FDTD MODELING OF TRANSMISSION LINES

Wave equation
To understand the general idea of integrating the differential equations in the time
domain, we start with the wave equation (for concreteness, written for the voltage wave)

 2V  z, t   2V  z, t 
 v2  z  . (1)
dt 2 dz 2
First, we expand the solution of this equation into a Taylor's series at point zk at the moment

tn :

V  zk , tn   z   2V  zk , tn 
2
V  zk  z, tn   V  zk , tn   z     , (2)
dz 2 dz 2

V  zk , tn   z   2V  zk , tn 
2
V  zk  z, tn   V  zk , tn   z     . (3)
dz 2 dz 2
Second, we approximate the sum by the second-order accurate formula

 2V  zk , tn 
V  zk  z, tn   V  zk  z, tn   2V  zk , tn    z  
2
 . (4)
dz 2
Rearranging the terms in the latter expression, we have for the second time derivative

 2V  zk , tn  1
 V  zk  z, tn   2V  zk , tn   V  zk  z , tn  . (5)
dz 2
 z 2
Now, introduce a short notation: a superscript will denote the time moment, and a subscript
will mean the position along the z-axis. Thus,

 2V  zk , tn 
dz 2

1
 z  2 Vkn1  2Vkn  Vkn1  . (6)

Analogously, by using a similar procedure we obtain

 2V  zk , tn 
dt 2

1
 t  2 Vkn1  2Vkn  Vkn1  . (7)

Substituting Eqs. (6) and (7) into the wave equation (1), we arrive at the final solution in the
form of an update equation


Vkn 1   vk t z  Vkn1  2Vkn  Vkn1  2Vkn  Vkn 1 .
2
 (8)

Step of the spatial grid, z , should satisfy the condition z   20 . To ensure stability, the
time step t should satisfy the Courant–Friedrichs–Lewy (CFL) condition,
 2

t  z vmax , (9)

where vmax is the maximum propagation velocity.

For homogeneous line, and the so-called magic time step t  z v , we arrive at

Vkn 1  Vkn1  Vkn 1  Vkn1 . (10)

Vkn 1
n 1

n
Vkn1 Vkn1
n 1
Vkn 1

k 1 k k 1 z
Fig. 1. Leapfrog integration of the wave equation.

Telegraph equations
In most cases, a better strategy is to integrate in time the original coupled first order
telegraph equations. Rewrite the telegraph equations in the form
V  z, t  1 I  z, t  I  z, t  1 V  z, t 
 ,  . (11)
t C  z  z t L  z  z

These equations are supplemented with the initial values of both voltage and current
distributed along the line. A numerical algorithm that solves these coupled equations must
involve a discretization procedure for both time and space derivatives. We use the Yee’s
scheme proposed in 1966 and adapted here to this 1D problem.
Let us start with the second of this set of equations. Replacing the derivatives by central
differences at point z k at the moment t n , we get

I  zk , tn  t 2   I  zk , tn  t 2  1 V  zk  z 2, tn   V  zk  z 2, tn 
 . (12)
t L  zk  z

Rearranging the terms, we arrive at

Using the short notation, we rewrite the latter equation in a more compact form

I kn 1 2  I kn 1 2 
t
Lk z

Vkn1 2  Vkn1 2 .  (14)

Analogously, the discretization of the first equation leads to

Vkn112  Vkn1 2 
t
Ck z

I kn 1 2  I kn11 2 .  (15)

The two latter expressions are the update equations for the voltage and current waves. These
equations can be rewritten in terms of the secondary parameters,

I kn 1 2  I kn 1 2  Yk
vk t n
z

Vk 1 2  Vkn1 2 ,  (16)

Vkn112  Vkn1 2  Z k
vk t n 1 2
z
Ik  I kn11 2 ,  (17)

where vk is the phase velocity evaluated at point z k , and Z k ( Yk ) is the characteristic

t
Vkn112
n 1
I kn 1 2
n 1 2

n
Vkn1 2 Vkn1 2
n 1 2
I kn 1 2
n 1
z
k 1 2
k 1 2
k 1

k 1
k

Fig. 2. Leapfrog integration of the telegraph equations with staggered grid.

For a homogeneous line, we can use the magic time step, which simplifies the update
equations:


I kn 1 2  I kn 1 2  Y0 Vkn1 2  Vkn1 2 ,  (18)

Vkn112  Vkn1 2  Z0  I kn 1 2  I kn11 2  . (19)