04/13199 12:31 From Motorola Mfax Ph : 602-?44-6591 Fax: 602-244-6693 To MOTOROLA CUSTOMER EXT 53532 02113
• I AN-546 Application Note {
SOLID-STATE LINEAR POWER AMPLIFIER DESIGN
Prepared by Roy C. Hejhall
Linear amplifier design techniques and new R F power transistors developed specifically for HF (2-30 MHz) linear amplifier service are discussed.
MOTOROLA ARCHIVE FILE COPY INFORMATION ONlY PARTS MENTIONED IN THIS DOCUMENT r.1AY NO LONGER BE AVAILABLE
MOTOROLA Sen~iconductor Products Inc. 04/13199 12:31 From Motorola Mtax Ph: 602-244-6591 Fax: 602-244-6693 To MOTOROLA CUSTOMER EXT 53532 03/13
I- SOLID-STATE LINEAR POWER AMPLIFIER DESIGN
INTRODUCTION L Althl>Ugh most RF power amplifiers operate in the signals, which :rrc generated by odd order non-... , , Class (• mode, 3 few applications IC4Uirc linc;~r operation . arc given in Table I. The prinrary areas rc4.uiring linear amplification arc suppressed carrier single sideband (SSO) and low level amplitude modulation (AM). The advantages of SSB over other modulation methods for high frc4ucnc.:y commu nications have been widely discussed in other literature 1,2 Jrd 29 .999 Ml-lz, 30.002 MHz and will not be repeated here. 5th 29.998 MHz. 30.003 MHz A number of linear amplifier design techniques will be 7th 29.997 MHz, 30.004 MHz discussed in this note. Several parameters used to specify linear RF amplifier perfonnance and methods for measur· The above listing contains only the spurious signals ing these parameters will be explained. RF power transis which arc dose lo the desired frequencies f) and f2 and tors specifically designed for HF (2-3 0 MHz) linear therefore difficult to filter. The non-linearities will also amplifier operation will be discussed . generate the sum terms 2f) + f2, etc., but these like the WHY LINEAR AMPLIFIERS? harmonics of f1 and f2. are far removed in frequency. lt is beyond the scope of this report to show The majority of RF power amplifiers in solid stale theoretically how the various spurious signals arc gener transmitters are operated in a class C mode. Class C ated. Rather the intent is simply to review the trouble operation is widely used for CW, FM , AM , and pulsed some spurious signals and the non-linearities causing them. transmitters for numerous reasons such as its high The interested reader is referred to Chapter 12 of efticiency and zero no signal dissipation. Reference I for a theoretical treatment of spurious signal Class C amplifiers arc inherently non-linear. However, generation. in applications where only a single frequency signal is These spurious signals arc not really a problem to a processed, the advantages of Class C operation outweigh selective receiver tuned on channel. The problem lies in the disadvantage of non-linearity. adjacent-channel intcrft:rence which is the major reason When only a single frequency signal is being amplified, for liucarity requirements being placed on SSB transmit any non-linearity will generate only harmonics of that ters. signal, regardless of the exponential order of the non Linearity Measurements linearity. These harmonics ca n be filtered with passive L-\ circuitry. Before proceeding with the subject of linear amplifier In the case of the multiple frc4uency signal, amplifier design, a review of measurement conditions and techniques linearity becomes more important. Th~ reason for this is is in order. that a multiple frequency signal makes it possible for 1l1ese subjects will be covered to clarify the linearity non-lincaritics to generate spurious signals a I frequencies data to be discussed later and to provide background which a1e very close to the fre4uency of the desired information for those who may not be familiar with signal. These spurious signals cannot be conveniently power-amplifier linearity measurements. allenuatcd by filtering. There are several methods that can be used to measun: The tyrkal SSB RF signal com;ists of multirlc frequen amplifier linearity . 1l1ese include multiple tone (fre des with a bandwidth of ahout 3 kllz. In thrs case, the quc m:y) tests and noise te\ts. odd order exponential non-lincaritrcs generate spurious Both methmls t:onsist of driving the amplifier under signals. which will be amplified and transmitted along test with a complex signal. The nature of the test signal is with the desired SSB signal. su-:h that distortion in 1he amplifier produces spurious The rroblcm is often illri~li .. .. Cicsss C ", as used in th 1s repo rt refer~ to operauon with no stgnal condi t ions lc 0 and VeE:: 0 , and a rheoret ic..al conduction angle o f less than 180° regardless ot the actual conduction anQie ------Circuit diagrams eJCternal to Motorola products are incluc:Jed as a mean5 of Illustrating typicaf 5emiconduc:tor applications; co"scqucntly , complete information sufficient for construction purposes Is not neces5arllv given. The Information in this, Application Note has been care· fully checked end is befleved to be entirely reUabte. However, no respon5ibilitv is auumed for inaccuracies. Furthermore, such Information does n ot convey to the ourchaser of the semiconductor devices described any license under the patent rights of Motoro1a Inc. or othttrs. 04/13199 12:31 From Motorola Mfax Ph: 602-244-6591 Fax: 602-244-6693 To MOTOROLA CUSTOMER EXT 53532 04/13 The am plitude of these distortion produ.:ts at the Since a linear-passive t.:omponent does not contribute ;unplificr output provides a convenient measurement of to the distortion , the problem of linearity focuses on the amplifier linearity. Sut.:h a method of specifying am pi slier distortion-producing clement which is the transistor. linc<~nty is very pnsclical. The measurement is fcassble to William Orr, who has do11e extensive work Power Output nod Linearity of the 2N5941 and 2N5942 Transistors 100 ...... 60 - ._ ...... ~ ...... r-...... ~ c: ,...... ,.._ ~ 70 ' ..... · -~~~~ -...... ::: w 40 ...... r-... J ...... ~ ...... ~~ "~">... .. r---... t-... ' ~~ . ~~~ ...... it ...... t'- 50 """ 30 4oo;;;::... . ~ ...... · ~~~ ,:·~~ ...... a: i'.. 40 w P;n • 150 mw' '.~:~ ); ['.. !'..... w~ . "< ,,, r---... ~ 20 ' 30 f----(PEP) !'-. II. r---.t-... 16Q~ fv: .·· ;...... I ...... 1- ...... :::l """ 0.. Vee • 28 Vdc Vee • 28 Vdc 20 """ I · ~ lea • 40 mA ·" " :> lea • 20 rnA P;,•100~ " 0 10 (PEP) r-· I I ~ .l I ...... - 8 .0 0 0." " 10 "~""' ~ 6.0 I l •: ...... 3.0 5 .0 7.0 10 20 30 3.0 5.0 7.0 10 ,.3.0 30 ""' (MH~) .. -:- . f, FREQUENCY (N!Hzl f , FREQUENCY -~. ,'. ··. ·-~=:. ·.·· FIGURE 1 - 2N5942 FIGURE 2- 2N5941 -10 z z 0 Vee ~ 28 Vdc 0 Vee • 28 Vdc i=- -20 - lca = 40rnA j: -20 - lea b 20 rnA +--+--+--t--:~--1----1 <(ID <( ..J~ t = 30, 30.001 MHz ..JID f • 30, 30,001 MHz ::;)- :J~ oz 0 z 0~ -30 oo ·:! 1- ~ ~Jor--t--1--+---+---1-~r--t~,. ~~--r---; . a: a: a: t wo ...... ~ wa:: 1-1- ...... t-0 ZVl ~ zt · ~"': v 3rd Order _tll -.o -40 1-- / ~-3rd0rder ---~ ...... 0 o·o ~ ·· ::t ~ _v~--"'" :!; ~ ... :. 5th Order -5th brder . ~ -- -50 ~ - I J 20-- 40 60 80 100 0 - ·. Pout, RF OUTPUT POWER WIPEP) Pout, AF OUTPUT POWER W (PEP). fiGURE 3- 2N5942 FIGURE 4- 2N5941 MOTOROLA ARCHIVE AlE COPt INFORMATION ONLY PARTS MENTIONED IN THIS DOCUMENT · II (l l'..: l' . . . MAY NO I.CNG~R Bt; . 1Yf11 GI y I 0 watt~ " ~ L'L' J{,• lo..' ll.' ll\.. 1..' - ~ lei [ .1 ~l'll\'t, d , 11 ..... tJ -..'\ I tll l 1\ l Rl · p(l\\ 'l' l tilt' Jl;n :olkl ltl:t d IL'\I'I ;IIIt:l' . 1~1 .'· c· :nt\rd •t h'l , ,~.' '" ~ ~~~ - . !.·, , ~" \ o..·ltl t\.' t t l · ....· 111;'"l I ~ I lt L: i~ l tll i:-' tl1 ' l\\ , 11k cnn!"ig.ur;•tlull I \ 'et l ~ J{ I .. 1.' I · H IH 11l il 11111!1lt~ "l f~ lillt~.. · ,t lHI : , , j ( ,· ~,. t, l l tll': llll~'· l'ilCil' t'~tiL' , -~ I' 04/13199 12:31 From Motorola Mfax Ph: 602-244.S591 Fax: 602-244-6693 To MOTOROLA CUSTOMER EXT 53532 06/13 wlc c'll' V('(' · de colln·icq supph veoll:if:c' I' = Rl ; PllW<'l 111 IIH· c; I~L' Ill' I he llltL';H ;tlliplllw c, lilt' dl'si1cd v:duc of pc1k l'IIVl'litpC pitWL'I i ~ used in <''lli.llicct l .1 In c'Oillplli l' I he lo;HI ~c: 'l'l;ull: c . Tl11t s th,· ,· .. lkL'I f, FREQUENCY (IIIIH.tl h:1sc :llld emitter cotlllc.:t ed Ill lk ground. Ti nt,, tile 11:11m slm 1s cotnplcicly "oil" wllt·n nn driving sign:tl ts FIGURE 6- 2N5942 prcsCIII . The linearity ot' a solid stale jl<~WCI ;unplificr may be gccally itnprovcu by npcr;llioll Willi i'LlrWard bias. "FIH ward bi:1s· · rci'crs to opL'Ialioll Willi a linil e Ill' signal colle ctor current. The extreme (;;tSC or f 0 ~~ - })(' Sl ipply V llll:t~· · - 2X Vik 0" 3.0 4 .0 5 .0 7 .0 10 20 30 u Thh :1111p lit'tcr \\':IS dl''i)!ll c'd Icc , ,.,. wh:11 JWiictllll;i\ cc ·<· f, FREQUENCY (MH>) ,·c11 dd hc uhLIII Il'd :II :1 Stll )! l<' ill'ljllt'l ll'\' with t:ccl lljlkl<'h FIGURE 7- 2N5942 lk.\lhk 1\i dl' 1;111 )-!L' llllpnl:nrc·,· ll l:il The output network is a dnuhlc pi de signed for MO 3 mA. When driven to 80 watts PEP output with a two watts 1'1.::1'. tone sigrwl. the average de base current incrl~ascs to The input network uses a ccntcr-taprwd t r;rnsfonner. approximately 200 rnA. Therefore the b:rse bias supply The basis for the design is the 2NS942 largc-\ignal input must be <.:apablc of furnishing base currents from 3 to impeJann: (Figure 5 and 6 ). The circuit dues provide an 200 rnA with a negligible shift in voltage . It is clc 50 100 20W v.,~;'b~ >-R•2---' IN4719 RFC -2 01 4 Watts 0 .001 1-1F 80 Watts PE' P Input >-+--+------:. E----8-< PEP Output 170-780 pf T AfcO 469 - 110 1ao pr Arco 469 30 MH z L1 . L2 = 21/2 Turns 1!114 AWG . l / 4 "" 1.0 .. J / 8 "" L Tt .- 20 Turr' ~ o f =20 Enameled Wtre Wound on M•crometals T 47 -6 Toroid , Center l3 • 41 12 Turns ;;14 AWG. 1 / 4 "" 1.0 .. 1 /2"" l . Tapped RFC · 2 ~ Ferroxcube · VK200 19/ 46 F erfitc Chol FIGURE 8 - 80 Watt PEP L inear Amplifillf ( , 04/13199 12:31 From Motorola Mfax Ph: 602-244~591 Fax: 602-244~693 To MOTOROLA CUSTO!'JIEA EXT 53532 08/13 .. ' 50 20 w AFC 2 .MOTOROLA '· ARcHIVE FILE COPY INFORMATION ONLY PARTS W.ENTIONt=O IN THIS DOCUM~NT MAY r~o LONGER CE o, 80 Wotts PEP Output AVA!LAfLE IN4719 AF I nput pF Area 469 = T1 : 4 : 1 6 Turns of 2 Twisted Pa irs of .;12G AWG [nameled T3: 1 : 9 10 Turns of 3 Twisted Pairs of #28 AWG Enameled Wire W~r e 16 Crest Per Inch) Wound on Stackpole 57 -9322 #1 1 Toroi d . 18 Cre•t Per Inch) Wound on T wo Stack pale 579074 11'11 T2 · 4 : 1 4 Turns of 4 Twisted Pairs of #26 AWG Enomeled Toroid Wire 16 Crest Per Inch) Wound on Stackpole 57-9322 #11 Toroid. AdJu FIGURE 9 - 80 Watt 2-30 MHz Broad Band Linear Amplifier about 260 rnA, and this is the source of the increased At practical levels of quiescent collector currents, in transis tor de base current at full power. Class B amplifiers, IMD generally degrades at low-power If there were no de resistance between Dl and the base output levels. Thus, a typical plot of the IMD ratio for an of the transistor, the no-signal current through Dl would amplifier may show -30 dB at full power, -35 dB at half be only about 120 rnA. Under these couditions, the power, -31 dB at one-tenth power, rising to perhaps transistor would steal all of the diode current at full -25 dB at less than one-tenth power. power. 01 then would shut off and the additional base About the only solution lo low distortion at low puwet current could only come from the bias source (which fur a power amplitler is tn in c r c a s~ the yuiescent cullcc!Or approximates a constant-current source) by reducing the bias current. Howt:vcr. if lll1e wishes to prevent the !MD current through Rl as the base voltage shifts downward. ratio from evet being. worse than it is at full power, bias The net result of all this would be a de base voltage shift currents approaching Class A operation may be required . of more than 0.5 V from zero to full power, which would A practical solution ca n be achieved by considering the severely degrade the amplifier lint:arity. absolu tc magnitude of the dis tortion produ<.:ts , and There is an additional benefit derived from the base insuring that they remain below thrir full-power output circuit operation. The reduction of D I current with RF levels. drive level results in additional temperature compensation For example, if an amplifier which has an IMD ratio of beyond the simple reduction of diode drop w1th tempera - 30 dB at full power exhibits an IMD ratio of -25 dB ture. The voltage across Dl drops with RF drive level fm when the pllWel output is reduced by 15 dB . the two reasons the heating of Dl and ;1 reduction in Dl Jislwlion products arc still 10 dll below what they were current . at full -puwer output. This i~ desp tl c the fa<.:t that th e IMD Thb rc~ults in excellent tcmp.:ra tu rc Ctllllpcnsattun . r<~ ti o has degraded . and th e amplifier ha~ absolutt:ly no ll'IHlcnctcs toward The criterion for low-power IMD used in the design of tl wrmal runilway at full power \l'lih heat si nk lcmpcra tlt c amplifiers described in this re pllrt is that the full lllll:' 111 e xecs~ uf IIOUC. power 11\10 ratio rating shall he maintained for all power Resistor I{ I lu~ a dual fun..:1 1u 11 1-" 11,1. 11 cau'c' c"Utl <: til uulpul lcveb fro111 full power output down to 10 dB tluw thtough RfT I when nu IU-' J1iV111!! signal i~ present. hdow full power. This is tlte key to the ditlc1cnL·c in vuiiJ~e bctwecn the SO Watt Broadband Linear Amplifier anode uf D I and th~ 2 N5'-J4 2 bJse . Second. R I reduces th e toLd RF 1111pcdancc from ba><' l< l )!rottnd and thereby Figure 9 shows a :! -.~0 Milt hH1adband HO watt PEP impruvcs the stabilit y of th e a1np l11 1cr. linear a1nplilier . The hia~ ..:ir..:url is idcnti<.:alto the HOwatt A~ Slalcd 111 bta, ing 1eqtmc1m'nt ~ . nu1n1 :dning lin e I111 Ca1 amplifier first J <.:scnbc d . The ke y to broadband ;nlt y 111 an amphl't c1 over the enure dvna ntic· 1a11g.c of lite ••pcration of this sct:onJ amplilicr lies wllh transformers amplifier presents a bta,in!! prohkn1. lt . T~ . and T3 . 04/13199 12:31 From Motorola Mfax Ph : 602-244~591 Fax: 602-244~693 To MOTOROLA CUSTOMER EXT 53532 09/13 ,; These t ra nsforrtlCI s a 1e the I ransm ission-lin<' h roadbaud opposite end of tl1e line unconnected . Then determine the typ.: dc~<.:rihcd hy Ruthroff) ;111d Pitzalis6. Tltey <.:omi~t frequencies at which the open-circuit line is exactly a of co111binations of wires which approxim<~tc a trans quarter w~vc·length long. This will be indicated by a very mission-line wouud \Ill a toroid magnetic core. They have low impedance reading with zero phase angle . The theo a much wider frcqucnc.:y response than conventional core retical input impedance of the line is zero, but line losses ( coupled or air-coupled tr band transformers is : t.han the line's Z0 and the input impedance of the line is First, the desired impedance step up/down rati~J is again measured . The Z0 is then computed from the selected . Usually this will be either 4 ; I or 9 ; I . Although a expression 16 : I ratio was attempted better results were obtained (4) with two series connected 4 : I transformers. where RL ~ the terminal ing resistance Second, the twisted-wire "transmission line" is pre- Rin "' the line input 1csistan..:e pared . This is nothing more than the name indi~ates a transmission line <.:onsi~ting nf twisted wires. In order to usc equation 4, RL should be a pure lu prcpa1ing the twisted-wire lines it is cnuvenient to resrstan~:e. Rin will also be a pure resistance when the line usc enameled wire of two different colors. ll1e required length is equal to · ~ number of wire pairs as stated on the schematics such as 4 "three twisted pairs", means 3 wires of each color arc then The se<.:o nd method consists of making input imped twisted to achieve the required number of crests per inch . am;e measurements on a section of line at a frequem;y A drill motor makes a convenient "wire twister". A single where the line length is less than ~ - with first an open "crest" is formed by all the wires of one color. 4 The key parameters for the transmission li e · · then a short circuit termination. Under these characteristic impedance (Zo) and length. These aram..MOTOR()tAditw , the line input impedance will be a pure eters are optimized for transformer bandwidth. ARcHIVE FflE~nc and Z0 may be computed from the expression: The optimum characteristic impedance of the wis~fiORMATION ONLY . . . . f ,.hAATS W.ENTIONr=o (5) Wires of a transmiSSIOn 1IJle trans tlrmer is given Y IN ...... n r.• nl n -, ,,.,un.lrl\1 rr.:trl~ltlrrn\.~ 1~ 04/13!99 12:31 From Motorola Mfax Ph: 602-244-6591 Fax: 602-244-6693 To MOTOROLA CUSTOMER EXT 53532 10/13 VSWI< l• •. ul .tl :ollph:"c :lll)..d es . 'I he 2N'i1l42 lr:tll\l ~.t"' w:" (l h• "' 1(, 1 / o t f-14t11f' q IIIII d,llll;lj.!l'd . I t f' Oilct ls lo;; h1 •• nunll11!• of 1\ieiWIII~ dt'\1)!11 tin il111:tdh ~ 111d lilll':ll :tlll)llilt<.:IS I\ lwi"n d 1';11n, wou•HJ no a IIJrt) ld I 4 l ujhl \."'l ll f ''t l11li ht· 1 ''"''Piicakd hy lille:1111y u>llsidnaliuns. sr)h lt•t .... I l f hj1'1h~r ' lhl' 1.1111cal :m.·a j, ihe :1111plilicr outpulllelwork. SIIIO:<' llll' co111pk\ ,:olkcllll load i111pcda11L"c has a siglllflc;llll l'lk ~ l 11 11 li11callly . "llierc·lorc. llclworl-.: wl11ch p1ovides :1 4 d rtl h. \o\1 111 '\ a sn i der~ I I I HJettu~ r ·-· ~~... -;.---+----• '•"•l,"laciiii"Y l o~ 1d 1"10111 :1 ).::till \I:IIHIJlOIIII 111ay still prt'Sl' lll I rl clllil<'' 111 achieving. "(lllllllllll hnearily pcrrollll:tll<.:<' "I he 11 <' 1\l'lllk\ of ihi\ :llll)lliric•l achil'VCd ;1 J.:c:adi: or 4 1o< jhl "" '"" 1 : """eel IO\Icth or · 1dwid11i al 1l1e npeme 111' Sllllll' linearity pcrforman.:c . AI KO walls PEP output. the IMD ratio at 30 Mll;r. 1s 1k ~ dly -.12 dB. while al >lHilc' lowc1 frequencies I he IMD l FIGURE 10 - Broadband Transfo IS ·ltghll ) Wtl"C al -2'i Ill - JU dB. of Transmisston - ltne Type. hg.urc 12 IS a pl u l uf IMD versus power oulplll :tl 4 . I 'i . :tnd 30 Mill' . IMD pcrfonnancc of - 30 JB c<.tn he two cores are prepared scpurulely and then wired togctlic:1 ad1ievcd ,,vcr the entire· opcr:11ing rrcqucucy range of tho:: in the lll s.::c R..: i"c l cn..:c~ 5 and 6 . 0 1 Tw~ Tone Test · R ~ tu rning specifically to the t: lr C: IIII of Figure 9 . tho:: z performa nce spcciflc:tttons :tnd d..: sign features will be· 0"'- v 1- - 20 d i Sl:llS ~ Ctl. pr FIGURE 12- IMD vs Power Output of Amplifier 3 0 shown in Figure 9 1 1 \<11..: IIi:~! f.:cdhad, Ita~ not been used and thai no Paut = ao w tPEPl- r- :~11 <: 111(11 has lwcn 111:1(1l' lu Ill<~ I-.<' 1h.: t-:ain of this amplilicr :'?"' 2 - ...... Jlal \\ 11h l!Cljlll'IIO.:)' 2U () t\ h11oadhand lr ~ III SmiiiCI ulili/.111); suc:h an arnpltl1cr 5 10 20 :1 0 w,ntiJ tn.:litde a ga in coni rol. I , F REOlJI" NCY IMH.•) ,\n l' :\Oilllpk ol .1 h• oadto aml ;~nrplllk r at a lower puwe r FIGURE 11 - Power Gaon Ve rsos Frequency Performance k vc·l 1111111111g h.: cdh ac· ~ Itt :ll · IIIL'\'~ a t:l> IISianl g.ain v•· 1s11~ of Amplifi~r Shown in Ftgure 9 lll'tjllc'lll'~ ' L'liOir;ICil'rl\11 0.: I ~ ~ 1\ ' <'lli:lll'r ill IIJi ~ iCpoll. 25 \\a If llruadb:ur d Lim:ar 1\ 111pl il"ic•r 11 .. - ''1' 1111111111 cha!at:l c' il\ lh· 111 1p.: d ~1 11-X ''" I 1 .~" ' '' I I ~ lh>W ~ :1 plo t "' t'ol\'<'1 ~ .111 1 1 <'I'"" t"rcqucnl'y lint:,ll :1111plil"ic·1 milt~ 111 ..: ~ 1'\ ) 11 7 ll11 ~ lll\istnr . lo ll ,-;I) \\ ,il l\ 1'1-"1' lllllplll J'"\\"c'f l'lll' 1\ · pic· :~l L" llik.: ltll i\ ~. 1111 !Ill' 1--.,·~ - ( I'IIIJh >il<'ll h Ill !his :nnplilit:l :IIC: il l<' l' ll i• l<'llc·\ .1 1 _\Q ~tilt Wil li:\() ll .tl b J'l)' .,ll lpiil l, -1 .\' . j",,, h, .. ,, .th.u"l 11 :t1 hl • '' llll'f> I I . I 2. :II HI 13. all nf which :111' .I 11 11, 1111p illi ,·l ~ I '·'"" · IJ. ,· ·""l'ltlicl w~, ill "lll.llc"l - "'' ;,·,1 :II:-,() \\ ;llh I'll' ""' .ollljl lil lt: l al "' illll,lt:lll'' lhc' l'!:ll"liL: IIit y or d,.,,,_ \ \ 1!1 .1 :\\' \ 1 h\llt.' IL' :'-1 1 -.1g 11. tl h \ '.!l hll'l 1111 ;.! II Ill .I ll llliiiHi l' 'I 04/13199 12:31 From Motorola Mfax Ph : 602-244-6591 Fax: 602-244-6693 To MOTOROLA CUSTOMER EXT 53532 11/13 MOTOROLA 100 ARCHIVE FILE COPY 1 -; w INFORMAT!Ot~ O~Ly PARTS MENTIONED ~~ ~---- IN THIS 00Ct:MWT 1N4719 ~.r :~.~ MAY NO LONGER OE l d AVAILABLE I I .1 V I I' ' -- -- u 1 IJ[· - ~1-(-- - -+H-1---<< 2 ~ Will .. 1'1. I' R f '1: (J ,,,, _,, ,, lllf) \ 11 T I . T 3 4 . 1 b l 11 n 1~ 1d 2 T wt"'it!CI Pa .rs o l 1 : '}( ~ AWG F. n. H II • ~ I t ! d 70 W trt•I G Cr ~t s l fl l' r lru .h l W11 11 nd o n S I.ICI ------~ FIGURE 13 - 25 Watt 2 -50 MHl Broadband Amplifier .. buildiug blocks'' ft•r g~l1l'l ; d 11 se. T l ;111 d ·1 ~ar e it.krllil:al. Class A Linear Amplifier with Feedback re spectively to Tl and '12 in tilL' lllJllll cirL'llll ot' th~ Fi!(Uie 16 show~ a low kvl'l b1uauband l111 ear amplifier 1 N5lJ4 2 arnpli fier just Liese 11h.:d . A11d uul put Ira nsi'orrne r of tile type whit:h might he used 111 the pre-Jrivcr stage' o l T3 i~ iucntiL:al toT I. ;1 I-IF SSB transmitter. This amplifier utilizes Class A bias The bias network b the· S 30 .. (!) . I i I ··-----+- -- - ~ ------:-- ·-r -~ i ~ ---1 ·· -t -~ -r - -- ~ - T · ----·--. - - I 20 50 f , Frequency tMHz) Pout· Powe< Output, Watu PEP FIGURE 14- Power Gain Versus Frequency Performance FIGURE 15 IMO versus Power Output of Amplifier of Amplifier Shown in Figure 13 Shown in Figure 13. Ill 04/13199 12:31 From Motorola Mfax Ph: 602-244-6591 Fax: 602-244-6693 To MOTOROLA CUSTOMER EXT 53532 12/13 uB1•H 111 c 1 ~ - r-- l_~ I aV e(. IOljJl O.li-JI· J0. 11!1· BH110f\ I-= I"::"" : i~ · tU ,, nr ...._.-----H(-----IQr-+-<( 2oo Mw r·~cr Input I Output MOTOROLA ARCHIVE FILE COPV INFORMATION ONLY --=- PARTS MENTIONED IN THIS DOCUMENT 11 4 : 1 3 Turns of One Twisted Pair of #26 AWG Enameled MAY NO LONGER BE w;re 15 Crest Par Inch) Wound on Stackpole 57 -1514 #24 H Toroid AVAILABLE RFC - 1 Ferooxcube VK 20019/46 Ferr;te Choko FIGURE 16- 200 mW .5-50 MHz Class A Linear Amptifittr The 111m. I implll tant amplifier design ~:unsider:.~litlll> for li11!:arity :.~rc the tic bia~ system and proper ..:ollc..:tor lo:.~d impedan<.:c. 25 II Ill II Ill Acknowlcdgmenrs: 20 Pout N 200 mW (PEP) Tile :.~uthor wishes It> a<.:knll\vlcdgc the l"ullowing "' <.:Ont nbutors to thi-. work . 15 / Brent Trout desi~;:ncd the amplifier shown in Figure I (1_ EJwanl Lt>u pc c-ons I rue ted 1lie am pli lie rs . performed 10 all I he <:lllllpkx llll!iiSllrCiliL'IliS. :JIIU 11TUVIdCU lllally cx..:l!i ietlt suggestions both on th.: Jlllplilicr designs and th.: prl!par:•tion uf tins rcpml. 0 .2 .6 1.0 5 10 so 100 l{cfcrcnces: f, FREQUENCY (MHz) I) l'appt•nlus. Bruc11c . and sc~,.,.;nikc. "S111glt- Silkh:•llll FIGURE 17- Power Gain versus Frequency Performance of Amplifier Shown in Figure16. l'llll<.:iple, ami C11·..:uits". l\h:(;raw-lldL 2) lluney JIHI Wcavcr. "An lntrudul'li I . 04/13,g9 12:31 From Motorola Mfax Ph : 602-244~591 Fax: 602-244~693 To MOTOROLA CUSTOMER EXT 53532 13/13 ( MOTOROLA ARCHIVE FiLE COPY IN::ORM:,TICN ONLY PARTS f.·:E'NT/O~~Eu IN Tn:S D~ ::;•_;r.t.t: NT MAY NO I.ONGEilBE AVAILABLE l MOTOROLA Semiconduc-tor Products Inc. 80 X 20912 • PhOENIX . ARIZO NA i=-S •"."•t • A SUBSIOI ARV Ot '.40TOf::OL A I NC ,.- I ) Two of the surplus transmit h. wP.re connected in parallel ntenna half rai sed. Also the ith respect to the mast can be ·est SWR. The length of the nds upon where you are going 1, of course. Mine. has given 1e lower end 25 feet from ! TV masting for the support nsy and will bend and buckle tical. The lower ends of the rom 10 to 4 feet from the !p about 45 degrees of droop. tiona/ Notes bracket to hold the mast to ·ave raised this assembly by one radial, first anchoring the BY J. H. iOHNSON,* Ex K4WYQ, Part I event side swing. I have used AND R. ARTIGO,* W6GFS kW) and RG-58/U for 300 ODAY IMPROVED rf power transistors are best antenna but for a cheap, withstand infinite VSW~ . ' Fig. 1 also gives soma T available at moderate prices. The technically portable or permament an· prices of popular vhf 'tubes in the 100-wati~~ lid inclined amateur should begin to experiment with 1 a 60-foot lot, it is hard tu under class. As you can see, the cost of going solid solid-state power amplifiers, if he hasn't already. state has never been lower! ·le to work four bands. You Designing a solid-state PA stage is a great deal ed with the effective radiated easier than designing a tube stage. Fewer parts are I on 40 and 15, good on 20, required, power supplies are not needed for mobile Design Philosophy For A'mateur A-mplifiers 1got it to work on 80, using a applications, there is less sheet-metal work to do, come out very well in DX There are many ways ··to design ' ~f power and wide-band performance can be achieved using ' the big boys. When I travel, amplifiers - it sometimes .'seems that each engineer transistors. Along with ease of construction, tran has his own philosophy. Designing amplifiers for lith ..... ~ JQST I sistors offer some advantages that tubes cannot the amateur bands is different in many' ways from match: those intended for industrial or military applica 1) If a transistor is selected with care and is tions. A few good reasons for.~sirg transistors have used properly, it will probably last a minimum of been presented above; but, af:wfiilt. poWer levels are 100,000 hours. transistors practical? If ;fou are inte!ested)n 2) Wide-band amplifiers can be constructed building a 1-kW final amplifier, a tube) tage is sdll ·our radials (or more) cut for easily using transistors. Octave and even decade the "way to go," unless utmost riliability is : frequency . Such a system bandwidths are possible. · required. Fig. 2 shows ,the Class · C '·power levels rformance. 3) The small size of a solid-state PA stage is easily achieved using two transistors in paraliC! in a 1t matter is that of the earth attractive for portable and mobile designs. PA stage. s are mounted at ground level The basic idea behind this article is to convince When you begin making design ·decisions, re be very important. Too many the reader that solid-state amplifier design is easy, view Fig. 1, plus any new transistors\ and 'use the verticals, get a five-foot long to demonstra.te how easy it can be, and to illustrate solid-state approach if the power level you seek can and drive it into the earth at basic principles ·with two practical designs for the be achieved using one or two transistors in the na. They think this provides a 2-meter band. The emphasis will be on vhf· and output stage. Leave the circuits .forthre¢/four or :ion. As a matter of fact, the uhf-circuit design at relatively high power levels for more transistors to the experienced engineer. It is are practically worthless for transistors (40 to 160 watts). Solid-state amplifier d ground rod is the type used design for lower frequencies or lower power levels 1y for home installations. This does not require many of the precautions prescrib 1vily galvanized, 5/8 inch in ed in this article. (See references 13 and 17 for a Transistors are now avail;ble . .for _hf, 10 feet long. The amateur discussion of hf transistorized amplifiers.) vhf and uhf amplifiers whiCh ri,i/izl buy these rods from any An important aspect of any amplifier design is tubes in cost an{[ performance. 'J'JJis ;upply house. If possible, tic cost. For amplifiers in the 100-watt-output-and :ction to the water-system under class, you no longer pay a premium to use article describes the design and .' ~on· 1! piping is used. transistors. Of course, you must consider the total struction techniques-'used at, SO MHz ;tatement from fellow hams cost, including that of the power supply, when and above, with practical examples · r antennas and radiate poorly making a valid comparison. Fig. 1 shows some including a 16()-watt-output .amplifier ; isn't true because a vertical recent prices for vhf and uhf transistors which are for 146 MH ST for September 1972 31 TRANSISTOR JUNCTION The 12- and 40-v have titvo capaci base circuit, one TRANSISTOR Underwood I ow-l JUNCTION TO 3!8" F R1 (¢jc) CASE THERMAL ·•--++-+---i-t-+-+---J RESISTANCE Several other SPECIFIED BY using transistors 1 ~ TRANSISTOR HANIJI'ACTIJRER 1) Transistors extent that a tu ..._"J Vi'S CASE TO HEAT \.!) ······•-+-+----l-+-+----l R.2 fJ>cs) SINK THERMAL constant of a RESISTANCE milli~econd and " be several minute ~ %'s 2) Because of ~ ·····•-+--+-+-+----1 LATERAL-HEAT· TRANS· a transistor shoul R3 FER-TO-FINNED AREA· OF·HEAT SINK THERMAL level well bela\\< RESISTANCE. usuALLY rating, not iwo t< SPECIFIED AS ONE TERM as with a tube 0 .2 .4 .6 .8 1.0 8Y HEAT SINK MANUFACTURER dissipation capat ¢CS-THERMAL\4"5········· RESISTANCE-CASE TO HEAT SINK loads. A good ru: R4 HEAT 51 N K- Fl NS is to keep the rf Fig. 6 - Thermal resistance of the case of a TO-AIR THERMAL RESISTANCE power dissipation transistor of the associated heat sink varies accord 3) Transistors ing to the stud (S) or flange (F) size. than manufactu1 voitage spikes. HEAT SINK 4) WHAT ABOUT SILICONE GREASE - A 12.6-volt mobile high quality silicone grease like GE Insulgrease2 or Fig. 7 - Resistances to heat flow encountered stand the 15- equivalent should always be used on both stud and form a transistor junction to a heat sink. ail to motive syste flange devices. The use of silicone grease will select has similar · improve the interface thermal resistance by at least 4) Ensure YOl 0.20 C/watt. apply full power. thermal resistance from case to heat sink is present, they ean 5) RELATIONSHIP OF LEADS TO CIRCUIT negligible. This is not so ... typical values for the - One of the most important aspects of transistor various packages bel ted to a heat sink using the mounting is assuring that the transistor rests on the manufacturer's specified torque are shown in Fig. circuit board without stressing the leads when it is 6. All of the individual thermal-resistance terms Transistor am1 bolted to the heat sink. See Fig. 8. must be added together as though they are suitable impedan Other important aspects of power-amplifier series-connected resistors. coupling compon• design are thermal considerations. For optimum The impedance reliability, the transistor chip must be kept as cool (R1 + R2 + R3 + R4) °C/watt x co~tstructed using act as a low-pa: as possible. There are several thermal resistances of Total Power Dissipation (Watt) = importance, as shown in Fig. 7. The thermal jtarmonic energy resistance value specified by the transistor manu Junction Temperature (° Celsius) matchirig . networ bandwidth of the facturer is 0JC only and does not include 0CS. For maximum reliability, the operating junction Because of this, many circuit designers assume the enough to permit temperature should be less than 1SO degrees or uhf amateur 2 G-641 Insulgrease, General Electric, Silicone Celsius (C).3 referred to is tl Product Dept., Waterford, NY 12188. 3 For additional details, see White, "Thermal network, not the Design of Transistor Circuits," QST, April, 1972. component. . Keel course, to minirr bandwidth, low-Q (because circulatil PC !equire critical cc BOARD the L, C, and Q m Smith chart once have been determi for instructions a optimum networl (octave or greater (c) IMPROPER design details. The input and transistor are usu: Fig. 8 - Proper (A) and improper (B and C) ways data sheet. . These to mount a power transistor. At B the transistor leads are forced up to meet the pc board, which series or parallel may fracture the leads at the edge of the case. At C transformed into 1 (MPROPER the lead inductance will be high, reducing stage 4 Grammer, "~ gain. Matchin& )l!etwor April and May, 19 QST for 32 September The 12- and 40-watt stages in this 3-stage amplifier have two capacitors connected in parallel in the base circuit, one returning to each emitter lead. OR Undervilood low-loss mica capacitors are used. TO , -< ERMAL Several other precautions should be taken when RESISTANCE SPECIFIEO BY using transistors to \(Void damage: VSISTOIIHANIJFACTIJRER 1) Transistors will not tolei:ate overloads to the CASE TO HEAT extent that a tube will, because the thermal time SiNK THERMAL constant of a transistor is in the order of 1 RESISTANCE milli&econd and the time constant for a tube may be several minutes. 2) Because of the short thermal time constant, rERAL -HEAT· TRANS· a transistor should be operated at an output power R-TO-FINNED AREA· 1 1 ...... level well below its maXimum power-dissipation . ... ~ j ~I' 1' 1·1' ' ( 1 'j ,... il: I,, \I ' - ·HEAT SINK THERMAL r' ,.. 'I ' ' rating, not two to three times the dissipation rating d £ 5 ;ISTANCE. IJSIJALLY ' 'FIEO AS ONE TERM as with a tube. · A transistor must have extra TSIN!OfANIIFACWRER dissipation capability to handle momentary over loads. A good rule of thumb for Class C operation sheets, make sure you know whether series or EAT SINK- FINS is-to keep the rf output below 50 percent of the )· AIR THERMAL parallel impedance equivalents are specified. power dissipation rating. :ESISTANCE 3) Transistors will not tolerate voltages higher When designing a matching network, always than manufacturer's rating. If possible, prevent work from the transistor to the termination. If the voitage. spikes. Most transistors intended for first matching component is a shunt element, the 12.6-volt mobile applications are designed to with parallel equivalent impedance should be used. Use to heat flow encountered stand the 15- to 16-volt transients found in the series equivalent when the first matching m to a heat sink. automotive systems. Make sure the transistor you component is a series element. A typical input select has similar specifications. matching network is shown in Fig. 11A along with 4) Ensure your .amplifier is stable before you some typical impedance values. The following is a apply full power. If low-frequency oscillations are step- by-step method for calculating the compon Jm case to heat sink is present, they can destroy your transistor. ent values required for the input and output o .. . typical values for the matching networks: d to a heat sink using the Circuit Design 1) The . transistor input impedance is usually d torque are shown in Fig. inductive because of lead inductance inside the a! thermal-resistance terms Transistor amplifier design aiso involves chasing package. In order to maintain the lowest loaded Q thp· · though they are suitable impedance-matching networks and de for the first matching section, the first component coupling components for the de feeder ; see Fig.?· used should be a shunt capacitor equal in imped- The impedance-matching networks are usually .-J' /watt x copstructed using L sections. These L sectiorts also act as a low-pass filter reducing the level of OUTPUT •issipation (Watt) = harmonic energy (Fig. 10). If the Q of each MATCHING NETWORK Jerature (° Celsius) inatchirig . network is kept low (2 to 3), the bandwidth of the resulting amplifier will be wide ity, the operating junction enough to permit operation across any of the vhf . Je less than 15 0 degrees or uhf amateur bands without retuning. The Q referred to is the loaded Q of the matching .etails, see White, "Thermal network, not the unloaded Q of any individual :ircuits," QST, April, 1972. component. Keep the component Q high, of COLLECTOR course, to minimize losses. In addition to wide FEED bandwidth; low-Q matching networks have low loss (because circulating currents are low) and do not require critical component values. The values for the L, C, and Q may be quickly determined using a +V Smith chart once the impedances to be matched have been determined. See reference 15 , 20 and 22 Fig. 9 - Networks that must be designed for a for instructions about using a Smith chart. If an power-transistor stage. optimum network for a very wide bandwidth (octave or greater) is desired, see reference 18 for HIGH IMPEDANCE IMPROPER design details. The input and load impedances for a power transistor are usually given in the manufacturer's 1d improper (B and C) ways data sheet.. These impedances may be either the 1nsistor. At B the transistor series or parallel equivalent. Either is readily > pc ~LOW IMPEDANCE meet the board, which at the edge of the case. At C transformed into the other.4 When comparing data viii be high, reducing stage 4 Grammer, "Simplified Design of Impedance 1 Matching ~etworks, in three parts, QST, March, April and May, 1957. Fig. 1 0 -The L Network. QST for September 1972 33 Both the intermediate imPedance point and the Fig. 12 - Smith-cl number bf L sections chosen are not critical, unless Fig. 11 networks. ma:Ximum bandwidth is.required. 3) After selecting the intermediate impedance point, L1 and C2 can be calculated using a Smith chart. The best choice of strip-line impedance for calculation is a value equal to Zm (15 ohms). Use a Smith chart normalized to 15 ohms to make the calculation. Start at Zl (5 ohms) on the chart nnd (A) progress clockwise on a circular path, with the chart center the origin, until you reach an admitt ance circle Which also passes through the desired output impedance (Zm, 15 ohms). See Fig. 12A. Note that the value of Ll, Cl, and Q can be read directly. 4) If bne· wishes the length of Ll to be shorter, then a higher vahl! of strip-line impedance can be chosen with a slight sacrifice in Q. Always normal ize the Smith chart to the strip-line impedance value. 5) Additional L sections can be calculated In (B) the same manner. An output matching network might look like the example shown in Fig. liB. 6) The transistor manufacturer usually specifies the load impedance required to obtain rated specifications (4 + ji). The impedance to start from on the Smith chart is the complex conjugate of the load (4 - j2). When working on the Smith chart, always begin with the ioad that the network sees on transistor end. The final value you obtain at the other end of the network is the impedance you require "looking into" the network. Other wise, the calculations are the same as the input (c) network. See Fig. 12B. Another proble Another interesting matching problem is the involves decou plinr Fig. 11 - The (A) input, (B) output and (C) design of the interstage network between two coupling networks interstage networks described in the text. transistors. A typical network is shown in Fig. 11 C. order to prevent I This network is plotted on a Smith chart in Fig. tion. Low-frequenc: 12C. The optimum line impedance for Ll and L2 erated by commo1 ance to Lp making the impedance real and equal to is quite low. At some sacrifice in Q, a higher because of the ex Rp. If the impedance of C1 is less than about 8 impedance is used to obtain a practical strip-line low-frequency gain ohms, it is often best to use two capacitors in width. two techniques to 1 parallel, one connected back to each transistor tions: emitter lead to minimize inductance and to equal There are several other ways to provide Imped ize ground currents. ance matching using strip lines: 1} Use source a1 1) QUARTER-WAVE MATCHING TRANS not sustain oscillati 2) If Rp is high (15 ohms, for example) then ance to prevent low the matching network may require only one L FORMERS - Two real impedances can be match is a low value of pu: section. If Rp is low (2 to 5 ohms), two L sections ed using quarter-wave strip-line with Zo calculated in Fig. 13A provide ~ probably will be required. The larger the imped as follows : of Fig. 13A should ance change, the higher the Q. Keeping the Q low for the carrier freqt improves the bandwidth and lowers power loss. If • large a value as pos1 two L sections are required, select an intermediate to handle the requ i.J impedance point approximately 2) EIGHTH-WAVE STUBS - An eikbth-wave value at 146 MH; stub may be used as a shunt capacitor or inductor. capacitor at the car If the end of the stub is open, the stub looks like a as small as possible capacitor with a reactance equal to the impedance power. C2 and C3 Using the values in Fig. llA: of the strip-line used as the stub. An inductive bypass at all freque reactance equal to that of the line is obtained when nel to de. (Capacit< the end is shorted. Using stubs in matching good choices.) At 1· Zm = 'J"5XSS networks provides excellent harmonic suppression. transistor will "see Zm= iiTO 3) QUARTER-WAVE STUBS- A quarter-wave "see" R2. Rl and such as 10 to 15 oh1 Zm = 15.8 stub which is shorted on one end looks like an infmite impedance on the other. Thus, the quarter 2) A second me > Zm = 15 ohms, rounded wave stub makes an excellent rf choke at vhf and oscillation involves ' to a convenient number above. · feedback to lower 34 QST for September nee point and the Fig. 12 - Smith-chart presentations of ·•cal, unless Fig. 11 networks. (' .1pedance l<> , ~ 'g a Smith -line- impedance for m (15 ohms). Use a ohms to make the s) on the chart lind .lar path, with the )U reach an admitt :hrough the desired ms). See Fig. 12A. and Q can be read of Ll to be shorter, e impedance can be 1 Q. Always normal trip-line impedance an be calculated in matching network :>wn in Fig. !lB. Jrer usually specifies d to obtain rated impedance to start ~ complex conjuga ~e orking on the Smith )ad that the network Jal value you obtain 1rk is the impedance :he network. Other : same as the input Another problem facing the circuit designer J; 1em is the involves decoupling the de supply line. The de 1 r""\ reen two coupling networks inust be designed carefully in Fig. llC. order to prevent low-frequency spurious oscilla ~ " art in Fig. tion. Low-frequency spurious energy is often gen :lance !or L1 and L2 erated by common-emitter rf power amplifiers ,flee in Q, a higher because of the extremely high (30 to 40 dB) a practical strip-line " low-frequency gain of vhf transistors. There are two techniques to prevent these unwanted oscilla .ys to provide lmped- tions: 1) Use source and load impedances which will !lATCHING TRANS not sustain oscillation. The best choice of imped iances can be ni atch ance to prevent low-frequency spurious generation Je with Zo calculated is a low value of pure resistance. The circuit shown in Fig. 13A provides such a termination. Ll and L3 of Fig. 13A should be low-value rf chokes chosen for the carrier frequency. L2 and L4 should be as large a value as possible consistant with the ability to handle the required current. (Ten J.LH is a good s - An eighth-wave value at 146 Mllz.) C1 is a low-value bypass capacitor or inductor. capacitor at the carrier frequency. Choose a value t, the stub looks like a as small as possible without reducing the output [Ual to the impedance power. C2 and C3 must provide a low-impedance e stub. An inductive bypass at all frequencies from the operating chan line is obtained when nel to de. (Capacitors of 0. 22 IJF and 10 IJF are ; stubs in matching good choices.) At low frequencies the base of the harmonic suppression. transistor will "see" R1 and the collector will UBS - A quarter-wave "see" R2. R1 and R2 should be some low value such as 10 to 15 ohms. Before mounting the power transistors on this ci r me end looks like an cuit board, straps were installed on either side of her. Thus, the quarter- 2) A second method of preventing If spurious each transistor mounting hole and on the outside 1t rf choke at vhf and oscillation involves using negative collector-to-base edges of the board to connect the ground foils to· feedback to lower the gain of the stage below a gether. See Fig. 4. QST for September 1972 36 Top side of 1 3) Use quality (high-Q) components at low- the prototypE impedance points. .. indicated in I 4) Use an etched .circuit board, with ship-line this model. Tl construction of inductors whenever possible. layout which 1 5) Keep the Q of matching networks low for non foil side minimum losses and maximum amplifier band otherwise ide width. one end of th for 02 and Q. 6) Choose a rugged transistor that is rated for assembled on the type of operation intended. Smaller heat ' the builder wi If you keep these key points in mind, following the assembly . the instructions outlined in this article, an rf power amplifier can be constructed and adjusted without difficulty. (The other parts of this article will appear in subsequent issues of QST). +V L1 BIBLIOGRAPHY 1) Black, "Electro Migration Failure Modes in Aluminum Metalization for Semiconductor De· vices," IEEE, September, 1969. 2) Reich and Hakim, "Hot Spot Thermal Resis· tance in Transistors," IEEE Transactions on Elec tron Devices, February, 1966. 3) Steffe and Moutoux, "Avoiding Second Breakdowns in Power Transistors," The Electronic Engineer, De~emb .er, 1967. BY l (B) tV 4) Gri, "Micowave Transistor - From Small Signal to Hi&h Power," Microwave Journal, Febru· Fig. 13 - Methods of eliminating low-frequency ary,1971. N AN EFF spurious oscillations. (See text.) 5) Gundcach, ·"Rx for RF Power Transistors," I solid-state Electronics, May 26, 1969. . QST, and in t 6) Lee, "Microwave Power Transistors," Micro wave Journal, February, 1969. necessary to ' selected frequency. The connection of the feed 7) Magill, ·"Watt ·- Megahertz Ratings Run Se and to reduce cond to High Reliability in Forei~ RF Power clipper. The back network is shown in Fig. 13B. Make the value Transistors," Electronics, April 27, 1970. of Ll large enough so that installation of the 8) Tatum, "Fingers in the Die," Electronics, schematically feedback network has no effect at the operating February 19, 1969. A slightly 9) Carley, McGou&h and O'Brien, "The Overlay frequency. The lead inductances of R1 and C1 are Transistor," Electronics, August 23, 196.5. taken, wherei often sufficient without additional inductance. C.l 10) Carley, "A Worthy Challenger for RF deliver a fair should be a value large enough for good coupling at Power Honors," Electronics, February 19, 1968. output from ( the lowest frequency of interest. A value of 10 to 11) Johnson and Mallinger, "You Can Depend stage ahead on Today's RF Transistors," Electronics, Septemb· 100 ohms is usually selected for Rl. Both the base er 13, 1971_. necessary to p and collector rf chokes must be low values of 12) Nelson, "Higher Power From Transistors on of all network inductance to obtain maximum benefit from the Six Meters," Ham Radio, October, 1968. selectivity to 13) Hejhall, "Solid State Power Amplifiers for the feedback network. Ssb Service," QST, March, 1972. put frequ encit Using one or the other of these tech 14) Franson, "How to Use RF Power Tran· ed by a spectr niques, most transistors can be stabilized. sistors," Ham Radio, January, 1970. MK-11 version However, a transistor can have several features 15) Linville and Gibbons, Transistors and Active Circuit, Chapter 14, McGraw-Hill Book A logical a built in which make the device easier to Company, 1961. ed by the sp• stabilize. The low frequency gain of the tran 16) Perna, "Don't Let the Wrong Input Q employment o sistor should be as low as possible. Any re Wreck Your Rf Power Amplifier," EDNjEEE, Nov ember, 1971. the use of di sistance or inductance in the transistor emit 17) Schultz, "Solid-State 100-W Amplifier nique was usee ter lead provides negative feedback which de Operates from a 12-Volt Source," EDNjEEE, creases the gain of the device, making . the March, 1972. 18) PitzaliS and Gibson, "Tables of Impedance transistor inherently more stable. A transistor Matching Networks which Approximate Prescribed with large-value emitter resistors is easier to Attenuation Versus Frequency Slopes," IEEE Transactions on Microwave Theory and Tech Generally, stabilize. A transistor with low emitter-lead classic Sonob1 inductance (such as the improved strip-line niques, April, 1971. 19) Hejhall, "Solid State Linear Power Amplifi· Circuits, DC 1 package) is slightly more difficult to stabilize. er Design," Motorola Seriliconductor Application changes made Note, 1971. 20) Smith, Electronic Applications of the subtle. The bc Reminders Smith Chart, McGraw-Hill Book Company, 1969. teur-band perf There are several key points which the designer 21) Wheeler, "Transmission-Line Properties of monly availabl should remember: Parallel Strips Separated by a Dielectric Sheet," IEEE Transactions on Microwave Theory and * Technica 1) Keep lead inductances as low as possible. Techniques, March, 1966. 1 Recomm 2) Make sure all ground paths are short and of · 22) Solid-State Power Circuits, RCA Designer's from local ra• low inductance. Handbook,1972. · IQST I Components, l 36 QST for Septemb POWER OUTPUT VERSUS POWER INPUT The best way to demonstrate the so fundamentals of solid-state power am y plifier design, as discussed in Part I, QST for September 1972, is with a 40 X / Vcc .. 12 .5V practical example. A single-stage 1---1: 175 MHz ,...--- / 40-watt 2-meter amplifier is the 30 choice for Part II. This amplifier uses 1/ many of the techniques presented in [/ 20 Part I and illustrates some interesting v variations. 10 12 15 PIN -POWER INPUT -WATTS (A) INPUT REACTANCE & RESISTANCE Fundamentals of VERSUS FREQUENCY Solid-State Fig. 15 - Smit Vcc •12 .5V - POU~ · 40W_ Power-Amplifier .'\ .-h XL Design show a numb( I .. B40-12. The resistance ra tir I the thermal re ,...---f-~ I Part II this specificat 2.0 •. output and d· ·2 - 1. 5 that will be ne• I Rl -+ -3 I I .0 Other B40- BY J. H. JOHNSON,* Ex K4WYQ, are the input -4 used as the l AND R. ARTIGO,* W6GFS obtained from 140 150 160 170 should be notl (B) 1-FREOUENCY -MHz is for maximu HE FIRST STEP in designing an rf power would have to lOAD IMPEDANCE VERSUS FREQUENCY T amplifier is a careful analysis of design objec From the po\1 tives. These objectives will ultimately define the 19, we see tha be approximat 5 characteristics of the transistor that is selected. The Vtt "' 12 .5V watts of drive. POUT •40W authors' list of objectives for a 40-watt 2-meter 4 r-- Rl power amplifier is as follows: 3 1) The amplifier must have a minimum output De. of 40 watts when driven by 10 watts and operated 2 As outlined from a 12.6-volt de supply. 1 the input-mate 2) The amplifier must be capable of operating power level wt into an open or short circuit without damage to 0 the one showr the transistor. using five trim I 3) The design must be simple to construct and work well up ~ easy to put into operation. 2 However, as th• 4) All objectives listed above must be met 3 go down, and , using an economical transistor. lossy becau se o 4 to demonstrat• 5 Selecting The Proper Transistor 140 150 160 170 power, the desj There are a number of rf power transistors 1-FREOUENCY -MHz ( C ) available that can be used. For this design the authors selected a B40-12 manufactured by Com Fig. 14- (A) Power, (B) input-impedance and (C) munic_ations Transistor Corp. The charts of Fig. 14 Bottom view o 2-inch cabinet output-impedance characteristics of the CTC * Communications Transistor Corp., 301 In B40-12. dustrial Way, San Carlos, CA 94070. follows Fig. 18 have been adde• 16 QST for Novembe · the ram art I, 1ith a -stage the y I ~ lCI : 1.8•0.12" j_ I _loopF: 68pFl'2 of Fig. 15- Smith-chart design of the input circuit. Fig. 16- Output-circuit design shown on a Smith ter chart. show a number of important characteristics of the be used. This technique lends itself to high or low B40-12. The transistor has an excellent thermal power, as well as broadband designs, by using resistance rating. Special attention must be given to lumped-element and distributed-element low-Q the thermal resistance of a power amplifier because matching networks. this specification indicates the limit for power The distributed inductance element takes the output and determines the size of the heat sink form of a microstrip line which can be easily that will be needed. calculated, and the calculated value is dependable. Other B40-12 parameters of immediate interest The capacitance value can also be easily calculated; are the input and load impedances which will be however, it may not be as dependable. The VYQ, used as the basis for network design ; they are capacitance value is not accurate at high frequen :s obtained from the impedance curves of Fig. 14. It cies because of lead inductance. For low-imped should be noted that the load impedance specified ance points in the circuit, only quality capacitors is for maximum gain. To optimize efficiency, we with minimum lead inductance can be used, and would have to design for a higher load impedance. they should be strip-line compatible. Input and an ""wer output matching networks for the 40-watt ampli ·c- From the power-gain-versus-frequency curve, Fig. 19, we see that at 147 MHz the output power will fier are designed using a Smith chart. Two charts re be approximately 45 watts (6.8 dB gain) with 10 are required, one impedance chart for series ele- ne watts of drive. -wat. _ ,oeter ll [I 1imum output Designing The Amplifier ; and operated As outlined in Part I, the first task is to design the input-matching circuits. For a 40-watt-output e of operating power level we could employ a circuit similar to >ut damage to the one shown in the manufacturer's data sheet, using five trimmer capacitors. Such a circuit will . construct and work well up to approximately 50 watts output. However, as the power output goes up, impedances must be met go down, and circuit components can become very lo lossy because of high circulation currents. In order ll to demonstrate techniques required for higher :s nsistor power, the design approach described in Part I will •r li wer transistors IT this design the ·c :tured by Com Bottom view of the amplifier housed in a 4 X 4 X j harts of Fig. 14 2-inch cabinet (LMB 143). Layout of components 3! Corp., 301 In- follows Fig. 18. A pilot lamp and metering jacks have been added to this version . QST for November 1972 ~flon Pc Board Fig. 17 - Diagram of the RFC3 amplifier. Except as other R1 wise noted, capacitors are disk ceramic and the resis ,. Glass Teflon Glass tor is 1/2-watt composi 200 tion. Components not . thick .030 in. thick f---t.::9PUT listed below are marked for "_t"" 337 0.61 circuit-board location pur g5o 0.385 poses. 68l 609 0.277 EXCEPT AS INDICATED, DECIMAL VALUES OF 466 0.211 CAPACITANCE ARE IN MICROFARADS t JIF I ; 371 0.168 OTHERS ARE IN PICOFARADS ( pF OR J.IJIFI: RESISTANCES ARE IN OHM$; 253 0.11 5 k• I 000 . 172 .078 C1 - 2 Underwood 200-pF mica capacitors con 01 - CTC power transistor fitted with Thermalloy nected as shown in Fig. 5 (Underwood Electric, 6151 B heatsink. 148 S. 8th Ave, Maywood, IL 60153; order RFC1 - Etched on pc board; see Fig. 5. .e admittance chart for shunt ele type J1 01 and specify values desi red for C1 RFC2 - 0.3 3-.UH molded rf choke (Nytronics harts are available from most college through C5). SWD0.33). ok stores. C2-C5, incl.- Underwood J101 mica . RFC3 - 6 turns No.18enam.wire, 1/4-inch 10,1 L1-L3, incl. - See Figs. 15, 16, and 18. inch long. Input Circuit ibed the importance of selecting the Output Circuit Q. For this design example, we will strip lines with a characteristic impedance of 50 Other important considerations are ohms. From Table I we find that a 50-ohm strip From the load-impedance chart in Fig. 14 we see :t I and should be reviewed to aid in line on glass-epoxy board is approximately 0.125 that the optimum load for best power gain is 3.2- :his design. inch wide. We next refer to Table II for the j 1.2. From this starting point, the complete dimensions of one wavelength corrected for dielec output circuit design is given in Fig. 16. All Jr design by referring to Fig. 15 to tric constant (e) of 4.8. The length of L1 is: component values were derived using the same tt · 'Ce. The input impedance procedure as given above for the input circuit. C4 " 2 + j1.8 ohms. Since the 37.5 X 0.042 = 1.58 inches (Fig. 17) is used for de blocking and is designed into ;ified in the series form, The value of all shunt capacitors is rounded off to the output network. Notice that the value of C4 is tl, on the impedance chart the nearest standard value available. Lengths of Ll o0tained from the impedance chart since it is a t t adamonal calculations. We should and L2 are not particularly critical. series element. 1 constant-Q line as shown. put-circuit design by adding a shunt ected from base to emitter of the tune out the base-lead inductive i1.8 ohms. To obtain the value of 0 - and the resultant real value - it is use an admittance overlay chart. ..-RFC1 c drawn on the impedance chart to Fig. 18 - Full-scale rlay admittance lines used. Then pc-board foil pat 0 :ther or not 50 ohms (the input tern and parts· 1 be reached in one step by drawing layout diagram (top ' Hance line from 3.8 ohms to the view) for the ampli fier. The bottom Q = 5 and a similar line from 50 0 dmittance overlay. If the two lines side of the pc board is a continuous cop :1 the area limited by Q = 5, the per plane which has tsformation is possible in oniy one not been etched. :ep is to determine the length and he strip-line inductor for the Input )mplish this we must first decide on thickness of pc board to be used. common pc board for rf use is G-10 xy having a thickness (w) of .062 se it. In this design we will use only 81INPUT 0 3ntenna-switching relay is construct n?' ·,oard which is mounted 9;f-OUTPUT ~ submounted using two a ng screw to assure no ls' ,plied to the stud of the tor. Part I of this article for 144MHz Is about mounting of transistors. QST for November 1972 19 Fig. 19 shows output-power performance of the 60 amplifier for a wide range of drive power. Pin•15W ...~50 Pjn•10W Final Circuit !40 P;n•7.5W The amplifier must key on automatically when in use. The simple rf-powered relay circuit shown ; 30 P;n•5.0 in Fig. 20 is used. This circuit was constructed from Q. -- r- ; 20 - junk-box parts and was found to work satisfactori 0 ly. Another circuit which was described recently in 10 QST5 will work equally as welL Other features 0 such as an indicator lamp or relative rf output 144 145 146 147 148 indicator can be used if desired. FREQ. (MHz) When using an open-frame relay for switching Fig . 19- Test results of the 840-12 amplifier. the rf amplifier in and out of the circuit, any mismatch introduced by the relay must be elimi nated . K1 was modified in a manner similar to that shown in the article by K7QWR (see footnote 5). In addition, two series-connected capacitors were Emplo Circuit Layout used to tune out reactance introduced by the relay . Fig. 18 shows the pc-board layout and compo This is required since the amplifier is fixed tuned nent placement. Placement of shunt capacitors has and thus has no adjustments to compensate for the relay inductance. · been varied to optimize performance. Capacitor C3 N THE R is the most sensitive as to position and variation of The amplifier was evaluated with a Swan I mo st irr 1/8 inch either side will change the output power FM-2X and a Standard SR-C826M. Most imported the receive by 10 percent. Special attention should be given to transceivers have VSWR protection circuits which quencies to shunt capacitor Cl. It consists of two 200-pF are adjusted to be sensitive to any mismatch to (TU). If, fo capacitors mounted close to the transistor case as protect the ouwut transistor. Do not attempt to print from shown in Fig. 18. Optimum performance can be abort or modify a protection circuit. 850-Hz shif achieved only if the recommended shunt capacitors A number of amplifiers were built using the at 2125 anc are used . RFC1 is etched on the pc board. circuit of Fig. 18, indicating the unit is reproduce audio Pane To evaluate the design the amplifier was tested able. It is the author's hope that this introduction audio spect without accessories attached. When power was first to microstrip-line power circuit design will stimu from 100 applied, the output level was 38 watts with 10 late further investigations by amateurs. To en analyzer or watts of drive. The output power was raised to 45 courage work with rf power transistors, the Under hams use, watts when the proper location for C3 was wood capacitors, pc boards, transistor, and com spectrum an determined. The amplifier is very stable for drive plete kits, for the 40-watt amplifier are available Spectru r levels from 0 to 15 watts and supply voltages from from Power Kits, P.O. Box 693, Cupertino, CA ways, but 8 to 15. Harmonic attenuation measured: 95014. The price, $47.50, also includes a heat sink selectivity ~ 2nd -37 dB and a predrilled chassis. [g]Ej of the cry 3rd -40 dB 5Hejhall, "Some 2-Meter Solid-State rf Power however, a 4th -45 dB Amplifier Circuits," May, 1972, QST. analysis call which has b the digital c running thr analog-to-di1 OC AMP. a computer (samples) of wave form 1 higher math EXCEPT AS INDICATED , DECIMAL VAWES OF tude vs. f req CAPACITANCE ARE IN MICROFARADS I ,JJf l ; OTHERS ARE IN PICOFARADS I pf OR JIJ.IF'} ~ Using th RESISTANCES ARE IN OHMS; average ham i•l 000 . computer ar However, it transform sy Fig. 20 - Diagram of the rf-powered switching INPUT only one £, ;·~M.K18 ~ 7 ~lK1CUTPUT circuit. Resistors are 1/2-;Watt composition and S. M. * Stanfo AMPLIFIER I capacitors are disk ceramic, except where noted otherwise. Ave., Menlo CR1, CR2 - High-speed silicon switching diode, tJ'rf~~e maJ CA1 1N914 1 N914 or equiv. K1 - 4pdt relay, 10-A contacts, 12-V coil, ·modified per instructions in QST for May 1972, pg. 45. Fig. 1 - T RFC1 -Miniature rf choke. voltage at e (B) S1 - Spst toggle. input sinew 20 QST for Novernl 11 1• 115 MJI vn•12 5V Many excellent low-power portable " ___.. transceivers are being used ·on two v meters. These rigs have receivers that 1l / / are as good as most fixed installations. / The only problem with the little rigs is / their low power. This article gives the operators of such equipment some thing to think about. IS 10 " " " 'lfii ·PDWEIIIHI'UT ·WATTS (A) Fundamentals of 90 Solid-State .75 ....fJ) Power-Amplifier ~ v 3:: / ....I 60 Design ~ ....n. Fig. 2- 45 v A and B i5 Part III a: w Care for 3:: 30 7 0 n. I is a new 0 BM70-12 n. in rf po11 15 Vee •12.5V BY J. J. JOHNSON,* ex-K4WYQ, 1•175 MHz Let's AND R. R. ARTIGO,* W6GFS for the 1 3 6 9 12 15 IB curves, r watts in 1 PIN- POWER INPUT-WATTS watts ou o ENCOURAGE as much interest as possible, will resul T this article will cover amplifiers for two a 12.5-vo ( B} different power levels. The first will operate with with up 1 drive levels from 1 to 4 ·watts, to deliver an output MHz, we of up to 15 watts. The second amplifier takes from power, as 5 to 15 watts of drive and wm give up to 8Q watts Which Wl of output power. Each amplifier has 50-ohm input the 2-wa1 and output circuits. The two amplifiers can be local pat~ 4 connected together as shown, to give up to 80 1 r watts of output power, when driven by most Tech hand-held transceivers. An rf-activated relay circuit 3 J is provided for use with either or both amplifiers. a Design Objectives Q. IJ Design a power booster for a hand-held trans J ceiver. This power booster must be capable of reliable operation in mobile service and meet the ~ ~ following requirements; ~ "' 1) Supply voltage: 11.5 to 14.5 V de. 2) Drive level : 1 to 4 watts. o 10 20 30 40 50 so ro eo so 100 3) Output power level: SO to 80 watts. Po WATTS 4) Automatic switching. (C) To meet our design objectives we must first select a transistor designed specifically for Fig. 1 -Power input-output curves for the 812-12 (A) and BM70-12 (Band C) transistors. Curves A land-mobile applications. In this case J;o~e ..have and B are for 12.5 volts and 175-MHz operation. selected the CTC 812-12 and the BM70-12. The With 13.5 volts, and at 146 MHz, performance B 12-12 is a standard device used as a driver or improves as shown in Curve C. predriver in commercial applications. The BM70-12 28 QST for April r portable ·d on tu'o :eiw···· tbat ,.-. ·s. •. :s lt ._ ,be nent some- L2 tJ1.6 ~ ~ 1o-1r-o~on als of p 1.sn J.C2 ~te (A) plifier (8) 1 Fig. 2- Design information for the input and output circuits for both 144-MHz amplifier stages . Curves A and 9 are for the driver stage, using the 912-12 transistor. Curves C and Dare for the 9M70-12. A and Care for the input circuits and 9 and Dare for the output circuit. See Figs . 2C and 20 on next page. is a new device representing the latest technology By se lecting devices designed for mobile in rf power transistor design.l applications, we now need only to refer to im ex-K4WYQ, Let's first review the power input/output curves pedances and typical efficiencies to ensure a proper for the two devices. From the 12.5-volt 175-M Hz design. All other parameters are designed to exceed ),* W6GFS curves, Figs. I A and I B, we see that a nominal 2 minimum require ments se t by the manufacturers watts into the 812-12 will yield from 10 to 12 of land mobile equipment. The amplifier will use watts output. Applying this power to the BM70-12 all microstrip-line technology, with fixed-tuned elements. This technique was chosen because of its 1 in•· • as possible, will result in approximately 70 watts ou tput, with ~ for two a 12.5-volt de power supply. In mobile applications ruggedness and reli ability. 'ir :ate with with up to 13.5 volts on the collector, and at 146 . ;n output MHz, we can expect up to 80 watts of output Circuit Design ~· ~m~-- -· takes from power, as in ,Fig. 1C. This represents a 15-dB boost, A step-by-step design of mi cros trip-line circuits •ill give up to 80 watts which will make a substantial improvement over was covered in earlier parts. Therefore, this section ifier has 50-ohm input the 2-watt signal, in work over other than purely will be devoted to the design accuracy of this wo amplifiers can be local paths. technique. Detailed design information can be ;n, to give up to 80 lTechnical Topics, Feb., 1973, QST. found on the Smith charts, Fig. 2, for those who 1hen driven by most ·-activated relay circuit •er or both amplifiers. Table I- Comparison of Design Values With Actual Circuit Values ?ctives After Tuning for a hand-held trans Circuit Component Design Value Post Tuning Value Error r mu st be capable of C1 110 pF 100 pF - 10% ~ service and meet the 912·12 input L1 1.10 inches 1.40 inches +21% 612-1 2 C2 57 pF 47 pF - 19% to 14.5 V de. output L2 2.00 inches 2.70 inches +25% ttts. BM70·12 C3 510 pF 400 pF - 22% 50 to 80 watts. input L3 0.64 inches 0.54 inches - 15% C4 150 pF 136 pF - 09% bjectives we mu st first L4 2.50 inches 2.60 in ches +04% C5 36 pF 33 pF -08% 1gned specifica ll y for In this case )!V e ..have BM70·12 L5 0.50 inches 0.68 in ches +25% output C6 210 pF 200 pF - 05% and the BM7G-12. The L6 2.60 inches 3.02 inches +14% ice used as a drive r or C7 35 pF 47 pF +25% plica tions. The BM70-12 QST for April 1973 29 The W6GFS 1-1/2 watts amperes po~ right. Fig. 2C load can be or 15 perce of a dB in € for a 10-• Another eri BI2-I2 inpt reactance ir circuit. From T< a two-stage close the reactance sli (C) strip-line tee Booster amplifier b• Fig. 20 t . c ' See caption on preceeding page. (0) are interested. The values of the components on overlay chart with available capacitor values drawn the schematic do not reflect the design values, but in. This overlay chart will aid us in determining rather are the values after t\lning. Table I compares desired line lengths for each particular capacitor design values to post-tuning values. A discussion of selection. An admittance overlay with these guide •1 ·32 UNC 1 resulting errors follows : lines becomes necessary if the capacitor selection is 1) Error in Capacitor Design Values: If we very limited. .. eliminate the extreme vaiues and average the rest; we find a net error of I 0 percent on the low side. 2) Error in Microstrip-line Lengths: The two This error is mostly due to the narrow selection of largest errors in line lengths can be found in the components available and their tolerances (2:. 10 networks determining the load for the power percent). This sort o( imor is difficult to com- . transistors. This is primarily the result of the load pensate for in future designs; however, our impedances given on the da t~ sheet representing NO accuracy can be improved if we use an admittance optimum load for optimum gain. In most cases the 30 QST for April ] The W6GFS amplifier in actual operation, showing 1-1/2 watts drive at the left; approximately 9 amperes power drain, center; and 70 watts output, right. load can be adjusted to improve efficiency by 10 or 15 percent, while sacrificing only a few tenths of a dB in gain. In this case the load was adjusted for a 10-percent improvement in efficiency. Another ei:ror of interest can be found in the B12-12 input circuit. This is most likely due to the reactance introduced by the relay and switching when the transceiver is keyed . This is a must, since circuit. few transceivers have an external keyin~ circuit From Table I we see that it is possible to design available. A second problem encountered with a two-stage fixed-tuned amplifier and be fairly booster amplifier operation is the quality of the close the first time around. A Smith chart, match between the transceiver and the external reactance slide rule, and basic knowledge of micro amplifier. This is a particularly difficult problem strip-line techniques are all that is necessary. with. imported equipment. These rigs require sensitive mismatch detection circuits to protect RfActivated Relay their power transistors. This protective circuit will Booster amplifier operation requires that the not permit the transceiver to key with the slightest amplifier be automatically switched in the line mismatch. f------1 000 NOM .------t MATCHED 112" FLANGE .465 .485 EQ.OIA. .130 .475 T25 .465 .485 NOTE• ALL DIMENSIONS IN ! INCHES Jr values drawn ALL LEADS 5 MILS ! in determining THICK cular capacitor fi th these guide :itor selection is 1gths: The two •e found in the ( 8) for the power suit of the load NOTE All d•mtns•ons shown HI'" •11chts eet representing Fig. 3 - Mechanical details of the amplifier n most cases the ( A 1 transistors, 812-12 -(A).and BM70-12 (B). QST for April 1973 31 , · I I 11:: ... c' > 1.) ;;; Q 1.) Cl) > z "! z A numl in recent p opeR-frarr modificatio tune out tl circuits de~ techniques, pensation f, and its intf must use a compatible, All connect 2Hejhall Amplifier C 3Johnso Power Amp QST. 4 orr, R, 32 QST for April ] 50-ohm strip-lines.5 The input VSWR of the circuit 4sed in this article was better than 1.2: I for the entire 2-meter band. The circuit introduced by Hejhall in May, 1972, QST was used for rf sensing. The Motorola HEP S9100 was used in place of the MPS-13A without noticeable change in performance. A I OK-ohm resistor wa s used in the input to set the keying level for an rf drive level of 500 mW or more. The level is adjustable to less than 100 mW by using a lower va lu e of resistor. Construction Hints Standard G-1 0 glass-filled epoxy printed ci rcuit board was used for this assembly. Only this kind of pc board can be used for this design. The circuit board layout is available for those inte~ested in ";i etching their own boards.t This can be done with li any pc board etching kit found in most electronic il stores. Check the instructions carefully before I starting. After the board is completed, the ir microstrip-line width should be checked. This line b as 125 mils wide, (I mil = .00 I inch), and it can ;, vary .:t. 10 percent without noticeable effect on 0 circuit performance. ti c All holes shown from the top side of the board It should be drilled out by using a No. 48 drill . i. Device mounting holes should be cut into the f board by using the small center holes for guides. ,. Refer to the package drawing for hole size. AI\ :c ;, holes shown from the back side of the board are 0 used for mounting components and can be drilled rt out with a No . 56 drill. U~e the four corner holes I, for mounting the pc board with No. 4 machine If screws. :; Connections between the top and bottom r su rfaces of th'e board are made through eyelets. h· (USECO No. 5609.) One eyelet is installed in each 0 1 remaining No. 48 drill hole. To install the eyelets, 01 a\ use a Key stone No. 1714 eyelet punch . After all II, eyelets have qeen installed, carefully solder them on both sides of the board to ensure good :ri connections. 01 The printed circuit board must be mounted I 00 1e mils above the heat sink. This will allow the Ia ·Ia transistors to be mounted directly to the heat sink tt A number of good circuits have appeared in and have their leads lie flat on top of the pc board. in recent publications. 2, 3, 4 These circuits used Mo st No. 4 flat washers are 25 or 50 mils in Je open-frame relays and generally require thickness, and will serve well as spacers. Provisions 10 modifications and special circuit considerations to are included on the board for activating or de H tune out the reactance introduced. The amplifier activating the rf sensing circuit. An additional l, circuits designed in this article use microstrip-line circuit is included for an indicator lamp if desired. )f techniques, with limited variable elements or com· Capacitors Ca. C4 , and c6 in the circuit pensation for the reactance introduced by the relay diagram represent two capacitors each. One is and its interconnecting leads. For this reason, we mounted on each side of the microstrip-line, as must use a good-quality relay, which is strip-line shown in the photograph. Solder all these compatible, and a high-impedance sensing circuit. Underwood capacitors as close as possible to the All connections made between circuit and relay use microstrip without shorting it out. 2Hejhall, "50-Watt 2·Meter Solid·State Power 6 Dimensions for microstrip lines of 50-hhm Amplifier Circuits," May, 1972, QST. impedance are given in Table I of Part II, QST for 3Johnson·Artigo, "Fundamentals of Solid-State Novem'ber 1972, page 18. Power Amplifier Design," Part II, November, 1972, t Pc board layout and parts placement guide is QST. available for $1, from Power Kits. P 0 Box 693, 4orr, Radio Handbook, 19th Edition 19·8. Cupertino CA 95014. 33 QST for April 1973 'A ne vh t ( re rg lg ~k n' !er a A ox L~ - A"o- 11 Designing Solid -State RF Power Circuits Part 1: Solid-state rf power amplifiers present design problems unlike those encountered with vacuum tubes. But those problems are not insoluble. This three-part senes shows how the experimenter can "roll his own." By Richard K. Olsen,* N6NR A s much as some would hate to amplifier for the 450-MHz band. The - 20 admit, the influence of the vacuum tube ....VI vee •t2.5V questions that must be answered include j in rf power amplifier design has greatly .... input drive , output power, supply volt r---r-- PIN • 3.0W I diminished. Through the efforts of ! ~~ - age, frequency and operating band Q: - - i solid-state pioneers such as Helge Gran "-' r-- width. From these questions we can 3: iO c--- 2.0W berg, OH2ZE, the dream of a solid-state ~ now generate a list which will dictate kilowatt is now a reality .1 Solid-state .... - - the requirements that must be fulfilled :::> f-. t.ow ,design, coupled with advances in micro := 5 by our design. They are as follows. :::> strip and complex multilayer stripline o_ - techniques, has generated a new breed ~ b Available input drive = 1 to 2 watts ~ 400 420 440 460 480 500 520 Minimum output power = 10 watts of vhf and uhf land-mobile transceivers I, FREQUENCY (MHz) which are much smaller and efficient Supply voltage = 12 .5 to 13.6 volts than their old tube predecessors. Also Freq. and bandwidth= 430 to 450 MHz consider the lineup of available add-on Fig. 1 -Output power versus input power amplifiers which enable a person to and frequency for the Motorola MRF618 Now we must select the proper transistor. This information, and that of convert his hand-held radio to a mobile Figs. 2 through 4, would be available to transistor for our application. Nearly all r radio by merely slipping it into a sleeve the experimenter/designer through a data of the major semiconductor manu d housing. How many people remember sheet obtained from the manufacturer of facturers have tables which list categori s carrying around a hand-held tr sceiver, the device being considered . cally all of their devices as to supply d two thirds of was fille ith 10 voltage, frequency band, output.power f pounds of batte and gain. They can be obtained from This new naturally necessary data for the design of an either the marketing department or \\ generated a new ' power amplifier. Then we will go through a literature distribution centers of each of f: amplifier design, which y cases design exercise from start to finish, these companies. c the experimenter has exposed encompassing theoretical and practical After looking down the list of de p to. This new technology is not any more design, testing and evaluation. By the vices we come upon the MRF618 which difficult than that associated with tubes time we are through, there will be is made by Motorola. The sheet shows and there are those who feel (and I enough material for you to utilize the that the minimum gain at 470 MHz is 6 agree) that in most cases the design of basic design tools to create an amplifier dB at an output power of 15 watts and solid-state PAs is simpler. suiting your own needs. a supply voltage of 12 .5 volts . This In this series we will explore the device is the closest one to our needs. world of solid state from transistor to Design Criteria and the Next we obtain the data sheet and from antenna connector. We will examine the Data Sheet the data sheet we can obtain all of the Smith Chart for its role in rf design and The first thing you should do in dynamic and electrical characteristics learn how to use it as a tool for design designing an amplifier is to list your we need for our design. and analysis. We will learn to use the objectives or "design criteria." This will Fig. 1 shows a graph of output transistor data sheet, to select a proper aid you in choosing the proper transis power vs. frequency at 12.5 volts for device for the job, and to extract the tor for the job and clarify the minimum several input power levels. Since 450 requirements that the amplifier will MHz is the upper limit of our band *4292 Quapaw Ave., San Diego, CA 92117 ; have to meet during testing and evalua usage, we can see that at 2 watts input I_ Engineering Consultant to Swan Elec tion. the output power is about 11 watts, tronics, 9233 Balboa Ave., San Diego, CA. F 1 Our design exercise in this text will falling well within our output power See the bibliography listing at the end of this SL part of the series. be to design a hand-held transceiver requirements at minimum supply volt- tr 28 Q5T'L- Now we are ready to generate anoth ,------1 20 Vee •12.5VOC __...v er list which we will use in our design I-f •470 MHz ·exercise. I v / v Test Vee= 12.5 V de and 13.6 V de v Pout vs. P;n at 1 and 2 W input 5v P;n vs.Pout at 15 and lOW output Test Frequency= 450 MHz 2 3 4 Zin = 3.0 + j5.5 .s1 P1N,INPUT POWER (WATTS) ZoL = 3 .0 + j2.5 n There are more things that can be Fig. 2 -Output power versus input power learned from the data sheet. These will for the Motorola MRF618 transistor. be discussed later in the· text. Simplified Smith Chart Mapping Techniques Fig. 4- Series-equivalent impedance of the age. Figs. 2 and 3 are useful for giving us One of the most important tools Motorola MRF618 transistor versus frequency. a rough estimate of how much power used in rf circuit design is the Smith output can be obtained for various Chart. It functions very well as a "road supply voltages and input .power levels. map" for plotting the direction and circumference of this smaller circle the Fig. 4 is a portion of the Smith magnitude of impedance transformation resistive component of Zs remains con Chart which describes the input (Z;n) when designing or analyzing the com stant at 50 n. Therefore, this is called a and output (ZoL) load impedances ponents of an rf circuit. To understand "constant resistance circle." from 400 to 500 MHz. For the moment how to use the chart effectively, we The two semicircles that intersect we need only be concerned with the must first examine its mechanics and the 50-.sl points on the outside of the table in the lower right-hand corner. learn how to interpret the many bits of chart are plots of all points where Zs = This table tells us that at 450 MHz, Z;n data that can be extracted from it. Rs ± j50 .st . The points along the is approximately 3.0 + j5.5 ohms and right-hand semicircle represent a con that ZoL (Zout) is approximately 3.0 + Series-Equivalent Impedance stant reactance of +j50 .st. If these j2 .5 ohms. This information is critical as First of all, the type of Smith Chart semicircles were drawn to include all it is the starting point when we design displayed in many publications repre points outside the chart, they would our input and output transformation sents the impedance coordinates for a also be complete circles and therefore circuits. series-equivalent circuit. Fig. 5 is an are called "constant-reactance circles ." From the data sheet we also see that example of a series circuit containing a We can now use the simplified chart this device is capable of withstanding a given amount of resistance (Rs) and to locate or define a specific complex 20:1 load VSWR at all phase angles, at inductive or capacitive reactance (±jX8 ). impedance value. In this case we wish to rated power output and supply voltage. The relationship for the impedance (Zs) define points A and B which lie on What this means is that the MRF618 of a series-equivalent circuit can be opposite sides of the centerline. Starting can deliver rated power to any type of mathematically represented by the at Rs =50 nand moving left, we notice load, whether it be capacitive or induc formula that we intersect the line representing r tive or anywhere in between, which -j50 .s1 at point A. Tllis point is II represents a 20:1 VSWR, without being Zs = Rs ± !Xs (Eq. 1) therefore defined as Zs = 50 - j50 .st. I· damaged . This is important when con Starting from Rs = 50 S1 again, move to i sidering what can happen during day-to The small "j" or j opera tor, as it is the right this time, and intersect the line ly day use of the amplifier', such as a called, is very simply a symbol used to representing +j50 S1 at point B. Point B, ~r forgotten antenna connecti"on. indicate a value of reactive component therefore, is defined as 50+ j50 S1. This m Other bits of information which we in a complex impedance. The plus(+) or is the basic procedure to use when Jr will examine later are mechanical speci minus (-) sign indicates whether it is defining or locating any series-equiva ::>f fications. These are important when inductive or capacitive , the plus (+) lent impedance (Zs) on the Smith considering physical layout of the am indicating inductive reactance and the Chart. e plifier. minus (- ) indicating capacitive reac ;h tance. The R is pure resistance. The R Parallel-Equivalent Admittance vs and jX components, therefore , combine We now know how to evaluate series 6 to represent what is called "complex components in an rf circuit, but what of ld ;;; 20 impedance." tis f PIN •3.0W Fig. 6 is a version of the Smith Chart f- 1-t •470 MHz ls . ~ that has had all of the circles removed ~ 15 •m a: ./ except for those representing Zs = 50± he "'~ j5 0 .st. The centerline of the chart :i 10 v .cs / represents pure resistance from zero to 5 / infinity ohms, reading from top to •Ut :: 5 bottom. The perimeter of the circle or represents pure inductive and capacitive 5r I 6 8 10 12 14 16 reactance in ohms reading from 0 at the r r'\ 1'. Vee, SUPPLY VOLTAGE (Vee) top to infinity at the bottom. The full IU !, ______, circle intersecting the 50-.sl point on the :ts, resistance scale is a plot of all points in Fig. 3- Output power at 4 70 MHz versus ;er supply voltage for the Motorola MRF618 the chart where Zs =50 ±JXs .st. More >1t - transistor. simply, this means that all along the Fig. 5- Series-equivalent circuit. August 1977 29 / This is a very popular chart form used by rf circuit designers (Form ZY-01-N, Analog Instruments Company, Inc.*). The centerline now contains values of Rs and Gp and the perimeter contains values of not only Xs and Bp but wavelengths to and from the generator. Circuit Analysis Before we go ahead, we must first understand the meaning of normalized impedance. Normalized impedance (ZN) o-H-{) is defined as the actual impedance of the device (Zs) divided by the system impedance (Z A). Mathematically, - Zs ZN - - Further, (Eq. 6A) ZA R - Rs and (Eq. 6B) N- ZA X - Xs (Eq. 6C) N- ZA The system impedance can be simply defined for our use as the impedance represented by the center point on the Zs-Gp line. The advantage of a chart like Fig. 9 is that it may be used for a circuit employing any system impedance. Since we will be using a 50-rl system, our center point (which was defined in Fig. 6 as 50 rl) will equal 50/50 or 1.0. The Fig. 6- Simplified Smith Chart for impedances. center point in Fig. 9 is just that, 1.0. Hence the name, "Normalized Imped ance and Admittance Coordinates." parallel components? The Smith Chart Fig. 1OA shows a typical input can also be used to evaluate parallel transformation circuit consisting of a components but we must invert our way connector, parallel capacitor, series of thinking and utilize the concept of inductor and transistor base. Since this parallel admittance instead of imped is to be a hypothetical-case situation, Jet ance. Because admittance (Yp) is the us state that the input impedance for opposite of impedance, we can define it the circuit is to be 50 + jO n and the mathematic ally . Z;n of the transistor is 1 0 + jO n. C 1 Fig. 7 - Parallel -equivalent circuit. and Ll of Fig. 10 perform the required impedance transformation between ~· Yp---- 1 (Eq. 2) Zp these two impedances. Now that we have the impedance data, the circuit can Therefore, if we have a parallel-equiva -Xs be redrawn as in Fig. 1 OB, with the (Eq. 5) lent impedance of 50 n, the parallel Bp = Rs2 + Xs2 coaxial input port as the generator and equivalent admittance would be 1/50 n the Z;n of the device as the load. or 20 millimhos (mmhos). By using Yp Fig. 8 is similar to Fig. 6 but this At this point I would like to state convention we can now express our time with circles to represent Yp = 20 ± that the most commonly accepted con selves once again in the additive proper {20 mmhos. The reason why the chart vention for design and evaluation using ty. appears inverted will become apparent a Smith Char.t mapping techniques is to little later on. In this chart, Gp is start from the load and work back Yp = Gp ±jBp (Eq. 3) plotted along the centerline and Bp is toward the generator. Using Eqs. 6B and on the perimeter of the chart. The 6C , we can plot our starting point, Z;n, G is conductance, the inverse of resis center circle is the "constant-conduc on the Smith Chart as shown in Fig. 11. tance, and B is susceptance, the inverse tance circle" and the outer arcs are the The plotted value is 0.2 + jO fl. Our , of reactance. (Refer to Fig. 7 .) The sign "constant-susceptance circles." Point A value of Zc is 50 n, plotted as 1.0 + jO before the j operator has now taken on can, therefore, be defined in the same n. We can now calculate how much a new meaning, positive being capacitive manner as with Fig. 6 and is found to be reactance we need for each component, and negative being inductive. The fol 20 + j20 mmhos. And point B is defined Cl and Ll. lowing equations illustrate the change as 20 - j20 mmhos. Now we have a As shown in Fig. 10, we desire to from series-equivalent impedance to means by which we can define both *Smith Charts may be obtained at most parallel-equivalent admittance. series and parallel elements in a circuit. university book stores. They may also be l By taking Fig. 8 and laying it on top ordered {100 for $13 postpaid when re· Rs of Fig. 6 and adding most or all of the mittance is enclosed) from Analog Instru· I Gp and (Eq. 4) ments Co., P. 0. Box 808, New Provi R/ +X/ remaining circles, we arrive at Fig. 9. dence, NJ 07974. 30 05T.,._ sed -N, .*). 1 b tor. [lrSt zed ZN) : of tern 6A) Fig. 9- Smith Chart showing "Normalized Impedance and Admittance Coordinates" (Analog Instruments Co. ZY-01-N). 6B) 6C) o-H-o 11. In doing this we arrive at Zc ==~ 1.0 + jO D. , which is the value of Zc, so 1ply therefore the circuit is properly trans wee formed with the component values indi the cated on the chart at point A. We determine the correct circuit ig. 9 values by first reading the normalized .·cuit impedance and admittance values from :ince the Smith Chart at point A. The imped our ance may be read as 0.2 + j0.4 D., where Fig. the 0.4 represents the normalized induc Thr tive reactance of Ll . From Eq. 6C, where this 0.4 equates to XN, we may 1 Fig. 8 -Simplified Smith Chart for admittances. p determine that the required inductive 0 reactance (Xs) is 20 ohms. And from ,put- the usual reactance equation we may of a add an inductive reactance (Ll) in series do not know this value either, but we determine the required inductance at eries with Z;n to transform the impedance to do know that the plotted admittance our intended operating frequency. this 50 D. (normalized value of 1.0). We will lie somewhere on the 1.0 constant As just shown, the normalized im represent this on the Smith Chart (Fig. conductance circle. The solution to our pedance of L1 in series with the input 1, let : for 11) by plotting the value 0.2 + iXs D., problem, then, may be found from the impedance of Ql is 0.2 + j0.4 D.. But , the where Xs represents the inductive re Smith Chart by locating the intersection since Cl is a parallel element, we need . Cl actance of L1, as shown in the diagram of the 0.2 constant-resistance circle and to transform this impedance to an ad 1ired of Fig. 1OB. Iititially, we do not know the 1.0 constant-conductance circle. mittance. This may be done simply by vee n this value for X.s, but we do know that So, starting with ZN of 0.2 + jO D. reading the admittance coordinates at we the plotted impedance will lie some on the centerline, we move toward the point A in Fig. 11 , 1.0 - j2.0 mhos. :can where on the 0.2 ·constant-resistance region of +jX8 , as shown by the arrow This value represents the parallel the circle . And since Cl is an element in in F)g- 11 , to the circle representing a admittance equivalent Qf L1 and th• · and parallel with Zc, we must treat the constant conductance of 1.0. This series input impedance of Ql. as shown combined impedance (Cl and Zc) as an occurs at point A. Since Cl is a parallel in Fig. 1OA. Th.e purpose t\f· ijl, is to ;tate admittance. The following equations element we plot its transformation by cancel the susceptive - j2 .0 poP- of con- apply. following the constant-admittance circle this parallel equivalent, so the no~ Ising and moving toward the +jBp area, as ized capacitive susceptance of C 1 must Yp indicated by the second arrow in Fig. be +j2.0 mhos (BN)· From this ai:fd' Bq. is to YN =- (Eq. 7A) oack YA ~ and Gp Zin • GN = (Eq. 7B) . 11. YA Our Bp + jO BN (Eq. 7C) m ch YA !]Xs 1ent, where Y N represents the normalized YA ~"""""'\ admittance and the system admit tance (1/50 or 0.02 in. this case). From (A) (B) n. these, the combined admittance of Cl so t-- n re- and Zc is 1.0 + jBp mhos, where Bp Fig. 10- Hypothetical input circuit for a power amplifier. Components C1 and L 1 perform 1stru- represents the capacitive susceptance of the matching transformation from the desired circuit input-impedance value to the input >rovi· Cl, also shown in Fig. 1OB. Initially we impedance value of the transistor, represented at B by Rs ± iXs. I August 1977 31 ------, Danley , ''Mounting Stripline-Opposed ·Emitt er (SOE) Transistors," Applica tio n Nore A AI.._ 555, Motoro la S PD. Dav is, " Matching Network Ocsigns With Com puter Solutio ns," Applica tion Noll' AN--- 26 7, Motorola Semiconductors, In c. Cra11hcrg, " nroadhand Tran ~ for!lll'I' S a nd Power Cu m bi n ~t i o n T cc.: hni4u es for IU ··~ Applica tio n Note AN- 749, Motorola S I'D . Granb erg, " Broadband Linear Power Ampli fiers Using Push-Pull Transistors," Applic- atio n Note AN-593, Motorola S PD . Granberg , "A Two Stage I KW Solid State Linear Amplifier," Applica tio n Note AN'--- 758, Motorola SPD. Granberg, "One KW - Solid-State Style," in two parts, QST for April and May, 1976. By S Hejhall , "Systemizing RF Power Ampli fier_ Design ," Application Note AN282A , Motorola SPD. Hejhall , " Solid State Linear Power Amplifi er Design ," Application No te A N-546-: ''A Mo to ro la SP D. Hejhall , "Some 2-Meter Solid -Sta te R F Worl Power- Amplifier C ir cuits ," QST, May, !ems 1972. Johnson and Artigo, "Fun dament als o f Soli d inter State Po wer-Amplifier Design ," QST, to b September , 1972 . Koche n , " Prac ti cal VHF a nd UHF Coil Win d and ing Da ta," Ham Radio, Apr il , 197 1. n oi'l Matthaei, ' 'Tables of Ch ebysh ev Impedance hash T ransfor ming Networks o f Low I' ass Filt er Form ," Proceedings of th e !EEE, August , am~ I 1964 de vel· Olse n , " 100 Watt Solid S ta te Power A mplifi er Fig. 11 - Solution of mput-circuit transformatiOn problem w1th the Smith Chart. for 432 MHz," Ham Radio . Septembe r, thi s 1 19 7 5. llUW Perna and Klein , "To You It 's a Capacitor - 0 But What Does the Cir cuit See " EON, !\ Nove m ber 5, 1973. ON-( Sm ith , 1.:.'/ectronic A pplicatiou.\· of th e Snutli Chart, McGraw-H ill , 1969. 7C, th e required susceptance fo r CJ pie and calcul ate th e va lues required for Sobol, "Exte nding IC Techno logy t o Mi cro 1 Fran , (Bp) is 0.04 mho. We may convert Ll and Cl at 450 Ml-lz , the actual wave Equipment ," Electronics, Vol. 40, V./o March 20, !967, p. 112. para ll el sus ce ptance (B p) to parallel components may be difficult to obtain. Tatum, " VHF, UHF Power Transistor Ampli reactance (Xp) simply by tak ing th e This problem can be solved through the fier Design," Application No te AN-1-1 , * I 01'1 D. use of a mi cros trip transformation, ITT Semiconductors. reciprocal; l /0.04 = 2 5 From the Wheeler, " Transmission -Line Properties o f rea ctance equati on we may then deter which will be covered in Part 2 of this Parallel Wide Strips by a Conformal Map mine the required capacitance for Cl at se ries. Part 2 will appear in a subseq uent ping Approximatio n ," IEEE Trans. o n Microwave Theory and T echniques, Vol. our in tended operating freq uency. From issue of QST. [ll !n----1 MTT-3 , May, !964. this proced ure, we ha ve determined that Wh eeler, "Transmissio n-Line Properties o f Re fe re nce Material and Bibliography Parallel Wid e Strips Separa ted b y a Di elec an Inductive reactan ce of 20 ohm s for Ada m , Microwave The ory w1d Applications, tri c Shee t ." !F.F:F: Trans. on Mh·rrn-wn w Ll in Fig. lOA and a capaciti ve reac Prentice-Hall , 1969. Th eory and Techniques, Vol. MTT -3, tance of 25 ohms for Cl will provide A nd erson, "S-Parameter Techniques fo r March, 1965. Faster, More Accurate Ne twork Design," You ng, " Microstrip Design Techniques fo r the proper match for a 50-ohm circuit Application Note 95-1, Hewlett Packard UHF Amplifiers," Application No te AN- input to the input impedance of th e Corporatio n. 548A, Motorola SPD. Fig. 1 Young, "Bro adband Network Des ign for UH F Beccio lini , " Impeda nce Matching Networks SBA - 1 transistor. Appli ed to RF Power Transistors," Appli Amplifiers," A pplica tion Note AN-704, lf you fo ll ow through on this exam - ca tion No te AN- 721, Motoro la SPD. Mo to rola S PD. out. Emergen cy Coordina lor WBS EDO reported a total of 3 14 points, whi ch in cre ases the total of the Ohi o sec ti on to 6,924 poi nts. Al so, th e call sign of Paul Dan zer shou ld ha ve read N 111. = If you'd like a diagram showi ng th e interconnecting cables for the ISB adap ter described in the May QST article By A I entitled " Ind epe nd ent Siueba nd for Your Dr ake TR-4C." se nd an s.a .s.e . to League hq. This a = ln th e article "M ore PEP - Less very r. P ~ tinl .. in Hint s and Kink s (QST for o Good community relations paid off wish t' Jul y , 1977). th e configuration of th e rece nlly for Dr. Charles Greene, W2CPI , blank< bridge re ctifier shows the diodes of Clayton, NJ. Wh en high winds ceiver. C RJ -CR4 in co rrectly drawn. They damaged the guy wires securing his T A !a bora should be reversed. 36 at 7 0 feet, he called on the Washing WB 2D o In the Simulated Emergency Test ton Township fire department for as "AN sistance . The commissioner himself Work .~ results, Jul y QST, the report for th e Two me ters has h it Eu rope in a big way, as co untie s of Hardin , Mari on and went up to in spe ct th e damage - and Karl Zimmer's·"Monster" array testifies. Karl, I Wya ndot (0 1-! ) were inad vertently left decided to repl ace all six guys . DC2ZG, is an electronics manufacturer. ·2o Ca1 32 0 51'1- Designing Solid-State RF Power Circuits Part 2: Microstriplines neither capacitor nor induct or nor resistor, but a combination of all three: They are transmission I i nes etched on circuit boards, designed for specific impedance requirements.t By Richard K. Olsen,* N6N R ~t ircuits containing microstrips are l O G not uncommon at uhf. But to the uninitiated, a microstripline might not Ql look much different from any of the uther conductors running their various ways to interconnect the many parts on the circuit board. A bit wider, perhaps, and generally running in straight lines where other conductors might curve (A) ( B ) about. However, there is much more than meets the eye in these miniature Fig. 12- Input circuit using the microstrip technique. transmission lines. They can be used to transform impedances. A shorted or an open line can be used to simulate an ductor (t), (3) width of the line (W) and Because microstrip works in a modified inductor or a capacitor. Combinations its effective electrical width (Weff). (4) transverse electric mode (TEM), we con of microstriplines can be used to do all wavelength at the operating frequency tinue, sorts of clever designs. (/0 ) of this type of microstrip (Aw 1 ) , There are many advantages to using and (5) dielectric constant of board f...TEM = ~ = ~ = 0.936 meters microstrip as a series transformation rna terial ( €,). ' y€, y4.8 element. An important one is greater For this example, assume our f o is (Eq. 10) repeatability from circuit to circuit. The 146 MHz . The type of board commonly microstrip, however, cannot be thought used is GlO double-clad copper printed The correction factor for G 10 glass of as being just an inductor, capacitor or circuit board. It has a dielectric constant epoxy circuit-board material is resistor; it possesses the properties of all of approximately 4.8, a dielectric thick /2 three. What microstrip is, on the other ness of 59.2 mils (1 mil = 0.001 inch), €, , hand, is a . transmission line. Trans and a conductor thickness of 1.4 mils. K= ( · 0 . 1225 mission lines make very good transfor Let us say the line width is 100 mils . 1 + 0.63(e,- 1)~w~tt) mation components. If you were to cut The effective width is calculated from the formula . 1/2 a transmission line at a given point and - 4.8 measure its impedance at that point - ( 1 0.1225 with respect to the starting point, you I+ 0 63(4.8- I} ( %·:!l) y would observe a very definite trans formation from one point to the other. Fig. 12A shows a simple circuit con = 1.161 :i- taining the same basic components as Fig. lOA (Part 1) except that L1 has f...wt = (f...TEM) • K u see been replaced by a length of microstrip X on a (Wl). = 1 02 .42 mils (Eq. 8) = 0.936 1.161 "dig To evaluate properly the amount of ~ transformation represented by W1, we = 1.087 m or 1087 mm (Eq . 11) must first know the following things: Next we must determine the wavelength (1) dielectric thickness of the board of our board material at [ 0 . We do this We can now determine the amount material (h), (2) thickness of the con- with the following formula , where C = of transformation required and the elec 300 X 106 meters/second . trical length of the stripline from map ll tPart 1 of this article appeared in the August 06 ping on the Smith Chart. Referring to 1977 issue of QST. f...o = C = 300 X 1 = 2.05 meters Fig. 13 , our Zin is once again 10 + jO D., : RL 6 •4292 Quapaw Ave., San Diego, CA 92117 ; t 146 x 1 o so our starting ZN is again 0.2 + jO n. L Engineering Consultant to Swan Elec (Eq. 9) RL tronics, 9233 Balboa Ave., San Diego, CA Next take a drawing compass, and with September 1977 15 the point of the compass on the center when it is terminated in a g~ven complex point of the chart (1 + jO), draw an arc impedan<;e . Microstrip can also be used from 0.2 + jO in the +Xs direction to to synthesize a value of inductive or intersect the 1.0 constant-conductance capacitive reactance depending upon circle . This arc just drawn represents a whether it is terminated in a short or constant-VSWR circle and intersects the open. These properties can be extremely 1.0 conductance circle at point A. Now useful at microwave when a given L we obtain a straightedge and draw a line section requires a shunt capacitance of from the center point through 0.2 + jO 0.3 pF. Such a capacitor is difficult to and on to the outer edge of the chart. manufacture and consequently is very Fig . 16 - Draw another straight line from the expensive. A piece of microstrip, ter· intersection (point A) to the center minated in an open, attached to the L point of the chart, and extend that line network at that point, can also be used . to the edge of the chart. Next follow The means of calculating this value is· along the scale marked "wav.elengths provided in the following formulas. toward genera tor" and determine the scale distance between the two straight Short: XL = Z 0 tan 8 I 8 < 90° (Eq. 14) 501\.r- lines just drawn. This is found to be 0.066 f... and represents the electrical Open: Xc = Z0 (-cot 8) 18 < 90° length of the stripline. From Eq. 11 , Fig. 13- Sol"ution of microstrip-transforma (Eq. 15) L 'Aw 1 has been found to be 1087 mm . tion problem with the Smith Chart. The physical length of the stripline may where Z 0 = characteristic microstripline be found from impedance (ohms) and Fig. 17 - 8 = electrical line length L =n · A w 1 = 0.066 X 1087 = 71.7 mm (degrees) = 2.8 inches (Eq. 12) Transmission-line theory tells us that any given transmission line terminated We can now see from the Smith in a short will, at a quarter wavelength Chart that the line has transformed us down the line, exhibit the properties of from 0.2 + jO n to 0.24 + j0.42 . The an open circuit. The inverse also applies. · impedance (Zs) at this point is now 12 And at an eighth wavelength away from + j21 n. The value of Rs has increased either a short or open the line will along with X8 , which proves that the exhibit a reactance equal in value to the line does not contain pure rea cta nce but characteristic impedance of the line. We ra ther behaves as a transmission line will use Eq. 14 to demonstrate. containing distributed amounts of both . Rs and X8 We can now ascertain the XL = Z 0 tan 8 \8 = 90 · (~~~) degrees value cf C1 with the Smith Chart. Draw an arc from 0.2 4 + j0.42 S1 to LO + jO =50 tan 190.(0.125\l S1 along the constant-conductance circle L \b.250)j that intersects those points. Next we determine that the constant-susceptance =50 tan [90 (.5)] circle intersecting 0.24 + j0.42 S1 is that Fig. 18 - Fig. 14- Reactance versus electrical length of BN = 1.8. From Eq. 7C (Part 1) C1 =50 tan 45° has a BN = +j1.8 mhos or Bp = +j36 for shorted microstripline. mmhos. The capacitance can be cal =50 (1) =50 ohms culated by I C = Bp = 36 X 10-3 6 'I.e=- z--rr~C. w 2 X 1T X 146 X 1 0 Referring to Fig. 14 we can use the ' Smith Chart to demonstrate this very =39.2pF (Eq. 13) same thing. Starting at XN = 0 S1 and It is important to understand that moving 0.125 A toward the generator, when mapping the magnitude of trans we arrive at +iXN = 1 or XL =50 n. formation of microstrip, the center We may now use this technique to point (ZN = 1.0), which is the axis of design an open-ended line replacement • rotation, must correspond to the char for Cl. C1 = 39.2 pF soXc 1 = -27.8 S1 acteristic impedance (Z0 ) of the micro and XN = 0 .5 56. Referring to Fig. 15 stripline. This means that if we are using we start at XN = oo and move clockwise 35-D microstrip, the points 10 + to ·-jXN = 0.556 S1. This is equal to a jO, 12 + ]11, and 50+ jO ohms will have distance of 0.17 A toward the generator. to be renormalized to conform to the By using Eqs. 11 and 12 and a value of 35-D system requirement. It is therefore 0.17 for n, we arrive at a line length of necessary to ascertain the Z 0 of the line. 184.7 mm or 7.3 inches. Obviously, at Later in the design exercise we will see 146 MHz it is more practical to use a how this is done. shunt capacitor. We have now seen through these Microstrip Inductors and Capacitors examples how to plot or ascertain the What we have seen so far explains Fig. 15 - Reactance versus electrical length amount of transformation presented by ~ the nature of microstrip transformation for open microstripline. inductors, capacitors and microstrip- ' Fig. 19 16 05T~ lines:·There are many, many more facets between frequency and circuit Q. At complex 450 MHz, a circuit with a Q of2 would >be used to the Smith Chart that will not be 1r · examined here. What we have learned, have an effective 3-dB bandwidth of however, are the minimum facts neces 225 MHz. It is advisable to design all sary for the design and evaluation of an transformations in a broadband circuit ··. ZL rf circuit using R, L, C and microstrip. to fall within a Q of 2 or less . xt. -f We are now ready to move on the design Referring back to Fig. 4, Part 1, we given L of the input and output circuits in our find our device Z;n to be 3 + j5.5 n. ttance of · uhf amplifier. Plotting that point on the Smith Chart, 'ficult to Fig. 19, we notice that it is a very low is very Fig. 16 - Double-L network. Z;n andZoL impedance. If we were to come straight :rip, ter Circuit-Design Exercise off the base with a series line or to the L We must now address ourselves to inductor, we would soon propel the be used. the task of designing an input and an transformation to a point having a value is· W1 rather high circuit Q. Therefore, we use las. output circuit for our amplifier. In this design exercise we will dispense with C2 to bring us closer to 50 n without (Eq. 14) working the formulas in the text·and all compromising BW. This capacitor Cl impedances called out will be referenced should be a fixed value in most cases 10° to a 50-ohm system. These impedances and of a common value . Let's try a (Eq. 15) will be defined in "real-world" values 40-pF capacitor. C2 has a Bp of 113.1 rather than as a normalized value as mmhos. This transforms us to 11.5 + trip line defined on our Smith Chart. Work the j3 .1 n. Note our circuit inductance at nd Fig. 17- Input circuit. formulas out yourself. This design exer this point is only 0.27 n. cise is intended to solidify your know Next comes WI. Using the compass, ledge of the concepts brought forth in draw an arc from 11.5 + j3 .1 n to the earlier sections. constant-conductance circle intersecting s us that Let's start with the input circuit. 50 + jO n . The transformation of Wl ruinated Since we wish to have an amplifier with should intersect this circle for C1 to velength a medium broadband response, we will properly transform the impedance to 50 erties of use what is known as double-L net n . Using the outer scale we see that this applies. works. Fig. 16 shows a typical double-L line must be 0.06 wavelength long. For ay from network. This type of circuit has a the sake of simplicity we will use 50-.Q me will reasonably broad frequency response microstrip. If when laying out the cir e , and has the capability of tuning a broad cuit board it was found that the line was li ~ range of impedances. In this exercise we too short to reach to the antenna will use microstriplines instead of in connector, more line could be used ductors as they are somewhat easier to simply as a 50-.Q transmission line. A legrees deal with and reproduce at uhf. Fig. 17 line of about 100 mils wide will be is the basic schematic that we will deal suitable in this application. Its Z 0 can with. Note that only one L section is be approximated in the formula shown drawn. This is because the MRF618, in Table 1. Using Eqs. 8, 9, 10 and 11, being an internally matched device, has we can determine that A of our 50-.Q one L section inside the device itself. line is 353 mm. W1 is 0.06 A or 21.2 Fig. 18- Lines of constant Q (figure of merit). Before we go on, I must say some mm. Converting to inches, our line is thing about figure of merit, or "Q ," as it 0.835 inch long and 0.1 inch wide. is known. Q often plays a significant Using the Smith Chart we see that our role in amplifier design, especially in final transformation value must be 32 .6 a brmidband circuit. Fig. 18 is a Smith mmhos. From Eq. 13 our capacitor is Chart representation of lines of constant 11.5 pF. A 1- to 20-pF variable will do Q. ..fup.ce 0 = Xr./R we cl!!Ulefine an nicely. 1se the impedance of TO + /20 n as having a The collector side of the MRF618 is very circuit Q of 2 . Find 10 + j20 non the does not afford us ·the luxury of internal nand chart and you will see it fall on the Q = rna tching, so in the interest of band terator, 2 line. Q is important to us in the width, we will use two L sections in our n. formula ZoL transformation. que to Fig. 20 is the basic circuit we will :ement use in our output transformation. Re BW= fo (Eq. 16) ferring once again to Fig. 4, we find that 27.8 n Q :;ig. 15 our ZoL is 3.2 + j2.5 .Q. Once again we •ckwise will use a 40-pF capacitor. Fig. 21 is the al to a rna p of our output circuit. C 1, having a era tor. The bandwidth of the circuit, BW, is Bp of 113.1 mmhos, transforms us to alue of directly related to the relationship 4.9 + j0.65 n. Once again we will use tgtl- 1' Table 1 . Calculation of Microstrip Impedance 377 h (Eq. 17) these Zo tin the ._re;-. Weff r1 +1 .735 (erH'l~ (W~ffrHSHl ted by ostrip- September 1977 17 through 5 + jl 0 D. and read the amount of wavelength this transformation repre sents; 0.029 A equals approximately 0.4 inch long by 0.1 inch wide. For conveni ence sake we will allow the C3 trans formation to bring us back to the line W1 W2 through 5 + jl 0 D.. This represents a Bp of 60.8 mmhos, or converting to capaci tance , 21.5 pF. This brin~ us to the 21.5 + j6 .5 D.. For C2 we can then use a 2- to 40-pF variable. W2 must bring us to a point which will allow C3 to transform us to 50 + jO D. . Draw an arc from 21.5 + j6.5 n to Wi~ the constant-conductance circle which intersects 50 + jO D. and · draw a line we through the point of int~rsection. W2 must be 0.059 A or . 0 .82 inch. The bar remaining transformation is accom Fig. 20- Output circuit. plished with a capacitance ·of 6.5 pF. C3 Fig. 21 -Output transformation. may be a 1- to 10-pF variable. By 1 Don't be concerned with small varia 50-D. microstrip. We will allow Wl to tions in line length caused by a change transform us to a maximum circuit Q of in frequency. C2 and C3 have enough and the design will be complete. Part 3 2. Draw a line through 4.9 + j0.65 D.. range to manipulate the transformation of this article will deal with biasing, Next draw an arc from that point in a back to the desired load point. mechanical considerations and some. clockwise direction. Notice that this arc All we have to do now is to design construction limits and will appear in a catio1 passes through 5 + jl 0 D. . Draw a line our collector and base biasing circuits subsequent issue of QST. [g!EJ i-f b. partie the b kHz, and an RTTY section. Upstairs is our widtl dormitory and lecture room. This Hz. l mobile shack is now situated in with Dobriach on Lake Millstatt (Carinthia, levels Austria), where we do public relations other for amateur radio in a youth summer prom camp. We offer information to 1,800 ina de young people every year - and have I ha' done so for the _past 10 years. We also static run a license class there. 10 pt "We have a second station, They OE1XBC, operating in Vienna's largest two t youth center, where all training for ing fl social workers takes place. We rely -er. C entirely on donations and contributions, as olt both in time and money. And it seems furth we are here to stay." P:. Let's hope so. fixed filter can 1 HOWZAT? are work o The mailroom probably puzzled for a and while over a piece of mail addressed junk thusly: "American Radio Relay League, modi Attn.: QST Exciting Events Editor." signe Now, who could that be?- WI YL recei' mixi sche1 teclu The well-appointed ham shack aboard the double-decker bus that brings amateur radio to RATHER FREQ-Y thousands of Austrian youngsters each year. Rescued from the Vienna city dump, the bus be m serves as an example of what hard work and enthusiasm can accomplish. o Bob, WA2JDU, is a 36-year-old sales i-f fil representative for a drug company. So is Bob, WB2JDU. As a matter of fact, How o Over in Austria, an active and public converted into a mobile shack in count they've been best friends for 20 years. 1 spirited amateur radio club is setting an less hours of voluntary work. In the Their last names begin with the same widt example for others. As Wolf Harranth, basement we have a number of com letter, B, and they have similar speaking the : OEl WHC, described its operations: munications receivers (for be DXers and voices and habits. When anyone calls mod them, it's usually, "Is one of the JDUs I "Our QTH is a double-decker bus, SWLers) and our ham station, OE8XBC, *128 rescued from the Vienna city dump and consisting of an FT-101, some 2-m gear on freq?" 2: 18 [J5T'1... , esigning Solid -State Power Circuits Part 3: Biasing of transistors, mechanical construction considerations and more. With this conclusion of the series the experimenter can set out in earnest to "roll his own_"t By Richard K. Olsen,* N6N R Fi! tio fi: Si on ou LParts 1 and 2 of this article, we have The diameter of the wire must also ha seen how to plot via the Smith Chart be a large enough value so as not to be and ascertain the amount of transforma drop an excessive amount of de voltage pr• tion presented by inductors, capacitors during peak current conditions. In this sit and microstriplines. By a practical circuit, we will use four turns of No. 18 Wt example we have designed an input and AWG enameled wire, 1/4 inch ID. This se1 an output circuit for a solid-state am is about 25 nH or 70 n. isc TI . Fig. 22- Collector supply decoupling. plifier for 450 MHz . The final step in The capacitors that follow are used generating our schematic approximation to shunt the low-frequency components in of the ampiifier will be to come up with that return to the power supply. These pa a method of biasing the transistor. low-frequency components can cause pa There are many ways to accomplish spurious oscillations ·and generate a as this, but the primary objective is to great deal of adjacent-channel inter se apply a de potential to the transistor ference. We will use a 0.1-JlF capacitor without affecting the rf properties of followed by a 620-pF feedthrough fol ce the circuit. Essentially the bias circuits lowed by a 1-JlF tantalum. th must have a low impedance for de and a The base of the transistor must also Tl L high impedance for rf. be held at some fixed value of de an fe L A method of applying collector sup potential. The terms, Class C, B, AB and ply voltage (V cc) is demonstrated in A also apply to transistor amplifiers. 2: Fig. 22. Coming off the collector, the rf Fig. 23 shows two ways of biasing the er first sees a large inductive reactance. transistor at Class C. A much higher re This amount of reactance must, at impedance to rf is desired with this cr absolute minimum be 10 times the choke. A ferrite bead is used to further Fi (A) va ue o e Z L of e trans1 at impede any rf coupled through the di lowest operatmg requency. t choke via interwinding capacitance. Too ca MHz this would be an inductive reac low an impedance in this circuit can er Fig . 23 - Two ways of biasing a tramistor tance of about 35 ohms. This cor severely compromise the overall gain of si; for Class C operation . B = fernte bead. responds to an inductance of about 12.5 the amplifier. As a rule of thumb, at D nH. We must also be careful so as not to 450 MHz I use a 10-turn coil made of at cause the inductor to become self No. 22 enameled wire wound on 1/8- b' resonant at or near the operating fre inch form. A resistor can be added as in d; quency. The higher we go in frequency, Fig. 23B to reduce the conduction angle 0. the more effect we see from inter if desired . n• winding capacitance. This capacitance If the amplifier is to be used in ssb m coupled with the inductance of the service, a small amount of positive Cl choke can cause resonance in the V cc voltage can be applied to the base to Cl circuit. bias the transistor in a Class AB config b· *4292 Quapaw Ave., San Diego, CA 92117; uration. (See Fig. 24.) This voltage , Engineering Consultant to Swan Elec about 0.6 to 0 .7 volt, is developed pi tronics, 9233 Balboa Ave., San Diego, CA across the diode which is biased into M Fig . 24 -Method of biasing a transistor for tParts 1 and 2 of this article appeared in QST Class AS operation. for August and September, 1977. saturation. The series resistor merely Ia 22 05T'1- amplifier. Both microstriplines are 2.5 in the board and crimping the eyelet to 01 inches long by 0.1 inch wide. The the surface. A good solder connection separation between groundplane and must then be made. For this exercise we +Vee: microstrip is about equal to twice the will use eyelets placed in all the points dielectric thickness of the board. Also , around the board where a ground con RFtNo----j~X there is a small break in the output line tact will be needed. 1-5 pf 1N914 to allow for the series de isolation We must now mount the board on capacitor. The rea son why the lines are th e heat sin k. Remember when choosing (A) made long is to allow for any possible a heat sink to use on e which will error in design. Since we are using 50-D dissipate th e heat generated by the microstrip , the remaining line merely transistor during normal operation . acts as a 50-D waveguide. Thermal di ssipation in terms of 8j c Cc Once th e board has been etched , a per wa t 1) is in cluded in the da ta sheet. hol e mu st be made in the center of it to Fi g. 26 shuws two 20-pF Une lco al low for th e tran sistor nange. The dat;r capa citors in parall el at th e coll ec tor sheet provides information as to th e and base rath er than th e single 40-pF physical size of the package and al so capacitors used in the earlier examples. how much torque may be applied in This is because the Unelco capacitors mounting it to the heat sink. The next work well as supports for the tran sistor thing to do is to assure a good con tact leads. Mount the capacitors on opp osite between the groundplanes on both sid es sides of the line from each other and Fig. 25- Antenna relay circuit with varia tions. of the circuit board. This may be done solder their center tabs together. Use a by wrapping the edges of the board and small copper strap to make connection mounting hole with copper foil and with the micros trip. fixes a limit on that saturation current. sold ering it or by using eyelets. Eyelets Next we mount the transistor. Put a Since this amplifier is to be used only are put in place by drilling a small hole small amount of thermal compound on on fm we may use Class C operation in our amplifier. We also must consider what might so happen if the output connector were to C10 to be accidently shorted . By using our .ge present output configuration, it is pos 1 is sible to short out the collector supply. We must, therefore, put a capacitor in J/) l;U"' W2 :1 i .~ series with the output connector to isolate the V c c circuit from the output. Tllis capacitor must be less than an ohm in series Xc at the output frequency. A '"' pair of 0.018-.uF chip capacitors in C4 C5 'SC Cl JSC parall el with each other will work well d as they also have a minute amount of cr- series inductance at f0 . .or To use the amplifier with a trans ol - ceiver we must provide a return path for the received signal around the amplifier. Fig. 26 - MRF618 amplifier. C11 - 1-J.lF , 35- V tantalum . Is o The insertion loss back through the B - Ferrite bead, Ferroxcube. L 1 - Approx. 0.12 11H ; 10 turns No. 22 en am. amplifier would greatly reduce the ef C1 - 1- to 20-pF variable. wire closewound, 1/ 8-inch 10 air core. de C2-C5, incl . - 20 pF, Unelco. L2- Approx. 0.089 11H ; 4 turns No. 18 enam. nd fective sensitivity of the receiver. Fig. C6 - 2- to 40-pF variable. wire closewound, 1 /4-inch I 0 air core . rs. 25 shows an antenna relay circuit which C7 - 1- to 1 0-pF variable. 01- Motorola MRF618 transistor. the employs a Darlington amplifier as a C8 - 0.018-1-'F chip capacitor, Vitramon. W1- Microstripline, 0.833 X 0.1 inch . relay-driver amplifier and a "half C9 - 0.1 -J.lF capacitor, Erie Red Cap. W2 - Microstripline, 0.4 X 0 .1 inch . 1C I C10 - 680-pF fe edthrough. W3 - Microstripltne, 0.82 X 0 .1 inch . his crystal-can" relay as an antenna switch. her Fig. 25A shows the method I employ in the driving the relay amplifier . The series 00 capacitor is chosen to have an Xc great ;an enough to prevent loading of the input of signal and low enough to drive the at Darlington. About 3 pF will work well of at 450 MHz . Fig. 25B is a circuit used /8- by Roy Hejhall, K7 QWR, in some of his ; in designs. Both circuits represent about 1gle 0.4 to 0.6 dB of insertion loss. Con nections to the relay pins should be >S b made with 50-S1 coax to preven t in Jve curring more insertion loss in the cir cuit . The crystal can itself should als o ~ be well ground ed . Fi g. 26, co upled with Fig. 25, com pe,_ prises our schematic diagram of the nto MRF618 amplifier. Fig. 27 is the board rely layout I used in building the prototype Fig. 27 - Circuit-board layout used for the prototype MRF618 amplifier. October 1977 23 now begin to evaluate the amplifier to Ps AT 2 5 12. 0 8 see if it meets the original design >---- 11 . ~ cri teria . 20 T 11. 0 6 The first check is gain. Apply tw o ./ -- / 10. 5 5 >-15 w watts to the input, checking to make 1/ ~ 10.0 - 4 "0 sure the input VSWR has bee n mini a. 10 9. 5 mi zed, and observe the output power. 9.0 B In this case with 2 W of drive at 13 .5 I B. 5 1 volts , the output power is 17.6 watts. 8.0 0 430 440 450 The gain may be calculated in the 2345678 following formula PIN Ar Fig. 28- Graph of Gpe vs. frequency for the Fig. 29 - Graph of power in vs . power out at prototype MRF618 amplifier. Gpe(dB) = lOl og~ou t (Eq. 18) 450 MHz. Sin 1n The gain th erefore is 9.4 dB under those alE the transistor flan ge before sc rewing it conditions. Next we will check am plifier The next test is Pout vs . Pin · Since down to th e hea t sink . Always mount efficiency. Power up the amplifier again 450 MH z is our worst-case frequ ency, the transistor on the heat sink before to the previous conditions and observe we will measure the amplifier there . To By 1\ soldering it into the circuit! The mount Ie. Ie is 2.1 A. Efficiency may now be do this we will tune the amplifier at Psa t ing of the rest of the components is calculated in the formula and record the input power at several academic. Just remember when con output power levels. The data read as necting the collector and base chokes to ffi . Pout ( shown in Table 3. Fig. 29 is the graph of make the connection as close to the E IcJency =Vee X le X 100 Eq. 19) the data. By observing the plot of Pout transistor as possible. vs . P;n and G e versus frequency we see may Efficiency equals 62.1 percent. We can that the ampllfier easily meets the ori Testing and Evaluation shelt< safely say that the amplifier is operating ginal design criteria. Broa, The next phase of our exercise is to well beyond the requirement stated The final check is for stability and it A fe·· evaluate the circuit we have created. earlier. Now we will evaluate the am is done with the spectrum analyze r. shoul After construction and a good visual plifier under various dynamic conditions First, power up the amplifier at 15 cerne once-over to assure that all is according to determine its overall operating char watts out and observe the spectrum. E to our schematic , we are ready to tune acteristics. There are no spurious signals prese nt on a nat up and evaluate the amplifier. The first graph we will collect data the scope display . Now vary Vee from revise ~ The basic tools we will need are a for is gain vs. frequency. What we will 5 to 15 volts and check for spurious. when pectrum analyzer, input and output do is establish an output power and Now vary the output power from zero speci1 powe r meters, a current meter, a power Vee specification and meas ure the in to Psat and observe the display once In again. If you are brave you may now supply , a signal generator , and a good put power at 430, 440 and 450 MHz. ( tran ~ nonreactive 50-SG load. Some of th ese Test conditions are, therefore, Vee = power the amplifier into an open circuit warn comp onents are difficult to ob tain and 13.6 volts with Pout = 15 watts. Our and a short circuit to test for both sage) often you will need to rely on your own tes t re sults are show n in Table 2, and spurious emissions and ruggedness. Once consi~ versatility to provide a necessa ry sub Fig. 28 is a graphical representation of these tests have been made (and we follov stitute. The signal generator may be the gain versus operating frequency . hope with good results), the amplifier 15, 1 ( replaced, for example, by a transceiver. With this graph we may ascertain the may be pressed into service. two-t• 1l1e output power may be adjusted by gain that can be expected at any fre Summary Hz, t vary in g the supply voltage . You must be quency in the required band. This very careful, however , that the genera This study examines the very basic alarm tor does not generate spurious emissions elements of solid-state power amplifier tinuo because of low Vee· The spectrum design . Part 1 discusses the basic in broad analyze r is perhaps the most difficult to formation that must be extracted from tivitie find a substitute for. Regardless, we will Table 2 the data sheet in order to establish the the a1 in this section speak in terms of ideal Test Results for Gain-Vs.-Frequency necessary design criteria. Part 1 also comm tes ting conditions so as to acquaint you Measurements demonstrates a fundamental approach logica with some of the testing that is em to the Smith Chart and is carried further fo Gpe Th pl oyed in circuit evaluation. in Part 2 when working through the coder Tune-up in this case is relatively 430 MHz 1.35 w 10.5 dB input and output transformation de While simple . Apply a small amount of power 440 MHz 1.45 w 10.15 dB signs. Part 3 explains the basics of 450 MH1 1 47 w 10 1 dB requir to the input (approximately 1/2 the amplifier construction and layout. Also units required drive level) and tune the input included in Part 3 is information on detect capacitor until a small ri se in collector biasing the transistor for various classes set of current (I c) is observed. Then tune the of service and a means of switching rf sta ti o1 two output variables for a peak reading around the amplifier to allow a return the ol< of output power. Now tune the input Table 3 path for received signals. Part 3 further ca pacitor for a minumum amount of sets up a basic pr oce dure for evaluatin g The C Test Results for Measurements of Power reflected pow er while ke eping an eye on In Vs. Power Out amplifier performance . Th the coll ector tuning. As the output Much, much more can be said of mm p power increases , the Z0 L of the device solid-state amplifier design. At the end U2 a of Part 1 is a list of articles, application will also change slightly and will require 6.4 w 25 w These minor adjustments during the tune-up 3.0 w 20 w notes, and text books that will serve as ternal procedure. Once you have tuned the 1.5 w 15 w an excellent source of additional in puts : ci rcuit to the point where you feel the 1.03 w 10W formation to assist you in amplifier 0.6 w 5W which circuit is opera ting properly , you may design . illST I 0 205 w 24 () 5T"L.