Transmission lines

Shouri Chatterjee
October 20, 2024

The transmission line is a very commonly used distributed circuit: a pair

of wires. Unfortunately, a pair of wires used to apply a time-varying voltage
is not ideal when the wavelength of the signal applied is comparable to the
length of the wires. The following is a brief analysis of the transmission line.

1 The general distributed circuit

A small section, of length ∆x, of a transmission line is modeled as a combi-
nation of a series resistor, a series inductor, a shunt capacitor, and a shunt
conductor. A transmission line is a cascade of several of these small sec-
tions. The values of the resistor, inductor, capacitor, and conductor are
proportional to the length of the transmission line section. We typically
denote these quantities as per-unit-length parameters. We have shown a
transmission-line section in Fig. 1.
We could compute the transmission matrix for the two-port network in
Fig. 1 and cascade an infinite number of these. However, we will not pursue
this analysis style in this discussion.

i(x, t) R′ ∆x L′ ∆x i(x + ∆x, t)

v(x, t) G′ ∆x v(x + ∆x, t)

C ′ ∆x

∆x

Figure 1: A section of a transmission line, of length ∆x. The inductance,

capacitance, resistance, and conductance per unit length are L′ , C ′ , R′ , and
G′ respectively.

1
 One can construct other distributed circuits. The analyses of other dis-
tributed circuits are typically similar to the analysis we present here for
transmission lines.

2 Analysis technique
Let us first evaluate Kirchoff’s equations for the transmission-line section in
Fig. 1. Kirchoff’s voltage and current laws give the following equations, (1),
(2), respectively.

di(x, t)
v(x, t) − v(x + ∆x, t) = L′ ∆x + i(x, t)R′ ∆x (1)
dt
dv(x + ∆x, t)
i(x, t) − i(x + ∆x, t) = C ′ ∆x + v(x + ∆x, t)G′ ∆x (2)
dt
For this treatment of transmission lines, let us assume that they are loss-less,
i.e., R′ and G′ are zero. We will extend our analysis of loss-less transmission
lines to lossy transmission lines later. Further, let us analyze transmission-
line sections of infinitesimally small length, at the limit of ∆x tending to
zero. (1) then reduces to:

v(x, t) − v(x + ∆x, t) di(x, t)

lim = L′
∆x→0 ∆x dt
∂v ∂i
or, − = L′ (3)
∂x ∂t
We have started using partial derivatives because both i and v are functions
of two variables, x, and t. Reduction of (2) leads us to:

i(x, t) − i(x + ∆x, t) ∂v(x + ∆x, t)

lim = C′ (4)
∆x→0 ∆x ∂t
∂v(x+∆x,t) ∂v(x,t)
With ∆x tending to zero, the term ∂t
will be equal to ∂t
. (2) will,
therefore, further simplify to:
∂i ∂v
− = C′ (5)
∂x ∂t
(3), (5) together form a pair of coupled partial differential equations in two
variables. They are known as the Telegrapher’s equations. The solution of
the Telegrapher’s equations is the solution for the transmission-line problem.

2
 The Telegrapher’s equations:
∂v ∂i
= −L′ (6)
∂x ∂t
∂i ′ ∂v
= −C (7)
∂x ∂t
The solution to the Telegrapher’s equations is the solution to the
transmission-line problem.

3 The wave equation

The first step towards solving the Telegrapher’s equations is to decouple the
two equations. We can take the time derivative of (6) and compare it with
the spatial derivative of (7). The outcome of this is as follows:

∂ 2v ∂ 2i
= −L′ 2
∂x∂t ∂t
∂ 2i 2
′ ∂ v
and = −C
∂x2 ∂t∂x
2 2
∂ i ′ ′∂ i
This gives: = LC 2 (8)
∂x2 ∂t
Likewise, we can take the time derivative of (7) and compare it with the
spatial derivative of (6). This results in a similar equation:

∂ 2v 2
′ ′∂ v
= L C (9)
∂x2 ∂t2
(8) and (9) are not coupled to each other. (8) and (9) happen to be the
same equation, one with voltage as the variable and the other with time.
In mathematics literature, the form of this equation is known as the wave
equation.
The solution to the wave equation is the sum of a forward-moving wave
and a backward-moving wave. If f (x) is any function of the variable x, then
f (x − ct) is a forward-moving wave, f (x + ct) is a backward-moving wave. c
is the velocity of the wavefront.
Plugging in f (x − ct) as a solution to i(x, t) in (8), we get:

f ′′ (x − ct) = c2 (−1)2 f ′′ (x − ct)

3
 f (x − ct0 )

f (x − ct1 )

f (x − ct2 )

ct0 ct1 ct2 x

Figure 2: One can jerk one end of a rope to release a wavefront on the rope.
The wave is an example of f (x − ct).
√
In other words, the velocity of the wavefront, c, is given by 1/ L′ C ′ . Plug-
ging f (x+ct) as a solution gives an identical result. Working with (9) instead
of (8) gives the same result, but for v(x, t).
For a second-order differential equation, there should be two degrees of
freedom. The two degrees of freedom are accounted for when we indepen-
dently choose functions for the forward and backward moving waves.

The general solution for both, v(x, t) and i(x, t), is therefore:

f1 (x − ct) + f2 (x + ct)
√
f1 and f2 are arbitrary functions. c = 1/ L′ C ′ .

4 Putting it all together

So far, we have solved the wave equation for voltages and currents, but not
the pair of Telegrapher’s equations. In the solution for voltages and currents
with the help of the wave equation, f1 and f2 are arbitrary functions, where
f1 is the forward-moving wave, f2 is the backward-moving wave. Let us say
that v(x, t) is the sum of a forward-moving function called V+ (x − ct) and a
backward-moving function called V− (x + ct). Also, say i(x, t) is the sum of a
forward-moving I+ (x − ct) and a backward-moving I− (x + ct).
v(x, t) = V+ (x − ct) + V− (x + ct) (10)
i(x, t) = I+ (x − ct) + I− (x + ct) (11)

4
The equations (10), (11), now can be plugged into the Telegrapher’s equa-
tions, (6) and (7).

V+′ (x − ct) + V−′ (x + ct) = −L′ (−cI+′ (x − ct) + cI−′ (x + ct)) (12)
I+′ (x − ct) + I−′ (x + ct) = −C ′ (−cV+′ (x − ct) + cV−′ (x + ct)) (13)

In (12) and (13), we can solve for I√+′ and I−′ in terms of V+′ and V−′ .
Further, plugging in the value of c as 1/ L′ C ′ , we obtain the following.

I+′ (x − ct) = V+′ (x − ct)/Z0 (14)

I−′ (x + ct) = −V−′ (x + ct)/Z0 (15)
r
L′
where Z0 = (16)
C′
The function I+ is, therefore, V+ /Z0 with the possible addition of an
p −V− /Z0 with the possi-
arbitrary constant. The function I− is, therefore,
ble addition of an arbitrary constant. Z0 = L′ /C ′ is defined to be the
characteristic impedance of the transmission line. Note: Z0 is only a
mathematical quantity. It is not a resistor or an impedance.

5 Interpretation of the solution

The analysis of the earlier sections tells us that voltages and currents move
like waves over a transmission line. Kirchoff’s laws are valid at all locations at
all times; however, the current going into a transmission line at the input port
might not be the current coming out of the transmission line at the output
port because of intermediate paths through the distributed capacitors.
The voltage along a transmission line is the sum of a forward-moving volt-
age wave and a backward-moving voltage wave. The current travelling along
a transmission line is also the sum of a forward-moving current wave and a
backward-moving current wave. Z0 , the characteristic impedance, relates the
voltage
p and current waves. The velocity
√ of all the wavefronts is c. Z0 is given
by L′ /C ′ , and c is given by 1/ L′ C ′ .1 The forward-moving current wave
is the forward-moving voltage wave divided by Z0 . The backward-moving
current wave is the negative of the backward-moving voltage wave divided
by Z0 .
A computation of L′ , the inductance per unit length and C ′ , the capacitance per unit
1

length, for simple geometries of transmission lines, will reveal that c is the speed of light
in the given medium. Sometimes, this can be a quick way to estimate C ′ if one knows L′ ,
or vice-versa.

5
 ZS

Vin Z0 ZL

Figure 3: A source with source impedance ZS is being used to excite a load

impedance, ZL , over a transmission line of characteristic impedance Z0 .

We can summarize the discussion as follows. Over a transmission line,

there will be a forward-moving voltage wave, V+ , and a backward-moving
voltage wave, V− . At any point on the transmission line, at any instant, the
voltage is V+ + V− , and the current is (V+ − V− )/Z0 .

6 Reflection coefficient
Consider a transmission line of characteristic impedance Z0 , terminated by
a load, ZL , as shown in Fig. 3.
At the load end, if the forward and backward moving waves are V+ and
V− respectively, then the voltage is V+ + V− , and the current into the load
is (V+ + V− )/ZL . The current is also the sum of the forward and backward
moving current waves, (V+ − V− )/Z0 .
V+ + V− V+ − V−
=
ZL Z0
V+ + V− ZL
or =
V+ − V− Z0
V− ZL − Z0
or ΓL = = (17)
V+ ZL + Z0

In (17), we have defined ΓL to be the ratio of the backward-moving wave to

an impinging forward-moving wave at the load. If a forward-moving wave,
V+ , hits a load ZL through a transmission line of characteristic impedance
Z0 , then there will be a reflection from the load given by V− = ΓL V+ . ΓL is
called the reflection coefficient. (17) shows that the reflection coefficient is
0 if the load is also Z0 , the reflection coefficient is +1 if the load is an open
circuit, and the reflection coefficient is −1 if the load is a short circuit.

Corollary 1. If ZL = Z0 , there are no reflections from the load, and the

transmission line is transparent, i.e., the input impedance is nothing but
Z0 . Similarly, if ZS = Z0 , there are no reflections from the source, and the
transmission line is transparent.

6
 When ZL = Z0 , the transmission line is said to be matched at the load.
The transmission line is matched at the source if ZS = Z0 .

7 The response to a pulse

Let us now consider the situation where the voltage source in Fig. 3 is a step
function, Vin = V0 u(t). Before time t = 0, both the forward and backward
moving waves should be equal to 0. Things should start changing only at
time t = 0+ . At this instant, the input voltage is V0 . The voltage across the
input terminals of the transmission lines is unknown; let us assume this to be
Vx . The current going into the input terminals will then be (V0 −Vx )/ZS . The
voltage Vx is the sum of a forward-moving wave that has just started and a
backward-moving wave that is zero. The statement implies that the forward-
moving wave is Vx . The current, therefore, has to be Vx /Z0 . (Remember, the
current is (V+ − V− )/Z0 . V+ is Vx , V− is 0.) This gives:

V0 − Vx Vx Z0
= or, Vx = V0 (18)
ZS Z0 ZS + Z0
(18) shows that the voltage appearing at the input of the transmission line is
the result of a resistive division of the applied voltage as if the transmission
line is a resistor of value Z0 . The transmission line is not a resistor Z0 , and
the resistive division appears to hold only initially.
The forward-moving wave, Vx , moves forward at a velocity c. After a
duration of l/c (where l is the length of the transmission line), it reaches
the load. At the load end, the wave Vx impinges on the load and creates a
reflection, ΓL Vx , at time t = (c/l)+ . Now the voltage at the load is Vx (1+ΓL ).
The backward-moving wave is now ΓL Vx .
The backward-moving wave, ΓL Vx , moves backward and impinges on the
source resistor at t = 2l/c. The reflection coefficient at the source (now look
at the transmission line backward, the backward-moving wave is your new
−Z0
forward direction, the source resistor is your new load resistor) is ΓS = ZZSS +Z0
.
The impinging backward-moving wave will cause a reflection, a new forward-
moving wave, ΓS ΓL Vx , at time t = (2l/c)+ . The voltage at the source is now
going to be Vx (1 + ΓL + ΓL ΓS ).
The waves slosh back and forth forever - till the voltages settle to their
final values. Graphically, this process is described in the space-time chart
of Fig. 4. At any point in space-time, the voltage is the sum of the voltage
waves that have passed till then.
The voltage at the load as the system settles to steady-state is going to

7
 t=0 V+
ΓL V + t = l/c
t = 2l/c ΓL ΓS V +
Γ2L ΓS V+ t = 3l/c
t = 4l/c Γ2L Γ2S V+
Γ3L Γ2S V+ t = 5l/c
t = 6l/c Γ3L Γ3S V+
Γ4L Γ3S V+ t = 7l/c
t = 8l/c Γ4L Γ4S V+
x=0 x=l

Figure 4: Response to a pulse input. The x-axis is distance, the y-axis is

time, going downwards. At any point, x, one can draw a section. The voltage
at x at any time t will be the sum of all the waves that have passed x till
time t.

be given by:
∞
X ∞
X
lim V (l) = (Vx ΓiS ΓiL + Vx ΓiS Γi+1
L ) = Vx ΓiS ΓiL (1 + ΓL )
t→∞
i=0 i=0
∞
X 1 + ΓL
= Vx (1 + ΓL ) (ΓL ΓS )i = Vx (19)
i=0
1 − ΓL ΓS

We can plug values for ΓL , ΓS , and Vx into (19) as follows. ΓL = (ZL −

Z0 )/(ZL + Z0 ), ΓS = (ZS − Z0 )/(ZL + Z0 ), and Vx = Vin Z0 /(ZS + Z0 ). This
results in:
−Z0
Z0 1 + ZZLL +Z0
ZL
lim V (l) = Vin · · ZL −Z0 ZS −Z0
= Vin
t→∞ ZS + Z0 1 − Z +Z Z +Z ZL + ZS
L 0 S 0

The result is indeed intuitive. After all, at DC, the inductors and capacitors
in the transmission line are all short-circuits and open-circuits, respectively.
The short and open substitutions imply that at DC, the transmission line is
just a wire. Voltage division between ZL and ZS is therefore natural.

Example 1. What happens when ΓL = 1 and ΓS = −1?

8
 v(l)

0
v(l/2) T 3T 5T 7T 9T 11T 13T
2
1
0

Figure 5: Voltage waveforms at the far end of the transmission line and at
the middle of the transmission line when the load is an open circuit, and the
source resistance is 0. T is the time the wave takes to travel from one end of
the line to the other.

ΓL = 1 corresponds to ZL → ∞ when the transmission line is terminated

in an open circuit. ΓS = −1 corresponds to ZS = 0 when an ideal
voltage source is directly applied to the transmission line system. For
these values of ΓL and ΓS , (19) will not converge. One can still use a
chart like Fig. 4 to work out the waveform at the load for a voltage source
u(t). The voltage at the source is 1 for t > 0. At the load, the voltage
jumps from 0 to 2 at t = l/c = T because ΓL is 1. The reflected wave
moves to the source end and is further reflected as a −1. This reflection
brings the load voltage down to 0 at t = 3T , and so on, as shown in
Fig. 5. We have also shown the voltage seen at x = l/2 in Fig. 5. At
x = l/2, we see a forward voltage wave +1 at T /2, a reverse voltage wave
+1 at 3T /2, a forward wave −1 at 5T /2, a reverse wave −1 at 7T /2, and
so on.
On average, the voltage at all points along the transmission line will be
1 for t > 0. However, this will be sloshing back and forth between 0 and
2 with a period of 4T = 4l/c.

Example 2. A 220 V DC generator has a source impedance of close to

0 Ω. A transmission line of length 1 km and characteristic impedance of
10 Ω is being used to deliver power to a load. The load suddenly switches
OFF (i.e., it becomes an open circuit). There will be an instantaneous
reflected wave from the load of 220 V when the load switches OFF. The
reflection creates a transient at the load-switch of 440 V. The switch
used to switch the load off needs to be rated at least for 440 V.

9
 ZS L0.5 L1.5 L2.5

vin (t) C0 C1 C2 C3 ZL

Figure 6: We can model the transmission line as a cascade of K sections.

The model is more and more accurate as K becomes larger and larger. In
general, choose K such that the wavelength of the maximum frequency to be
observed is much larger than d/K, where d is the length of the transmission
line. This diagram illustrates the case where K = 3.

Exercise 1. Show that the voltage at the source end of the transmission
line also settles to Vin ZL /(ZL + ZS ) as time progresses. (i.e., the same
result as the load end of the transmission line.)

Exercise 2. Plot the voltage at the load end of the transmission line as
a function of time, when Z0 is 50 Ω, ZL is 100 Ω, and ZS is 200 Ω.

8 Numerical simulation
Numerically simulating transmission lines is not simple. However, we can
break the transmission line into discrete sections and simulate it like an ordi-
nary RLC circuit with techniques studied in Chapter num-. Unfortunately,
simulation is always heavy on resources and slow.
The first step for numerical simulation is to approximate the transmission
line as a cascade of K sections (Fig. 6). K, the number of sections, should
be chosen such that the length of each section is much smaller than the
maximum frequency of interest. The simulation will not be able to capture
frequencies higher than this.
In Fig. 6, all the inductors labeled Lx.5 are of value L′ d/K where d is
the length of the line, L′ is the inductance per unit length, K is the number
of sections. The capacitors C0 and CK are 12 C ′ d/K, all other capacitors are
C ′ d/K, where C ′ is the capacitance per unit length. We have broken up
the first and last capacitors into halves to add symmetry to the model. One
may visualize each section as a pi-network of 21 C ′ d/K, L′ d/K, and 12 C ′ d/K.
Other components, such as R′ and G′ , could be added into the model.

10
 The second step is to write out the A matrix to solve the circuit using
the state-space equations of (ttt-6). For this, we identify the state variables
as all the inductor currents (left to right) and all the capacitor voltages (top
to bottom). In our circuit, we will have 2K + 1 state variables. In the case
of K = 3,  ′
X = v0 i0.5 v1 i1.5 v2 i2.5 v3
Then we can write out the KVL/KCL equations to obtain the derivatives of
the state variables. This operation is readily done as follows.

C0 v̇0 = −i0.5 + vin /ZS − v0 /ZS Boundary condition

L0.5 i̇0.5 = v0 − v1
C1 v̇1 = i0.5 − i1.5
L1.5 i̇1.5 = v1 − v2
C2 v̇2 = i1.5 − i2.5
L2.5 i̇2.5 = v2 − v3
C3 v̇3 = i2.5 − v3 /ZL Boundary condition

The state-space form representation is now ready.

  − 1 − C10 0 0 0 0 0
  
vin 
v̇0 C0 ZS v0 C 0 ZS
1 1
i̇0.5    L0.5
0 − L0.5 0 0 0 0   i0.5   0 
1
− C11
      
 v̇1   0 C1
0 0 0 0   v1   0 
1 1
      
i̇1.5  =  0 0 0 − 0 0 i + 0
  
1.5

   L1.5 L1.5     
1 1
 v̇2   0 0 0 0 − 0  v2   0 

   C2 C2     
1 1  i   0 
i̇2.5   0 0 0 0 L2.5
0 − L2.5  2.5
v̇3 0 0 0 0 0 1
− C31ZL v3 0
C3

The state-transformation matrix can easily be generalized to an arbitrary

number of sections. We can use the Octave program in Prog. 1 to simulate,
visualize, and animate the propagation of waves through a transmission line.
The sample code of Prog. 1 has also incorporated R′ , which requires a minor
modification in the state-transformation matrix.

Program 1. simtxline.m: Script to simulate a transmission line

1 Z0 = 50; c = 3 e8 ;
2 Lp = Z0 / c ; Cp = 1/ c / Z0 ; % Standard c a b l e
3 Rp = 0.001; % 1 mOhm p e r meter .
4 d = 30; % Length i n m e t e r s
5
6 % Number o f s e c t i o n s
7 K = 400; % A r b i t r a r y

11
 8 Zs = 25; Zl = 100;
9
10 % I n d u c t a n c e and c a p a c i t a n c e p e r s e c t i o n
11 Lps = Lp * d / K ; Cps = Cp * d / K ;
12 Rps = Rp * d / K ;
13
14 % Number o f s t a t e v a r i a b l e s = 2K+1
15 Xinit = zeros (2* K +1 , 1);
16
17 % Time r e s o l u t i o n
18 t_res = d / c / K /100;
19 t_stop = 3* d / c ;
20
21 % Input
22 Vinput = zeros (1 ,round( t_stop / t_res )+1);
23 Vinput (1 ,round(10 e -9/ t_res ):end)=1; % S t e p f u n c t i o n
24 % Any o t h e r i n p u t waveform can be c r e a t e d . For example :
25 % Gaussian p u l s e a p p l i e d from 10 ns t o 30 ns
26 % t t = −10e −9: t r e s : 1 0 e −9;
27 % Vinput ( 1 , round (10 e−9/ t r e s ) : round (30 e−9/ t r e s ) ) = exp (−( t t /2 e − 9 ) . ˆ 2 ) ;
28
29 % Initialize
30 Xs = zeros (2* K +1 ,round( t_stop / t_res +1));
31 A = zeros (2* K +1 , 2* K +1); BU = zeros (2* K +1 ,1);
32
33 % Define A
34 f o r n =1: K
35 A (2* n ,2* n -1)=1/ Lps ;
36 A (2* n ,2* n )= - Rps / Lps ;
37 A (2* n ,2* n +1)= -1/ Lps ;
38 end;
39 f o r n =1: K -1
40 A (2* n +1 ,2* n )=1/ Cps ;
41 A (2* n +1 ,2* n +2)= -1/ Cps ;
42 end;
43 A (1 ,1)= -2/ Cps / Zs ; A (1 ,2)= -2/ Cps ;
44 A (2* K +1 ,2* K )=2/ Cps ; A (2* K +1 ,2* K +1) = -2/ Zl / Cps ;
45
46 % Prepare f o r s i m u l a t i o n
47 h = t_res ;
48 inv_be = inv (eye (2* K +1) - h * A ); % For BE
49 %i n v t r = i n v ( e y e (2∗K+1)−h /2∗A) ; % For t r a p e z o i d a l
50 %i n v g 2 = i n v (3∗ e y e (2∗K+1)−2∗h∗A) ; % For Gear2
51
52 % Simulate
53 f o r k =2:(round( t_stop / t_res ))
54 BU (1) = 2* Vinput (1 , k )/ Zs / Cps ;
55 %Xs ( : , k ) = i n v t r ∗ ( e y e (2∗K+1)+h /2∗A)∗ Xs ( : , k−1)+h∗BU; %Trap
56 %Xs ( : , k ) = i n v g 2 ∗ (4∗ Xs ( : , k−1)−Xs ( : , k −2)+2∗h∗BU) ; % Gear2
57 Xs (: , k ) = inv_be * ( Xs (: ,k -1)+ h * BU ); % BE
58 end;
59
60 % P l o t v o l t a g e s a t i n p u t , b e g i n n i n g o f t x l i n e , end o f t x l i n e
61 vbegin = Xs (1 ,:); % V o l t a g e w i t h time a t i n p u t
62 vend = Xs (2* K +1 ,:); % V o l t a g e w i t h time a t o u t p u t
63 f i g u r e (1); c l f ;
64 plot ([0: t_res : t_stop ] , Vinput , ’r ’ ); hold on ; grid on ;
65 plot ([0: t_res : t_stop ] , vend , ’b ’ ); hold on ; grid on ;
66 plot ([0: t_res : t_stop ] , vbegin , ’k ’ ); hold on ; grid on ;
67
68 % Animate , v o l t a g e w i t h d i s t a n c e
69 x = 0: d / K : d ;

12
 1 vin (t)
v0 (t)
0.8 vL (t)
Voltage (V)
0.6

0.4

0.2

0 50 100 150 200 250

Time (ns)
2
Figure 7: Simulated waveform for a Gaussian pulse, e−(t/T ) , where T is 10 ns.
The length of the transmission line is 30 m, Z0 is 50 Ω, R′ is 1 mΩ/m, ZS and
ZL are 25 Ω and 100 Ω respectively. The input voltage, vin (t), the voltage
at the input of the transmission line, v0 (t), and the voltage at the load end,
vL (t), have been shown.

70 f i g u r e (2); c l f ;
71 f o r k =1:100:round( t_stop / t_res +1)
72 plot (x , Xs (1:2:end, k ));
73 ylim ([ -0.2 , 1.2]);
74 pause (0.001);
75 end;

On simulation, one will notice that the trapezoidal method gives signif-
icant ringing; ringing with Gear-2 is somewhat lesser, and the least ringing
is in BE. However, this ringing is unavoidable. As soon as we have sectioned
the transmission line, we have limited the maximum observable simulation
frequency. As such, the frequencies higher than this limit are not observed
in the output of the simulation. Since a step input has infinite bandwidth,
the manifestation of this band-limiting is a ringing in the simulation. The
ringing behavior would be absent If the input waveform had been a more
gentle waveform with less bandwidth, such as a Gaussian pulse (Fig. 7) or a
sinusoid.
Numerical simulations are resource-intensive. However, we require nu-
merical simulations in most practical situations, such as where the solution
is not a simple wave equation, the transmission line is lossy, and the source
and load impedances are not resistive.

13
9 Steady-state sinusoid stimulation
Without any loss of generality, the forward and backward moving waves can
be phasors. In other words, V+ (x − ct) and V− (x + ct) may be V+ ej(ωt−kx)
and V− ej(ωt+kx) , respectively. Here k is the wavenumber and is given as
k = ω/c = 2πf /c = 2π/λ, where λ is the wavelength corresponding to the
frequency of excitation. The phasor corresponding to the forward-moving
wave is, then, V+ e−j2πx/λ . The backward-moving wave is V− ej2πx/λ . V+
and V− are complex numbers, in general. The current at any point is the
difference of the forward and backward moving voltage waves, divided by Z0 .
We are working entirely in the phasor domain.
The vital thing to note about analysis with phasors is the reflection co-
efficient. The reflection coefficient at the load is still given by:
ZL − Z0
ΓL =
ZL + Z0
ZL can now be any complex impedance. The same analysis as (17) holds.
The forward-moving wave at any point ahead of the load will be the
forward-moving wave at the load, only advanced in time. At any point
ahead of the load, the backward-moving wave will be delayed in time. For
example, at the source end, the forward-moving wave will be ej2πl/λ times the
forward-moving wave at the load. The backward-moving wave will be e−j2πl/λ
times the backward-moving wave at the load. The reflection coefficient at
the source is the ratio of the backward-moving to the forward-moving waves
at the source. This tells us that Γ(0) = ΓL e−j4πl/λ . Now we can compute
the reflection coefficient at any point in space over the entire length of the
transmission line.

Γ(x) = ΓL e−j4π(l−x)/λ (20)

With the help of the boundary conditions, we should work out the voltages
at the boundaries. With the reflection coefficient and the boundary voltages,
we can evaluate the forward and backward moving waves. Once we know the
forward and backward moving waves, we can evaluate the current.

Example 3. A source of impedance 40 Ω is used to transmit power over

a 50 Ω transmission line of length 2.5λ to a load of 60 Ω. VSWR (voltage
standing wave ratio) is the ratio of the maximum voltage amplitude
over the entire length of the transmission line to the minimum voltage
amplitude over the entire length of the transmission line. Let us evaluate
the VSWR.

14
 0.6

normalized |V (x)|
0.58
0.56
0.54
0.52
0.5
0 0.5 1 1.5 2 2.5
distance in λ

Figure 8: Voltage amplitude along a 2.5λ long transmission line. ZL is 60 Ω,

while Z0 is 50 Ω. The waveform, also known as a standing wave, has a VSWR
of 1.2.

We have given a small Octave routine showing all the steps to solve the
problem in Prog. 2.

Program 2. losslesstxline.m: Evaluate a lossless transmission line

1 ZS = 40; ZL = 60; Z0 = 50;
2 ll = 2.5; % Length r e l a t i v e t o lambda
3 xl = [0:0.01: ll ] ’; % D i s t a n c e r e l a t i v e t o lambda
4
5 GL = ( ZL - Z0 )/( ZL + Z0 ); % r e f l e c t i o n c o e f f a t l o a d end
6 G0 = GL * exp( - j *2* ll *2* pi ); % r e f l e c t i o n c o e f f a t s o u r c e end
7 Zin = Z0 * (1+ G0 )/(1 - G0 ); % i n p u t impedance
8
9 V1 = 1 * Zin /( Zin + ZS ); % I n p u t v o l t a g e ( n o r m a l i z e d )
10 V1p = V1 / (1+ G0 ); % Forward moving v o l t a g e wave
11 V1m = V1p * G0 ; % Backward moving v o l t a g e wave
12
13 V = V1p * exp( - j *2* pi * xl ) + V1m * exp( j *2* pi * xl ); % V e v e r y w h e r e
14 plot ( xl , abs( V )); % VSWR = max( a b s (V) ) / min ( a b s (V) )

We have shown a plot of the amplitude of the voltage along the trans-
mission line in Fig. 8. In a laboratory, one can experimentally estimate
the VSWR from the plot of the voltage amplitude along the length of
the transmission line. Notice that the amplitude is periodic, with a peri-
odicity of λ/2. There are two peaks and two valleys in every λ length of
the line. The amplitude along the transmission line is a standing wave
because of the (constructive and destructive) interference of the forward
and backward moving waves. The VSWR is 1.2.

15
 d

Z0 ZL
Zin

Figure 9: Input impedance looking into a transmission line terminated by

ZL .

Exercise 3. A 1 MV 50 Hz generator of source impedance 0.1 Ω is used

to transmit power over a 5 Ω 200 km high-tension cable to a load of
impedance 1 + j2 Ω. Assume that the velocity of the wave in the medium
is 3 × 107 meters per second. Find the rating of the insulation required
for the cable.

Example 4. An impedance ZL terminates a lossless transmission line of

characteristic impedance Z0 and length d. What is the input impedance,
Zin , as shown in Fig. 9?
At the load end, V+ and V− are related by ΓL . At the input side, the
forward-moving wave will be advanced in time by d/c. The backward-
moving wave will be delayed in time by d/c. The phase corresponding
to a time of d/c is 2πd/λ.

V(d) = V+ (d) + V− (d)

I(d) = V+ (d)/Z0 − V− (d)/Z0 = V(d)ZL
V(0) = V+ (d)ej2πd/λ + V− (d)e−j2πd/λ
I(0) = V+ (d)ej2πd/λ /Z0 − V− (d)e−j2πd/λ /Z0
∴ Zin = V(0)/I(0)
ej2πd/λ + Γ(d)e−j2πd/λ
= Z0
ej2πd/λ − Γ(d)e−j2πd/λ
(ZL + Z0 )ej2πd/λ + (ZL − Z0 )e−j2πd/λ
= Z0
(ZL + Z0 )ej2πd/λ − (ZL − Z0 )e−j2πd/λ
ZL cos 2πd/λ + jZ0 sin 2πd/λ
= Z0
jZL sin 2πd/λ + Z0 cos 2πd/λ
ZL + jZ0 tan 2πd/λ
= Z0 (21)
Z0 + jZL tan 2πd/λ

16
 30 mm Z0 = 500 Ω

Z0 = 5 Ω
input 50 Ω output

30 mm

Figure 10: A circuit with open circuit and short circuit stubs.

10 Quarter-wavelength lines
From the preceding discussion, it follows that transmission lines of length
λ/4 are unique. For such transmission lines, from (20), Γ(0) is given by −ΓL .
The input impedance, looking into the source end of the transmission line,
is given by:
V+ + V− 1 + Γ(0)
Zin = V /I = Z0 = Z0 (22)
V+ − V− 1 − Γ(0)
Given that Γ(0) is nothing but −ΓL , we obtain the input impedance as
1−ΓL
Z0 1+Γ L
. With the value of ΓL from (17), the input impedance is Z02 /ZL .
For an open-circuit load, the input impedance looks like a short-circuit.
For a short-circuit load, the input impedance looks like an open-circuit. For
a capacitive load, the input impedance looks like an inductor and vice-versa.
A λ/4 long transmission line inverts the load impedance, normalized to Z0 .
Note, this property is valid only at frequencies at which the length of the
transmission line is exactly an odd multiple of λ/4.

Example 5. A circuit is designed using an open circuit stub and a short

circuit stub, as shown in Fig. 10. (An open circuit stub is a transmission
line of characteristic impedance Z0 terminated with an open circuit.
Likewise, a short circuit stub is a transmission line terminated with a
short circuit.) Find the system transfer function, from the input to
the output, H(jω) as a function of ω. (The wave velocity is 3 × 108
meters/second.)
Applying the limit ZL → ∞ to (21), the input impedance of an open-
circuit stub is Z0 /j tan(2πd/λ). The input impedance of a short-circuit

17
 50 nH

output
input 20 pF 50 Ω

Figure 11: Lumped circuit model of the circuit in Fig. 10.

stub, from (21), is jZ0 tan(2πd/λ). In terms of angular frequency, ω,

these work out to:
Z0
Open-circuit stub: Zin = (23)
j tan(ωd/c)
Short-circuit stub: Zin = jZ0 tan(ωd/c) (24)
Now for ω such that ωd/c ≪ 1, or ω ≪ c/d, tan(ωd/c) ≈ ωd/c. In our
case, c/d for both the stubs is 10 G-rad/s. For ω ≪ 10 G-rad/s, the
open-circuit stub behaves like a capacitor of value d/(cZ0 ) = 20 pF, and
the short-circuit stub behaves like an inductor of value Z0 d/c = 50 nH.
Effectively, for ω ≪ 10 G-rad/s, the circuit can be modeled as shown in
Fig. 11.
The circuit model in Fig. 11 can now be easily analyzed. The analysis
results of the circuit model (dashed), as opposed to the actual transmis-
sion line (solid), are shown in Fig. 12. Divergence of the model from the
actual result is visible beyond 5 G-rad/s.

11 Lossy transmission lines

The analysis and discussion since Section 2 assume that the transmission line
is lossless, and R′ and G′ are both zero. Let us now consider non-zero R′
and G′ . Unfortunately, time-domain analysis of such a set of equations will
not be possible. We will now assume that all the voltages and currents are
steady-state sinusoids (or sums thereof) and proceed further. We are going
to represent the voltages and currents as phasors.
We have redrawn the transmission line section of Fig. 1 in Fig. 13 with
a few modification. Here V(x) and I(x) are phasors, i.e., V(x) is really
v(x, t) = |V (x)|ej(ωt+∠V ) .
If we analyze the same way as (1) and (2), then we obtain:
V(x) − V(x + ∆x) = (jωL′ ∆x + R′ ∆x)I(x) (25)
I(x) − I(x + ∆x) = (jωC ′ ∆x + G′ ∆x)V(x + ∆x) (26)

18
 0
Transfer function, |H(jω)|

LC-Model
−20

−40

−60
Actual
−80
107 108 109 1010
Angular frequency, ω, rad/s

Figure 12: The system transfer function may be computed using the exact
impedances of the open and short-circuit stubs. We can also compute the
system transfer function by approximating tan(ωd/c) to be ωd/c for ω ≪ c/d.
|H(jω)| for the circuit in Example 5 has been shown, with and without the
approximate LC-model.

I(x) R′ ∆x jωL′ ∆x I(x + ∆x)

V(x) G′ ∆x V(x + ∆x)

jωC ′ ∆x

∆x

Figure 13: A section of a transmission line, of length ∆x. The inductance,

capacitance, resistance, and conductance per unit length are L′ , C ′ , R′ , and
G′ , respectively. The voltages and currents are now phasors that are functions
of the distance, x, denoted as V(x) and I(x) respectively.

19
 The Telegrapher’s equations: In the limit, as ∆x tends to 0, (25) and
(26) can be simplified.

dV(x)
= −(jωL′ + R′ )I(x) (27)
dx
dI(x)
= −(jωC ′ + G′ )V(x) (28)
dx
The equations (27) and (28) are the Telegrapher’s equations. They rep-
resent the phasor forms of (6) and (7).

The two equations (27) and (28) can be solved by taking a second deriva-
tive and decoupling them.

d2 V(x)
= (jωL′ + R′ )(jωC ′ + G′ )V(x) (29)
dx2
d2 I(x)
2
= (jωL′ + R′ )(jωC ′ + G′ )I(x) (30)
dx
The solution to both equations is as follows:

V(x) = V+ e−γx + V− eγx (31)

I(x) = I+ e−γx + I− eγx (32)

where γ 2 = (jωL′ + R′ )(jωC ′ + G′ ). V+ and V− are forward and back-

ward moving voltage phasors, respectively. Similarly, I+ and I− are forward
and backward moving current phasors, respectively. (31) and (32) are the
solutions of (29) and (30), respectively.
Notice that γ = α + jβ is generally complex. It is purely imaginary in
the case of a lossless line, where R′ and G′ are zero. The imaginary part of
γ will lead to a rotation in the wave as it moves through the transmission
line. The real part of γ will attenuate the wave as it moves through the
transmission line. Because of the loss in the transmission line, the attenuation
is understandable. Since we are working with phasors, if x is the wavelength,
λ, the rotation in the wave should be 2π. This means λ is 2π/β. In other
words, the velocity of the wavefront is ω/β.
The original Telegraphers’ equations (27) and (28) need to be solved.
Plugging our solutions in (31) and (32) back into the Telegraphers’ equations
of (27) and (28), we can reduce two of the four unknown phasors.

20
 V(x) = V+ e−γx + V− eγx
V+ −γx V− γx
and, I(x) = e − e (33)
Z0 Z0
s
jωL′ + R′
where Z0 = (34)
jωC ′ + G′

Z0 is known as the characteristic impedance and is generally a complex

quantity.
p When R′ and G′ are 0, as expected, Z0 reduces to a real resistance
of L/C in the case of a lossless transmission line. Likewise, γ is also a
complex quantity. In case√ of a lossless transmission line, γ reduces to the
wavenumber, i.e., jω L′ C ′ , and is ω/c where c is the velocity of light in
the transmission line system.
Finally, we may impose the boundary conditions at the source and at the
load to solve for the last two unknown phasors, i.e., V+ and V− . For a load
impedance of ZL (Fig. 3), the ratio of V+ and V− at the load can be worked
out and is identical to (17). Further, we can define the reflection coefficient
at any point along the transmission line as the ratio of the backward-moving
voltage phasor wave to the forward-moving voltage phasor wave.

V− eγL ZL − Z0
Γ(L) = −γL
= = ΓL (35)
V+ e ZL + Z0
V− eγx
Γ(x) = = ΓL e2γ(x−L) (36)
V+ e−γx
Transmission line analysis using phasors is used all the time in power
transmission and microwave circuits. This analysis is necessary when the
wavelength is comparable or smaller than the overall length of the cable.
The electrical length of a cable is the number of wavelengths in the length
of the cable. If the electrical length of the cable is anything more significant
than 0.1, one should model the distributed effects of the transmission line.
Transmission line effects come up in long power lines2 , in data cables such
as USB and SATA3 , on the back-plane of a computer motherboard, and in
microwave systems.
2
The free-space λ is roughly 6000 km at 50 Hz. In the transmission-line medium, λ
might reduce to 2000 km. 200 km long cables may have an electrical length ≥ 0.1λ.
3
In USB-3.1, the data rate is 10 Gbit/s. At the speed of light, each bit will use a length
of 3 cm (bit-length, as opposed to wavelength). 10 bits will be traversing a 0.3 m cable at
a time.

21
 Example 6. Consider a 50 Hz power transmission system where the
source is at 1 MV and has a source resistance of 100 Ω. The transmission
line is 200 km long, with a copper resistance of 0.05 Ω/km, an insulation
loss of 100 nS/km, a series inductance of 4 mH/km, a capacitance of
10 nF/km. Let us assume a load resistance of 10 kΩ. Let us find the
voltage at the source end of the line, the voltage at the load-end of the
line, the power delivered to the load, the power input to the transmission
line. Let us also plot the magnitude of the voltage along the length of
the transmission line. We have solved the numerical in Prog. 3. We
have plotted the magnitude of the voltage along the transmission line in
Fig. 14.

Program 3. powertxline.m: Script to solve example 6.

1 ZS = 100; ZL = 10 e3 ; VS = 1 e6 ; w = 100* pi ;
2 Rp = 0.05/1 e3 ; Gp = 100 e -12/1 e3 ; Lp = 4e -3/1 e3 ; Cp = 10 e -9/1 e3 ;
3 d = 200 e3 ;
4
5 Z0 = sqrt (( j * w * Lp + Rp )/( j * w * Cp + Gp )); % c h a r impedance
6 g = sqrt (( j * w * Lp + Rp )*( j * w * Cp + Gp )); % gamma
7
8 GL = ( ZL - Z0 )/( ZL + Z0 ); % r e f l e c t i o n c o e f f a t l o a d end
9 G0 = GL * exp( -2* g * d ); % r e f l e c t i o n c o e f f a t s o u r c e end
10 Zin = Z0 * (1+ G0 )/(1 - G0 ); % i n p u t impedance
11
12 V1 = VS * Zin /( Zin + ZS ); % v o l t a g e a t s o u r c e −end
13 I1 = VS / ( Zin + ZS ); % c u r r e n t a t s o u r c e −end
14 V1p = V1 /(1+ G0 ); % v o l t a g e + wave
15 V1m = V1p * G0 ; % v o l t a g e − wave
16 V2p = V1p * exp( - g * d ); % v o l t a g e + wave a t l o a d −end
17 V2m = V1m * exp( g * d ); % v o l t a g e − wave a t l o a d −end
18 V2 = V2p + V2m ; % V o l t a g e a t l o a d −end
19 I2 = ( V2p - V2m )/ Z0 ; % Current a t l o a d −end
20 P2 = V2 * conj ( I2 ); % Power d e l i v e r e d t o l o a d
21 P1 = V1 * conj ( I1 ); % Power i n p u t t o t r a n s −l i n e
22
23 x = [0:100: d ] ’; % p o i n t x a l o n g t h e t r a n s −l i n e , i n s t e p s o f 100m
24 V = V1p * exp( - g * x ) + V1m * exp( g * x ); % v o l t a g e as a f u n c t i o n o f x
25 plot (x , abs( V ));

11.1 Loading coils

Oliver Heaviside noticed that in (34), Z0 could be maintained as a real
impedance if L′ : R′ = C ′ : G′ . Unfortunately, in electrical cables, G′
is negligible, but R′ , C ′ , and L′ are measurable quantities. The negligible
G′ will cause Z0 to be complex and cause significant distortion to signals

22
 1.06

|V (x)| in MV
1.04

1.02

0 50 100 150 200

x in km

Figure 14: Voltage along the transmission line in example-6

L L

M M

L L

Figure 15: Extremely low G′ can be compensated for by increasing L′ . Pe-

riodically inserted mutual inductances increased the overall L′ . The loading
coil was an invention required for long-distance telephony.

traveling through the transmission line. The distortion caused long-distance

telephone cables to be impractical beyond a few kilometers (e.g. λ10 kHz is
30 km).
The idea that was used to implement long-distance telephone cables (even
trans-Atlantic cables) was to increase L′ to minimize the imaginary part of
Z0 . Mihajlo Pupin used loading coils [1] periodically to add discrete induc-
tance into the transmission line (Fig. 15). For every section of the transmis-
sion line, the inductance effectively increases by 2L + 2M , where L is the
value of each discrete inductor, and M is the mutual inductance with the
dots, as shown in the figure. [2] provides a highly entertaining history of the
technology.

12 Smith charts
The Smith chart is named after Philip Smith. The Smith chart is a graph-
ical calculator for solving problems that involve transmission lines. Philip
Smith was an engineer at the Bell Laboratories when he came up with the
Smith Chart in 1936 [3, 4, 5]. Later he started his own company, Analog
Instruments, which sold aircraft navigational instruments and radar-related

23
 j1
j0.5 j2

j0.2 j5

j0.1 j10
0.2
0.1 0.5 1 2 5 10
−j0.1 −j10

−j0.2 −j5

−j0.5 −j2
−j1
(a) (b)

Figure 16: The graphs are maps from the impedance plane to the reflection-
coefficient plane. (a) The impedance is a constant resistance in series with a
variable reactance. (b) The impedance is a constant reactance in series with
a variable resistor. We have normalized all marked resistances and reactances
with respect to Z0 , the characteristic impedance of the transmission line. The
point marked as 1, in (a), corresponds to a resistance Z0 and is a reflection
coefficient of 0.

items. At its heart, the Smith chart is nothing but a plot of the complex
reflection coefficient, Γ, at different locations on the transmission line. Intri-
cate complex-number calculations reduce to operations that require a scale
and a compass on the Smith chart.
At the outset, we need to appreciate that the function Γ = (ZL (s) −
Z0 )/(ZL (s) + Z0 ) is a conformal mapping4 in the complex plane that takes
us from the Z-plane to the Γ-plane. We can move back and forth between
impedance and reflection coefficient.
The graphs in Fig. 16 are plots of the reflection coefficient, when (a) the
impedance is R0 ± jX, and (b) the impedance is R + jX0 . All the plots
are arcs of circles. These contours are the circles of constant resistance,
in the lower Γ-plane the circles of constant capacitance, and, in the
upper Γ-plane the circles of constant inductance respectively. The big
circle is for |Γ| = 1. The point at the center of the big circle marks Γ = 0
and ZL = Z0 . Regions outside |Γ| = 1 are irrelevant because the reflection
coefficient cannot be greater than 1 for passive loads.
4
A conformal mapping is a 1-to-1 mapping from the complex plane to the complex
plane. f : C → C.

24
 0.12 0.13 0.14
0.11
0.1 0.15
0.38 0.37 0.36 0.1
0.0
9 0.39 6
0.4 90 80 0.35
100
8 0.3 0.1
0.0 1 70 7
0.4 110 4

1
0.9

1.2
0.3

0.8
2 0.1
7 0.4 120 + j X /Z0
60 3
0.0

1.4
8

0.7
E NT
PON
COM 0.3

1.6
3 CE
0.4 0

0.6
TA
N 2
50

1.8
AC

0.
06

13 RE

19
0.

2
IV A
0.5

0.
N
44

CT

31
DU
0.

G
L
IN
5

0.2
0.0

O
14

F
4
5

0.3
0.
0.4

R
E
F
3

LE

0.2
CT
→

1
0.2
6
150
0.4

0.3

IO
OR

9
N
AT

CO

0.22
ER

EF

0.28
0.47
EN

FI
5
S G

160

0.2

CI
ARD

EN

0.23
0.27
0.48
T OW

T
IN
TH S

10
170

D EGR
0.1

0.24
→ WA V E L E N G

0.26
0.49

EE S
20
50
0.1

0.2

0.3

0.4

0.5

0.6

0.7
0.8
0.9

1.2

1.4
1.6
1.8

10

20
50

0.25
0.25
180

5
0.5

0
50
RESISTIVE COMPONENT, R/Z0
20
←

0.24

0.26
L OAD
0.49

−170

−10
10
0.1
S
ARD

0.23

0.27
0.48

T OW
0

−20 2
−16

5
0.2
S

0.2
H
GT

0.28
0.47

4
EN L
VE

50

0
−3
.
−1

2
0.3
A

1
0
W

0.2
6
0.4

3
←

9
5

0.2
0.0
0

4
−4 0

0.
06 −14
5

0.3
0 .19
0.4

0
Z
X/
−j
0.

2
0.5

T
EN −5
1.8

0.

30
44

ON
31

−1 7 MP 0.1 0
0.

CO
1.6
0.6

0.0 CTA
NC
E 8
R EA
1.4

0.3
0.7

3 20 I VE
C A P AC I T 0.1 −60
0.4 −1 0.08 2
1.2
0.8

7
0.9

9 0 −70 0.1 0.3

2
0.4 0.0 −11 6 3
−100 −80 0.15
1 0.1 −90 0.3
0.4 0.11 0.14 4
0.12 0.13
0.35
0.4
0.39 0.36
0.38 0.37

Figure 17: A professional Smith chart.

Exercise 4. Prove that the maps of R0 + jX and R + jX0 onto the

Γ-plane are circles.

Fig. 17 is an example of a Smith chart that may be printed and used for
transmission-line problem-solving.

25
 L

50 Ω, d 500 − j20 Ω

Figure 18: Possible circuit configuration for maximum power transfer, as

discussed in example 7.

Example 7. A load of 500 − j20 Ω is driven by a source with a source

resistance of 50 Ω. The circuit in Fig. 18 shows a possible configuration
with the help of which maximum power may be delivered from the source
to the load. The transmission line and the capacitor are not power-
consuming elements. Let us try to find the required inductor and the
transmission line segment such that the input impedance to the circuit
is 50 Ω.
We will use a Smith chart to work this out.

1. First, we will normalize our load impedance to the characteristic

impedance of the transmission line. The normalized load is 10 −
j0.4. Then we locate the point 10 − j0.4 on the Smith chart; it
turns out it is very close to 10.
2. Now we will add a transmission line in series towards the source.
When we add a transmission line, the value of Γ rotates. Use the
marker for “Wavelengths towards generator” on the Smith chart,
and rotate our point till we hit the circle of constant resistance
passing through 1.
3. There are two possibilities for the length of the transmission line
to be used; the shorter takes us from 0.25λ to approximately 0.3λ
and, therefore, is of length 0.05λ.
4. Now, this point is marked out and read as 1 − j2.8. A series nor-
malized inductance of j2.8 will take us along the circle of constant
resistance to 1. We have shown the process in the Smith chart of
Fig. 19.
5. One can readily perform further numerical refinement. Finally, the
length of the transmission line required is 0.04842λ, and the value
of the inductor required is 50 × j2.83 = j141.5 Ω.

Let us also try to solve this with Octave. We will iterate over many
possible lengths of added transmission line, and select the length at which
the real part of the input impedance is just right.

26
 0.12 0.13 0.14
0.11
0.1 0.15
0.38 0.37 0.36 0.1
0.0
9 0.39 6
0.4 90 80 0.35
100
8 0.3 0.1
0.0 1 70 7
0.4 110 4

1
0.9

1.2
0.3
7 2
0.4 120 0.8 + j X /Z0
60 3 0.1
0.0

1.4
8
0.7

E NT
PON
COM 0.3

1.6
3 E
0.4 0
0.6

A NC 2
CT 50

1.8

0.
06

A
13 RE

19
0.

2
IV A
0.5

0.
N
44

CT

31
U
0.

G
ND

L
I
5

0.2
0.0

O
14

F
4
5

0.3
0.
0.4

R
E
F
3

LE

0.2
CT
→

1
0.2
6
150
0.4

0.3

IO
R

9
TO

N
4
RA

CO

0.22
1 + jX
NE

EF

0.28
0.47
E

FI
5
S G

160

0.2

CI
ARD

EN

0.23
0.27
0.48
T OW

T
IN
TH S

10
170

D EGR
0.1

0.24
→ WA V E L E N G

0.26
0.49

EE
20

S
50
0.1

0.2

0.3

0.4

0.5

0.6

0.7
0.8
0.9

1.2

1.4
1.6
1.8

10

20
50

0.25
0.25
180

5
0.5

0
10 − j0.2
50
RESISTIVE COMPONENT, R/Z0
20
←

0.24

0.26
L OAD
0.49

−170

−10
10
0.1
S
ARD

0.23

0.27
0.48

T OW
0

−20 2
−16

5
0.2
S

0.04842λ
0.2
H
GT

0.28
0.47

4
1 − jX
N
LE E

50

0.2
−3
AV

−1

0.3
1
0
W

0.2
6
0.4

3
←

−j2.83
5

0.2
0.0
0

4
−4 0

0.
06 −14
5

0.3
0 .19
0.4

0
Z
X/
−j
0.

2
0.5

T
EN −5
1.8

0.

30
44

ON
31

−1 7 MP 0.1 0
0.

CO
1.6
0.6

0.0 CTA
NC
E 8
R EA
1.4

0.3
0.7

3 20 I VE
C A P AC I T 0.1 −60
0.4 −1 0.08 2
1.2
0.8

7
0.9

9 0 −70 0.1 0.3

0.4
2 0.0 −11 6 3
−100 −80 0.15
1 0.1 −90 0.3
0.4 0.11 0.14 4
0.12 0.13
0.35
0.4
0.39 0.36
0.38 0.37

Figure 19: Example 7 worked out with the help of a Smith chart. One can
readily obtain approximate values graphically. If required, one may further
refine these approximate values.

27
 Listing 4. Octave work in lieu of the Smith chart
octave:1> Z0 = 50; ZL = 500-j*20;
octave:2> GL = (ZL-Z0)/(ZL+Z0);
octave:3> l = 0:0.00001:0.5;
octave:4> Gin = GL*exp(-j*4*pi*l);
octave:5> Zin = (1+Gin)./(1-Gin);
octave:6> index = min(find((real(Zin)<1.0002) & (real(Zin)>0.9998)));
octave:7> length = l(index)
length = 0.048070
octave:8> real(Zin(index))
ans = 0.99988
octave:9> imag(Zin(index))
ans = -2.8487
octave:10> X = -imag(Zin(index)) * Z0
X = 142.43

In the Octave code we considered all possible lengths from 0 to 0.5λ in

steps. We computed the input impedance for all these transmission line
segments. Then we selected the length at which this input impedance
has a real part as close to 1 as possible. The find function is handy for
this search. Then we obtained the required transmission line length, the
input impedance, and the reactance that is to be added in series. It will
also be instructive to plot the real(Zin) for all l. There are two lengths
at which the real part of the impedance crosses 1, namely at 0.04807λ
and at 0.45065λ. This is the other point marked in the Smith chart of
Fig. 19 as 1 + jX.

Smith-chart-based problem solving is quite common in the domain of

microwave circuits. We presented this section to give a flavor of the Smith
chart. The Smith chart is a handy tool for solving transmission-line circuit
problems graphically. The answers from the Smith chart are approximate but
very close to the exact answer. The example 7 requires elaborate calculations
to obtain the solution using Octave, whereas it can be worked out graphically
in an elegant manner.

13 Important concepts
• The transmission line is a distributed circuit with an infinite cascade
of R-L-C-G sections. Analysis of a single section leads us to a pair of
coupled partial differential equations in current and voltage over space
and time. These are called the Telegrapher’s equations.

28
• The Telegrapher’s equations can be differentiated once more to arrive
at the wave equation. The wave equation has a simple mathematical
solution as the superposition of a forward-moving and a backward-
moving wave. The wave function can be any arbitrary function, as
long as it satisfies
√ the boundary conditions. The wave moves at the
velocity of 1/ L C ′ , which also happens to be the velocity of light in
′

the medium.

• The forward and backward moving waves are related to each other
through
p the reflection coefficient, Γ. The characteristic impedance is
Z0 = L′ /C ′ .

• The two boundaries are at the source and the load. The boundary
conditions constrain Γ to be (ZL − Z0 )/(ZL + Z0 ) at the load. At
the source, only if we are interested in a time-domain analysis, the
reflection coefficient is (ZS − Z0 )/(ZS + Z0 ). If we are considering
sinusoidal steady-state analysis instead, the reflection coefficient is not
so.

• We can compute the reflection coefficient at each location on the trans-

mission line for steady-state sinusoidal stimulation. When the forward-
moving wave or backward-moving wave moves by d, it gets multiplied
by e−j2πd/λ (or delayed). Overall, the reflection coefficient changes by
e−j4πd/λ .

• The input impedance of a quarter wavelength line is Z02 /ZL . A quarter

wavelength line is an impedance inverter. A capacitive load is converted
to an inductive impedance and vice-versa.

• For a lossy transmission line, Z0 = (jωL′ + R′ )/(jωC ′q

p
+ G′ ). The
jωL′ +R′
reflection coefficient also needs to be re-evaluated as Γ = jωC ′ +G′ .

• A Smith chart is a plot of the reflection coefficient on the complex

plane. It is often used to design matching networks for maximum power
transfer. The plots of R0 + jX (constant resistance) and R + jX0
(constant reactance) map to circles in the Γ-plane. All straight lines
in the impedance plane map to either circles or straight lines in the
reflection-coefficient plane. The mapping of straight lines to circles is
known as conformal mapping.

29
 12.5 nH
i(t) Z0 = 50 Ω

1 cos(109 t) 40 pF 200 Ω

λ/4

Figure 20: Transmission line circuit for exercise 4.

14 Exercises
1. A transmission line of characteristic impedance 200 Ω is of length such that it
takes a signal 1 ms to traverse from one end to the other. The transmission
line is terminated at the load-end in an open circuit. A voltage source,
v(t) = u(t), with a source impedance of 50 Ω is applied at the input end.
Draw a graph of the voltage at the load-end as a function of time, from
t = −1 ms to t = 10 ms.

2. A transmission line of characteristic impedance 50 Ω is 1 meter long. The

velocity of a wave traveling through the transmission line is 3 × 108 meters
per second. A step input, u(t), drives the transmission line through a source
impedance of 25 Ω. The transmission line drives a load of 100 Ω. Find the
time the output voltage takes to settle to 90% of the final value.

3. Consider an HVDC lossless transmission line of length 1 km. The veloc-

ity of a wave through this transmission line is 1 × 108 meters per second.
The characteristic impedance of the line is 2 Ω. A DC generator of source
impedance 0.01 Ω is supplying power through this line at a voltage of 1 kV.
The load is a 1-Ω resistor. At time t = 0, the customer, by mistake, applies
a short-circuit at the load. Deduce the voltage waveform at the middle of
the transmission line, at the load end, at the source end, as a function of
time.

4. In the circuit of Fig. 20, the transmission line is of length λ/4 at the given
operating frequency. Find the current i(t).

5. Consider a two-port network with just a lossless transmission line of length

λ/4 and characteristic impedance, Z0 = 50 Ω, as shown in Fig. 21. Work out
the Z-parameters for this network. Start with Z11 and Z22 . Next, compute
Z21 and Z12 by applying a current at P1 and estimating the voltage at P2 ,
when P2 is an open circuit.

6. Extend the question of exercise 5 to a lossless transmission line of arbitrary

length. Find the Z-parameters for a transmission line of arbitrary length.

30
 P1 Z0 = 50 Ω P2

λ/4

Figure 21: Two-port network for exercise 5.

L′ ∆x/2 L′ ∆x/2 L′ ∆x

C ′ ∆x C ′ ∆x/2 C ′ ∆x/2

(a) (b)

Figure 22: Other possible transmission line sections.

Further, obtain the transmission matrix (T-matrix) from the Z-parameters.

The T-matrix for a lossless transmission line should work out to:
 
cos(2πd/λ) jZ0 sin(2πd/λ)
T=
j sin(2πd/λ)/Z0 cos(2πd/λ)

7. Show that even if we start from any one of the transmission line sections
shown in Fig. 22(a), (b), we will reach the same Telegrapher’s equations, (6)
and (7).

8. A 50 Hz power transmission line has a length of 100 km and has L′ =

10 µH/m, C ′ = 10 pF/m, R′ = 100 µΩ/m, G′ = 0. The input is 1 MV with
a source resistance of 10 Ω. The load is 10 kΩ. Find the magnitude of the
voltage at the load. Find the power delivered to the load. Find the power
(active and reactive) that enters the transmission line at the source end.

9. Consider a cascade of three lossless transmission line sections with charac-

teristic impedances of 20 Ω, 120 Ω, and 20 Ω, and lengths of 2 mm, 6 mm,
and 2 mm, respectively. Assume the wave velocity is 3 × 108 meters per
second through all the transmission lines. The circuit is drawn in Fig. 23.
Find |VL (jω)/VS (jω)| as a function of ω. ω/(2π) should be in the range of
0.1-10 GHz.

10. Use a Smith chart or any other technique to match a (40 + j80) Ω load
to a 50 Ω source impedance for maximum power transfer. With the Smith
chart, first, plot the load impedance. Second, using a compass, draw a circle

31
 50 Ω Z0 = 20 Ω Z0 = 120 Ω Z0 = 20 Ω

VS (jω) 50 Ω VL

2 mm 6 mm 2 mm

Figure 23: Circuit schematic for exercise 9.

through the point with the center at the origin to cut the circle going through
the origin. The earlier step will give the length of the transmission line to
be added in series. Next, traverse to the center along the circle of constant
resistance. The last step will give the reactance to be added in series.

11. A lossless transmission line of length 0.7λ and characteristic impedance 1 Ω

is connected between a load of (1.5 + j0.4) Ω and a source with source-
impedance of 0.8 Ω. Plot the magnitude of the voltage along the transmission
line. Find the ratio of the maximum to the minimum voltage magnitude
along the transmission line. Prove that the VSWR is (1 + |ΓL |)/(1 − |ΓL |)
for a lossless line.

32
References
[1] M. Pupin, “Art of reducing attenuation of electrical waves and apparatus
therefore,” Jun. 1900, united States Patent No. 0 652 230.

[2] J. E. Brittain, “The introduction of the loading coil: George A. Campbell

and Michael I. Pupin,” Technology and Culture, Society for the History
of Technology, vol. 11, no. 1, pp. 36–57, Jan. 1970.

[3] P. H. Smith, “Transmission line calculator,” Electronics, vol. 12, pp. 29–
31, Jan. 1939.

[4] ——, “An improved transmission line calculator,” Electronics, vol. 17,
pp. 130–133, Jan. 1944.

[5] ——, Electronic applications of the Smith Chart in waveguide, circuit and
component analysis. Raleigh, NC, USA: SciTech Publishing, 1969.

33