Design of Highly Efficient Broadband Class-E Power Amplifier Using

12 Pages·2012·2.12 MB·English

by  Kenle Chen

3162 IEEETRANSACTIONSONMICROWAVETHEORYANDTECHNIQUES,VOL.59,NO.12,DECEMBER2011
Design of Highly Efficient Broadband Class-E
Power Amplifier Using Synthesized
Low-Pass Matching Networks
Kenle Chen,StudentMember,IEEE, and Dimitrios Peroulis,Member,IEEE
Abstract—Anewmethodologyfordesigningandimplementing TABLEI
high-efficiency broadband Class-E power amplifiers (PAs) using STATE-OF-THE-ARTHIGH-EFFICIENCYBROADBANDPAS
high-order low-pass filter-prototype is proposed in this paper. A
GaN transistor is used in this work, which is carefully modeled
and characterized to prescribe the optimal output impedance
for the broadband Class-E operation. A sixth-order low-pass
filter-matching network is designed and implemented for the
output matching, which provides optimized fundamental and
harmonic impedances within an octave bandwidth ( -band).
SimulationandexperimentalresultsshowthatanoptimalClass-E
PA is realized from 1.2 to 2 GHz (50%) with a measured effi-
ciencyof80%–89%,whichisthehighestreportedtodayforsuch aboveanoctave.Thedistributedamplifier(DA)isacommonso-
a bandwidth. An overall PA bandwidth of 0.9–2.2 GHz (84%)
lution for wideband amplification [5]. However, this approach
is measured with 10–20-W output power, 10–13-dB gain, and
requires multiple devices, which results in a low overall effi-
63%–89% efficiency throughout the band. Furthermore, the
Class-E PA is characterized through measurements using con- ciency.Anotherpossiblesolutionbasedontheoptimizationof
stant-envelop global system for mobile communications signals, thefundamentalharmonicimpedancehasbeenpresentedin[6]
indicatingafavorableadjacentchannelpowerratiofrom 40to and [7], achieving multioctave bandwidths. However, signifi-
50dBcwithintheentirebandwidth.
canthigherorderharmonicsremaininsuchdesigns,leadingto
Index Terms—Broadband, Class-E, GaN, high efficiency, high relatively low efficiencies ( 50% in [6] and 20% in [7]) and
power, low-pass matching network, power amplifier (PA), syn- harmonicdistortion.
thesis.
Thus, realizing high-power and high-efficiency PAs within
a broad bandwidth is a critical research area. To date, several
design techniques have been proposed and demonstrated with
I. INTRODUCTION
an optimization of bandwidth and efficiency [8]–[12]. Several
critical features of these PAs are summarized in Table I. It is
H IGH-POWER and high-efficiency power amplifiers
important to note that the matching network topologies used
(PAs) are very important for modern wireless commu-
inthosedesignsareequivalenttolow-pass prototypeswithan
nication systems, especially those in base-stations, to achieve
order .Higherorderlow-passprototypeshavenotyetbeen
low-cost and highly reliable transmitters. High-efficiency PAs
widelyappliedinthebroadbandPAdesign,althoughtheycan
are commonly realized using switch-mode topologies, such
potentially yield improved matching quality and wider band-
as Class-D, Class-E, and Class-F , or harmonic-tuned cir-
width.
cuitries,likeClass-FandClass-J[1].Thosehigh-efficiencyPAs
Herein, we present a practical method for designing a
usuallyrequireahighsaturationlevelofthedeviceandaccurate
broadbandhigh-efficiencyClass-EPAthatemployshigh-order
harmonicengineering,leadingtobandwidthrestrictionsintheir
low-pass circuit topologies at the input and output
frequencyresponse[2]–[4].
.ComparedtoClass-Ddesigns,thepresentedClass-E
Nevertheless, future communication systems, e.g., WiMax
PA exhibits simpler circuitry and reduced sensitivity to para-
andLTE,wouldrequirewiderbandwidths,notonlyforthecov-
sitics. Compared to Class-Fand Class-F designs, it is more
erageofmultiplefrequencybands,butalsoduetotheincreasing
forgivingtoharmonicterminationsrequirementsandallowsfor
effectivebandwidthofthesignal,i.e.,upto100MHz.Themil-
a broadband performance. The presented design is based on a
itarywirelesssystemsrequireevenwiderbandwidths,typically
commerciallyavailable25-WCreeGaNHEMT.Athree-stage
Chebyshev low-pass network is synthesized for
ManuscriptreceivedMay16,2011;acceptedAugust02,2011.Dateofpubli- theoutputmatching,anditisimplementedusingtransmission
cationOctober20,2011;dateofcurrentversionDecember14,2011.Thiswork lines,asthosein[13]–[15],whichyieldabetterPAperformance
wassupportedbyRockwellCollinsInc.
than that using lumped elements [15]. Optimal fundamental,
The authors are with the School of Electrical and Computer Engineering
andtheBirckNanoTechnologyCenter,PurdueUniversity,WestLafayette,IN second-harmonic, and third-harmonic impedance terminations
47906USA(e-mail:[email protected];[email protected]). are provided for the transistor output over a broad bandwidth.
Colorversionsofoneormoreofthefiguresinthispaperareavailableonline
TheoutputparasiticsofthispackagedGaNtransistorarecare-
athttp://ieeexplore.ieee.org.
DigitalObjectIdentifier10.1109/TMTT.2011.2169080 fullycharacterizedandde-embeddedinconfirmingawideband
0018-9480/$26.00©2011IEEE
Copyright © 2011 IEEE. Reprinted from IEEE Transactions on Microwave Theory and Techniques, Vol. 59, No. 12, December 2011.
This material is posted here with permission of the IEEE. Such permission of the IEEE does not in any way imply IEEE endorsement of any of
Cree’s products or services. Internal or personal use of this material is permitted. However, permission to reprint/republish this material for advertis-
ing or promotional purposes or for creating new collective works for resale or redistribution must be obtained from the IEEE by writing to
[email protected] By choosing to view this document, you agree to all provisions of the copyright laws protecting it.
CHENANDPEROULIS:DESIGNOFHIGHLYEFFICIENTBROADBANDCLASS-EPA 3163
startflowingafterthevoltagehasreachedzero(softswitching).
Thevalueof and canbedetermineduniquelybyapplying
thesetwoconstraints,leadingto
(4)
(5)
Consequently,asshownintherectangularboxinFig.1,thereis
nooverlapbetweenthetransientdrain(orcollector)currentand
voltage, which leads to a zero dc power dissipation and 100%
Fig.1. IdealClass-EPAtopology.
drainefficiency.
UsingFourier-seriesexpansions[13],[17],theoptimalfun-
damentalload,yieldingperfectClass-Eoperation,canbedeter-
Class-E behavior. This design achieves a state-of-the-art per-
minedby
formancecomparedtotherecentlypublishedresults(TableI).
In particular, it exhibits the highest broadband reported effi-
(6)
ciency. Furthermore, modulated testing using a global system
for mobile communications (GSM) signal shows a very good
potential of this Class-E PA for amplifying constant-envelop Thisimpedancepresentsinductive,whichisillustratedinFig.1.
signals. IntheidealClass-Emode,thetotalcurrentthroughthecom-
binedswitch-capacitor tankis a puresinusoidal wave,and the
harmonicsareentirelyduetothevoltage.Inturn,theidealim-
II. BROADBANDCLASS-EPA
pedancesathigherorderharmonicfrequenciesareinfinite
A. ReviewoftheClassicClass-EPATheory
(7)
The conventional ideal Class-E PA topology is depicted in
Fig. 1. For this PA circuitry [16], the transistor is considered
as aswitch and acapacitor is connectedinparallelwith B. BroadbandClass-EPAScheme
it. If the transistor switch is turned on, the current flows en-
The classic Class-E topology shown in Fig. 1 is indeed for
tirely through the switch (field-effect transistor (FET)’s drain
the narrowband design, as the series resonator has a lim-
andsource,orbipolarjunctiontransistor(BJT)’scollectorand
itedfrequencyresponse.Equation(6)alsoindicatesthattheop-
base)andthevoltageiszero.Thiscurrentcanbeexpressedas
timumloadimpedanceisfrequencyvarying.Toimplementthe
(1) broadbandClass-EPA,twoproblemsaretobeaddressed.First,
a matching network is needed to transform to for the
Whentheswitchisturnedoff,thecurrentflowsentirelyintothe switch-capacitortankoverasignificantbandwidth,e.g., 50%.
capacitor,whichischargedsimultaneously.Duringthisoff-state Second,abroadbandfilter,whosepassbandcoversthedesired
interval,thevoltageonthisparallelcapacitorisgivenby bandwidth,isneededtoprovideinfiniteimpedanceatharmonic
frequencies, replacing the resonator. In this research, the
matching network and filter is co-designed in order to reduce
thecircuitcomplexity.
Furthermore,theperfectswitchingbehaviorisnotachievable
(2)
atthegigahertzfrequencylevelbecauseofthenonidealeffects
ofthetransistorandnonsquarewavedrivensignalattheinput.
Practically, the transistor gate/base is biased at the pinch-off
TherearetwoboundaryconditionsfortheidealClass-Eoper-
point.Theinputsignalnowneedstobesufficientlylargetoturn
ation[13],[16],[17],referredaszerovoltageswitching(ZVS)
the transistor on during the entire positive half duty-cycle and
and zero voltage derivative switching (ZVDS) conditions. As-
subsequentlyturnitoffduringthenegativehalfduty-cycle[1].
sumetheswitchisturnedoffat andturnedonat ,
TheproposedrealisticbroadbandClass-EPAtopologyisshown
thosetwoconditionsaregivenby
inFig. 2.
It is important to note, however, that the bandwidth of this
matching network design method is limited to one octave [8]
becausethe secondharmonicofthelow-endfundamentalwill
(3) occur in the passband if the bandwidth is above one octave.
Thus, the original target frequency band is from 1 to 2 GHz
where denotesthetimeperiodofoneClass-Edutycycle.The ( -band).Toachieveamultioctavebandwidth,onecandesign
ZVSconditionpreventssimultaneousnonzerovoltageandcur- multiplematchingnetworksfordifferentbandsanduseswitches
rentacrosstheswitchdevice,andZVDSenforcesthecurrentto toselecttheproperone.
3164 IEEETRANSACTIONSONMICROWAVETHEORYANDTECHNIQUES,VOL.59,NO.12,DECEMBER2011
Fig.2. BroadbandClass-EPAtopology.
Fig. 4. Characterization of Cree GaN CGH40025 at package plane
throughout -band. (a) De-embedded optimal output impedance from ideal
Class-E model. (b) Optimum output and input impedances extracted from
load–pullandsource–pullsimulations.
values, i.e., , , , , , and , are estimated
usingthemethodpresentedin[19].Thepackagemodelispro-
videdbythemanufacturer.Usingthispackagedtransistortode-
sign a Class-E PA, these parasitic effects need to be carefully
consideredinordertoperformasufficientlyaccuratematching.
Fig.3. TypicalGaNtransistormodelshowingtheparasitics[18],[19]. Especially,theoutputcapacitanceofthetransistor canbe
takenadvantagetoreplacetheshuntcapacitor intheideal
topology,whichisthemainadvantageoftheClass-Etopology
C. DesignofBroadbandClass-EPAUsingaPackaged overtheotherswitch-modePA,Class-D.
GaNHEMT Fig. 4(a) shows the ideal Class-E impedance of this tran-
Recentdevelopmentsinsolid-statetechniqueshaveenabled sistoratthedevicepackageplane de-embeddedfromthe
wide bandgap devices, such as GaN and SiC HEMTs, which Class-E reference plane . and are illustrated in
significantlyimprovetransistorperformancefromconventional Fig. 3. However, the intrinsic parasitics are usually nonlinear
CMOS techniques. To date, GaN HEMTs have been applied (voltage dependent), which might lead to an accuracy degra-
innumeroushigh-efficiencybroadbandPAdesigns[8]–[12].In dationofthistheoretical modelinthe highpowercase.Fortu-
thisresearch,aGaNtransistorisselectedastheactivecompo- nately,averygoodnonlinearmodelofthistransistorisavailable
nenttoimplementthebroadbandClass-EPA.Ingeneral,GaN fromthemanufacturer,whichisalsooptimizedforswitch-mode
transistorsexhibithighbreakdownvoltage,lowandvoltage-in- PA designs [18]. Therefore, the load–pull simulation is con-
dependentoutputcapacitance ,andlowturn-onresistance ductedontheGaNtransistorusingAgilent’sAdvancedDesign
[18]. These characteristics are very important for real- System(ADS)1inordertofindamoreaccurateloadimpedance.
izinghigh-efficiencyswitch-modePAs.However,currentcom- Itisworthynotingthattheoptimalimpedanceobtainedhereis
mercially available GaN HEMTs that are appropriate for dis- referred to the package plane, as the package effect has been
crete designs are typically packaged. Their packages often in- incorporated in the transistor model. In the load–pull simula-
troducenonnegligibleparasitics.Fig.3showsatypicalequiva- tion,asufficientlylargeinputRFpower,i.e., 1W,isapplied
lentcircuitofa25-WCreeGaNHEMT(CGH40025)[18],[19], to drive the transistor as a switch. Fig. 4(b) shows the simu-
whichissealedintheCree440166package.Fig.3alsoliststhe lated efficiency-optimized load impedance of this transistor at
approximate values of those parasitics. The intrinsic parasitic 1–2GHz.Fig.4(a)and(b)indicatesthesimilarloadimpedance
values, i.e., , and , are approximated based on
1AgilentTechnol.Inc.,SantaClara,CA,2009.[Online].Available:http://
[18] and the manufacturer’s datasheet. The extrinsic parasitic www.agilent.com
CHENANDPEROULIS:DESIGNOFHIGHLYEFFICIENTBROADBANDCLASS-EPA 3165
Fig.7. Simulatedload–pullcontoursofthethirdharmonicimpedance:(a)at
GHzand(b)at GHz.
Fig.5. Optimalimpedanceofthesecondandthirdharmonicsextractedfrom
load–pullsimulation. to the third-harmonic impedance with a maximum efficiency
dropofonly10%inthesmallblueregions(inonlineversion).
Tosumup,there aretworequirements ofthematchingnet-
workforbroadbandClass-EPA:
• providetheappropriatefundamental-frequencyimpedance
across the continuum spectrum of frequencies across the
desiredbandwidth;
• providethe appropriate harmonic impedance which stays
inthehigh-efficiencyregion.
In the broadband PA design, compromises are always
unavoidable in matching the necessary impedances across
the desired bandwidth [20]. Also, as the second harmonic
impedanceofthefrequencynear1GHzwouldbeclosetothe
fundamentalmatchingimpedanceof2GHz,anoptimaldesign
within 1–2 GHz is hard to achieve in practice. Therefore, the
presented design strives for achieving optimal results in the
Fig.6. Simulatedload–pullcontoursofthesecondharmonicimpedance:(a)at 1.2–2-GHzregion.
GHzand(b)at GHz.
III. DESIGNANDIMPLEMENTATIONOFINPUTANDOUTPUT
MATCHINGNETWORKSUSINGLOW-PASSTOPOLOGY
region for the Class-E operation within this frequency range,
whiletheload–pullimpedancespreadslessthantheidealone.
A. MatchingNetworkTopologySelectionforBroadband
The broadband Class-E behavior will be further demonstrated
Class-EPA
usingwaveformengineeringintheschematicsimulationofthe
entirePA(SectionIV).Thesource–pullsimulationisalsocon- There are actually numerous circuit topologies to realize
ductedtoobtaintheoptimuminputimpedanceofthetransistor, broadband matching such as a impedance transformer.
whichisshowninFig.4(b). Fig.8showsthemostcommonbroadbandmatchingmethods,
Harmonicimpedancematchingisanotherveryimportantas- includingmultipletransmissionlinesectionssynthesizedusing
pectinthehigh-efficiencyPAdesign.Theoretically,infinitehar- the small reflection method [21], magnetic coupling network
monicimpedancesneedtobeprovidedattheClass-Ereference [21], and multistage ladder networks. However, the first two
plane,whichis,however,notachievableusinganyrealisticfilter networks [see Fig. 8(a) and (b)] give the same impedance
circuit. Moreover, the parasitics need to be considered. Thus, for the high-order harmonics as the fundamental one. Thus,
theload–pullsimulationiscarriedoutagaintoprescribetheop- they are not applicable for highly efficient PAs. From the
timalharmonicimpedancesatthepackageplane.Fig.5depicts ladder-basednetworks[seeFig.8(c)and(b)],onlythelow-pass
theoptimalimpedancesofsecondandthirdharmonicswhen and bandpass behaviors are useful. However, a conventional
variesfrom1to2GHz. low-pass filter or a conventional bandpass filter do not lead
Asexpressedin[13],thesecondharmonicimpedanceplays to the optimal behavior because of the restrictions imposed in
themostimportantroleinimpactingtheefficiencyofaClass-E their out-of-band behaviors, as is explained in Fig. 9, which
PA.Fig.6showsthesimulatedload–pullcontoursofthesecond do not yield the minimum possible in-band ripple. If these
harmonicat1and2GHz,indicatingthetolerableregionofthe restrictionsarelifted[e.g.,steepout-of-bandattenuationbelow
second harmonic impedance in which high efficiency can be thelowestfrequency,showninFig.9(b)],alowerripplecanbe
achieved.Fig.7illustratesthecontoursofthethirdharmonicat1 achievedwithinthedesiredbandwidth[23]foragivennumber
and2GHz,whichunderlinesthattheefficiencyislesssensitive of ladder steps. Consequently, the low-pass matching network
3166 IEEETRANSACTIONSONMICROWAVETHEORYANDTECHNIQUES,VOL.59,NO.12,DECEMBER2011
Fig.10. Three-stagelow-passmatchingnetwork:(a)prototype,(b)scaledto
the50- system,and1.6-GHzcenterfrequency.
Thelow-passmatchingnetworksynthesisusingaChebyshev
prototype has been theoretically presented in [23]. However,
this presentation only covers the real-to-real impedance trans-
former.Inthisdesign,theGaNtransistorrequiresaninductive
output impedance, as depicted in Fig. 4. Therefore, this theo-
reticalmethodisnotsufficient.Inthissection,amorepractical
Fig. 8. Common broadband matching networks. (a) Small reflection theory
[21].(b)Magneticcouplingstructure.(c)Multistagelow-passladdernetwork. method for synthesizing a real-to-complex low-pass matching
(d)Multistagebandpassladdernetwork. networkisintroduced.
B. SynthesisoftheMultistageLow-PassMatchingNetwork
Itiswellknowninfiltertheorythathigherorderdesignslead,
ingeneral,towiderbandwidthandsteeperstopbandattenuation
[22].Thisisalsotruetolow-passmatchingnetworkdesign.In
the -bandClass-EPAdesign,theoptimalloadimpedanceac-
tuallyvarieswithfrequency,asillustratedinFig.4.Therealpart
ofthisfrequency-varyingimpedanceisaround10
, leading to a transformation ratio of 5:1. To achieve
thisratiothroughoutanoctavebandwidthwithanin-bandripple
of 0.1 dB, at least three stages are needed for the low-pass
network, as expressed theoretically in [23]. One- or two-stage
low-passmatchingnetworksyieldanin-bandrippleof0.9–1.4
and0.2–0.3,respectively.Asthetargetimpedance isacom-
plexnumber,therearetwomainstepsinthedesignprocessthat
aredescribedinthefollowing.
Step 1) Thefirststepistodesigna5:1real-to-realChebyshev
low-passmatchingnetworkwithinthedesiredband-
width. A prototype three-stage 5:1 low-pass trans-
former with a bandwidth of 80% is extracted from
[23], as shown in Fig. 10(a). These -elements de-
notealow-passtransformerinanormalizedsystem
with1- impedanceand1-rad/sangularfrequency
[23].Thisprototypeisscaledtothe50- systemat
Fig.9. ComparisonofADSsimulatedfrequencyresponsesbetween:(a)con-
acenterfrequencyof GHzby
ventionallow-passfilter,(b)bandpassfilter,and(c)low-passmatchingnetwork
[23].
(8)
approach instead of conventional filters is selected to realize
(9)
thisbroadbandClass-EPA.
CHENANDPEROULIS:DESIGNOFHIGHLYEFFICIENTBROADBANDCLASS-EPA 3167
TABLEII
PARAMETERSOFTHETHREE-STAGELOW-PASSMATCHINGNETWORKSFOR
REAL-TO-REALANDREAL-TO-COMPLEXTRANSFORMATIONS
Fig.11. Simulatedinputimpedanceofthethree-stagelow-passmatchingnet-
work:(a)a1:5real-to-realChebyshevtransformer,(b)withalarger opti-
mizedforPAoutputmatching.
where and represent the normalized angular
frequency and impedance. Fig. 10(b) shows the
scaled low-pass matching network with the values
ofelementslistedinTableII.Thesimulatedresults
are shown in Fig. 11(a), indicating the frequency
responsewithadouble-knotshapearoundthetarget
Fig. 12. Synthesized three-stage low-pass network for output matching:
impedance.
(a) simulated -parameters when using port 1 and port 2 impedances of
Step 2) The second step is the post-optimization of the and50 and(b)simulatedinputimpedance.
real-to-real transformer in the first step, which
already yields a starting point, in order to form a
real-to-complex transformer. The simplest option and capacitors at the desired frequency range, the low-pass
is to simply increase to 1.8 nH, as shown in matchingnetworkisimplementedwithall-distributedelements
Fig.11(b).However,thisdoesnotyieldanoptimal inthispaper.
result. A computer-aided design (CAD) optimiza- The inductors are replaced by high-impedance (high- )
tion is needed, in which the input port impedance transmission-line sections and the capacitors are replaced by
is set to be (conjugate of the optimal low-impedance (low- ) open-circuit stubs. The implemen-
impedance at the center frequency). Each element tation of the matching network designed in Section III-B is
ofthisnetworkisgraduallyadjusteduntilaCheby- shown in Fig. 13. Due to fabrication tolerances in our setting,
shev response is observed (flat in-band and theallowedsmallestlinewidthis20mil,whichleadstoa96-
equal-ripplein-band ).Theachieved -parame- microstrip line on the 20-mil-thick Rogers Duroid 5880LZ
ters are shown in Fig. 12(a). The input impedance printedcircuitboard(PCB) .2Usingshort-line-ap-
of this optimized matching network is shown in proximation theory expressed in [21], the length of the short
Fig.12(b).TheelementvaluesareshowninTableII high- transmissionlinesarecalculatedby
incomparisonwiththereal-to-realcase.
(10)
C. Distributed-ElementImplementationoftheLow-Pass
OutputMatchingNetwork where isthecharacteristicimpedanceoftheline,and and
The most straightforward way to implement the designed arethepropagationconstantandphasevelocityofthetrans-
three-stagematchingnetworkistobuilditwithlumpedinduc- missionline,respectively.Fortheopen-circuitstubs,whichare
torsandcapacitors,asshowninthecircuitschematic(Fig.12).
2Rogers Corporation, Rogers, CT. [Online]. Available: http://www.roger-
However, giventhe difficulty in getting high-quality inductors scorp.com/
3168 IEEETRANSACTIONSONMICROWAVETHEORYANDTECHNIQUES,VOL.59,NO.12,DECEMBER2011
Fig.13. Implementedlow-passoutputmatchingnetworkusingdistributedel-
ements.
Fig.14. Fundamentalandharmonicimpedancesoftheimplementedoutput
matching network between fundamental bandwidth of 1–2.2 GHz extracted
identicallyplacedonbothsidesofeachnode,thelengthofeach from:(a)schematicsimulationand(b)full-wavesimulation.
stubiscalculatedby
(11)
where isthestubcharacteristicimpedance,whichis36 in
thisdesign.
Equations(10)and(11)provideastartingpointtorunaCAD-
based post-optimization.Theparameters of thismatchingnet-
work is optimized using ADS together with the transistor in
ordertoachieveanoptimalClass-EPAperformance.Thefinal-
izeddesignofthisoutputmatchingnetworkisshowninFig.13.
ThismatchingnetworkisevaluatedinadvanceoftheentirePA.
Fig. 14 shows the simulated input impedances of this imple-
mented matching network from schematic and full-wave sim-
ulations throughout the band of 1–2.2 GHz, including funda-
mental, second, and third harmonics. A good agreement be-
tween schematic and full-wave simulations is observed. The
fundamentalimpedanceregion(redsolidlineinonlineversion)
indicatesagoodmatchingqualityfrom1.2to2GHz,compared
to the load–pull results plotted in Fig. 4. It is also seen from
Fig.15. Designandimplementationoftheinputmatchingnetwork.(a)Eighth-
Fig. 14 that the second-harmonic impedance is very close to orderlow-passprototype.(b)Implementedcircuitusingtransmissionlines.
theoptimumregionshowninFig.5when GHz.The
thirdharmonicimpedancemovesalongtheboundaryofSmith
Chart,whichfallsinthehigh-efficiencyregionformostofthe follows a similar path to the one shown before for the output
frequencies, except around GHz. However, the effi- matching network using distributed elements (Section III-C).
ciency drop due to the third harmonic around GHz The final circuit schematic and its parameters are shown in
is 10%.Consideringthefundamentalandharmonicmatching Fig.15(b).Itisimportanttonotethattheimplementedschmatic
conditions,asillustratedinSectionII,anoptimizedClass-Eop- initially yields a 0.65-mm 96- transmission line ( ) for
erationcanachievedfrom1.2to2GHz. implementing , implying that the stubs surrounding this
line are too close to each other, leading to an nonnegligible
D. BroadbandInputMatchingNetworkDesign
couplingeffect.Asexpressedin(10),thelengthofthislinecan
ThedesiredinputimpedanceoftheGaNtransistorisshown beincreasedbydecreasingtheimpedance.However,anormal
in Fig. 4(b). This impedance has a real part varying from 4.5 higher impedance transmission line leads to an unbalanced
to 2 and an imaginary part varying from 3.9 to 3.1 . As becauseofthedifferentwidthsof and .Thus,
the imaginary part has a variable polarity and a small ampli- a 1.2-mm tapered line is used here to replace the 96- trans-
tude,thisbroadbandmatchingproblemisconsideredasa20:1 mission line. Simulation using the open-ended coupled-line
real-to-realimpedancetransformation(from50to2.5 )within modelisconductedtoinvestigatethecouplingeffectsbetween
1–2.2GHz.Comparedtothecaseofoutputmatching,thistrans- the stubs. Three pairs of the closest stubs of input and output
formationratioismuchgreatersoafourstagelow-passnetwork matchingnetworksarecharacterized,andtheresultsareshown
isdesignedtoimplementtheinputmatching. inFig.16,indicatingaweakcouplingof 30dBwithinthe
The Chebyshev low-pass prototype of a 20:1 impedance desiredbandwidth.Fig.17plotsthematchingimpedanceofthis
transformer with 80% fractional bandwidth [23], shown in inputmatchingnetworkextractedfromschematicandfull-wave
Fig.15(a),isappliedandscaledtothedesiredfrequencyband simulations, showing a good agreement between them. The
and50- systemimpedance.Theimplementationofthiscircuit simulation results further demonstrate that the weak coupling
CHENANDPEROULIS:DESIGNOFHIGHLYEFFICIENTBROADBANDCLASS-EPA 3169
Fig.18. CircuitschematicofthebroadbandClass-EPA.
Fig.16. Investigationofthecouplingeffectbetweenthecloseststubsusingthe
coupled-linemodel.
Fig. 19. Simulated – characteristics of the GaN transistor and dynamic
loadlineatitsintrinsicdrainplacewhenthePAisoperatingat1.2GHz.
anearopencircuit.Two30-pFcapacitorsareconnectedatthe
inputandoutputportsasdcblocks.
Fig.17. Simulatedimpedanceoftheimplementedinputmatchingnetwork.
(a)Schematicsimulation.(b)Full-wavesimulation. The entire PA is simulated in ADS using the harmonic-bal-
ancesimulatorincludingtheeffectsofrealisticbiascircuitsand
dcblocks.Fig.19showsthesimulateddynamicloadlineofthe
effects have no significant impact on the low-pass matching Class-EPAat1.2GHzand15-Woutputpower,indicatingafa-
network. A good coverage of the optimal input impedance is vorable switching behavior. The density of the circle markers
alsoobservedcomparedtothatshowninFig.4(b). implies the amount of time that the transistor is operating in
differentregions.Fig.20showsthesimulatedvoltageandcur-
IV. PADESIGNANDIMPLEMENTATION rentwaveformsattheintrinsicdrainplane,whenthebroadband
Class-EPAisoperatingat1.2,1.6,and2GHzwith15-Woutput
TheClass-EPAisrealizedbyconnectingtheGaNtransistor
power.Comparedtotheidealwaveforms(Fig.1),thesimulated
totheinputandoutputmatchingnetworksdesignedintheabove
resultsshowasimilarshapeandthedifferenceismainlydueto
sections.ThecircuitschematicdiagramofthedesignedClass-E
thenonidealswitchingeffect,whichcausesa 100%drainef-
PA is shown in Fig. 18. In the PA implementation, the trans-
ficiency.
mission line ( ) of the output matching network is modi-
ThePAisfabricatedontheRogers5880LZsubstrate(same
fied tofit the package lead ofthe transistor, whilekeeping the
astheonesusedtoimplementthematchingnetworks).Fig.21
samefrequencyresponseastheoriginalversion.Thegatebias
shows the fabricated PA mounted on a copper substrate with
networkisrealizedusinga22-nHinductor followedby
connectingscrews.Thebottomcopperplateactsasnotonlythe
a 0.5- H conical inductor together with shunt capac-
ground plane, but also as a heat sink. The dimensions of the
itors presenting a “short circuit” for both RF and lower fre-
entirePAtestingboardarealsogiveninFig.21.
quency(megahertzlevel).Thisnetworkissimulatedusingreal-
isticmodelsfrommanufacturers,yieldingalargeimpedanceof
atthegigahertzlevel,whichcanbeconsideredas V. EXPERIMENTALRESULTS
an ideal open. Thedrain bias circuit is realized using a 25-nH
A. Continuous-Wave(CW)Measurements
inductor with a large current handling capability of 3.5 A and
shuntcapacitors.Thisnetworkpresentsasimulatedimpedance ThePAistestedusingthesingle-toneCWsignalfrom0.9to
of at1GHzand at2GHz,whichpresents 2.2GHzwith0.1-GHzstep.Thetransistorgateisbiasedatthe
3170 IEEETRANSACTIONSONMICROWAVETHEORYANDTECHNIQUES,VOL.59,NO.12,DECEMBER2011
Fig.22. Measuredandsimulatedoutputpowerandgainacrosstheentireband-
width.
Fig.23. Measuredandsimulatedharmonicsovertheentirebandwidth.
Fig.20. De-embeddedintrinsicdrainwaveformsofvoltageandcurrentfrom
ADSsimulation:(a)at1.2GHz,(a)at1.6GHz,and(c)at2GHz. acommercialwidebandpre-amplifier(Mini-Circuits,3ZHL-16
W-43 )toprovideasufficientlylargedrivingpowerofaround
31dBmforthebroadbandtesting.ThePAoutputpowerismea-
suredusinganAgilentE4448spectrumanalyzer.Fig.22shows
the measured and simulated fundamental output powers from
0.9 to 2.2 GHz, which are within the range of 10–20 W. The
drainbiasvoltage(bluelineinonlineversionofFig.22)applied
in CW testing is optimized for PAE. The same gate and drain
bias voltages are applied in the simulation and measurements.
ThepowergainofthisbroadbandClass-EPAisalsoplottedin
Fig.22,whichrangesfrom10to13dBwithintheentireband-
width.Thesecondandthirdharmonicsarealsomeasuredusing
thespectrumanalyzer,whichareshowninFig.23.Whenoper-
atingbelow1.2GHz,thesecondharmonicremainssignificant
becauseitfallsin(orcloseto)thefundamentalbandwidth.From
1.3to2GHz,theharmoniclevelsfallfrom 40to 60dBcas
thefavorableharmonicimpedancesareprovidedbytheoutput
matchingnetwork.
Fig.21. FabricatedcircuitofbroadbandClass-EPA. Fig. 24 shows the PA efficiencies within the entire band
obtained from measurements and simulation. The measured
thresholdof 3.3Vinthemeasurement.TheCWsignalisgen-
3Mini-CircuitsCorporation,Brooklyn,NY.[Online].Available:http://www.
erated by an Agilent E4433B signal generator and boosted by minicircuits.com/
CHENANDPEROULIS:DESIGNOFHIGHLYEFFICIENTBROADBANDCLASS-EPA 3171
Fig.24. Measuredandsimulateddrainefficiencywithin0.9–2.2GHz.
Fig.26. Measuredoutputpower,gain,efficiency,andPAEversusinputpower
at1.6GHz.
Fig.25. MeasuredandsimulatedPAEovertheentirebandwidth.
PA efficiency is above 80% between 1.2–2 GHz, which is the
bandwidth for the optimal Class-E operation, as discussed
Fig.27. Measuredefficiencyand outputpower versusdrainbiasvoltageat
in Section III-C and proven in the above Section IV. The 1.6GHz.
maximum measured value is 89% at 1.3 GHz. The measured
efficiency is over 60% throughout the overall bandwidth from
0.9to2.2GHz.Thisisthehighestaverageefficiency 80% and2)burst-modetransmitters[26],[27],whichtypicallyapply
achieved at -band for such a wide bandwidth and for this switch-mode PAs for their efficient performance at the satu-
power level today. In order to explain the gradual drop of rated power level. To prove this, we tested this PA under dif-
efficiency from approximately 80%–89% to 60%–70% at ferent drain bias voltages. Fig. 27 shows the maximum mea-
1.2 and 2 GHz, we need to consider the matching net- suredoutputpowerandefficiencyversusdrainbiasat1.6GHz.
work performance. Specifically, the output matching network It is seen that a high efficiency of 70% can be maintained
provides a suboptimal match at the fundamental frequency withinalargevoltagesupplyingrange,indicatingthatthisPAis
below1.2GHz.Furthermore,asalreadymentioned,thesecond especiallysuitablefortheEER/burst-modetransmitters.Amax-
harmoniclevelisquitehighbelow1.2GHz( 15and 20dBc imumpowerof44.5dBmisfoundatthebiasvoltageof26V.
for 1- and 1.1-GHz CW signals, respectively). The same is Adynamicpowerrangeof16.5dBisachievedwiththesupply
true above 2 GHz, except that the third harmonic becomes voltagevaryingfrom6to26V.ItisalsoseenfromFig.27that
more significant, which is possibly due to the self-resonance the maximum PAE (79%) is yielded not at the maximum bias
of the implementedcomponents. Themeasured and simulated voltage, but at lower points of 22–23 V. This phenomenon is
broadbandPAEsareshowninFig.25. alsoobservedin[12]and[15].
ThePAisalsocharacterizedunderdifferentdrivinglevelsto
B. GSMModulated-SignalMeasurements
evaluateitsdynamicperformance.Fig.26showsthemeasured
gain,outputpower,drainefficiency,andPAEat1.6GHz,while To evaluate the PA performance in an actual GSM commu-
the input power is swept from 15 to 31 dBm. The gain com- nication system, this Class-E PA is tested using a minimum-
presses at dBm and power-added efficiency (PAE) shiftkeying(MSK)-modulatedGSMsignalwithamodulation
reachesitsmaximumvalueattheinputpowerof30dBm.The symbol rate of 500 ks\s, which has a constant envelop. This
PAEdropsfrom79%to23%whentheoutputpowerdecrease GSMsignalisgeneratedbyanAgilentE4438Csignalgenerator
from43.5to30dBm.AnimprovedPAperformanceatback-off andisamplifiedtoasufficientlylargedrivinglevelbythecom-
power levels can be obtained by techniques such as: 1) enve- mercial PA already mentioned in Section V-A. The bias con-
lope elimination and restoration (EER) transmitters [24], [25] ditionusedinthismodulatedmeasurementissameasthatap-