How to calculate the input impedance from FDTD simulation?
I have seen that the input impedance may be calculated by the fourier transform of the line voltage and the line current and then dividing both inputs. But how can this be done? At what time step shoud I stop the simulation to calculate the voltage and the current???
I have a little exp with simulation of microstrip by fdtd. The voltage is the integral of E along a specified path, from the trace to the ground. The current is the integral of H around the trace. The time you stop simulate is the E and H of your concerned point converged to zero.
Thanx for the advice...
Concerning the E, i just have to calculate it taking values from the ground to the strip on a single line.
But concerning the H, do I need to calculate it taking values surrounding that line at each height, that is, from the ground to the strip? Or do I calculate is from values surrounding the line at a particular height in between the ground and the strip?
Then, to find the input impedance, I just divide the voltage by the current? Or do I need to fourrier transform before?
Of course, fourier transform should be done to the voltage and current, because impedance is a frequency-domain parameter.
To calculate current, use the equation
I=integral along a close path (Hdl)
set a close path that include the trace(in the cross secton view). apply the above equation to obtain the current in the trace.
Ok... but the integration can be aproximated to the sum of the fields etc. Then I obtain a value for the voltage and a value for the current!
So when do I use the fourier transform? and How?
Thanking you in advance!
The voltage and current you mentioned are the values at one time spot. Calculate V and I at every time on your concerned place.
Now you've got the voltages and currents at every time steps. i.e. you got v(t) and i(t), where t=n*deltaT. deltaT is time step. n=0,1,2,...
Then V(f)=FFT(v(t)) and I(f)=FFT(i(t)). You can add zero data after v(t) converged to zero in order to improve the resolution in frequency domain.
Finally Z(f)=V(f)/I(f)
Oh thanks...
I'll try it!!!
Hi...
I've tried it... but the values obtained is not as expected!
I think the problem is with my gaussian source which is:
Ey(i,j,k) = ( cos (2 * pi * f * (N*dT - t ) ) ) * exp( -( (N*dT - t )^2 ) / 2*T^2)
where f is the frequency
N is the number of time step
dT is the time step
T controls the pulse width
t (I don't exactly know what is t but I've usd it as 100*dT)
My values are
f = 2.0e+9
N = 50
dT = 3.85e-12
T = 0.001
t = 100*dT
Are these values ok to excite my antenna which dimensions are 22x14x8mm (length x width x height)?
Thank you!
dt is constrained by the spacial grids and satisfied with stable condition.
t is the time when gauss pulse reaches to the peak. The greater of t, the more flat of exciting fields from the start time. I mean, the E(y) is start from zero,not suddenly reach a high value.
You can try to simulate the structure to determine N. Keep simulating until the fields you concern converge to zero.
Thanks...
But what about 'T'??? How does it control the pulse width?
Concerning the equation
Ey(i,j,k) = ( cos (2 * pi * f * (N*dT - t ) ) ) * exp( -( (N*dT - t )^2 ) / 2*T^2)
what values should i j and k take???
the smaller the T, the narrower the pulse width.
you can plot some curves with different T.
because a smaller T,the impulse rise time become more short.so when transform into spectral domain,its 3dB bandwidth become more wide.you could read some paper refer to FDTD to find the relation between T and BW.