Stats

  • Locked Locked
  • Replies 4
  • Subscribers 126
  • Views 8214
  • 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

SpectreRF: Can we force no pss iterations? Or just take the 1st one

StephanWeber

Hi,

pss is great e.g. also because of its sweep capabilities and powerful output measurement options. However, sometimes the pss iterations bother and take much time to execute. Is there a way to shut off the iterations? I think of a simple amplifier or comparator or latch. Here usually the tran sim is already very close to the steady state. I could use the initial tran run only, but this way I would still waste time with iterations, and I cannot use the powerful measurements like P1dB, spectrum output, etc.

Bye Stephan

  • No. The clue is in the "already very close to the steady state". Very close is not good enough - plus there's additional data that needs to be computed in the analysis which is more than just a transient. If it's really close enough, it would only take one or two iterations anyway. I would have thought you'd know this from your background...

    Andrew

  • You can use the steadyratio parameter to fake pss convergence (which is usually not a good idea, as Andrew mentioned).

  • in reply to Andrew Beckett

    I am mainly interested in things like P1dB or harmonic distortions. For digital circuits like latches, comparators, etc. my observation is that they usually do not have long time constants, which might impact the results like they do in many AC coupled amplifiers or control loops.

    Indeed using steadyratio is something I should try. Main need is having powerful sweeps also in a transient-like analysis, e.g. I have an 200mA sine pulse overlaid to a DC value, and need to check how an amplifier behaves for varying DC values, like if approaching the saturation limits. So I need to run a sweep on DC value, but also as usual vs T, VDD, etc. Doing some sweeps by spectre itself is usually faster. As long as we do not trigger pss convergence problems. I could also use a very long tran analysis, in which the DC value is swept in a staircase style, but setting this up is more effort, and the results are much harder to extract.

  • in reply to Frank Wiedmann

    Dear StephanWeber,

    I was about to respond this morning when I saw Frank's response - which was exactly what I was going to suggest - both Frank and Andrew are way ahead of me as usual!

    Just to provide a bit more information concerning Frank's response if it is useful...

    From Andrew's comment StephanWeber, you probably know this already for which I apologize. However, for others who might view this post, the convergence criteria and the parameters that impact PSS convergence are provided in the RF Analysis Assembler/Explorer User Guide (21.1) at URL:

    https://support.cadence.com/apex/techpubDocViewerPage?path=spectreRFinExplorer/spectreRFinExplorer21.1/Chap4.html#pss_analysis

    where it states:

    Hence, you might consider relaxing your simulation accuracy requirements "very carefully" as a possible means of reducing the number of iterations to achieve PSS convergence. However, achieving your goal may result in simulation errors that are not tolerable in your application.

    Shawn

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.