S - Parameter Basics PDF

Download as pdf or txt
Download as pdf or txt
You are on page 1of 24
At a glance
Powered by AI
The document discusses scattering parameters (S-parameters) which characterize linear two-port networks at microwave frequencies. S-parameters relate incoming and outgoing wave amplitudes and can be directly measured using network analyzers.

Linear two-port networks can be characterized by their transfer matrix, impedance matrix, admittance matrix, and scattering matrix. The transfer matrix relates voltages and currents between ports while the impedance matrix relates voltages to currents.

Scattering parameters (S-parameters) relate the outgoing wave amplitudes to the incoming wave amplitudes at each port. S11 and S22 represent reflection coefficients while S21 and S12 represent transmission coefficients. S-parameters can be directly measured using network analyzers.

13

S-Parameters

13.1 Scattering Parameters


Linear two-port (and multi-port) networks are characterized by a number of equivalent
circuit parameters, such as their transfer matrix, impedance matrix, admittance matrix,
and scattering matrix. Fig. 13.1.1 shows a typical two-port network.

Fig. 13.1.1 Two-port network.

The transfer matrix, also known as the ABCD matrix, relates the voltage and current
at port 1 to those at port 2, whereas the impedance matrix relates the two voltages
V1 , V2 to the two currents I1 , I2 :†
    
V1 A B V2
= (transfer matrix)
I1 C D I2
     (13.1.1)
V1 Z11 Z12 I1
= (impedance matrix)
V2 Z21 Z22 −I2

Thus, the transfer and impedance matrices are the 2×2 matrices:
   
A B Z11 Z12
T= , Z= (13.1.2)
C D Z21 Z22

The admittance matrix is simply the inverse of the impedance matrix, Y = Z−1 . The
scattering matrix relates the outgoing waves b1 , b2 to the incoming waves a1 , a2 that
are incident on the two-port:
† In the figure, I flows out of port 2, and hence −I flows into it. In the usual convention, both currents
2 2
I1 , I2 are taken to flow into their respective ports.
13.1. Scattering Parameters 523 524 13. S-Parameters

40 GHz. Frequency resolution is typically 1 Hz and the results can be displayed either
       on a Smith chart or as a conventional gain versus frequency graph.
b1 S11 S12 a1 S11 S12
= , S= (scattering matrix) (13.1.3)
b2 S21 S22 a2 S21 S22

The matrix elements S11 , S12 , S21 , S22 are referred to as the scattering parameters or
the S-parameters. The parameters S11 , S22 have the meaning of reflection coefficients,
and S21 , S12 , the meaning of transmission coefficients.
The many properties and uses of the S-parameters in applications are discussed in
[956–995]. One particularly nice overview is the HP application note AN-95-1 by Ander-
son [971] and is available on the web [1323].
We have already seen several examples of transfer, impedance, and scattering ma-
trices. Eq. (10.7.6) or (10.7.7) is an example of a transfer matrix and (10.8.1) is the
corresponding impedance matrix. The transfer and scattering matrices of multilayer
Fig. 13.1.2 Device under test connected to network analyzer.
structures, Eqs. (6.6.23) and (6.6.37), are more complicated examples.
The traveling wave variables a1 , b1 at port 1 and a2 , b2 at port 2 are defined in terms Fig. 13.1.3 shows more details of the connection. The generator and load impedances
of V1 , I1 and V2 , I2 and a real-valued positive reference impedance Z0 as follows: are configured by the network analyzer. The connections can be reversed, with the
generator connected to port 2 and the load to port 1.
V1 + Z0 I1 V2 − Z0 I2
a1 =  a2 = 
2 Z0 2 Z0
(traveling waves) (13.1.4)
V1 − Z0 I1 V2 + Z0 I2
b1 =  b2 = 
2 Z0 2 Z0

The definitions at port 2 appear different from those at port 1, but they are really
the same if expressed in terms of the incoming current −I2 :

V2 − Z0 I2 V2 + Z0 (−I2 )
a2 =  = 
2 Z0 2 Z0
V2 + Z0 I2 V2 − Z0 (−I2 )
b2 =  = 
2 Z0 2 Z0 Fig. 13.1.3 Two-port network under test.

The term traveling waves is justified below. Eqs. (13.1.4) may be inverted to express The two line segments of lengths l1 , l2 are assumed to have characteristic impedance
the voltages and currents in terms of the wave variables: equal to the reference impedance Z0 . Then, the wave variables a1 , b1 and a2 , b2 are
  recognized as normalized versions of forward and backward traveling waves. Indeed,
V1 = Z0 (a1 + b1 ) V2 = Z0 (a2 + b2 ) according to Eq. (10.7.8), we have:
(13.1.5) V1 + Z0 I1 1 V2 − Z0 I2 1
1 1
I1 =  (a1 − b1 ) I2 =  (b2 − a2 ) a1 =  =  V1+ a2 =  =  V2−
Z0 Z0 2 Z0 Z0 2 Z0 Z0
(13.1.6)
In practice, the reference impedance is chosen to be Z0 = 50 ohm. At lower fre- V1 − Z0 I1 1 V2 + Z0 I2 1
b1 =  =  V1− b2 =  =  V2+
quencies the transfer and impedance matrices are commonly used, but at microwave 2 Z0 Z0 2 Z0 Z0
frequencies they become difficult to measure and therefore, the scattering matrix de- Thus, a1 is essentially the incident wave at port 1 and b1 the corresponding reflected
scription is preferred. wave. Similarly, a2 is incident from the right onto port 2 and b2 is the reflected wave
The S-parameters can be measured by embedding the two-port network (the device- from port 2.
under-test, or, DUT) in a transmission line whose ends are connected to a network ana- The network analyzer measures the waves a1 , b1 and a2 , b2 at the generator and
lyzer. Fig. 13.1.2 shows the experimental setup. load ends of the line segments, as shown in Fig. 13.1.3. From these, the waves at the
A typical network analyzer can measure S-parameters over a large frequency range, inputs of the two-port can be determined. Assuming lossless segments and using the
for example, the HP 8720D vector network analyzer covers the range from 50 MHz to propagation matrices (10.7.7), we have:
13.1. Scattering Parameters 525 526 13. S-Parameters

The actual measurements of the S-parameters are made by connecting to a matched


          load, ZL = Z0 . Then, there will be no reflected waves from the load, a2 = 0, and the
a1 e−jδ1 0 a1 a2 e−jδ2 0 a2 S-matrix equations will give:
= , = (13.1.7)
b1 0 e jδ1 b1 b2 0 e jδ2 b2

b1 

b1 = S11 a1 + S12 a2 = S11 a1 ⇒ S11 = = reflection coefficient
where δ1 = βll and δ2 = βl2 are the phase lengths of the segments. Eqs. (13.1.7) can be a1 ZL =Z0
rearranged into the forms: 
b2 

          b2 = S21 a1 + S22 a2 = S21 a1 ⇒ S21 = = transmission coefficient
b1 b1 a1 a1 ejδ1 0 a1 ZL =Z0
=D , =D , D=
b2 b2 a2 a2 0 jδ2
e
Reversing the roles of the generator and load, one can measure in the same way the
The network analyzer measures the corresponding S-parameters of the primed vari- parameters S12 and S22 .
ables, that is,

       13.2 Power Flow


b1 
S11 
S12 a1 

S11 
S12
= , S = (measured S-matrix) (13.1.8)
b2 
S21 
S22 a2 
S21 
S22 Power flow into and out of the two-port is expressed very simply in terms of the traveling
wave amplitudes. Using the inverse relationships (13.1.5), we find:
The S-matrix of the two-port can be obtained then from:
1 1 1
        Re[V1∗ I1 ] = |a1 |2 − |b1 |2
b1 b1 a1 a1 2 2 2
=D = DS = DS D ⇒ S = DS D (13.2.1)
b2 b2 a2 a2 1 1 1
− Re[V2∗ I2 ] = |a2 |2 − |b2 |2
2 2 2
or, more explicitly,
The left-hand sides represent the power flow into ports 1 and 2. The right-hand sides
     
jδ1   jδ1 represent the difference between the power incident on a port and the power reflected
S11 S12 e 0 S11 S12 e 0
=   from it. The quantity Re[V2∗ I2 ]/2 represents the power transferred to the load.
S21 S22 0 ejδ2 S21 S22 0 ejδ2
  (13.1.9) Another way of phrasing these is to say that part of the incident power on a port
 
S11 e2jδ1 S12 ej(δ1 +δ2 ) gets reflected and part enters the port:
=  
S21 ej(δ1 +δ2 ) S22 e2jδ2
1 1 1
Thus, changing the points along the transmission lines at which the S-parameters |a1 |2 = |b1 |2 + Re[V1∗ I1 ]
2 2 2
are measured introduces only phase changes in the parameters. (13.2.2)
1 1 1
Without loss of generality, we may replace the extended circuit of Fig. 13.1.3 with the |a2 |2 = |b2 |2 + Re[V2∗ (−I2 )]
2 2 2
one shown in Fig. 13.1.4 with the understanding that either we are using the extended 
two-port parameters S , or, equivalently, the generator and segment l1 have been re- One of the reasons for normalizing the traveling wave amplitudes by Z0 in the
placed by their Thévenin equivalents, and the load impedance has been replaced by its definitions (13.1.4) was precisely this simple way of expressing the incident and reflected
propagated version to distance l2 . powers from a port.
If the two-port is lossy, the power lost in it will be the difference between the power
entering port 1 and the power leaving port 2, that is,

1 1 1 1 1 1
Ploss = Re[V1∗ I1 ]− Re[V2∗ I2 ]= |a1 |2 + |a2 |2 − |b1 |2 − |b2 |2
2 2 2 2 2 2

Noting that a† a = |a1 |2 + |a2 |2 and b† b = |b1 |2 + |b2 |2 , and writing b† b = a† S† Sa,
we may express this relationship in terms of the scattering matrix:

1 † 1 1 1 1
Ploss = a a − b† b = a† a − a† S† Sa = a† (I − S† S)a (13.2.3)
Fig. 13.1.4 Two-port network connected to generator and load. 2 2 2 2 2
13.3. Parameter Conversions 527 528 13. S-Parameters

For a lossy two-port, the power loss is positive, which implies that the matrix I − S† S Explicitly, we have:
must be positive definite. If the two-port is lossless, Ploss = 0, the S-matrix will be   −1
unitary, that is, S† S = I. Z11 − Z0 Z12 Z11 + Z0 Z12
S=
We already saw examples of such unitary scattering matrices in the cases of the equal Z21 Z22 − Z0 Z21 Z22 + Z0
travel-time multilayer dielectric structures and their equivalent quarter wavelength mul-    
Z11 − Z0 Z12 1 Z22 + Z0 −Z12
tisection transformers. =
Z21 Z22 − Z0 Dz −Z21 Z11 + Z0

where Dz = det(Z + Z0 I)= (Z11 + Z0 )(Z22 + Z0 )−Z12 Z21 . Multiplying the matrix
13.3 Parameter Conversions factors, we obtain:

It is straightforward to derive the relationships that allow one to pass from one param-  
eter set to another. For example, starting with the transfer matrix, we have: 1 (Z11 − Z0 )(Z22 + Z0 )−Z12 Z21 2Z12 Z0
S= (13.3.5)
Dz 2Z21 Z0 (Z11 + Z0 )(Z22 − Z0 )−Z12 Z21
D 
1 A AD − BC
V1 = AV2 + BI2 V1 = A I1 − I2 + BI2 = I1 − I2
C C C C Similarly, the inverse relationship gives:

I1 = CV2 + DI2 D
1
V2 = I1 − I2  
C C Z0 (1 + S11 )(1 − S22 )+S12 S21 2S12
Z= (13.3.6)
The coefficients of I1 , I2 are the impedance matrix elements. The steps are reversible, Ds 2S21 (1 − S11 )(1 + S22 )+S12 S21
and we summarize the final relationships below: where Ds = det(I − S)= (1 − S11 )(1 − S22 )−S12 S21 . Expressing the impedance param-
    eters in terms of the transfer matrix parameters, we also find:
Z11 Z12 1 A AD − BC
Z= = ⎡ ⎤
Z21 Z22 C 1 D B
    (13.3.1) 1 ⎢
A+ − CZ0 − D 2(AD − BC) ⎥
Z11 Z22 − Z12 Z21 S= ⎢ Z0 ⎥ (13.3.7)
A B 1 Z11 ⎣ B ⎦
T= = Da 2 −A + − CZ0 + D
C D Z21 1 Z22 Z0
We note the determinants det(T)= AD − BC and det(Z)= Z11 Z22 − Z12 Z21 . The B
where Da = A + + CZ0 + D.
relationship between the scattering and impedance matrices is also straightforward to Z0
derive. We define the 2×1 vectors:
        13.4 Input and Output Reflection Coefficients
V1 I1 a1 b1
V= , I= , a= , b= (13.3.2)
V2 −I2 a2 b2 When the two-port is connected to a generator and load as in Fig. 13.1.4, the impedance
Then, the definitions (13.1.4) can be written compactly as: and scattering matrix equations take the simpler forms:

1 1 V1 = Zin I1 b1 = Γin a1
a=  (V + Z0 I)=  (Z + Z0 I)I
2 Z0 2 Z0  (13.4.1)
(13.3.3) V2 = ZL I2 a2 = ΓL b2
1 1
b=  (V − Z0 I)=  (Z − Z0 I)I
2 Z0 2 Z0 where Zin is the input impedance at port 1, and Γin , ΓL are the reflection coefficients at
where we used the impedance matrix relationship V = ZI and defined the 2×2 unit port 1 and at the load:
matrix I. It follows then, Zin − Z0 ZL − Z0
Γin = , ΓL = (13.4.2)
1 1 Zin + Z0 ZL + Z0
 I = (Z + Z0 I)−1 a ⇒ b=  (Z − Z0 I)I = (Z − Z0 I)(Z + Z0 I)−1 a
2 Z0 2 Z0 The input impedance and input reflection coefficient can be expressed in terms of
the Z- and S-parameters, as follows:
Thus, the scattering matrix S will be related to the impedance matrix Z by
Z12 Z21 S12 S21 ΓL
S = (Z − Z0 I)(Z + Z0 I)−1  Z = (I − S)−1 (I + S)Z0 (13.3.4) Zin = Z11 −  Γin = S11 + (13.4.3)
Z22 + ZL 1 − S22 ΓL
13.4. Input and Output Reflection Coefficients 529 530 13. S-Parameters

The equivalence of these two expressions can be shown by using the parameter
conversion formulas of Eqs. (13.3.5) and (13.3.6), or they can be shown indirectly, as
follows. Starting with V2 = ZL I2 and using the second impedance matrix equation, we
can solve for I2 in terms of I1 :

Z21
V2 = Z21 I1 − Z22 I2 = ZL I2 ⇒ I2 = I1 (13.4.4)
Z22 + ZL
Then, the first impedance matrix equation implies:
Fig. 13.4.1 Input and output equivalent circuits.
 
Z12 Z21
V1 = Z11 I1 − Z12 I2 = Z11 − I1 = Zin I1
Z22 + ZL
Similarly, the generator and the two-port can be replaced by a Thévenin equivalent
Starting again with V2 = ZL I2 we find for the traveling waves at port 2: circuit connected at port 2. By determining the open-circuit voltage and short-circuit
current at port 2, we find the corresponding Thévenin parameters in terms of the impe-
V2 − Z0 I2 ZL − Z0
a2 =  =  I2 dance parameters:
2 Z0 2 Z0 ZL − Z0
⇒ a2 = b2 = ΓL b2
V2 + Z0 I2 ZL + Z0 ZL + Z0 Z21 VG Z12 Z21
b2 =  =  I2 Vth = , Zth = Zout = Z22 − (13.4.8)
2 Z0 2 Z0 Z11 + ZG Z11 + ZG

Using V1 = Zin I1 , a similar argument implies for the waves at port 1:


13.5 Stability Circles
V1 + Z0 I1 Zin + Z0
a1 =  =  I1
2 Z0 2 Z0 Zin − Z0 In discussing the stability conditions of a two-port in terms of S-parameters, the follow-
⇒ b1 = a1 = Γin a1
V1 − Z0 I1 Zin − Z0 Zin + Z0 ing definitions of constants are often used:
b1 =  =  I1
2 Z0 2 Z0

It follows then from the scattering matrix equations that: Δ = det(S)= S11 S22 − S12 S21

1 − |S11 |2 − |S22 |2 + |Δ|2


K= (Rollett stability factor)
S21 2|S12 S21 |
b2 = S21 a1 + S22 a2 = S22 a1 + S22 ΓL b2 ⇒ b2 = a1 (13.4.5)
1 − S22 ΓL 1 − |S11 |2
μ1 = ∗ (Edwards-Sinsky stability parameter)
|S22 − ΔS11 | + |S12 S21 |
which implies for b1 :
  1 − |S22 |2 (13.5.1)
S12 S21 ΓL μ2 = ∗
b1 = S11 a1 + S12 a2 = S11 a1 + S12 ΓL b2 = S11 + a1 = Γin a1 |S11 − ΔS22 | + |S12 S21 |
1 − S22 ΓL
B1 = 1 + |S11 |2 − |S22 |2 − |Δ|2
Reversing the roles of generator and load, we obtain the impedance and reflection
coefficients from the output side of the two-port: B2 = 1 + |S22 |2 − |S11 |2 − |Δ|2

C1 = S11 − ΔS22 , D1 = |S11 |2 − |Δ|2
Z12 Z21 S12 S21 ΓG
Zout = Z22 −  Γout = S22 + (13.4.6) ∗
Z11 + ZG 1 − S11 ΓG C2 = S22 − ΔS11 , D2 = |S22 |2 − |Δ|2

where The quantity K is the Rollett stability factor [967], and μ1 , μ2 , the Edwards-Sinsky
stability parameters [970]. The following identities hold among these constants:
Zout − Z0 ZG − Z0
Γout = , ΓG = (13.4.7)
Zout + Z0 ZG + Z0 B21 − 4|C1 |2 = B22 − 4|C2 |2 = 4|S12 S21 |2 (K2 − 1)
The input and output impedances allow one to replace the original two-port circuit  
|C1 |2 = |S12 S21 |2 + 1 − |S22 |2 D1 (13.5.2)
of Fig. 13.1.4 by simpler equivalent circuits. For example, the two-port and the load can
 
be replaced by the input impedance Zin connected at port 1, as shown in Fig. 13.4.1. |C2 |2 = |S12 S21 |2 + 1 − |S11 |2 D2
13.5. Stability Circles 531 532 13. S-Parameters

For example, noting that S12 S21 = S11 S22 − Δ, the last of Eqs. (13.5.2) is a direct portions of the unit-circles may lie within the stability regions and such ΓG , ΓL will lead
consequence of the identity: to a stable input and output impedances.
The connection of the stability regions to the stability circles is brought about by the
  
|A − BC|2 − |B − AC∗ |2 = 1 − |C|2 |A|2 − |B|2 (13.5.3) following identities, which can be proved easily using Eqs. (13.5.1)–(13.5.8):

We define also the following parameters, which will be recognized as the centers and |ΓL − cL |2 − rL2
radii of the source and load stability circles: 1 − |Γin |2 = D2
|1 − S22 ΓL |2
(13.5.9)
C∗
1 |S12 S21 | |ΓG − cG |2 − rG 2
cG = , rG = (source stability circle) (13.5.4) 1 − |Γout | = 2
D1
D1 |D1 | |1 − S11 ΓG |2
For example, the first can be shown starting with Eq. (13.5.8) and using the definitions
C∗
2 |S12 S21 |
cL = , rL = (load stability circle) (13.5.5) (13.5.5) and the relationship (13.5.6):
D2 |D2 |
 
 S11 − ΔΓL 2
 = |S11 − ΔΓL | − |1 − S22 ΓL |
2 2
They satisfy the following relationships, which are consequences of the last two of
1 − |Γin |2 = 1 − 
 
Eqs. (13.5.2) and the definitions (13.5.4) and (13.5.5): 1−S Γ22 L |1 − S Γ |2 22 L
 
  |S22 | − |Δ| ∗ ∗
|ΓL | − (S22 − ΔS11 )ΓL − (S22 − Δ∗ S11 )Γ∗
L + 1 − |S11 |
2 2 2 2
2
1 − |S11 |2 = |cL |2 − rL D2 =
  (13.5.6) |1 − S22 ΓL |2
2
1 − |S22 |2 = |cG |2 − rG D1
D2 |ΓL |2 − C2 ΓL − C∗ ∗
2 ΓL + 1 − |S11 |
2
=
We note also that using Eqs. (13.5.6), the stability parameters μ1 , μ2 can be written as: |1 − S22 ΓL |2

     
μ1 = |cL | − rL sign(D2 ) D2 |ΓL | − c∗
2 ∗ ∗
L ΓL − cL ΓL + |cL | − rL
2 2
D2 |ΓL − cL |2 − rL2
= =
  (13.5.7) |1 − S22 ΓL |2 |1 − S22 ΓL |2
μ2 = |cG | − rG sign(D1 )
It follows from Eq. (13.5.9) that the load stability region is defined by the conditions:
For example, we have:
 
    1 − |Γin |2 > 0  |ΓL − cL |2 − rL2 D2 > 0
1 − |S11 | 2
D2 |cL | − rL2 2
D2 |cL | − rL 2
D2 
2 
μ1 = = =   = |cL | − rL
|C2 | + |S12 S21 | |D2 ||cL | + |D2 |rL |D2 | |cL | + rL |D2 | Depending on the sign of D2 , these are equivalent to the outside or the inside of the
load stability circle of center cL and radius rL :
We finally note that the input and output reflection coefficients can be written in the
alternative forms: |ΓL − cL | > rL , if D2 > 0
(load stability region) (13.5.10)
S12 S21 ΓL S11 − ΔΓL |ΓL − cL | < rL , if D2 < 0
Γin = S11 + =
1 − S22 ΓL 1 − S22 ΓL
(13.5.8) The boundary of the circle |ΓL − cL | = rL corresponds to |Γin | = 1. The complement
S12 S21 ΓG S22 − ΔΓG of these regions corresponds to the unstable region with |Γin | > 1. Similarly, we find
Γout = S22 + =
1 − S22 ΓG 1 − S11 ΓG for the source stability region:
Next, we discuss the stability conditions. The two-port is unconditionally stable if
any generator and load impedances with positive resistive parts RG , RL , will always lead |ΓG − cG | > rG , if D1 > 0
(source stability region) (13.5.11)
to input and output impedances with positive resistive parts Rin , Rout . |ΓG − cG | < rG , if D1 < 0
Equivalently, unconditional stability requires that any load and generator with |ΓL | <
1 and |ΓG | < 1 will result into |Γin | < 1 and |Γout | < 1. In order to have unconditional stability, the stability regions must contain the unit-
The two-port is termed potentially or conditionally unstable if there are |ΓL | < 1 and circle in its entirety. If D2 > 0, the unit-circle and load stability circle must not overlap
|ΓG | < 1 resulting into |Γin | ≥ 1 and/or |Γout | ≥ 1. at all, as shown in Fig. 13.5.1. Geometrically, the distance between the points O and A in
The load stability region is the set of all ΓL that result into |Γin | < 1, and the source the figure is (OA)= |cL | − rL . The non-overlapping of the circles requires the condition
stability region, the set of all ΓG that result into |Γout | < 1. (OA)> 1, or, |cL | − rL > 1.
In the unconditionally stable case, the load and source stability regions contain the If D2 < 0, the stability region is the inside of the stability circle, and therefore, the
entire unit-circles |ΓL | < 1 or |ΓG | < 1. However, in the potentially unstable case, only unit-circle must lie within that circle. This requires that (OA)= rL − |cL | > 1, as shown
in Fig. 13.5.1.
13.5. Stability Circles 533 534 13. S-Parameters

circle, as shown in Fig. 13.5.2. The portion of the unit-circle that does not lie within the
stability region will correspond to an unstable Zin .
There exist several other unconditional stability criteria that are equivalent to the
single criterion μ1 > 1. They all require that the Rollett stability factor K be greater
than unity, K > 1, as well as one other condition. Any one of the following criteria are
necessary and sufficient for unconditional stability [968]:

K >1 and |Δ| < 1


K >1 and B1 > 0
K >1 and B2 > 0 (stability conditions) (13.5.13)
K >1 and |S12 S21 | < 1 − |S11 |2
K >1 and |S12 S21 | < 1 − |S22 |2
Fig. 13.5.1 Load stability regions in the unconditionally stable case. Their equivalence to μ1 > 1 has been shown in [970]. In particular, it follows from
the last two conditions that unconditional stability requires |S11 | < 1 and |S22 | < 1.
  These are necessary but not sufficient for stability.
These two conditions can be combined into sign(D2 ) |cL | − rL > 1. But, that is
A very common circumstance in practice is to have a potentially unstable two-port,
equivalent to μ1 > 1 according to Eq. (13.5.7). Geometrically, the parameter μ1 repre- 
but with |S11 | < 1 and |S22 | < 1. In such cases, Eq. (13.5.6) implies D2 |cL |2 − rL2 )> 0,
sents the distance (OA). Thus, the condition for the unconditional stability of the input  2
and the lack of stability requires μ1 = sign(D2 ) |cL | − rL )< 1.
2
is equivalent to:
Therefore, if D2 > 0, then we must have |cL |2 − rL2 > 0 and |cL | − rL < 1, which
combine into the inequality rL < |cL | < rL + 1. This is depicted in the left picture of
μ1 > 1 (unconditional stability condition) (13.5.12)
Fig. 13.5.2. The geometrical distance (OA)= |cL | − rL satisfies 0 < (OA)< 1, so that
It has been shown by Edwards and Sinsky [970] that this single condition (or, alter- stability circle partially overlaps with the unit-circle but does not enclose its center.
natively, the single condition μ2 > 1) is necessary and sufficient for the unconditional On the other hand, if D2 < 0, the two conditions require |cL |2 −rL2 < 0 and rL −|cL | <
stability of both the input and output impedances of the two-port. Clearly, the source 1, which imply |cL | < rL < |cL | + 1. This is depicted in the right Fig. 13.5.2. The
stability regions will be similar to those of Fig. 13.5.1. geometrical distance (OA)= rL − |cL | again satisfies 0 < (OA)< 1, but now the center
If the stability condition is not satisfied, that is, μ1 < 1, then only that portion of the of the unit-circle lies within the stability circle, which is also the stability region.
unit-circle that lies within the stability region will be stable and will lead to stable input We have written a number of MATLAB functions that facilitate working with S-
and output impedances. Fig. 13.5.2 illustrates such a potentially unstable case. parameters. They are described in detail later on:

smat reshape S-parameters into S-matrix


sparam calculate stability parameters
sgain calculate transducer, available, operating, and unilateral power gains
smatch calculate simultaneous conjugate match for generator and load
gin,gout calculate input and output reflection coefficients
smith draw a basic Smith chart
smithcir draw a stability or gain circle on Smith chart
sgcirc determine stability and gain circles
nfcirc determine noise figure circles
nfig calculate noise figure

The MATLAB function sparam calculates the stability parameters μ1 , K, |Δ|, B1 , B2 ,


as well as the parameters C1 , C2 , D1 , D2 . It has usage:
Fig. 13.5.2 Load stability regions in potentially unstable case.
[K,mu,D,B1,B2,C1,C2,D1,D2] = sparam(S); % stability parameters

If D2 > 0, then μ1 < 1 is equivalent to |cL | − rL < 1, and if D2 < 0, it is equivalent


The function sgcirc calculates the centers and radii of the source and load stability
to rL − |cL | < 1. In either case, the unit-circle is partially overlapping with the stability
circles. It also calculates gain circles to be discussed later on. Its usage is:
13.5. Stability Circles 535 536 13. S-Parameters

[cL,rL] = sgcirc(S,’l’); % load or Zin stability circle The computed stability parameters at 1 GHz were:
[cG,rG] = sgcirc(S,’s’); % source or Zout stability circle

[K, μ1 , |Δ|, B1 , B2 , D1 , D2 ]= [0.781, 0.847, 0.250, 0.928, 0.947, 0.168, 0.178]


The MATLAB function smith draws a basic Smith chart, and the function smithcir
draws the stability circles: The transistor is potentially unstable because K < 1 even though |Δ| < 1, B1 > 0, and
B2 > 0. The load and source stability circle centers and radii were:
smith(n); % draw four basic types of Smith charts, n = 1, 2, 3, 4
smith; % default Smith chart corresponding to n = 3
cL = 2.978∠51.75o , rL = 2.131
smithcir(c,r,max,width); % draw circle of center c and radius r cG = 3.098∠162.24 , o
rG = 2.254
smithcir(c,r,max); % equivalent to linewidth width=1
smithcir(c,r); % draw full circle with linewidth width=1 Because both D1 and D2 are positive, both stability regions will be the portion of the Smith
chart that lies outside the stability circles. For 2 GHz, we find:
The parameter max controls the portion of the stability circle that is visible outside
the Smith chart. For example, max = 1.1 will display only that portion of the circle that
[K, μ1 , |Δ|, B1 , B2 , D1 , D2 ]= [1.089, 1.056, 0.103, 1.025, 0.954, 0.201, 0.166]
has |Γ| < 1.1.

Example 13.5.1: The Hewlett-Packard AT-41511 NPN bipolar transistor has the following S- cL = 2.779∠50.12o , rL = 1.723
parameters at 1 GHz and 2 GHz [1324]:
cG = 2.473∠−159.36o , rG = 1.421
S11 = 0.48∠−149o , S21 = 5.189∠89o , S12 = 0.073∠43o , S22 = 0.49∠−39o
S11 = 0.46∠162o , S21 = 2.774∠59o , S12 = 0.103∠45o , S22 = 0.42∠−47o The transistor is stable at 2 GHz, with both load and source stability circles being com-
pletely outside the unit-circle. 

Determine the stability parameters, stability circles, and stability regions.
Problem 13.2 presents an example for which the D2 parameter is negative, so that
Solution: The transistor is potentially unstable at 1 GHz, but unconditionally stable at 2 GHz. the stability regions will be the insides of the stability circles. At one frequency, the
The source and load stability circles at 1 GHz are shown in Fig. 13.5.3.
unit-circle is partially overlapping with the stability circle, while at another frequency,
it lies entirely within the stability circle.

13.6 Power Gains


The amplification (or attenuation) properties of the two-port can be deduced by com-
paring the power Pin going into the two-port to the power PL coming out of the two-port
and going into the load. These were given in Eq. (13.2.1) and we rewrite them as:

1 1
Pin = Re[V1∗ I1 ]= Rin |I1 |2 (power into two-port)
Fig. 13.5.3 Load and source stability circles at 1 GHz. 2 2
(13.6.1)
1 1
The MATLAB code used to generate this graph was: PL = Re[V2∗ I2 ]= RL |I2 |2 (power out of two-port and into load)
2 2

S = smat([0.48 -149 5.189 89 0.073 43 0.49 -39]); % form S-matrix where we used V1 = Zin I1 , V2 = ZL I2 , and defined the real parts of the input and
load impedances by Rin = Re(Zin ) and RL = Re(ZL ). Using the equivalent circuits of
[K,mu,D,B1,B2,C1,C2,D1,D2] = sparam(S); % stability parameters Fig. 13.4.1, we may write I1 , I2 in terms of the generator voltage VG and obtain:
[cL,rL] = sgcirc(S,’l’); % stability circles
1 |VG |2 Rin
[cG,rG] = sgcirc(S,’s’); Pin =
2 |Zin + ZG |2
smith; % draw basic Smith chart (13.6.2)
smithcir(cL, rL, 1.1, 1.5); 1 |Vth |2 RL 1 |VG |2 RL |Z21 |2
% draw stability circles
PL = =  
smithcir(cG, rG, 1.1, 1.5); 2 |Zout + ZL |2 2 (Z11 + ZG )(Zout + ZL )2

13.6. Power Gains 537 538 13. S-Parameters

Using the identities of Problem 13.1, PL can also be written in the alternative forms: Three widely used definitions for the power gain of the two-port network are the
transducer power gain GT , the available power gain Ga , and the power gain Gp , also
called the operating gain. They are defined as follows:
1 |VG |2 RL |Z21 |2 1 |VG |2 RL |Z21 |2
PL =   =   (13.6.3)
2 (Z22 + ZL )(Zin + ZG )2 2 (Z11 + ZG )(Z22 + ZL )−Z12 Z21 2
power out of network PL
GT = = (transducer power gain)
The maximum power that can be delivered by the generator to a connected load maximum power in PavG
is called the available power of the generator, PavG , and is obtained when the load is maximum power out PavN
∗ Ga = = (available power gain) (13.6.8)
conjugate-matched to the generator, that is, PavG = Pin when Zin = ZG . maximum power in PavG
Similarly, the available power from the two-port network, PavN , is the maximum power out of network PL
power that can be delivered by the Thévenin-equivalent circuit of Fig. 13.4.1 to a con- Gp = = (operating power gain)
∗ ∗
power into network Pin
nected load, that is, PavN = PL when ZL = Zth = Zout . It follows then from Eq. (13.6.2)
that the available powers will be: Each gain is expressible either in terms of the Z-parameters of the two-port, or in
terms of its S-parameters. In terms of Z-parameters, the transducer gain is given by the
|VG |2 following forms, obtained from the three forms of PL in Eqs. (13.6.2) and (13.6.3):
PavG = max Pin = (available power from generator)
8RG
(13.6.4) 4RG RL |Z21 |2
|Vth |2 GT =  
PavN = max PL = (available power from network) (Z22 + ZL )(Zin + ZG )2
8Rout
Using Eq. (13.4.8), PavN can also be written as: 4RG RL |Z21 |2
=   (13.6.9)
(Z11 + ZG )(Zout + ZL )2
 2
|VG | 
2
 Z21 

PavN =   (13.6.5) 4RG RL |Z21 |2
8Rout Z11 + ZG =  
(Z11 + ZG )(Z22 + ZL )−Z12 Z21 2
The powers can be expressed completely in terms of the S-parameters of the two-
And, in terms of the S-parameters:
port and the input and output reflection coefficients. With the help of the identities of
Problem 13.1, we find the alternative expressions for Pin and PL :
1 − |ΓG |2 1 − |ΓL |2
  GT = |S21 |2
|VG |2 1 − |Γin |2 |1 − ΓG |2 |1 − Γin ΓG |2 |1 − S22 ΓL |2
Pin =
8Z0 |1 − Γin ΓG |2 1 − |ΓG |2 1 − |ΓL |2
  = |S21 |2 (13.6.10)
|VG | 1 − |ΓL |2 |1 − ΓG |2 |S21 |2
2 |1 − S11 ΓG |2 |1 − Γout ΓL |2
PL =  
8Z0 (1 − Γin ΓG )(1 − S22 ΓL )2 (1 − |ΓG |2 )|S21 |2 (1 − |ΓL |2 )
  (13.6.6) =  
|VG |2 1 − |ΓL |2 |1 − ΓG |2 |S21 |2 (1 − S11 ΓG )(1 − S22 ΓL )−S12 S21 ΓG ΓL 2
=  
8Z0 (1 − Γout ΓL )(1 − S11 ΓG )2 Similarly, we have for Ga and Gp :
 
|VG |2 1 − |ΓL |2 |1 − ΓG |2 |S21 |2  2
=   RG  
 Z21  = 1 − |ΓG | |S21 |2
2
8Z0 (1 − S11 ΓG )(1 − S22 ΓL )−S12 S21 ΓG ΓL 2 1
Ga =  
Rout Z11 + ZG |1 − S11 ΓG |2 1 − |Γout |2
Similarly, we have for PavG and PavN :  2 (13.6.11)
RL  Z21  =
 1 1 − |ΓL |2
Gp = |S21 |2
|VG |2 |1 − ΓG |2 Rin Z22 + ZL 
 1 − |Γin |2 |1 − S22 ΓL |2
PavG =
8Z0 1 − |ΓG |2
(13.6.7) The transducer gain GT is, perhaps, the most representative measure of gain for
|VG |2 |1 − ΓG |2 |S21 |2 the two-port because it incorporates the effects of both the load and generator impe-
PavN =  
8Z0 1 − |Γout |2 |1 − S11 ΓG |2 dances, whereas Ga depends only on the generator impedance and Gp only on the load
impedance.
It is evident that PavG , PavN are obtained from Pin , PL by setting Γin = Γ∗
G and ΓL =
If the generator and load impedances are matched to the reference impedance Z0 ,
Γ∗
out , which are equivalent to the conjugate-match conditions.
so that ZG = ZL = Z0 and ΓG = ΓL = 0, and Γin = S11 , Γout = S22 , then the power gains
reduce to:
13.6. Power Gains 539 540 13. S-Parameters

Simultaneous matching is discussed in Sec. 13.8. The necessary and sufficient con-
2 |S21 |2 |S21 |2 dition for simultaneous matching is K ≥ 1, where K is the Rollett stability factor. It can
GT = |S21 | , Ga = , Gp = (13.6.12)
1 − |S22 |2 1 − |S11 |2 be shown that the MAG can be expressed as:
A unilateral two-port has by definition zero reverse transmission coefficient, that is,
S12 = 0. In this case, the input and output reflection coefficients simplify into: |S21 |   
GMAG = K − K2 − 1 (maximum available gain) (13.6.20)
|S12 |
Γin = S11 , Γout = S22 (unilateral two-port) (13.6.13)
The maximum stable gain (MSG) is the maximum value GMAG can have, which is
The expressions of the power gains simplify somewhat in this case: achievable when K = 1:

1 − |ΓG |2 1 − |ΓL |2 |S21 |


GTu = |S21 |2 GMSG = (maximum stable gain) (13.6.21)
|1 − S11 ΓG |2 |1 − S22 ΓL |2 |S12 |
1 − |ΓG |2 1 In the unilateral case, the MAG is obtained either by setting ΓG = Γ∗ ∗
Gau = |S21 |2 (unilateral gains) (13.6.14) in = S11 and
|1 − S11 ΓG |2 1 − |S22 |2 ∗ ∗
ΓL = Γout = S22 in Eq. (13.6.14), or by a careful√limiting process in Eq. (13.6.20), in which
1 1 − |ΓL |2 K → ∞ so that both the numerator factor K − K2 − 1 and the denominator factor |S12 |
Gpu = |S21 |2 tend to zero. With either method, we find the unilateral MAG:
1 − |S11 |2 |1 − S22 ΓL |2
For both the bilateral and unilateral cases, the gains Ga , Gp are obtainable from GT

by setting ΓL = Γ∗out and Γin = ΓG , respectively, as was the case for PavN and PavG . |S21 |2
GMAG,u =    = G1 |S21 |2 G2 (unilateral MAG) (13.6.22)
The relative power ratios Pin /PavG and PL /PavN measure the mismatching between 1 − |S11 |2 1 − |S22 |2
the generator and the two-port and between the load and the two-port. Using the defi-
The maximum unilateral input and output gain factors are:
nitions for the power gains, we obtain the input and output mismatch factors:
   1 1
Pin GT 4Rin RG 1 − |Γin |2 1 − |ΓG |2 G1 = , G2 = (13.6.23)
Min = = = = (13.6.15) 1 − |S11 |2 1 − |S22 |2
PavG Gp |Zin + ZG |2 |1 − Γin ΓG |2
They are the maxima of the input and output gain factors in Eq. (13.6.14) realized
   ∗ ∗
PL GT 4Rout RL 1 − |Γout |2 1 − |ΓL |2 with conjugate matching, that is, with ΓG = S11 and ΓL = S22 . For any other values
Mout = = = = (13.6.16)
PavN Ga |Zout + ZL |2 |1 − Γout ΓL |2 of the reflection coefficients (such that |ΓG | < 1 and ΓL | < 1), we have the following
The mismatch factors are always less than or equal to unity (for positive Rin and inequalities, which follow from the identity (13.6.18):

Rout .) Clearly, Min = 1 under the conjugate-match condition Zin = ZG or Γin = Γ∗G , and

1 − |ΓG |2 1 1 − |ΓL |2 1
Mout = 1 if ZL = Zout or ΓL = Γ∗ . The mismatch factors can also be written in the ≤ , ≤ (13.6.24)
out
|1 − S11 ΓG |2 1 − |S11 |2 |1 − S22 ΓL |2 1 − |S22 |2
following forms, which show more explicitly the mismatch properties:
Often two-ports, such as most microwave transistor amplifiers, are approximately
   
 Γ − Γ∗ 2  Γ − Γ∗ 2 unilateral, that is, the measured S-parameters satisfy |S12 | |S21 |. To decide whether
 in G   out L 
Min = 1 −   , Mout =1−  (13.6.17) the two-port should be treated as unilateral, a figure of merit is used, which is essentially
 1 − Γin ΓG   1 − Γout ΓL 
the comparison of the maximum unilateral gain to the transducer gain of the actual
These follow from the identity: ∗ ∗
device under the same matching conditions, that is, ΓG = S11 and ΓL = S22 .
   For these matched values of ΓG , ΓL , the ratio of the bilateral and unilateral transducer
|1 − Γ1 Γ2 |2 − |Γ1 − Γ∗ 2
2 | = 1 − |Γ1 |
2
1 − |Γ2 |2 (13.6.18) gains can be shown to have the form:
The transducer gain is maximized when the two-port is simultaneously conjugate ∗ ∗
GT 1 S12 S21 S11 S22
matched, that is, when Γin = Γ∗ ∗
G and ΓL = Γout . Then, Min = Mout = 1 and the three
gu = = , U=    (13.6.25)
GTu |1 − U|2 1 − |S11 |2 1 − |S22 |2
gains become equal. The common maximum gain achieved by simultaneous matching
is called the maximum available gain (MAG): The quantity |U| is known as the unilateral figure of merit. If the relative gain ratio
gu is near unity (typically, within 10 percent of unity), the two-port may be treated as
GT,max = Ga,max = Gp,max = GMAG (13.6.19) unilateral.
The MATLAB function sgain computes the transducer, available, and operating
power gains, given the S-parameters and the reflection coefficients ΓG , ΓL . In addition,
13.6. Power Gains 541 542 13. S-Parameters

it computes the unilateral gains, the maximum available gain, and the maximum stable The amplifier cannot be considered to be unilateral as the unilateral figure of merit ratio
gain. It also computes the unilateral figure of merit ratio (13.6.25). It has usage: gu = 1.23 is fairly large (larger than 10 percent from unity.)
The amplifier is operating at a gain of GT = 6.73 dB, which is far from the maximum value
Gt = sgain(S,gG,gL); transducer power gain at given ΓG , ΓL
of GMAG = 16.18 dB. This is because it is mismatched with the given generator and load
Ga = sgain(S,gG,’a’); available power gain at given ΓG with ΓL = Γ∗
out
Gp = sgain(S,gL,’p’); operating power gain at given ΓL with ΓG = Γ∗ impedances.
in
To realize the optimum gain GMAG the amplifier must ‘see’ certain optimum generator
Gmag = sgain(S); maximum available gain (MAG)
and load impedances or reflection coefficients. These can be calculated by the MATLAB
Gmsg = sgain(S,’msg’); maximum stable gain (MSG)
function smatch and are found to be:
Gu = sgain(S,’u’); maximum unilateral gain, Eq. (13.6.22)
G1 = sgain(S,’ui’); maximum unilateral input gain, Eq. (13.6.23) ΓG = 0.82∠−162.67o ⇒ ZG = g2z(ZG , Z0 )= 5.12 − 7.54j Ω
G2 = sgain(S,’uo’); maximum unilateral output gain, Eq. (13.6.23)
ΓL = 0.75∠52.57o
⇒ ZL = g2z(ZL , Z0 )= 33.66 + 91.48j Ω
gu = sgain(S,’ufm’); unilateral figure of merit gain ratio, Eq. (13.6.25)

The design of such optimum matching terminations and the function smatch are discussed
The MATLAB functions gin and gout compute the input and output reflection coef-
in Sec. 13.8. The functions g2z and z2g were discussed in Sec. 10.7 . 

ficients from S and ΓG , ΓL . They have usage:

Gin = gin(S,gL); input reflection coefficient, Eq. (13.4.3)


Gout = gout(S,gG); output reflection coefficient, Eq. (13.4.6)
13.7 Generalized S-Parameters and Power Waves
The practical usefulness of the S-parameters lies in the fact that the definitions (13.1.4)
Example 13.6.1: A microwave transistor amplifier uses the Hewlett-Packard AT-41410 NPN represent forward and backward traveling waves, which can be measured remotely by
bipolar transistor with the following S-parameters at 2 GHz [1324]: connecting a network analyzer to the two-port with transmission lines of characteristic
impedance equal to the normalization impedance Z0 . This was depicted in Fig. 13.1.3.
S11 = 0.61∠165o , S21 = 3.72∠59o , S12 = 0.05∠42o , S22 = 0.45∠−48o A generalized definition of S-parameters and wave variables can be given by using
in Eq. (13.1.4) two different normalization impedances for the input and output ports.
Calculate the input and output reflection coefficients and the various power gains, if the Anticipating that the two-port will be connected to a generator and load of impedan-
amplifier is connected to a generator and load with impedances ZG = 10 − 20j and ZL = ces ZG and ZL , a particularly convenient choice is to use ZG for the input normalization
30 + 40j ohm. impedance and ZL for the output one, leading to the definition of the power waves (as
opposed to traveling waves) [958–960,962]:
Solution: The following MATLAB code will calculate all the required gains:

V1 + ZG I1 V2 − ZL I2
Z0 = 50; % normalization impedance a1 =  a2 = 
ZG = 10+20j; gG = z2g(ZG,Z0); % ΓG = −0.50 + 0.50j = 0.71∠135o 2 RG 2 RL
ZL = 30-40j; gL = z2g(ZL,Z0); % ΓL = −0.41 − 0.43j = 0.59∠−133.15o ∗
(power waves) (13.7.1)
V1 − ZG I1 V2 + ZL∗ I2
b1 =  b2 = 
S = smat([0.61 165 3.72 59 0.05 42 0.45 -48]); % reshape S into matrix 2 RG 2 RL

Gin = gin(S,gL); % Γin = 0.54∠162.30o We note that the b-waves involve the complex-conjugates of the impedances. The
Gout = gout(S,gG); % Γout = 0.45∠−67.46o quantities RG , RL are the resistive parts of ZG , ZL and are assumed to be positive. These
Gt = sgain(S,gG,gL); % GT = 4.71, or, 6.73 dB
definitions reduce to the conventional traveling ones if ZG = ZL = Z0 .
Ga = sgain(S,gG,’a’); % Ga = 11.44, or, 10.58 dB These “wave” variables can no longer be interpreted as incoming and outgoing waves
Gp = sgain(S,gL,’p’); % Gp = 10.51, or, 10.22 dB from the two sides of the two-port. However, as we see below, they have a nice interpre-
tation in terms of power transfer to and from the two-port and simplify the expressions
Gu = sgain(S,’u’); % Gu = 27.64, or, 14.41 dB
G1 = sgain(S,’ui’); % G1 = 1.59, or, 2.02 dB
for the power gains. Inverting Eqs. (13.7.1), we have:
G2 = sgain(S,’uo’); % G2 = 1.25, or, 0.98 dB
gu = sgain(S,’ufm’); % gu = 1.23, or, 0.89 dB 1 1
V1 =  (Z∗ a + ZG b1 ) V2 =  (ZL∗ a2 + ZL b2 )
RG G 1 RL
Gmag = sgain(S); % GMAG = 41.50, or, 16.18 dB (13.7.2)
Gmsg = sgain(S,’msg’); % GMSG = 74.40, or, 18.72 dB 1 1
I1 =  (a − b1 ) I2 =  (b2 − a2 )
RG 1 RL
13.7. Generalized S-Parameters and Power Waves 543 544 13. S-Parameters

The power waves can be related directly to the traveling waves. For example, ex-
pressing Eqs. (13.7.1) and (13.1.5) in matrix form, we have for port-1: ⎡ ⎤
  ejφG  
          ⎢ 0 ⎥
a1 ΓG ⎢ ⎥ FG
F = ⎢ 1 − |ΓG |
1 1 ZG V1 V1 1 Z0 Z0 a1 0 2 0
=  , =  Γ= , ⎥= (13.7.9)
b1 1 ∗
−ZG I1 I1 1 −1 b1 0 ΓL ⎣ e jφL ⎦ 0 FL
2 RG Z0 0 
1 − |ΓL |2
It follows that:
      Using these matrices, it follows from Eqs. (13.7.4) and (13.7.6):
a1 1 1 ZG Z0 Z0 a1
=  or, a1 = FG (a1 − ΓG b1 )
b1 2 RG Z0 1 ∗
−ZG 1 −1 b1
⇒ a = F(a − Γ b) (13.7.10)
     a2 = FL (a2 − ΓL b2 )
a1 1 Z0 + ZG Z0 − ZG a1
=  (13.7.3) b1 = FG

(b1 − Γ∗
G a1 )
b1 2 RG Z0

Z0 − ZG ∗
Z0 + ZG b1 ⇒ b = F∗ (b − Γ∗ a) (13.7.11)
b2 = FL∗ (b2 − Γ∗
L a2 )
The entries of this matrix can be expressed directly in terms of the reflection coeffi-
cient ΓG . Using the identities of Problem 13.3, we may rewrite Eq. (13.7.3) and its inverse Using b = Sa, we find
as follows::
a = F(a − Γ b)= F(I − ΓS)a ⇒ a = (I − ΓS)−1 F−1 a
    
 
a1 1 e jφG
−ΓG e jφG
a1 b = F (S − Γ )a = F (S − Γ )(I − ΓS)−1 F−1 a = S a
∗ ∗ ∗ ∗
= 
b1 1 − |ΓG |2 −Γ∗
Ge
−jφG
e−jφG b1
     (13.7.4) where I is the 2×2 unit matrix. Thus, the generalized S-matrix is:
a1 1 e−jφG ΓG ejφG a1
=  ∗ −jφG
b1 1 − |ΓG |2 ΓG e ejφG
b1 S = F∗ (S − Γ∗ )(I − ΓS)−1 F−1 (13.7.12)
where, noting that the quantity |1 − ΓG |/(1 − ΓG ) is a pure phase factor, we defined: We note that S = S when ZG = ZL = Z0 , that is, when ΓG = ΓL = 0. The explicit
expressions for the matrix elements of S can be derived as follows:
ZG − Z0 |1 − ΓG | 1 − Γ∗
G
ΓG = , ejφG = = (13.7.5)
ZG + Z0 1 − ΓG |1 − ΓG |  (S11 − Γ∗G )(1 − S22 ΓL )+S21 S12 ΓL
S11 = e−2jφG
Similarly, we have for the power and traveling waves at port-2: (1 − S11 ΓG )(1 − S22 ΓL )−S12 S21 ΓG ΓL
(13.7.13a)
     (S22 − Γ∗L )(1 − S11 ΓG )+S21 S12 ΓG
a2 1 ejφL −ΓL ejφL a2 
S22 = e−2jφL
=  (1 − S11 ΓG )(1 − S22 ΓL )−S12 S21 ΓG ΓL
b2 1 − |ΓL |2 −Γ∗
Le
−jφL
e−jφL b2
 
     (13.7.6)
 1 − |ΓG |2 S21 1 − |ΓL |2
a2 1 e−jφL ΓL ejφL a2 S21 = e−j(φG +φL )
=  (1 − S11 ΓG )(1 − S22 ΓL )−S12 S21 ΓG ΓL
b2 1 − |ΓL |2 Γ∗
Le
−jφL
e jφL
b2   (13.7.13b)
 1 − |ΓL |2 S12 1 − |ΓG |2
where S12 = e−j(φL +φG )
(1 − S11 ΓG )(1 − S22 ΓL )−S12 S21 ΓG ΓL
ZL − Z0 |1 − ΓL | 1 − Γ∗
L

The S11 
, S22 parameters can be rewritten in terms of the input and output reflection
ΓL = , ejφL = = (13.7.7)
ZL + Z0 1 − ΓL |1 − ΓL | coefficients by using Eq. (13.13.2) and the following factorization identities:
The generalized S-parameters are the scattering parameters with respect to the
(S11 − Γ∗ ∗
G )(1 − S22 ΓL )+S21 S12 ΓL = (Γin − ΓG )(1 − S22 ΓL )
power wave variables, that is,
(S22 − ΓL )(1 − S11 ΓG )+S21 S12 ΓG = (Γout − Γ∗

L )(1 − S11 ΓG )
    
b1 
S11 
S12 a1
= ⇒ b = S a (13.7.8) It then follows from Eq. (13.7.13) that:
b2 
S21 
S22 a2

To relate S to the conventional scattering matrix S, we define the following diagonal  Γin − Γ∗  Γout − Γ∗
S11 = G
e−2jφG , S22 = L
e−2jφL (13.7.14)
matrices: 1 − Γin ΓG 1 − Γout ΓL
Therefore, the mismatch factors (13.6.17) are recognized to be:
13.7. Generalized S-Parameters and Power Waves 545 546 13. S-Parameters

 2  2  2  2
MG = 1 − |S11 | , ML = 1 − |S22 | (13.7.15)  2 |S21 | |S21 |
GT = |S21 | , Ga =  2 , Gp =  2 (13.7.21)
1 − |S22 | 1 − |S11 |
The power flow relations (13.2.1) into and out of the two-port are also valid in terms
of the power wave variables. Using Eq. (13.7.2), it can be shown that: When the load and generator are matched to the network, that is, Γin = Γ∗ G and
 
1 1 1 ΓL = Γ∗
out , the generalized reflections coefficients vanish, S11 = S22 = 0, making all the
Pin = Re[V1∗ I1 ]= |a1 |2 − |b1 |2 gains equal to each other.
2 2 2
(13.7.16)
1 1 1
PL = Re[V2∗ I2 ]= |b2 |2 − |a2 |2
2 2 2 13.8 Simultaneous Conjugate Matching
In the definitions (13.7.1), the impedances ZG , ZL are arbitrary normalization param-
eters. However, if the two-port is actually connected to a generator VG with impedance We saw that the transducer, available, and operating power gains become equal to the
ZG and a load ZL , then the power waves take particularly simple forms. maximum available gain GMAG when both the generator and the load are conjugately
It follows from Fig. 13.1.4 that VG = V1 +ZG I1 and V2 = ZL I2 . Therefore, definitions matched to the two-port, that is, Γin = Γ∗ ∗
G and ΓL = Γout . Using Eq. (13.5.8), these
Eq. (13.7.1) give: conditions read explicitly:

V1 + ZG I1 VG S12 S21 ΓL S11 − ΔΓL


a1 =  =  Γ∗
G = S11 + =
2 RG 2 RG 1 − S22 ΓL 1 − S22 ΓL
(13.8.1)
V2 − ZL I2 S12 S21 ΓG S22 − ΔΓG

a2 =  =0 (13.7.17) Γ∗
L = S22 + =
2 RL 1 − S22 ΓG 1 − S11 ΓG

V2 + ZL∗ I2 ZL + ZL∗ 2RL  Assuming a bilateral two-port, Eqs. (13.8.1) can be solved in the two unknowns ΓG , ΓL
b2 =  =  I2 =  I2 = RL I2 (eliminating one of the unknowns gives a quadratic equation for the other.) The resulting
2 RL 2 RL 2 RL
solutions can be expressed in terms of the parameters (13.5.1):
It follows that the available power from the generator and the power delivered to
the load are given simply by: 
B1 ∓ B21 − 4|C1 |2
ΓG =
|VG |2 1 2C1
PavG = = |a1 |2  (simultaneous conjugate match) (13.8.2)
8RG 2
(13.7.18) B2 ∓ B22 − 4|C2 |2
1 1 ΓL =
PL = RL |I2 |2 = |b2 |2 2C2
2 2
Because a2 = 0, the generalized scattering matrix gives, b1 = S11

a1 and b2 = S21

a1 . where the minus signs are used when B1 > 0 and B2 > 0, and the plus signs, otherwise.
The power expressions (13.7.16) then become: A necessary and sufficient condition for these solutions to have magnitudes |ΓG | < 1
and |ΓL | < 1 is that the Rollett stability factor be greater than unity, K > 1. This is
satisfied when the two-port is unconditionally stable, which implies that K > 1 and
1  2 1   2 1
  
Pin = |a | − |b1 |2 = 1 − |S11 | |a |2 = 1 − |S11
 2
| PavG B1 > 0, B2 > 0.
2 1 2 2 1 A conjugate match exists also when the two-port is potentially unstable, but with
(13.7.19)
1  2 1 1  21 K > 1. Necessarily, this means that B1 < 0, B2 < 0, and also |Δ| > 1. Such cases are
PL = |b | − |a2 |2 = |b2 |2 = |S21 | |a1 |2 = |S21
 2
| PavG
2 2 2 2 2 rare in practice. For example, most microwave transistors have either K > 1 and are
stable, or, they are potentially unstable with K < 1 and |Δ| < 1.
It follows that the transducer and operating power gains are:
If the two-port is unilateral, S12 = 0, then the two equations (13.8.1) decouple, so
 2
PL  2 PL |S21 | that the optimum conjugately matched terminations are:
GT = = |S21 | , Gp = =  2 (13.7.20)
PavG Pin 1 − |S11 |
∗ ∗
ΓG = S11 , ΓL = S22 (unilateral conjugate match) (13.8.3)
These also follow from the explicit expressions (13.7.13) and Eqs. (13.6.10) and
(13.6.11). We can also express the available power gain in terms of the generalized The MATLAB function smatch implements Eqs. (13.8.2). It works only if K > 1. Its
 2
  2

S-parameters, that is, Ga = |S21 | / 1 − |S22 | . Thus, we summarize: usage is as follows:
13.8. Simultaneous Conjugate Matching 547 548 13. S-Parameters

[gG,gL] = smatch(S); % conjugate matched terminations ΓG , ΓL

To realize such optimum conjugately matched terminations, matching networks S11 = 0.61∠165o , S21 = 3.72∠59o , S12 = 0.05∠42o , S22 = 0.45∠−48o
must be used at the input and output of the two-port as shown in Fig. 13.8.1.
The input matching network can be thought as being effectively connected to the Determine the optimum conjugately matched source and load terminations, and design

impedance Zin = ZG at its output terminals. It must transform Zin into the actual appropriate input and output matching networks.
impedance of the connected generator, typically, Z0 = 50 ohm.
Solution: This is the continuation of Example 13.6.1. The transistor is stable with K = 1.1752
The output matching network must transform the actual load impedance, here Z0 , and |Δ| = 0.1086. The function smatch gives:

into the optimum load impedance ZL = Zout .
[ΓG , ΓL ]= smatch(S) ⇒ ΓG = 0.8179∠−162.6697o , ΓL = 0.7495∠52.5658o

The corresponding source, load, input, and output impedances are (with Z0 = 50):

∗ ∗
ZG = Zin = 5.1241 − 7.5417j Ω , ZL = Zout = 33.6758 + 91.4816j Ω

The locations of the optimum reflection coefficients on the Smith chart are shown in
Fig. 13.8.1 Input and output matching networks. Fig. 13.8.3. For comparison, the unilateral solutions of Eq. (13.8.3) are also shown.

The matching networks may be realized in several possible ways, as discussed in


Chap. 12. Stub matching, quarter-wavelength matching, or lumped L-section or Π-
section networks may be used. In designing the matching networks, it proves convenient
to first design the reverse network as mentioned in Sec. 12.13.
Fig. 13.8.2 shows the procedure for designing the output matching network using
a reversed stub matching transformer or a reversed quarter-wave transformer with a
parallel stub. In both cases the reversed network is designed to transform the load
impedance ZL∗ into Z0 .

Fig. 13.8.3 Optimum load and source reflection coefficients.

We consider three types of matching networks: (a) microstrip single-stub matching net-
works with open shunt stubs, shown in Fig. 13.8.4, (b) microstrip quarter-wavelength
matching networks with open λ/8 or 3λ/8 stubs, shown in Fig. 13.8.5, and (c) L-section
matching networks, shown in 13.8.6.

Fig. 13.8.4 Input and output stub matching networks.


Fig. 13.8.2 Two types of output matching networks and their reversed networks..

In Fig. 13.8.4, the input stub must transform Zin to Z0 . It can be designed with the help of
Example 13.8.1: A microwave transistor amplifier uses the Hewlett-Packard AT-41410 NPN the function stub1, which gives the two solutions:
bipolar transistor having S-parameters at 2 GHz [1324]:
13.8. Simultaneous Conjugate Matching 549 550 13. S-Parameters

 
0.3038 0.4271
dl = stub1(Zin /Z0 , ’po’)=
0.1962 0.0247

We choose the lower one, which has the shortest lengths. Thus, the stub length is d =
0.1962λ and the segment length l = 0.0247λ. Both segments can be realized with mi-
crostrips of characteristic impedance Z0 = 50 ohm. Similarly, the output matching net-
work can be designed by:
Fig. 13.8.5 Quarter-wavelength matching networks with λ/8-stubs.
 
0.3162 0.1194
dl = stub1(Zout /Z0 , ’po’)=
0.1838 0.2346

Again, we choose the lower solutions, d = 0.1838λ and l = 0.2346λ. The solutions using
shorted shunt stubs are:
   
0.0538 0.4271 0.0662 0.1194
stub1(Zin /Z0 )= , stub1(Zout /Z0 )=
0.4462 0.0247 0.4338 0.2346

Using microstrip lines with alumina substrate ( r = 9.8), we obtain the following values Fig. 13.8.6 Input and output matching with L-sections.
for the width-to-height ratio, effective permittivity, and wavelength:

w
u= = mstripr( r , Z0 )= 0.9711
h  
16.8955 −22.7058
eff = mstripa( r , u)= 6.5630 [X1 , X2 ]= lmatch(Z0 , Zin , ’n’)=
−16.8955 7.6223
λ0  
λ= √ = 5.8552 cm 57.9268 −107.7472
eff [X1 , X2 ]= lmatch(Zout , Z0 , ’n’)=
502.4796 7.6223
where λ0 = 15 cm is the free-space wavelength at 2 GHz. It follows that the actual segment

lengths are d = 1.1486 cm, l = 0.1447 cm for the input network, and d = 1.0763 cm, According to the usage of lmatch, the output network transforms Z0 into Zout , but that is
l = 1.3734 cm for the output network. equal to ZL as required.

In the quarter-wavelength method shown in Fig. 13.8.5, we use the function qwt2 to carry Choosing the first rows as the solutions in both cases, the shunt part X1 will be inductive
out the design of the required impedances of the microstrip segments. We have for the and the series part X2 , capacitive. At 2 GHz, we find the element values:
input and output networks:
X1 1
L1 = = 1.3445 nH, C1 = − = 3.5047 pF
[Z1 , Z2 ]= qwt2(Zin , Z0 )= [28.4817, −11.0232] Ω ω ωX2
X1 1
[Z1 , Z2 ]= qwt2(Zout , Z0 )= [118.7832, 103.8782] Ω L2 = = 4.6097 nH, C2 = − = 0.7386 pF
ω ωX2

For the input case, we find Z2 = −11.0232 Ω, which means that we should use either a The output network, but not the input one, also admits a reversed L-section solution:
3λ/8-shorted stub or a λ/8-opened one. We choose the latter. Similarly, for the output
 
case, we have Z2 = 103.8782 Ω, and we choose a 3λ/8-opened stub. The parameters of 71.8148 68.0353
each microstrip segment are: [X1 , X2 ]= lmatch(Zout , Z0 , ’r’)=
−71.8148 114.9280

Z1 = 28.4817 Ω, u = 2.5832, eff = 7.2325, λ = 5.578 cm, λ/4 = 1.394 cm


The essential MATLAB code used to generate the above results was as follows:
Z2 = 11.0232 Ω, u = 8.9424, eff = 8.2974, λ = 5.207 cm, λ/8 = 0.651 cm
Z1 = 118.7832 Ω, u = 0.0656, eff = 5.8790, λ = 6.186 cm, λ/4 = 1.547 cm
Z0 = 50; f = 2; w=2*pi*f; la0 = 30/f; er = 9.8; % f in GHz
Z2 = 103.8782 Ω, u = 0.1169, eff = 7.9503, λ = 6.149 cm, 3λ/8 = 2.306 cm
S = smat([0.61 165 3.72 59 0.05 42 0.45 -48]); % S-matrix
Finally, the designs using L-sections shown in Fig. 13.8.6, can be carried out with the help
of the function lmatch. We have the dual solutions for the input and output networks: [gG,gL] = smatch(S); % simultaneous conjugate match
13.9. Power Gain Circles 551 552 13. S-Parameters

where in the last four cases G is the desired gain in dB.


smith; % draw Fig. 13.8.3
plot(gG, ’.’); plot(conj(S(1,1)), ’o’);
plot(gL, ’.’); plot(conj(S(2,2)), ’o’);
13.10 Unilateral Gain Circles
ZG = g2z(gG,Z0); Zin = conj(ZG);
ZL = g2z(gL,Z0); Zout = conj(ZL); We consider only the unconditionally stable unilateral case, which has |S11 | < 1 and
|S22 | < 1. The dependence of the transducer power gain on ΓG and ΓL decouples and
dl = stub1(Zin/Z0, ’po’); % single-stub design
the value of the gain may be adjusted by separately choosing ΓG and ΓL . We have from
dl = stub1(Zout/Z0, ’po’);
Eq. (13.6.14):
u = mstripr(er,Z0); % microstrip w/h ratio
eff = mstripa(er,u); % effective permittivity 1 − |ΓG |2 1 − |ΓL |2
GT = |S21 |2 = GG |S21 |2 GL (13.10.1)
la = la0/sqrt(eff); % wavelength within microstrip |1 − S11 ΓG |2 |1 − S22 ΓL |2
[Z1,Z2] = qwt2(Zin, Z0); % quarter-wavelength with λ/8 stub The input and output gain factors GG , GL satisfy the inequalities (13.6.24). Concen-
[Z1,Z2] = qwt2(Zout, Z0); trating on the output gain factor, the corresponding gain circle is obtained as the locus
of points ΓL that will lead to a fixed value, say GL = G, which necessarily must be less
X12 = lmatch(Z0,Zin,’n’); L1 = X12(1,1)/w; C1 = -1/(w * X12(1,2))*1e3;
X12 = lmatch(Zout,Z0,’n’); L2 = X12(1,1)/w; C2 = -1/(w * X12(1,2))*1e3; than the maximum G2 given in Eq. (13.6.23), that is,
X12 = lmatch(Zout,Z0,’r’); % L, C in units of nH and pF
1 − |ΓL |2 1
= G ≤ G2 = (13.10.2)
One could replace the stubs with balanced stubs, as discussed in Sec. 12.9, or use Π- or
|1 − S22 ΓL |2 1 − |S22 |2
 
T-sections instead of L-sections. 
 Normalizing the gain G to its maximum value g = G/G2 = G 1 − |S22 |2 , we may
rewrite (13.10.2) in the form:
  
13.9 Power Gain Circles 1 − |ΓL |2 1 − |S22 |2
=g≤1 (13.10.3)
|1 − S22 ΓL |2
For a stable two-port, the maximum transducer gain is achieved at single pair of points
This equation can easily be rearranged into the equation of a circle |ΓL − c| = r , with
ΓG , ΓL . When the gain G is required to be less than GMAG , there will be many possible
center and radius given by:
pairs ΓG , ΓL at which the gain G is realized. The locus of such points ΓG and ΓL on the
Γ-plane is typically a circle of the form: ∗
  
gS22 1 − g 1 − |S22 |2
c= , r= (13.10.4)
1 − (1 − g)|S22 |2 1 − (1 − g)|S22 |2
|Γ − c| = r (13.9.1)

where c, r are the center and radius of the circle and depend on the desired value of the When g = 1 or G = G2 , the gain circle collapses onto a single point, that is, the

gain G. optimum point ΓL = S22 . Similarly, we find for the constant gain circles of the input
In practice, several types of such circles are used, such as unilateral, operating, and gain factor:
available power gain circles, as well as constant noise figure circles, constant SWR circles,

  
and others. gS11 1 − g 1 − |S11 |2
c= , r = (13.10.5)
The gain circles allow one to select appropriate values for ΓG , ΓL that, in addition to 1 − (1 − g)|S11 |2 1 − (1 − g)|S11 |2
providing the desired gain, also satisfy other requirements, such as striking a balance  
where here, g = G/G1 = G 1 − |S11 |2 and the circles are |ΓG − c| = r .
between minimizing the noise figure and maximizing the gain.
Both sets of c, r satisfy the conditions |c| < 1 and |c| + r < 1, the latter implying
The MATLAB function sgcirc calculates the stability circles as well as the operating,
that the circles lie entirely within the unit circle |Γ| < 1, that is, within the Smith chart.
available, and unilateral gain circles. Its complete usage is:
Example 13.10.1: A unilateral microwave transistor has S-parameters:
[c,r] = sgcirc(S,’s’); % source stability circle
[c,r] = sgcirc(S,’l’); % load stability circle
[c,r] = sgcirc(S,’p’,G); % operating power gain circle
S11 = 0.8∠120o , S21 = 4∠60o , S12 = 0, S22 = 0.2∠−30o
[c,r] = sgcirc(S,’a’,G); % available power gain circle
[c,r] = sgcirc(S,’ui’,G); % unilateral input gain circle The unilateral MAG and the maximum input and output gains are obtained as follows:
[c,r] = sgcirc(S,’uo’,G); % unilateral output gain circle
13.10. Unilateral Gain Circles 553 554 13. S-Parameters

The input and output matching networks can be designed using open shunt stubs as in
GMAG,u = sgain(S, ’u’)= 16.66 dB Fig. 13.8.4. The stub lengths are found to be (with Z0 = 50 Ω):
 
G1 = sgain(S, ’ui’)= 4.44 dB 0.3704 0.3304

dl = stub1(ZG /Z0 , ’po’)=
G2 = sgain(S, ’uo’)= 0.18 dB 0.1296 0.0029
 
0.4383 0.0994
dl = stub1(ZL∗ /Z0 , ’po’)=
Most of the gain is accounted for by the factor |S21 | , which is 12.04 dB. The constant input
2
0.0617 0.3173
gain circles for GG = 1, 2, 3 dB are shown in Fig. 13.10.1. Their centers lie along the ray to

S11 . For example, the center and radius of the 3-dB case were computed by Choosing the shortest lengths, we have for the input network d = 0.1296λ, l = 0.0029λ,
and for the output network, d = 0.0617λ, l = 0.3173λ. Fig. 13.10.2 depicts the complete
[c3 , r3 ]= sgcirc(S, ’ui’, 3) ⇒ c3 = 0.701∠−120o , r3 = 0.233 matching circuit. 


Fig. 13.10.2 Input and output stub matching networks.


Fig. 13.10.1 Unilateral input gain circles.

Because the output does not provide much gain, we may choose the optimum value ΓL =

S22 = 0.2∠30o . Then, with any point ΓG along the 3-dB input gain circle the total trans- 13.11 Operating and Available Power Gain Circles
ducer gain will be in dB:
Because the transducer power gain GT depends on two independent parameters—the
GT = GG + |S21 |2 + GL = 3 + 12.04 + 0.18 = 15.22 dB source and load reflection coefficients—it is difficult to find the simultaneous locus of
points for ΓG , ΓL that will result in a given value for the gain.
Points along the 3-dB circle are parametrized as ΓG = c3 + r3 ejφ , where φ is any angle.

If the generator is matched, Γin = Γ∗ G , then the transducer gain becomes equal to
Choosing φ = arg(S11 )−π will correspond to the point on the circle that lies closest to the
the operating gain GT = Gp and depends only on the load reflection coefficient ΓL .
origin, that is, ΓG = 0.468∠−120o , as shown in Fig. 13.10.1. The corresponding generator
The locus of points ΓL that result in fixed values of Gp are the operating power gain
and load impedances will be:
circles. Similarly, the available power gain circles are obtained by matching the load
ZG = 69.21 + 14.42j Ω, ZL = 23.15 − 24.02j Ω end, ΓL = Γ∗ out , and varying ΓG to achieve fixed values of the available power gain.
Using Eqs. (13.6.11) and (13.5.8), the conditions for achieving a constant value, say
The MATLAB code used to generate these circles was: G, for the operating or the available power gains are:
S = smat([0.8, 120, 4, 60, 0, 0, 0.2, -30]);
1 1 − |ΓL |2 S11 − ΔΓL
[c1,r1] = sgcirc(S,’ui’,1); Gp = |S21 |2 = G, Γ∗
G = Γin =
[c2,r2] = sgcirc(S,’ui’,2); 1 − |Γin |2 |1 − S22 ΓL |2 1 − S22 ΓL
[c3,r3] = sgcirc(S,’ui’,3);
(13.11.1)
1 − |ΓG |2 1 S22 − ΔΓG
Ga = |S21 |2 = G, Γ∗
L = Γout =
smith; smithcir(c1,r1); smithcir(c2,r2); smithcir(c3,r3); |1 − S11 ΓG |2 1 − |Γout |2 1 − S11 ΓG

c = exp(-j*angle(S(1,1))); line([0,real(c)], [0,imag(c)]); We consider the operating gain first. Defining the normalized gain g = G/|S21 |2 ,
substituting Γin , and using the definitions (13.5.1), we obtain the condition:
gG = c3 - r3*exp(j*angle(c3));

plot(conj(S(1,1)),’.’); plot(conj(S(2,2)),’.’); plot(gG,’.’);


13.11. Operating and Available Power Gain Circles 555 556 13. S-Parameters

roots in the definitions of the radii r in Eqs. (13.11.3) and (13.11.4) are non-negative.
The gain circles lie inside the unit circle for all such values of G. The radii r vanish
1 − |ΓL |2
g= when G = GMAG , that is, the circles collapse into single points corresponding to the
|1 − S22 ΓL |2 − |S11 − ΔΓL |2 simultaneous conjugate matched solutions of Eq. (13.8.2).
1 − |ΓL |2 The MATLAB function sgcirc calculates the center and radii c, r of the operating
=   ∗ ∗
|S22 |2 − |Δ|2 |ΓL |2 − (S22 − ΔS11 )ΓL − (S22 − Δ∗ S11 )Γ∗
L + 1 − |S11 |
2 and available power gain circles. It has usage, where G must be entered in dB:

1 − |ΓL |2 [c,r] = sgcirc(S,’p’,G); operating power gain circle


=
D2 |ΓL |2 − C2 ΓL − C∗ ∗
2 ΓL + 1 − |S11 |
2 [c,r] = sgcirc(S,’a’,G); available power gain circle

Example 13.11.1: A microwave transistor amplifier uses the Hewlett-Packard AT-41410 NPN
This can be rearranged into the form: bipolar transistor with the following S-parameters at 2 GHz [1324]:
 
gC2 gC2∗ 1 − g 1 − |S11 |2 S11 = 0.61∠165o , S21 = 3.72∠59o , S12 = 0.05∠42o , S22 = 0.45∠−48o
|ΓL |2 − ΓL − Γ∗
L =
1 + gD2 1 + gD2 1 + gD2
Calculate GMAG and plot the operating and available power gain circles for G = 13, 14, 15
and then into the circle form: dB. Then, design source and load matching circuits for the case G = 15 dB by choosing
 2   the reflection coefficient that has the smallest magnitude.
 gC∗  g2 |C2 |2 1 − g 1 − |S11 |2
 
ΓL − 2
 = + Solution: The MAG was calculated in Example 13.6.1, GMAG = 16.18 dB. The gain circles and the
 1 + gD2  (1 + gD2 )2 1 + gD2
corresponding load and source stability circles are shown in Fig. 13.11.1. The operating
Using the identities (13.5.2) and 1 − |S11 |2 = 2K|S12 S21 | + D2 , which follows from gain and load stability circles were computed and plotted by the MATLAB statements:
(13.5.1), the right-hand side of the above circle form can be written as:
[c1,r1] = sgcirc(S,’p’,13); % c1 = 0.4443∠52.56o , r1 = 0.5212
[c2,r2] = sgcirc(S,’p’,14); % c2 = 0.5297∠52.56o , r2 = 0.4205
  [c3,r3] = sgcirc(S,’p’,15); % c3 = 0.6253∠52.56o , r3 = 0.2968
g |C2 |
2 2
1 − g 1 − |S11 | 2
g |S12 S21 | − 2gK|S12 S21 | + 1
2 2
[cL,rL] = sgcirc(S,’l’); % cL = 2.0600∠52.56o , rL = 0.9753
+ = (13.11.2)
(1 + gD2 )2 1 + gD2 (1 + gD2 )2
smith; smithcir(cL,rL,1.7); % display portion of circle with |ΓL | ≤ 1.7
Thus, the operating power gain circle will be |ΓL − c|2 = r 2 with center and radius: smithcir(c1,r1); smithcir(c2,r2); smithcir(c3,r3);


gC∗2
g2 |S12 S21 |2 − 2gK|S12 S21 | + 1
c= , r= (13.11.3)
1 + gD2 |1 + gD2 |

The points ΓL on this circle result into the value Gp = G for the operating gain.
Such points can be parametrized as ΓL = c + rejφ , where 0 ≤ φ ≤ 2π. As ΓL traces
this circle, the conjugately matched source coefficient ΓG = Γ∗ in will also trace a circle
because Γin is related to ΓL by the bilinear transformation (13.5.8).
In a similar fashion, we find the available power gain circles to be |ΓG − c|2 = r 2 ,
where g = G/|S21 |2 and:


gC∗1
g2 |S12 S21 |2 − 2gK|S12 S21 | + 1
c= , r= (13.11.4)
1 + gD1 |1 + gD1 |

We recall from Sec. 13.5 that the centers of the load and source stability circles were
Fig. 13.11.1 Operating and available power gain circles.
cL = C∗ ∗
2 /D2 and cG = C1 /D1 . It follows that the centers of the operating power gain
circles are along the same ray as cL , and the centers of the available gain circles are The gain circles lie entirely within the unit circle, for example, we have r3 +|c3 | = 0.9221 <
along the same ray as cG . 1, and their centers lie along the ray of cL . As ΓL traces the 15-dB circle, the corresponding
For an unconditionally stable two-port, the gain G must be 0 ≤ G ≤ GMAG , with ΓG = Γ ∗in traces its own circle, also lying within the unit circle. The following MATLAB code
GMAG given by Eq. (13.6.20). It can be shown easily that the quantities under the square computes and adds that circle to the above Smith chart plots:
13.11. Operating and Available Power Gain Circles 557 558 13. S-Parameters

phi = linspace(0,2*pi,361); % equally spaced angles at 1o intervals In particular, the point ΓG = c3 − r3 ej arg c3 = 0.4774∠−162.67o lies closest to the origin.
The corresponding matched load will have ΓL = Γ∗ out = 0.5728∠50.76 . The resulting
gammaL = c3 + r3 * exp(j*phi); % points on 15-dB operating gain circle o

gammaG = conj(gin(S,gammaL)); % circle of conjugate matched source points normalized impedances are:
plot(gammaG);
ZG ZL
zG = = 0.3609 − 0.1329j , zL = = 1.1135 + 1.4704j
In particular, the point ΓL on the 15-dB circle that lies closest to the origin is ΓL = Z0 Z0

c3 − r3 ej arg c3 = 0.3285∠52.56o . The corresponding matched load will be ΓG = Γin =
0.6805∠−163.88 . These and the corresponding source and load impedances were com-
o and the corresponding stub matching networks will have lengths:
puted by the MATLAB statements:    
0.3684 0.3905 0.3488 0.1030
stub1(z∗
G , ’po’)= , stub1(z∗
L , ’po’)=
gL = c3 - r3*exp(j*angle(c3)); zL = g2z(gL); 0.1316 0.0613 0.1512 0.2560
gG = conj(gin(S,gL)); zG = g2z(gG);
The lower solutions have the shortest lengths. For both the operating and available gain
The source and load impedances normalized to Z0 = 50 ohm are: cases, the stub matching circuits will be similar to those in Fig. 13.8.4. 


ZG ZL When the two-port is potentially unstable (but with |S11 | < 1 and |S22 | < 1,) the
zG = = 0.1938 − 0.1363j , zL = = 1.2590 + 0.7361j stability circles intersect with the unit-circle, as shown in Fig. 13.5.2. In this case, the
Z0 Z0
operating and available power gain circles also intersect the unit-circle and at the same
The matching circuits can be designed in a variety of ways as in Example 13.8.1. Using points as the stability circles.
open shunt stubs, we can determine the stub and line segment lengths with the help of We demonstrate this in the specific case of K < 1, |S11 | < 1, |S22 | < 1, but with
the function stub1: D2 > 0, an example of which is shown in Fig. 13.11.2. The intersection of an operating
  gain circle with the unit-circle is obtained by setting |ΓL | = 1 in the circle equation
0.3286 0.4122
dl = stub1(z∗
G , ’po’)= |ΓL − c| = r . Writing ΓL = ejθL and c = |c|ejθc , we have:
0.1714 0.0431
  1 + |c|2 − r 2
0.4033 0.0786 r 2 = |ΓL − c|2 = 1 − 2|c| cos(θL − θc )+|c|2 ⇒ cos(θL − θc )=
dl = stub1(z∗
L , ’po’)= 2|c|
0.0967 0.2754

Similarly, the intersection of the load stability circle with the unit-circle leads to the
In both cases, we may choose the lower solutions as they have shorter total length d + l.
relationship:
The available power gain circles can be determined in a similar fashion with the help of
the MATLAB statements:
1 + |cL |2 − rL2
rL2 = |ΓL − cL |2 = 1 − 2|cL | cos(θL − θcL )+|cL |2 ⇒ cos(θL − θcL )=
2|cL |
[c1,r1] = sgcirc(S,’a’,13); % c1 = 0.5384∠−162.67o , r1 = 0.4373
% c2 = 0.6227∠
[c2,r2] = sgcirc(S,’a’,14); −162.67o , r2 = 0.3422
Because c = gC∗ ∗
2 /(1 + gD2 ), cL = C2 /D2 , and D2 > 0, it follows that the phase
[c3,r3] = sgcirc(S,’a’,15); % c3 = 0.7111∠ −162.67o , r3 = 0.2337
[cG,rG] = sgcirc(S,’s’); % cG = 1.5748∠ −162.67o , rG = 0.5162
angles of c and cL will be equal, θc = θcL . Therefore, in order for the load stability
circle and the gain circle to intersect the unit-circle at the same ΓL = ejθL , the following
smith; smithcir(cG,rG); % plot entire source stability circle condition must be satisfied:
smithcir(c1,r1); smithcir(c2,r2); smithcir(c3,r3);
1 + |c|2 − r 2 1 + |cL |2 − rL2
cos(θL − θc )= = (13.11.5)
Again, the circles lie entirely within the unit circle. As ΓG traces the 15-dB circle, the 2|c| 2|cL |
corresponding matched load ΓL = Γ∗ out traces its own circle on the Γ-plane. It can be
 2

Using the identities 1 − |S11 |2 = B2 − D2 and 1 − |S11 |2 = |cL |2 − rL D2 , which
plotted with:
follow from Eqs. (13.5.1) and (13.5.6), we obtain:
phi = linspace(0,2*pi,361); % equally spaced angles at 1o intervals
1 + |cL |2 − rL2 1 + (B2 − D2 )/D2 B2
gammaG = c3 + r3 * exp(j*phi); % points on 15-dB available gain circle = =
gammaL = conj(gout(S,gammaG)); % circle of conjugate matched loads 2|cL | 2|C2 |/|D2 | 2|C2 |
plot(gammaL);
where we used D2 > 0. Similarly, Eq. (13.11.2) can be written in the form:
   
1 − g 1 − |S11 |2 g 1 − |S11 |2 − 1 g(B2 − D2 )−1
r 2 = |c|2 + ⇒ |c|2 − r 2 = =
1 + gD2 1 + gD2 1 + gD2
13.11. Operating and Available Power Gain Circles 559 560 13. S-Parameters

Therefore, we have:
 
1 + |c|2 − r 2 1 + g(B2 − D2 )−1 /(1 + gD2 ) B2
= =
2|c| 2g|C2 |/|1 + gD2 | 2|C2 |

Thus, Eq. (13.11.5) is satisfied. This condition has two solutions for θL that cor-
respond to the two points of intersection with the unit-circle. When D2 > 0, we have
arg c = arg C∗
2 = − arg C2 . Therefore, the two solutions for ΓL = e
jθL
will be:
 
B2
ΓL = ejθL , θL = − arg(C2 )± acos (13.11.6)
2|C2 |
Similarly, the points of intersection of the unit-circle and the available gain circles
and source stability circle are:
 
B1
ΓG = ejθG , θG = − arg(C1 )± acos (13.11.7) Fig. 13.11.2 Operating and available power gain circles.
2|C1 |
Actually, these expressions work also when D2 < 0 or D1 < 0.
% available power gain circles:
Example 13.11.2: The microwave transistor Hewlett-Packard AT-41410 NPN is potentially un- [c1,r1] = sgcirc(S,’a’,20); % c1 = 0.6809∠171.69o , r1 = 0.4137
stable at 1 GHz with the following S-parameters [1324]: [c2,r2] = sgcirc(S,’a’,21); % c2 = 0.7786∠171.69o , r2 = 0.3582
[c3,r3] = sgcirc(S,’a’,22); % c3 = 0.8787∠171.69o , r3 = 0.3228

S11 = 0.6∠−163 , o
S21 = 7.12∠86 , o
S12 = 0.039∠35 , o
S22 = 0.50∠−38 o
figure;
smith; smithcir(cL,rL,1.5); smithcir(cG,rG,1.5);
Calculate GMSG and plot the operating and available power gain circles for G = 20, 21, 22 smithcir(c1,r1); smithcir(c2,r2); smithcir(c3,r3);
dB. Then, design source and load matching circuits for the 22-dB case by choosing the
reflection coefficients that have the smallest magnitudes. gG = c3 - r3*exp(j*angle(c3)); % ΓG of smallest magnitude
gL = conj(gout(S,gG)); % corresponding matched ΓL
Solution: The MSG computed from Eq. (13.6.21) is GMSG = 22.61 dB. Fig. 13.11.2 depicts the plot(gL,’.’); plot(gG,’.’);
operating and available power gain circles as well as the load and source stability circles.
The stability parameters are: K = 0.7667, μ1 = 0.8643, |Δ| = 0.1893, D1 = 0.3242, D2 = Because D1 > 0 and D2 > 0, the stability regions are the portions of the unit-circle that
0.2142. The computations and plots are done with the following MATLAB code:† lie outside the source and load stability circles. We note that the operating gain circles
intersect the unit-circle at exactly the same points as the load stability circle, and the
S = smat([0.60, -163, 7.12, 86, 0.039, 35, 0.50, -38]); % S-parameters available gain circles intersect it at the same points as the source stability circle.
The value of ΓL on the 22-dB operating gain circle that lies closest to the origin is ΓL =
[K,mu,D,B1,B2,C1,C2,D1,D2] = sparam(S); % stability parameters
c3 − r3 ej arg c3 = 0.4773∠50.80o and the corresponding matched source is ΓG = Γ∗ in =
0.7632∠167.69o . We note that both ΓL and ΓG lie in their respective stability regions.
Gmsg = db(sgain(S,’msg’)); % GMSG = 22.61 dB
% operating power gain circles: For the 22-dB available gain circle (also denoted by c3 , r3 ), the closest ΓG to the origin will
[c1,r1] = sgcirc(S,’p’,20); % c1 = 0.6418∠50.80o , r1 = 0.4768 be ΓG = c3 − r3 ej arg c3 = 0.5559∠171.69o with a corresponding matched load ΓL = Γ∗ out =
[c2,r2] = sgcirc(S,’p’,21); % c2 = 0.7502∠ 50.80o , r2 = 0.4221
0.7147∠45.81o . Again, both ΓL , ΓG lie in their stable regions.
[c3,r3] = sgcirc(S,’p’,22); % c3 = 0.8666∠ 50.80o , r3 = 0.3893
% load and source stability circles: Once the ΓG , ΓL have been determined, the corresponding matching input and output
[cL,rL] = sgcirc(S,’l’); % cL = 2.1608∠50.80o , rL = 1.2965 networks can be designed with the methods of Example 13.8.1. 

[cG,rG] = sgcirc(S,’s’); % cG = 1.7456∠171.69o , rG = 0.8566

smith; smithcir(cL,rL,1.5); smithcir(cG,rG,1.5); % plot Smith charts


smithcir(c1,r1); smithcir(c2,r2); smithcir(c3,r3); % plot gain circles
13.12 Noise Figure Circles
gL = c3 - r3*exp(j*angle(c3)); % ΓL of smallest magnitude Every device is a source of internally generated noise. The noise entering the device and
gG = conj(gin(S,gL)); % corresponding matched ΓG the internal noise must be added to obtain the total input system noise. If the device is
plot(gL,’.’); plot(gG,’.’);
an amplifier, the total system noise power will be amplified at the output by the gain of
† The function db converts absolute scales to dB. The function ab converts from dB to absolute units. the device. If the output load is matched, this gain will be the available gain.
13.12. Noise Figure Circles 561 562 13. S-Parameters

The internally generated noise is quantified in practice either by the effective noise The MATLAB function nfcirc implements Eq. (13.12.5). Its inputs are the noise
temperature Te , or by the noise figure F of the device. The internal noise power is given parameters Fmin , rn , ΓGopt , and the desired value of F in dB, and its outputs are c, r :
by Pn = kTe B, where k is the Boltzmann constant and B the bandwidth in Hz. These
concepts are discussed further in Sec. 15.8. The relationship between Te and F is defined [c,r] = nfcirc(F,Fmin,rn,gGopt); % noise figure circles

in terms of a standard reference temperature T0 = 290 K (degrees Kelvin):


The function nfig implements Eq. (13.12.3). Its inputs are Fmin , rn , ΓGopt , and a
Te vector of values of ΓG , and its output is the corresponding vector of values of F:
F =1+ (13.12.1)
T0
F = nfig(Fmin, rn, gGopt, gG); % calculate noise figure F in dB
The noise figure is usually quoted in dB, FdB = 10 log10 F. Because the available gain
of a two-port depends on the source impedance ZG , or the source reflection coefficient Example 13.12.1: The microwave transistor of Example 13.11.1 has the following noise param-
ΓG , so will the noise figure. eters at 2 GHz [1324]: Fmin = 1.6 dB, rn = 0.16, and ΓGopt = 0.26∠172o .
The optimum source impedance ZGopt corresponds to the minimum noise figure Determine the matched load ΓLopt corresponding to ΓGopt and calculate the available gain.
Fmin that can be achieved by the two-port. For other values of ZG , the noise figure F is Then, plot the noise figure circles for F = 1.7, 1.8, 1.9, 2.0 dB.
greater than Fmin and is given by [94–97]:
For the 1.8-dB noise figure circle, determine ΓG , ΓL that correspond to the maximum pos-
Rn sible available gain and design appropriate input and output matching networks.
F = Fmin + |ZG − ZGopt |2 (13.12.2)
RG |ZGopt |2 Solution: The conjugate matched load corresponding to ΓGopt is:
where RG = Re(ZG ) and Rn is an equivalent noise resistance. We note that F = Fmin
 ∗
when ZG = ZGopt . Defining the normalized noise resistance rn = Rn /Z0 , where Z0 = S22 − ΔΓGopt
ΓLopt = Γ∗
out = = 0.4927∠52.50o
50 ohm, we may write Eq. (13.12.2) in terms of the corresponding source reflection 1 − S11 ΓGopt
coefficients:
The value of the available gain at ΓGopt is Ga,opt = 13.66 dB. This is to be compared with the
|ΓG − ΓGopt |2 MAG of 16.18 dB determined in Example 13.11.1. To increase the available gain, we must
F = Fmin + 4rn   (13.12.3) also increase the noise figure. Fig. 13.12.1 shows the locations of the optimum reflection
|1 + ΓGopt |2 1 − |ΓG |2
coefficients, as well as several noise figure circles.
The parameters Fmin , rn , and ΓGopt characterize the noise properties of the two-port The MATLAB code for generating this graph was:†
and are usually known.
In designing low-noise microwave amplifiers, one would want to achieve the mini-
mum noise figure and the maximum gain. Unfortunately, the optimum source reflection
coefficient ΓGopt does not necessarily correspond to the maximum available gain.
The noise figure circles and the available gain circles are useful tools that allow one
to obtain a balance between low noise and high gain designs. The noise figure circles
are the locus of points ΓG that correspond to fixed values of F. They are obtained by
rewriting Eq. (13.12.3) as the equation of a circle |ΓG − c|2 = r 2 . We write Eq. (13.12.3)
in the form:

|ΓG − ΓGopt |2 (F − Fmin )|1 + ΓGopt |2


= N, where N = (13.12.4)
1 − |ΓG |2 4rn
which can be rearranged into the circle equation:
 2   Fig. 13.12.1 Noise figure circles.
 
ΓG − ΓGopt  = N + N 1 − |ΓGopt |
2 2
 N + 1 (N + 1)2
S = smat([0.61, 165, 3.72, 59, 0.05, 42, 0.45, -48]);
Thus, the center and radius of the noise figure circle are:

  
Fmin = 1.6; rn = 0.16; gGopt = p2c(0.26, 172);

ΓGopt N2 + N 1 − |ΓGopt |2 † The function p2c converts from phasor form to cartesian complex form, and the function c2p, from
c= , r= (13.12.5)
N+1 N+1 cartesian to phasor form.
13.12. Noise Figure Circles 563 564 13. S-Parameters

Gmag = db(sgain(S,’mag’)); % maximum available gain


Gaopt = db(sgain(S,gGopt,’a’)) % available gain at ΓGopt

gLopt = conj(gout(S,gGopt)); % matched load

[c1,r1] = nfcirc(1.7,Fmin,rn,gGopt); % noise figure circles


[c2,r2] = nfcirc(1.8,Fmin,rn,gGopt);
[c3,r3] = nfcirc(1.9,Fmin,rn,gGopt);
[c4,r4] = nfcirc(2.0,Fmin,rn,gGopt);

smith; plot([gGopt, gLopt],’.’);


smithcir(c1,r1); smithcir(c2,r2); smithcir(c3,r3); smithcir(c4,r4);

The larger the noise figure F, the larger the radius of its circle. As F increases, so does
the available gain. But as the gain increases, the radius of its circle decreases. Thus, for a Fig. 13.12.3 Maximum available gain for given noise figure.
fixed value of F, there will be a maximum value of the available gain corresponding to that
gain circle that has the smallest radius and is tangent to the noise figure circle.
Ga = 14.81 dB gain circle, which is tangential to the 1.8-dB noise figure circle, are shown
In the extreme case of the maximum available gain, the available gain circle collapses in Fig. 13.12.3.
to a point—the simultaneous conjugate matched point ΓG = 0.8179∠−162.67o — with a
The following MATLAB code performs these calculations and plots:
corresponding noise figure of F = 4.28 dB. These results can be calculated by the MATLAB
statements:
phi = linspace(0,2*pi,721); % angle in 1/2o increments
gG = c2 + r2*exp(j*phi); % ΓG around the c2 , r2 circle
gG = smatch(S); G = db(sgain(S,gG,’a’)); % available gain in dB
F = nfig(Fmin, rn, gopt, gG); plot(phi*180/pi, G);

Thus, we see that increasing the gain comes at the price of increasing the noise figure. [Ga,i] = max(G); % maximum available gain

As ΓG traces the F = 1.8 dB circle, the available gain Ga varies as shown in Fig. 13.12.2.
gammaG = gG(i); % ΓG for maximum gain
Points around this circle can be parametrized as ΓG = c2 + r2 ejφ , with 0 ≤ φ ≤ 2π. gammaL = conj(gout(S,gammaG)); % matched load ΓL
Fig. 13.12.2 plots Ga versus the angle φ. We note that the gain varies between the limits
12.22 ≤ Ga ≤ 14.81 dB. [ca,ra] = sgcirc(S,’a’,Ga); % available gain circle

Available Gain for F = 1.8 dB smith; smithcir(c2,r2); smithcir(ca,ra);


15 plot([gammaG,gammaL],’.’);

The maximum gain and the point of tangency with the noise figure circle are determined by
14 direct search, that is, evaluating the gain around the 1.8-dB noise figure circle and finding
where it reaches a maximum.
Ga (dB)

The input and output stub matching networks can be designed with the help of the function
13
stub1. The normalized source and load impedances are:

1 + ΓG 1 + ΓL
zG = = 0.3840 − 0.0767j , zL = = 1.0904 + 1.3993j
1 − ΓG 1 − ΓL
12
0 90 180 270 360
φ (degrees) The stub matching networks have lengths:
   
Fig. 13.12.2 Variation of available gain around the noise figure circle F = 1.8 dB. 0.3749 0.3977 0.3519 0.0991
stub1(z∗
G , ’po’)= , stub1(z∗
L , ’po’)=
0.1251 0.0738 0.1481 0.2250
The maximum value, Ga = 14.81 dB, is reached when ΓG = 0.4478∠−169.73o , with a
The lower solutions have shorter total lengths d + l. The implementation of the matching
out = 0.5574∠52.50 . The two points ΓG , ΓL , as well as the
resulting matched load ΓL = Γ∗ o

networks with microstrip lines will be similar to that in Fig. 13.8.4. 



13.13. Problems 565 566 13. S-Parameters

If the two-port is potentially unstable, one must be check that the resulting solutions 13.7 Computer Experiment. The Hewlett-Packard ATF-36163 pseudomorphic high electron mo-
for ΓG , ΓL both lie in their respective stability regions. Problems 13.6 and 13.7 illustrate bility transistor (PHEMT) has the following S- and noise parameters at 6 GHz [1324]:
the design of such potentially unstable low noise microwave amplifiers.
S11 = 0.75∠−131o , S21 = 3.95∠55o , S12 = 0.13∠−12o , S22 = 0.27∠−116o

Fmin = 0.66 dB, rn = 0.15, ΓGopt = 0.55∠88o


13.13 Problems
Plot the F = 0.7, 0.8, 0.9 dB noise figure circles. On the 0.7-dB circle, determine the source
13.1 Using the relationships (13.4.3) and (13.4.6), derive the following identities: reflection coefficient ΓG that corresponds to maximum available gain, and then determine
the corresponding matched load coefficient ΓL .
(Z11 + ZG )(Z22 + ZL )−Z12 Z21 =
(13.13.1) Design microstrip stub matching circuits for the computed values of ΓG , ΓL .
(Z22 + ZL )(Zin + ZG )= (Z11 + ZG )(Zout + ZL ) 13.8 Computer Experiment. In this experiment, you will carry out two low-noise microwave am-
(1 − S11 ΓG )(1 − S22 ΓL )−S12 S21 ΓG ΓL = plifier designs, including the corresponding input and output matching networks. The first
(13.13.2) design fixes the noise figure and finds the maximum gain that can be used. The second
(1 − S22 ΓL )(1 − Γin ΓG )= (1 − S11 ΓG )(1 − Γout ΓL ) design fixes the desired gain and finds the minimum noise figure that may be achieved.
Using Eqs. (13.4.4) and (13.4.5), show that: The Hewlett-Packard Agilent ATF-34143 PHEMT transistor is suitable for low-noise ampli-
fiers in cellular/PCS base stations, low-earth-orbit and multipoint microwave distribution
Z21 S21 1 − ΓL Z21 S21 1 − ΓG systems, and other low-noise applications.
= , = (13.13.3)
Z22 + ZL 1 − S22 ΓL 1 − Γin Z11 + ZG 1 − S11 ΓG 1 − Γout At 2 GHz, its S-parameters and noise-figure data are as follows, for biasing conditions of
VDS = 4 V and IDS = 40 mA:
2 Z0 (1 − Γin )(1 − ΓG ) 2Z 0 (1 − Γout )(1 − ΓL )
= , = (13.13.4)
Zin + ZG 1 − Γin ΓG Zout + ZL 1 − Γout ΓL
S11 = 0.700∠−150o , S12 = 0.081∠19o
Finally, for the real part RL = Re(ZL ), show that: S21 = 6.002∠73o , S22 = 0.210∠−150o
1 + ΓL 1 − |ΓL |2
ZL = Z 0 ⇒ RL = Z 0 (13.13.5) Fmin = 0.22 dB, rn = 0.09, ΓGopt = 0.66∠67o
1 − ΓL |1 − Γ L |2
a. At 2 GHz, the transistor is potentially unstable. Calculate the stability parameters
13.2 Computer Experiment. The Hewlett-Packard ATF-10136 GaAs FET transistor has the follow-
K, μ, Δ, D1 , D2 . Calculate the MSG in dB.
ing S-parameters at 4 GHz and 8 GHz [1324]:
Draw a basic Smith chart and place on it the source and load stability circles (display
S11 = 0.54∠−120o , S21 = 3.60∠61o , S12 = 0.137∠31o , S22 = 0.22∠−49o only a small portion of each circle outside the Smith chart.)
S11 = 0.60∠87o , S21 = 2.09∠−32o , S12 = 0.21∠−36o , S22 = 0.32∠−48o Then, determine the parts of the Smith chart that correspond to the source and load
stability regions.
Determine the stability parameters, stability circles, and stability regions at the two frequen-
b. For the given optimum reflection coefficient ΓGopt , calculate the corresponding load
cies.
reflection coefficient ΓLopt assuming a matched load.
13.3 Derive the following relationships, where RG = Re(ZG ):
Place the two points ΓGopt , ΓLopt on the above Smith chart and determine whether they
Z0 + Z G 1 |1 − Γ G | Z0 − Z G ΓG |1 − ΓG | lie in their respective stability regions.
 =  ,  = −
2 R G Z0 1 − |ΓG |2 1 − ΓG 2 RG Z0 1 − |ΓG |2 1 − ΓG c. Calculate the available gain Ga,opt in dB that corresponds to ΓGopt .

13.4 Derive Eqs. (13.7.13) relating the generalized S-parameters of power waves to the conven- Add the corresponding available gain circle to the above Smith chart. (Note that the
tional S-parameters. source stability circle and the available gain circles intersect the Smith chart at the
same points.)
13.5 Derive the expression Eq. (13.6.20) for the maximum available gain GMAG , and show that it
is the maximum of all three gains, that is, transducer, available, and operating gains. d. Add to your Smith chart the noise figure circles corresponding to the noise figure
values of F = 0.25, 0.30, 0.35 dB.
13.6 Computer Experiment. The microwave transistor of Example 13.11.2 has the following noise
For the case F = 0.35 dB, calculate and plot the available gain Ga in dB as ΓG traces
parameters at a frequency of 1 GHz [1324]: Fmin = 1.3 dB, rn = 0.16, and ΓGopt = 0.06∠49o .
the noise-figure circle. Determine the maximum value of Ga and the corresponding
Determine the matched load ΓLopt corresponding to ΓGopt and calculate the available gain.
value of ΓG .
Then, plot the noise figure circles for F = 1.4, 1.5, 1.6 dB.
Place on your Smith chart the available gain circle corresponding to this maximum Ga .
For the 1.5-dB noise figure circle, determine the values of ΓG , ΓL that correspond to the
Place also the corresponding point ΓG , which should be the point of tangency between
maximum possible available gain.
the gain and noise figure circles.
Design microstrip stub matching circuits for the computed values of ΓG , ΓL .
13.13. Problems 567

Calculate and place on the Smith chart the corresponding load reflection coefficient
ΓL = Γ∗out . Verify that the two points ΓG , ΓL lie in their respective stability regions.
In addition, for comparison purposes, place on your Smith chart the available gain
circles corresponding to the values Ga = 15 and 16 dB.
e. The points ΓG and ΓL determined in the previous question achieve the maximum gain
for the given noise figure of F = 0.35 dB.
Design input and output stub matching networks that match the amplifier to a 50-ohm
generator and a 50-ohm load. Use “parallel/open” microstrip stubs having 50-ohm
characteristic impedance and alumina substrate of relative permittivity of r = 9.8.
Determine the stub lengths d, l in units of λ, the wavelength inside the microstrip lines.
Choose always the solution with the shortest total length d + l.
Determine the effective permittivity eff of the stubs, the stub wavelength λ in cm, and
the width/height ratio, w/h. Then, determine the stub lengths d, l in cm.
Finally, make a schematic of your final design that shows both the input and output
matching networks (as in Fig.10.8.3.)
f. The above design sets F = 0.35 dB and finds the maximum achievable gain. Carry out
an alternative design as follows. Start with a desired available gain of Ga = 16 dB and
draw the corresponding available gain circle on your Smith chart.
As ΓG traces the portion of this circle that lies inside the Smith chart, compute the
corresponding noise figure F. (Points on the circle can be parametrized by ΓG = c +
rejφ , but you must keep only those that have |ΓG | < 1.)
Find the minimum among these values of F in dB and calculate the corresponding
value of ΓG . Calculate the corresponding matched ΓL .
Add to your Smith chart the corresponding noise figure circle and place on it the points
ΓG and ΓL .
g. Design the appropriate stub matching networks as in part 13.8.

You might also like