Ad0850232 PDF
Ad0850232 PDF
Ad0850232 PDF
AD NUMBER
AD850232
FROM
Distribution authorized to U.S. Gov't. agencies and their contractors; Administrative/Operational Use; Mar 1969. Other requests shall be referred to Commanding Officer, Naval Weapons Center, Corona Lab., Attn: Code 2035, Corona, CA 91720.
AUTHORITY
US Naval Weapons Center ltr dtd 16 Apr 1969
,,4 .r
NWCCL TP 845
0(
SO00
4by
D. Rubin
&.,
ODOD
--I~rFA r r{
4
5
,.Research Department
ABSTRACT. This report discusses a method-known as theinvill method-for determining the terminations that a transistor amplifier should have for a specified value of power gain and bandwidth. Basically, the Linvill method makes use of measured transistor parameters to develop charts from which one can read power gain and input impedance or admittance as functions of the load terminations. This report gives a complete geometrical deri-v-ation of the Linville "stability factor," whose value is an indication of the stability of the amplifier under various load conditions. In addition, procedure steps are given for using the charts developed for determining input and load admittances.
NAVAL
WEAPONS
CENTER
DISTRIBUTION STAT!UMENT FOREIGN THIS DOCUMENT IS SUBJECT TO SPECIAL EXPORT CONTROLS AND EACH TRANSMITTAL TO COMCR FOREIGN NATIONALS MAY BE MADE ONLY WITH PRIOR APPROVAL OF THE (CODE 2035). CORONA. MANDING OfFICER OF THE NAVAL ',WEAPONS CENTER CORONA LABORATORIES CAL;FORNIA 91720.
"GOVERNMENVS
CFMl
* WhITE SECTION
Doc FCAi........
lUf
SECT
VAL
EAPONS
............................
CENTER
Commanding
CORONA
... R. M. Fo-bis, CDR.. USN
LABORATORIES
Officer
...............
FOREWORD In designing high frequency receivers-, personnel of the Electronics Division of the Naval Weapons Center Corona Laboratories must devise sensitive amplifiers which have very little noise and wide bandwidth. Such circuits require highly effective ust of transistors. This report describes a method by which. the measured admittance parameters of a transistor are used to obtain power gain as a function of frequency and load admittances of an amplifier. This work was accomplished under AIRTASK A36533/216/69F17343604.
Released by C. P. HABER
Acting Associate Head of Research Department, Corona
26 March 1969
Collation
. Cover,
ii
NWCCL TP 845
i.
CONTENTS Introduction .................................... Power Relationships .......................... Stability Factor ....................................... Gain Relationships ............................. Application Steps ............................... Appendixe s: A. B. C. Conversion to h Parameters ...... ............. .. ... 25 ... .. .. .. 1 2 10 .11 18
Circle Intersection Points of C > 1 ............ Determination of Input Admittance from G2 and B 2 Values ...........................
.. ..
27 30
References ...................................
'I
ii
9.T.
NWCCL TP 845
INTRODUCTION It is not reasonable to use transistor models for the design of amplifiers in the UHF region and above. Actual measurement of the admittance (y), hybrid (h), or scattering (s) parameters of a transistor at the frequencies of interest is practical with laboratory instruments and leads to precise calculations of gain, bandwidth, input and output admittances, etc. In 1956 Linvill and Schimpf (Ref. 1) described a method by which gain and input impedance could be characterized graphically for any value of load admittance. All that was required was a knowledge of the h parameters at the frequencies of interest. From these, radii of various circles (gain circles) were calculated and positioned on a regular Smith Chart. The regions of stability were also clearly indicated. The method, which was reiterated in a book (Ref. 2), was revie- ed by various writers (Ref. 3 through 6), most of whom preferred the use of y parameters. An exceedingly important quantity known as the stability factor C (called the critical factor by Linvill) was defined by Linvill in his 1956 paper. The C factor, a function only of the device parameters, determined if the device would be stable for all values of load admittance, or if not, would determine the area of the Linvill chart (and hence the range of load values) to be avoided. Linvill and his reviewers define the C factor, but do not give explicit proof regarding its evaluation, although it was suggested that the C factor could be evaluated by way of geometry. In this report it will be shown that the geometry used in the discussion of the Linvill method could indeed be used to determine the value of the C factor. The explanation of the Linvill method can then be made without it being accepted on definition alone or requiring additional references. It will finally be shown how charts obtained by the Linvill method can be used to determine power gain and input admittance as a function of load admittance.
!1
P0W1~il 11 I11LAT1ONS1 11Ptig uiw~ 1hpjkt JPI) m~id power otitl,.At (P,,) rAr ubtatined fruiv the relattomhhipo botwoon Ninath Voltage liati aurrUVIt ph$ OOPO (Pl) Mcud (110 y OV1 PA PKIIietes11Of the dOViO -1, 8'ettering (o) pipikimptret may he directly convaredtnt to itheu' of thooo tw Not#, The The derivktion of fo11tjwftng analyoto wil be donte uiahi y patrmetro th.o po~wer' aut~utinN uping h pirametSIr In giveti in Appo-ndim A, ty
rq1iqHi~tNu
valitti
Vol
whorp~
V L t thie Wlla
chtti auikkvo
Tluwlaful,4e
Not Nert
y,
h "v~
"Y41l14
\"I.,y
YA -
NWCCL, TP 845
e-(L e 1Re
(2)
P. = Re Pile,
+ yi~e2)e
= Re
1i1
+ yiz(L + jM)
)I ell
1 p v. _._
lJZ
:~V2 "
Then (Y12Y21 21 LReI
lYl +M.ImI
Y2 1
P'=Rey
1 zi
+) lel + Y2 2 e 2
1
e2] (3) 3
= -Re[
i e(L+ 1
jM) (R
Y) Ne
+ Y2-
but
NWCCL TP
845
_------------------
~e 12
(L.)L+ =I.
L2 + M2 4(Rey,,) .1 l je 1
Py=
-Re L.(L+
[Mo
-I
2) Ih2Z1 (L2 + M
M)
(22
Therefore,
_ _ _ _
2Re YM2
__ _
__
(4)
y 4Re(e
I)(a
Y2R1y 2
2 vRey
Z then fromeq.
pI
y2 2
2Re
3 and 4,
= Re yl 2 La2 + Ma2 (6)
N2)2 - , (L+ L Me 1b 2b
(7)
NWCCL TP 845
or
MRe Y22
L L+ jM
T
'
Y22
(8)
That is, a given value of L and M represents a specific YL. For YL = Y22' we have from Eq. 8 Y
L 22-
2ReY 2
L+ jM
-jY
(Re y2 2
)
+ Y2 )
?' (R e
=
(Re Y 22
Y.Ym 2
Re
Y2 2 (L+ M
I)~r
j2
Therefore
2 jML+ L jMi
or L = 1 and M= 0.
P! 1
=
B'.
10
(9)
P'
00 10
is therefore,
NWCCLTP 845
P'
=2P' L
0
-(L
+ M 2 )?'
-
00
oo
[(L
1)2
M(o
The equation for Pl is that of a paraboloid whose axis is along the line L = 1, M = 0, and whose maximum height is P'0o. Its intersection with the LM-plane (see Fig. 2) occurs when P' = 0. There we have 2 1 M = 0). 1, i.e., a circle of radius I centered at (L 1, (L2 ._)2 + MI
I!/\
.'<,<.'.
",
SIlk
~\
I\.\
I
I 0
.
..-'-; ,
-N.
M'=L
FIG. Z.
The equation for Pt is that of a plane, intersecting the LM-plane along the line given when P1 = 0, i.e., when 0 = ReYlLaI + Ma 2 , or
NWCCL TP 845
aI M = a
Re -(01) a2
There is another line, M' (called the gradient line), which lies ir the I,Mplane, is perpendicular to the Pt = 0 line, and intersects (L = 1, M = 0). This line is the projection of a line in the Pt plane along the maximum slope of the plane. The gradient line may be written in the slope-irintercept form M = mL + d. For this line to be perpendicular to the Pt = 0 line, the slopes of the two lines must be negative recipricals of each other, i.e., a a1
I a a2 so that a2 M= -a1 L+
so that d = a 2 /a
(L
1)
(12)
The P! = 0 line and the gradient line intersect at (ho, Mo). I Eq. 11 and 12,
From
Mo
(R4e \a 2 /
" 2I
\al/Jo
a1
we have
NWCCL TP 845
a.a
(Val)
a1
a1
\a/
The Po = 0 circle, the P= 0 line, and the gradient line are shcwn in Fig. 3. From the figure'
tan0 = -tan
1a 2 /ai; therefore
(14)
tane
---
a2 a.I
(15)
and since
(-1 2
2 1)
=~Ry
12y 2 1)
im(yl,2 y 2 1 )]'
Re (Y1 2 Y2 1 ) + jim(yl 2 y 2 1 )
-~(Yl
22
l)
(y1 2 y 2 1 ) or
= tan 1Arg(-yl 2 yl)" ""
(16)
_NWCCL
TP 845
\
GRADIENT LINE
FIG. 3.
LM-Plane Traces.
Any line perpendicular to the LM-plane which intersects the P plane and P_ paraboloid will give the gain G = Po/Pt_ =~ o/PI' for a particular value of L and M, and hence for a particular YL' For a given output power (a circle parallel to the LM-plane), the point of least power input must lie along the line of steepest descent of the Pt plane. To find the maximum gain, we find the point along the M' axis
which gives the largest P, over P ratio.
NWCCL TP 845
STABILITY FACTOR Sighting down the L' axis of Fig. 2, we obtain the projection shown in Fig. 4. The P! plane is reduced to a line and the P paraboloid reduced to a parabola. Note from Fig. 2 tnat instability can occur for some values of L and M if the Pt = 0 line intersects the Po = 0 circle. Within the sector thus formed, there will be some values of P even though Pt is zero or negative, i.e., the device will oscillate for some values of Y1 ,. In Fig. 4, if -r < -1 the dev'ce will be stable for all YL" From this figure it is clear that for potential instability C > 1; this is the C referred to earlier as the "stability factor." It will be shown that C depends only on the device parameters (not YL). In general there will be some frequency above which C will be less than one and (because of low inherent gain) the device cannot oscillate.
P?
100
+1
J(L
)'
(Mi
1)2(a )
)2
1 2)
=
and since from Eq. 13, (LO - W
~22
(a,/a?)ZMo
10
NWCCL TP 845
(a
a,,
Rey i 2
R y i
"1
F2[
1)+1--Il
y IIRe Y2 2 1
(17)
r Therefore
j2Rey, Re y 2? -Re(y 1 Zy
GAIN RELATIONSHIPS We now find maximum device gain, Gmax. From Fig. 4, P' 1 - M' 2 ) and P!- P! (M'/r + 1); then the device gain, G, is 0 00( 1 10 ' p, p, i' G oi oo(
io
P=
~0
-MI
,2)=G
00
IK
-
(19)
= KGG G 00
Then (dG/dM')M,=- = 0.
11
NWCCL TP 845
__
4t
-r M'
or since M' > -r
*=
-: I
r<
I+ CM' we get
-L
-
I(
12c
J1-
2\
(22)
Note in Eq. 22 that the maximum value of KG, given C < 1, occurs when C = 1. There KG = 2. The maximum gain Gmax, therefore, cannot be greater than 3 dB over the gain Goo given when YL = Y22" Also, KG cannot be defined for C > 1 since some values of YL will then cause oscillation. From Eq. 9 and 5,
Po
G 00
4Rey
z22
00
Pi
y ReYll - Re(yl Me 2 .2 1 )
j
or
NWCCL TP 845
G
00
IYZ 1 21
2[2Me Y,,Re Y2 2
-
(23)
Re (Y1 Z2Y2 I
1 + CM,
or M' 2 + CKM' = 1
-
K.
Therefore
S=
- K + (2)
(24)
Thcse are circles centered on the M' axis. Each value of K gives a circle whose center and radius, respectively, are given by CK MK - --K 2C
"PK
[IK+(.-" G max)
(25)
Specifically, for K = KG (G
CKG
"MK
PKG =0
Mt
and for K = I (G = G MK
).
C
C
S"
PK=l
13
NWCCL TP 845
Figures 5 and. 6 show constant gain circles constructed from the y parameters of a 2N4957 transistor. A few constant gain circles are shown at 1,000 MHz (Fig. 5) and 450 MHz (Fig. 6). Note particularly that for C < 1, 1. A gain of KG Goo occurs at only one point (a particular YL).
2. A gain of I Goo occurs for a circle which includes the point L= 1, M= 0. 3. Gain circles are contained within one another-the largest circles correspond to the least gain.
and for C > 1, 1. As can be seen from Fig. 4, gain corresponding to values of MI < -r are indeterminate. Areas to one side of the M' = 1/C line in Fig. 6 correspond to values of YL which will lead to oscillation. Figure 7 shows the gain diagram for such a case. 2. Equation 2Z is meaningless; however, Eq. Z3, 24, and 25 hold, just as for C < 1. 3. All circles intersect at M' Appendix B). -I/C, L'= A- - I/CZ (sec
We have now found the gain of the device as a function of L and M. Since the L and M values relate to YL, we now find gain as a function of YL' Equation 8 gave ZRe Y 2, L+ jM
= G + jB 2
(26)
14
NWCCL TP 845
100
5~
NWCCL TP 845
f
G
450 MHz
318
1.20
K=2-
M=M, 0 L
K=0,
K=0 .6
K="0.0
..
( 0
NWCCL TP 845
p P
00 P.0
UNSTABLE
FIG. 7.
jM) 2
L2
L
or 2 .,
Ley2M
2M R e
2
+G L2 S+ 2
M
t.(L
2e
Y2MR2
eY~ G 2
(2e
G
17
NWCCL TP 845
BZ (++M .Y2~2)2
/ L =\BZ ( 2 a) /2Z
(27)
APPLICATION STEPS Equations 27 represent orthogonal circles, very similar to those obtained by relating reflection coefficient to load impedances of a transmission line (Smith Chart). The circles obtained from Eq. 27 are plotted in Fig. 8 and 9 over the gain circles of Fig. 6 and 7, respectively. Of course, oneneed only to use the ready-made circles on a Smith Chart. The design process is therefore as follows: 1. Normalize a Smith Chart in terms of Rey 2 2 (or Reh 2 2 ) as shown in Fig. 9. Be sure to add 1.0 to all values of the abscissa of the Smith Chart. 2. Obtain sets of Y parameters for the center of the frequency range to be used. For broadband amplifiers several sets may be required (for each set plot one chart). 3. 4. Draw the gradient line at an angle
5. If C > 1, draw a line perpendicular to the gradient line at a distance -1/C from the center of the chart.
6.
C
IRey 2 2
-
MkCK/2.
9. Determine ranges of L and M values needed for given bandwidth. This determines required YL and also input admittance (Appendix C). 10. Devise matching networks for source and load impedances.
Steps 9 and 10 are best carried out on a digital computer, especially when a large bandwidth is required. Several sets of G2, B 2 values may
NWCCL
TP 845
f =1000 MHz
G =6.72
K=O8 LnilChr
C = 0.76.
K=0.
119
ItI
NWCCL TP 845-
G
00
1.20
9. FIG.
Chrtfo2C0.0
=il
20=
NWCCL TP 845
be lected, e.g., from within the K = 1 (Gain = Goo) circle of a Linvill cha. t near the highest frequency of interest. Each GZ, Bz set requires a particular output admittance. At this frequency a matching network is devised which makes the load appear as the required output admittance. The input admittance of the amplifier is also determined by the G 2 , B 2 set under consideration. An input network is then devised that makes this input admittance appear as the conjugate of the source admittance. We now have one frequency at which we know that all available power will be absorbed by the amplifier and a given power gain will appear at the load. At other frequencies the output network and load, togethor with the now values of y parameters, determine new G2, B2 valuts and henet new device pi1n. Tht resulting transistor input admittance and input network will determine how much available power will be absorbed by the amplifier. The network attenuation multiplied by the device gain yields total power gain of the amplifier as a function of frequency. ror some of the original GZ, B2 values selected at the high frequency, the resultant mismatches at lower frequencies will yield a rather level total-gain-versus-frequency curve.
21
NWCCL TP 845
Appendix A
CONVERSION TO h PARAMETERS In this appendix the procedure is shown for deriving the transistor power equations using the h parameters. By definition e i Therefore, e_ h2 1 hz +YL 22 +YL h22, 11
h h
-h
=
i1 + h 1 2 e 2
h 2iI + h
e2 = -e2YL
i-.-I
if YL w 0 l',
1 m t
121 h 2 2 + h22
h 21 h2 2Reh 22
e2
Now
22
NWCCL TP 845
Re{l[il+ h(lh
YMh
ll1
1)
ehise =
ReJh
12
21~
__
Reh+ +
"\
jJ 8)2
P.
/hzhl
FM/hhz
/h
Threo
em 1h
2
hpaaeesi
"1
/hi
2h
(=28
jM )+
from which
12
Re Lh2((2Reh
Ih (h
21
he
22
2(28hl
22
23
NWCCL TP 845
or P'=L 2 -2Reh o
V 1
22
2I)h
(L + M )zReh
22
(29)2
Equations 28 and 29, it can be noted, are identical to Eq. 3 and 4 if we set Y12Yzl M. e y 22 h12 h21 2Ke h 22
a-
2ReY12
2Rehl2
The analysis follows through with hij zation of the Smith Charts to Reh 2 .i
24
NWCCL TP 845
Appendix
CIRCLE INTERSECTION POINTS FOR C > 1 In this appendix it is shown that the Dower gain circles for C > 1 all intersect at (M' = -1/C, L' = 41 - 1/C 2 ). Equations 25 gives 2K
The equations for circles of different K values (see Fig. 10) can be written
(L') 2 + (M' , 2 =) 2
/
K
FIG. 10.
a!
NWCCL TP 845
_
Mk,
we get
If the intersection of two circles due to K1 and K 2 occur at L'1 and MW, we will have (L) (M')+ CMIK = 1 -K
(L-,)Z + (MI.)
+ CM!K 2 = 1 - K
i
L
M1 -C
-(30)
Since M! and L! are independent of K values, all circles for C > 1 will intersect at 1 %1 ' M'
-
L=
1C-C2
and (L')- + (M.)2 = 1, i.e., the intersections are those points where the M' = -1/C line intersects the extremities of the Smith Chart.
NWCCL TP 845
Appendix
DETERMINATION OF INPUT ADMITTANCE FROM G AND B VALUES 2 2 The following procedure shows how the input admittance Yin can be determined from the coordinates of a point (G 2 , Bz) on the Linvill chart. Since, by Eq. 1,
eI = YlleI + Y12e2
e2 = el(L + iM)
2
=
/ 2Rey y 1
2
Then
Y.
(L-+
M)(2' 2y2)
(31)
As in Eq. 5, let
a a, + ja:
A
2Rey22!
ale
al cosk
j al
sinfl
From Fig. 3,
jsinO)
27
NWCCL TP 845
Y in
or
-JisinO)
Y~s
- Yl
lal
(32)
= (LcosO+ MsinO)+ j(McosO - LsinO) = M"+ jL" where M", L" are orthogonal axes rotated an angle 0 from the L, M axes. The projections M", L" of the point (GZ, BZ) onto these axes thereby give (see Fig. 11) Y in = Yll + a(M" + jL"-.)
C.in+ jB.
L" AXIS
M AXIS
(G 2,
B)
I ",l ; .1t
Ax f~i t s
IN L AXIS (1,0) L
FIG. 11.
28
NWCCL TP 845
where Gin and Bin, the input conductance and susceptance, are G in Rey
1
respectively,
+ laIM"
B.in = Iayl
and
aIL" l
MRe y22
29
NWCCL TP 845
REFERENCES 1. Linvill, J. G., and L. G. Schinmpf. "The Design of Tetrode Transistor Amplifiers," BELL SYST TECH J, Vol. 35 (July 1956), pp. 813-840. Linvill, J. G., and J. F. Gibbons. New York, McGraw-Hill, 1961. Transistors and Active Circuits.
2..
3.
Johnson, G. "High Frequency Amplifier Design Using Admittance Parameters," ELECTRO-TECHNOL (November 1963), pp. 74-79. Johnson, G., P. Norris, and F. Opp. "High Frequency Amplifier Design Using Admittance Parameters," ELECTRO-TECHNOL (December 1963), pp. 66-74. Lauchner, J., and M. Silverstein. "Design High Frequency Amplifiers Graphically," ELECTRON DESIGN (April 1966), pp. 48-54. Texas Instruments, Inc. Solid State Communication, by the Engineering Staff. New York, ,IcGraw-Hill, 1966.
4.
5.
6.
30
s..turtt, Clsmca.,,o.
(Security classificationof title, body of abstract nlidindexi.p annorficmn n u.t hb entered when the overall repoFI is (i.,sili,?d)
I OF41INATING ACTIVITY (Corporate author) 20. REPORT SECURI TY CLASSIFICA TION
UNCLASSIFIED
2b. GRo
AU THOSiS1 (First
Jase nvee)
David Rubin
, REPORT'OATE 7s, 7OTAL NO, OF PAGES
7b. NO. OF R EF 5
March 1969
0a. CONTRACT OR GRANT NO. go.
31
ORIGINATOR'S REPORT NUM9TERISI
b. PROJECT
NO.
NWCCL TP 845
,. OTER REPORT NOISI (Any other numbers that may be assigned this report)
C. AIRTASK A36533/216/69FI7343604
d.
10
DISTRIBUTION
STATEMET' -
DRTTHIS
DOCUMENT IS SUBJECT TO SPECIAL EXPORT CONTROLS AND EACH TRANSMITTAL TO FOREIGN GOVERNMENTS OR FOREIGN NATIONALS MAY BE MADE ONLY WITH PRIOR APPROVAL OF THE COMMANDING OFFICER OF THE NAVAL WEAPONS CENTER CORONA LABORATORIES (CODE 2035), CORONA, CALIFORNIA 91720.
12. SPONSORING MILITAN4Y ACTIVITY
SUPPLEMENTARY
NOTES
Washington,
D. C.
20360
13
AB
TRACT
This report discusses a method-known as the Linvill method-for determining the terminations that a transistor amplifier should have for a specified value of power gain and bandwidth. Basically, the Linvill method makes use of measured transistor parameters to develop charts from which one can read power gain and input impedance or admittance as fun..tions of the load termination. The report gives a complete geometrical derivation of the Linvill "stability factor,'1 whose value is an indication of the stability of the amplifier under various load conditions. In addition, procedure steps are given for using the charts developed for determining input and load admittances.
DD
*
"j I1OR
473
0l
UNCLASSIFIED
Securilv Cl. ssification
Security Clanfcaition
14
UNCLASSIFIED
LINK A KE WORDS ROLE
.-
LINK OLE
B WT
LINK ROLE
C WT
Linvill method transistor amplifiers admittance parameters Smith Chart stability factor
rfr%
FOR"M 1AA71
(BACK)
UNCLASSIFIED
___~~A~ SUPPLEMETR
-
A__
k-~
A&I
INFORMATI ON
AD-850 232
No limitation
USNWC Itr,
16 Apr 69
Coronat, crtiif.