Paul
|
Sorry,
I completely forgot to update this post as we got around the problem of getting different simulation results using Spectre (with Spectre model), Hspice and Eldo (same model, except level=53 for Eldo).
There is a paper on sourcelink which helped a lot: BSIM3V3.X MOSFET MODEL COMPATIBILITY HSPICE LEVEL 49 -Spectre BSIM3V3.X which explains the different ways of calculating source and drain parasitic capacitance for Spectre compared to Hspice. With respect to Eldo vs Hspice, the point to consider is that Hspice has one parameter defining the calculation of the related diffusion parasitics, while Eldo has several. So we had to be careful that both models are compatible on that point. CJGATE is conditionnally used by Eldo, depending on the value of DCAPLEV. In our case, DCAPLEV has the default value and thus CJGATE is *not* used, but this doesn't seem to alter the simulation results with respect to Hspice results.
Thanks for your help.
Paul
|