Geoffrey_Coram
|
I don't believe spice3 (spice3f5 from Berkeley) allows an ac analysis with a swept dc value. You could run a sequence of ac analyses with different Vg values and post process the results.
Otherwise, you probably have to make some changes to the C code of Spice. In particular, say for BSIM3, in the file "b3.c" you will find that the only operating-point parameters that you can ask for are gmbs, gm, gds, vdsat, id, vbs, vgs, vds. You will need to add lines like OP( "cgg", BSIM3_CGG, IF_REAL, "Cgg"), OP( "cgd", BSIM3_CGD, IF_REAL, "Cgd"), OP( "cgs", BSIM3_CGS, IF_REAL, "Cgs"), and re-compile Spice. (MOS level 1 has lines for Cgs and Cgd, but they are commented out.)
Then you will be able to save @m1[cgg] @m1[cgd] @m1[cgs] and these should appear in your output. (You might try it now with save @m1[gm] to see how it comes out.)
Note that these are the internal transcapacitances of the channel charge and do not include the overlap or junction capacitances. Also, the frequency is not used; these are raw capacitances (and if you have gate resistance, the effective impedance ought to depend on frequency). Contrast this with the ac analysis, where you stimulate the gate with an ac voltage and measure the ac current; the imaginary gate current is (2 pi freq) * cgg -- assuming no gate resistance.
|