PPV values from pss/pnoise simulation in spectreRF

Why do you want to see this? It's part of a flow that is now discontinued, as you mentioned.

There is an option on the pnoise options form called "enable osc ppv" which does save a results database (accessible via the results database after simulation), but I've no idea if it would be of any use to you or even if the results are still useful (given the discontinuation of the rest of the flow, I doubt very much whether it's actively tested).

Regards,

Andrew.

Thanks Andrew! I tried it and the quantities are as expected except for some scaling factors which I need to figure out. Somehow I had either not noticed this option on the form or realized that it is for saving PPV waveforms. 

PPVs are essential for oscillator analysis and are calculated to obtain noise or other sensitivities in the oscillator (e.g. in pnoise) independently of the noise aware PLL package. The latter was merely using PPVs to accurately macromodel the nonlinear effects in the VCO. PPVs can provide useful insights during oscillator circuit design. There are papers (below) written about how to calculate PPVs from PXF results. But if PPVs are available from the simulation results, there is no need to re-calculate them. (After all, PXF for an oscillator uses PPVs and the calculations in the papers use PXF to back calculate the PPVs :)

I noticed that MMSIM12 manual has some description of the enable ppv option, relating it to noise aware PLL flow, whereas MMSIM14 says that the enable ppv option is obsolete. Please don't get rid of the option altogether!

Levantino, S.; Maffezzoni, P., "Computing the Perturbation Projection Vector of Oscillators via Frequency Domain Analysis," in Computer-Aided Design of Integrated Circuits and Systems, IEEE Transactions on , vol.31, no.10, pp.1499-1507, Oct. 2012

Levantino, S.; Maffezzoni, P.; Pepe, F.; Bonfanti, A.; Samori, C.; Lacaita, A.L., "Efficient Calculation of the Impulse Sensitivity Function in Oscillators," in Circuits and Systems II: Express Briefs, IEEE Transactions on , vol.59, no.10, pp.628-632, Oct. 2012

hi, I checked the ppv=yes box, but the pnoise result are worse when this box is checked.

I compared the input.scs for the 2 run. they are the same except for the ppv=yes part. 

Andrew, can  you confirm that the ppv=yes will affect pnoise result? it shouldn't I hope? thank you for your help.

Frank

Frank,

Which spectre version are you using? (spectre -W will report this, but the subversion is also shown at the top of the log file). Are you using shooting or harmonic balance?

A while back a new method was introduced in pnoise to improve accuracy and reliability, and initially ppv was not supported with the new method (so I think it switched back to the old method). That was addressed (via CCR 1775821 - in case I ever need to look it up again) in SPECTRE17.1 ISR2.

Regards,

Andrew

hi Andrew,

thank you for replying. the version is "Version 15.1.0.644.isr13 64bit -- 22 Oct 2016"

I am using harmonic balance. Thank you. 

Frank

Hi Frank,

I suggest you start by checking with SPECTRE17.1 (ideally the latest ISR). The version you're using is soon after some of the noise enhancements happened - so important to check it's OK in current version. If it's still a problem in the latest version, you should contact customer support. PPV output may not be a mainstream flow nowadays, but it does seem that we aim to fix issues with it.

Regards,

Andrew.

hi Andrew, so SPECTRE17.1 ISR2. has the ppv supported and checking ppv=yes still give the correct result? thanks

Frank

Hi Frank,

I can't say that for certain, because it may be a specific issue with your simulation data, or it may be related to the updates that I found R&D had made in that version. However, it does seem that a number of updates to PPV have been made to align them with the newer noise improvements, so it absolutely makes sense to use the newer version. As I said, if you still see issues, you should contact customer support.

Regards,

Andrew.

Thank you Andrew. have a good night.

This is very strange. We have not seen pnoise results being affected by ppv setting. I had thought that this was just a reporting option, i.e., ppvs are used for calculating phase noise and this option just saves them so that we can view them. The latest manual seems to list it in "convergence parameters". Not sure if it has any effect on actual convergence.

I haven't used it recently. I'll try a run when I get some time and let you know.

Nagendra