10-1

Solutions for Chapter 10 Problems

1. Lumped Element Matching Networks

P10.1: A matching network consists of a length of a T-Line through section in series with
a capacitor. Determine the length (in wavelengths) required of the through section and
the capacitor value needed (at 1.0 GHz) to match a 10 - j35  load impedance to the 50
 line.

Fig. P6.32a
This problem is identical to P6.32:
We find the normalized load, zL = 0.2 –
j0.7, located at point a (WTG = 0.400).
Now we move from point a clockwise
(towards the generator) until we reach
point b, where we have z = 1 + j2.4. Fig. P6.32b
Moving from a to b corresponds to d =
0.500+0.194-0.400 = 0.294. For the series capacitance we have
j
 j 2.4  ,
CZ o
1
or C   1.33 pF
2  1x109   50   2.4 

P10.2: Design an L-section matching

network to match a 10 + j15  load to a
50  line. Determine specific values of
the lumped elements at a 1.0 GHz
operating frequency.

Referring to the figure, we normalize the

load (point a) at zL = 0.2 + j0.3. We then
add series +j0.1 to move along the real =
0.2 circle to the point zb = 0.2+j0.4 (point
b). We next jump to the admittance chart,
at yb = 1.0-j2.0 (point b’). Finally, we add
shunt +j2.0 to reach the center of the chart
at point c’. Fig. P10.2(a)
 10-2

Realizing these components, we have for

the series element:

Fig. P10.2b

j 0.1 
j L
, or L 
 0.1  50   0.796nH .
So L = 0.80 nH. Then, the shunt element is:
Zo 2  1x109 
j 2.0  jCZ o ,
2.0
or C   6.37 pF .
2  1x109  50
So C = 6.4 pF. The final circuit is shown in Fig P10.2 (b). An alternate solution is also
shown.

P10.3: Design an L-section matching network to match a 80 - j50  load to a 50  line.

Determine specific values of the lumped elements at a 1.0 GHz operating frequency.

We first normalize and locate the load, zL

= 1.6 – j1.0 (point a). Then we move to
the admittance chart (point a’) at yL =
0.44+j0.28. We add +j0.22 (shunt
element) to get to the rotated circle (point
b’) where we have yb = 0.44 + j0.50.
Next, we move back to the impedance
chart, at zb = 1 – j1.1 (point b). Finally,
adding a series +j1.1 brings us to the
matched condition.
Realizing these components, we have for
the shunt +j0.22:
j 0.22  jCZ o ,
0.22
or C   0.7 pF .
2  1x109   50 
For the series +j1.1 element:
j L
j1.1  ,
Zo

or L 
 1.1  50   8.7 nH . Fig. P10.3
2  1x109 
 10-3

The final circuit is shown in Fig P10.3. An alternate solution replaces the shunt capacitor
with a shunt inductor of value 10 nH, and the series inductor is replaced with a series
capacitor of value 2.9 pF.

P10.4: Design an L-section matching network to match a 30 + j70  load to a 50  line.

Determine specific values of the lumped elements at a 2.5 GHz operating frequency.

The normalized load is zL = 0.6+j1.4, at

point a. We move to the rotated circle by
adding –j0.9 (series element) to get to zb =
0.6+j0.5 (point b). We then move to the
admittance chart (point b’) at yb = 1 – j0.8.
Finally, adding a shunt element of value
+j0.8 brings us to the matched condition.
Realizing these components, for the series
–j0.9 we have:
j
 j 0.9  , or
CZ o
1
C  1.4 pF .
2  2.5 x109   50   0.9 
For the shunt +j0.8 we have:
j 0.8  jCZ o , or

C
 0.8  1.02 pF .
2  2.5 x109   50 
The figure shows the final matching Fig. P10.4
network. An alternate solution replaces
the series capacitance with another series capacitance of value 0.68pF, and the shunt
capacitance with a shunt inductor of value 4.1 nH.

P10.5: Suppose you want to match a 20 + j50  load to a 50  line. For the design of an
L-section matching network, you notice the normalized load impedance lies outside both
the 1 ± jx circle and the rotated 1 ± jx circle. Find all four possible solutions, and for
each one determine specific values of the lumped elements at a 2.5 GHz operating
frequency.

In each solution, we proceed from a to b to c, with primes denoting points in the

admittance chart.

Conversion of normalized element values to components values shown in the figures is

accomplished using Table 10.2.
 10-4

Fig. P10.5a Fig. P10.5b

Fig. P10.5c Fig. P10.5d

 10-5

P10.6: Suppose you want to match a 100  line to a load ZL = 200 - j100  (a resistor in
series with a capacitor) at a frequency of 500 MHz. (a) Determine the element values for
the load. (b) Design a shorted shunt stub matching network. (c) Design an L-section
matching network.

Fig. P10.6b

Fig. P10.6a

The load is a series resistance/capacitance combination, where the capacitance is given

by:
j 1
 j100  , or C   3.18 pF .
C 2  500 x106   100 
For the stub matching network, we first locate the normalized load zL on the Smith Chart,
move to the normalized admittance yL, and then proceed along the constant || circle to
the point y = 1+j1, traveling a through-line length of 0.125G. Then, we travel from the
normalized admittance of a short circuit (at 0.250G on the WTG scale) to y = 0 - j1,
traveling a stub line length of 0.125G. So for this particular problem the through and
stub lengths are the same.

Next, we turn to the lumped element matching problem. We locate the load at zL = 2 – j1
(point a), and then move to yL = 0.4 + j0.2 (point a’). Then we move a distance +j0.3
 10-6

(corresponding to a shunt element) to yb = 0.4 + j0.5 (point b’). Switching to the

impedance chart, or zb = 1 – j1.2, we add a series +j1.2 to reach the matched condition
(point c).
Realizing these components, we have for the shunt j0.3:
j 0.3  jCZ o , or C 
 0.3  0.95 pF .
2  500 x106   100 
For the series j1.2:
j1.2 
j L
, or L 
 1.2   100   38nH .
Zo 2  500 x106 

P10.7: (JustAsk): Design an L-section matching network to match a load ZL = 100 + j80
 to a 50  line. Find the lumped element values at an operating frequency of 11.18
GHz. Further, your design should allow for DC biasing the load element through the
matching network.

Allowing DC bias through the network

requires use of a shunt capacitor and a
series inductor. Referring to the figure,
we first normalize the load, zL = 2 + j1.6
(point a), and find the corresponding point
in the normalized admittance chart, yL = .
31 – j.25 (point a’). To get a shunt
capacitance, we must add a positive
admittance. So we move to the point yb = .
31 + j.47 (point b’) by adding shunt
+j0.72. Then we move to the impedance
chart, at zb = 1 – j1.45. Adding a series
+j1.45 gives us the necessary series
inductance. The element value for the
shunt capacitance is:
j 0.72  jCZ o ,

or C 
 0.72   0.20 pF .E
2  11.18 x109   50 
The series inductance is:
j L
j1.45  ,
Zo Fig. P10.7

or L 
 1.45  50   1.03nH .
2  11.18 x109 
 10-7

P10.8: Design an L-section matching

network for a load that has a 25.0 
resistor in series with a 1.061 pF capacitor.
Assume a 50  system impedance at 3.0
GHz.

We normalize the load to find zL = 0.5 –

j1.0 (point a), and then add series j0.5 to
reach zb = 0.5 –j0.5 (point b). Moving to
the admittance chart, yb = 1 + j1, so we
add a shunt –j1.0 to reach the matched
condition.
The element values are:
j L
j 0.5  , or
Zo

L
 0.5   50   1.33nH .
2  3 x109 
and
 jZ o
 j1  , or
L
Fig. P10.8
L
 50   2.65nH .
2  3 x109 

P10.9: MATLAB: There are two

fundamental solutions for the L-section
matching network of P10.2. Develop a
routine to plot the || versus frequency for
both solutions from 0.1 GHz to 10 GHz.
(Hint: this is somewhat similar to
MATLAB 9.1).

The two fundamental solutions are shown

in Figure P10.9a. For the top solution, we
have:
j
Z in1   j Lm1  Z L 
Cm1
And for the bottom solution:
 j 
Z in 2  j Lm 2   ZL 
  Cm 2 
Once Zin is found, then Fig. P10.9a
 10-8

Z in  Z o
 ,
Z in  Z o
and we can plot || versus frequency.
From P10.2 we have ZL = 10 + j15 , Lm1 = 0.80 nH, Cm1 = 6.4 pF, Lm2 = 4.0 nH and Cm2
= 4.5 pF. The following MATLAB routine generates the plot.

% MLP1009
%
% Plot magnitude of reflection coefficient
% from 0.1GHz to 10 GHz looking into each
% matching network from P10.02.
%
clc
clear
ZL=10+j*15;
Zo=50;
f=0.1:0.01:10;
w=2*pi.*f*1e9;
Lm1=0.80e-9;
XL1=j*w.*Lm1;
Cm1=6.4e-12;
XC1=-j./(w.*Cm1);
Zin1=parallel(XC1,ZL+XL1);
Ref1=abs((Zin1-Zo)./(Zin1+Zo));

Lm2=4.0e-9;
Cm2=4.5e-12;
XL2=j*w.*Lm2;
XC2=-j./(w.*Cm2);
Zin2=parallel(XL2,ZL+XC2);
Ref2=abs((Zin2-Zo)./(Zin2+Zo));

semilogx(f,Ref1,'-*',f,Ref2,'-o')
Legend('L=.8nH,C=6.4pF','C=4.5pF,L=4nH')
xlabel('frequency(GHz)')
ylabel('magnitude of reflection coefficient')
grid on
 10-9

Fig P10.9b

P10.10: MATLAB: There are two

fundamental solutions for the L-section
matching network of P10.3. Develop a
routine to plot the || versus frequency for
both solutions from 0.1 GHz to 10 GHz.

The two fundamental solutions are shown in

Figure P10.10a. For the top solution, we
have:
j
Z in1  j Lm1  Z
Cm1 L
And for the bottom solution:
j
Z in 2   j Lm 2 Z L
 Cm 2
Once Zin is found, then
Z  Zo
  in ,
Z in  Z o Fig. P10.10a
and we can plot || versus frequency.
From P10.3 we have ZL = 80 – j50 , Lm1 = 8.7 nH, Cm1 = 0.7 pF, Lm2 = 10 nH and Cm2 =
2.9 pF. The following MATLAB routine generates the plot.

% MLP1010
% Plot magnitude of reflection coefficient
% from 0.1GHz to 10 GHz looking into each
 10-10

% matching network from P10.03.

clc
clear
ZL=80-j*50;
Zo=50;
f=0.1:0.01:10;
w=2*pi.*f*1e9;

Lm1=8.7e-9;
XL1=j*w.*Lm1;
Cm1=0.7e-12;
XC1=-j./(w.*Cm1);
Zin1=XL1+parallel(ZL,XC1);
Ref1=abs((Zin1-Zo)./(Zin1+Zo));

Lm2=10e-9;
Cm2=2.9e-12;
XL2=j*w.*Lm2;
XC2=-j./(w.*Cm2);
Zin2=XC2+parallel(ZL,XL2);
Ref2=abs((Zin2-Zo)./(Zin2+Zo));

semilogx(f,Ref1,'-*',f,Ref2,'-o')
Legend('L=8.7nH,C=.6pF','C=2.9pF,L=10nH')
xlabel('frequency(GHz)')
ylabel('magnitude of reflection coefficient')
grid on

Fig. P10.10b
 10-11

P10.11: MATLAB: Suppose the L-section matching network of Example 10.1 is realized
with a capacitor that can be characterized by the circuit model of Figure 9.13(b), where
Rx = 0.010  and Lx = 7.2 nH, and an inductor that can be characterized by the circuit
model of Figure 9.10 where Rx = 0.10  and Cx = 5.2 pF. Compare plots of || versus
frequency for the ideal case to the case where parasitics are included. The frequency
range is from 0.1 GHz to 10 GHz.

Replacing the shunt capacitance with its circuit model we get the circuit of Fig p10.11a.
We now have:
 j 
Z in  j Lm   Rx   j Lx  Z L
 C 
Then,
Z  Zo
  in ,
Z in  Z o
and we can plot || versus frequency.

% MLP1011
%
% Plot magnitude of reflection coefficient
% from 0.1GHz to 10 GHz looking into each
% matching network of Ex10.1. Compare with
% plot including parasitic effects.
%
clc
clear
ZL=250-j*250;
Zo=50;
f=0.1:0.01:10;
w=2*pi.*f*1e9;

Lm1=30e-9;
XL1=j*w.*Lm1;
Cm1=0.8e-12;
XC1=-j./(w.*Cm1);
Zin1=XL1+parallel(ZL,XC1);
Ref1=abs((Zin1-Zo)./(Zin1+Zo));

Rx=.010;
Lx=7.2e-9;
XC2=XC1+j*w*Lx+Rx;
Zin2=XL1+parallel(ZL,XC2);
Ref2=abs((Zin2-Zo)./(Zin2+Zo));
 10-12

semilogx(f,Ref1,'-*',f,Ref2,'-o')
Legend('ideal cap','cap with parasitics')
xlabel('frequency(GHz)')
ylabel('magnitude of reflection coefficient')
grid on

Fig. P10.11a

In this problem, the parasitics

introduce only a slight change
in the resonance frequency.
Fig. P10.11b

2. Scattering Parameters
P10.12: Find the scattering matrices for the simple two-port networks shown in Figure
10.50.

A port terminated in a short circuit will have a reflection coefficient of -1. An open
termination will have a +1 reflection coefficient. The ideal through connections will have
no reflection. By inspection, then, the scattering matrices for the circuits shown are:
0 1  1 0 
( a)  S     (b )  S    
1 0  0 1 
(a)
 1 0  1 0 
(c )  S     (d )  S    
 0 1  0 1

P10.13: Cut a 50  T-Line and insert a series 50  resistor followed by a shunt 50 

resistor. Determine the scattering matrix for this 2-port network. Is the network lossless?
Is it reciprocal? Calculate the insertion loss.
 10-13

The circuit (Figure P10.13(a)) is redrawn

in (b) to calculate S11 and S21. Here,
75  50 1
S11      0.2
75  50 5
V  25   1  1   2 
S 21  2        1    
V1  75   5  3   5 
S 21  0.40
Figure P10.13(c) is used to calculate S22
and S12:
Z L  50 100  33.3,
33.3  50
S 22     0.2
33.3  50
V  50  1
S12  1      0.8    0.4
V2  100  2
Fig. P10.13 0.20 0.40 
So:  S     , and we see by
0.40 0.20 
inspection that the scattering matrix is reciprocal. It is also not lossless (summing the
squares of any column yields a number less than one). Finally, to calculate the insertion
loss:
IL21  20 log S 21  20 log 0.4  8dB

P10.14: In a 50  system, a two-port network consists of a 25  series resistor followed

by a 50  shunt resistor (see Figure 10.11(a)). Calculate the return loss looking into port
1 of this network if port 2 is terminated in a 100  resistor.

The difficult approach:

We have the scattering matrix from
example 10.5 for the circuit of Fig
P10.14(a):
0 1 
2
 S  

1 1 
 2 4
By staring with
V1  S11V1  S12V2
V2  S21V1  S22V2
Fig. P10.14
and considering that
100  50  1 
V2  V2  V2  V2 ,
100  50 3
we can manipulate the scattering parameters to arrive at:
 10-14

V1
 S 
S12 S21
 0 
1
2   
1
2  0.07692
V1 11
3  S22 3 1 
4
The return loss is then:
RL  20 log 0.07692  22.3dB.

The easier approach:

All we need is the reflection coefficient looking into the circuit of Fig P10.14(b). We
have
Z L  25  50 100  58.33,
Z L  Z o 58.33  50
L    0.07693,
Z L  Z o 58.33  50
and then
RL  20log 0.07693  22.3dB

P10.15: A series capacitor of value C = 2.0 pF is inserted in a 50  T-Line. At 1.0 GHz,

determine [S], the return loss and the insertion loss.

The impedance of the capacitor at 1 GHz

is
j j
Zc     j80
C 2  1x10   2 x1012 
9

To find S11,
50  j80  50
S11   
50  j80  50
 j80
  0.625e  j 51


100  j80
And for S21,
Fig. P10.15
V 50
S 21  2  
V2 50  j80


 1  0.625e  j 51


 50 50j80  0.78e j 39

So
0.625e  j 51 0.780e j 39 
 

  
S  ,
 0.780e j 39  j 51


0.625e
RL  20 log S11  4.1dB
IL  20 log S 21  2.2dB
 10-15

P10.16: A series inductor of value L = 3.5 nH is inserted in a 50  T-Line. At 1.0 GHz,

determine [S], the return loss and the insertion loss.

The impedance of the inductor at 1 GHz is

Fig. P10.16

Z L  j L  j 2  1x109   3.5 x109   j 22

To find S11,
50  j 22  50
S11   
50  j 22  50
 0.215e j 77.6


And for S21,

V 50
S 21  2  
V2 50  j 22


 1  0.215e j 77.6


 50 50j 22  0.977e  j12.4

So
0.215e j 78 0.98e  j12  RL  20 log S11  13.4dB
 

 S    , and
 0.98e  j12 0.215e j 78  IL  20 log S 21  0.18dB
 

P10.17: The scattering matrix for a 3-port network is

 0.60 0 j 0.80 
 
 S    0 1.0e j 90

0 .
 j 0.80 0 0.60 
 
 10-16

(a) Is this network reciprocal? (b) Is it lossless? (c) Determine the return loss at port 1 if
ports 2 and 3 are connected together by a matched T-Line of electrical length 45.
(a) by inspection, we see the network is reciprocal.
(b) Each column’s values, when squared, add to one, so the network appears to be
lossless. This is confirmed by evaluation of  S   S  which indeed results in a unitary
t *

matrix.
(c) To determine the return loss, we need to determine V1 V1 .
We start with:
(1)V1  0.6V1  j 0.8V3
(2)V2  1e j 90 V2  jV2


(3)V3  j 0.8V1  0.6V3

Examining Figure P10.17(b), we have
(4)V2  V3 e  j 45


(5)V3  V2 e  j 45 .


Inserting (5) and (2) into (3), we have

 
(6)V3  j 0.8V1  0.6 V2 e j 45  j 0.8V1  0.6e  j 45 jV2
 

Now inserting (4) into (6),

 
(7)V3  j 0.8V1  0.6e  j 45 j V3e  j 45  j 0.8V1  0.6V3
 


and solving (7) for V3
(8)V3  1  0.6   j 0.8V1 , so V3  j 2V1 .
Next, inserting (8) into (4) we have
(9)V2   j 2V1  e j 45  j 2e  j 45 V1 ,
 

and plugging this into (2),

 
(10)V2  j j 2e  j 45 V1  2e  j 45 V1
 

(a) (b)
Fig. P10.17
 10-17

Finally, inserting this value into (5):

 
(11)V3  2e  j 45 V1 e  j 45  j 2V1 .
 

Inserting (11) into (1) leads us to the desired ratio:

(12)V1  0.6V1  j 0.8  j 2V1   1V1
V1
or   1 , and RL = 0 dB.
V1

P10.18: (JustAsk): The scattering matrix

(assuming a 50  impedance system) for
a 2-port network is:
 0.5 0.5e j 45 


 S    j 45 .
 0.5e 0.5e j 90 


(a) Is this network reciprocal? (b) Is this Fig. P10.18

network lossless? (c) Determine the
return loss looking into port 1 if port 2 is
terminated in an open-ended Zo stub of electrical length 45.

By inspection it is reciprocal but not lossless. For the scattering matrix:

(1)V1  S11V1  S12V2  0.5V1  0.5e j 45 V2


(2)V2  S21V1  S22V2  0.5e j 45 V1  0.5e j 90 V2

 

Finding the return loss we will need the relationship between V2 and V2 resulting from
the 45 line. There is a round trip delay of 90, and at the open end the reflected wave
equals the incident wave. So we have
(3)V2  V2 e  j 90 .


Inserting (3) into (2) we can solve for V2 :

 S21V1
(4)V 
2
1  S22 e  j 90


(4) is now inserted into (3) to find

S21e  j 90 V1



(5)V2   .
1  S22 e  j 90
Finally, inserting (5) into (1) we can find the ratio needed:
V1 S12 S21e  j 90


 S11 
V1 1  S22 e  j 90


 0.5 
 0.5e   0.5e  e
j 45 j 45  j 90

1
1   0.5e  e j 90  j 90

 RL  20log  1  0dB
 10-18

This is an interesting result. Note that terminating port 2 in a short circuit results in a
7dB return loss.

P10.19: Three T-Lines with the same characteristic impedance Zo are connected as shown
in Figure 10.51. Determine the scattering matrix that represents this 3-port network. Is
this network reciprocal? Is it lossless?

The load seen by each line is a pair of Zo impedances in parallel. So

1
1 1
S11   L  2  ,
1 3
1
2
2
S21    1   L  .
3
So the scattering matrix is:
 1 2 2 
1
 S    2 1 2  .
3
 2 2 1

3. Couplers and Dividers

P10.20: Consider a 3-port network that is matched at all ports (S11 = S22 = S33 = 0).
Show that it is impossible to construct a reciprocal network that is lossless for this case.

For a 3-port scattering matrix matched at all ports and reciprocal we have
 0 S21 S31 
 S    S21 0 S32  .
 S31 S32 0 
If it is also lossless, then  S   S    U  , which gives the following equations:
t *

 S21  2   S31  2  1
 S21  2   S32  2  1
 S31  2   S32  2  1
and
*
S31S32 0
*
S21S32 0
*
S21S31  0.
 10-19

If we choose any pair of S31, S21 and S32 to be zero to satisfy the bottom set of equations,
then one of the top set of equations will be zero. Therefore, the network cannot be
lossless, reciprocal and matched at all ports.

P10.21: A circulator referenced to a 50  impedance is characterized by

 0.50 0.75e j 60 
 
0.050e j 60
 
 S    0.75e j 60 0.050e j 60  .
 
0.50
 j 60  
 0.050e 0.75e j 60 0.50 

Is this network (a) reciprocal? (b) Lossless? Calculate (c) insertion loss, (d) isolation and
(e) VSWR.

(a) not reciprocal, since (for instance)

S12  S21.
(b) not lossless, since summing the
squared column 1 elements gives a value
less than one.
(c)
IL  20log S21  20log 0.75  2.5dB
(d) I  20log S31  20log 0.05  26dB
1   1  S11 1  0.5
(e) VSWR    3
1   1  S11 1  0.5
Fig. P10.22
P10.22: Calculate the insertion loss and
the VSWR for the previous problem if the isolated port is terminated in a short circuit.

The insertion loss will be

V
IL  20log 2 ,
V1
and
1 
VSWR  ,
1 
V1
where   .
V1
For a 3 port scattering matrix we have
V3  S31V1  S32V2  S33V3 ,
and since port 2 is terminated in a matched load we therefore have
(1)V3  S31V1  S33V3 .
 10-20

 
With port 3 terminated in a short circuit, V3  V3 , and using (1) we find
  S31V1
(2)V 
3 .
1  S33
Now, for a 3 port scattering matrix matched at port 2 we have
(3)V1  S11V1  S12V2  S13V3  S11V1  S13V3
Inserting (2) into (3), we find
V1
 S11 
S13 S31
  0.5 
 


0.75e j 60 0.05e j 60



 0.513e  j 2.2 ,



V1 1  S33 1  0.5
1   1  0.513
so VSWR    3.1
1   1  0.513
Finally, for a 3 port scattering matrix matched at port 2 we also have
(4)V2  S21V1  S22V2  S23V3  S21V1  S23V3 ,
Inserting (2) into (4), we find

0.05e j 60  0.5



V2
V1
 S21 
S31S33
1  S33

 0.75e j 60
 1  0.5
 0.733e  j 60 ,


V2
and IL  20log  20log 0.733  2.7dB
V1

P10.23: The following information is supplied for a commercial L-Band circulator: ILmax
= 0.60 dB, Imin = 18 dB, and VSWRmax = 1.35. Calculate the worst-case magnitudes for
the scattering matrix. Assume a symmetrical circulator.

ILmax  20log S21  0.6dB , S 21  100.6 20  0.933

I min  20log S31  18dB, S31  10 18 20  0.126
1  S11 VSWR  1 1.35  1
VSWRmax  , or S11    0.15
1  S11 VSWR  1 1.35  1
So, for the symmetrical network we have the following magnitudes for the S-matrix:
0.15 0.13 0.93
 S    0.93 0.15 0.13
 
 0.13 0.93 0.15

P10.24: Verify the scattering matrix (equation (10.27)) for the resistive power divider of
Figure 10.16.

The impedance seen looking into, say, port 1 is Z o/3 plus the series combination of a pair
of 4Zo/3 impedances. So, with a total load impedance of Zo, there is no reflection.
 10-21

S21 is found using a pair of resistive divider circuits:

 4 4  
 Zo Zo 
3 3  Z  23 1 1
S21      o    1 
 4 Z 4 Z  1 Z  4 Z  1 43 2
 3 o 3 o 3 o  3 o 
 
Here, the transmission coefficient  is simply 1+ = 1. The first ratio in parenthesis
determines how much of the voltage at port 1 is dropped across the pair of parallel 4Z o/3
impedances, and the second ratio determines how much of this voltage drops across the
port 2 termination.
So,
0 1 1 
 2 2
 S    1 2 0 1 2  .
1 1 0
 2 2 

P10.25: Suppose 10.0 mW of microwave

power is fed into port 1 of the resistive
divider shown in Figure 10.16. With ports
2 and 3 terminated in matched loads,
determine how much power is transmitted
to each port and how much is dissipated in Fig. P10.25
the divider.


P2  P3  10mW S21
2

2 Pdiss  10mW  2(2.5mW )  5mW
1
 10mW    2.5mW
2

P10.26: Repeat problem P10.25 for a Wilkinson power divider.

 10-22

P2  P3  10mW S21  2

2
 1 
 10mW    5mW
 2
Pdiss  10mW  2(5mW )  0mW

P10.27: A 4-port “20-dB Coupler” is specified as having 20 dB coupling, 50 dB isolation,

and 0.25 dB of insertion loss. If 100 mW of power is input, calculate the power out of
the other three ports. Assume all ports are terminated in matched loads.

P 
C  20dB  10log  3  , so P3   P1   1020 10   1mW
 P1 
I  50dB  10log  S41  , so S41  10 50 10  0.00316
2
P4  S41 P  1W
IL  0.25dB  10 log  S 21  , so S 21  10 0.25 10  0.9716
2
P2  S21 P  94.4mW

P10.28: Suppose the coupling for an ideal symmetrical 4-port coupler is 3 dB. Find the
scattering matrix and determine the insertion loss.

In the ideal coupler, all the power entering port 1 is split and exits ports 2 and 3 with none
reflected and none to the isolated port, and none lost to dissipation.
C  3dB  20log S31 , or S31  10 3 20  0.707,
S21  S31  0.707 and IL  20log S21  3dB.
The scattering matrix is:
 0 .707 .707 0 
.707 0 0 .707 
 S   
.707 0 0 .707 
 
 0 .707 .707 0 

P10.29: (JustAsk): Suppose to port 1 of an ideal ring hybrid coupler we apply the
appropriate frequency voltage
V1  1.0e j 0 V .


If port 2 is terminated in a short circuit, determine the voltage exiting ports 1, 3 and 4.

For the ideal ring hybrid, we have

 10-23

0 1 1 0
 0 1
j 1 0
 S   .
2 1 0 0 1
 
 0 1 1 0
 
Assuming ports 3 and 4 are terminated in matched loads ( V3  V4  0 ), then we have
V1  S12V2
V2  S21V1
V3  S31V1
V4  S42V2
Now, at port 2 terminated in a short circuit, V2  V2 so
V1  S12V2   S12V2   S12 S21V1

V1    
 2  2 

  j    j  j 0
 1e 
 0.5e j 0 V


 2

  j  j 0
V3  S31V1    1e 
 0.707e  j 90 V


V4  S 42V2   S42V2   S 42 S21V1

 


 2  2 
 1e 
  j   j  j 0
V4    
 0.5e j 0 V


P10.30: Consider an ideal ring hybrid coupler, with all ports terminated in matched loads.
A signal 5.0ej30V is injected into port 2, and 3.0ej30V is injected into port 3. Determine
the signals exiting ports 1 and 4.

For the ideal ring hybrid, we have

0 1 1 0 
 
 j 1 0 0 1
 
S  .
2 1 0 0 1 
 
 0 1 1 0 

With all ports terminated in matched loads, all the incident voltages ( Vk ) will be zero,
except, of course, for the injected signals into ports 2 and 3. So we have
V1  S12V2  S13V3
V4  S42V2  S43V3 .
j
 0.707e  j 90 , we have

With
2
 10-24

V1  S12V2  S13V3


 0.707e  j 90


  5e    0.707e   3e 
j 30  j 90 j 30

 3.535e  j 60  2.121e  j 60  5.66e  j 60 V

  

and
V4  S42V2  S43V3


 0.707e  j 90


  5e    0.707e   3e 
j 30  j 90 j 30

 1.414e  j 60  1.414e j120 V

 

P10.31: Suppose to port 1 of a quadrature hybrid coupler we apply the appropriate

frequency voltage
V1  1.0e j 0 V .


If port 2 is terminated in a short circuit, determine the voltage exiting ports 1, 3 and 4.

The scattering matrix for the quadrature hybrid coupler is

0 j 1 0
 
1  j 0 0 1 
 S   .
2 1 0 0 j
 
0 1 j 0
With ports 3 and 4 terminated in matched loads  V3  V4  0  and port 2 terminated in a
 

short  Vv  V2  ,we then have

 

V1   S12 S21V1

V2  S21V1
V3  S31V1
V4   S 42 S21V1
Evaluating,
  j   j  j 0
V1    
 2  2 
 1e 
 0.5e j 0 V



 1  j 0
V3  S31V1  
 2
 1e 
 0.707e j 0 V



 1   j  j 0
V4    
 2  2 
 1e 
 0.5e  j 90 V


P10.32: Given a 50.0 mil thick Teflon substrate, design a quadrature hybrid coupler for
2.50 GHz operation.
 10-25

Fig. P10.32

For h = 50 mils and r = 2.1 for Teflon, and assuming a copper conductor where we may
neglect the conductor thickness, we must design quarter wave microstrip at 50  and at
50/ 2 . We use ML0605 to find:

Zo = 50  Zo = 35.4 
w=159 mils w=258 mils
 = 0.0893 m = 3520 mils  = 3470 mils
 /4 = 879 mils  /4 = 867 mils

The resulting design is shown in Figure P10.32.

P10.33: Suppose you join a pair of quadrature hybrid couplers (port 2 of coupler 1
attached to port 1 of coupler 2, port 3 of coupler 1 attached to port 2 of coupler 2). The
resulting network will have 4 ports: ports 1 and 3 from the first coupler, and ports 2 and 4
from the second coupler. Determine the overall scattering matrix.

Each individual coupler in Figure P10.33 has the matrix:

0 j 1 0
 
1  j 0 0 1 
 S   .
2 1 0 0 j
 
0 1 j 0
 10-26

We can expand the matrix to yield the following equations, where we’ve also indicated
       
the equivalent terms VA 2  VB1 , VA2  VB1 , VA3  VB 4 , and VA3  VB 4 :
For the A matrix:
VA1  S12VA2  S13VA3  S21VB1  S13VB4
VA2  S21VA1  S24VA4
VA3  S31VA1  S34VA4
VA4  S42VA2  S43VA3  S 42VB1  S43VB4
For the B matrix:
VB1  S12VB2  S13VB3
VB2  S21VB1  S24VB4  S 21VA2  S24VA3
VB3  S31VB1  S34VB4  S31VA2  S34VA3
VB4  S42VB2  S43VB3
   
Note that for the overall matrix, we are only interested in VA1 , VA 4 , VB 2 , and VB 3 . So we
insert reflected values into those equations. For example,
VA1  S21VB1  S13VB4  S21  S 21VB2  S13VB3   S13  S42VB2  S43VB3  ,
which expanding and rearranging becomes
VA1   S 21S 21  S13 S42  VB2   S21 S13  S13 S43  VB3
Inserting the scattering parameter values, we find
   j    j   1  1       j  1   1   j   
VA1        VB 2      

  VB 3   jVB 3
  2   2   2  2     2  2   2  2  
Likewise, we find
VA4  jVB2
VB2  jVA4
VB3   jVA1.
The overall scattering matrix is then:

Fig. P10.33
 10-27

4. Filters
P10.34: Derive an insertion loss expression for a high-pass filter realized using a shunt
inductor inserted in a Zo = Ro system. What inductance value is needed for a 1.0 GHz
cutoff frequency if Ro = 50 ?

1
With no filter present, we have vL  vs . With the filter in place,
2
Ro j L j LRo
vLf  vs , where Ro j L  .
Ro j L  Ro Ro  j L
Expanding and rearranging:
1
vLf  .
jR
2 o
L
vL 1  jR  jR
Now we have   2  o   1 o ,
vLf 2  L  2 L

vL  Ro 
2
  Ro 2 
 1   , so IL  10 log 1   2 L   .
vLf  2 L     
Now, for a 1 GHz cutoff frequency,
Ro R 50
 1, or L  o   3.98nH .
2  2 f c  L 4 f c 4  1x109 
So L = 4.0 nH.

P10.35: Derive an insertion loss expression for a high-pass filter realized using a series
capacitor inserted in a Zo = Ro system. What capacitance value is needed for a 1.0 GHz
cutoff frequency if Ro = 50 ?

1
With no filter present, we have vL  vs . With the filter in place,
2
Ro
vLf  vs .
 j 
2 Ro   
 C 
 j 
 2
vL 1  C  j
  1 ,
vLf 2 1 2 Ro C

 
2   1 
2

vL 1 IL  10 log  1  .
 1 
  4 fRoC 

 , and

vLf  4 fRoC   
For a 1 GHz cutoff frequency,
 10-28

(a)

(b)
Fig. P10.36 a&b

1 1 1
 1, so C    1.6 pF .
4 f c RoC 4 f c Ro 4  1x109   50 

P10.36: Design a 5th order low-pass filter with f c = 2.0 GHz for a 50  system where we
will allow only 0.1 dB of ripple. Use the Figure 10.28(b) circuit configuration, and
compare your insertion loss plot with that of Figure 10.31.

For N = 5 and 0.1 dB ripple, we read the appropriate values from Table 10.3 as indicated
in Figure P10.36(a).

Impedance Transformation:
L1 '  g1 Ro   1.1468   50   57.3H
L5 '  L1 '
g 1.3712
C2 '  2   27.4mF
Ro 50
C4 '  C 2 '
L3 '  g3 Ro   1.975   50   98.75 H

Frequency Transformation:
L' 57.3
L1 ''  1   4.6nH
c 2  2 x109 
L5 ''  L1 ''
 10-29

C2 ' 27.4mF
C2 ''    2.18 pF , C4 ''  C2 ''
c 2  2 x109 
L3 ' 98.75
L3 ''    7.86nH .
c 2  2 x109 
These values are indicated on the filter design shown in Figure P10.36(b).

For generating the insertion loss plot, we make use of the equivalent circuits shown in
Figure P10.36(c)-(e). We have
j  R 
Z1   R  j L1  , vLf  v1  
C1  R  j L1 
j  Z1 
Z2   Z1  j L2  , v1  v2  
C1  Z1  j L2 
 Z2 
v2    vs
 Z 2  R  j L1 

(c) (d)

(e)
Fig. P10.36 c,d & e

So finally we have,
vLf 1  R  Z1  Z2 
 ,    ,
vs 2  R  j L1  Z1  j L2  Z 2  R  j L1 
and with
vL 1 v vLf v  v 
 , we divide the two ratios L  L to arrive at IL  20 log  L  .
vs 2 vs vs vLf  vLf 
 
 10-30

These equations are used in MLP1036 to generate the insertion loss plot of Figure
P10.36(f).

% MLP1036
% Design N=5 LPF, 0.1 dB ripple
% Compare with N=3 LPF of Example 10.14.
%
clear;clc;clf

R=50;
f=0.1:0.1:10;
w=2*pi*f*1e9;

% N=3
L31=4.1e-9;
C31=1.83e-12;
ZL31=i*w*L31;
ZC31=-i./(w*C31);
Z131=R+ZL31;
Z132=parallel(Z131,ZC31);
Z133=Z132+R+ZL31;
sqrtPLR3=Z131.*Z133./
(2*R*Z132);

Fig. P10.36(f)

IL3=10*LOG10(abs(sqrtPLR3));

% N=5
L51=4.6e-9;
C52=2.2e-12;
L53=7.9e-9;
XL51=j*w*L51;
 10-31

XC52=-j./(w*C52);
XL53=j*w*L53;
Z51=parallel(XC52,R+XL51);
Z52=parallel(XC52,Z51+XL53);
A=(R./(R+XL51)).*(Z51./(Z51+XL53)).*(Z52./(R+Z52+XL51));
IL5=20*log10(abs(1./(2*A)));

semilogx(f,IL3,'--b',f,IL5,'.-k')
Legend('N=3','N=5')
grid on
xlabel('frequency(GHz)')
ylabel('IL(dB)')
axis([0.1 10 0 40])

P10.37: Design a 3rd order low-pass filter with f c = 1.0 GHz for a 50  system starting
with the Figure 10.28(a) circuit configuration. Determine component values for each
amount of ripple (0.1, 1 and 3 dB) and compare the three insertion loss responses.

Referring to Figure 10.28(a), we have the following equations based on values of g 1, g2

and g3 read from the appropriate portion of Table 10.3, after applying both impedance
and frequency transformations:
g g R
C1 ''  1  C3 '', L2 ''  2 .
c R c
For the MATLAB routine used to compare the insertion loss responses, we let
j
XC  and XL  j L.
C
Considering v1 is the voltage across C1, we define Z1 as the impedance seen looking to the
right of C1 and Z2 as the parallel impedance of Z1 and C1. Then,
 R XC 
vLf  Z1v1 , where Z1    ,
 R XC  XL 
 Z2 
and v1    vs , where Z 2  Z1 XC.
 Z2  R 
Then, with vL = vs/2, we arrive at
vL R  Z 2
 .
vLf 2 Z1Z 2

% MLP1037
%
% Compare N=3 LPF with different ripples
%
 10-32

R=50;
fc=1e9;
wc=2*pi*fc;

% 0.1 dB ripple
g1=1.0315;
g2=1.1474;

C=g1/(wc*R);
L=g2*R/wc;

f=0.1:0.01:10;
w=2*pi*f*1e9;
XC=-j./(w.*C);
XL=j.*w.*L;
Z1=parallel(R,XC)./(parallel(R,XC)+XL);
Z2=parallel(Z1,XC);
A=abs((R+Z2)./(2.*Z1.*Z2));
ILa=20*log10(A);

% 1 dB ripple
g1=2.0236;
g2=.9941;

C=g1/(wc*R);
L=g2*R/wc;

f=0.1:0.01:10;
w=2*pi*f*1e9;
XC=-j./(w.*C);
XL=j.*w.*L;
Z1=parallel(R,XC)./
(parallel(R,XC)
+XL);
Z2=parallel(Z1,XC);
A=abs((R+Z2)./
(2.*Z1.*Z2));
ILb=20*log10(A);

% 3 dB ripple
g1=3.3487; Fig. P10.37
g2=.7117;
 10-33

C=g1/(wc*R);
L=g2*R/wc;

f=0.1:0.01:10;
w=2*pi*f*1e9;
XC=-j./(w.*C);
XL=j.*w.*L;
Z1=parallel(R,XC)./(parallel(R,XC)+XL);
Z2=parallel(Z1,XC);
A=abs((R+Z2)./(2.*Z1.*Z2));
ILc=20*log10(A);

semilogx(f,ILa,'--',f,ILb,'-',f,ILc)
legend('0.1 dB','1 dB','3 dB')
grid on
xlabel('frequency (GHz)')
ylabel('IL (dB)')

P10.38: (JustAsk): Starting with the Figure 10.28(b) circuit configuration, design a 3rd
order high-pass filter with fc = 2.4 GHz for a 50  system where we will allow only 1 dB
of ripple. Plot the insertion loss.

For the 3rd order, 1 dB ripple, high-pass filter, we first select the component values from
Table 10.3:
g1  g3  2.0236, and g 2  0.9941,
as shown in Figure P10.38(a).

Impedance Transformation:
g
L1 '  g1 Ro  101.2, C2 '  2  0.0199.
Ro

Frequency Transformation:
(using Table 10.4)
1
C1 '' 
c L1 '
1
  0.655 pF
2  2.4 x109   101.2 
 10-34

1
L2 '' 
c C2 '
1
  3.33nH .
2  2.4 x109   0.0199 
See Figure P10.38(b).

To develop an expression for the insertion, (a)

we consider v2 to be the voltage across the
inductor L2’’, and also we let
Z1  XL  R  XC  , where XL = jwL and
j
XC  .
C

Then,
(b)
Fig. P10.38a&b

R Z1 v R Z1
vLf  v1 , v1  vs , so Lf  .
R  XC Z1  R  XC vs R  XC Z1  R  XC
Now, with vL=vs/2, we have
vL  R  XC   Z1  R  XC 
 .
vLf 2 RZ1
We then generate the insertion loss plot via ML1038:
% MLP1038
% Design N=3 high pass filter.%
fc=2.4; %corner frequency in GHz
wc=2*pi*fc*1e9;
g1=2.0236;
g2=.9941;
R=50;

L=g1*R;
C=g2/R;

C1=1/(wc*L);
L2=1/(wc*C);

f=0.1:0.1:10;
w=2*pi*f*1e9;
XC1=-j./(w.*C1); Fig. P10.38c
XL2=j.*w.*L2;
 10-35

Z1=parallel(XL2,R+XC1);
A=((R+XC1).*(R+XC1+Z1))./(2.*R.*Z1);
IL=20*log10(abs(A));

semilogx(f,IL)
grid on
xlabel('frequency (GHz)')
ylabel('IL (dB)')
title('2.4 GHz High Pass Filter')

P10.39: Starting with the Figure 10.28(a) circuit configuration, design a 5th order high-
pass filter with fc = 1.0 GHz for a 50  system where we will allow 3 dB of ripple. Plot
the insertion loss.

Assuming 3dB of ripple, from Table 10.3 we have:

g1  g5  3.4817, g 2  g 4  0.7618, and g3  4.5381
(see Figure P10.39(a))

Now we perform the impedance transformation:

g g
C1 '  1  0.06963, L2 '  g 2 R  38.09, and C3 '  3  0.09076 .
R R
The frequency transformation is performed using Table 10.4:
1 1 1
L1 ''   2.29nH , C2 ''   4.18 pF , and L3 ''   1.75nH .
c C1 ' c L2 ' c C3 '
The final circuit is shown in Figure P10.39(b).

The MATLAB insertion loss plot is based on analysis of the Fig P10.39(b) circuit.
Assuming the reactances
j
XL1  j L1 '', XC2  , and XL3  j L3 '',
C2 ''
the impedance to the right of L3 is:

(a) (b)
Fig. P10.39a&b
 10-36

We’ll also define Z2 to be the impedance of Z1 in parallel with XL3, and Z3 to be:
Z 3   Z 2  XC2  XL1.
Now we can find the voltage ratio relations:
 R XL1   Z2   Z3 
vLF    v1 , v1    v2 , and v2    vs .
 Z 1  Z
 2  XL2  Z
 3  R 
Combining, and realizing vL = vs/2, we arrive at:
vL Z1  Z 2  XC2   Z 3  R 
 .
vLf 2 Z 2 Z 3  R XL1 
The insertion loss is then:
 v 
IL  20 log  L  .
 vLf 
 
This is plotted by ML1039 in Figure P10.39(c).

% MLP1039
%
% Design N=5 high pass filter, 3 dB ripple.
%
fc=1; %corner frequency in GHz
wc=2*pi*fc*1e9;
g1=3.4817;
g2=.7618;
g3=4.5381;
R=50;

Ca=g1/R;
L=g2*R;
Cb=g3/R;

L1=1/(wc*Ca);
C2=1/(wc*L);
L3=1/(wc*Cb);

f=0.1:0.01:10;
w=2*pi*f*1e9;

XL1=j.*w.*L1;
XC2=-j./(w.*C2);
XL3=j.*w.*L3; Fig. P10.39(c)

Z1=parallel(R,XL1)+XC2;
Z2=parallel(Z1,XL3);
 10-37

Z3=parallel(Z2+XC2,XL1);
A=(Z1.*(Z2+XC2).*(Z3+R))./(2.*parallel(R,XL1).*Z2.*Z3);
IL=20*log10(abs(A));

semilogx(f,IL)
grid on
xlabel('frequency (GHz)')
ylabel('IL (dB)')
title('1 GHz High Pass Filter: N=5')

P10.40: Starting with the Figure 10.28(b) circuit configuration, design a 3rd order
Chebyshev bandpass filter for a 50  system. The passband is to be from 900 MHz to
1100 MHz with only 1 dB of ripple allowed.

Assuming 1dB of ripple, from Table 10.3 we have:

g1  g3  2.0236, and g 2  0.9941.
(see Figure P10.28(b))
Now we perform the impedance transformation:
g
L1 '  L3 '  g1 R  101.2, and C2 '  2  0.01988
R
The frequency transformation is performed using Table 10.4, where L1 ' is transformed
into a series circuit containing L1 '' and C1 '' . C2 ' is transformed into a parallel circuit
containing L2 '' and C2 '' . This is shown in Figure P10.40(a). To perform the
transformation, we calculate BW and wo:
BW  2  200 MHz 
o  2  900MHz   1100 MHz   2  995MHz 
We then have:

(a) (b)
Fig. P10.40a&b
 10-38

To plot the insertion loss using ML1040, we use the impedances shown in Figure
P10.40(b), given as
j j
Z1  j L1 '' , Z 2  j L2 ''
C1 '' C2 ''
We also define Z3 as the impedance seen by the voltage v1:
Z 3  Z 2  R  Z1  .
Now we can find the voltage ratio relations:
 R   Z3 
vLF    v1 , and v1    vs .
 R  Z1   Z 3  R  Z1 
Combining, and realizing vL = vs/2, we arrive at:
vL  R  Z1   R  Z1  Z 3 
 .
vLf 2 RZ 3
The insertion loss is then:
 v 
IL  20 log  L  .
 vLf 
 
This is plotted by ML1040 in Figure P10.40(c).

% MLP1040
% Design N=3 band pass filter, 1 dB ripple.
%
clc
clear

R=50;
flo=900e6;fhi=1100e6;
BW=2*pi*(fhi-flo);
wo=2*pi*sqrt(flo*fhi);

g1=2.0236;g2=.9941;
 10-39

Fig. P10.40(c)

L=g1*R;C=g2/R;

L1=L/BW;
C1=BW/((wo^2)*L);
L2=BW/((wo^2)*C);
C2=C/BW;

f=0.1:0.01:10; %f in GHz
w=2*pi.*f*1e9;
Z1=j.*(w.*L1-1./(w.*C1));
Z2=parallel(j.*w.*L2,-j./(w.*C2));
Z3=parallel(Z2,R+Z1);

A=((R+Z1).*(R+Z1+Z3))./(2.*R.*Z3);
IL=20*log10(abs(A));

semilogx(f,IL)
grid on
xlabel('frequency (GHz)')
ylabel('IL (dB)')
title('N=3 BPF: 900-1100 MHz')

P10.41: Starting with the Figure 10.28(b) circuit configuration, design a 3rd order
Chebyshev bandpass filter for a 50  system. The passband is to be from 900 MHz to
1100 MHz with 3 dB of ripple allowed. Plot the insertion loss and calculate the shape
factor.
 10-40

(a) (b)
Fig. P10.41a&b

Assuming 3 dB of ripple, from Table 10.3 we have:

g1  g3  3.3487, and g 2  0.7117.
(see Figure P10.28(b))
Now we perform the impedance transformation:
g
L1 '  L3 '  g1 R  167.4, and C2 '  2  0.01423
R
The frequency transformation is performed using Table 10.4, where L1 ' is transformed
into a series circuit containing L1 '' and C1 '' . C2 ' is transformed into a parallel circuit
containing L2 '' and C2 '' . This is shown in Figure P10.41(a). To perform the
transformation, we calculate BW and wo:
BW  2  200 MHz 
o  2  900MHz   1100 MHz   2  995MHz 
We then have:
L' 167.4
L1 ''  1   133.2nH
BW 2  200 x106 
BW 2  200 MHz 
C1 ''    0.1921 pF
o L1 '  2   995MHz  2  167.4 
2 2

BW 2  200MHz 
L2 ''    2.259nH
o C2 '  2   995MHz  2  0.01423
2 2

C2 ' 0.01423
C2 ''    11.32 pF
BW 2  200 x106 

To plot the insertion loss using ML1041, we use the impedances shown in Figure
P10.41(b), given as
j j
Z1  j L1 '' , Z 2  j L2 ''
C1 '' C2 ''
We also define Z3 as the impedance seen by the voltage v1:
 10-41

Z 3  Z 2  R  Z1  .
Now we can find the voltage ratio relations:
 R   Z3 
vLF    v1 , and v1    vs .
 R  Z 1  Z
 3  R  Z1 

Combining, and realizing vL = vs/2, we arrive at:

vL  R  Z1   R  Z1  Z 3 
 .
vLf 2 RZ 3
The insertion loss is then:
 v 
IL  20 log  L  .
 vLf 
 
This is plotted by ML1041 in Figure P10.40(c).

% MLP1041
%
% Design N=3 BPF, 3 dB ripple.
%
clc
clear

R=50;
flo=900e6;fhi=1100e6;
BW=2*pi*(fhi-flo);
wo=2*pi*sqrt(flo*fhi);

g1=3.3487;g2=.9941;

L=g1*R;C=g2/R;

L1=L/BW;
C1=BW/((wo^2)*L); Fig. P10.41 (c)
L2=BW/((wo^2)*C);
C2=C/BW;

f=0.5e9:0.005e9:2e9; %f in GHz
w=2*pi.*f;
Z1=j.*(w.*L1-1./(w.*C1));
Z2=parallel(j.*w.*L2,-j./(w.*C2));
Z3=parallel(Z2,R+Z1);

A=((R+Z1).*(R+Z1+Z3))./(2.*R.*Z3);
IL=20*log10(abs(A));
 10-42

plot(f,IL)
axis([.5e9 1.8e9 0 70])
grid on
xlabel('frequency (GHz)')
ylabel('IL (dB)')
title('N=3 BPF: 900-1100 MHz')

Figure 10.41(c) also shows lines marking the frequencies at 60 dB insertion loss for
calculation of shape factor. We have
BW60 dB  1.7  0.57  GHz
SF    5.6
BW3dB 0.2GHz

P10.42: Starting with the Figure 10.28(b) circuit configuration, design a 3rd order
Chebyshev bandpass filter for a 50  system. The passband is to be from 2.2 GHz to 2.6
GHz with only 1 dB of ripple allowed.

Assuming 1dB of ripple, from Table 10.3 we have:

g1  g3  2.0236, and g 2  0.9941.
(see Figure P10.28(b))
Now we perform the impedance transformation:
g
L1 '  L3 '  g1 R  101.2, and C2 '  2  0.01988
R
The frequency transformation is performed using Table 10.4, where L1 ' is transformed
into a series circuit containing L1 '' and C1 '' . C2 ' is transformed into a parallel circuit
containing L2 '' and C2 '' . This is shown in Figure P10.42(a). To perform the
transformation, we calculate BW and wo:
BW  2  400 MHz 
o  2  2200 MHz   2600 MHz   2  2392MHz 
We then have:

(a) (b)
Fig. P10.42a&b
 10-43

To plot the insertion loss using ML1043, we use the impedances shown in Figure
P10.42(b), given as
j j
Z1  j L1 '' , Z 2  j L2 ''
C1 '' C2 ''
We also define Z3 as the impedance seen by the voltage v1:
Z 3  Z 2  R  Z1  .
Now we can find the voltage ratio relations:
 R   Z3 
vLF    v1 , and v1    vs .
 R  Z1   Z 3  R  Z1 
Combining, and realizing vL = vs/2, we arrive at:
vL  R  Z1   R  Z1  Z 3 
 .
vLf 2 RZ 3
The insertion loss is then:
 v 
IL  20 log  L  .
 vLf 
 
This is plotted by ML1042 in Figure P10.42(c).

% MLP1042%
% Design N=3 BPF, 1 dB ripple.
%
clc
clear

R=50;
flo=2.2e9;fhi=2.6e9;
BW=2*pi*(fhi-flo);
wo=2*pi*sqrt(flo*fhi);
 10-44

Fig. P10.42(c)

g1=2.0236;g2=.9941;

L=g1*R;C=g2/R;

L1=L/BW;
C1=BW/((wo^2)*L);
L2=BW/((wo^2)*C);
C2=C/BW;

f=1:0.01:5; %f in GHz
w=2*pi.*f*1e9;
Z1=j.*(w.*L1-1./(w.*C1));
Z2=parallel(j.*w.*L2,-j./(w.*C2));
Z3=parallel(Z2,R+Z1);

A=((R+Z1).*(R+Z1+Z3))./(2.*R.*Z3);
IL=20*log10(abs(A));

plot(f,IL)
grid on
xlabel('frequency (GHz)')
ylabel('IL (dB)')
title('N=3 BPF: 900-1100 MHz')
 10-45

P10.43: Starting with the Figure 10.28(b) circuit configuration, design a 3rd order
Chebyshev bandstop filter for a 50  system. The stopband is to be from 900 MHz to
1100 MHz with 3 dB of ripple allowed. Plot the insertion loss.

(a) (b)
Fig. P10.43a&b

Assuming 3 dB of ripple, from Table 10.3 we have:

g1  g3  3.3487, and g 2  0.7117.
(see Figure P10.28(b))
Now we perform the impedance transformation:
g
L1 '  L3 '  g1 R  167.4, and C2 '  2  0.01423
R
The frequency transformation is performed using Table 10.4, where L1 ' is transformed
into a parallel circuit containing L1 '' and C1 '' . C2 ' is transformed into a series circuit
containing L2 '' and C2 '' . This is shown in Figure P10.43(a). To perform the
transformation, we calculate BW and wo:
BW  2  200 MHz 
o  2  900MHz   1100 MHz   2  995MHz 
We then have:
1 1
C1 ''    4.75 pF
BW L1 ' 2  200 x106   167.4 
BW L1 ' 2  200 MHz   167.4 
L1 ''    5.39nH
o2  2   995MHz 
2 2

1 1
L2 ''    55.9nH
BW C2 ' 2  200 MHz   0.01423
BW C2 ' 2  200 x10   0.01423
6

C2 ''    0.458 pF
o2  2   995MHz 
2 2
 10-46

To plot the insertion loss using ML1043, we use the impedances shown in Figure
P10.43(b), given as
j j
Z 2  j L2 '' , Z1  j L1 ''
C2 '' C1 ''
We also define Z3 as the impedance seen by the voltage v1:
Z 3  Z 2  R  Z1  .
Now we can find the voltage ratio relations:
 R   Z3 
vLF    v1 , and v1    vs .
 R  Z1   Z 3  R  Z1 
Combining, and realizing vL = vs/2, we arrive at:
vL  R  Z1   R  Z1  Z 3 
 .
vLf 2 RZ 3
The insertion loss is then:
 v 
IL  20 log  L  .
 vLf 
 
This is plotted by ML1043 in Figure P10.43(c).

% MLP1043
%
% Design N=3 band-stop filter, 3 dB ripple.
%
clc;clear

R=50;
flo=900e6;fhi=1100e6;
BW=2*pi*(fhi-flo);
wo=2*pi*sqrt(flo*fhi);

g1=3.3487;g2=.7117;

L=g1*R;C=g2/R;

C1=1/(BW*L);
L1=BW*L/((wo^2));
C2=BW*C/((wo^2)); Fig. P10.43c
L2=1/(BW*C);

f=.8:0.001:1.2; %f in GHz
w=2*pi.*f*1e9;
Z2=j.*(w.*L2-1./(w.*C2));
Z1=parallel(j.*w.*L1,-j./(w.*C1));
 10-47

Z3=parallel(Z2,R+Z1);

A=((R+Z1).*(R+Z1+Z3))./(2.*R.*Z3);
IL=20*log10(abs(A));

plot(f,IL)
grid on
xlabel('frequency (GHz)')
ylabel('IL (dB)')
title('N=3 Band-Stop Filter: 900-1100 MHz')
axis([0.8 1.2 0 50])

P10.44: Starting with the Figure 10.28(a) circuit configuration, design a 5th order
Chebyshev bandstop filter for a 50  system. The stopband is to be from 2.3 GHz to 2.5
GHz with 1 dB of ripple allowed. Plot the insertion loss.

From Table 10.3 for 1 dB of ripple and N

= 5 we find:
g1  g5  2.1349
g 2  g 4  1.0911
g3  3.0009

We first perform the impedance (a)

transformation:
g
C1 '  1  0.0427,
Ro
L2 '  g 2 Ro  54.56,
g3
C3 '   0.0602
Ro

Figure P10.44(b) shows the filter portion

of the circuit after transforming based on (b)
Table 10.4. Fig. P10.44a&b
To perform the frequency transformation,
we first calculate:
BW  2  2.5GHz  2.3GHz   2  200 MHz 
o  2  2.3GHz   2.5GHz   2  2.398GHz 
Then,
1 1
L1 ''    18.6nH
BW C1 ' 2  200 x106   0.0427 
 10-48

BW C1 ' 2  200 MHz   0.0427 

C1 ''    0.236 pF
o2  2   2.398GHz 
2 2

1 1
C2 ''    14.6nH
BW L2 ' 2  200MHz   54.56 
BW L2 ' 2  200 x10   54.56 
6

L2 ''    0.302nH
o2  2   2.398GHz 
2 2

1 1
L3 ''    13.3nH
BW C3 ' 2  200MHz   0.06002 
BW C3 ' 2  200 x10   0.06002 
6

C3 ''    0.332 pF
o2  2   2.398GHz 
2 2

The voltage ratios needed to determine

insertion loss are found with the aid of
Fig P10.44(c). Here we have
j
Z1  j L1 ' ,
C1 '
j
Z 2  j L2 '
 C2 '
j
Z 3  j L3 '
C3 ' Fig. P10.44(c)
Some of these impedances are also
lumped together:
Z a  Z 3  Z 2  Z1 R  and Zb  Z1  Z 2  Z a  .
The voltage ratios are then:
 Z1 R   Za   Zb 
vLf    v1 , v1    v2 , v2    vs ,
 Z1 R  Z 2   Za  Z2   R  Zb 
so
vLf  Z1 R   Z a   Z b 
   .
vs  Z1 R  Z 2   Z a  Z 2   R  Z b 
Now, since vL=vs/2, then after manipulation we have
vL  Z1 R  Z 2   Z a  Z 2   R  Z b 

vLf 2 Z a Z b  Z1 R 
vL
and this is used to find IL  20 log using MLP1044:
vLf
 10-49

% MLP1044
% Design N=5 Band-Stop filter, 1 dB ripple.
%
clc;clear
R=50;
flo=2.3e9;fhi=2.5e9;
BW=2*pi*(fhi-flo);
wo=2*pi*sqrt(flo*fhi);

g1=2.1349;g2=1.0911;g3=3.0009;
Ca=g1/R;L=g2*R;Cb=g3/R;
L1=1/(BW*Ca);
C1=BW*Ca/((wo^2));
L2=BW*L/((wo^2));
C2=1/(BW*L);
L3=1/(BW*Cb);
C3=BW*Cb/((wo^2));

f=1.8:0.0001:3; %f in GHz
w=2*pi.*f*1e9;

Z1=j.*w.*L1-j./(w.*C1);
Z2=parallel(j.*w.*L2,-j./(w.*C2));
Z3=j.*w.*L3-j./(w.*C3);

Za=parallel(Z1,R);
Zb=parallel(Z3,Z2+Za);
Zc=parallel(Z1,Z2+Zb);

A=(Za+Z2).*(Zb+Z2).*(Zc+R);
B=2.*Za.*Zb.*Zc;
IL=20*log10(abs(A./B));

plot(f,IL)
grid on
xlabel('frequency (GHz)')
ylabel('IL (dB)')
title('N=5 Band-Stop Filter: 2.3 - 2.5 GHz')
axis([1.8 3 0 50])
 10-50

Fig. P10.44(d)

5. Amplifiers
P10.45: (JustAsk): The following S-Parameters were measured at 2.0 GHz in a 50 
system:
S11  0.68e j125 S12  0


S 21  3.6e j 40 S22  0.86e  j 74 .

 

(a) Determine the gain, in dB, without any matching networks. (b) Determine the
maximum gain, assuming optimized matching networks.

(a) With out matching networks, we have S = L = 0 as indicated in Figure P10.45. So

our transducer gain term is:
2 2
1  S 2 1 L 2
GT  2
S 21 2
 S21  12.96
1   IN  S 1  S 22  L
So GT (dB) = 10 log(12.96) = 11.1 dB.

(b) Now assuming optimized matching, the transducer gain term reduces to:

Fig. P10.45
 10-51

and GT (dB) = 10 log(92.6) = 19.7dB.

P10.46: For P10.45, (a) design open-ended shunt stub matching networks. (b) You are to
realize the matching networks in microstrip constructed on 25.0 mil thick Teflon.
Determine the required microstrip width, and provide a labeled sketch of your network
similar to Figure 10.42.

Referring to Figure P10.46(a) we find the open-ended shunt stub matching network to
achieve  L  S 22  0.86e j 74 . We move from the open-ended stub in the admittance chart

*

(point a’) a distance 0.206 to reach the point b’. The normalized admittance at this
point, added to the matched load admittance of y = 1+j0, puts us at point c’. From here,
we move a distance (0.397– 0.209 = 0.188) along the constant || circle to point d’.
Transforming this to the impedance chart, we reach our desired impedance point d.

The same procedure is used for the source matching network. Referring to Figure
P10.46(b) we find the open-ended shunt stub matching network to achieve
 S  S11*  0.68e  j125 . We move from the open-ended stub in the admittance chart (point


a’) a distance 0.173 to reach the point b’. The normalized admittance at this point,
added to the matched load admittance of y = 1+j0, puts us at point c’. From here, we
move a distance (0.5000.174 – 0.185 = 0.489) along the constant || circle to
point d’. Transforming this to the impedance chart, we reach our desired impedance
point d.

The final stub matching network is indicated in Figure P10.46(c).

To realize the stub matching network in microstrip, we employ ML0605 for microstrip
design:

Microstrip Design
width & thickness will be in the same units

enter the desired impedance: 50

enter the substrate thickness: 25
enter substrate rel permittivity: 2.1
w = 79.3268
eeff = 1.8015
up = 223363676.0028m/s
>>

The length of a guide wavelength is

 10-52

(a) (b)

(c)

(d)
Fig. P10.46

c 3 x108 m / s  1mil 
G    6   4400mils
f  eff 2 x10 1.8015  25.4 x10 m 
9

This is used to find the physical lengths required of the stubs, as shown in Figure
10.46(d).
 10-53

P10.47: For P10.45, design a matching network using lumped elements. In the sketch of
your solution, indicate line lengths in terms of wavelength.

We first design the lumped element

matching network to achieve
 L  S 22  0.86e j 74 .

*

For this type of problem we actually work

backwards, starting from point c and
working back to point a, to identify the
location of the points. Then we work the
matching network forwards to determine
the component values. From the matched
condition at point a, we add a shunt
inductance to point b’ (-j2.0). Then,
transforming to the impedance chart, we
add a series inductance to reach point c
(j1.3 – j0.4 = j0.9).
For the shunt inductance we have:
 jZ o
 j 2.0  , or
L
50
L  2nH
Fig P10.47(a) 2  2   2 x109 

For the series inductance,

j L
j 0.9  , or
Zo

L
 0.9   50   3.6nH .
2  2 x109 
Now we design the lumped element
matching network to achieve
 S  S11*  0.68e  j125 .


From the matched condition at point a, we

add a shunt capacitance to point b’ (j1.8).
Then, transforming to the impedance
chart, we add a series capacitance to reach
point c (-j0.49 + j0.42 = -j0.07).
For the shunt capacitance we have:
j1.8  jCZ o , or
1.8
C  2.9 pF
Fig. P10.47(b) 2  2 x109   50 
For the series capacitance,
 10-54

Fig. P10.47(c)

j
 j 0.07  , or
CZ o
1
C  23 pF .
 0.07  2  2 x109   50 
The overall circuit is given in Figure P10.47(c).

P10.48: The following S-Parameters were measured at 10 GHz in a 50  system:

S11  0.72e j 76 S12  0


S 21  4.4e j125 S22  0.58e  j 30 .

 

(a) Determine the gain, in dB, without any matching networks. (b) Determine the
maximum gain, assuming optimized matching networks.

(a) With out matching networks, we have S = L = 0. So our transducer gain term is:
2 2
1  S 2 1  L 2
GT  2 S 21 2  S21  19.36
1   IN  S 1  S22  L
So GT (dB) = 10 log(19.36) = 12.9 dB.

(b) Now assuming optimized matching, the transducer gain term reduces to:
1 2 1
GT  2
S21 2
 60.58
1  S21 1  S22
and GT (dB) = 10 log(60.59) = 17.8dB.

P10.49: For P10.48, design shorted shunt stub matching networks with the overall line
lengths minimized. In the sketch of your solution, indicate line lengths in terms of
wavelength.

Referring to Figure P10.49(a) we find the shorted shunt stub matching network to achieve
 L  S22  0.58e j 30 . We move from the open-ended stub in the admittance chart (point

*

a’) a distance 0.099 to reach the point b’. The normalized admittance at this point,
added to the matched load admittance of y = 1+j0, puts us at point c’. From here, we
 10-55

(a) (b)

(c)

Fig. P10.49

move a distance (0.458– 0.326 = 0.132) along the constant || circle to point d’.
Transforming this to the impedance chart, we reach our desired impedance point d.

The same procedure is used for the source matching network. Referring to Figure
P10.49(b) we find the shorted shunt stub matching network to achieve
 S  S11*  0.72e  j 76 . We move from the open-ended stub in the admittance chart (point


a’) a distance 0.072 to reach the point b’. The normalized admittance at this point,
added to the matched load admittance of y = 1+j0, puts us at point c’. From here, we
move a distance (0.500+ 0.106– 0.311 = 0.295) along the constant || circle to
point d’. Transforming this to the impedance chart, we reach our desired impedance
point d.

Figure P10.49(c) shows the completed circuit. Note that this pair of matching networks
results in the smallest overall sum of stub lengths.
 10-56

P10.50: For P10.48, design the matching networks using lumped elements.

We first design the lumped element matching network to achieve

 L  S22  0.58e j 30 .

*

For this type of problem we actually work backwards, starting from point c and working
back to point a, to identify the location of the points. Then we work the matching
network forwards to determine the component values. From the matched condition at
point a, we add a series inductance to point b (+j1.6). Then, transforming to the
admittance chart, we add a shunt capacitance to reach point c’ (-j.25 –- j.44 = j0.19).
For the series inductance we have:
 j L 50  1.6 
j1.6  , or L   1.27nH
Zo  2   10 x109 
For the shunt capacitance,
j 0.19  j CZ o , or C 
 0.19   .060 pF
2  10 x109   50 

(a) (b)

Fig. P10.50
 10-57

Now we design the lumped element matching network to achieve

 S  S11*  0.72e  j 76 .


From the matched condition at point a, we add a shunt capacitance to point b’ (+j1.2).
Then, transforming to the impedance chart, we add a series capacitance to reach point c (-
j1.2 -- j.48 = -j0.72).
For the shunt capacitance we have:
j1.25  j CZ o , or C 
 1.25  0.40 pF
 2   10 x109   50 
For the series capacitance,
j 1
 j 0.72  , or C   .44 pF
 CZ o 2  10 x10   50   0.72 
9

6. Receiver Design
P10.51: Determine the IF power, in watts, exiting a mixer that has a 6.0 dB conversion
loss if 0 dBm of RF power and of LO power enters the mixer.

P 
CL  10log  RF   6dB,
 PIF 
PRF
 100.6  3.981
PIF
PRF 1mW
PIF    0.25mW
3.981 3.981

P10.52: (JustAsk):
Referring to Example 10.21
and Figure 10.48, suppose
you require a 100 W
output power level and the
antenna receives -80 dBm.
If you have several of each
amplifier available, design
the receiver. You are also
allowed to insert a fixed
value attenuator.

One design is shown in

Figure P10.52.
Fig. P10.52