Zero Diagonal found in Jacobian

[Jimmy Blue]

Get this message sometimes, it seems to have no
documentation (Help search fails) and no sense
to it. The same circuit flat, and hierarchical,
will call out different quantities of complaint.
Overlaying a resistor mesh does not fix it (nor
is that a desirable thing, instantiating a large
quantity of resistors). If it is severe enough
the "reordering Jacobian" fails to satisfy it and
the simulation bombs.

What is this whining about and what does the
circuit need, to get happy?

 

On Tuesday, January 27, 2004 11:45:10 AM UTC-5, Jimmy Blue wrote:

Get this message sometimes, it seems to have no
documentation (Help search fails) and no sense
to it. The same circuit flat, and hierarchical,
will call out different quantities of complaint.
Overlaying a resistor mesh does not fix it (nor
is that a desirable thing, instantiating a large
quantity of resistors). If it is severe enough
the "reordering Jacobian" fails to satisfy it and
the simulation bombs.

What is this whining about and what does the
circuit need, to get happy?

I'm seeing that behavior as well. I'm using an out-of-date version of spectre (10.1.1 from Aug 2011) and I've yet to encounter that on my new version (12.1.0 from April 2013). I was able to get the old simulations to run by loosening up tolerances and quantities such as minr.