how to calculate transmission and reflection spectra

[qestchen]

how to calculate reflection

The structure is air-glass-air.
The incident plane wave, Ei(t) and Hi(t), is from the left side.
And the transmission wave, Et(t) and Ht(t), is on the right side.
Now I get E(t) and H(t) from FDTD, and I take descrete fourier transform to get E(w) and H(w).
However, I find Et(w)/Ei(w) is a function of measured position.
Et(w) has different ampitude at different points.
Please help me.

 

[iyami]

how to calculate transmission intensity

You could make answering your question easier by answering a few questions first:

1) Is the incoming planewave normal to the glass air interface
2) how do you insert the planewave
3) what ABC's are you using
4) how much do the results differ
5) what type of source do you use (pulse, sine wave, sine wave + gaussian envelop,...)
6) what meshsize do you use
7) what wavelength do you use
8) are the two glass/air interfaces parallel
9) Do you use 1d/2d/3d FDTD

 

[qestchen]

fdtd transmission spectra

1) Is the incoming planewave normal to the glass air interface
I use 1d FDTD.
2) how do you insert the planewave
soft source. E[x]=E[x]+...
3) what ABC's are you using
I have checked that the absorption in PML is very good.
4) how much do the results differ
I upload the picture, Re(Et(w)) vs wavelength in different three positions.
5) what type of source do you use (pulse, sine wave, sine wave + gaussian envelop,...)
sine wave + gaussian envelop
6) what meshsize do you use
2.1nm
7) what wavelength do you use
the sin wave is about 600nm and I want to plot spectra from 380-700nm.
8 ) are the two glass/air interfaces parallel
yes
9) Do you use 1d/2d/3d FDTD
1d FDTD.

Should I calculate the transmission spectra by abs(Et(w))/abs(Ei(w))?
But why? thanks a lot!

 

[iyami]

how to calculate reflectivity

ABC:
====
PML is fine and a good idea if you also want to go to higher dimensions.
But if 1d-fdtd is all your interested in than the 'perfect' absorbing boundary
condition is easier to implement (it only works for the time step given by
the Courant condition)

I also use the 'big enough enough' absorbing boundary condition for 1d-fdtd.
Just make the model big enough so that no reflection from the PEC boundaries
could have possibly reached the measurement points when you stop you simulation.
It gives 0 reflection for all choices of dt in return for some increase in the model size.

GRID:
=====
should not be a problem

SOURCE
=========
1) If you use a sine wave (600nm with Gaussian envelope) then you should only compute
the transmission coefficient for that wavelength
2) a gaussian pulse may be more efficient
3) you should keep the source far enough away from the glass air interface that you can
distinguish the inserted signal from the reflected signal.

RESULTS
==========
Once you have measured the signals I (inserted), T (transmitted) and R (reflected)
you should computed there spectrum using DFT or FFT for the frequencies you are interested in
before you compute the transmission coefficient T_f/I_f for a frequency f.

Finally, the transmission coefficient can refer to either the intensity or amplitude of the waves.
Depending on which you need you have to compute different quotients

 

[sainathmunavalli]

hi... i am working onthe similar id fdtd problem...i have a gaussian pulse originating in air on left side...it hits the dielectric and prt of it gets transmitted and part of it gets reflected, since i have used ABC, at end of 300 time iterations i just have plane line indicating end... i need to calculate the transmission n reflection coefficient, kindly help me how to measure incident wave, transmitted wave and reflectedd wave parameters.???
help plz