Re: hspice level 49 and .lib/.param?
Dan McMahill <[email protected]>
| Newsgroups | gmane.comp.gnu.gnucap.devel |
|---|---|
| Message-ID | <[email protected]> |
Al Davis wrote: > On Friday 11 November 2005 12:35 am, Dan McMahill wrote: > >>anyone working on an Hspice level 49 MOS compatible model for >>gnucap? >> >>mosis, for example at: >> >>http://www.mosis.org/cgi-bin/cgiwrap/umosis/swp/params/tsmc-0 >>18/t58f_mm_non_epi_thk-params.txt >> >>has level 49 models for various CMOS processes. > > > That is HSPICE specific. I think it is just BSIM3 with a > different name, Try level 7. it doesn't quite get you there. level 49 in hspice also looks at a VERSION parameter along with the HSPVER parameter to pick out various bsim3v3 versions. The manual I have gives a table showing how to set these 2 parameters to pickout bsim3v3.0 - oct 95 bsim3v3.0 - oct 95 with june 98 fixes bsim3v3.1 - dec 96 bsim3v3.1 - dec 96 with june 98 fixes bsim3v3.2 - june 98 bsim3v3.2.1 - april 99 bsim3v3.2.2.- april 99 It looks like the hspice changes on top of bsim is: ACM -- junction diode model and area calculation method. ACM = 0,1,2,3 for the hspice model ACM = 10,11,12,13 for the berkeley model There are also a handful of parameters related to the junctions which are a bit different. So, where to go from here? I'd be willing to try and code up something for gnucap which would let you directly use a level 49 model file, but I could use a few pointers in the right direction. For example, should I create a d_mos49.model and try and put everything in there (even though it would duplicate stuff from d_mos7.model)? In terms of how to deal with the VERSION and HSPVER parameters, should I just look at how the LEVEL parameter is currently dealt with in gnucap as an example of how to switch models based on a parameter? Since mosis seems to mostly supply level 49 models, this should be a huge benefit to students who want to be able to take a model from mosis and just run with it. I guess I need to first read and understand how gnucap implements the mos models to begin with. Maybe then I'll have a clearer vision of how to actually implement this stuff. Thanks -Dan