IEEE MTT-V032-I03 (1984-03)

118 Pages·1984·15.613 MB·English

by  FANTOMASPING

//-,
~i~,p’,
IEEE MI CROVv’AVE THEORY AA-D TECHNIQUES SOCIETY
~w
@ J :
Ill
ACENTURY OFELECTRICAL PROGRESS
The M]crowave Theory and Techniques Society ISan organization, wlthln the framework of the IEEE. of members with prtnclpal professiorml interest in
the field of microwave theory and techniques. All members of the IEEE are eligible for membership In the Society and WIII receive this TRANSA.CTiO\S
upon payment of the annual Society membership feeof $8,00, Affii]ate membership ISavailable upon payment of the annual afflllate fee of $2200. P!US
the Society fee of $8.00. For information onjoining write to the IIEEE at the addressbelow
ADMIrWISTRATIVE COMMITTEE
H. G.OLTMAN, JR.,Prestdent H. HOWE, JR., Vice President J.E.RAUE, Secretary-Treasurer
N. ‘IN.COX T. ITOH H, J.KUNO F.J.ROSENBAUM*
J.E.DEGENFORD,JR. F.IVANEK S.L. MARCH C.T. RUCKER*
V. G.GELNOVATCH G.JERINIC D, N. MCQUIDDY, JR R. A. SPARKS*
P.T. GREILING IL KAGIWADA E.C. NIEHENKE B. E,SPIELMAN
R. B.HICKS R. H, KNERR J,IM.ROE
* E.r officio (pa~rpresidents ) HonorarJ, Life Members Dlstingu/ rhed Lecturers
A. C. BECK D. D. KING A A. OLINER K. TOM IYASU J. A GIORDMAINE
S. B. COHN W. W. MU MFORD T. S. SAAD L. YOUNG S. ADAM
S..MTT Chapter Chairmen
Albuquerque: R. L. GARDNER Houston S,LONG Phll~delphm. C. C.,\LLEN
Atlanta: G.K. HUDDLESTON Huntswlle INACTIVE Phoenix LEX AKERS
Baltimore: PETERD. HRYCAK India. B.BHAT Portland. INACTIVE
Benelux: A. GUKSSARD Israel A MADJAR Prrnceton WALTER SLUSARK
Boston: CARL D. BERGLUND Kltchener-Waterloo, Y. L, CHOW San Diego. J.H. ZICKGAF
Boulder/Denver: C.T. JOHNK Los Angeles: F J BERNUES Santa Clara Valley: P T HO
Buffalo: INACTIVE Milwaukee C J. KOTLARZ Schenectady. J.BORREGO
Canaveral: G.G,RASSWEILER Montreal: J. L. LEIZEROWICZ Seattle C. K. CHO1
Central Illinois: G. E.STILLMAN New JerseyCoast: RUSSELL A GILSON Southeastern Michigan. P.[. PRESSEL
Chicago: S.S.SAAD New York/Long Island: J HAUSNER St. Louis. CURTIS E LARSON
Columbus: N. WANG North Jersey, M SCHNEIDER Syracuse B.K. MITCHELL
Connecticut. INACTIVE Orange County: INACTIVE Tokyo. T. OKOSHI
Dallas: R. E.LEHMANN Orlando: F.F’.WILCOX Tucson. IhACTIVE
Florida West Coast: R. E,HENNING Ottawa: J. WIGHT Washington, DC: J. H DOtJGLAS
IEEE TRANSACTIONS@ ON MICROWAVE THEORY AND TECHNIQUES
Editor Associate Editors
T. ITOH N. R. DIETRICH F. lVANEK E YAMASHITA
(Patent Abstracts) (Abstracts Edilor–Asia)
Address all manuscripts tothe Editor, T Itoh, Dept. of Electrical Englneerlng. University ofTexasatAust]rr, Aust]n, TX 78712. Submission offour copies
of manuscripts. includlng figures, will expedite therewew. Pub//ca/iorr Policy. All papersw]II bereviewed for their technical merit, anddecisionstopublish
will bemadeindependently ofan,author’s ablllty orwillingness topaycharges.Voluntary pagechargesof$95perprinted pagewill berequestedfor papersof
five pagesorless.PagechargesofS100perpagearemandatory for eachme inexcessof five printed pages,Also, the MTT-S Administrate Committee
hasestablished aquota for the number of pagesprinted Ineachissueof this TRANSACTIONS whosecostsare not defrayed by payment of pagecharges.
Papersnot covered by page charges maybe delayed until space Inan Issue isavailtile, The Editor can waive the quota requirement forexceptional papersor
becauseof other extenuating circumstances.
THE INSTITUTE OF ELECTRICAL AND ELECTRONICS ENGINEERS, INC.
Officers
RICHARD J.GOWEN, President J.BARRVOAKES, Vice President, Educational Actiuilies
DONALD D. KING, President-Elect RUSSEL C. DREW, Vice President, Professional Activities
HENRY L. BACHMAN, Executive VccePresident JOSEB,CRUZ, JR., Vice President, Publication Activities
CYRIL J.TUNIS, Treasurer MERRILt. W. BUCKLEY, JR., Vice President, Regional Actil,ities
V. PRASAD KODAL1, Secretary STEPHAN KAHNE, Vice President, Technical Activities
EIVrERSON W. PUGH, DIL1siOn JV Direr{or
Headquarters Staff
ERIC HERZ, Executive D[recror and General Manager
ELWOOD K, GANNETT, Deputy General Manager
THOMAS W. BARTLETT, Controller DAVID L. STAIGER, Staff Director, Publlshing Seroices
DONALD CHRISTIANSEN, Editor of Spectrum CHARLES F. STEWART, JR., Staff Director, Administration Seroices
1RVING ENGELSON, Staff Director, Technical Activities DONALD L SUPPERS, Staff Director, Field Services
LEO FANNING, Staff Director, Professional Act!tiilies THOMAS C WHITE, Staff Director, Public Information
SAVA SHERR,S~a~~Director, Standards Joi+ N F. W1LHi3_kI, Slaff Director, Professional Acllcities
Publications Department
Production Managers: ANN H. BIJRGMEYER, CAROLYNE ELENOWITZ, GAIL S. FERENC
Associate Editor: WILLIAM J.HAG EN
IEEE TRANSACTIONS ON MICROWAVE THEORY AND TECHNIQUES ispublished monthly byThe Institute of Electrical and Electronics Engineers,
Inc. Headquarters: 345 East47Street, New York, NY 10017. Rtisponsibility for thecontents restsupon the authors andnot upon the IEEE, the Society,
or itsmembers. IEEE ServiceCenter (for orders,subscriptions. addresschanges,Region/SectIon/Student Services):445HoesLane, Piscataway, NJ 08854.
Telephones: Headquarters 212-705 + extension: Information -7900, General Manager -7910, Controller -7748, Educational Services -7860, Publishing
Services -7560, Standards -7960, Techmcal Services-7890, IEEE Service Center 201-981-0060 Professional Services:Washington Office 202-785-00 i7.
NY Telecopie~ 212-752-4929. Telex 236-411 (International messagesonly). Individual copies:IEEE members $6.00 (first copyonly), nonmembers $12.00
percopy.Annual subscription price:IEEE members,duesplusSocietyfee.Priceforno~membersonrequest.Available inmicrofiche andmicrofilm. Copyright
andReprint Permission: Abstracting ispermitted with credit to thesource.Libraries arepermitted tophotocopy beyond the limits of U.S. Copyright law
for private useofpatrons:(1) thosepost-1977 articles that carry acodeatthebottom ofthefirst page,provided theper-copy feeindicated inthecodeISpaid
~hrough theCopyright Clearance Center, 29CongressStreet. Salem, MA 01970; (2) pre-1978 articles without fee.Instructors arepermitted tophotocopy
]solated articles for noncommercial classroomusewithout fee.Forother copying. reprint orrepubhcation permission, write toDirector, Publishing Services
at IEEE Headquarters. All rights reserved.Copyright @1984byIrhe Institute ofElectr]cal andElectronics Engineers, Inc. Printed inU.S.A. Second-class
postagepaid atNew York, NY andat additional mailing offices, lPostrmaste~SendAddress changestoIEEE, 445 HoesLane, Piscataway, NJ 08854.
IEEETRANSACTIONSONMICROWAVSTHEORYAND TECHNIQUES, VOL.MTT-32,NO.3,MARCH1984 225
Foreword
P ONSIDERABLE progress has been made over the nal modeling are the subject of the other three papers. In
~ past several years in power and low-noise GaAs FET’s the topic on power amplifiers, special power combining
and circuits. Power FET’s with output powers of up to techniques are utilized to achieve high output powers at C-
30 W at S– C bands and a few hundred milliwatts at CanalK-band for communication applications.
K-band have been achieved. Ultra-low-noise FET’s operat- The subjects covered in this Special Issue reflect the
ing at 20 GHz and beyond have been reported. Increasing current technological trend in. GaAs FET’s. We hope that
interest in GaAs monolithic IC technology has also stimu- this Speeial Issue will provide important teehnical informa-
lated the steady improvements in discrete power and low- tion for stimulating further development in GaAs FET
noise FET performance. Special circuit techniques such as technology into the late 1980’s and beyond. We sincerely
large-signal characterization, modeling, and power combin- appreciate the help of the following reviewers for selecting
ing have become increasingly important for realizing the the best papers for this Special Issue.
ultimate performance potentials of this important and
versatile solid-state microwave device.
Y. Ayasli H. Macksey P. Saunier
This Special Issue covers current developments in power
R. Coats R. Minasian W, Schroeder
and low-noise GaAs FET circuit technology and applica-
J. Goel K. Niclaus F. Sechi
tions. Specifically, those technological areas relating to
J. Higgins S.Perlow V. Sokolov
low-noise, and power amplifiers, dual-gate devices, and
H. Huang W. Peterson Y. Tajima
broad-band amplifiers are covered. Characterization and
B. Kim D. Poulin H. Willing
performance of low-noise FET’s are presented in three
W. Ku R. Pucel H. Yamasaki
papers on low-noise amplifiers. Three papers are devoted
R. Lehmann C. Rauscher B. Yarman
to the subject of dual-gate FET’s with emphasis on model-
ing, mixer application, and power performance. Five papers
cover the important issue of broad-band amplifier design HUA QUEN TSERNG
and performance. Of these five papers, two are devoted to , CHARLES C. HUANG
the distributed amplifiers. Design technique and large-sig- GuestEditors
Hua Quen Tserng (M70–SM83) received the B.S. degree in electrical engineering from
National Taiwan University, Taipei, Taiwan, in 1962, and the M.S. and Ph.D. degrees in
electrical engineering from Rice University, Houston, TX, in 1966 and 1968, respectively.
He joined the Central Research Laboratories of Texas Instruments Incorporated in
Dallas, TX, in 1968. From 1964 to 1968, at Rice University, he was engaged in research
work on transport phenomena in semiconductors and optimization of thermoelectric
power generators and refrigerators. From 1968 to 1969, he carried out work on thermal
physics and characterization of semiconductor devices, including failure analysis and
temperature-dependent properties of semiconductor devices. From 1969 to 1975, he
worked on GaAs IMPATT diodes for high-power, high-efficiency rnicrostrip oscillator
and amplifier applications. Since 1975, he has been responsible for the development of
microstrip and monolithic GaAs power FET amplifiers and oscillators at TL His work has
appeared in a number of scientific publications.
Charles C. Huang received the B.S. degree in 1969 from National Taiwan University and
the M.S. degree in 1971 from the University of Alabama, Tuscaloosa, in electrical
engineering. In 1975, he received the Ph.D. degree in electrical engineering and computer
science from the University of California, Berkeley.
From 1975 to 1980, he was a member of the technical staff at Hewlett-Packard
Company, San Jose, CA, where he was engaged ip the development of sub-micron GaAs
FET’s. Since 1980, he has been employed at Avantek, Inc., Santa Clara, CA, where he is
presently Manager of GaAs FET device development. As such, he is responsible for the
design and development of all new gallium arsenide FET’s and monolithic IC’S.
Dr. Huang is a member of Eta Kappa Nu.
226 IEEETRANSACTIONSONMICROWAVETHEORYANDTECHNIQUES,VOL.MTT-32,NO.3,MARCH1984
A 22–24-GHZ Cryogenically Cooled GaAs
FET Amplifier
ANTHONY CAPPELLOI, MEMBER,IEEE,AND JOHN PIERRO, MEMBER, IEEE
Abstract —This paper describes the desig]fland performance of acryo- -i”t--’-i”t-,.
genically cooled low-noise FET amplifier operating in the 22-24-GHz
range. The amplifier employs five cascadedsingle-ended gain stagesandan
integraf bandpass filter. Noise temperatures in the 200 K range with an
associated gain of 2ffdB aretypical for the nine cooled units built to date.
Fig. 1. Coplanar waveguide.
I. lNTRCIDUCT [ON
ground planes by the equal gaps (W) forms the waveguid-
T HE USE OF cryogenic refrigeration to reduce the
ing structure. The characteristic impedance of a transmis-
noise temperature of C&As FET amplifiers has been
sion line in this configuration is proportional to the aspect
shown to be quite beneficial. Noise temperatures close to
ratio S/(S + 2W).
or below those achievable with parametric amplifiers are
believed to be possible [1]. Results have been reported at
II. TRANSISTOR CHARACTERIZATION
lower frequencies [2], [3]. This is the first K-band amplifier
reported that employs cryogenic cooling. Since K-band S-parameter data was not available for the
The design relies on an electrica I model derived from a Mitsubishi MGFC-1403 transistor, S-parameter measure-
physical model of the device. S-parameter measurements ments were a critical first step in the design. Direct mea-
made at lower microwave frequencies were used to obtain surement of S-parameters at K-band frequencies could not
element values for the device model. This model enabled be done because of equipment limitations. Therefore,
the design of a cascadable amplifier stage with gain over 4 extrapolation of data measured at lower frequencies was
dB and noise figure under 4.6 dB at room temperature. considered.
The design prerequisites include the following: In order to reliably extrapolate such data, a comprehen-
sive model was created for the transistor. This was done by
1) selection of a guiding structure, and
mounting the FET in a coplanar 50-fl system and measur-
2) device characterization.
ing the S-parameters of the device. Automatic network
The guiding structure selected for the hybridl amplifier analyzer measurements provided reliable, de-embedded de-
and filter circuits is coplanar waveguide. This structure was vice data from 2–15 GHz.
chosen over microstrip for the following reasons [4], [5]. This information was then entered in a computer file
containing the circuit model shown in Fig. 2. The program
1) It allows the circuit designer to realize both low- and
varied several key elements in the file striving to make the
high-impedance transmission lines without the need for
circuit analysis match l:he measured S-parameters. After
excessively wide or narrow conductor strips.
optimization, the model tracked the measured S-parameter
2) Both series and shunt elements can be realized easily.
data quite closely. The model was then analyzed up to 30
3) Parasitic source grounding inductance can be mini-
GHz, yielding reliable S-parameter data for circuit design.
mized since the need for wraparound grounding ribbons or
To verify the validity of the device model, slotted-line
via holes is eliminated.
measurements of Sll and S22 were performed at 22–24
4) Coplanar waveguide is less likely to propagate spuri-
GHz. These measurements agreed quite closely with the
ous modes than rnicrostrip. This feature, along with the
predicted Sll and S22.
ability to minimize source inductance, enables one to
achieve circuits with high reverse, isolation (Slz). This is III. NOISE MODEL
essential for a cascadable amplifier stage.
After the device model was obtained, it was analyzed to
A cross-sectional view of the guiding structure is shown predict the optimum source admittance for the minimum
in Fig. 1. The outside conductor strips are electrically noise figure. This was done by creating a simplified noise
grounded. The center strip which is separated from the model from the device model already obtained. The model
is shown in Fig. 3, and the corresponding equations are
Manuscript received August 11,1983; rewsed December 19,1983.
The authors are with the Eaton Corporation AIL Division, Commack (1)
Road, Deer Park, NY 11729.
0018 -9480/’84/0300-0226$01 .00 01984 IEEE
CAPPELLOANDPIERRO:CRYOGENICALLYCOOLEDGaAs FETAMPLIFIER 227
A&u—o
o
TL 2
&Q
-in
r7-
Rz
10
rl 0.15nH 0.15nH 10rL
*0.4 PF$WPFVD
“ rpF20”4pF
Fig. 4. Amplifier circuit.
o 0
Fig. 2. Transistor model. TABLE I
TRANSISTORPARAMETERS
Fre-
quency
(GHz) 511 521 512 522 ropt
22 0.63~ 1.15&~ o.137~ o.35~
23 0.64,Q5&0 l.lo~ o.141fi 0.36~ .63~
Fig. 3. Noise model.
24 0.65~ 1.06~ o.145~ 0.36.&l&
and
TABLE II
Q:cgs AMPLIFIER-CIRCUITANALYSIS
“= (2)
(Q; +l)
szl(dB)
R1=Rg+Ri+Rz (3) Fquree-ncy NFoliscejure
(GHz) S~~ 521 512 522 I (dB)
and
22 o.33&J 1.60~ o.19~ o.19pJ& 4.06 --
I30”
“ = (4) 23 0.23@ 1.61= o.zl~ o.07@lQ I 4.13
2Tfc;1cg. “ 24 0.32~ 1.51= o.zl~ “14* I 355 I --
The values Rg, R,, Rz, and Cg~,needed to calculate RI and
Ql, are element values obtained from the transistor model were used to design input and output matching networks.
shown in Fig. 2. Once values gl and Cl were obtained, the Computer optimization was then used to obtain minimum
optimum source admittance (Y,OPt= gJOPt+ @,Opt) was noise figure and flat gain. During a subsequent out-of-band
calculated using (5) and (6) analysis, it was found that a gain peak existed at around 6
GHz. A decoupling network was incorporated in the gate
(5) and drain bias circuits to reduce the gain peak. In order to
achieve additional rejection at 6 GHz, a bandpass filter
- Q~ ~c, module was designed. The filter is cascaded with the
b –c;= — (6)
Sopt= amplifier modules in the complete assembly. The final,
()Q;+l “
optimized circuit is shown in Fig. 4 and results of the
The quantity A is derived from the minimum attainable circuit analysis appear in Table II. The noise-figure analy-
noise figure (Fti ) of the device at the frequency of interest sis at 23 GHz is included in this table.
through (7)
V; CONSTRUCTION
A= (Fti-1)2
(7) The amplifier circuit was built with distributed and
4Fti “
lumped elements. The capacitors selected were low-loss
The derivations for (l)–(7) can be found along with a parallel plate types. Required inductances were achieved
complete description of the noise modeling technique in with bond wires or ribbons. Tran,smission lines and resis-
[6]. The Fti at 23 GHz was estimated to be 3.75 dB by tors were photoetched on a 0.015-in Au–Cr-deposited
extrapolating the manufacturer’s data. alumina substrate. The chrome adhesion layer provided
adequate sheet resistance for the thin-film resistors. Fig. 5
IV. CIRCUIT DESIGN
is aphotograph of an assembled amplifier stage.
A computer analysis of the circuit in Fig. 2 yielded the Kovar was chosen as the carrier and amplifier housing
device S-parameters shown in Table I. Included in the material because of its excellent thermal stability and close
table is the optimum generator reflection coefficient for compatibility with alumina over the wide temperature
the minimum noise figure at 23 GHz. This was found from range. Kovar also lends itself to electron-beam welding,
the noise model already described. Smith Chart techniques which is used to hermetically seal the amplifier.
228 IEEETRANSACTIONSONMICROWAVETHEORYANDTECHNIQUES,VOL.MTT-32,NO.3,MARCH1984
~~
22 23 24
FREOUENCY (GHz)
Fig. 7. Single-stage noise figure.
1
6
~o
%
0;5 T
K — — e y
3
L 10
E
i . ~~~~, ~~~~
22 24
FREQUENCY (GHz)
Fig. 8. Single-stage input return loss.
s
=0
30
Fig. 5. Assembled amplifier stage. ; 10
K
3
L 20 —
u
t
4 L / ~ ‘ \ / ~ % . R; 22 T - ‘; :24 ‘:
~ FREQUENCY (GHz)
Z3 T Fig. 9. Single-stage output return loss.
<
c1
2
JJJJ I111 I111 111 I I,I! 11!i ,,II 1I,! ,,II I,11
22 24
FREQUENCY (GHz)
Fig. 6. Measured single-stage gain. 6PJP:}IH
,SOLATOR CCNNECTOR STAGE I sTAGE 2 STAGE 3 LINE STAGE d ST.GE 5 ~OTAL
:~7NMW ,,,, ,,, .8 .40.8 .40d6 ,40dB 05dB .40dB ‘40dB lS75.8
VI. SINGLE-STAGE PERFORMANCE N(REOTOMG.,.,, M,, 0“ 0’4 25 25 2S 089 25 25 1,28,s ,8
.,0,o,OsM, ,,,,, M,, 358 185 5460 5460 5460 358 5460 5660
Gain and return loss were measured using standard C.0O1sN6TRIBUTION 358 20’ ’52 ‘ 2’10 ‘M’ 2‘ 469 ’88 II*W
swept measurement techniques. The noise figure was (ROOM ,,. ?,
.:~lK~!N ,189 ,)9, 3S 354 354 089 354 3% 4,3g.
measured using an extremely low-noise parametric ampli- 262 d,
..y,cp: (., ’35 4’3 1092 1C92 7G92 95 lmz l@32
fier as a second stage to minimize measurement uncer-
tainty. The AIL-developed parametric amplifier provided a :.,0y(;s,., (who. 95 55 1035 522 209 07 94 38 2325.
noise figure of 2.6 and 28 dB of gain at 23 GHz, at 290 K .GAAPIPNROXAITMA7T7ELKYISE1S5TIMdA8TEHDIGHEFRROMTHEPNASTTHEEXPREOROIEMNCTEEMP7E0RAwTuRE
GAIN
..No,*, ,EM, ERATURC ,SEs,,MA,,. FROM ,.s, EXPER,EWE ,0 SE
ambient temperature. 20%OFTHE ROOM TEMPERATURE NOISE TEMPERATURE
Fig. 6 shows the measured gain versus frequency for a Fig. 10. Cascadedgain andnoise-temperature prediction.
typical amplifier stage with corrections for test fixture
losses. Connector losses were estimated to be 0.25 dB each
at 23 GHz. Fig. 7 shows the measured noise figure of the line module. The through line is replaced with a sixth gain
stage with corrections made for connector loss and stage on units intended for room-temperature use only.
second-stage contribution. Figs. 8 and 9 are the measured Fig. 10 shows the block diagram and accompanying gain
input and output return losses in the test fixture. Reverse and noise-temperature budget for the cascaded amplifier.
isolation measured was in excess of 17 dB over the 22–24- External isolators are placed at the input and output ports
GHz band. of the amplifier, although only the input isolator was
included in the noise-temperature budget.
The performance indicated in Fig. 10 is for a single
VII. CASCADED-AMPLIFIER PREDICTED PERFORMANCE midband frequency. A single-stage unit with a midband
The complete amplifier consists of five FET modules gain of 4 dB and noise figure of 4.6 dB at room tempera-
cascaded with one bandpass filter module, and a through- ture was assumed for this calculation. The through-line loss
CAPPELLOAND PIERRO:CRYOGENICALLY COOLED (h% FETAMPLIFIER 229
L=0.2dB@3Cil K ~————— - —_________ q
,1.7..A.-,.-”,’? E1.S!WITCH h ! La ‘3
~,
Tc .
... . ,.” . .“. . I
Fig. 12. Cryogenic measurement setup.
‘“~
29
28 n
i~i (
Z 27 /
<
c1 1
26 I
25
24
22.0 23,o 24,0
FREOUENCY (GHz)
Fig. 11. Integrated amplifier assembly. Fig. 13. Amplifier gain at Tmbient= 77 K.
was neglected in this estimate. Including isolator loss, the IX. CRYOGENIC AMPLIFIER MEASUREMENTS
predicted gain is 26.2 dB, and the predicted noise tempera-
Gain and noise-temperature measurements were made
ture is 232 K for the cryogenically cooled amplifier.
using a CTI Refrigerator Model No. 102OR, The block
diagram in Fig. 12 describes the measurement setup. The
VIII. INTEGRATED AMPLIFIER CONSTRUCTION area inside the dashed line represents the vacuum ves-
sel which has waveguide input/output ports. Separate
The complete amplifier is housed in a Kovar steel chassis
slotted-line measurements were made on the sections
as shown in Fig. 11. As it appears in the photograph, the
labeled LI through Lb to determine their individual losses.
top section contains the cascaded gain and filter stages,
The diagram shows the physical temperature at which each
located beneath a subcover which provides continuous
section was maintained during the cryogenic measure-
waveguide walls from input to output. This prevents un-
ments. The 189 K temperature assigned to Lz and L5 is the
wanted spurious responses and enhances reverse isolation.
estimated average temperature of the stainless steel wave-
The module occupying the lower section of the chassis is a
guide sections, since they sustained a temperature gradient
bias assembly. This circuit provides de-overvoltage protec-
along their lengths. The reference for the gain measure-
tion, as well as individual device bias adjustment via the
ment was taken outside the vessel (essentially at the input
terminals along the lower edge of the assembly. The bias
to LI ), and the gain measurement was corrected for the
terminals located on the same surface as the output
losses listed in Fig. 12. The noise temperature was measured
connector are for positive and negative supply voltage. The
using standard Hot/Cold Y-Factor measurement tech-
unit features an electron-beam welded top cover for a
niques and corrected for the losses described in Fig. 12,
hermetically sealed package and field-replaceable SMA
connectors.
X. DISCUSSION OF MEASURED AMPLIFIER
Special consideration was given to the mechanical design
PERFORMANCE
because of the extremely low-temperature environment in
which the amplifier must operate. Kovar steel was chosen Figs. 13 and 14 show the measured gain and noise
for the housing material as well as for the individual temperature for a typical unit at an ambient temperature of
module carriers. The welded-in connector shells which 77 K. The noise temperature climbs rapidly at the band
make up the fixed part of the two-piece connectors are also edges. This is to be expected in a narrow-band design. The
made of Kovar Steel. Kovar is an alloy that is thermally increase in gain due to reducing the physical temperature
stable and has a temperature coefficient that closely tracks of the unit is seen in Fig. 15. This unit exhibits an average
the temperature coefficient of alumina. Tolerancing was increase of approximately 8-dB gain relative to the room-
kept very close because of the high frequency of operation, temperature gain. This is a little better than our 1.5-dB per
but some strain relief was allowed on gold ribbons connect- stage prediction. A room-temperature noise figure is not
ing adjacent stages. plotted here, but it was measured. This unit averaged
230 IEEETRANSACTIONS ONMICROWAVE THEORYAND TECHMQUES, VOL. MTT-32,NO.3,MARCH1984
iii sensitive receiving systems that were built for the Tokyo
g 500 -
> Astronomical Observatory.
i
EIll 400 . The authors would also like to thank the following AIL
E
2 Division staff members: C. Booth and A. Kunze for sub-
; 300
U strate fabrication, D. Broadhurst and A. Rees for MIC
: P
~ 200 assembly work, and F. Winter for his assistance in the
1- I i
. computer analysis and substrate layouts, R. Niebling was
g 100
z 21 22 23 24 25 responsible for the fabrication drawings. B. Reinheimer
FRECOJENCY (GHz) was responsible fbr testing and alignment of the 9 units
Fig. 14. Amplifier nOiSetemperature at T~bi~n~= 77K.
shipped to date.
30 1. REFERENCES
4 - - ~
[1] S.Wienreb, “Low noise technology, 1982 state of the m%” in 1982
25
/ Ml’-z’-s Irrt. Symp. Dig., pp. 10,11.
L ~ [2] J. PierTo, “Cryogenically cooled GGASFET amplifier with a noise
20
/ . . temperature under 70 K al. 5.0 GHz,” IEEE Trans. Microwave
I~Z .~ ‘i’YreoryTech., vol. MTT-24, p. 972, Dec. 1976.
z 15 i+++ -+ti+ ++++ * [3] J. Pierro and K. Louie, “Low temperature performance of GRAS
z
cl MESFETS at L-band,” in 1979MTT- S Int. Symp. Dig.
10
[4] K. C. Gupta, R. Garg, and I. J.Bahl, Microstrip Lines and Slotlines.
\ Dedham, MA: Artech House, 1979,pp. 257-302.
5 - A - ROOM TEMPERATURE :: \ [5] J. B. Knorr and K.-D. Duchler, “Analysis of coupled slots and
—77K \\ coplanar strips on dielectric substrate,” IEEE Trans. Microwave
o.21 22 23 24 25 Theo~ Tech., vol. MTT-23, pp. 541-547, Jnfy 1975.
FREOUENCY (GHz) [6] A. F. Podell, “A functional GGASFET noise model,” IEEE Trans.
Fig, 15. Amplifier gain versusfrequency and temperature. Electron Devices, vol. ED-28, pp. 511-517, May 1981.
approximately 5.5 dB, or equivalently 1030 K, at room *
ambient. This also agrees quite closely with the prediction
shown in Fig. 10.
XI. CONCLUSION Anthony Cappello (M82) was born in Rocktille
Centre, NTY,on June 16, 1958. He received the
Measured data confirms that K-Band GaAs FET ampli- Bachelor of Sciencedegreein electrical engineer-
fiers with noise temperatures approaching those previously ing from the Polytechnic Institute of New York
in January of 1980. He is currently pursuing an
achievable only with parametric amplifiers are possible.
M.S.E.E. degreeat the sameinstitution.
This performance is a direct result of cryogenically cooling He has worked at the ATL Division in the
the sealed amplifier. Coplanar waveguide has been shown Solid-State and Microwave Subsystems Section
of the Advanced Technology Division since be-
to be a useful and advantageous circuit medium for use at
ginning his career in 1980. He has been respon-
high microwave frequencies. sible for the design and develo~ment of hifi-
it should be noted that this design employs 0.5-~m gate performance low-noise transistor amplifi~rs for militz@ and scient~fic
applications. His present work includes design and development work on
length devices which can no longer be considered state-of- microwave oscillators, filters, and switches, aswell asmonolithic microw-
the-art. The new 0.25-pm device technology will un- ave integrated circuits.
doubtedly result in extremely low-noise temperatures at
K-band frequencies and higher. It should be noted also
that this design exhibits ample performance margin at
*
77 K ambient for the application for which it was devel-
oped. Cooling to lower temperatures (20 K or less) is
possible.
Lower temperature cooling, and the use of 0.25-pm FET John Pierro (M78) received the B.E. (E.E.) de-
greefrom City College of New York in 1967 and
devices, will probably result in low-noise transistor ampli-
the M.S.13.E.(system science) degree from the
fiers operating at K-band with noise temperatures in the Polytechnic Institute of New York in 1978.
1OO-K range. At the present, Eaton Corporation AIL Divi- Hejoined theAIL Ditision of Cutler–Hammer
(now Eaton Corporation) in 1967 asanengineer.
sion is investigating these ideas. The AIL Division hopes to
He presently is a Section Head in the Receiver
publish favorable results in the near future. Systemsand Technology Department of the Ad-
vanced Technology Systems Division. For the
ACKNOWLEDGMENT past eight years, he hasbeen responsible for the
development of low-noise GaAs FET rmmlifiers
This work was performed at Eaton Corporation AIL for critical receiver appfication~. He hasauthored and coauthored ~everaf
Division in the Advanced Technology Systems Division papers on this work. Previous to this, hewasresponsible for the design of
passive radiometry systemsfor earth-resources experiments and avariety
under B. J. Peyton, Director, and J. Whelehan, Department
of anoJog components, including discriminators and logarithmic ampli-
Head. The amplifier was developed for a group of ultra- fiers.
IEEETRANSACTIONS ON MICROWAVE THEORYAND TECHNIQUES, VOL. MTT-32,NO.3,MARCH1984 231
Characterization of GaAs FET’s in Terms of
Noise, Gain, and Scattering Parameters
Through a Noise Parameter Test Set
ENRICO F. CALANDRA, MEMBER,IEEE,GIOVANNI MARTINES, AND MARIO SANNINO
Abstract —A method for the complete characterization of GaAs FET’s making use of the well-known Friis formula
in terms of noiseparameters (F’, rO., R.), gainparameters (Ga., r.g, Rg),
and of those scattering parameters (SII, SZZ,ISIZ1,I% l,~lz %) mat we q.(s’’2)-l
Fm(r.)=F(r$)+ (2)
needed for low-noise microwave amplifier design is presented. The instm-
Ga(rs)
mentatfon employed, i.e., anoise-figure measuring systemequipped with a
vectoriaf reffectometer, as well as the time consumption, are the same where Fm(r, ) is the measured noise figure, G.(r, ) is the
required for the determination of noise parameters only through conven- DUT available power gain, and F,(Sjz ) is the noise figure
tional methods. The measuring setup and the experimental procedure are
of the measuring stages when input is terminated on the
described in detail. Considerations about the computer-aided dataprocess-
ing technique are also provided. As an experimental result, the char- DUT output reflection coefficient S~z(r, ). If an output
acterization of a sample device versns frequency (4-12 GHz) and drain matching network is employed, S~2can be tuned lo zero for
current is reported. A comparison between the scattering parameters each r.; then F,(S~2 ) reduces to the constant value F,(O).
provided by the method and those measured by means of a network
The gain Ga(r,) can be computed through the scattering
anafyzer isalso included.
parameters or measured as a power ratio by means either
of a gain measuring system or the same instrumentation
I. INTRODUCTION used for noise measurements [l].l After measuring some
o (redundant) values of Ga(r, ) the gain parameters GaO,rOg
PTIMIZATION OF noise figure, gain, and input
(magnitude and phase), and N. defined by
and output VSWR in designing wide-band, low-noise
MESFET amplifiers requires a complete character~ation 1 1 Ir. - roglz
of the device in terms of noise, gain, and scattering param- ‘=—+4Ng(~_lr,l,)(~-lr0,12) ‘3)
% (r.) G..
eters versus frequency and drain current.
Scattering parameters are usually measured through a can be derived through the same (computer-aided) data
(automatic) network analyzer. processing procedure as above for the noise parameters.
Noise and gain parameters cannot be measured through The described procedure, now commonly applied, refers
an instrument, but their determination requires time-con- to the state-of-the-art in the field of device characterization
suming experimental and data-processing procedures. as assessed in 1969, when Lane first proposed to substitute
To determine noise parameters, it is necessary to per- with a computer-aided data processing technique the
form measurements of the device noise figure F(ll,) for graphic procedure established ten years before by the IRE
some (redundant, i.e., more than four, for accuracy) values Standards [2], [3]. This procedure is, however, time-con-
of the input termination reflection coefficient r,. Solving suming and requires different measuring systems for the
then the set of equations derived from the following rela- determination of all the parameter sets.
tionship (or an equivalent one): In this paper, a method is presented which allows the
simultaneous determination of the noise and gain para-
jr, - ronjz
F(r. )= Fo+4~n (1) meters and of those scattering parameters that are needed
(l-lr,t’)(wrony) for the design and analysis of microwave amplifiers
the four noise parameters FO,rOn(magnitude and phase), (%1, S22,1S121l&, ll, L S12$1) b means of a single measur-
and N. are obtained. ing system. The instrumentation employed, i.e., a noise
In order to evaluate the device noise figure F(r, ), it is characterization setup equipped with a vectorial reflectom-
necessary to account for the noise contribution of the
measuring stages following the device under test (DUT),
1Commercial instruments for the simultaneous measurement of noise
figure and gain of adevice driven by anoise sourcearealsoavailable (e.g.,
AILTECH mod.7380 and Hewlett-Packard rnod.8970). Theseinstruments
Manuscript received June 2, 1983; revised January 5, 1984. This work are very useful for measurements on matched devices; they are not
was supported in part by the Nationaf Research Council (CNR), under convenient, however, for transistor characterization because in this case
the MONOMIC Program. the useof amatching network at the DUT output port isrequired, which
The authors are with the Istituto di Elettrotecnica and Elettronica, in turn implies time-consuming tuning adjustments as r, varies and
University di Palermo, viale delle Scienze,90128 Palerrno, Italy. increased risk of oscillations.
0018-9480/84/0300-0231 $01.00 01984 IEEE
232 IEEETRANSACTIONS ONMICROWAVE THEORYAND TECHNIQUES, VOL. MTr-32, NO.3,MARCH
eter, and the time--consumption are the same required for - Fm - F,
, I
the determination of noise parameters only through con-
IIHII+EI+-EEIE
ventional methods. The chosen test-set topology and the
particular data processing procedures devised, together with
the fact that measurements of signal sensitive parameters
m m
are performed at noise level, assure good repeatability and
accuracy. A further advantage of the proposed measuring
system is that the possibility of oscillation build-up during r,,‘ s;, S;2~r,
characterization of potentially unstable devices is strongly (
Fig>.1. Simplified block diagram illustrating themeasurement principle.
reduced.
The theoretical analysis of the method, the measuring
setup, the step-by-step experimental procedure, and the measured for each r~ to compute Fr. From the value of S22,
(computer-aided) data processing technique are fully dis- obtained by measuring S;’ for r, = O,and the relationship
cussed. Measurement procedures for testing the accuracy of the device gain under .ZO-terminated (usually 50 Q) input
of the results obtained are also described. conditions G=(0), ISzlIgiven by
As an experimental result, the complete characterization
1s2,12= Ga(o)(l - 1s2212) (5)
of a packaged GaAs MESFET versus frequency (4–12
is derived. From the relationship that expresses the depen-
GHz) and drain current (5-30 percent 1~~~) is reported. A
dence of G.(r,) from the input termination reflection
comparison between the scattering parameters computed
coefficient in terms of the scattering parameters we have
through the proposed method and the ones measured by a
network analyzer is also included. IS211T-V)
‘1-s’lr’’2=’ (6)
II. ANALYSIS OFTHE METHOD (1- p~2y)Ga(r$)
The method presented here is the improvement of a which allows the calculation, through proper algorithms, of
method for the simultaneous determination of noise and the magnitude and phase of S1l.
gain parameters through noise-figure measurements only, For example, by putting x = Re {Sll } and y = Im {S1l}
already successfully applied for the characterization of from (6), we get
bipolar transistors up to 4 GHz [4], [5].
x2+y2-}ax+by+c=0 (7)
The principle of the method can be discussed referring to
the simplified block diagram shown in Fig. 1. where
As compared with conventional measuring systems, it
2Cos r, 2sin ~
can be observed that a matching network (tuner) at the ~=– L = b=
output of the DIJT is not used because a) it requires Ir$l Ir,l
seeking for a careful tuning in order to maintain the
matching for each r,, and b) it may cause device oscillation
(also outside of the measuring band) which falsifies the
measurements. Instead of a tuner, an isolator and a step
The set of equations obtained from (7) for some (redun-
attenuator are used. The isolator allows the separation of
dant, i.e., more than two) values of I’, can be solved by
the DUT from the stages following it, allowing us to use,
means of the least-squares method and a (computer-aided)
for evaluating F,(s..z ), the expression
successive approximation procedure.
p-s:2rr12 From the computed value of Sll and the relationship
(4)
E(%/2)=~ r(o)
l–p:’lz
S12S21= ~(sj2 – S22)(1 – sllr, ) (9)
which permits the computation of F, from the measured s
values of S~2, F,(O), and I’,. The step attenuator is em- obtained from the expression of S;2(r, ), the product S12Szl
ployed in order to easily obtain several different values of is then computed (magnitude and phase).
F, for the same value of r,. It is noteworthy that the computed parameters (S1l,
Once a set of measurements of Fw is performed for a IS121,IS211,and S12S21)are derived with high accuracy due
fixed value of r. and for several (redundant, i.e., more than to the redun / ancy in the processed data; only S22 is
two) values of F,, F(r,) and G.(r,) are derived from (2). directly measured. In addition, the former set of parame-
By repeating this measurement cycle for some (redundant, ters are determined at noise level, thus eliminating nonlin-
i.e., more than four) values of r,, the noise and gain earity effects.
parameters of the DUT are derived from (1) and (3),
III. MEASURING SETUPAND EXPERIMENTAL
respectively.
PROCEDURE
From the above measurements, the scattering parameters
Sll, IS211,IS121, SIZSZI may also be derived by computa- The detailed block diagram of the measuring system
L
tion. used for both C- and X-band measurements is shown in
The DUT output reflection coefficient S~2 has been Fig. 2.