Re: [SI-LIST] : Linpar 2-D field solver experiences?

About this list Date view Thread view Subject view Author view

From: [email protected]
Date: Thu Nov 30 2000 - 10:46:58 PST


In a message dated 11/21/00 9:58:22 PM Pacific Standard Time,
[email protected] writes:

> Has anybody used the Linpar CAD 2-D field solver?
> I've used ansoft, CALIF, Greenfield, Pacific Numerix,
> and have recently come across Linpar.
>
> Linpar seems to be a very easy to use CAD tool, and
> seems very cost effective (sub $500) for a general
> purpose 2-D field solving tool for simulating impedance
> and crosstalk.
>
>

Dave:

I have not tried the latest Windows version of LINPAR, but I have used MSDOS
versions of both LINPAR and its companion program LINRES for almost a decade
with very positive results/success. I also have used Agilent's APPCAD,
Polar's CITS25, and UltraCAD's T-Line Calculator. Relative to the real world
(i.e., documented by measurement on known physical layups), LINPAR has been
one of the best.

A MAJOR ADVANTAGE of LINPAR over the other free (or low cost) programs is
that it can address multiple traces, including different sizes,
simultaneously. Because of the coupling of adjacent traces in today's densely
routed boards, the individual trace loading and crosstalk is important to all
SI engineers. Most pay an arm and a leg for CAD programs that (rightly)
address this issue, but their accuracy (as you might have noticed) has often
been questioned in this SI forum.

The (purely text) output is in the form of L-R-C-G matrices, as well as Er
(effective), propagation speed, and individual trace impedances. For
crosstalk and waveform simulations, these matrices must then be imported into
the LINRES program, sources and loads defined and saved into additional
files, and then run. The simulation plots from LINRES are very basic, but I
have compared the results with HSPICE W-element simulations and FOUND
NEGLIGIBLE DIFFERENCE (except in the HORRENDOUS difference in price).

However, LINPAR/LINRES (MSDOS) has definite limitations. The ease of use is
cumbersome to execute several combinations of trace widths, trace spaces,
dielectric thicknesses, Er values, etceteras. Because of the original MSDOS
program size limitations (from a decade ago), more complex structures are
limited in size. And, with today's thin dielectrics, stripline trace
thickness and shape has an impact on the impedances and crosstalk that is not
accounted for (infinitely thin traces are the default for stripline), unless
you draw a full 2D picture of the configuration. This latter approach is
cumbersome and time consuming, but can be done. The infinitely thin traces
yield higher (~5%) Zo predictions than reality unless the dielectric
thickness is >5 x trace thickness. However, microstrip trace thicknesses ARE
accounted for.

NOTE: The newer Windows version of LINPAR may have corrected all of the above
negatives/limitations. Further comments from any who have tried it are kindly
solicited.

Dave, if you have received other responses NOT sent to the SI List, could you
summarize them and post them?

Mike
Michael L. Conn
Owner/Principal Consultant
Mikon Consulting
(408)727-5697

        *** Serving Your Needs with Technical Excellence ***

**** To unsubscribe from si-list or si-list-digest: send e-mail to
[email protected]. In the BODY of message put: UNSUBSCRIBE
si-list or UNSUBSCRIBE si-list-digest, for more help, put HELP.
si-list archives are accessible at http://www.qsl.net/wb6tpu
****


About this list Date view Thread view Subject view Author view

This archive was generated by hypermail 2b29 : Tue May 08 2001 - 14:30:18 PDT