You are on page 1of 11

EEE G581

RF & Microwave Engineering


Dr. Praveen Kumar A.V.
BITS Pilani EEE Department
BITS Pilani, Pilani
Pilani Campus
FDTD

BITS Pilani
Pilani Campus
Finite Difference Method (FDM)
Given a geometry with certain boundary conditions, where a differential
equation needs to be evaluated using FDM

It involves three steps

(1)Divide the solution region into a grid of nodes

(2)Approximate the given differential equation by finite difference


equivalent (Taylor expansion method)

(3)Solve the difference equations subject to the prescribed boundary


conditions.

BITS Pilani, Pilani Campus


FDM
1 2 1
F ( x0  x)  F ( x0 )  xF ( x0 )  x F ( x0 )  x 3 F ( x0 )  ........
2! 3!
1 2 1 3
 
F ( x0  x)  F ( x0 )  xF ( x0 )  x F ( x0 )  x F ( x0 )  ........
2! 3!
F ( x0  x)  F ( x0  x) 1
 F ( x0 )  x 2 F ( x0 )  ........
2x 3!

dF ( x) F ( x0  x)  F ( x0  x)
F ( x0 )    O(x 2 )
dx x  x0 2x

BITS Pilani, Pilani Campus


FDM

• when x0
dF ( x) F ( x  x)  F ( x  x)
 or
dx 2x
dF ( x) F ( x  x / 2)  F ( x  x / 2)

dx x

• This is called the central difference scheme, having second order


accuracy

BITS Pilani, Pilani Campus


Partial differential equations using FDM
Partial differential equations with 1st order variations in space and time

F ( x, t )  F ( x  x, t )  F ( x  x, t ) 

x 2x

F ( x, t )  F ( x, t  t )  F ( x, t  t ) 

t 2t

BITS Pilani, Pilani Campus


Space and Time indexing
• x=ix so that x+x i+1, xx i1
• similarly t±t n±1
• Also the temporal index is superscripted

F ( x, t )  F ( x  x, t )  F ( x  x, t ) 

x 2x
 F (i  1, n)  F (i  1, n)   F (i  1)  F (i  1) 
n n

 
2x 2x

F ( x, t )  F ( x, t  t )  F ( x, t  t ) 

t 2t
 F (i, n  1)  F (i, n  1)   F (i )  F (i ) 
n 1 n 1

 
2t 2t

BITS Pilani, Pilani Campus


FDTD method
• Formulation of the FDTD method begins by considering the differential form of
Maxwell's curl equations which govern the propagation of electric and magnetic
fields in a medium.
• For a simple, lossless medium H
xE    (1)
t
E
xH   (2)
t

• The geometry is sampled into cells and the above equations are evaluated at these
discrete cells (index i, j, k) at discrete intervals of time (index n)  time stepping

• Eq (1) is used to compute H-field


• Then Eq(2) is used to compute E-field using previously computed H-field data

• The time stepping is continued until a steady state is reached

BITS Pilani, Pilani Campus


Maxwell’s curl equations
H E
 xE    xH  
t t

H x 1  E y Ez  Ex 1  H z H y 
       
t   z y  t   y z 
H y 1  Ez Ex  E y 1  H x H z
       
t   x z  t   z x 
H z 1  Ex E y  Ez 1  H y H x 
       
t   y x  t   x y 

BITS Pilani, Pilani Campus


Sampling and Indexing

Er ( x, y, z, t )  Er (i, j , k , n)  Ern (i, j, k )


r  x, y or z

Eg :
n  0.5
Ex ( x  5x, y  2y, z, t  0.5t )  E x (i  5, j  2, k )

BITS Pilani, Pilani Campus


Finite Difference Form

Ex 1  H z H y 
   
t   y z 

Exn 1 (i, j , k )  Exn 1 (i, j , k ) 1  H zn 1 (i, j  1, k )  H zn 1 (i, j  1, k )   H y (i, j , k  1)  H y (i, j , k  1)  


n 1 n 1

      
2t   2y   2 z  
t t
Exn 1 (i, j , k )  Exn 1 (i, j , k ) 
y
 H zn 1 (i, j  1, k )  H zn 1 (i, j  1, k )  
z
 H yn 1 (i, j , k  1)  H yn 1 (i, j , k  1) 

BITS Pilani, Pilani Campus

You might also like