Re: [Help-gnucap] Strange behaviour on .model definition, a bug?

[al davis]

On Monday 12 March 2007 18:00, Rubén Gómez Antolí wrote:
> Ok, I understand, but, output circuit is "correct" when I
> comment these lines, any ideas?

This is a different bug, one I didn't know about.


> open circuit: internal node 2

This alone doesn't necessarily indicate a problem.  It says it 
had some trouble solving, but might still be successful.  Put 
it in the same class as messages like "pivot change on the fly" 
or "source stepping used" that you sometimes get from Spice.

I confirmed that there really is a bug that was introduced 
between the 0.34 and 0.35 releases.

If you do ".option noincmode" it gives correct results.

This is definitely a bug.  If ".option noincmode" ever makes a 
nontrivial difference in results, there is a serious bug.

If this is the case, it may also make sense to check ".option 
nobypass", which also should never make a nontrivial 
difference.  In this case, with ".option nobypass", results are 
still wrong.