Question about S-parameter analysis

[tjuby]

Hi, I just have somewhat of a quick question about simulating with S-parameters.

Say you have a cascade of 2-port devices and you do a S-parameter analysis, perhaps with CST or matlab or any other program.

What I have noticed is that you seldom get S-parameters for all devices that lies within the same frequency range, (typically one ranges from ie 1 - 5 GHz, while another goes from 2 - 4 Ghz) moreover they may not have the same step in frequency between measures.( ie, one component has data values (dB and phase) at 1GHz, 1.2 Ghz, 1.4 Ghz etc , while another component has data values at 0.9Ghz, 1.1Ghz, 1.3Ghz etc.)

So my question is, does this in any way impede the analysis of the simulation? Or will most simulators deal with this well enough?

 

[acrislg]

Hi
I used CST to make this kind of analysis and this program allows to use different S-parameter file formats.

Only, be sure that the simulation frequency range is within both files.

Bye

 

[tjuby]

Yup, I too have used CST the same way.

Really I just wondered if it would be preferable to have the same frequency steps in all S2P-files. It does seem it doesn't matter, but it could be that since the simulator must estimate to get data at correct points, some (probably minor) error will be introduced.. I don't know, so I just though I'd see if someone does, cause I never thought about it before

Thanks for replying!

 

[volker_muehlhaus]

Really I just wondered if it would be preferable to have the same frequency steps in all S2P-files. It does seem it doesn't matter, but it could be that since the simulator must estimate to get data at correct points, some (probably minor) error will be introduced..

The simulators will do data interpolation, so the frequency spacing in the data files does not really matter if you have "enough" data points. In some tools, you can choose the interpolation method (linear, spline) and the interpolation domain (rectangular, polar, dB).

Best regards
Volker

 

[tjuby]

Thanks Volker

I guess they are advanced enough to make pretty good estimates. Just got this idea in my head that maybe it could cause problems. But as you say, it would probably only be the case if there are few samples.