Stats

  • Locked Locked
  • Replies 5
  • Subscribers 124
  • Views 12054
  • 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

DC Convergence Problem When Simulating a Post-Layout r_c and r_c_cc Extraction

shbmsra12

Hi All,

I'm simulating a post layout r_c and r_c_cc extraction of a transmitter circuit for transient simulation using config view, The problem is when simulation starts it checks for the "Trying 'homotopy = dpptran' for initial conditions" and made several attempts before it fails like after 4-5 hours. I also tried with loosening the vabstol and iabstol to relaxed values than the default value. But nothing works. For c_cc extraction the post layout simulation worked fine and no such error appeared for the same circuit. My transient simulation run setting is "Liberal" and there is one ideal constant current source apart from supply and control signals. Im using cadence IC6.1.8-64b.83 amd MMSIM 20.1.0.068. I have used Calibre v2021.3_15.9 for the c_cc, r_c and r_c_cc extraction. The snippet of the output log is attached. Can anyone help me with it.

Thanks!

Parents
  • Andrew Beckett

    It's pretty hard to diagnose from a picture of a portion of the log file. Are you using APS or Spectre X? (If not, try those, and with APS try using +postlayout or +postlayout=hpa; with Spectre X start with +preset=lx).

    Other than that, I suggest contacting customer support as it will be much easier to support you that way!

    Andrew

  • shbmsra12
    in reply to Andrew Beckett

    HI Andrew,

    I'm using Spectre. Is this problem is something simulation specific? Could you please help me with where to choose "Spectre X start with +preset=lx" as told by you. And Let me know what else is needed to diagnose the root-cause of the problem.

  • Andrew Beckett
    in reply to shbmsra12

    Generally convergence issues are a consequence of the circuit or the models. There are however strategies in APS and Spectre X which will further aid convergence of difficult circuits (spectre classic will also try a number of strategies too, but there are additional algorithms in APS and X), and especially the postlayout options available in both (using the High Performance Simulation options in ADE allows you to pick APS or Spectre X) - that's why I was suggesting trying those to see if it will help. They almost certainly will make it simulate faster too...

    Regards,

    Andrew

  • shbmsra12
    in reply to Andrew Beckett

    Hi Andrew,

    Now I'm running the whole simulation in ADE. It works well with when high performance simulation is chosen in ADE and APS is selected.. But I believe the accuracy has been compromised upon choosing the APS option. If I want to run with SPECTRE only, it ends up showing DC convergence issue. I don't understand why it's option specific.

  • Andrew Beckett
    in reply to shbmsra12

    The accuracy should not be compromised when picking APS. Why do you think that? The whole point of APS is that it should give the same accuracy as Spectre (not necessarily identical answers, but it should giving results within the same accuracy tolerances as Spectre classic).

    To investigate why it has problems with spectre only, that would need customer support as I said earlier, and even then I'm not sure why you would be worried about using APS if that works. The reason why it can be easier to converge with APS is because there are specific additional strategies involved to help with DC convergence with post-layout designs, there are optimisations done to the circuit to reduce the complexity of the problem whilst keeping accuracy, and you're using more advanced technology to help with convergence in general.

    Regards,

    Andrew

Reply
  • Andrew Beckett
    in reply to shbmsra12

    The accuracy should not be compromised when picking APS. Why do you think that? The whole point of APS is that it should give the same accuracy as Spectre (not necessarily identical answers, but it should giving results within the same accuracy tolerances as Spectre classic).

    To investigate why it has problems with spectre only, that would need customer support as I said earlier, and even then I'm not sure why you would be worried about using APS if that works. The reason why it can be easier to converge with APS is because there are specific additional strategies involved to help with DC convergence with post-layout designs, there are optimisations done to the circuit to reduce the complexity of the problem whilst keeping accuracy, and you're using more advanced technology to help with convergence in general.

    Regards,

    Andrew

Children
No Data

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.