Stats

  • Locked Locked
  • Replies 2
  • Subscribers 125
  • Views 2185
  • 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 SP analysis -- noise correlation matrix

Yevgeny

Good day.

I am running the following simple simulation (spectre 17.1.0): 

rx (xx yy) resistor r=50
port11 (xx vssa) port r=50 isnoisy=no
port12 (yy vssa) port r=50 isnoisy=no
sp sp ports=[port11 port12] start=100M stop=1G dec=80 \
    donoise=yes file="res.s2p"  datafmt=spectre paramtype=s noisedata=cy annotate=status

This is a 50Ohm resistor, connected between two 50 Ohm noiseless ports. When I look at the resulting file "res.s2p" I see the S-parameter noise correlation matrix: 

0.023      -0.023

-0.023      0.023

I would like very much to understand how this matrix defined in SPECTRE, since I fail to arrive at these numbers no matter how I try to define the matrix.

The S matrix itself is, by the way, perfectly sound:

1/3 2/3

2/3 1/3

 

I also tried to use "paramtype=y" (i.e. Y params instead of S params), the noise correlation matrix becomes:

0.02    -0.02

-0.02   0.02

which makes perfect sense (Y-param noise correlation, i.e. 1/R, if I divide by 4kT df). 

I apologize ahead if this is some stupid numerical error on my behalf; have spent quite some time on this, and could find nothing in docs. I filed cadence support request, but perhaps knowledgeable folk will answer faster Slight smile

Help is much needed. 

Thanks!

  • Yevgeny

    UPD: I continue working on this. I run SP analysis on a real circuit (quite complex, transistors, inductors, using Intel proprietary modes through CMI, and such). 

    I run it once with "paramtype=y noisedata=cy" and the second time with "paramtype=s noisedata=cy". I look at noise correlation matrix (CY) and discover something disturbing: 

    with "paramtype=s" I get non-negative eigenvalues, but with "paramtype=y" I get both positive and negative eigenvalues. The negative ones are large; can not be attributed to floating-point arithmetic errors. 

    To make my point clear: Correlation matrix is supposed to be positive semidefinite, and since it is Hermitian by construction, it should have had only non-negative eigenvalues. I have to conclude that the matrix provided by "paramtype=y" is broken!

    So far, one good thing has come out of it: I have refreshed some undergrad algebra. 

  • Andrew Beckett
    in reply to Yevgeny

    I would suggest you provide this info to the AE dealing with your case rather than posting it here. I'm not going to duplicate work done by a colleague (I don't have the time, for a start), so I suggest you follow up via the official channel. 

    Andrew.

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.