Stats

  • Locked Locked
  • Replies 2
  • Subscribers 125
  • Views 3035
  • 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

Spectre gives different DC simulation results for the same circuit (even some results don't satisfy KCL)

zuiying

When I simulated the same two-stage op amp circuit in spectre many times, spectre gave different dc operating point simulation results (the following three pictures are the dc operating point simulation results of the same circuit, each time M6 and M7 Ids are different).

The last two simulation results do not satisfy KCL (the id of M6 and M7 are not equal). Are these phenomena caused by errors in the simulator itself? Or are there other reasons? Is there any way to avoid simulation results that do not meet KCL? Thank you very much.

Parents
  • Andrew Beckett

    The differences are likely to be a combination of numerical tolerances and also the effect of the gmin impedances added to aid convergence around potentially off devices. In this case It's more likely to just be due to reltol because gmin is normally 1e-12 Siemens (so 1TOhm) and so the current through gmin is likely to be at most a couple of picoamps. Given that KCL is checked (like everything else) using reltol - the sum of currents into each node has to be less than reltol*(largest current into node) to be seen as converged, it's likely that. With reltol=1e-3 then the current differences are well below reltol (about 100 times better than reltol would assume). You could of course tighten reltol, but do you really need that level of accuracy (which would slow down the simulation of course). All the other results are also going to be using reltol (and vabstol/iabstol) to determine numerical accuracy too.

    Regards,

    Andrew

Reply
  • Andrew Beckett

    The differences are likely to be a combination of numerical tolerances and also the effect of the gmin impedances added to aid convergence around potentially off devices. In this case It's more likely to just be due to reltol because gmin is normally 1e-12 Siemens (so 1TOhm) and so the current through gmin is likely to be at most a couple of picoamps. Given that KCL is checked (like everything else) using reltol - the sum of currents into each node has to be less than reltol*(largest current into node) to be seen as converged, it's likely that. With reltol=1e-3 then the current differences are well below reltol (about 100 times better than reltol would assume). You could of course tighten reltol, but do you really need that level of accuracy (which would slow down the simulation of course). All the other results are also going to be using reltol (and vabstol/iabstol) to determine numerical accuracy too.

    Regards,

    Andrew

Children

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.