Spectre S-parameter based modeling- Stability issues

[madhu]

Hi All,

I am using n-port component in spectre to perform time domain
simulation using touchstone s-parameter files for power planes which
are obtained 3D full wave tools. In many cases, the simulation explodes
due to stabiltiy reasons. What are the possible causes of this and how
can one solve this issue. Does Spectre n-port component use convolution
methods to perform time domain simulation from frequency domain
s-parameter data? if so how does it enforce passivity in the model. Is
it possible to enforce passivity for the data and then try in spectre
again. Please share your ideas and views on this. Thanks in advance.


Madhu

 

[Andrew Beckett]

On 28 Nov 2006 20:59:54 -0800, "madhu" <Madhu.Sampath@gmail.com> wrote:

Hi All,

I am using n-port component in spectre to perform time domain
simulation using touchstone s-parameter files for power planes which
are obtained 3D full wave tools. In many cases, the simulation explodes
due to stabiltiy reasons. What are the possible causes of this and how
can one solve this issue. Does Spectre n-port component use convolution
methods to perform time domain simulation from frequency domain
s-parameter data? if so how does it enforce passivity in the model. Is
it possible to enforce passivity for the data and then try in spectre
again. Please share your ideas and views on this. Thanks in advance.


Madhu

There is an application note on sourcelink (I think if you search for
"nport app note" or "nport application note" you should find it) which
covers the various settings in more detail.

Most problems are caused by insufficient data in the s-parameter file. You need
to go to sufficiently low frequency (the s-parameters want to be smooth if you
extrapolate them), and also high frequency - it is silly simulating at (say)
6GHz if your s-parameters are only up to 6GHz, as there is no information
as to what happens to the harmonics. Extrapolation might be non-physical
and hence non-causal.

You also need to ensure that there are sufficient points to give a smooth
response within the s-parameter range.

When using interpolation method of linear or spline, it uses a convolution
based approach; with rational, it uses a rational (lumped) model based
approach - usually convolution is quick and more accurate - note I would
nearly always recommend setting usewindow=yes, BTW.

But more is in the app note.

Regards,

Andrew.
--
Andrew Beckett
Principal European Technology Leader
Cadence Design Systems, UK.