CADENCE Noise analysis

Discussion in 'Cadence' started by Stephane Mazurie, May 15, 2004.

  1. Hi, All!

    Here's a new "problem": during CADENCE exploration,
    I've used the S-Parameter Analysis, what is very useful, to obtain
    Noise Figure, for example. Thanks to this analysis, I can guess and
    plot the noise figure (NF) for various variable, such as frequency,
    and the minimum noise figure (NFmin)...
    (in my case, and in this question, only this aspect is mentioned)
    When I launch a Noise Analysis, in Results->Direct Plot menu, I can
    also draw the Noise Figure, that is rather logical...
    But my problem is when I compare the results obtained with the
    S-Parameter Analysis-> NF-> Plot and Noise Analysis->Results->Direct
    Plot: they're not the same!

    I guess there are differences between these two simulations, but I
    can't find useful things about this.
    Maybe you could answer this new "problem"?

    Thanks in advance,
    Regards

    S. Mazurie
     
    Stephane Mazurie, May 15, 2004
    #1
  2. Where you using the output set to "probe" for the noise analysis,
    and the input set to "port"? And pointing at the same ports that you
    were using in the sp analysis?

    It's important that the input is a port (because it generates noise), and
    that the output is a probe (so that it can remove the noise in the load
    in the noise figure calculations).

    Also, I recall there being some errors with SP analysis in some earlier
    versions - what subversion are you using (type "icfb -W")?

    The results should be the same between the two otherwise.

    Regards,

    Andrew.
     
    Andrew Beckett, May 18, 2004
    #2
  3. It was true that the output was a port, and trying to change this port
    to a probe!
    I've found that it must be a "oprobe": just changing the Instance name
    "PORT1" to "oprobe1" would be ok, and is ok!
    And hopefully, the two simulation results are now the same !
    Tanks a lot!

    S. Mazurie
     
    Stephane Mazurie, May 24, 2004
    #3
  4. The instance name should make absolutely no difference whatsoever. It
    is important that it you use an oprobe to measure the output noise, or
    alternatively you ensure that the load is noiseless some other way
    (like having a resistor with isnoisy=no, or a port with a noise temperature
    of -273.152 degrees), but oprobe should be enough.

    There certainly is no need to change the instance name to oprobe1.

    Andrew.
     
    Andrew Beckett, May 25, 2004
    #4
Ask a Question

Want to reply to this thread or ask your own question?

You'll need to choose a username for the site, which only take a couple of moments (here). After that, you can post your question and our members will help you out.