Designer’s Guide Community

‹ Previous topic | Next topic ›

Pages: 1

Silvaco's models (Read 46 times)

August West

Community Member

Offline

I'll get up and fly
away

Posts: 71

Silvaco's models
Mar 01^st, 2004, 11:56am

Has anyone taken a serious look at the Silvaco models (http://www.silvaco.com/downloads/verilogADownloads.html). If so, how do they look?

I took a careful look and the diode, and it does not seem ready for prime time. It had a couple of issues.
1. Several standard parameters are missing (eg, ibv, xti)
2. vj parameter is misnamed phi, and given wrong default value.
3. noise model is missing.
4. The current used to determine the size of the diffusion capacitance includes the current through the junction and diffusion capacitances, which it should not.

Are the other models is a similar state? If so, it would seem it indicate that Silvaco is not yet using these models as their base models.

-August

IP Logged

Geoffrey_Coram

Senior Fellow

Offline

Posts: 1998
Massachusetts, USA

Re: Silvaco's models
Reply #1 - Jul 9^th, 2004, 8:33am

August -
You are correct that most of the models are not ready for prime-time.

Many of the problems arise from the lack of certain features in Verilog-A. For example, the BSIM3 model is also lacking noise, because you can't get the derivatives for the noise equation 8/3 kT (gm+gds+gmbs) without manually computing them -- defeating the purpose of a higher-level language. BSIM3 did seem useable for transient, though.

There is a subcommittee (of which I am the chairman) proposing extensions to Verilog-AMS to handle these features.

Another sort of problem is that found in the MOS level=3 model. It uses a capacitance-based model, which can result in lack of charge conservation. The subcommittee is not proposing extensions to handle this, since "everyone knows" that's a bad way to write a model now.

-Geoffrey

If at first you do succeed, STOP, raise your standards, and stop wasting your time.

IP Logged

Pages: 1

‹ Previous topic | Next topic ›