The Designer's Guide Community Forum

The Designer's Guide Community Forum
https://designers-guide.org/forum/YaBB.pl
Simulators >> RF Simulators >> Setting of "number of harmonics" in PSS
https://designers-guide.org/forum/YaBB.pl?num=1142607529

Message started by rintel on Mar 17^th, 2006, 6:58am

Title: Setting of "number of harmonics" in PSS
Post by rintel on Mar 17^th, 2006, 6:58am

Hello all,

I am simulating a mixer with PSS and PXF. The setting of "number of harmonics" in PSS seems to affect a lot on the spectrum ploted by PXF voltage gain. There can be 20dB difference on the LO harmonic response, if I choose the "number of harmonics" to be 0 or 20. The output of mixer is zero-IF. I think the maximum sideband option in PXF should matter, but why the "number of harmonics" in PSS also affect in this case? Any discussion on related topics will be appreciated!

Best regards,
Rintel

Title: Re: Setting of "number of harmonics" in PSS
Post by Ken Kundert on Mar 17^th, 2006, 9:13am

Changing the number of harmonics in the initial PSS analysis should only indirectly affect the results from a subsequent PXF analysis. It can affect it because setting the number of harmonics places a bound on how large the timestep can be in the PSS analysis. If changing the number of harmonics strongly affects the PXF results it implies that the PSS analysis is not producing accurate results. This might be because you have widely spaced frequencies and signal sizes present in the circuit. You might consider tightening tolerances a bit in the PSS.

-Ken

Title: Re: Setting of "number of harmonics" in PSS
Post by rintel on Mar 17^th, 2006, 11:27am

Thank you Ken!
I have tried your suggestions. Set errpreset in PSS to conservative, and maxstep to 1/10 of LO period. Also I set tolerance in PXF to 1e-5. There are some change on the results at least. But the difference is still big. It seems to me that when "number of harmonics" is 0, the result is less accurate. Is it fair to guess that larger "number of harmonics" tends to give more accurate simulation? Is there any other way to make sure which is the accurate one?

rintel

Title: Re: Setting of "number of harmonics" in PSS
Post by Ken Kundert on Mar 17^th, 2006, 2:17pm

Something does not seem right. Take a look at the value used for maxstep as reported in the "important parameters" section of the log file. If you manually set maxstep in the 0 harmonics case to be smaller than that used by Spectre in the 30 harmonics case, then changing the number of harmonics should not affect PXF. It it does, something else is going on. IN that case you should probably contact Cadence for an explanation.

-Ken

Title: Re: Setting of "number of harmonics" in PSS
Post by rintel on Mar 19^th, 2006, 10:31am

I tried your suggestion again. The setting of important parameters, including 'maximum step', in both cases, are exactly the same. It seems no way to resolve the problem. Also to contact Cadence is unlikely to be realistic, because the process as I read online is really complicated...
Anyway, thanks for help!

-Rintel

Title: Re: Setting of "number of harmonics" in PSS
Post by Andrew Beckett on Mar 30^th, 2006, 5:20am

Why are you setting the tolerance in PXF to 1e-5? That's much larger than the default of 1e-9. Leave it alone...

You've probably degraded the PXF accuracy considerably.

Andrew.