Re: [ng-spice] pb with spice3f4 and ngspice


To ng-spice@ieee.ing.uniroma1.it
From Al Davis <aldavis@ieee.org>
Date Sun, 15 Apr 2001 11:59:30 -0700
Delivered-To mailing list ng-spice@ieee.ing.uniroma1.it
In-Reply-To <Pine.LNX.3.96.1010415201213.9532A-100000@ieee.ing.uniroma1.it >
Mailing-List contact ng-spice-help@ieee.ing.uniroma1.it; run by ezmlm
References <Pine.LNX.3.96.1010415201213.9532A-100000@ieee.ing.uniroma1.it >
Reply-To ng-spice@ieee.ing.uniroma1.it

On Sun, 15 Apr 2001, pnenzi@ieee.ing.uniroma1.it wrote:
> On Sun, 15 Apr 2001, Al Davis wrote:
> > On Sun, 15 Apr 2001, pnenzi@ieee.ing.uniroma1.it wrote:
> > > On Thu, 15 Mar 2001, routoure Jean-Marc wrote:
> > > > Hi,
> > > >
> > > > I find a little bug in ngspice ans spice3f4.
> > >
> > > Correct, I have checked that on the latest CVS code. The bug
> > > seems to be related with the way that spice uses for deriving a
> > > pmos model from an nmos: it multiply some values by MOS1type
> > > which means: -1 for pmos and 1 for nmos. May be that the "show
> > > all" command do not show vds and vgs but MOS1type*vgs and
> > > MOS1type*vds. This can be the cause of the error.
> >
> > "It's not a bug.  It's a feature!"
>
> Are you joking ??? ;-) It seems a Micro**** way...
>
> > It shows the values of the internal model, after coercing into a
> > "standard" form.  This is the way a modeler would want to see it.
>
> Al, getting seriously, do you think that this "feature" should be
> removed ?

The way it is implemented is minimum code.

Right or wrong, it has been that way forever.  Even Spice-2 did it 
that way.  If you change it, some scripts will break.  Long time 
users who have used it that way for years will find it foreign.  I 
consider Spice to be primarily a legacy program, and the legacy must 
be maintained and preserved.  So, NO, this "feature" should not be 
removed.

However, in development of a new simulator, that doesn't claim 100% 
Spice compatibility, you are no longer bound to carry all of the 
mistakes^h^h^h^h^h^h^h^h features of the past.


Partial thread listing: