Analysis of Microwave Networks

Debapratim Ghosh

School of Electrical Sciences

Indian Institute of Technology Bhubaneswar

e-mail: [email protected]

Debapratim Ghosh School of Electrical Sciences, IIT Bhubaneswar

 Overview

I Microwave networks, like any electrical network, consist of interconnection of

various electrical elements.

I Each network has a number of ports for input(s) and/or output(s).

I In essence, microwave networks are no different from electrical networks, as the

usual circuit laws do apply to them too, under certain conditions!

I There exist network parameters that relate the voltage/currents at one port, with
those at other port(s).

I You are already aware of some of these parameters through your network theory
course; nevertheless, let us revisit them briefly.

Debapratim Ghosh School of Electrical Sciences, IIT Bhubaneswar

 Impedance and Admittance Parameters

Consider the two-port network as shown. The voltages and currents V1 , I1 , V2 , and I2
may originate from sources, or may be developed as a result of sources at other ports or
inside the network.
I1 I2

+ 2-port +
V1 network V2
- -

Impedance (Z ) parameters Admittance (Y ) parameters

         
V1 Z11 Z12 I1 I1 Y11 Y12 V1
= (1) = (4)
V2 Z21 Z22 I2 I2 Y21 Y22 V2

   
V1  V1  I1  I1 
Z11 = Z12 = (2) Y11 = Y12 = (5)
I1 I2 =0 I2 I1 =0 V1 V2 =0 V2 V1 =0
   
V2  V2  I2  I2 
Z21 = Z22 = (3) Y21 = Y22 = (6)
I1 I2 =0 I2 I1 =0 V1 V2 =0 V2 V1 =0

Debapratim Ghosh School of Electrical Sciences, IIT Bhubaneswar

 Some Important Points about Z and Y Parameters

I If one or more elements in the Z or Y matrix of a network have real values, that
means the network has resistive or conductive elements.

I This means that the network is lossy, as it has elements that dissipate power.

I If the Z or Y matrix of a network is symmetric, i.e. [Z ]T = [Z ], that means the

network is reciprocal.

I Reciprocity of a 2-port network is defined as: if a current I at Port 1 produces a

voltage V at Port 2, then the network is reciprocal if the same current I fed to Port 2
produces voltage V at Port 1.

Debapratim Ghosh School of Electrical Sciences, IIT Bhubaneswar

 The Drawback of Z and Y Parameters at Microwave Frequencies

I Each Z or Y parameter can be measured by measuring voltage/current at particular

ports.

I However, at microwave frequencies, measurement of voltage and current is

not easy; it is easier to measure power!

I The measurement of Z and Y are contingent to making a particular port

open-circuit, or short-circuit, respectively.

I However, an open- or a short-circuit port at microwave frequencies does not

guarantee that the network sees the same; this is because of impedance
transformation along the port feeding lines!

I Moreover, making open- or short-circuits at ports for certain types of networks may
damage the network!

I e.g. what happens when you short-circuit the output of an op-amp to

ground? That is undesirable!

Debapratim Ghosh School of Electrical Sciences, IIT Bhubaneswar

 The Preferred Approach for Microwave Frequencies
I As per transmission line theory, an open- or a short-circuit signifies it only at that
point, and not at all points on the line!
I Hence, we need to think in terms of waves. In a transmission line, in general, there
are two travelling waves in opposite directions.
I Rather than trying to make the voltage or current zero, we can try and make any
one of the two waves zero.
I Consider the following N-port network, where each port is fed with a lossless line of
impedance Z0 .

V2- Port 2

Vi+ denotes the wave entering the ith port

Vi- denotes the wave leaving the ith port
V2+
Port N
Port 1
VN -
V1+ N - port VN +
V1- network

Debapratim Ghosh School of Electrical Sciences, IIT Bhubaneswar

 Scattering Parameters

I Just like we do for transmission lines, let us try to relate the outgoing waves Vi− at
all ports (1 ≤ i ≤ N) with all incoming waves Vi+ .
I Here, these waves are of the form of Vi+ = Vf e−γx and Vi− = Vb eγx , as per the
solution of Telegrapher’s equations.
I Let [V − ] be an N × 1 matrix that denotes the outgoing waves at all ports, and [V + ]
an N × 1 matrix denoting incoming waves i.e.
 −  +
V1 V1
V −  V + 
2  2 
[V − ] =  +
 
 .  and [V ] =  .  (7)
  
 .   . 
VN− VN+
I Then, we define the scattering (or S) parameter matrix as
 −    +
V1 S11 S12 .. S1N V1
−
V   S21 S22 .. S2N  V + 
 2    2 
 . = .
   . .. .  . 
  (8)
 .   . . .. .  . 
VN− SN1 SN2 .. SNN VN+

Debapratim Ghosh School of Electrical Sciences, IIT Bhubaneswar

 Scattering Parameters (cont’d..)

I Thus, a single S-parameter can be obtained as

Vi− 

Sij = +  (9)
Vj Vn+ =0,n6=j

I What this means is, an S-parameter is expressed as a ratio of outgoing voltage

wave to incoming voltage wave (any single or two ports), while all incoming waves at
others ports are nullified.

I How can we nullify an incoming wave at each port? Simply by connecting that port
with a matched load (equal to feeding line impedance Z0 ).

I This technique overcomes the limitation of open- and/or short-circuit techniques to

measure network parameters by effectively nullifying a wave!

I Since the pair of waves at any port can result from other ports due to interaction of
waves inside the network, the name ‘scattering’ was given to these parameters.

Debapratim Ghosh School of Electrical Sciences, IIT Bhubaneswar

 Conversion from Z to S-Parameters

I As discussed, the voltage and current matrices for an N-port network can be related
using its Z -parameters, viz.

[V ] = [Z ][I] (10)
I In terms of wave equations as solutions for the Telegraphers’ equations, the voltage
and current matrices can be expressed as a sum of forward and backward wave
matrices as

[V ] = [V + ] + [V − ] (11)
1
[I] = ([V + ] − [V − ]) (12)
Z0
I Substituting these in (10), we get
1
[V + ] + [V − ] = [Z ]([V + ] − [V − ]) (13)
Z0
I Grouping the forward-wave and backward-wave terms separately, and defining
1
normalized impedance matrix as [z] = [Z ], (13) becomes
Z0

Debapratim Ghosh School of Electrical Sciences, IIT Bhubaneswar

 Conversion from Z to S-Parameters (cont’d..)

([z] + [I])[V − ] = ([z] − [I])[V + ] (14)

where [I] is the N × N identity matrix.

I Multiplying the L.H.S. by the inverse of the coefficient matrix, we get

[V − ] = ([z] + [I])−1 ([z] − [I])[V + ] (15)

I Recall the definition of S-parameters from (8) as [V − ] = [S][V + ], and relating this
with the above result, it follows that

[S] = ([z] + [I])−1 ([z] − [I]) (16)

I Does this result look familiar? Notice that this is very similar to the equation for
reflection coefficient in a single transmission line!
I In fact, reflection coefficients are subsets of S-parameters, as we shall soon see.
I Conversely, the normalized impedance matrix can be obtained from the
S-parameters as (the proof is left as an exercise)

[z] = ([I] + [S])([I] − [S])−1 (17)

Debapratim Ghosh School of Electrical Sciences, IIT Bhubaneswar

 Establishing the Condition for Reciprocity for S-Matrix

I The condition for reciprocity through Z -matrix is [Z ]T = [Z ]. If a network is

reciprocal, then what condition will the S-matrix obey?

I Recall from (16) that [S] = ([z] + [I])−1 ([z] − [I])

I Now, if the network is reciprocal, [z]T = [z], and [I] being the identity matrix, is
already a symmetric matrix.

I Therefore, as both matrices are symmetric, the product of their sums will also be
symmetric.

I Hence, the condition for reciprocity can also be written as

[S]T = [S] (18)

Debapratim Ghosh School of Electrical Sciences, IIT Bhubaneswar

 The Condition for Lossless Property for S-Matrix

I It was discussed that if a network is lossless, then the Z matrix must be purely
imaginary. However, it is difficult to generalize a lossless S matrix using this
condition alone.
I A lossless network is defined as one where the algebraic sum of the average (real)
power at all ports equals zero i.e. the real power entering the network should equal
the real power leaving it. Mathematically, this means
1
Pavg = Re{VI ∗ } (19)
2
I Denoting the voltages and currents at all ports in the form of matrices, this becomes
1 1
Re{[V ]T [I]∗ } = Re{([V + ] + [V − ])T ([V + ] − [V − ])∗ }
2 2Z0
1
= Re{[V + ]T [V + ]∗ − [V − ]T [V − ]∗ + [V − ]T [V + ]∗ − [V + ]T [V − ]∗ }
2Z0
I Notice that the first two terms inside the bracket would denote squared magnitude
terms, hence they will be real. The remaining two terms denote matrix operation of
the form [A] − [A]∗ , hence they would be purely imaginary and thus, not needed.

Debapratim Ghosh School of Electrical Sciences, IIT Bhubaneswar

 The Condition for Lossless Property for S-Matrix (cont’d..)

I The average power expression then becomes as follows, equated to zero for
lossless property.
1
([V + ]T [V + ]∗ − [V − ]T [V − ]∗ ) = 0
2Z0
∴[V − ]T [V − ]∗ = [V + ]T [V + ]∗
∴[S]T [V + ]T [S]∗ [V + ]∗ = [V + ]T [V + ]∗

Note that transposes are used in order to multiply the single row/column matrices.

I To maintain the above equality, it requires that

[S]T [S]∗ = [I] (20)

I A matrix that satisfies (20) is called a unitiary matrix.

I Thus, the S matrix for a lossless network must obey the Unitiary Property.

Debapratim Ghosh School of Electrical Sciences, IIT Bhubaneswar

 More About S-parameters

I Consider the diagonal elements of any N × N S-parameter matrix, defined as

V− 

Sii = i +  (21)
Vi V + =0,j6=i
j

where 1 ≤ i, j ≤ N
I Does the above expression look familar? It should, because it is similar to the
reflection coefficient Γ at a point on the transmission line connected to the i th port.
I But, we had defined the reflection coefficient simply as the ratio of the backward to
forward wave, without any additional condition. This is because the analysis at that
time was for a single transmission line connecting a source and load.
I Now, for any N-port network, this same ratio of the backward to forward wave at the
i th port will depend on the load connected to the other ports.
I Hence, the terms Sii define the reflection coefficient at the i th port, provided all other
ports are matched. Like Γ, for stable networks, |Sii | ≤ 1.
I The terms Sij , with i 6= j indicate transmission coefficients. It is possible to have
|Sij | ≤ 1, in case the network has active elements like transistors.

Debapratim Ghosh School of Electrical Sciences, IIT Bhubaneswar

 An Example– Two-Port Network

I A two-port network has four S-parameters S11 , S12 , S21 , S22 .

I The return loss (in dB) at each port is defined as

RL at Port 1 = −20 log10 |S11 | RL at Port 2 = −20 log10 |S22 | (22)

I The insertion losses (in dB), corresponding to the transmision coefficients are
defined as

Insertion loss from Port 1 to 2 = −20 log10 |S21 | (23)

Insertion loss from Port 2 to 1 = −20 log10 |S12 | (24)

I For low-frequency networks, we would question if they are– (i) reciprocal and (ii)
lossless.

I For microwave networks, we question if they are– (i) reciprocal (ii) lossless and (iii)
matched.

I If a network is matched at the i th port, then by definition, Sii = 0. The S-parameter

criteria for lossless and reciprocal networks are already discussed.

Debapratim Ghosh School of Electrical Sciences, IIT Bhubaneswar

 An Example– Two-Port Network (cont’d..)

I Can a 2-port network meet all three conditions– reciprocal, lossless, and matched?
I If all ports are matched, then S11 = S22 = 0.
I If the network is reciprocal, then S21 = S12 = S. The matrix then becomes
 
0 S
[S] =
S 0
I If the network has to be lossless, then it must satisfy the unitiary property, i.e.
[S]T [S]∗ = [I]. Thus,

0 S∗
  
0 S
[S]T [S]∗ = ∗
S 0 S 0
 2 
|S| 0
=
0 |S|2
I This is can be equal to the identity matrix if |S| = 1. Thus, it is indeed possible for a
2-port network to satisfy all three conditions!
I Exercise: Prove that, for any three-port network, all of these three conditions can
never be satisfied!
Debapratim Ghosh School of Electrical Sciences, IIT Bhubaneswar
 Transmission (ABCD) Parameters

I Another set of highly useful circuit parameters called transmission (ABCD)

parameters are defined for a 2-port network, which relate the voltage and current at
one port to those at the other port, as shown.
I1 I2

+ 2-port +
V1 network V2
- -

    
V1 A B V2
= (25)
I1 C D I2

Notice the change in direction of I2 , when compared with Z or Y parameters!

I The usefulness of ABCD parameters lies in the fact that, the equivalent ABCD
matrix of individual series-cascaded two-port networks equals the product of the
individual ABCD matrices in order.
I A and D are voltage and current ratios respectively, while B and C are impedance
and admittance quantities, respectively. It can be shown that the condition for
reciprocity is AD − BC = 1.
Debapratim Ghosh School of Electrical Sciences, IIT Bhubaneswar
 Conversion from ABCD to Z -parameters
I In terms of Z and ABCD parameters of a two-port network, the port voltages and
currents can be related as
ABCD Parameters Z Parameters

V1 = AV2 − BI2 V1 = Z11 I1 + Z12 I2

I1 = CV2 − DI2 V2 = Z21 I1 + Z22 I2
I If Port 2 is made open-circuit (i.e. I2 = 0) then
V1 A
Z11 = = (26)
I1 C
V2 1
Z21 = = (27)
I1 C
I Now, if Port 2 is short-circuited (i.e. V2 = 0) then we obtain the following equations
V1 = −BI2 Z21 I1 + Z22 I2 = 0
I1 = −DI2 Z11 I1 + Z12 I2 = V1 = −BI2
I From these, we obtain
AD
Z12 = −B (28)
C
D
Z22 = (29)
C
Debapratim Ghosh School of Electrical Sciences, IIT Bhubaneswar
 Conversion to S-Parameters

I Once the Z -parameters are obtained, they can be converted into S-parameters,
using (16), and we obtain
A − D + BY0 − CZ0
S11 = (30)
A + D + BY0 + CZ0
2(AD − BC)
S12 = (31)
A + D + BY0 + CZ0
2
S21 = (32)
A + D + BY0 + CZ0
−A + D + BY0 − CZ0
S11 = (33)
A + D + BY0 + CZ0

Debapratim Ghosh School of Electrical Sciences, IIT Bhubaneswar