The Designer's Guide Community Forum

The Designer's Guide Community Forum
https://designers-guide.org/forum/YaBB.pl
Simulators >> RF Simulators >> hidden state in rfLib and SpectreRF
https://designers-guide.org/forum/YaBB.pl?num=1053365004

Message started by Geoffrey_Coram on May 19^th, 2003, 10:23am

Title: hidden state in rfLib and SpectreRF
Post by Geoffrey_Coram on May 19^th, 2003, 10:23am

Does anyone else find it ironic that the "rfLib" distributed by Cadence has elements with hidden state such that they cannot be used in a PSS analysis in Cadence's own SpectreRF?

Title: Re: hidden state in rfLib and SpectreRF
Post by Jess_Chen on May 19^th, 2003, 4:42pm

It does indeed seem strange that some models in rfLib have hidden state and therefore do not work with SpectreRF but there is an explanation. The only elements in rfLib with hidden state are the instrumentation blocks. The instrumentation blocks generate common baseband signals and compute statistics on baseband signals. Those blocks were designed for use with the baseband equivalent models also found in rfLib. The baseband equivalent models accelerate run times, at the cost of some accuracy of course, to the point where eye-diagram and evm simulations have practical run times. Many of the models have baseband and passband views. The idea was that you could use the faster baseband view with the instrumentation blocks in Spectre to study RF effects on baseband metrics then switch to the passband view and SpectreRF to compute the associated RF specifications.

-Jess Chen

Title: Re: hidden state in rfLib and SpectreRF
Post by Geoffrey_Coram on May 27^th, 2003, 2:29pm

What about trying to use the baseband signal generators to modulate the carrier in an Envelope-Following analysis?

Envelope-Following has the same hidden-state restriction.

Title: Re: hidden state in rfLib and SpectreRF
Post by Andrew Beckett on May 27^th, 2003, 10:09pm

Yes, for this purpose it is rather annoying.

There's a relatively simple workaround for the generators
though.

Place an instance of the signal generator in a schematic,
and run a normal transient analysis.

Then use the printvs function in the calculator on the
output of the block (don't use start, stop and step in
the printvs form), and then use File->Print to save the
output data to a file.

You can then use a vsource with type pwl (or vpwlf),
or port with source type pwl, to read the data you
saved in the previous sim and use that in an envlp
analysis.

A bit of a hack, bit it works.

I believe there is an initiative to solve the hidden state
problem (probably by fixing the models in the short
term) .

Regards,

Andrew.

Title: Re: hidden state in rfLib and SpectreRF
Post by Jess_Chen on Jun 5^th, 2003, 7:51am

In a real design exercise, I think you will eventually find it necessary to use pwlf sources in any event. In designing WLAN transmitters for example, the baseband signal is often digitally filtered to meet spectral mask and EVM requirements. The digital filter is not easy to design. Some pulse shapes are more susceptible to spectral regrowth than others. You will probably need a design tool like SPW or Matlab to design the digital filter. That being the case, it is much simpler to generate the baseband signal in a tool like SPW or Matlab then import it into Spectre or SpectreRF using pwlf sources for envelope simulation. In many cases, different design groups are repsonsible for the baseband and RF sections. The pwlf provides a convenient interface between two groups that often use very different design tools.

-Jess