How to simulate a microstrip model losslessly in IE3D?

[flywke]

Hi everyone
I have a question in simulating a two-port microstrip network in IE3D.
The lossless simulation should be done because the zero real parts of the Y-parameters of the ports are needed. Now I want to know which parameters should be set in the IE3D to obtain the result that I need. I have modified the parameter of conductivity of the metal and ground but there is no evident influence at all.

Thanks in advance!
flywke

 

[jian]

Hi, Flywke:

Yes. Setting sigma to a very large value normally should remove the conductor loss. However, you may not be able to get Re(Y11) = 0 because there is normally some radiation loss.

Bes tregards,

 

[flywke]

Hi, Flywke:

Yes. Setting sigma to a very large value normally should remove the conductor loss. However, you may not be able to get Re(Y11) = 0 because there is normally some radiation loss.

Bes tregards,

Thanks for ur suggestion and Could you tell me which EM simulator can get the zero real part of characteristic impedance/admittance.

flywke

 

[jian]

Hi, Fywke:

I am not sure whether you can get exact Im(Yc) = 0 for any numerical code. You will see Im(Yc) != 0 even you set all the loss related parameters to be lossless.

Best regards,

 

[flywke]

Thank u jian.
Now all I can do is to give up. But some design example utilizing this and It is a puzzle for me.

flywke

 

[jian]

Hi, flywke:

It is possible you can get Im(Yc) very small. However, it is impossible to get it to be exactly 0 numerically.

Best regards,

 

[rautio]

Hi Flywke -- In shielded EM tools, it is possible to get Im(Zo) = 0, and to get exactly lossless results. Two tools are AWR Emsight, and Sonnet (I work for Sonnet). For the lossless case, these tools (at least Sonnet, I am not sure about Emsight) use only the imaginary part of the MoM matrix. The real part is zero. This is exactly like the Y or Z parameter matrix for a network that contains only capacitors and inductors, no resistors. This is nice because only half the memory is used for the lossless case. There is no need to use up memory for the real (lossy) part of the matrix because it is all exactly zero. However, keep in mind that an exactly lossless circuit can not be built.

 

[flywke]

Thank u Rautio for your suggestion. I have no experience in using sonnet, but I will have a try on it.
flywke