8A Small-Signal Amplifier

Design: Design for Maximum

Power Gain and Design for
Fixed Transducer Power
Gain
The information in this work has been obtained from sources believed to be reliable.
The author does not guarantee the accuracy or completeness of any information
presented herein, and shall not be responsible for any errors, omissions or damages
as a result of the use of this information.

March 08

2006 by Fabian Kung Wai Lee

References

[1]* D.M. Pozar, Microwave engineering, 3rd Edition, 2005 JohnWiley & Sons.
[2] R.E. Collin, Foundations for microwave engineering, 2nd
Edition, 1992 McGraw-Hill.
[3] R. Ludwig, P. Bretchko, RF circuit design - theory and
applications, 2000 Prentice-Hall.
[4]* G. Gonzalez, Microwave transistor amplifiers - analysis and
design, 2nd Edition 1997, Prentice-Hall.
[5] Gilmore R., Besser L.,Practical RF circuit design for modern
wireless systems, Vol. 1 & 2, 2003, Artech House.

March 08

2006 by Fabian Kung Wai Lee

Basic Single-Stage Small-Signal

Amplifier Design Flow
Define
Center frequency fo
Power gain GT
Noise figure F
Source/load impedance
Bandwidth
VSWR

Tuning and
Optimization
Large-signal
analysis:
Gain compression
TOI
Efficiency

Not
Satisfied

Select
active component

Design d.c.
biasing circuit
Get S-parameters

Stability analysis and

stabilization

Some graphical tools used:

Constant GP circles
Constant F circles
Constant GA circles
Constant VSWR/mismatch
circles

Design input and output

Impedance transformation
Networks with Smith Charts

Check GT, noise figure Satisfied

VSWR, bandwidth

March 08

Prototyping

2006 by Fabian Kung Wai Lee

Linear/Quasi-Linear Amplifier Types

Amplifier
Types

Required s

Required L

Graphical tools

Maximum
Unilateral GT

s11*

s22*

None

Maximum
GT(max)

sm

Lm

None

Fixed GT
Tunable
with user
defined F and
input mismatch

Tunable

Constant GP circle
Constant F circle
Constant
VSWR/Mismatch
circle
Constant GA circle

Lowest noise

Tunable
(typically = 2*)

Constant F circle

Maximum
Output Power

sm

OL

Constant PL
contour

March 08

2006 by Fabian Kung Wai Lee

Good for lower

frequency,
fo < 500MHz.
Not discussed
here, see [1],
[4] and [5]

Discussed
in High Power
Circuits
4

1.0 Maximum GT Amplifier Simultaneous Conjugate

Match

March 08

2006 by Fabian Kung Wai Lee

Simultaneous Conjugate Match in

Amplifier (1)

To maximize power transfer in amplifier, both input and output ports

of the amplifier have to be conjugately matched.
When both input and output impedance of the amplifier are
conjugately matched to the source and load impedance, the
amplifier is called Simultaneous Conjugately Match (SCM).
It is not easy to achieve SCM, as L affects 1 , and s affects 2 .
Only under special condition can SCM occurs.
S D
1 =

2 =

11

1 S 22L
S 22 Ds
1 S11s

Under SCM, Gp= GA= GT, and the gain obtained from the amplifier is
the maximum, GT(max).
Most of the time the condition for SCM is not fulfilled and we can
only optimized the power gain of the amplifier while GT(max) is not
attainable.
March 08

2006 by Fabian Kung Wai Lee

Simultaneous Conjugate Match in

Amplifier (2)
Amplifier
(2 port)

Source

s 1*

Non SCM

2* L

Amplifier
(2 port)

Source

s = 1*
March 08

Load

SCM Condition

Load

2* = L

2006 by Fabian Kung Wai Lee

Load and Source Impedance for

Simultaneous Conjugate Match (1)

The input and output reflection coefficient of an amplifier are given by:

S DL
1 = 11
= Input Reflection Coeff.
1 S 22L

When we want the input and output port of the amplifier to be

conjugately matched to source and load impedance, s= 1* and L=
2*.
Thus we need to solve the following simultaneous equations:
S DL
s* = 1 = 11
1 S 22L

S Ds
2 = 22
= Output Reflection Coeff.
1 S11s

S Ds
L* = 2 = 22
1 S11s

This can only happen for a particular set of source and load impedance
(as s affects 2 and L affects 1 ).
March 08

2006 by Fabian Kung Wai Lee

Load and Source Impedance for

Simultaneous Conjugate Match (2)

Solving both equations simultaneously results in (1.1a) and (1.1b),

Example 9-10 of Ludwig, Bretchko [3] provide the details.

Sm =

1
A1 A12 4 B1 2
2 B1

A1 = 1 + S11 S 22 D

)
1
2

Lm =

1
A2 A2 2 4 B2 2
2 B2

A2 = 1 + S 22 S11 D

)
1
2

*
B2 = S 22 DS11

*
B1 = S11 DS 22

(1.1a)

March 08

(1.1b)

2006 by Fabian Kung Wai Lee

Power Gain During Simultaneous

Conjugate Match (1)

When we connect source and load impedance having Sm and Lm to

an amplifier, Pin = PAs , PL= PAo.
As a result GP = GT = GA .
Also note that transducer power gain GT under this condition is the
maximum power gain! Commonly denoted by GT(max).
PRs = 0

PAs

Pro = 0

PAo

PL = PAo

Pin = PAs
Input
Transform
-ation
Network

Zs

Vs

Sm

Output
Transform
-ation
Network

Amplifier
K>1
Sm*

Lm*

March 08

PAs

ZL

Lm

P
P
P
GP = L = L = GT = Ao = G A = GT (max )
Pin

PL

PAs

2006 by Fabian Kung Wai Lee

10

Power Gains During Simultaneous

Conjugate Match (2)

Substituting (1.1a) and (1.1b) into the expression for GP will give us
GT(max).
GP(max)

Gp = GT = GA = GT(max)
2

GP =

Lm

S 21 1 L
2

1 S 22L 1 1

GMS
Maximum
stable
gain

S DL
1 = 11
1 S 22 L

S 21
K
S12

GT (max) =
2

Roulette Stability Factor

March 08

K=

1 S 11 S 22

+D

2 S12 S 21

K 2 1

(1.2a)

(1.2b)

2006 by Fabian Kung Wai Lee

11

Summary for SCM Amplifier Design

SCM to obtain maximum power gain, can only be applied when the
amplifier is unconditionally stable with input and output port conjugately
matched.
And when this happens, the maximum power gain that can be obtained
is:
S
S
GT (max) =

21
2
K K 1
S12

GMS =

21

S12

(1.3)

The limit for GT(max) is GMS. It is the largest maximum stable power gain
that can be achieved for an amplifier.
From the above relations for GT(max), it is only real if K 1. Thus an
amplifier can only be simultaneously conjugate matched when it is
unconditionally stable (pass K factor test).
If the amplifier is only conditionally stable, then SCM or maximum power
gain cannot be achieved. The required values for Lm and Sm will falls
outside the unit disc of the Smith Chart. We can only optimize GT up to
a certain point.
March 08

2006 by Fabian Kung Wai Lee

12

Example 1.1 - SCM Amplifier Design

A BJT with IC=10mA and VCE=6V is operated at 2.4GHz. The

corresponding S-parameters are S11=0.3<30o, S12=0.2<-60o, S21=2.5<80o, S22=0.2<-15o. Determine whether the transistor is unconditionally
stable and find the values for s and L that provide maximum power
gain. Sketch the schematics of the system (use a box to represent the
amplifier).

March 08

2006 by Fabian Kung Wai Lee

13

Solution to Example 1.1

S11 = 0.260 + j 0.150
S 21 = 0.434 j 2.462
S12 = 0.100 j 0.173
S 22 = 0.193 j 0.052
Stability Test:
2

K=

1 S11 S22 + D
2 S12 S21

= 1.178

D = S11S 22 S12 S 21 = 0.555

Since K > 1 and |D| < 1, the amplifier is unconditionally stable.

We can maximize the power gain Gp by carrying out simultaneous
conjugate match design.

March 08

2006 by Fabian Kung Wai Lee

14

Solution to Example 1.1 Cont...

Use (1.1a) to find

2
Lm = 2 1B A2 A22 4 B2 = 0.044 + j 0.116
2

Use (1.1b) to find

2
sm = 21B A1 A12 4 B1 = 0.281 j 0.091
1

Convert to impedance:

1+

Z Lm = Z o 1Lm = 53.085 + j12.508

Lm

1+
Z sm = Z o 1sm
sm
March 08

= 86.889 j17.325
2006 by Fabian Kung Wai Lee

15

Solution to Example 1.1 Cont...

Finding maximum power gain:

G p (max) =

S 21
S12

86.889j17.325

March 08

K K 2 1 = 6.942

Amplifier

2006 by Fabian Kung Wai Lee

53.085+
j12.508

16

Appendix 1
Derivation of Source and Load
Impedance for Simultaneous
Conjugate Match and Constant
Power Gain Circle

March 08

2006 by Fabian Kung Wai Lee

17

The Condition for Maximum GP (1)

We start by looking at the expression for power gain GP.

1 =

S11 DL
1 S 22L

2 =

S 22 Ds
1 S11s
2

GP =

S 21 1 L
2

1 S 22L 1 1

g2 =

March 08

Even though the method presented

here is much longer, it gives more
insight into the characteristics of
the power gain GP.

1 L
2

(1 )S
2

L
2

2
21

1 S 22L S11 DL

1 S 22L S11 DL

= g 2 S 21

We call this the normalized

power gain. Notice that it is
dependent on L.

2006 by Fabian Kung Wai Lee

18

The Condition for Maximum GP (2)

If we could maximize |g2|, then GP can be maximized.

Using
*

L = U 2 + jV2

C2 = S 22 DS11

Because |S21| is
fixed, it is an
g2 =
important para2
2
2
2
1 S11 + L S 22 D 2 Re(L C2 )
meter in amplifier design. It
Using
sets the limit
2
L = U 22 + V22 of power gain.
Re(LC2 ) = U 2 Re (C2 ) V2 Im(C2 )

U 22

1 L

We can rewrite the above expression as (See Chapter 1 of [3]):

+ V22

2U 2 g 2 Re(C2 )

1 S 22 g 2 + D g 2

March 08

2V2 g 2 Im(C2 )
2

1 S 22 g 2 + D g 2

1 + g 2 S11 g 2
2

1 S 22 g 2 + D g 2

2006 by Fabian Kung Wai Lee

19

The Condition for Maximum GP (3)

This can be further expanded into (see extra notes) a circle equation in
the complex plane:
2

g 2 Re(C2 )
g 2 Im(C2 )
U 2 +
+ V2
= 22c (A.1)
2
2
2
2

1 S 22 g 2 + D g 2
1 S 22 g 2 + D g 2

Where

Coordinate for center of a circle

2

22c =

March 08

1 + g 2 S11 g 2
2

1 S 22 g 2 + D g 2

*
g 2 2 S 22 DS11

2
1 S 2 g + D 2 g
22
2
2

2006 by Fabian Kung Wai Lee

The radius
of a circle

(A.2)

20

10

The Condition for Maximum GP (4)

2c2 can be further simplified using the identity

2

Re 2 (C2 ) + Im 2 (C2 ) = S 22 DS11* = S12 S 21 + (1 S11 ) S 22 D

To

22c =

Where

1 2 K S12 S 21 g 2 + S12 S 21 g 2 2
2
1 S 2 g + D 2 g
22
2
2

K=

1 S 11 S 22 + D

2 S12 S 21

March 08

(A.3)

(A.4)

2006 by Fabian Kung Wai Lee

21

The Condition for Maximum GP (5)

Expression (A.1) again represents a circle in the complex L = U2+jV2

plane.
For each value of g2 , a circle can be drawn. Thus each circle
represents the contour of constant power gain GP for a given load ZL.
When 2c2=0, the power gain is the maximum.
Im( L )

Take note that there

is no minimum power
gain! GP can only
approach 0.

g0
g4

g3

g1
g2

What happens if we set g2 > go ?

Re( L )

March 08

2006 by Fabian Kung Wai Lee

22

11

The Condition for Maximum GP (6)

From (A.3), when 2c2=0:

g 2(max ) =

1
S12 S 21

K K 2 1

Since g2(max) only exist if K > 1, this implies maximum gain only exist
when the amplifier is unconditionally stable.
At K=1,
1
GP(max)
Maximum of the
g 2(max ) =
S12 S 21
maximum !!!
GMS

S
1
2
GP (max ) =
S 21 = 21
S12 S 21
S12

And this power gain is known as Maximum Stable Gain, GMS. This is
the maximum power gain that is achievable !!! It is often specified in
datasheet of an amplifier or amplification device and gives the
theoretical maximum gain achievable.
March 08

2006 by Fabian Kung Wai Lee

23

Corresponding Load Impedance for

Maximum Power Gain

From (A.1) and putting 2c2=0:

g 2 Re(C2 )
g 2 Im(C2 )
U 2m +
+ V2m
=0
2
2
2
2

S
g
+
D
g

S
g
+
D
g
22
2
2
22
2
2

g 2 Re(C2 )
U 2m +
=0
(A.5a)
2
2
1 S 22 g 2 + D g 2
V2m

March 08

g 2 Im(C2 )
2

=0

(A.5b)

1 S 22 g 2 + D g 2

2006 by Fabian Kung Wai Lee

24

12

Load Impedance for Maximum Power

Gain

By expanding (A.5a) and (A.5b) using g2(max) , we can show that:

Or

Lm = U 2m + jV2m

Lm =

1
A2 A2 2 4 B2 2
2 B2

A2 = 1 + S 22 S11 D
B2 =

)
1
2

(A.6a)

*
S 22 DS11

The proof is extremely tedious and will not be shown here. You can
refer to the extra hand-written notes for the proof.

March 08

2006 by Fabian Kung Wai Lee

25

Source Impedance for Maximum Power

Gain

Using

1 =

S11 DL
1 S 22L

We can obtain the corresponding 1 .

And requiring that S = 1 * , the source reflection coefficient can be
obtained.

Sm

1
A1 A12 4 B1 2
=
2 B1

A1 = 1 + S11 S 22 D

)
1
2

(A.6b)

*
B1 = S11 DS 22
March 08

2006 by Fabian Kung Wai Lee

26

13

2.0 Design for Fixed

Transducer Power Gain GT
The Constant GP Circle
Approach

March 08

2006 by Fabian Kung Wai Lee

27

Constant Power Gain Circles (1)

From the Dependency Diagram of previous chapter we see that GP

only depends on L. Indeed it is possible to determine the range of L
which would give a fixed GP, and this range actually falls on the locus
of a circle on Smith Chart.
From the equation (A.1) of Appendix 1, the radius and center of the
Constant Power Gain Circle are given by:
Normalized GP

C 2 = s22 Ds11*

g2 =

(2.1a)

GP

s21

(2.1b)

g 2 Re(C2 )
g 2 Im(C2 )
Center
T
=
+j
2
2
of circle G P 1 s 2 g + D 2 g
1 s22 g 2 + D g 2
22
2
2
Radius
of circle
March 08

(2.1c)

RG P =

1 2 K s12 s21 g 2 + s12 s21 g 2 2

2

1 s22 g 2 + D g 2
2006 by Fabian Kung Wai Lee

(2.1d)
28

14

Constant Power Gain Circles (2)

Im

ZL which fall on
this circle will
give similar GP
when connected
to the amplifiers
output
Re
0
TGp
L Plane

RGp
Constant
Gp circle

March 08

2006 by Fabian Kung Wai Lee

29

The Constant Gp Circle Approach

1. Use constant Gp circles together with LSC (if amplifier is

conditionally stable) to select the appropriate L.
2. From the selected L determine 1.
3. Conjugate match the input port (make sure s is in the stable
region).
Pro
4. Thus GT = Gp.

Zs

Pin = PAs
Input
Transform
-ation
Network

Vs

P
P
GT = L = L = G p
PAs Pin
March 08

PAo

PRs = 0

PAs

PL
PL
Output
Transform
-ation
Network

Amplifier

S*

2006 by Fabian Kung Wai Lee

ZL

30

15

Example 2.1 - Constant GP Amplifier

Design
A microwave BJT has S11=0.9<60o, S12=0.06<60o, S21=3<120o,
S22=0.82<-30o at 1.2GHz. Design a single-stage amplifier having 50
input and output transmission lines, with a required transducer power
gain between 40-50.

Solution

K = 0.902 and |D|=0.898, so the amplifier is conditionally stable.

S-parameters at 1.2GHz:

S11 = 0.45 + 0.7794i

S12 = 0.03 + 0.052i

S21 = 1.5 + 2.5981i

S22 = 0.7101 0.41i

D = 0.8191 + 0.369i

March 08

2006 by Fabian Kung Wai Lee

31

Example 2.1 Cont...

Plot Load Stability Circle
LSC
Since |S11|=0.9 < 1
and the LSC encloses
the origin, the stable region
is as indicated.
Plane for L

(S22 D S11
)

CL :=

)2 (

S22

)2

CL = 0.4003 + 0.4631i
S12 S21

RL :=

S22

)2 (

)2

Stable region

RL = 1.336
March 08

2006 by Fabian Kung Wai Lee

32

16

Example 2.1 Cont...

Constant GP Circle for Gp = 36

C2 := S22 D S11

Constant GP Circle for Gp = 45

C2 := S22 D S11

C2 = 0.0539 + 0.0624i

36

g2 :=

S21

g2 :=
g2 = 4

)2

Cgp :=
1 ( S22

)2 g2 + (

)2 g2

( g2 Re( C2) + i g2 Im( C2) )

K :=

1 ( S11

( S22

Rgp :=

1 ( S22

)2 g2 + (

1 2 K S12 S21 g2 +
1 (

g2 = 5

)2
1

Cgp :=

) 2 g2 + (

) 2 g2

( g2 Re( C2) + i g2 Im( C2) )

Cgp = 0.8262 0.956i

2

K :=

2 S21 S12

K = 0.902

S21

1 ( S22

Cgp = 0.4678 0.5413i

2

C2 = 0.0539+ 0.0624i

45

2
S22 ) g2 + (

)2 g2 = 0.4611

S12 S21 g2
D

S11

Rgp :=

)2 (

S22

)2 + (

)2

2 S21 S12

K = 0.902

)2

)2 g2

) 2 g2 + (

S22

1 2 K S12 S21 g2 +
1 ( S22

Rgp = 1.0161
March 08

g2 +

(
(

) 2 g2 = 0.3264

S12 S21 g2
D

)2

g2

Rgp = 1.3228
2006 by Fabian Kung Wai Lee

33

Example 2.1 Cont...

Draw the Constant Power Gain
Circles

LSC

Constant GP circles
on the L plane
GP=36

L=0.05<0.6177rad

GP=45
This point is
the farthest
away from the
LSC, it is less
susceptible to
parameter
changes.
March 08

2006 by Fabian Kung Wai Lee

34

17

Example 2.1 Cont...

Finding the corresponding load impedance

L = 0.05 < 0.6177rad = 0.041 j 0.029

Z L = Zo

1 + L
1 L

Z o = 50

Z L = 54.154 j 3.144

March 08

2006 by Fabian Kung Wai Lee

35

Example 2.1 Cont...

Finding the corresponding input impedance and taking its complex

conjugate:

1 =

Z1 =

S 11 D L
= 0 . 4472
1 S 22 L

+ j 0 . 785

1+ 1
Z = 10.129 + j84.704
1 1 o

Z s = Z1* = 10.129 j84.704

March 08

2006 by Fabian Kung Wai Lee

36

18

Example 2.1 Cont...

The block diagram...

10.13

54.154

-j84.70
Amplifier

Pin

-j3.144

PL

GT = PL/Pin = 45

March 08

2006 by Fabian Kung Wai Lee

37

Example 2.1 Cont...

Since actual Zs and ZL are 50 , impedance transformation network is

needed to change the 50 load impedance into 54.154 - j3.144, and
to 10.13-j84.70 for source impedance. We may use either 2 or 3
elements impedance transformation network. This is shown below for
the load network.
jX

RL=50, XL=0, Rs=54.154, Xs = -3.144

R
X = X L RL (Rs RL ) + L X s2 = 14.725
Rs

54.154-j3.144

jB

50 L = X = 1.95nH
2f
Rs R L
= 6.488 10 3
Rs X L + R L X s + Rs X
B
C=
= 0.86 pF
2f

B=

March 08

2006 by Fabian Kung Wai Lee

38

19

Example 2.1 Cont...

The final network...

1.95nH

1.26pF
50
3.38nH

0.86pF

Amplifier

ZS = 10.13-j84.70

50

ZL = 54.154-j3.144

GT = PL/Pin= 45
March 08

2006 by Fabian Kung Wai Lee

39

3.0 Design for Fixed

Transducer Power Gain GT
The Constant GA Circle
Approach

March 08

2006 by Fabian Kung Wai Lee

40

20

Preliminary

The approach is just the dual of the Constant GP Approach.

Instead of tuning L for fixed GP we tune s for fixed GA, therefore
constant GA circles and s plane for the Smith Chart are employed.
This approach is useful we need to take into account the final
transducer power gain GT and noise figure of the amplifier.

March 08

2006 by Fabian Kung Wai Lee

41

Constant Available Power Gain Circles

(1)

From the Dependency Diagram of previous chapter we see that GA

only depends on s. Indeed it is possible to determine the range of s
which would give a fixed GA, and this range actually falls on the locus
of a circle on Smith Chart.
The derivation is similar to the procedures in Appendix 1, see [4] for
Normalized GA
the details.

C1 = s11 Ds22*
Center
of circle

TG A =

Radius
of circle

RG A =

g1 =

(3.1a)

g1 Re(C1 )
2

s21

g1 Im(C1 )

1 + s11 g1 D g1

GA

1 + s11 g1 D g1

(3.1b)
(3.1c)

March 08

1 2 K s12 s21 g1 + s12 s21 g12

2

1 + s11 g1 D g1
2006 by Fabian Kung Wai Lee

(3.1d)
42

21

Constant Available Power Gain Circles

(2)
Im

Zs which fall on
this circle will
give similar GA
when connected
to the amplifiers
output
Re
0
TGA
s Plane

RGA
Constant
GA circle

March 08

2006 by Fabian Kung Wai Lee

43

The Constant GA Circle Approach

1. Use constant GA circles together with SSC (if amplifier is

conditionally stable) to select the appropriate s.
2. From the selected s determine 2.
3. Conjugate match the output port (make sure L is in stable region).
Pro = 0
4. GT = GA.

Zs

PL = PAo

Pin
Input
Transform
-ation
Network

Vs

GT =
March 08

PAo

PRs

PAs

Output
Transform
-ation
Network

Amplifier

PL

2*

ZL

P
PL
= Ao = G A
PAs
PAs
2006 by Fabian Kung Wai Lee

44

22