substrate. However, the exact εr and tan δ still remains unknown, and experimental tests

need to be done to determine them. Large quantities of this substrate were unavailable during

this master thesis, which made it impossible to characterise the substrate.

A last available substrate is a 0.4 mm thick aramid fabric, that is often used in protective

clothing. It has a permitivitty εr = 1.84, and a tan(δ) of 0.015 [12]. An overview of the

available materials and their characteristics is given in Table 3.3.

The effect of the substrate thickness and dielectric constant on the antenna characteristics

was already described in the previous section. For the patch, one of the thick substrates will

be chosen, to have a large bandwidth. The thickest substrate available is the 11 mm thick

black foam. This material has a dielectric constant of εr = 1.15, which is very low. A low εr

is favorable for both the bandwidth and the efficiency. This knowledge makes this substrate

an attractive choice for the patch substrate, and will be used for the design further on. The

3D textile is only slightly thinner, and will also have a very low εr, so can be considered as

an alternative for the black foam substrate.

For the antenna feed structure, a thinner material will be used. Using the 3.94 mm substrate

would make the total antenna thickness 15 mm, which is rather thick. Thick substrates also

result in larger spurious emissions of the feedline, which can cause problems to meet the C/S

requirements on spurious emissions. Besides, it could also excite higher order modes in the

microstrip feedline. In addition, integrating electronic circuitry on such a thick substrate

would require high vias, introducing a lot of undesirable parasitic inductance. Therefore, the

34

aramid fabric will be used instead. However, this substrate will result in a ver thin feedline,

and will thus have a low power handling capacity. Furthermore, to obtain good coupling, the

substrate thickness must be a certain fraction of the wavelength. This is why it was opted to

use two layers of aramid.

Not only the substrate must be flexible, but also the conductive parts of the antenna. A pure

copper polyester taffeta fabric was available for realising the conductive parts. This material

has a surface resistivity of 3mΩ/sq ([23]), and is very flexible, so will not affect the flexibility

of the entire antenna.

The conductive layers will be glued to the substrates by means of thermally-activated adhesive

sheets, called vlies-o-fix.

Power amplifier materials and components.

The goal of this masterthesis is to integrate the power amplifier on the antenna feed substrate,

by reusing the antenna feed substrate as the substrate for the power amplifier. For a first

prototype, the power amplifier will be designed on a standard 1 mm FR4 substrate, and if

the design works, it could be redesigned for 0.8 mm aramid. The copper paths can be etched

on a very thin polyimide layer, which can be glued to the aramid. The possibility exists to

also etch the antenna feedline on the polyimide, so the power amplifier can immediately be

connected to the antenna without the need for extra soldering

The most important component for the power amplifier is the transistor. The transistor used

in the final design is the “AFT05MS031NR1” of freescale semiconductor. However, the

reason why this transistor is chosen, will be explained in section 3.4.2.

The choice for the other components was mainly based on their footprint, having a self

resonance frequency well above 406 MHz, and being capable of handling the applied power.

The Q-factor and the parasitics of the components were also considered, and chosen in such

a way to obtain optimal efficiency.

3.4 Computer Aided Design of the Active Antenna

Two different simulation programs were used to design the active antenna, namely CST mi-

crowave studio and Agilent’s Advanced Design System (ADS). The antenna will be simulated

with the ADS momentum solver as well as with the full 3D CST simulator. ADS momentum

is actually a planar wave solver, but is capable of handling more than one layer and is there-

fore called a planar 3D solver. It will calculate the EM-fields by using a frequency-domain

method of moments. CST is a 3D full wave simulator, that will calculate S-parameters as

well as near- and far-fields very accurate. It has both a frequency and time domain solver.

35

Figure 3.8: Topology of an aperture coupled shorted patch antenna with a H-shape slot.

The drawback of CST are the long calculation times compared to the planar 3D momentum

solver of ADS. This is why the antenna will first be designed in ADS, and then further be

optimised in CST.

The power amplifier will be completely designed in ADS. First, only the circuit will be drawn

and optimised, and afterwards the PCB layout will be drawn and simulated in ADS momen-

tum. By modelling the passive interconnections in the ADS Momentum full-wave environment

and the lumped elements in the ADS circuit simulator, a joint circuit/full-wave co-simulation

was set-up to obtain accurate simulation results.

3.4.1 Design of the Flexible Antenna

The design of the antenna was based on the principles of [29]. The antenna topology and

parameters can be seen in Figure 3.8. Since this design was made for a centre frequency of

1715 MHz, and with different substrates, the length of the patch was recalculated for the

black foam substrate of 11mm. According to formula 2.1, and presuming that the length

can be reduced by half when a shorting plane is used, this results in a patch length of 172.2

mm. The initial width of the patch was chosen 230 mm, in order have the same length/width

proportion as in the book. The feed line width was calculated to have a 50 Ω impedance on

36

350 360 370 380 390 400 410 420 430 440 450−50

−45

−40

−35

−30

−25

−20

−15

−10

−5

0

frequency [MHz]

|S11

| (dB

)

7.8 MHz

Figure 3.9: Simulated |S11| of the initial wide patch

antenna.

Patch Wp 150

Lp 146.4

Aperture Ha 74.8

S1 34.1

S2 37.9

Wa 55.9

Th 1.9

Tv 3.6

feed line t 40.5

Wf 2.4

Table 3.4: Parameters of the initial

wide aperture coupled

patch antenna (mm).

the chosen substrate, resulting in a width of 2.6 mm. The length of the feed line was tuned

by performing parameter sweeps and looking at the antenna input impedance on the Smith

chart. The length was chosen in such a way that it made the input impedance real at the C/S

centre frequency. The H-aperture was centered at the middle of the patch, and the dimensions

of the slot were initially scaled by a factor of 1715/406, according to the desired frequency

shift.

The first design was made in ADS momentum with an infinite groundplane, because this is

much faster than simulating an antenna with finite groundplane in CST microwave studio.

The initial patch with the dimensions above was drawn, and the design was further optimised

by doing parameter sweeps on all variables. The resulting parameters and simulated |S11| for

this design can be found in Table 3.4 and Figure 3.9, respectively.

The resulting bandwidth, for which |S11| < −10 dB, is 7.8 MHz, which is much smaller than

was hoped for. The bandwidth can be increased by increasing the height of the substrate.

The height at this moment is 0.015λ, and the rule of thumb is that the height can be increased

to 0.05λ to have a good bandwidth. Two layers of the black foam substrate could be glued

together and used as a substrate. The substrate would then have a height of 22 mm, which

is too thick for wearable applications. Hence, the substrate height is kept on 11mm.

Furthermore, the width of the patch is now 150 mm, which is too big to fit into a life jacket.

This means that the patch width still needs to be narrowed. Making the patch narrower

will decrease the bandwidth even more, since it will decrease the antenna volume, and the

surface of the radiating planes. For this reason, the design was iteratively made more narrow,

adapting the other parameters to keep the resonance frequency at 406 MHz. When making

37

350 360 370 380 390 400 410 420 430 440 450−30

−25

−20

−15

−10

−5

0

frequency [MHz]

|S11

| (dB

)

4.2 MHz

Figure 3.10: Simulated |S11| of the narrow patch

antenna.

Patch Wp 50

Lp 145.8

Aperture Ha 82

S1 25.7

S2 35.1

Wa 40

Th 1.7

Tv 3.6

feed line t 28.3

Wf 2.4

Table 3.5: Parameters of the narrow

PIFA antenna (mm).

the patch more narrow, it could be noticed that the increase in bandwidth that was obtained

using an aperture coupled feed the antenna was partially lost. By requiring a minimum

bandwidth of 4 MHz, the final parameters can be found in Table 3.5. The simulates |S11| for

the narrowed antenna can be seen in Figure 3.10.

Due to the small bandwidth, other techniques to increase the bandwidth were explored. Most

of these techniques consist of making the antenna asymmetric to introduce other resonance

modes. For example, the shorting plane was made smaller, to a width of 10 mm, placed at

the left of the antenna. Another technique was cutting out a triangle of the patch. However,

these techniques did not improve the antenna performance significantly, and they would only

increase the complexity of the design, and thus the amount of parameters. Hence, the design

of antenna of Figure 3.8 with the parameters of Table 3.5 was chosen.

Because ADS works with an infinite groundplane, the current ADS design was drawn in CST,

with a finite grounplane. Normally it is preferred to have a square groundplane with length at

least 0.8λ ([36]) to have a good bandwidth, but this would mean dimensions of almost 60 cm

in this case, which is too big for wearable applications, and would make the miniaturising of

the patch pointless. To have no influence on the resonance frequency, the groundplane should

have a length bigger than 0.2λ ([36]), which corresponds with 15 cm. A square groundplane

of 15x15cm2 would again be too large, so the effect of the finite groundplane should be taken

into account. To simulate the effect of the groundplane on the antenna characteristics, three

different groundplanes sizes were tested.[36]

The simulation results of these three simulations can be seen in Figure 3.11. It can be noticed

that none of these results are the same as the simulation results of ADS. For the groundplane

sizes of 300x300mm2 and 500x500mm2, the resonance frequency is the same, namely 401

38

350 360 370 380 390 400 410 420 430 440 450−45

−40

−35

−30

−25

−20

−15

−10

−5

0

frequency [MHz]

|S11

| (dB

)

Groundplane = 80x190mm2



Figure 3.11: ADS antenna design simulated in CST with different groundplane sizes.

MHz. It can be asumed that from a certain size, increasing the surface of the groundplane

will have little or no effect on the input reflection coefficient. But even for large groundplane

dimensions, there is still a mismatch between the results from ADS and the time domain

solver of CST, although they are close to each other. The differences can be explained by

the different solving mechanisms between the simulators. The comparison between the ADS

antenna and the CST antenna with groundplane size 500x500mm2 can be seen in Figure 3.12.

The third simulation that can be seen in Figure 3.11, is for groundplane dimensions of

80x190mm2. These are the dimensions that are desired for this applications, because they

allow integration into a life jacket. It can be seen that for these dimensions, the central fre-

quency is shifted to 388 MHz. With these dimensions, the groundplane cannot be assumed

big enough anymore to have no effect on the resonance frequency.

However, since the antenna in CST only had a frequency shift compared to the ADS simu-

lations, but has a bandwidth of 6 MHz, this design was used as a start for optimising the

antenna in CST. The results after optimisation can be seen in Figure 3.13. The dimensions

of the antenna can be found in Table 3.6. The Obtained bandwidth with these parameters is

6.5 MHz, and the |S11| at 406 MHz is -18 dB, which satisfies the C/S requirements.

Besides the |S11|, the radiation pattern, radiation efficiency and total efficiency were also

calculated in CST. The radiation efficiency is 62.2 % for 406 MHz, and the total efficiency,

which includes the losses due to impedance mismatch, is 61.5 %.

The 3D simulated radiation pattern can be seen in Figure 3.14. As expected, the area where

the largest gain is reached, is the direction pointing away from the patch, which is the z-

39

350 360 370 380 390 400 410 420 430 440 450−30

−25

−20

−15

−10

−5

0

frequency [MHz]

|S11

| (dB

)

ADS antennaCST antenna

Figure 3.12: ADS antenna design compared to CST simulation with groundplane of size

500x500mm2.

direction on the plot. However, the maximum gain is not reached in the z-direction, but is

tilted slightly towards the positive y-axis, away from the shorting plane. The gain in that

direction is 0.27 dBi. The required gain for 90% of the elevation angles between 5 and 60

is minimum -3 dBi and maximum 4 dBi. Since the maximum gain is below the upper bound,

this condition will not be of concern. To check whether the requirement on the minimum gain

is met, 2D plots of radiation pattern can be seen in Figure 4.7a and 4.7b, showing the gain

in dBi for cutting planes xz and yz in the way they are defined in Figure 3.14.

For the radiation pattern in the xz-plane, the gain is above -3 dBi for all angels, so these angles

are not a problem. However, for the yz-plane, the gain drops below -3 dBi for elevation angles

between 64 and 156 , and between 238 and 321 . This angles are part of the specified

region for which the gain must meet the requirements. In total, the radiation pattern goes

below -3 dBi for the specified area during 57 , or thus over 52% of the area. This is below the

required 90% of C/S. However, this 52% will be partially compensated by the full coverage in

the xz-plane. To know what the complete coverage for these angles is, the average over more

values of φ should be calculated.

Because the design will be made by hand, the optimised antenna was also simulated for small

variations in parameters, as will be shown and discussed hereunder. Small deviations on the

parameters shouldn’t affect the antenna performance too much, since they are likely to happen

when the antenna is made by hand, even when this is done with greatest care. Simulation

results for small variations in the dimensions of the patch can be seen in Figure 3.17. Since

the goal was set to have a |S11| < -10 dB, this boundary is depicted on the graphs by a gray

line. However, since C/S requires a |S11| < -14 dB, this boundary is also added on the graph

40

350 360 370 380 390 400 410 420 430 440 450−40

−35

−30

−25

−20

−15

−10

−5

0

frequency [MHz]

|S11

| (dB

)6.5 MHz

Figure 3.13: Simulated |S11| of the antenna in CST.

Patch Wp 48

Lp 140

groundplane Wg 80

Lg 190

Aperture Ha 72

S1 25.7

S2 35.1

Wa 40

Th 1.8

Tv 3.6

feed line t 29

Wf 2.7

Table 3.6: Dimensions of the final

design in CST (mm).

Figure 3.14: 3D radiation pattern simulated in CST.

41

Figure 3.15: Simulated radiation pattern in the xz-plane.

in black. 406 MHz is also annotated on the graphs, by means of a vertical line.

As can be seen in Figure 3.17a, changing the length of the patch, changes the resonance

frequency, as expected. For variations op to 1 mm, the S11 at 406MHz stays below -10 dB.

The reflection coefficient doesn’t stay below the -14 dB that is required for C/S, if the length

is increased by 1 mm. It can also be seen that the matching becomes worse when variations

become larger, which could also be predicted, since the feed structure is designed for a certain

patch length. Variations in Wp (Fig. 3.17b), do not influence the resonance frequency, but

only the matching. Besides the width of the patch, the length and width of the feedline, and

the position of the aperture, will also influence matching, as was also expected. However, as

can be seen in Figures 3.17c, 3.17d and 3.17f, none of the small variations, resulted in a |S11|above -14 dB. The last parameter that was varied is the εr of the black foam substrate. It can

be seen that small changes in εr will result in considerable shifts of the resonance frequency.

It is thus very important to have a good characterisation of the substrate. Besides, the εr can

vary when the substrate absorbs moisture. This can be avoided by chosing materials with a

low moisture regain. In the case of integration into a life jacket, the antenna will be isolated,

and moisture will not cause problems.

Hence, except for variations in εr and patch length, which will cause a frequency shift, small

variations in the design are tolerated, since they all result in a |S11| < -14, thus meeting the

C/S requirements.

42

Figure 3.16: Simulated radiation pattern in the yz-plane.

3.4.2 Design of the Power Amplifier

The type of amplifier that will be used (class-E) and its operation principle was already

discussed in the section 3.2.2. One of the most important things to choose when designing a

power amplifier, is choosing the transistor. A good amplifier can only be designed if a proper

transistor is chosen. Being able to deliver 5 W output power and switching at a frequency

of 406 MHz are the necessary requirements the transistor must fulfill. Next to that, a list

of other characteristics of transistors will be used to select the best transistor. The overall

system needs to be kept as small as possible, so we want a small transistor. In literature,

a lot of examples with GaN HEMT transistors are found, because of their fast switching

capabilities. GaN HEMT transistors are very expensive, and most of them are not capable of

handling high powers. Since only a frequency of 406 MHz is required, and LDMOS transistors

are available that have a higher maximum operation frequency then 406 MHz, an LDMOS

tranistor was opted. However, the choice of transistors that is able to switch at a frequency

of 406 MHz, and is also able to withstand power ratings above 5 Watt, seemed to be very

limited.

The smallest transistor found, that meets all requirements is the PD54008l-e transistor from

ST microelectronics, which is a LDMOS RF power transistor. The transistors dimensions are

5x5x1 mm. According to the datasheet, the transistor can achieve a gain of 15 dB, giving an

output power of 8 W at 500 MHz, and has excellent thermal stability. The maximum ratings

are 15 V for VGS , 25 V for VDS , and 5 A for ID. More information can be found in the

datasheet [37].

With this transistor, an output power of 36.5 at an input power of 18 dBm, and a PAE of

43

380 385 390 395 400 405 410 415 420 425−60

−50

−40

−30

−20

−10

0

frequency [MHz]

|S11

| (dB

)

Lp= 138 mm

Lp = 139 mm

Lp = 140 mm

Lp = 141 mm

Lp = 142 mm

(a) Lp

394 396 398 400 402 404 406 408 410 412 414 416−60

−50

−40

−30

−20

−10

0

frequency [MHz]

|S11

| (dB

)

Wp= 47 mm

Wp = 48 mm

Wp = 49 mm

(b) Wp.

394 396 398 400 402 404 406 408 410 412 414 416−60

−50

−40

−30

−20

−10

0

frequency [MHz]

|S11

| (dB

)

t = 28 mmt = 29 mmt = 30 mm

(c) t

394 396 398 400 402 404 406 408 410 412 414 416−60

−50

−40

−30

−20

−10

0

frequency [MHz]

|S11

| (dB

)

Wf = 2.5 mm

Wf = 2.7 mm

Wf = 2.9 mm

(d) Wf .

380 385 390 395 400 405 410 415 420 425−60

−50

−40

−30

−20

−10

0

frequency [MHz]

|S11

| (dB

)

εr = 1.1

εr = 1.15

εr = 1.2

(e) εr

394 396 398 400 402 404 406 408 410 412 414 416−60

−50

−40

−30

−20

−10

0

frequency [MHz]

|S11

| (dB

)

Ha = 67 mm

Ha = 67.5 mm

Ha = 68 mm

(f) Ha.

Figure 3.17: Simulated reflection coefficients for small variations in design parameters.

58 % , was obtained in simulations. However, when this circuit was fabricated and tested,

non of the transistors seemed to work. It turned out that current already was flowing trough

the gate, when only applying the bias voltage, and while the in and output were terminated

with 50 Ω . Therefore, a new design iteration was performed with a new transistor and more

decoupling and protection. The design steps for the second amplifier will be discussed in more

44

0 1 2 3 4 5 6 7 8 9 100

1

2

3

4

5

6

7

VGS

= 4.75 V

VGS

= 4.5 V

VGS

= 4.25 V

VGS

= 4 V

VGS

= 3.75 V

VGS

= 3.5 V

VGS

= 3.25 V

VDS

[V]

I DS[A

]

Figure 3.18: I-V characteristic of the MOSFET, with chosen bias point

detail hereunder.

The second transistor that was chosen, was the AFT05MS031N RF power LDMOS transistor

from Freescale Semiconductor. This transistor is larger compared to the previous one, with

dimensions of ± 10x11 mm. However, it can handle a higher voltages ratings at the drain,

having a maximum VDS of 40V, which makes it more robust against voltage and current spikes.

This transistor can deliver a gain of 18.3 dB with an efficiency of 64.1 %for frequencies [38].

The most relevant information from the datasheet can be found in appendix A.

Now that the transistor is chosen, it still needs to be biased. It is very important to choose

a good bias point to be able to drive the transistor into switching mode. This bias point is

chosen based on the input-output characteristics of the transistor, depicted in Figure 3.18.

VGS must be chosen in such a way that when an input signal is applied, the voltage swing is

high enough to force the transistor in on-state for the first half period, and in cutoff for the

second half. VDS must be chosen so that for the chosen VGS , the required power is delivered

to the output. The chosen bias point is VGS = 3.5V and VDS = 7.5V . This point is also

shown in figure 3.18. [39]. At this bias point, the drain current will be 0.737A, so the power

taken from the DC source is 5.53 W. For low power input signals, the transistor will work in

its linear region, and there will be a high gain, but low efficiency. For higher input power, the

variation in VGS , will be high enough to force the transistor into cut-off mode and saturation.

The second step is to chose the appropriate component values to drive the transistor in class

E. For this, the basic design of Figure 3.19 is used. The formulas mentioned above, 3.4 to

3.7, will be used to calculate the correct component values, given a required output power of

5 W, and the transistors output capacitance, 49.5pF [38].

45

VGS

LRF,choke

CDC,block

sin

C

L0C0

ZL

L1

VDS

sout

Figure 3.19: Basic class E circuit with MOSFET to function as a switch.

VGS 3.5 V

VDS 7.5 V

LRF,choke 1µH

L1 2.8 nH

C 49.9 pF

C0 10.8 pF

L0 9.5 nH

ZL 4.8 Ω

Table 3.7: Component values for QL

= 5 and Ld = 2.8 nH.

VGS 3.5 V

VDS 7.5 V

LRF,choke 1µH

L1 6 nH

C 33.3 pF

C0 26.4 pF

L0 7.3 nH

ZL 4.8 Ω

Table 3.8: Component values for QL

= 4 and Ld = 6 nH.

However, for the calculation of the components, an output power of 6 W was chosen. This

power is still below the maximum output power of the C/S requirements, and is chosen to

have a certain safety margin while still using ideal components for simulations. QL was chosen

to be 5, to have good damping after switching. The value of L1 was tuned until the value

of C was equal to the output capacitance of the transistor. The resulting values can be seen

in Table 3.7. At the input, the chosen VGS is applied, and an inductor is chosen to prevent

the RF signal from leaking to the DC voltage source, and a decoupling capacitor is placed

between the input and the gate.

When a transient simulation was performed with these values, the signals at the drain were

not as expected, as can be seen in Figure 3.20. The first possible explanation could be that

the transistors output capacitance is not correct. The output capacitance is given for a very

low VGS , and can be different when a gate voltage is applied. L1 was changed in the formulas,

which affected the values of C and C0. Simulations were done for different values of these

46

80 82 84 86 88 90 92 94 96 98 1000

5

10

15

20

25

time (ns)

Vol

tage

(V)/

Cur

rent

(A)

VDS

IDS

Figure 3.20: Current and voltage waveforms at the drain with component values of Table 3.7, and

an input power of 100 mW (20dBm)

components, corresponding with different values for the transistors output capacitance. The

results where best when L1 was 6 nH, and C was 32.37 pF. Still, a big overshoot in VDS

could be seen at turn on, so QL was decreased. However, when taking a QL lower than four,

the output waveforms weren’t sinusoidal anymore, so QL was chosen four. The values for

the resulting circuit can be seen in Table 3.8, and the waveforms at the drain can be seen in

Figure 3.21. The voltage and current waveforms are still not ideal, since they don’t resemble

square wave forms, and the voltage never drops to zero. This is due to a finite resistance

in the on state, and because the input power is not sufficient to make the transistor go into

complete cutoff. The transfer function of the output network is not ideal either, resulting in

an oscillation in the voltage waveform during the on-state.

80 82 84 86 88 90 92 94 96 98 1000

2

4

6

8

10

12

14

16

18

20

time (ns)

Vol

tage

(V)/

Cur

rent

(A)

VDS

IDS

Figure 3.21: Current and voltage waveforms at the drain with component values of Table 3.8, and

an input power of 100 mW (20dBm).

47

80 82 84 86 88 90 92 94 96 98 1000

2

4

6

8

10

12

14

16

18

20

time (ns)

Vol

tage

(V)/

Cur

rent

(A)

VDS

IDS

Figure 3.22: Drain source voltage and current of the class-E amplifier with input and output

matching, at an input power of 18 dBm.

The next step to be taken is designing an output matching network to transform the output

impedance of 4.66 Ω to 50 Ω. This is necessary because the power delivered to the output is

dependent on the load resistance. Since the antenna is designed to have an input impedance

of 50 Ω at this moment, the load applied to the circuit will also be 50 Ω. Best power transfer,

on the other hand, is achieved when the output load is 4.66 Ω. If a matching network is placed

between the power amplifier and the 50 Ω load impedance, it is possible for the amplifier to

still see an impedance of 4.66 Ω at his output ports, while the load is 50 Ω. To choose the

matching network, Smith charts are a very handy tool. The required impedance at the two

sides of the matching network is plotted, as can be seen in Figure 3.23. In this case, for

the matching network, a simple parallel resistor could be used, since this would decrease the

impedance seen by the amplifier. However, this would introduce too much losses. This is

why a C-L matching circuit was chosen. An inductance in parallel with the output, will move

the impedance to the inductive part (upper part of the Smith chart). Afterwards, a series

capacitor will move the impedance back into the direction of the capacitive part. If proper

values are chosen, the impedance seen at the input of the matching network is the required

impedance. The values of the C-L network could be calculated by working out the impedance

of the network. Another way is to use tools like the ADS smith chart tool, that are specially

made for this purpose. The chosen values will depend on the frequency, so good matching

will only be obtained for a limited frequency range.

At the input port, the goal is to have maximum power transfer from the source to the circuit.

It can be proven that this is obtained when the source reflection coefficient is conjugate

matched to the input reflection coefficient of the circuit. This is equivalent with choosing

the source impedance equal to the complex conjugate of the input impedance of the circuit.

If the device is considered to be unilateral, which means that S12 = 0, the input reflection

48

Figure 3.23: Impedance transformation on the Smith chart for output matching.

coefficient equals S11. From this S11, the input impedance of the circuit can be calculated. In

this case, the S12 will be very low, so unilaterality is assumed. The input impedance now is

50x(0.022 -0.03j) Ω. Because the output power of the signal generator is also for a reference

impedance of 50 Ω, an input matching network was provided as well, to transform 50x(0.022

+0.03j) Ω to 50Ω. The entire circuit is shown in Figure 3.24, with the component values given

in Figure 3.9. C4-L4 is the output matching network, and C3-L3 the input matching network.

The S-parameters of the circuit are shown in 3.25.

On the S-parameter curves of Figure 3.25, it can be seen that there is a good input matching,

VGS

L2

C2

L1

7.5 V

C1

L0C0 C4

L4 50Ω

+

−

C5L3

C350Ω

+

−

Figure 3.24: Class E-type amplifier with input and output matching.

49

VGS 3.5 V

VDS 7.5 V

L0 7.3 nH

L1 6 nH

L2 100 nH

L3 3.65 nH

L4 7.04 nH

C0 26.4 pF

C1 10 nF

C2 10 nF

C3 47.9 pF

C4 24.4 pF

C5 10 nF

Table 3.9: Component values of

circuit with matching

network at in- and

output.

VGS 3.5 V

VDS 7.5 V

L0 9.45 nH

L1 2.8 nH

L2 100 nH

L3 3.27 nH

L4 7.04 nH

C0 17 pF

C1 10 nF

C2 10 nF

C3 61.2 pF

C4 24.4 pF

C5 10 nF

Table 3.10: Component values for

circuit after tuning.

as the |S11| is very low, this is because the source impedance was chosen to be conjugate

matched to the amplifiers input impedance. However, the |S22| is very high, so power that is

injected at the output ports, will be reflected. This is because the output was first matched

to obtain an ideal gain, and afterwards, the input impedance was conjugate matched to have

ideal power transfer at the input. It was not possible to obtain good input and output

matching and still obtaining the same amount of gain. However, a good output matching

is less important than a good input matching, because this power will not be reflected to

the source, but to the antenna terminals. This power can result in distortion of the signal,

but this is acceptable as long as the antenna is matched to the output of the amplifier. If

the antenna would not be matched, a signal could go back and forth between the amplifier

and antenna, and this could damage the connection. However, measurements on the antenna

proved that the good matching is obtained. The |S21| at 406 MHz is 24.19 dB, which is very

high. This is good since the power gain is proportional to the |S21|. The |S12|, on the other

hand, is -27.8 dB, which means that the assumption of a unilateral device was justified.

To know the efficiency of the amplifier, and the power gain at 18 dBm input power, a Large

Signal S-Parameter (LSSP) simulation was performed. The resulting gain and PAE for 18

dBm input power was 36.4 dB and 64.3 % respectively. The component values were tuned to

obtain better results. The resulting component values after tuning can be seen in Table 3.10,

and the power gain and efficiency as a function of the input power can be seen in Figure 3.26

and 3.27.

50

300 320 340 360 380 400 420 440 460 480 500−50

−40

−30

−20

−10

0

10

20

30

frequency [MHz]

S−

para

met

ers

(dB

)

−19.6744

24.1926

−1.9126

−27.8724

|S11

|

|S12

|

|S21

|

|S22

|

Figure 3.25: S-parameters simulation for amplifier including matching networks.

Because the voltage and current waveforms at the drain are very important to have good class-

E operation, these were also simulated, and can be seen in Figure 3.28. To check whether the

input matching is still good with the given values, an S-parameter simulation with the new

component values was performed, and can be seen in Figure 3.29. The |S11| is higher now,

but is still smaller than -10 dB, which was the requirement. The |S21| is higher now, which

explains the higher power gain.

It can be observed from the transient simulation that the voltage waveforms are not ideal,

because the drain voltage never drops to zero, but stays at a voltage of ± 3 V while the

transistor is in its on state, and the current isn’t zero either during the time interval in which

transistor is in cutoff. This is because the voltage sweep of VGS is not big enough. A solution

would be to choose a lower VDS , since this would require a smaller variation in VGS , but then

it would be impossible to obtain the required output power with this transistor. A tradeoff

had to be made between enough output power and a good efficiency. The output power with

this circuit is 37.5 dBm, and the PAE is 75% which is still far above the efficiency (65%) that

can be obtained according to the transistors datasheet.

In the output power graph, it can be seen that the transistor operates in its nonlinear region

at an input power of 18 dBm, since the output power doesn’t increase linear with the input

power anymore. This was expected, and even required for obtaining class-E operation. It can

also be seen that the efficiency increases with increasing input power, which is logical because

a higher input signal will cause the transistor to saturate for a longer period of time, reducing

the power dissipation in the transistor.

The circuit is now designed with ideal lumped components. To obtain more accurate results,

51

0 2 4 6 8 10 12 14 16 18 20 2220

22

24

26

28

30

32

34

36

38

40

Pin

[dBm]

Pou

t[dB

m]

Figure 3.26: Output power as a function of

input power with ideal

components, after tuning.

0 2 4 6 8 10 12 14 16 18 20 220

10

20

30

40

50

60

70

80

90

100

Pin

[dBm]

PA

E [%

]

Figure 3.27: PAE of the amplifier as a

function of input power, with

ideal components, after tuning..

these where changed to real component models. The task was to find components with values

close to the ideal ones now in the circuit, that are small, preferably SMD components, that

have a self resonance frequency well above 406 MHz, a high Q-factor, and are able to handle

the power required for the circuit. Another option would be to replace the capacitors and

inductors by their transmission line equivalent. This would be easier, because variation in

component values can have severe effect, and these will be less when using transmission lines,

and there would be no problems with self resonance frequencies. But since the circuit needs to

be small, and transmission lines at a frequency of 406 MHZ will be long, lumped components

were chosen. Moreover, when using transmission lines in wearable applications, it is possible

that they will bent, which can cause failure of the circuit.

At this moment, only one decoupling capacitor for gate and drain is used in the ideal circuit.

However, in reality, all capacitors have a self resonance frequency (SRF), above which they

start to behave as an inductor, and thus get a lower impedance with increasing frequency,

losing their decoupling capacities. Since decoupling must be obtained for a large bandwidth, it

is replaced by a series of decoupling capacitors, placed in parallel. When using more capacitors,

with different values and SRF, decoupling can be achieved over a large bandwidth. To be

certain of a good decoupling, the same values and types that are used to decouple a power

amplifier example in the datasheet, are chosen. All the noise at the DC sources, that is in

the frequency band covered by the capacitors, will be shunted trough these capacitors, which

avoids damage and interference. Since a high current is delivered from the bias source, and the

cables have a large L, signal variations can occur in DC power supply, which will be avoided

by using decoupling capacitors.

At the gate and source, small capacitor values are added to avoid that high voltage spikes

52

80 82 84 86 88 90 92 94 96 98 1000

2

4

6

8

10

12

14

16

18

20

time (ns)

Vol

tage

(V)/

Cur

rent

(A)

VDS

IDS

Figure 3.28: Current and voltage waveforms at the drain, after tuning and with ideal components.

damage the transistor. These capacitances will short circuit all high frequency components of

the voltage spikes to ground, while letting desired RF signal trough. The problem with voltag

spikes could also be solved by using TVS diodes, but since these are bigger components,

capacitors were chosen. Adding these capacitors also increased the stability, because they

induce losses at the gate and drain.

After replacing the ideal components by real ones, and adding the small capacitances at the

gate and drain, the input matching was gone. The solution was to remove the input matching

network that is currently used, and replace it by a new one. This time, to get good matching

while using real component models, more elements were necessary, as can be seen in Figure

3.30, that shows the entire circuit. Besides low reflections at the input, input matching was

also necessary to obtain the required gain.

Now that all components of the circuit are changed to real components, transmission lines

were added to represent the PCB layout. Although the frequency is only 406 MHz, and the

total PCB length with this microstrip lines would only be 4 cm, which is electrically small

compared to the wavelength of 70 cm, the microstrip lines had an influence on the power

amplifier’s performance, probably because of the added inductance of the microstrip lines

connecting the components. Losses in the copper traces at the applied power will also be non

negliable. To even out the effect of adding the microstrip lines, tuning of the components

was done by adding small ideal components, to check their effect, and then changing the real

components accordingly. Tuning was necessary, because using real component values and

adding the microstrip lines, lowered the power gain to less than 17 dB at an input power of

18 dBm.

53

300 320 340 360 380 400 420 440 460 480 500−50

−40

−30

−20

−10

0

10

20

30

frequency [MHz]

S−

para

met

ers

(dB

)

−13.1108

24.9014

−1.0350

−27.1636

|S11

|

|S12

|

|S21

|

|S22

|

Figure 3.29: S-parameters with matching networks, after tuning.

VGS

LRF,choke,2

CG5CG4CG3CG2CG2

CG7CG6

LRF,choke,1

VDS

CD1 CD2 CD3 CD4

CD5

L0C0 Co1

Lo1 50Ω

+

−

CDC,blockLi1

Ci1Ci2Li2

Ci3

50Ω

+

−

Figure 3.30: Power Amplifier circuit topology.

After elaborate tuning, an output power of 36.5 dBm, at an input power of 18 dBm, and

an efficiency of 57% was obtained. These results are poor compared to those with ideal

components, but are still acceptable and satisfy the requirements. The reduced performance

of the amplifier can be explained by losses in the components, that will decrease the gain as

well as the efficiency.

The last thing that was done, was adding a voltage divider to be able to use only one battery,

and derive the gate voltage from the drain voltage. By choosing good resistor values, the

current flowing trough them is negligible, and doesn’t affect the amplifiers efficiency. However,

this voltage divider was not used in measurements, to make it possible to adjust the gate and

drain voltage separately.

When good results were achieved in the circuit simulator, a PCB layout was drawn in ADS

54

momentum, where the copper traces between the components had the same dimensions as

the transmission lines that were used in the circuit simulator. The difference between the

simulations with transmission lines in the circuit simulator, and the one with the layout

simulated in ADS momentum used in the circuit simulator, are not substantial, which means

that electromagnetic interference between different traces on the PCB is negligible.

Figure 3.31: Layout of the final design.

The layout that was used can be seen in Figure 3.31. In the middle, a rectangle needs to

be milled out of the PCB to be able to insert the transistor, according to the transistors

application note. The transistor structure can be seen in Figure 3.32 The resulting output

power and efficiency can be seen in Figure 3.34 and Figure 3.33, and the component values

are listed in Table 3.11

However, when the design was assembled, it turned out to be unstable, oscillating at the

harmonics of 113 MHz. When searching for the origin of the problem, a parameter sweep

was done over different component values for LRF,choke,1, and it could be seen that for low

values, there was a peak in the |S11| at 130 MHz, which can be seen in figure 3.35. This

Figure 3.32: Transistor case style [38].

55

0 2 4 6 8 10 12 14 16 18 20 220

5

10

15

20

25

30

35

40

45

50

55

60

65

70

75

Pin

[dBm]

PA

E [%

]

54.8119

Figure 3.33: Efficiency of the power amplifier.

peak disappears according to the simulations when using an inductance greater than 10 nH.

During measurements, it turned out that the circuit only became stable when using values for

LRF,choke,1 of 100 nH and higher. Therefore, this value was changed in the circuit. Because

changing this inductance influenced the power gain and efficiency, tuning was again necessary

to fulfill the requirements again. The final component used in simulation are listed in Table

3.12.

After the circuit was tuned, necessary simulations were carried out to test all specifications

imposed by the COSPAS-SARSAT system. In Figure 3.36 the output power as a function

of the input power can be seen. The output power, at an input power of 18 dBm, is 35.7

dBm, which is almost the same as for lower values of LRF,choke,1. On the other hand, the

efficiency is decreased to 39%. This is probably because the value of LRF,choke,1 determines

the required output capacitance of the transistor. Since this capacitance cannot be changed,

the transistors output network is probably not ideal to get good Class-E operation anymore.

Higher inductors also have a higher parasitic resistance, and since the current flowing trough

this inductance is very high, this can cause a lot of losses. The PAE in function of the input

power can be seen in figure 3.37

However, since the PCB layout could not be changed anymore, and the only requirement

imposed by C/S is a minimum output power of 35 dBm, these results were accepted. These

results could be improved in a new design iteration.

The results of the S-parameter simulation can be seen in Figure 3.38. It can be seen that the

input matching is still good, having an |S11| of -18.5 dB. The output reflection is even lower

than for ideal components, having a value of -5.2 dB. However, the transmission coefficient,

|S21| is now much lower, being only 18 dB.

56

0 2 4 6 8 10 12 14 16 18 20 2220

22

24

26

28

30

32

34

36

38

Pin

[dBm]

Pou

t[dB

m]

35.7105

Figure 3.34: Output power versus input power.

100 150 200 250 300 350 400 450 500 550 600−40

−30

−20

−10

0

10

20

30

frequency [MHz]

S−

Par

amet

ers

(dB

)

L = 1 nH, |S

11|

L = 3.2 nH, |S11

|

L = 10 nH, |S11

|

L = 1 nH, |S21

|

L = 3.2 nH, |S21

|

L = 10 nH, |S21

|

Figure 3.35: S-parameters for different values of LRF,choke1.

57

Component Value Component Value

LRF,choke2 100 nH CG1 180 pF

LRF,choke1 2.1 nH CG2 200 pF

CDc,block 22 nF CG3 0.01 µ F

L0 6 nH CG4 0.1 µF

C0 18 pF CG5 22 µF

Lo1 6.8 nH CG6 10 pF

Co1 27 pF CG7 27 pF

Li1 3.3 nH CD1 180 pf

Ci1 27 pF CD2 0.01 µF

Li2 5.1 nH CD3 0.1 µF

Ci2 33 pF CD4 470 µF

Ci3 22 pF CD5 27 pF

Table 3.11: Component values after adding the layout.

Obtaining the input and output stability circles, it could be seen that the amplifier is stable for

all terminations. By doing an harmonic balance simulation, the power of the higher harmonics

in the output signal could be obtained. The power of the second harmonic is 5.5 dBm, or

thus more than 30 dB lower then the fundamental tone. This suppression is good, when it

is considered that the Q-factor of the output filter was chosen low at the beginning of the

design.

Furthermore, C/S also imposes constraints on the spurious emissions of the amplifier. How-

ever, the component models were not suitable to perform these simulations, so this condition

needs to be verified by means of measurements.

58

0 2 4 6 8 10 12 14 16 18 20 22 2418

20

22

24

26

28

30

32

34

36

38

40

Pin

[dBm]

Pou

t[dB

m]

35.71

Figure 3.36: Output power versus input power of the final design.

0 2 4 6 8 10 12 14 16 18 20 22 240

5

10

15

20

25

30

35

40

45

50

55

60

Pin

[dBm]

PA

E [%

]

38.12

Figure 3.37: PAE versus input power of the final design.

59

300 320 340 360 380 400 420 440 460 480 500−50

−40

−30

−20

−10

0

10

20

30

frequency [MHz]

S−

para

met

ers

(dB

)

−18.5168

18.0335

−33.7456

−5.2348

|S11

|

|S12

|

|S21

|

|S22

|

Figure 3.38: S-parameter simulation.

Component Value Part number Manufacturer

LRF,choke1 111 nH Coilcraft 0805LS-111XJLB

LRF,choke1 100 nH Coilcraft 0603HP-R10

CDc,block 22 nF Murata GRM188R61H103KA01

L0,Co1 5.1 nH Coilcraft 0603HP-5N1

C0, Ci3 18 pF Murata GQM1885C1H180JB01

Lo1 39 pF Murata GQM1885C1H390JB01

Li1 3.3 nH Coilcraft 0603CS-3N3

Ci1 22 pF Murata GQM1885C1H220JB01

Li2 6.8 nH Coilcraft 0603HP-6N8


CG1, CD1 180 pF AVX 06035A181JAT2A

CG2 200 pF AVX 06035A201FAT2A

CG3, CD2 0.01 µ F MULTICOMP MC000252

CG4, CD3 0.1 µF MURATA GRM188R71H103KA01D.

CG5 22 µF KEMET T491X226K035AT

CG6 10 pF Murata GQM1885C1H100JB01

CG7, CD5 27 pF Murata GQM1885C1H270JB01

CD4 470 µF Electrolytic capacitor

Table 3.12: Component values of the amplifier, with a AFT05MS031N RF power transistor.

60

Chapter 4

Measurements and Results

4.1 Antenna Measurements

4.1.1 Input Reflection Coefficient

After obtaining satisfying simulation results in CST, the antenna was manufactured. All

conductor layers were made out of copper-coated nylon taffeta electro-textile and were glued

onto the substrates. Because of the manual fabrication process, small differences between

different prototypes and the design are possible. The measured |S11| of the first antenna can

be seen in figure 4.1, together with the simulation results of CST. Figure 4.1 shows that the

resonance peak is shifted from 406 MHz to 382 MHz.

330 340 350 360 370 380 390 400 410 420 430 440 450−40

−35

−30

−25

−20

−15

−10

−5

0

frequency [MHz]

|S11

| (dB

)

MeasurementSimulation result from CST

Figure 4.1: Measured and simulated reflection coefficients of the aperture coupled shorted patch

antenna.

61

Because the resonance frequency is shifted downwards, the resonance wavelength is longer

than the wavelength corresponding with 406 MHz. Hence, the antenna dimensions are too

large. The antenna dimension that will mainly determine the resonance frequency is the patch

length. Hence, in order to shift the resonance frequency of the first prototype to the desired

frequency, the patch length will be decreased in steps of 2 mm by cutting the patch. Figure

4.2 depicts the measured reflection coefficients for different patch lengths.

300 320 340 360 380 400 420 440 460 480 500−40

−35

−30

−25

−20

−15

−10

−5

0

frequency [MHz]

|S11

| (dB

)

Patch length = 140mmPatch length = 138 mmPatch length = 136 mmPatch length = 134 mm

Figure 4.2: Measured reflection coefficients of the aperture coupled shorted patch antenna for

different patch lengths.

Figure 4.2 demonstrates that the resonance frequency increases by decreasing the patch length,

as earlier assumed. For a patch length of 134mm, the resonance frequency is 404 MHZ, and

the |S11| equals -12.58 at 406 MHz. However, this does not satisfy the C/S specifications of

having a |S11| < -14 dB, and can be attributed to a poor matching of the antenna impedance

to 50 Ω. This is also demonstrated in figure 4.3, where the S11 is plotted on the Smith chart.

One way to improve matching, consists of changing the length of the feedline. Since decreasing

the length of the feedline decreased matching, a new antenna was made with again a patch

length of 134 mm, and an excess on the length of the feedline. Best matching was obtained

with a feed length 7 mm longer than the simulated length, which corresponds with a length t

= 36 mm (as depicted in figure 3.8) . The resulting |S11| can be seen in figure 4.4. The |S11|at 406 MHz is -19.12 dB, and the -10 dB resonance band is from 395.5 to 415.2 MHz. This

results satisfy the design goals for the |S11| set for this thesis, as well as the requirements of

C/S.

To verify these results, a second prototype was constructed with the same patch length and

feed line length. A quasi similar |S11| curve was measured, implying that the adjustments

that had to be made are probably because of a wrong characterisation of the black foam

62

0.2

0.5

1.0

2.0

5.0

+j0.2

−j0.2

+j0.5

−j0.5

+j1.0

−j1.0

+j2.0

−j2.0

+j5.0

−j5.0

0.0 ∞

Figure 4.3: S11 of the aperture coupled shorted patch antenna with a patch length of 134 mm.

substrate, and not because of manufacturing mistakes. The black foam material is also not

specifically made to be used as antenna material, and the εr is not always exactly the same.

Hence it is possible that the characteristics are slightly different for other batches. Since the

actual length is shorter than the length of the simulated antenna, the εr, being 1.15, used for

the simulations is probably too low. On the simulation results, that could be seen in figure

3.17e, it can be seen that changing the εr of the black foam from 1.15 to 1.2 already results

in a frequency shift of 7 MHz down.

A second remark that can be made are that the losses of the antenna are rather high. Even

far from the resonance band, the |S11| is below -2 dB. It is very unlikely that the antenna will

radiate at these frequencies, because the efficiency is only high enough in a small frequency

band around 406 MHz. Hence, it can be assumed that the power that is not reflected, is lost

in the antenna.

The most important cause of high losses are probably the losses in the copper polyester taffeta

fabric, which has a non-negligible sheet resistance. The feed structure of the antenna consists

of a very thin trace of copper-coated nylon taffeta electro-textile, which will introduce losses

before the power is even transferred to the patch. Other causes of losses can be the substrate

materials that are used, but because they both have a low loss tangent and dielectric constant,

they are not likely to introduce such high losses. Since the losses in the substrates, as well

as the resistivity of the opper polyester taffeta fabric was already included in the antenna

simulations, the actual losses will probably be higher than the ones estimated in [23].

To reduce losses, the antenna could be made with a conductor with a lower sheet resistance.

To confirm that the losses are due to losses in the copper polyester taffeta fabric, an antenna

was made where all the conductive sheets consisted are replaced by copper foil. However, this

63

300 320 340 360 380 400 420 440 460 480 500−30

−25

−20

−15

−10

−5

0

frequency [MHz]

|S11

| (dB

)

19.7 MHz

Figure 4.4: Measurement results for an antenna with a patch length = 134 mm and t = 36 mm .

reduced the flexibility of the antenna, and is not a practical solution for wearable applications.

Since most of the losses are supposed to be in the thin feedline, an antenna where only the

feedline was replaced by copper foil, and the patch and groundplane are made out of copper

polyester taffeta fabric, was also fabricated. This antenna could be an acceptable solution,

since the feedline is very thin, and will only be on a small part of the antenna surface, and

wouldn’t affect the antenna’s flexibility. The results of the |S11| measurement results can be

seen in figure 4.5.

These three measurements were performed on antennas that were made on the same day,

and measured close after each other, so differences due to environmental conditions can be

excluded. It can be seen that the assumption that most of the losses are contained in the

thin feedline is correct, since the difference in reflection coefficient for frequencies out of the

antenna resonance band, between the antenna entirely made out of copper and the one with

only a copper feed line, is negligible. The resonance frequency is slightly different for the three

measurements. This shift is partially due to the different conductive sheets that are used, but

can also be partially due to fabrication inaccuracies. Another observation that can be made,

is that the resonance bandwidth is much smaller for the antennas where copper is used. This

will however not necessarily mean that less power will be radiated, but can be attributed

to a reduction in antenna losses. An other solution would be to use a thicker feedline, and

compensate the shift in the input impedance by varying other antenna dimensions, or by using

a matching network, or the power could directly be matched to the antenna input impedance.

Because the antenna where copper foil is used for the feedline, and copper polyester taffeta

fabric for the patch and groundplane has the best performance concerning flexibility and

64

300 320 340 360 380 400 420 440 460 480 500−40

−35

−30

−25

−20

−15

−10

−5

0

frequency [MHz]

|S11

| (dB

)

FlectronCopperFeedline in Copper

Figure 4.5: Measured reflection coefficients of the aperture coupled shorted patch antenna for

different conductive materials.

losses, this antenna will be used to perform other tests. The |S11| of this antenna, measured

over a larger frequency range, can be seen in figure 4.6a. At the double of the resonance

frequency, a small dip can be observed, but because of bad matching, no significant radiation

will occur. Figure 4.6b, zooms in on the resonance band, and the resonance band is annotated

on the graph. The [406 406.1] MHz band is also annotated on the graph.

0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 1.1 1.2 1.3−25

−20

−15

−10

−5

0

frequency [GHz]

|S11

| (dB

)

(a) Measured over a large frequency band.

394 396 398 400 402 404 406 408 410 412−25

−20

−15

−10

−5

0

frequency [MHz]

|S11

| (dB

)

12.8 MHz

(b) Zoom on resonance band.

Figure 4.6: Measured reflection coefficients of the aperture coupled shorted patch antenna with

copper foil feed line and copper-coated tafetta patch and ground plane.

65

4.1.2 Radiation Pattern

According to the C/S specifications, the radiation pattern must fulfill certain conditions,

that are depicted in Figure 2.5. Besides fitting these conditions, the radiation towards the

groundplane is also a point of interest, because this indicates how much power is radiated

towards the human body, and needs to be minimised.

To measure the radiation pattern in free space conditions, the antenna will be measured in an

anechoic chamber. Two antennas will be set up in the room at a certain distance from each

other. One of the antennas will radiate a certain power, and the power received by the other

antenna will be measured. If the antenna gain of one of the antennas is known, as well as the

distance between the antennas, the power gain of the second antenna can be calculated. To

obtain the radiation pattern, the antenna of interest can be rotated to different angles.

An anechoic room is a room that is designed to avoid all reflections against the walls, in

order to reproduce a free space environment. However, the anechoic room in which the

measurements were performed, only has complete absorption for frequencies above 1 GHz,

so the possibility to have reflections needs to be considered. Furthermore, the farfield of an

antenna is considered to start when the distance df between the antennas is larger than 2d2

λ

(Fraunhofer), with d the largest dimension of the antenna. For the patch antenna, d = 190

mm, so this would mean a distance larger than 10cm, which is achieved. Besides being larger

than this, two additional requirements apply, which are df >> d and df >> λ [40]. The

second of these conditions, is hardly met, because the distance between the two antennas is

only in the order of 4λ. However, because the measurements can still give a good idea of the

radiation pattern and the antenna gain, they will be discussed hereunder.

To obtain more accurate results, a three antenna measurement was performed. Two out of the

three antennas are positioned in the anechoic chamber, and the S-parameters are measured.

One of the antennas is afterwards turned 90 , to have the transmission for both polarisations.

If this is done for the three combinations of two antennas, the gain of every antenna in the

main direction can be calculated as follows.

According to the Friis formula, given in a absorption less medium, we get:

PtPr

= GrGt

(λ

4πR

)2

(1− |Γt|2)(1− |Γr|2), (4.1)

Where the ratio of transmit and receive power, Pt/Pr, can be calculated with the measured

S21, and the impedance mismatch losses due to the reflection coefficients Γt and Γr correspond

with S11 and S22. The free-space path loss,(

λ4πR

)2, can be calculated by filling in the distance

between the antennas. This leaves only two unkown parameters in the antenna, Gt and Gr.

66

If this measurement is done three times, we get three unknown parameters, namely the gain

of each of the three antennas, and three equations. This will lead to the solution for the gain

of each antenna separately.

The three antennas that were used were a series 15 adjustable standard gain dipole, a PIFA

design of [23], and the aperture coupled patch antenna with a copper feedline. The resulting

antenna gain at 406 MHz are 6.38, -2.26 and -2.13 dBi respectively, in the direction perpen-

dicular to the patch. The resulting gain for the antenna designed in this master thesis is

-2.13dBi in the direction perpendicular to the patch. This gain is almos 2 dB lower than the

simulated gain in that direction. This is probably caused by extra losses in the antenna, as

discussed above. The radiation pattern in the xz-plane, as is depicted in figure 4.8, can be

seen in figure 4.7a, where the gain is given in dBi.

−180 −140 −100 −60 −20 20 60 100 140 180−12

−10

−8

−6

−4

−2

0

2

Theta [degr]

Gai

n (d

Bi)

MeasuredSimulated

(a) ZX-plane, φ = 0 .

−180 −140 −100 −60 −20 20 60 100 140 180−12

−10

−8

−6

−4

−2

0

2

Theta [degr]

Gai

n (d

Bi)

MeasuredSimulated

(b) ZY-plane, φ = 90 .

Figure 4.7: Measured radiation pattern of the aperture coupled shorted patch antenna with copper

foil feed line in the xz- and yz-plane at 406 MHz.

According to the C/S specifications, the gain for 90 % of the elevation angles between 5

and 60 must be above -3 dBi. This corresponds with the angles between 30 and 85 , and

between −30 and −85 . This condition is not met, since the antenna gain is only above

-3 dBi from −15 to 69 , for the xz-plane, and for the yz-plane this is from −18 to 93 .

Figure 4.8: Illustration of the axial system used.

67

The low antenna gain could be compensated by increasing the power at the antenna input

terminals.

Besides, the low gain for small elevation angles, there is a significant backwards radiation.

This is due to the small groundplane that is used, which is not large enough to shield radiation

towards this direction.

4.1.3 Influence of the Human Body and Bending of the Antenna.

Because it is very likely that the antenna will be bent at the moment it sends out a distress

signal, the antenna reflection coefficients were measured while the antenna is bent under

different bending radii. If the antenna is bent, it will probably be in the direction of the

groundplane, across the length, as is illustrated in Figure 4.10. The reflection coefficients for

three different bending radii were measured, and can be seen in figure 4.9

320 340 360 380 400 420 440 460 480−30

−25

−20

−15

−10

−5

0

frequency [MHz]

|S11

| (dB

)

No bendingBending radius 90mmBending radius 75 mmBending radius 50 mm

Figure 4.9: Measured reflection coefficients for three

different bending radii..

R

Figure 4.10: Direction in

which the

antenna is

bent.

It can be seen on the figure that the influence of bending is not critical. The antenna reflection

coefficient at 406 MHz is below -14 dB, for bending up to a radius of 75mm, meeting the C/S

requirements. For a radius of 50mm, the |S11| equals -12.36 at 406 MHz, which doesn’t satisfy

the C/S specifications, but is still acceptable.

Furthermore, the antenna will be used in a life jacket, that will be worn by somebody. This

means that the antenna will be deployed in the vicinity of the human body. This is why

the antenna was also measured when deployed on the human body. Since the space between

the antenna and the bod still remains unknown, measurements were performed in a situation

where the antenna is placed directly on the human body, with only a t-shirt in between, and

in the situation where the antenna is deployed at a distance of 110 mm from the human body.

68

320 340 360 380 400 420 440 460 480−30

−25

−20

−15

−10

−5

0

frequency [MHz]

|S11

| (dB

)

No Human bodyOn t−shirt110 mm spacer

Figure 4.11: Reflection coefficient when a human body is present.

Both times, the antenna was place on the chest.

From these results, it can be seen that the presence of the human body doesn’t deteriorate the

antenna performance, and the antenna will thus be suitable for applications where a human

body is present.

Overall, it can be concluded that the |S11| of the antenna stays below -10 dB for measurements

under different conditions. This means that little power will be reflected. In chapter 3 it was

assumed that a high |S22| of the power amplifier will not cause trouble in case that the

matching with the antenna is good. Here it is proven that this will be the case.

4.2 Power Amplifier Measurements

A first design was made with the “PD54008l − e” transistor of ST microelectronics. All

the components were soldered, including the transistor and connectors. First of all, the

biasing was checked while the in- and output were terminated by a 50 Ω impedance. The

drain current seemed to be identical to those of the simulations when the gate and drain

voltage were set. However, when a gate voltage was applied a second time, current was going

trough the transistors gate, while the transistor didn’t start to conduct when increasing the

gate voltage, implicating the transistor stopped working. Unfortunately, this happened to

all transistors, so no further tests could be done. Because only assumptions could be made

to explain the failure of the transistor, the decision was made to make a new design with a

different transistor, and more protection against voltage spikes at the gate and drain.

69

The second amplifier design was made with the “AFT05MS031N” RF power LDMOS transis-

tor from freescale semiconductor, which is capable of handling a higher drain voltage. Because

of the transistors structure, where the source terminal is 1mm lower than the gate and drain,

as shown in figure 3.32, a hole was foreseen in the PCB to place the transistor in. Like this,

the source plane of the transistor was on the same height as the bottom of the PCB. The

source was connected to the groundplane by soldering a copper plane on the source plane,

that extended to the PCB’s groundplane. Afterwards, the other components were soldered

on the board.

Again, the bias drain and gate voltage were applied to test whether the biasing corresponds

with the simulated values. First, it could be remarked that the drain current was not equal

for the different transistors that were tested. For a gate voltage of 3.4V and a drain voltage

of 7.5V, the drain current varied between 1.1A and 2A for different transistors. As will be

proved later on, it didn’t alter the S-parameters or power gain, but a higher bias current

results in a lower efficiency.

The second measurement that was done, was measuring the S-parameters. This was done

with a PNAX N5242A from Agilent technologies, which is the same as used for the antenna

measurements.

When using the component values of Table 3.11, the circuit started to oscillate when increasing

the gate voltage above 2.7V. When slowly increasing the gate voltage, the transistor began to

conduct, and the |S21| started to increase, and got above zero, which indicates an amplification

of the signal. From a certain point, around VGS = 2.7 V, the |S21| dropped below zero again,

and peaks in the S-parameters curves could be observed. This peaks continued to exist when

lowering VGS again, until the gate voltage was below 2 V. Explanations for this phenomena

are given in [41]. However, the solutions they proposed, didn’t solve the problem. The S-

parameters for a gate voltage of 2.7 and 2.9V can be seen in figure 4.12 and 4.13 respectively.

For a gate voltage of 2.7 V, the amplifier didn’t oscillate. To be able to already compare

the measurement results with simulation results, a simulation was done with a gate voltage

of 2.7V. The S-parameters obtained with the simulation can be seen in figure 4.14. It can

be seen that the |S21| in the simulations is much lower than in the measurements. This is

probably because according to the simulations there is no significant current flowing at a gate

voltage of 2.7V, while in the measurements, already a current of 0.13 A was flowing trough the

transistor. The input matching in the simulated values is shifted towards higher frequencies

when lowering the gate voltage, while the resonance peak of the measured values is shifted

towards lower frequencies.

To confirm that the “noisy” S-parameter curves are caused by an oscillation, the output was

measured with a spectrum analyser. Peaks in the spectrum could indeed be seen, and they

70

300 320 340 360 380 400 420 440 460 480 500−50

−40

−30

−20

−10

0

10

20

30

frequency [MHz]

S−

para

met

ers

(dB

)

−3.5154

6.1255

|S11

|

|S12

|

|S21

|

|S22

|

Figure 4.12: Measured S-parameters with component values from Table 3.11, and VGS = 2.7V.

300 320 340 360 380 400 420 440 460 480 500−50

−40

−30

−20

−10

0

10

20

30

frequency [MHz]

S−

para

met

ers

(dB

)

|S11

|

|S12

|

|S21

|

|S22

|

Figure 4.13: Measured S-parameters with component values from Table 3.11, and VGS = 2.9V.

were located at every harmonic of 113 MHZ.

While looking for a possible explanation, simulations were done to check whether they could

give an answer to the problem. When a parameter sweep was done for different values of

LRF,choke,1, a peak in the |S11| curve at 130 MHz could be seen (Figure 3.35), for values from

3 nH to 10 nH. Since this peak is close to the fundamental oscillation frequency, 113MHz , the

current LRF,choke,1 on the PCB was replaced by a value of 100 nH. This indeed stopped the

oscillation. Because according to simulations, higher efficiency could be obtained when using

smaller values for the inductor, measurements were done with different inductors. However,

all values lower than 100 nH resulted in an oscillation, so the circuit was optimised in the

simulations for this value of LRF,choke,1 .

71

300 320 340 360 380 400 420 440 460 480 500−40

−35

−30

−25

−20

−15

−10

−5

0

5

frequency [MHz]

S−

para

met

ers

(dB

)

−5.4862

−28.567

−9.5235

−1.6195

|S11

|

|S12

|

|S21

|

|S22

|

Figure 4.14: Simulated S-parameters, compared to the measured S-parameters with components

from Table 3.11, both for a VGS = 2.7V.

New boards were soldered with this new component values as listed in table 3.12. Slightly

different component values had to be chosen for Ci3 and Li2 to get better input matching.

The best input matching was obtained when Ci3 = 15pF and Li2 = 5 nH. The complete

list of the components that were used on this circuit is given in table 4.1, with the names

corresponding with how they are defined in figure 3.30. In Figure 4.15, the S-parameters can

be seen at a gate voltage of 2.7V, when there is no significant drain current.

When slowly increasing the gate voltage, the |S21| started to increase as well. The S-parameter

curves for VGS 3.03V and 3.21V can be seen in figures 4.16 and 4.17 respectively. It can be

seen that the S-parameters don’t vary a lot, which makes it a better choice to use a gate

voltage of 3.03V, since this results in a higher efficiency. In Figure 4.18, the simulated and

measured S-parameters are compared, when VGS = 3.21V.

After obtaining good S-parameters, the gain of the power amplifier was measured. This was

done by using a USB power sensor U2000B from Agilent Technologies. This devices measures

the amount of power that is received at a given frequency. The PNA-X is used to generate a

one tone signal at 406 MHz. First, the losses in the cables that will be used were measured,

by applying and input power of -5dBm in the cable at 406 MHz, and looking at the measured

value at the power sensor. For the cable, used for applying the input signal to the power

amplifier, a loss of 0.78 dB was measured. For the cable connecting the power amplifier to

the power sensor, a loss of 0.66 dB was measured. These values were taken into account when

steering the power amplifier, and when analysing the results.

Afterwards, the gain was measured for different input power. Because of heat generation in

the device, and because it took some time to change the input power, the power was turned

72

300 320 340 360 380 400 420 440 460 480 500−50

−40

−30

−20

−10

0

10

20

30

frequency [MHz]

S−

para

met

ers

(dB

)

S11

S12

S21

S22

Figure 4.15: Measured S-parameters with component values from table 4.1, when VGS = 2.7V and

IDS = 0.044 A.

300 320 340 360 380 400 420 440 460 480 500−50

−40

−30

−20

−10

0

10

20

30

frequency [MHz]

S−

para

met

ers

(dB

)

−7.3061

17.5906

−10.6307

−33.8186

S11

S12

S21

S22


IDS = 0.79 A.

off after every measurement. Because the used DC voltage source was not very accurate, it

must be remarked that the gate voltage was not always equal for the different measurements.

Because the drain current was much higher than in simulations, namely 2 A, when applying a

gate voltage of 3.4 volt, measurements were done for a gate voltage of 3.1 and 3.3 volt, where

the DC current is 0.85 A, and 1.4 A respectively. A list with all the measured values can be

found in appendix B, and the measured values for two prototypes can be seen in Figure 4.19

and 4.20.

The last test performed was measuring the spectrum of the signal, to see if the requirement

on the spurious emissions is fulfilled. The PNA-X is again used to generate a 406 MHz tone,

73

300 320 340 360 380 400 420 440 460 480 500−50

−40

−30

−20

−10

0

10

20

30

frequency [MHz]

S−

para

met

ers

(dB

)

19.3517

−8.6403−9.22333

−35.5815

|S11

|

|S12

|

|S21

|

|S22

|


IDS = 1.88 A.

300 320 340 360 380 400 420 440 460 480 500

−40

−30

−20

−10

0

10

20

frequency [MHz]

S−

para

met

ers

(dB

)

Measured |S11

|

Simulated|S11

|

Measured |S21

|

Simulated|S21

|

Measured |S22

|

Simulated|S22

|


IDS = 1.36 A, compared with the simulated results.

and the frequency spectrum is captured using a Rhode & Schwarz FSV40 signal analyser.

However, when these measurements were performed, no power amplifier was available that

was capable of delivering 18 dBm at the input terminals of the amplifier that should be

meausered, and the PNA-X can only deliver 15 dBm. Hence, the spurious emissions had

to measured at an input power of 15 dBm, and not the 18 dBm, for which the amplifier is

designed. The spectrum can be seen in figure 4.21. To test wether there is good suppression

of the harmonics, the spectrum was also measured over a larger frequency band, and can be

seen in figure 4.22.

A final remark that needs to be made is that one of the decoupling capacitors used at the

74

−10 −8 −6 −4 −2 0 2 4 6 8 10 12 14 16 180

5

10

15

20

25

30

35

40

Pin

[dBm]

Pou

t[dB

m]

first measurementsecond measurement

Figure 4.19: Measured output power versus input power for VGS = 3.1V.

−10 −8 −6 −4 −2 0 2 4 6 8 10 12 14 16 185

10

15

20

25

30

35

40

Pin

[dBm]

Pou

t[dB

m]

Simulated output powerMeasurement on first prototypeMeasurement on second prototype

Figure 4.20: Measured output power versus input power for VGS = 3.3V, compared to the

simulated output power.

drain is an electrolytic capacitor of 470µF. For this value, no small capacitors were found,

and the one used has a diameter of 6 mm and a height of 11 mm. This height is not suit-

able for wearable applications, so the measurements were also performed without using this

capacitor. Since no failure occurred, it can be assumed that this component isn’t crucial for

the power amplifier. Furthermore, the surface temperature of the transistor was measured.

Since without the use of a heat sink, and when applying a VGS of 3.3V, which resulted in a

IDS of 1.4A, the temperature only increased till 55 C after 50 seconds, and the transistor

only needs to work during 0.5 seconds, a heat sink will not be required.

75

−50 −40 −30 −20 −10 0 10 20 30 40 50−50

−45

−40

−35

−30

−25

−20

−15

−10

−5

0

5

Relative frequency [kHz]

Rel

ativ

e P

ower

(dB

c)

Figure 4.21: Measured spectrum at an input power of 15 dBm, with the frequency given relative to

406 MHz.

100 300 500 700 900 1100 1300 1500−10

−5

0

5

10

15

20

25

30

35

40

Frequency [MHz]

Pow

er (

dBm

)

Figure 4.22: Measured spectrum at an input power of 18 dBm.

76

Component Value Part number Manufacturer

LRF,choke1 111 nH Coilcraft 0805LS-111XJLB

LRF,choke1 100 nH Coilcraft 0603HP-R10

CDc,block 22 nF Murata GRM188R61H103KA01

L0,Co1 5.1 nH Coilcraft 0603HP-5N1

C0 18 pF Murata GQM1885C1H180JB01

Lo1 39 pF Murata GQM1885C1H390JB01

Li1 3.3 nH Coilcraft 0603CS-3N3


Li2 5 nH Coilcraft 0604HQ-5N0XJLB


Ci3 15 pF Murata GRM1885C1H180JA01

CG1, CD1 180 pF AVX 06035A181JAT2A

CG2 200 pF AVX 06035A201FAT2A

CG3, CD2 0.01 µ F MULTICOMP MC000252

CG4, CD3 0.1 µF MURATA GRM188R71H103KA01D.

CG5 22 µF KEMET T491X226K035AT

CG6 10 pF Murata GQM1885C1H100JB01

CG7, CD5 27 pF Murata GQM1885C1H270JB01

CD4 - - -

Table 4.1: Component values used for measurements, with a AFT05MS031N RF power transistor.

77

Chapter 5

Conclusions and Future Work

The goal of this master thesis was to design an active antenna for a Cospas-Sarsat distress

beacon, that can be integrated into a life jacket. This active antenna will consist of a power

amplifier and an antenna. The message sent by Cospas-Sarsat beacons, will be phase modu-

lated on a carrier frequency within the [406-406.1] MHz band.

The design of an antenna that resonates at a frequency of 406 MHz, and is suitable for

integration into a life jacket resulted in some design challenges. A frequency of 406 MHz

corresponds with a wavelength of 74 cm. Because the length of most antennas is in the

order of λ/2, which would result in a very long antenna, using miniaturising techniques were

necessary. Furthermore, the antenna should be light-weighted, flexible and planar, to obtain

unobtrusive integration into the life jacket. This reduced the amount of antenna topologies

and materials that could be used. When designing a flexible antenna for wearable applications,

bending of the antenna and the presence of the human body need to be considered, since this

can cause a shift of the antennas resonance frequency.

The antenna topology that was chosen, is an aperture coupled shorted patch antenna. This

topology has a planar structure, and the shorting plane will result in a reduced length of

the patch. Aperture coupling will increase the impedance matching compared to other feed-

ing techniques, and will make the integration of electronics onto the antenna possible. The

dimensions of the final design of the antenna are 80x190x11.8mm, and entirely exists of flex-

ible materials, so is suitable for integration into a life jacket. According the simulations, a

bandwidth of 6.6 MHz was achieved, and an |S11| of -20.9 at 406 MHz.

Measurements on multiple antennas confirmed these results, although the patch length had

to be reduced from 140 mm to 134 mm, to achieve the desired resonance frequency. The

antenna also sustained on-body tests, and when bent under different radii, keeping an |S11|below -10 dB for the desired frequency band.

78

The simulated antenna efficiency is 0.622, and a maximum gain of 0.27 dBi. The required

gain of -3dBi was not achieved in the required 90 % of the region defined by Cospas-Sarsat.

However, the gain is too low in only a very small area, and is almost -3 dBi.

The measured maximum gain is obtained by performing a three antenna measurement, and

resulted in -1.8 dBi. This gain is much lower than the simulated one, but it should be

remarked that the measurement was performed in an anechoic chamber that only has optimal

absorption for frequencies higher then 2 GHz, so nu accurate results could be obtained.

For the antenna radiation pattern, improvements are still possible. Since the gain specifi-

cations are almost met according to the simulations, the low gain could be compensated by

delivering a slightly higher power to the antenna terminals. To confirm the simulation re-

sults on the antenna radiation pattern and efficiency, the antenna should be measured in an

anechoic chamber that is appropriate for 406 MHz, or could be measured outside, where the

conditions are close to free space. The influence of the presence of a human body on the

antenna radiation pattern should also be investigated.

To satisfy the Cospas-Sarsat specifications, the power amplifier needs to deliver between 35

and 39 dBm output power to the antenna terminals. In this master thesis, it was assumed that

a signal generator will be used that will deliver 18 dBm output power, so a gain between 17

and 21 dB is required. Because the system will depend on a battery, and to avoid too much

heat generation in the transistor, a high efficiency Class-E amplifier topology was chosen.

An amplifier having dimensions of 36.5x15.4 mm was designed that had a gain of 17.6 dB

according to simulations. A measured gain of 18.1 dB was obtained, after changing two

components of the input matching network to get an |S11| below -10 dB at 406 MHz. The

simulated power added efficiency was 38.12 % and the measured efficiency was 37.13 %.

Besides the input matching, it can be stated that the measured results correspond well with

the simulations.

Because of problems with stability when testing the amplifier, changes had to be made, which

resulted in an efficiency that was lower then the desired 50%, at an input power of 18 dBm.

The cause of the instability that was encountered when using a small inductance for the RF

choke at the drain, should be further investigated, which may lead to better results and a

higher efficiency. Since the amplifiers efficiency increases with a higher input power, a small

amplifier stage could be used between the signal generator and this power amplifier, to increase

the input power, and thus the efficiency. A higher input power would also result in higher

output power of the amplifier, that could be used to compensate the low gain of the antenna.

Furthermore, the drain current of the amplifier varied over different transistors. However, it

could be noticed that the amplification was sufficient if the bias current was 0.8 A, which is

the same drain current of the simulations. Because a simple voltage devider could not be used,

79

because the bias gate voltage needed to be different to obtain this current over the different

tested transistors. Therefore, it could be useful to build a system that sets the gate voltage

in such a way that the DC drain current equals 0.8A.

The conditions in which the amplifier will be used can cause the amplifier to bend, changing

the S-parameters slightly, or the antenna can be bent, changing the input impedance of the

antenna, and thus the termination impedance of the power amplifier. This can seriously

affect the power amplifiers performance, and even worse, it can cause the amplifier to become

unstable. Because the applied power is very large, the signal generator or antenna can be

damaged if the input or output reflection coefficient of the amplifier is very high, or if an

oscillation occurs. Therefore, it can be useful to check build in a VSWR alarm and output

level detector that can turn of the beacon if something goes wrong.

The power amplifier is currently designed on rigid FR4 substrate. It should be redesigned for

the same substrate as the antenna feed structure such that it can be integrated on the antenna,

and can be directly connected to antenna feed line. However, given the small dimensions

relatively to the wavelength of the signal, the influence of the substrates dielectric permittivity

will not alter the power amplifiers operation substantially, making it possible to use the same

topology as is used for the FR4 substrate.

It can thus be concluded that, although there is still room for improvements, a robust wearable

antenna, resonating at a frequency of 406 MHz, was succesfully designed, as well as a power

amplifier which satisfies the Cospas-Sarsat specifications.

80

Appendix A

AFT05MS031N RF power LDMOS,

Freescale Semiconducter

81

AFT05MS031NR1 AFT05MS031GNR1

1RF Device DataFreescale Semiconductor, Inc.

RF Power LDMOS TransistorsHigh Ruggedness N--ChannelEnhancement--Mode Lateral MOSFETsDesigned for mobile two--way radio applications with frequencies from

136 to 520 MHz. The high gain, ruggedness and broadband performance ofthese devices make them ideal for large--signal, common source amplifierapplications in mobile radio equipment.

Typical Performance: (13.6 Vdc, TA = 25C, CW)

Frequency(MHz)

Gps(dB)

D(%)

P1dB(W)

136--174 (1,4) 23.2 62.0 31

380--450 (2,4) 18.3 64.1 31

450--520 (3,4) 17.7 62.0 31

520 (5) 17.7 71.4 33

Load Mismatch/Ruggedness

Frequency(MHz)

SignalType VSWR

Pin(W)

TestVoltage Result

155 (1) CW >65:1 at allPhase Angles

0.55(3 dB Overdrive)

17 No DeviceDegradation

420 (2) 1.6(3 dB Overdrive)



1. Measured in 136--174 MHz VHF broadband reference circuit.2. Measured in 380--450 MHz UHF broadband reference circuit.3. Measured in 450--520 MHz UHF broadband reference circuit.4. The values shown are the minimum measured performance numbers across the

indicated frequency range.5. Measured in 520 MHz narrowband test circuit.

Features Characterized for Operation from 136 to 520 MHz Unmatched Input and Output Allowing Wide Frequency Range Utilization Integrated ESD Protection Integrated Stability Enhancements Wideband — Full Power Across the Band:

136--174 MHz 380--450 MHz 450--520 MHz

225C Capable Plastic Package Exceptional Thermal Performance High Linearity for: TETRA, SSB, LTE Cost--effective Over--molded Plastic Packaging In Tape and Reel. R1 Suffix = 500 Units, 24 mm Tape Width, 13 inch Reel.Typical Applications Output Stage VHF Band Mobile Radio Output Stage UHF Band Mobile Radio

Document Number: AFT05MS031NRev. 1, 4/2013

Freescale SemiconductorTechnical Data

136--520 MHz, 31 W, 13.6 VWIDEBAND

RF POWER LDMOS TRANSISTORS

AFT05MS031NR1AFT05MS031GNR1

TO--270--2PLASTIC

AFT05MS031NR1

Figure 1. Pin Connections

(Top View)

DrainGate

Note: The backside of the package is thesource terminal for the transistor.

TO--270--2 GULLPLASTIC

AFT05MS031GNR1

Freescale Semiconductor, Inc., 2012--2013. All rights reserved.

2RF Device Data

Freescale Semiconductor, Inc.


Table 1. Maximum Ratings

Rating Symbol Value Unit

Drain--Source Voltage VDSS --0.5, +40 Vdc

Gate--Source Voltage VGS --6.0, +12 Vdc

Operating Voltage VDD 17, +0 Vdc

Storage Temperature Range Tstg --65 to +150 C

Case Operating Temperature Range TC --40 to +150 C

Operating Junction Temperature Range (1,2) TJ --40 to +225 C

Total Device Dissipation @ TC = 25CDerate above 25C

PD 2941.47

WW/C

Table 2. Thermal Characteristics

Characteristic Symbol Value (2,3) Unit

Thermal Resistance, Junction to CaseCase Temperature 79C, 31 W CW, 13.6 Vdc, IDQ = 10 mA, 520 MHz

RJC 0.67 C/W

Table 3. ESD Protection Characteristics

Test Methodology Class

Human Body Model (per JESD22--A114) 2, passes 2500 V

Machine Model (per EIA/JESD22--A115) A, passes 100 V

Charge Device Model (per JESD22--C101) IV, passes 2000 V

Table 4. Moisture Sensitivity Level

Test Methodology Rating Package Peak Temperature Unit

Per JESD22--A113, IPC/JEDEC J--STD--020 3 260 C

Table 5. Electrical Characteristics (TA = 25C unless otherwise noted)

Characteristic Symbol Min Typ Max Unit

Off Characteristics

Zero Gate Voltage Drain Leakage Current(VDS = 40 Vdc, VGS = 0 Vdc)

IDSS — — 2 Adc

Zero Gate Voltage Drain Leakage Current(VDS = 13.6 Vdc, VGS = 0 Vdc)

IDSS — — 1 Adc

Gate--Source Leakage Current(VGS = 5 Vdc, VDS = 0 Vdc)

IGSS — — 600 nAdc

On Characteristics

Gate Threshold Voltage(VDS = 10 Vdc, ID = 115 Adc)

VGS(th) 1.6 2.1 2.6 Vdc

Drain--Source On--Voltage(VGS = 10 Vdc, ID = 1.2 Adc)

VDS(on) — 0.13 — Vdc

Forward Transconductance(VDS = 10 Vdc, ID = 7.5 Adc)

gfs — 5.8 — S

1. Continuous use at maximum temperature will affect MTTF.2. MTTF calculator available at http://www.freescale.com/rf. Select Software & Tools/Development Tools/Calculators to access MTTF

calculators by product.3. Refer to AN1955, Thermal Measurement Methodology of RF Power Amplifiers. Go to http://www.freescale.com/rf.

Select Documentation/Application Notes -- AN1955.(continued)



Table 5. Electrical Characteristics (TA = 25C unless otherwise noted) (continued)

Characteristic Symbol Min Typ Max Unit

Dynamic Characteristics

Reverse Transfer Capacitance(VDS = 13.6 Vdc 30 mV(rms)ac @ 1 MHz, VGS = 0 Vdc)

Crss — 1.6 — pF

Output Capacitance(VDS = 13.6 Vdc 30 mV(rms)ac @ 1 MHz, VGS = 0 Vdc)

Coss — 49.5 — pF

Input Capacitance(VDS = 13.6 Vdc, VGS = 0 Vdc 30 mV(rms)ac @ 1 MHz)

Ciss — 109 — pF

Functional Tests (1) (In Freescale Narrowband Test Fixture, 50 ohm system) VDD = 13.6 Vdc, IDQ = 10 mA, Pout = 31 W, f = 520 MHz

Common--Source Amplifier Power Gain Gps 16.5 17.7 19.0 dB

Drain Efficiency D 70.0 71.4 — %

Load Mismatch/Ruggedness (In Freescale Test Fixture, 50 ohm system) IDQ = 10 mA

Frequency(MHz)

SignalType VSWR

Pin(W) Test Voltage, VDD Result

520 CW >65:1 at all Phase Angles 1.1(3 dB Overdrive)

17 No Device Degradation

1. Measurements made with device in straight lead configuration before any lead forming operation is applied. Lead forming is used for gullwing (GN) parts.

4RF Device Data



TYPICAL CHARACTERISTICS

201

1000

0 84

VDS, DRAIN--SOURCE VOLTAGE (VOLTS)

Figure 2. Capacitance versus Drain--Source Voltage

C,CAPACITANCE(pF)

12

10

160

5

4

VDS, DRAIN--SOURCE VOLTAGE (VOLTS)

Figure 3. Drain Current versus Drain--Source Voltage

2

VGS = 4.25 Vdc

4

3

1

8 12 16 20

I DS,DRAINCURRENT(AMPS)

3.5 Vdc

3.25 Vdc

3 Vdc

TA = 25C

2.75 Vdc

250

109

90

TJ, JUNCTION TEMPERATURE (C)

Figure 4. MTTF versus Junction Temperature -- CW

Note: MTTF value represents the total cumulative operating timeunder indicated test conditions.

MTTF calculator available at http://www.freescale.com/rf. SelectSoftware & Tools/Development Tools/Calculators to access MTTFcalculators by product.

107

106

104

110 130 150 170 190

MTTF(HOURS)

210 230

108

105

VDD = 13.6 Vdc

0

100

Measured with 30 mV(rms)ac @ 1 MHz, VGS = 0 Vdc

Crss

Ciss

Coss3.75 Vdc

6

7

4 Vdc

ID = 2.5 Amps

3.2 Amps

3.9 Amps



520 MHz NARROWBAND PRODUCTION TEST FIXTURE

Figure 5. AFT05MS031NR1 Narrowband Test Circuit Component Layout — 520 MHz

C10

C9

C13 C14

C15

C16

C11

C12

C7

C2 C8L2C5

C3

L1

C4

C6

AFT05MS031NRev. 1

C1

B2

C17

C18

B1B3

CUTOUTAREA

Table 6. AFT05MS031NR1 Narrowband Test Circuit Component Designations and Values — 520 MHzPart Description Part Number Manufacturer

B1, B2, B3 RF Beads, Long 2743021447 Fair--Rite

C1 22 F, 35 V Tantalum Capacitor T491X226K035AT Kemet

C2, C14 0.01 F Chip Capacitors C0805C103K5RAC Kemet

C3, C13 0.1 F Chip Capacitors CDR33BX104AKWS Kemet

C4 200 pF Chip Capacitor ATC100B201JT300XT ATC

C5 6.2 pF Chip Capacitor ATC100B6R2JT500XT ATC


C7, C16 180 pF Chip Capacitors ATC100B181JT200XT ATC


C9, C10, C11, C12 36 pF Chip Capacitors ATC100B360JT500XT ATC



C18 470 F, 63 V Electrolytic Capacitor SME63V471M12X25LL United Chemi--Con

L1 43 nH, 10 Turn Inductor B10TJLC Coilcraft

L2 56 nH Inductor 1812SMS--56NJLC Coilcraft

PCB 0.030, r = 2.55 AD255A Arlon

6RF Device Data



RFINPUT

RFOUTPUT

Z1

0.199

0.082

Microstrip

Z2

0.017

0.082

Microstrip

Z3*

0.670

0.082

Microstrip

Z4*

0.560

0.060

Microstrip

Z5*

0.370

0.082

Microstrip

Z6

0.079

0.082

Microstrip

Z7

0.352

0.082

Microstrip

*Line

lengthincludes

microstrip

bends

Z8

0.190

0.270

Microstrip

Z9

0.257

0.275

Microstrip

Z10

0.145

0.275

Microstrip

Z11

0.091

0.082

Microstrip

Z12*

0.1322

0.082

Microstrip

Z13*

0.1420

0.082

Microstrip

Z14

0.315

0.082

Microstrip

Figure

6.AFT05MS031N

R1NarrowbandTestCircuitSchem

atic—

520MHz

Table7.AFT05MS031N

R1NarrowbandTestCircuitMicrostrips—

520MHz

Description

Microstrip

Description

Microstrip

L1

Z4Z3

C5Z1

Z2Z8

Z7Z6

Z10

Z9

L2

V DS

V GS

Z5

Z12

Z11

Z13

C11

Z14

C2C3

C4C7

C6

C8C9

C12

C16

C13

C14

C15

C17

C1+

C10

C18

+

Appendix B

List of measurements on the ouput

power of the amplifier

88

Input Output IDS VGS Output IDS VGS

power power power

- 10 6.70 0.56 3.08 9.16 1.4 3.32

-5 11.51 0.66 3.12 13.68 1.3 3.35

0 17.8 0.66 3.13 18.6 1.2 3.3

5 22.05 0.81 3.16 23.33 1.17 3.22

10 27.82 0.97 3.18 29.0 1.3 3.24

15 29.94 0.69 2.94 33.7 1.57 3.14

18 34.8 0.89 3.16 36.34 1.14 3.28

Table B.1: Measured output power of prototype 1.

Input Output IDS(A) VGS(A) Output IDS(A) VGS(A)

power power power

- 15 4.05 0.79 3.11 4.91 1.68 3.32

-10 9.08 0.74 8.42 9.89 1.29 3.23

-5 14.06 0.75 3.1 16.46 11.94 3.40

0 18.96 0.92 3.03 19.86 19.2 1.49

5 23.86 0.88 3.10 24.72 1.14 24.06

10 26.91 0.53 2.95 29.51 1.33 3.24

11 29.49 0.89 3.12 30.51 1.77 3.26

12 29.34 0.73 3.03 31.13 1.71 3.29

13 30.83 0.91 3.09 31.83 1.22 3.21

14 31.54 0.93 3.06 32.61 1.66 3.29

15 31.65 0.82 3.04 33.46 1.62 3.31

16 32.78 0.99 3.09 33.61 1.19 3.17

17 32.6 0.62 2.98 34.29 1.29 3.26

Table B.2: Measured output power of prototype 2.

89

Bibliography

[1] Cospas-sarsat participants. https://www.cospas-sarsat.org/en/

about-cospas-sarsat/participating-countries-organisations.

[2] K.K. Ivanov Yu.G. Zurabov and A.D. Kuropyatnikov. Cospas-sarsat satellite system.

In The Third International Conference on Satellite Communications, 1998., volume 1,

pages 156 – 158, Moscow, September 1998. Print ISBN 5-93184-002-8.

[3] D. Levesque. Cospas-Sarsat Information Bulletin Enclosure -30th Anniversary Special

Issue, pages 1–9. Cospas-Sarsat Secretariat, February 2009.

[4] Cospas-Sarsat Secretariat. Cospas-Sarsat Information Bulletin issue 25, page 4. 2009.

[5] The cospas-sarsat low-altitude earth orbit (leosar) system for search and rescue. https:

//www.cospas-sarsat.org/en/system/systemoverview/leosar-system.

[6] The cospas-sarsat geostationary search and rescue (geosar) system. https://www.

cospas-sarsat.org/en/system/systemoverview/geosar-system.

[7] B. Niehoefer A. Lewandowski and C. Wietfeld. Performance evaluation of satellite-based

search and rescue services: Galileo vs. cospas-sarsat. In Vehicular Technology Conference,

2008. VTC 2008-Fall. IEEE 68th., pages 1–5, Calgary, BC, September 2008. ISSN 1090-

3038.

[8] Cospas-Sarsat Secretariat. Cospas-sarsat 406 mhz frequency management plan. C/S

T.012, Issue 1 - Revision 6, October 2009.

[9] Cospas-Sarsat Secretariat. Specifications for cospas-sarsat 406 mhz distress beacons. C/S

T.001, Issue 3 - Revision 13, October 2012.

[10] MRB1002 data sheet.

[11] P. ; Scudeller Y. ; Toutain Serge Alnukari, A. ; Guillemet. Active heatsink antenna for

radio-frequency transmitter. Advanced Packaging, IEEE Transactions on, 33(1):139–146,

February 2010. ISSN 1090-3038.

90

https://www.cospas-sarsat.org/en/about-cospas-sarsat/participating-countries-organisations

https://www.cospas-sarsat.org/en/about-cospas-sarsat/participating-countries-organisations

https://www.cospas-sarsat.org/en/system/systemoverview/leosar-system

https://www.cospas-sarsat.org/en/system/systemoverview/leosar-system

https://www.cospas-sarsat.org/en/system/systemoverview/geosar-system

https://www.cospas-sarsat.org/en/system/systemoverview/geosar-system

[12] F. ; Rogier H. Dierck, A. ; Declercq. Review of active textile antenna co-design and

optimization strategies. In RFID-Technologies and Applications (RFID-TA), 2011 IEEE

International Conference on, pages 194–201, Sitges, September 2011. IEEE. Digital

Object Identifier :10.1109/RFID-TA.2011.6068637.

[13] Xuan Hui Wu; Zhi Ning Chen. Comparison of planar dipoles in uwb applications.

IEEE TRANSACTIONS ON ANTENNAS AND PROPAGATION, 53(6):1973–1983,

June 2005.

[14] Yuh-Yih Lu ; Shih-Chiang Wei ; Chun-Yi Li ; Hsiang-Cheh Huang. Design of 2.45ghz

planar meander dipole antenna. In Intelligent Information Hiding and Multimedia Signal

Processing (IIH-MSP), 2011 Seventh International Conference on, pages 5–8, Dalian,

October 2011. IEEE. Digital Object Identifier :10.1109/IIHMSP.2011.17.

[15] Christian Sturm Werner Wiesbeck, Grzegorz Adamiuk. Basic properties and design

principles of uwb antennas. Proceedings of the IEEE, 97(2):372–385, February 2009.

[16] Dr. V.S. Tripathi Indrasen Singh. Micro strip patch antenna and its applications: a

survey. Int. J. Comp. Tech. Appl., pages 1595–1599, September-October 2011.

[17] N. Merabtine A. Boualleg. Analysis of radiation patterns of rectangular microstrip an-

tennas with uniform substrate. Semiconductor Physics, Quantum Electronics & Opto-

electronics., pages 88–91, volume = 8, number = 3,, 2005.

[18] Hendrik Rogier. Antennas and propagation. Ghent university.

[19] T. Sumanth N. Mohana Rao R. Anil Kumar Y.Harish K. Praveen Kumar, K. San-

jeeva Rao. Effect of feeding techniques on the radiation characteristics of patch antenna:

Design and analysis. International Journal of Advanced Research in Computer and Com-

munication Engineering, 2(2):1276–1281, February 2013.

[20] D.B.O. Konditi K.V. Rop. Performance analysis of a rectangular microstrip patcha

antenna on different dielectric substrates. Innovative Systems Design and Engineering,

3(8):7–14, 2012.

[21] Muhammad Kamran Ishfaq Stephen J. Boyes Hassan Tariq Chattha, Yi Huang. A

comprehensive parametric study of planar inverted-f antenna. Wireless Engineering and

Technology, pages 1–11, 2012.

[22] Sudhakar Rao and Nuria Llombart. Body-worn antennas making a splash: Lifejacket-

integrated antennas for global search and rescue satellite system. IEEE Antennas and

Propagation Magazine, Vol. 55 No. 2, pages 324–341, April 2013.

[23] Bram Cuvelier. Design of an integrated textile antenna for the cospas-sarsat system.

Master’s thesis, Ghent university, 2012-2013.

91

[24] G. Manara Andrea A. Serra, P. Nepa. A wearable multi antenna system on a life jacket

for cospas sarsat rescue applications. 2011 IEEE International Symposium on Antennas

and Propagation (APSURSI), pages 1319–1322, July 2011.

[25] Nathan O. Sokal. Class-e high-efficiency rf/microwave power amplifiers: Principles of

operation, design procedures, and experimental verification. Analog Circuit Design, pages

269–301, 2002.

[26] Dusan Milosevic. High-Efficiency Linear RF Power Amplification: A Class-E Based EER

Study Case. PhD thesis, Technische Universiteit Eindhoven, 2009.

[27] Stephen D. Targonski and David M. Pozar. Design of wideband circularly polarized

aperture coupled microstrip antennas. Microstrip Antennas: The analysis and Design of

Microstrip Antennas and Arrays, pages 130–133, 1995.

[28] Alexander Kuchar. Aperture coupled microstrip patch antenna array. Master’s thesis,

Technische Universitat Wien, 1996.

[29] Kin-Lu Wong. Compact and Broadband Microstrip Antennas, pages 48–49. Wiley, 2002.

[30] S.C.Shrivastava Zarreen Aijaz. Coupling effects of aperture coupled microstrip antenna.

nternational Journal of Engineering Trends and Technology, pages 130–133, July to Aug

2011.

[31] David M. Pozar. A review of aperture coupled microstrip antennas: History, operation,

development, and applications. May 1996.

[32] Alan D. Sokal Nathan O. Sokal. Class e- a new class of high-efficiency tuned single-ended

switching power amplifiers. IEEE Journal of Solid-State Circuits, SC-10(3):168–176,

June 1975.

[33] Firas Mohammed Ali Al-Raie. Design of input matching networks for class-e rf power

amplifiers. High frequency electronics, pages 40–48, January 2011.

[34] Saad Al-Shahrani. Design of Class -E Radio Frequency Power Amplifier. PhD thesis,

Faculty of the Virginia Polytechnic Institute and State University, 2001.

[35] H. Rogier F. Declercq, I. Couckuyt and T. Dhaene. Environmental high frequency char-

acterization of fabrics based on a novel surrogate modelling antenna technique. IEEE

Trans. Antennas Propag., 61(10):5200–5213, October 2013.

[36] M.-C. Huynh and W. Stutzman. Ground plane effects on planar inverted-f antenna (pifa)

performance. IEE Proc.-Microw. Antennas Propag, 150:209–213, August 2003.

[37] PD54008L-E data sheet.

92

[38] AFT05MS031 data sheet.

[39] Jordi Ambros Moreno. Design and assembly of a class e power amplifier 2ghz, 2011.

[40] Theodore S. Rappaport. Wireless Communications Principles and Practice Second Edi-

tion., page 108. Prentice-Hall, Inc., 2010.

[41] David B. Rutledge Sanggeun Jeon, Almudena SuA¡rez. Global stability analysis and

stabilization of a class-e/f amplifier. Submitted to IEEE Transactions on Microwave

Theory and Technique.

93

List of Figures

2.1 Cospas-Sarsat (C/S) principle of operation. . . . . . . . . . . . . . . . . . . . . 5

2.2 Message format. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8

2.3 Biphase L encoding principle. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9

2.4 Architecuture of the “MRB1002” signal generator of Syrlinks. [10] . . . . . . . 9

2.5 Vertical cross section of the radiation pattern with areas where the gain must

be between -3 and 4 dBi. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11

2.6 Possible places in the life jacket to integrate the beacon. . . . . . . . . . . . . . 12

2.7 Planar dipole antenna. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14

2.8 Planar meandered dipole antenna . . . . . . . . . . . . . . . . . . . . . . . . . . 14

2.9 Planar bowtie antenna . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14

2.10 Two examples of planar monopole antennas. . . . . . . . . . . . . . . . . . . . . 15

2.11 Topology of a microstrip patch antenna. . . . . . . . . . . . . . . . . . . . . . . 15

2.12 Electrical field distribution in a patch antenna. . . . . . . . . . . . . . . . . . . 16

2.13 Different feeding techniques for microstrip patch antennas. . . . . . . . . . . . . 17

2.14 Shorted patch antenna. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18

2.15 Electrical field and current magnitude in patch antenna. . . . . . . . . . . . . . 18

2.16 Current flowing in a meandered patch antenna. . . . . . . . . . . . . . . . . . . 19

2.17 Principle of Class-A power amplifier. . . . . . . . . . . . . . . . . . . . . . . . . 20

2.18 Principle of Class-B power amplifier. . . . . . . . . . . . . . . . . . . . . . . . . 21

94

2.19 Principle of Class-E power amplifier. . . . . . . . . . . . . . . . . . . . . . . . . 21

3.1 Spurious emission mask for the [406.0-406.1] MHz band that is required for

each individual channel [9] . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24

3.2 Topology of a planar inverted-F antenna (PIFA) with its electrical field distri-

bution. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27

3.3 Measured |S11| of a coaxial fed PIFA antenna. . . . . . . . . . . . . . . . . . . . 28

3.4 Structure of an aperture coupled shorted patch antenna. . . . . . . . . . . . . . 29

3.5 Different shapes of apertures that can be used to increase coupling. . . . . . . . 29

3.6 Ideal waveforms for obtaining highly efficient class-E operation . . . . . . . . . 31

3.7 Schematic representation of the operation principle of class E amplifiers. . . . . 32

3.8 Topology of an aperture coupled shorted patch antenna with a H-shape slot. . 36

3.9 Simulated |S11| of the initial wide patch antenna. . . . . . . . . . . . . . . . . . 37

3.10 Simulated |S11| of the narrow patch antenna. . . . . . . . . . . . . . . . . . . . 38

3.11 ADS antenna design simulated in CST with different groundplane sizes. . . . . 39

3.12 ADS antenna design compared to CST simulation with groundplane of size

500x500mm2. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40

3.13 Simulated |S11| of the antenna in CST. . . . . . . . . . . . . . . . . . . . . . . . 41

3.14 3D radiation pattern simulated in CST. . . . . . . . . . . . . . . . . . . . . . . 41

3.15 Simulated radiation pattern in the xz-plane. . . . . . . . . . . . . . . . . . . . . 42

3.16 Simulated radiation pattern in the yz-plane. . . . . . . . . . . . . . . . . . . . . 43

3.17 Simulated reflection coefficients for small variations in design parameters. . . . 44

3.18 I-V characteristic of the MOSFET, with chosen bias point . . . . . . . . . . . . 45

3.19 Basic class E circuit with MOSFET to function as a switch. . . . . . . . . . . . 46

3.20 Current and voltage waveforms at the drain with component values of Table

3.7, and an input power of 100 mW (20dBm) . . . . . . . . . . . . . . . . . . . 47

95

3.21 Current and voltage waveforms at the drain with component values of Table

3.8, and an input power of 100 mW (20dBm). . . . . . . . . . . . . . . . . . . . 47

3.22 Drain source voltage and current of the class-E amplifier with input and output

matching, at an input power of 18 dBm. . . . . . . . . . . . . . . . . . . . . . . 48

3.23 Impedance transformation on the Smith chart for output matching. . . . . . . . 49

3.24 Class E-type amplifier with input and output matching. . . . . . . . . . . . . . 49

3.25 S-parameters simulation for amplifier including matching networks. . . . . . . . 51

3.26 Output power as a function of input power with ideal components, after tuning. 52

3.27 PAE of the amplifier as a function of input power, with ideal components, after

tuning.. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52

3.28 Current and voltage waveforms at the drain, after tuning and with ideal com-

ponents. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53

3.29 S-parameters with matching networks, after tuning. . . . . . . . . . . . . . . . 54

3.30 Power Amplifier circuit topology. . . . . . . . . . . . . . . . . . . . . . . . . . . 54

3.31 Layout of the final design. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55

3.32 Transistor case style [38]. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55

3.33 Efficiency of the power amplifier. . . . . . . . . . . . . . . . . . . . . . . . . . . 56

3.34 Output power versus input power. . . . . . . . . . . . . . . . . . . . . . . . . . 57

3.35 S-parameters for different values of LRF,choke1. . . . . . . . . . . . . . . . . . . 57

3.36 Output power versus input power of the final design. . . . . . . . . . . . . . . . 59

3.37 PAE versus input power of the final design. . . . . . . . . . . . . . . . . . . . . 59

3.38 S-parameter simulation. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60

4.1 Measured and simulated reflection coefficients of the aperture coupled shorted

patch antenna. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61

4.2 Measured reflection coefficients of the aperture coupled shorted patch antenna

for different patch lengths. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62

96

4.3 S11 of the aperture coupled shorted patch antenna with a patch length of 134

mm. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63

4.4 Measurement results for an antenna with a patch length = 134 mm and t = 36

mm . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64


for different conductive materials. . . . . . . . . . . . . . . . . . . . . . . . . . . 65


with copper foil feed line and copper-coated tafetta patch and ground plane. . . 65

4.7 Measured radiation pattern of the aperture coupled shorted patch antenna with

copper foil feed line in the xz- and yz-plane at 406 MHz. . . . . . . . . . . . . . 67

4.8 Illustration of the axial system used. . . . . . . . . . . . . . . . . . . . . . . . . 67

4.9 Measured reflection coefficients for three different bending radii.. . . . . . . . . 68

4.10 Direction in which the antenna is bent. . . . . . . . . . . . . . . . . . . . . . . . 68

4.11 Reflection coefficient when a human body is present. . . . . . . . . . . . . . . . 69

4.12 Measured S-parameters with component values from Table 3.11, and VGS =

2.7V. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71

4.13 Measured S-parameters with component values from Table 3.11, and VGS =

2.9V. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71

4.14 Simulated S-parameters, compared to the measured S-parameters with compo-

nents from Table 3.11, both for a VGS = 2.7V. . . . . . . . . . . . . . . . . . . 72

4.15 Measured S-parameters with component values from table 4.1, when VGS =

2.7V and IDS = 0.044 A. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73


3.03V and IDS = 0.79 A. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73


3.21V and IDS = 1.88 A. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74


3.21V and IDS = 1.36 A, compared with the simulated results. . . . . . . . . . 74

4.19 Measured output power versus input power for VGS = 3.1V. . . . . . . . . . . . 75

97

4.20 Measured output power versus input power for VGS = 3.3V, compared to the

simulated output power. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75

4.21 Measured spectrum at an input power of 15 dBm, with the frequency given

relative to 406 MHz. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76

4.22 Measured spectrum at an input power of 18 dBm. . . . . . . . . . . . . . . . . 76

98

List of Tables

3.1 Design goals of the power amplifier, given an input power of 18 dBm. . . . . . 25

3.2 Design goals for the antenna. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26

3.3 Overview of available materials. . . . . . . . . . . . . . . . . . . . . . . . . . . . 34

3.4 Parameters of the initial wide aperture coupled patch antenna (mm). . . . . . . 37

3.5 Parameters of the narrow PIFA antenna (mm). . . . . . . . . . . . . . . . . . . 38

3.6 Dimensions of the final design in CST (mm). . . . . . . . . . . . . . . . . . . . 41

3.7 Component values for QL = 5 and Ld = 2.8 nH. . . . . . . . . . . . . . . . . . 46

3.8 Component values for QL = 4 and Ld = 6 nH. . . . . . . . . . . . . . . . . . . 46

3.9 Component values of circuit with matching network at in- and output. . . . . . 50

3.10 Component values for circuit after tuning. . . . . . . . . . . . . . . . . . . . . . 50

3.11 Component values after adding the layout. . . . . . . . . . . . . . . . . . . . . . 58

3.12 Component values of the amplifier, with a AFT05MS031N RF power transistor. 60

4.1 Component values used for measurements, with a AFT05MS031N RF power

transistor. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 77

B.1 Measured output power of prototype 1. . . . . . . . . . . . . . . . . . . . . . . 89

B.2 Measured output power of prototype 2. . . . . . . . . . . . . . . . . . . . . . . 89

99

design of a wearable active cospas-sarsat beacon …

Documents