MAGIC Magic Mailing List
 
 
From: cfk (cfk AT pacbell DOT net)
Date: Sun Apr 20 2003 - 13:35:20 EDT

  • Next message: R. Timothy Edwards: "Re: tech files & transistor geometry extraction" 

    Thank you for your comment Erwin, I have read the maint files, but I suspect
my problem is more in understanding if I am using magic wrong or if there is
an error in calculation. Towards that end, I went searching for an example
on the net that would allow common ground. There is a set of library pads in
magic format along with the technology file and tha tanner documentation
that begat the original pad at Univ Dallas
(http://www.utdallas.edu/~poras/courses/ee6325/cadtools.html). At the bottom
of this page is a reference to 0.5UM CMOS Pad Library Documentation and the
0.5um CMOS Pad Library itself.

I took the library, and loaded its various .mag files along with the
technology file onto my computer and ran "magic -w PadARef -TSCN3ME_SUBM.30"
as the PadARef appears to be the simplest pad with schematically 2
transistors. The documentation indicates this is a W=100*1, L=3*1 and M=6. I
am trying to interpret that as two multiple transistors with a total W of
600u and a total L of 18u.

When I do extract followed by exttospice from within magic, I get 12 nfet's
and 12 pfet's each wtih a W=61.8u and a L=90u. So, 12 * 61.8 = 741.6u Wide
and 12 * 90 = 1080u Long. This is from a cell whose documentation says is
90u wide by 300u high and is geometrically improbable. There must be
something wrong with the extracted spice Lengths and Widths. The questions
is, is it the technology file, the use of said file or magic itself.

I went to this new example in hopes that by using a commonly available set
of files on the net that someone else could duplicate my efforts and see
where I am going awry.

With Puzzlement, Charles

Kinda as a side note to Tim, and along with the cursor position idea of a
while back, I think that a ruler that can be placed on a layout in either
vertical or horizontal dimensions in microns would be a wonderful addition
to magic. I find I am continuously counting grid squares trying to keep
track of whats going on and when designing a transistor(s), it would be
quite helpful to keep the actual geometry in mind instead of just grid
squares and lambda. This should be a ruler that can be placed and then
erased when no longer needed.


----- Original Message -----
From: "Erwin J. Prinz" <ejprinz AT austin DOT rr.com>
To: "cfk" <cfk AT pacbell DOT net>
Cc: <magic-dev AT csl DOT cornell.edu>
Sent: Saturday, April 19, 2003 4:31 PM
Subject: Re: More on lambda & SCN6M_SUBM.10.tech27


> Dear Charles:
>
> The link:
> http://bach.ece.jhu.edu/~tim/programs/magic/
> has a section about technology files.
> You can make a subdirectory "current" in
> $CAD_HOME/lib/magic/sys/current
> and put the files formerly in "2001a" in there. Then, Magic can find them.
>
> The reason I post to the list is that there seems to be a newer version
> "2002a" also in
> ftp://ftp.mosis.edu/pub/sondeen/magic/new/beta/
> so maybe someone could comment whether the "2002a" files are more
> current that the "2001a" files.
>
> Either way, I don't think any other scripts or install procedures are
> needed.
>
> Best regards, Erwin

    •  
     
    Questions? Contact Rajit Manohar     		cornell logo