In looking at MWO and ADS, it seems to me that most of the nonlinear transistor models in their libraries are simply Spice models (either as ..model or .subckt files). Is this pretty much all harmonic balance simulators use; or are there actually some true HB transistor models out there?
Thank You,
Bill
Yet I always hear how inaccurate Spice models are at microwave frequencies; so I guess I just assumed there were actually HB models out there so as to more accurately take into account all of the high frequency effects of an active device...
Cheers,
Bill
What would you expect to be in a 'true HB transistor model' that is not also accessible from the code comprising a full spice model?
Charles.
There's some interesting stuff (research) going on with nonlinear black-box models, but basically if you want a nonlinear time or frequency-domain model, you're going to be using a spice-like model. In the frequency domain case the HB simulators just fft and ifft, but the equations are often those used in spice.
This isn't all that bad... For instance, some work under Prof. Mayaram at Oregon State showed that the BSIM3 model works pretty good, with some extra parasitics, into multiple GHz. There are a number of 'high frequency' device models that are basically low-frequency spice models with extra parasitics which become important as the frequency rises.
As far as "true HB transistor models" -- what would that look like?
Most high frequency models are lumped approximations for the passive stuff and spice-like nonlinear models with (apparently kluged-up) parasitics added that kinda make the frequency response go the right way.
If the spice models are bad... -- I've always wondered how anything worked even semi-efficiently before HB. Dropping nonlinear devices into a microwave network and modeling it using linear models... ugh.
Have something to add? Share your thoughts — no account required.
Ask the community — no account required