CHAPTER 3 AMPLIFIER DESIGN TECHNIQUES

Transcription

1 CHAPTER 3 AMPLIFIER DEIGN TECHNIQUE 3.0 Introduction olid-stte microwve mplifiers ply n importnt role in communiction where it hs different pplictions, including low noise, high gin, nd high power mplifiers. The high gin nd low noise mplifiers re smll signl low power mplifiers nd re mostly used in the receiver side where the signl level is low. The smll signl - prmeter cn e used in designing these low power mplifiers. The high power mplifier is used in the trnsmitter side where the signl should e t high level to cross the desired distnce. The intent of this chpter is to give n overview of some sic principles used in the nlysis nd design of the microwve trnsistor mplifier. 5

2 6 3. mll ignl Amplifier Design The design procedures for smll signl microwve mplifier consist of selecting the dc is point for the trnsistor, mesuring the -prmeters of the trnsistor, studying the stility, designing the input nd output mtching network to chieve the desired gols, uilding the mplifier, nd performing the mesurements. The dc is point of the trnsistor should e determined first. The selection of the dc quiescent for the trnsistor mplifier depends on the prticulr ppliction. Figure 3. shows trnsistor chrcteristics with four quiescent points locted t A, B, C, nd D. For low-noise, high power gin, high output power, or high efficiency pplictions, the quiescent point locted t A, B, C, or D receptively is recommended. A network s ehvior t microwve frequencies cn e chrcterized using the scttering prmeters (-prmeters). These prmeters re defined in terms of trviling wves. Figure 3. shows the incident nd reflected wves for two-ports network (common-source FET). The reltionship etween the -prmeters nd the incident nd reflected wves cn e expressed s follows. + + (3.) (3.) where, i nd i re the incident nd reflected wves respectively t port i nd cn e defined in terms of the voltge wve, s shown in equtions (3.3) nd (3.4):

3 i i V + ( Z ) 0.5 o V ( Z ) 0.5 o (3.3) (3.4) where, Z 0 is the reference impednce. 0.9 I D B VG5 Id (ma) C VG4 VG VG 0. I D A D VG Vds (V) Figure 3. Chrcteristics nd recommended quiescent points for trnsistor mplifier

4 8 The -prmeters represent the trnsmission or reflection coefficients nd cn e otined s follows: 0 (input reflection coefficient with output properly terminted) 0 (forwrd trnsmission coefficient with output properly terminted) 0 (output reflection coefficient with output properly terminted) 0 (reverse trnsmission coefficient with output properly terminted).

5 9 Figure 3. A trnsistor s two-port network. The stility of n mplifier is very importnt considertion in microwve circuit design. tility or resistnce to oscilltion in microwve circuit cn e determined y the -prmeters. Oscilltions re possile in two-port network if either or oth the input nd the output port hve negtive resistnce. This condition occurs when the mgnitude of the input or output reflection coefficients is greter thn one, Γ in > or Γ out >. There re two types of mplifier stility, unconditionlly stle nd conditionlly stle. In the former, the rel prt of the input nd output impednces of the mplifier is greter thn zero for ll pssive lod nd source impednces. However, the mplifier is sid to e conditionlly stle or potentilly unstle if the rel prt of the input or output impednces of the mplifier is less thn zero for t lest pssive lod or source impednces. The stility test should e done for every frequency in the desired rnge.

6 30 Figure 3.3 shows the source, lod, input, nd output reflection coefficients for two-port network. In terms of reflection coefficients, the necessry conditions for unconditionl stility t given frequency re Γ <, (3.5) Γ L < (3.6) Γ IN Γ L + < Γ L, (3.7) Γ OUT Γ + < Γ. (3.8) Γ OUT Γ L Γ Γ IN Z V Two-port network Z L Z IN Z OUT Figure 3.3 tility of two-port network

7 3 The necessry nd sufficient conditions for two-port network to e unconditionl stile re [D. Woods, 976] K + >, (3.9) <. (3.0) In prctice, most of the microwve trnsistor mplifiers re potentilly unstle ecuse of the internl feedck. There re two wys to overcome the stility prolem of the trnsistor mplifier. The first is to use some form of feedck to stilize the mplifier. The second is to use grphicl nlysis to determine the regions where the vlues of Γ nd Γ L (source nd lod reflection coefficients) re less thn one, which mens the rel prts of Z IN nd Z OUT re positive. ustituting the vlues of Γ IN nd Γ OUT in equtions (3.7) nd (3.8) nd solving for Γ nd Γ L result in the stility circles. The rdii nd centers of the circles re given y [G. Gonzlez, 984] Output tility Circle, r L (3.) C L ( * ) *

8 3 And input tility Circle, r L (3.) C L ( * ) *. Then the stility circles need to e plotted in the mith chrt to determine the stle regions or in other words, the regions where vlues of Γ nd Γ L produce Γ OUT < nd Γ IN <. Most of the time, microwve mplifiers used for nrrownd or widend pplictions fce stility prolems t certin frequency rnges. Instility is primrily cused y three phenomen: internl feedck round the trnsistor, externl feedck round the trnsistor cused y n externl circuit, or excess of gin t frequencies outside of the nd of opertion. Figure 3.4 shows six pssive feedck-networks tht re used usully to stilize mplifier circuits.

9 33 () () (c) (d) (e) (f) Figure.3.4 Pssive-networks stiliztion

10 34 Amplifiers usully need mtching networks to chieve the desired gols. The input-mtching network of the low noise mplifier (LNA) is designed to trnsform the 50 Ω impednce of the preceding stge to the impednce required to chieve the minimum noise figure, nd the output-mtching network is designed to chieve high power gin. In the high output power mplifier, the input-mtching network presents conjugte impednce to the input impednce of the mplifier, nd the output-mtching network is designed to chieve the desired output power. The input nd output mtching networks of the high gin mplifier re designed to e conjuglly mtched to the input nd output impednce of the mplifier. Mtching networks cn e implemented using lumped elements, distriuted elements (trnsmission lines), or comintion of lumped nd distriuted elements. The lumped elements cn e used s long s their dimensions re much smller thn the electricl wvelength. With modern microwve integrted circuits, the lumped elements cn e used up to 60 GHz. The ZY mith chrt cn e used conveniently in the design of mtching networks. The process of choosing the source nd lod impednces for the high gin mplifier nd LNA is well estlished using the smll-signl -prmeters [G. Gonzlez, 984, H, 98].

11 35 3. High Power Amplifier Design In the cse of the power mplifier, the input signl level is often high, nd consequently the output current is either in the cutoff or sturtion region during portion of the input signl cycle. The smll signl -prmeters cn e used if the lrge-signl mplifier is operting in clss-a. However, for other clsses (B, C, etc), the smll signl -prmeters re of little use, nd it ecomes necessry to use other techniques. The design procedures of the high power mplifier re similr to those of the smll signl mplifier. The min difference is tht the high power mplifier uses other techniques to otin the proper impednces t the input nd output ports of the mplifier. The min techniques used in the high power mplifier design re conventionl lod-pull, ctive lod-pull, nd two-port lrge signl chrcteriztion. 3.. Lod-pull Techniques The conventionl nd ctive lod-pull techniques hve the sme concept of opertion; they provide informtion for the source nd lod reflection coefficients s function of output power nd gin. Figure 3.5 illustrtes the sic concept of the lodpull techniques. The trnsistor under test is plced in mesuring setup where the dc is nd c input signl re fixed. The output tuner is djusted until the power meter C mesures given power level nd the input tuner is djusted for zero reflected power (red t power meter B). The power meter A reds the incident power, nd the power gin cn e otined. From this informtion gin, power dded efficiency (PAE), nd output power contours cn e generted nd drwn in the mith chrt s function of the output lod. While the conventionl lod-pull technique uses pssive tuner (tuning stus) to vry the output lod, the ctive lod-pull technique vries the lod ctively y injecting power wve with vrile mgnitude nd phse towrd the trnsistor output. The lod

12 36 vrition cn tke plce t the fundmentl frequency f o lone or t the fundmentl frequency nd numer of the hrmonics (multihrmonic lod-pull). The hrmonic lod impednce ffects the performnce of some opertionl clsses (e.g., clsses-c) [. R. Mzumder, A. Azizi, F. E. Grdiol, 979]. A pssive tuner is commercilly ville, cn hndle high powers, is esy to use, nd is of comprtively low cost. Despite these dvntges, it hs inherent losses tht impose limittions in the reflection-coefficient mgnitude. The mximum mgnitude decreses with the frequency nd with the numer of elements nd cles connected etween the mesurement pln nd tuner. The ctive lod-pull principle does not suffer from limittions in the lod reflection-coefficient mgnitude. Two sic principles of lod vrition re commonly used: either the signl injected into the device output is synchronized with the one pplied to the input [Tkym, 976] or the signl generted y the device itself is fed ck to its output with vrile phse nd mgnitude [G. P. Bv, U. Pisni, nd V. Pozzolo, 98].

13 37 Meter A Meter B ignl genertor Input tuner Directionl Coupler Trnsistor under test Lod Output tuner Directionl Coupler Meter C Figure 3.5 Two-port lod-pull mesurement system

14 Two-Port Lrge ignl Techniques The two-port lrge signl chrcteriztion is siclly mking -prmeter mesurements with respect to 50Ω reference impednces. There re two min methods: the direct extension of smll-signl mesurement to lrge signl [W. Leighton, R. Chffin, nd J. We, 973], nd qusi-lrge-signl pproch wherey the lrge-signl -prmeters re mesured y simultneous ppliction of two coherent signls t the sme frequency to the input nd output of the device [P.D. vn, nd.r. Mzumder, 978]. Although it is not useful with nonliner mplifier, the ovious dvntge of the former method is its simplicity. Although the -prmeters of nonliner mplifier cn e otined using the qusi-lrge-signl method, they re of little use if the non-linerity is severe. Equtions (3.) nd (3.) cn e rewritten in the form: + (3.3) + (3.4) + (3.5) +. (3.6)

15 39 In the qusi-lrge-signl technique, nd re the two pplied signls. The four - prmeters cn e determined for different vlues of nd. Once the -prmeters of the device re otined, the next step is to pply the smll signl -prmeter techniques [G. Gonzlez, 984]..