Re: [ng-spice-devel] Device bug in spice rework 5_2
On Tue, 25 Jan 2000, Anthony Wilson wrote:
> Well, in implementing the bsimsoi model, I thought I could just put all
> three version in the executable. But although I was able to get it to
> compile, the sample spice decks did not run correctly. Evidently because
> all the version used most to the same code(i.e. equations) that the
> simulator was not distinguishing among the variations. In looking at the
> model more closely, one would have to differentiate all three bsimsoi from
> one another so that the correct initialization is done and so that the
> correct equations are called during simulation. How would one do this I
> asked? Not quite sure. So that's why I only implemented the dd bsimsoi
> model in MacSpice.
Berkeley's people say that each module is separate from the others. I
made a version of spice with the three soi models and got some strange
behaviour, the I renamed all the B3SOI to B3SOIDD, B3SOIFD, etc, and got
DD and FD working. There is another fix, I think that there is function
called bsim3soi... that should be change to in bsim3soidd.... and so on.
What relations exists between models, they are intended to be accessed via
the SPICedev structure. Some issue remains on the state vector, which
should be independent but I still do no figure how it works.
> I would think so, this was happening to me also. You may want to check you
> link order also. Because spice is statically linked, the order has made a
> difference in device setup in previous versions of MacSpice.
Yes, this is an issue, the new model is always linked after bsim3, I will
check
> did you run a different spice deck against the level 49 model?
Yes, same result
Paolo
Partial thread listing: