ECE 604, Lecture 11

Mon, Feb 4, 2019

Contents
1 Transmission Line Theory 2
1.1 Time-Domain Analysis . . . . . . . . . . . . . . . . . . . . . . . . 3
1.2 Frequency-Domain Analysis . . . . . . . . . . . . . . . . . . . . . 6

2 Lossy Transmission Line 7

Printed on March 24, 2019 at 16 : 21: W.C. Chew and D. Jiao.

1
ECE 604, Lecture 11 Mon, Feb 4, 2019

1 Transmission Line Theory

Figure 1:

Transmission lines were the first electromagnetic waveguides ever invented. The
were driven by the needs in telegraphy technology. It is best to introduce trans-
mission line theory from the viewpoint of circuit theory.
Circuit theory is robust and is not sensitive to the detail shapes of the com-
ponents involved such as capacitors or inductors. Moreover, many transmission
line problems cannot be analyzed with the full form of Maxwell’s equations,1
but approximate solutions can be obtained using circuit theory in the long-
wavelength limit.
Examples of transmission lines are shown in Figure 1. The symbol for a
transmission line is usually represented by two pieces of parallel wires, but in
practice, these wires need not be parallel.
1 Usually called full-wave analysis.

2
ECE 604, Lecture 11 Mon, Feb 4, 2019

Figure 2: Courtesy of slides by A. Wadhwa, A.L. Dal, N. Malhotra.

Circuit theory also explains why waveguides can be made sloppily when
wavelength is long or the frequency low. For instance, in the long-wavelength
limit, we can make twisted-pair waveguides with abandon, and they still work
well (see Figure 2). Hence, we shall first explain the propagation of electromag-
netic signal on a transmission line using circuit analysis.

1.1 Time-Domain Analysis

We will start with performing the time-domain analysis of a simple, infinitely
long transmission line. Remember that two pieces of metal can accumulate
attractive charges between them, giving rise to capacitive coupling, electric field,
and hence stored energy in the electric field. Moreover, a piece of wire carrying
a current generates a magnetic field, and hence, yielding stored energy in the
magnetic field. These stored energies are the sources of the capacitive and
inductive effects. But these capacitive and inductive effects are distributed over
the spatial dimension of the transmission line. Therefore, it is helpful to think
of the two pieces of metal as consisting of small segments of metal connected
together. Each of this segment will have a small inductance, as well as a small
capacitive coupling between them. Hence, we can model two pieces of metal
with a distributed lumped element model as shown in Figure 3. For simplicity,
we assume the other conductor to be a ground plane, so that it need not be
approximated with lumped elements.
In the transmission line, the voltage V (z, t) and I(z, t) are functions of both
space z and time t, but we will model the space variation of the voltage and
current with discrete step approximation. The voltage varies from node to node
while the current varies from branch to branch of the lump-element model.

3
ECE 604, Lecture 11 Mon, Feb 4, 2019

Figure 3:

First, we recall that the I-V relation of an inductor is

dI0
V0 = L0 (1.1)
dt
where L0 is the inductor, V0 is the time-varying voltage drop across the inductor,
and I0 is the current through the inductor. Then using this relation between
node 1 and node 2, we have
∂I
V − (V + ∆V ) = L∆z (1.2)
∂t
The left-hand side is the voltage drop across the inductor, while the right-hand
side follows from the aforementioned V-I relation of an inductor, but we have
replaced L0 = L∆z. Here, L is the inductance per unit length (line inductance)
of the transmission line. And L∆z is the incremental inductance due to the
small segment of metal of length ∆z. Then the above can be simplified to
∂I
∆V = −L∆z (1.3)
∂t
Next, we make use of the V-I relation for a capacitor, which is
dV0
I0 = C 0 (1.4)
dt
where C0 is the capacitor, I0 is the current through the capacitor, and V0 is a
time-varying voltage drop across the capacitor. Thus, applying this relation at
node 2 gives
∂ ∂V
−∆I = C∆z (V + ∆V ) ≈ C∆z (1.5)
∂t ∂t
where C is the capacitance per unit length, and C∆z is the incremental capac-
itance between the small piece of metal and the ground plane. In the above,

4
ECE 604, Lecture 11 Mon, Feb 4, 2019

we have used Kirchhoff current law to surmise that the current through the
capacitor is −∆I, where ∆I = I(z + ∆z, t) − I(z, t). In the last approximation
in (1.5), we have dropped a term involving the product of ∆z and ∆V , since it
will be very small or second order in magnitude.
In the limit when ∆z → 0, one gets from (1.3) and (1.5) that
∂V (z, t) ∂I(z, t)
= −L (1.6)
∂z ∂t
∂I(z, t) ∂V (z, t)
= −C (1.7)
∂z ∂t
The above are the telegrapher’s equations. They are two coupled first-order
equations, and can be converted into second-order equations easily. Therefore,
∂2V ∂2V
− LC =0 (1.8)
∂z 2 ∂t2
∂2I ∂2I
2
− LC 2 = 0 (1.9)
∂z ∂t
The above are wave equations that we have previously studied, where the ve-
locity of the wave is given by
1
v=√ (1.10)
LC
Furthermore, if we assume that

V (z, t) = f+ (z − vt) (1.11)

a right-traveling wave, and substituting it into (1.6) yields

∂I 0
−L = f+ (z − vt) (1.12)
∂t
or that
r
1 C
I= f+ (z − vt) = f+ (z − vt) (1.13)
Lv L
Consequently,
r
V (z, t) L
= = Z0 (1.14)
I(z, t) C
where Z0 is the characteristic impedance of the transmission line. The above
ratio is only true for one-way traveling wave, in this case, one that propagates in
the +z direction. In the above, an integration constant could have been added
to (1.13), but this constant is non-causal, and hence, can be eliminated.
For a wave that travels in the negative z direction, i.e.,

V (z, t) = f− (z + vt) (1.15)

5
ECE 604, Lecture 11 Mon, Feb 4, 2019

with the corresponding I(z, t) derived, one can show that

r
V (z, t) L
=− = −Z0 (1.16)
I(z, t) C

Time-domain analysis is very useful for transient analysis of transmission

lines, especially when nonlinear elements are coupled to the transmission line.
Another major strength of transmission line model is that it is a simple way
to introduce time-delay in a circuit. Time delay is a wave propagation effect,
and hence, it is harder to incorporate in circuit theory or a pure circuit model
consisting of R, L, and C.

1.2 Frequency-Domain Analysis

Frequency domain analysis is very popular as it makes the transmission line
equations very simple. Moreover, generalization to a lossy system is quite
straight forward. Furthermore, for linear time invariant systems, the time-
domain signals can be obtained from the frequency-domain data by performing
a Fourier transform.
For a time-harmonic signal on a transmission line, one can analyze the prob-
lem in the frequency domain using phasor technique. The telegrapher’s equa-
tions (1.6) and (1.7) then become

d
V (z, ω) = −jωLI(z, ω) (1.17)
dz
d
I(z, ω) = −jωCV (z, ω) (1.18)
dz
The corresponding Helmholtz equations are then

d2 V
+ ω 2 LCV = 0 (1.19)
dz 2
d2 I
+ ω 2 LCI = 0 (1.20)
dz 2
The general solutions to the above are

V (z) = V+ e−jβz + V− ejβz (1.21)

I(z) = I+ e−jβz + I− ejβz (1.22)
√
where β = ω LC. This is similar to what we have seen previously for plane
waves in the one-dimensional wave equation in free space, where

Ex (z) = E0+ e−jk0 z + E0− ejk0 z (1.23)

√
where k0 = ω µ0 0 . We see a much similarity between (1.21), (1.22), and
(1.23).

6
ECE 604, Lecture 11 Mon, Feb 4, 2019

To see the solution in the time domain, we let V± = |V± |ejφ± , and the
voltage signal above can be converted back to the time domain as
V (z, t) = <e{V (z, ω)ejωt } (1.24)
= |V+ | cos(ωt − βz + φ+ ) + |V− | cos(ωt + βz + φ− ) (1.25)
As can be seen, the first term corresponds to a right-traveling wave, while the
second term is a left-traveling wave.
Furthermore, if we assume only a one-way traveling wave to the right by
letting V− = I− = 0, then it can be shown that, for a right-traveling wave
r
V (z) V+ L
= = = Z0 (1.26)
I(z) I+ C
In the above, the telegrapher’s equations, (1.17) or (1.18) have been used to find
a relationship between I+ and V+ .
Similarly, applying the same process for a left-traveling wave only, by letting
V+ = I+ = 0, then
r
V (z) V− L
= =− = −Z0 (1.27)
I(z) I− C

2 Lossy Transmission Line

Figure 4:

The power of frequency domain analysis is demonstrated in the study of lossy

transmission lines. The previous analysis, which is valid for lossless transmission
line, can be easily generalized to the lossy case. In using frequency domain and
phasor technique, impedances will become complex numbers as shall be shown.
To include loss, we use the lumped-element model as shown in Figure 4. One
thing to note is that jωL is actually the series line impedance of the transmission
line, while jωC is the shunt line admittance of the line. First, we can rewrite
the expressions for the telegrapher’s equations in (1.17) and (1.18) in terms of
series line impedance and shunt line admittance to arrive at
d
V = −ZI (2.1)
dz

7
ECE 604, Lecture 11 Mon, Feb 4, 2019

d
I = −Y V (2.2)
dz
where Z = jωL and Y = jωC. The above can be generalized to the lossy case
as shall be shown.
The geometry in Figure 4 is homomorphic2 to the lossless case in Figure
3. Hence, when lossy elements are added in the geometry, we can surmise that
the corresponding telegrapher’s equations are similar to those above. But to
include loss, we generalize the series line impedance and shunt admittance from
the lossless case to lossy case as follows:

Z = jωL → Z = jωL + R (2.3)

Y = jωC → Y = jωC + G (2.4)

where R is the series line resistance, and G is the shunt line conductance, and
now Z and Y are the series impedance and shunt admittance, respectively.
Then, the corresponding Helmholtz equations are

d2 V
− ZY V = 0 (2.5)
dz 2
d2 I
− ZY I = 0 (2.6)
dz 2
or
d2 V
− γ2V = 0 (2.7)
dz 2
d2 I
− γ2I = 0 (2.8)
dz 2
where γ 2 = ZY , or that one can also think of γ 2 = −β 2 . Then the above is
homomorphic to the lossless case except that now, β is a complex number. As
before, the above are second order one-dimensional Helmholtz equations where
the general solutions are

V (z) = V+ e−γz + V− eγz (2.9)

−γz γz
I(z) = I+ e + I− e (2.10)

where
√ p
γ= ZY = (jωL + R)(jωC + G) = jβ (2.11)

where β = β 0 − jβ 00 is now a complex number. In other words,

0
z−β 00 z
e−γz = e−jβ
2 A math term for “similar in structure”. The term is even used in computer science

describing a emerging field of homomorphic computing.

8
ECE 604, Lecture 11 Mon, Feb 4, 2019

is an oscillatory and decaying wave. Or focusing on the voltage case,

00
z−jβ 0 z 00
z+jβ 0 z
V (z) = V+ e−β + V− eβ (2.12)

Again, letting V± = |V± |ejφ± , the above can be converted back to the time
domain as

V (z, t) = <e{V (z, ω)ejωt } (2.13)

00 00
= |V+ |e−β z
cos(ωt − β 0 z + φ+ ) + |V− |eβ z
cos(ωt + β 0 z + φ− ) (2.14)

The first term corresponds to a decaying wave moving to the right while the
second term is also a decaying wave moving to the left. When there is no loss,
or R = G = √ 0, and from (2.11), we retrieve the lossless case where β 00 = 0 and
γ = jβ = jω LC.
Notice that for the lossy case, the characteristic impedance, which is the
ratio of the voltage to the current for a one-way wave, can be derived using
homomorphism:
r s r s
V+ V− L jωL Z jωL + R
Z0 = =− = = → Z0 = = (2.15)
I+ I− C jωC Y jωC + G

The above Z0 is manifestly a complex number. Here, Z0 is the ratio of the

phasors of the one-way traveling waves, and apparently, they current phasor
and the voltage phasor will not be in phase for lossy transmission line.
In the absence of loss, the above becomes
r
L
Z0 = (2.16)
C
the characteristic impedance for the lossless case previously derived.