Stats

  • Locked Locked
  • Replies 19
  • Subscribers 63
  • Views 11849
  • Members are here 0
This discussion has been locked.
You can no longer post new replies to this discussion. If you have a question you can start a new discussion

PPV values from pss/pnoise simulation in spectreRF

VLSIiitm

I would like to see the perturbation projection vector(PPV) which is used, e.g. to calculate phase noise in an oscillator.

I read an old presentation on "Noise aware PLL plugin" (www.cadence.com/.../ctp_cdnlivesv2007_Thibier...). Apparently this has been discontinued(according to this post community.cadence.com/.../31327). Anyway, in the presentation, it says that perturbation projection vector(PPV) values are stored in the simulation results directory(slide 17). Is this true of a regular pss/pnoise simulation in SpectreRF? If so, how would I make it save the PPVs and access the results? Is there some argument that can be given to save? (like for time varying operating points as mentioned in this post community.cadence.com/.../21507) 

(SpectreRF certainly calculates all these, but I don't know how to access these)

  • in reply to VLSIiitm

    Nagendra,

    I agree, it's odd. From the checks I made, it looked as if the issue was only related to shooting, and even then I wasn't sure from the issue reported whether ppv made it switch back to the old noise flow and affect the results. So I'd be interested in knowing if there are any remaining issues in current versions.

    Regards,

    Andrew.

  • in reply to Andrew Beckett

    Frank is indeed right. Now, "enable osc ppv" changes the phase noise results. I simulated an ideal oscillator and it gave wildly different phase noise with and without this option. I have filed a case (#46294758). In MMSIM141(which is what we used in earlier work, about the time of my original post) it worked fine. Now I tried MMSIM151 and SPECTRE171(sub-version 17.1.0.307.isr6 ) and both give the same wild result. The phase noise is correct without "enable osc ppv". The ppv itself is correct with "enable osc ppv".

    Nagendra

  • in reply to VLSIiitm

    Thanks for testing this. I should have tried it myself - I just tried with one example and I too see a difference (not anywhere near as severe as yours). Clearly something has broken when PPV is enabled with harmonic balance. I tried the same example with shooting, and in that case it didn't affect the noise result.

    So I suspect that when a customer reported an issue with the PPV for shooting, we fixed that - but clearly the harmonic balance part is misbehaving. The trouble with this not being a mainstream flow - it's probably not in the test suite any more.

    I'll keep an eye on the case and resulting CCR.

    Regards,

    Andrew.

  • in reply to Andrew Beckett

    Thanks. I checked with shooting and it is fine for my circuit as well. I also checked pss with harmonic balance(as opposed to hb). It is also messed up, but pnoise output in this case is different from that with hb.

    ppvs are the same in all 3 cases when enabled(pss+shooting, pss+hb, hb).

    Nagendra

  • in reply to Andrew Beckett

    An update(I've also updated the CCR). Looks like "AM" output of hbnoise is equal to "PM" noise from shooting and "PM" output of hbnoise looks nearly, but not exactly, like "AM" noise from shooting. The noise from shooting appears to be correct(matches calculations). USB or LSB(which show the total noise) seem to be correct, at least for my test case. Probably the total noise computation is fine, but decomposition into AM and PM is incorrect.

    Nagendra

  • in reply to VLSIiitm

    An additional hbnoise problem that we recently found is that LSB noise is wrong for multitone simulations (CCR 1958596). This is not related to PPV and will be fixed in Spectre 17.1 ISR8.

  • in reply to Frank Wiedmann

    As a follow-up, CCR 1974906 was filed, and will be fixed for the ppv issue with hbnoise in Spectre 17.1 ISR9.

    Regards,

    Andrew.

  • in reply to Andrew Beckett

    This problem has been fixed in "Version 17.1.0.433.isr9 64bit -- 25 Oct 2018". I assume future releases of v18 will also have the fix. Thanks all.

    Nagendra

  • in reply to VLSIiitm

    The fix should make it into SPECTRE18.1 ISR3, currently due around mid-December.

    Regards,

    Andrew.

Community Guidelines

The Cadence Design Communities support Cadence users and technologists interacting to exchange ideas, news, technical information, and best practices to solve problems and get the most from Cadence technology. The community is open to everyone, and to provide the most value, we require participants to follow our Community Guidelines that facilitate a quality exchange of ideas and information. By accessing, contributing, using or downloading any materials from the site, you agree to be bound by the full Community Guidelines.