|
|
Magic Mailing List |
|
From: Jeff Sondeen (sondeen AT ISI DOT EDU) Date: Fri Feb 13 2004 - 15:24:18 EST
not sure if this helps, but historically we simply omit the diff->substrate caps to make sure they're not double counted (but allow user to uncomment them if they want, maybe to calculate total parasitics for some reason). /jeff eg: (from ftp://ftp.isi.edu/pub/sondeen/magic/new/beta/2002a/SCN6M_SUBM.10.light.tech27) #ndiff # MODEL HANDLES THIS: areacap (ndiff,ndc,ndm12c)/active 9.990 # MODEL HANDLES THIS: overlap (ndiff,ndc,ndm12c)/active ~space/w 9.990 # MODEL HANDLES THIS: perimc (ndiff,ndc,ndm12c)/active ~(ndiff,ndc,ndm12c,nfet,pfet,fet)/active 25.900 # MODEL HANDLES THIS: sideoverlap (ndiff,ndc,ndm12c)/active ~(ndiff,ndc,ndm12c,nfet,pfet,fet)/active ~space/w 25.900 Tan, Eric L. writes: > Hi Tim, > > I downloaded magic-7.2.62.tgz and installed with non-TCL option. When I do > ext2spice, all the area and perimeter of fets are all zero. I used > 'ext2spice -J flat myfile', with all extract options in magic. Does this > happen to you? I tried older (7.1) ext2spice on the newly extracted .ext > file and it gives the area and perimeter normally, with a remark that says > the cell was extracted by 7.2 and should be re-extracted. Should I use this > old ext2spice instead? > > Another related question that confuses me is: From my understanding (by > reading magic man page), the capacitance option for :ext enables node > capacitances to substrate (perimeter and area) computation. Sounds like > junction cap to me. If so, then the juction cap will be computed twice if > the simulator(hspice) also computes it with params CJS ans CJSW in the FET > model cards. Is turning off capacitance option the right thing to do? > > Thanks! > Eric
|
|
|
|