Re[2]: IBIS connector models

     Greetings,
     
     
     RE: Including mounting foot prints in the model...
        We reviewed this a while ago. A problem we've seen with this
     approach is that there are not standard routing methods or layer
     definitions. This precludes the creation of a general connector
     model. In our business making a special connector model for each
     application could become un-manageable. As such, our standard model
     only provides a connector model from where the current enters the
     connector to where the current exits the connector. This allows pcb
     models to be added to the connector model.
     
     RE: 30+ cross sections...
     There are sometimes reasons for multiple cross section models.
     Technical explanation is too long for this email. But generally, I
     would suggest that most models do not exceed 15 sections. (Maybe I
     should confirm what you mean by "cross sections" first. I might have
     a different definition than to what you are referring."
     
     RE: Justification for circuit grounds.
     Many connectors models do not have defined grounds. CURRENT RETURN
     PATHS can be defined in SPICE. We supply these type of models on a
     regular basis.
     
     RE: Accounting for losses...
     What is meant by not very well? I think this is relative to
     wavelength... right??
     
     RE: correlating Maxwell matrices to measured data.
     It can be done. It has been done. It does take some work. It does
     take some special measurement and matrix extraction techniques.
     Again, technical explanation is too long for this email.
     
     RE: ... N-port circuit measurement ...
     One can use the same VNA to measure L, C and Z. These results can be
     correlated to matrix results. It does take some special measurement
     and matrix extraction techniques. Again, technical explanation is too
     long for this email. However... I believe that, N-Port measurements
     are seen as most appropriate for first order effects (??), sometimes
     (for better accuracy) further effects need to be reviewed.
     
     Gus Panella
     Molex, Inc.
     Ph: 630-527-4617

______________________________ Reply Separator _________________________________
Subject: Re: IBIS connector models
Author: "Dr. Edward P. Sayre" <esayre@nesa.com> at INTERNET
Date: 97/12/15 1:41 PM

John & others:
     
Do you do your measurements with connector mounted on a circuit card with
ground planes or is the connect probed at the connection points with 50
ohm probing? Using your differential measurement, how do you account for
signal/ground positions or ratios?
     
I favor mounting connectors on circuit cards to include the mounting and
footprint effects. I do not favor Maxwell matrices in IBIS unless they can
be measured and characterized. Many Maxwell matrices include as many as
30+ crossections with no real justification on circuit element grounds for
the large number. They often do not account for losses very well either.
     
Aside from that, Maxwell matrices are not theoretically what one measures
in a coupled N-port circuit measurement.
     
Ed Sayre
     
     
     
     
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~+
| NORTH EAST SYSTEMS ASSOCIATES, INC. |
| ------------------------------------- |
| "High Performance Engineering & Design" |
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~+
| Dr. Ed Sayre e-mail: esayre@nesa.com |
| NESA, Inc. http://www.nesa.com/ |
| 636 Great Road Tel +1.508.897-8787 |
| Stow, MA 01775 USA Fax +1.508.897-5359 |
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~+
     
     
 
Received on Mon Dec 15 14:38:40 1997