Xref: utzoo comp.lsi:789 comp.lsi.cad:215
Path: utzoo!utgpu!jarvis.csri.toronto.edu!rutgers!gatech!mcnc!kenkel
From: kenkel@mcnc.org (Stephen Kenkel)
Newsgroups: comp.lsi,comp.lsi.cad
Subject: Spice Level 3 charge modeling
Keywords: spice, xqc, charge
Message-ID: <5009@alvin.mcnc.org>
Date: 3 Aug 89 17:49:55 GMT
Organization: Microelectronics Center of NC; RTP, NC
Lines: 24

Does anyone have a publicly distributed version of Spice (TM)
which has accurate charge modeling for the Level III MOS device
model?  Our copy of Spice2g6 (the "standard" Fortran version)
has the charge-modeling code commented out.  Spice3a7 and Spice3b2
both seem to be missing this code, and have comments indictating
"bypass charge modeling."

One test is the paramter xqc.  Spice2g6 ignore this parameter, without
complaining.  Spice3* complains, and then ignores the parameter.
(This paramter effects the partitioning of the channel charge, 
and is also used as a flag to choose between the Meyer and 
the Ward/Dutton charge modeling.  The charge effects can most
easily be seen in the high frequency response with AC analysis)

I tried to re-activate the code (set ICHARGE=1) in Spice2g6, but in our
distribution the charge function (mos3q) is not passed the appropriate
parameters, and I would prefer not to hack up foreign code.

Can anyone shed light on this problem?

(We are trying to verify our implementation of the Level III model
in our own circuit level simulator, CazM)

Thanks