|
|
Magic Mailing List |
|
From: cfk (cfk AT pacbell DOT net) Date: Sat Jun 21 2003 - 18:24:42 EDT
I am having a puzzling time with models and tclspice. If I go to http://bear.ces.cwru.edu/eecs_cad/cad_spice.html and click on the first link to tut_spice3_invertor.html and use the model there, I can extract and simulate from magic and get a plot with tclspice from within magic of an invertor. This model says it is level 2 and has a modest number of parameters (about 25). The problems start when I go to http://www.mosis.org, click on electrical parameters, followed by TSMC 0.18 micron and select the first parametric test result for "T43B_MM_NON_EPI". The model there, which is supposedly obtained from wafer runs has level 8 with about 100 parameters. When I use this model in tclspice, I get an error for 'XL' and 'XW'. If I delete 'XL' & 'XW", then the spice run doesnt converge. What I am trying to do is correlate and invertor drawn in magic, extracted and simulated with tclspice with the model parameters from an actual wafer run from mosis. It seems to me that should be possible. It should be meaningful, especially if the invertor is drawn with the SCN6M_DEEP.09.TSMC technology file which should correlate reasonably well with 0.18 micron wafers fabbed by TSMC. It should be possible to not only simulate the rise and fall time of the invertor, but also the frequency of a 31stage ring oscillator. The frequency (about 320 Mhz), is also listed in the mosis parametric results. Certainly both an invertor and a ring oscillator can be drawn, extracted and simulated. The simulated result from magic should come out on the order of 320Mhz. I think we have all the information to accomplish this, but I dont seem to be able to get past a model issue from mosis with tclspice. Any suggestions or help would be greatly appreciated. Charles
|
|
|
|