Path: utzoo!utgpu!news-server.csri.toronto.edu!rutgers!mcnc!ge
From: ge@mcnc.org (George Entenman)
Newsgroups: comp.lsi.cad
Subject: Resistance extraction bug in magic v.6
Message-ID: <2743@speedy.mcnc.org>
Date: 14 Dec 90 20:58:11 GMT
Distribution: usa
Organization: MCNC; RTP, NC
Lines: 31


We seem to have found a small bug in the magic version 6.0
resistance extraction.  In the technology file the contact
resistance is set using the command:

	contact type size R

The documentation indicates that type is the type of contact,
size is the length of a minimum contact, and R is the
resistance of a single contact in milliohms.  We entered this
information for our technology and tried a few simple
examples.  The results were not correct and the problem was
traced to the contact resistance.  After poking through the
Magic 6 resistance code we found that in the file ResMakeRes.c
that the resistance of a contact was computed as follows

	(area of the contact)/size

instead of

	(area of the contact)/size*size

The code would work if the definition of "size" was changed to
be the area of a minimum contact instead of the length of one
side.  The best thing to do is probably to change the code
since a documentation change depends on the user noticing the
change instead of the installer making an update.  Who, if
anyone is responsible for accumulating and distributing bug
fixes for magic version 6.0?

				Fred Heaton and George Entenman