Using the results of FEM or MoM to obtain time domain

[adel_48]

Hello All.

I am trying to obtain time domain response of two antennas. I am sending one pule from one of them and receiving it from the other antenna. Using time domain simulation gives acceptable results (CST Microwave Studio). However, I was exploring using frequency domain simulations as well (HFSS, IE3D). The problem is that I can obtain the frequency domain S21, but to obtain the received pulse, I have to extrapolate the results towards 0 and inf. Using a simple rectangular windows (i.e. setting S21 to zero around the simulation band) gives imaginary parts in the resulting time domain signal. I am not sure how to deal with them (use the absolute, use only the real part .. ). Is there other types of windows that does not generate imaginary parts, and does not cause distortion in the received pulse ?

thanks in Advance,
Adel

 

[jian]

Hi, Adel_48:

On IE3D, you can do the following: (1) Simulate your structure for the s-parameters from very low to high enough frequency with dense enough frequency points; (2) Use MDSPICE (another simulator of ZELAND) to do the time transient analysis directly from s-parameters. If you are solving the Tx and Rx antenna problems, you will have 4 steps: (1) You can find the Tx antenna's pattern; (2) Find the Rx antenna's pattern with planewave excitation; (3) From PATTERNVIEW of IE3D, you can find the 2-port s-parameters from Tx and Rx patterns; (4) You put the s-parameters into MDSPICE for time-transient analysis.

Best regards,

 

[Wiley]

It sounds like you are applying the Fourier transform incorrectly. For a signal to be completely real, the real part of it's Fourier transform must be an even function and the imaginary part must be an odd function. Typically one forces the S parameters at negative frequencies to be the complex conjugate of those at positive frequencies; this forces the time domain S parameters to be completely real.

Applying a rectangular window will add ripple into the time domain response, but should not create an imaginary part.

 

[adel_48]

Hi Mr. Jian,

This way I never knew about it to obtain time domain data

I really like IE3D, it has a lot of things that make doing tasks much easier, and it also gives excellent accuracy/time.

May be my second problem with IE3D (my first was simulation of finite substrates, and it is solved in the newer versions) is that it is based on MoM using multi-layer green's functions. I mean it is still a 2.5D simulator, not a full 3D one. This gives it great advantage when it comes to speed and accuracy when simulating planar structures. But in a lot of cases, 3D structures is also required. Aren't you trying to implement surface equivalence method in IE3D, like FEKO and WIPL-D ?

Also, I have another question about IE3D. If you heard about WIPL-D, they use higher order basis functions employing a much larger mesh (according to the site, it can use mesh size of up to 2 times the wavelength). Why didn't you try to implement it in IE3D ?

Hello Wiley,

Thank you very much. This is a very clever idea though it is very simple. It looks like I am forgetting what I learned

thank you both and best regards,
Adel

 

[jian]

Hi, Adel:

Thank you very much.

Certainly, we are still developing it. There will be more capabilities implemented into IE3D. Thank you very much for your suggestions.

Best regards,