Stats

  • Locked Locked
  • Replies 5
  • Subscribers 63
  • Views 15068
  • 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

Is it possible to save the rational model of the S-parameter block?

Aldo2

Hello,

I'm doing a parametric AC simulation (mmsim 11.1). I need a lot of steps, but the single simulations are fast.

When I add an S-parameter block the parametric simulations slows down because a rational fitting process is required at each simulation step ("Starting rational fitting process." message appears in the output log form).

Is it possible to save the rational model of the S-parameter block? I would like spectre could calculate the rational model only at the beginning of the sweep and then use it on the following sweep steps.

Thank you

Best regards

Aldo

  • Aldo,

    First of all, we normally don't recommend using rational (see Tawna's latest version of her nport app note) but if you really did want to, the ROM data file field on the nport component allows you to specify the file in which it is saved.

    Regards,

    Andrew.

  •  Hi Andrew,

    After my post "Enforcing causality of S parameter in file..." (Thu, Dec 1 2011 2:20 AM)
    ===================================================================
    "Hello, I'm using harmonic balance analysis (mmsim 11.1) with a schematic containing an
    S-parameter model. In the NPORT component I set Causality Correction = no During the
    simulation I get the following message: Notice from spectre during periodic steady
    state analysis `hb'. Enforcing causality of S parameter in file Is it possible to
    avoid this time-consuming step? Thank you very much Aldo"
    ===================================================================
    I opened a service request with customer support.It was suggested to me to use
    rational fitting with a low order like 3 or 4 (I used 4).

    However now, with AC, I see no problem to use SPLINE, but with HB I should continue to
    use RATIONAL for the issue explained in my previous post.

    Do you see any drawback (I've not read yet the application note on NPORT)?

    Thank you

    Best Regards

    Aldo

  • Aldo,

    Using the rational model is often even more time consuming than the causality correction...

    The main problem with the rational method is that it really has not had much attention over the last few years - the focus has been on the convolution-based approaches (spline and linear). So I generally would advise against it, especially for larger number of ports or higher order systems. I have used it to check results in certain specific situations, but it would not be my first choice. Hopefully the specific issue you were having was followed up with R&D to help ensure that the s-parameter modelling is improved to handle the s-parameter data you were having problems with. Although of course we'd need to have your data to do that - not sure (I've not checked the service request).

    I'm sure Tawna will chime in when she sees this.

    Regards,

    Andrew.

  • Hi Aldo,
     
    What Andrew has said is correct.   And you really do want to read the nport appNote.  (/blogs/rf/archive/2012/01/03/nport-application-note-has-been-updated-and-re-released.aspx?postID=1306677 )   Specifically the section on "Using the nport in SpectreRF simulations" if you are using SpectreRF.  There are some situations where you must use rational interpolation in SpectreRF.
     
     
     
    Rational interpolation can be a reasonable choice if you have lumped elements.  However, the rational model is not as robust and accurate as the convolution based model (linear or spline), particularly when the s-parameter data models distributed systems with long propagation delay, or the DUT is a multi-port system with the number of ports > 4.   
     
    It is important to use the latest MMSIM ISR when simulating a circuit containing nports.  R&D has made significant improvements to the nport, and more are in the works.
     
    If you have a specific SR/CCR open with Customer Support/R&D, I can take a look at it briefly.  But I would need to know the SR number.
     
    Best regards,
     
    Tawna
     
  • Hi Aldo,
     
    I took a look at your SR.    With Harmonic Balance, use spline or linear interpolation.  Causality correction should be set to fmax. (auto takes forever and only should be used only if fmax doesn't work.)     If the data is non-causal, you *want* causality correction!   The same goes for passivity.  I always turn on passivity checking ...and if the data is non-passive, then I set passivity to enforce.  If the data is passive, then I turn passivity checking off.

    If you follow the guidelines in the nport appNote, you will be fine. Without seeing your s-parameter data, it is difficult to give more solid advice.  (I'm the appNote author and worked with R&D on the most recent update for MMSIM 11.1).

    Best regards,
     
    Tawna

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.