Monte Carlo DEV simulation in ADVance-MS

[microx]

Anyone around here tried to do Monte Carlo simulation in Eldo through
ADVance-MS? I'm trying to do instance specific (DEV) variation and have
been randomly getting the following message:

# #E Error: (eldo - 1520) Parameter SOMETHING_OR_OTHER found in `Top
Spice'
# depends directly or undirectly on DEV specification..
# This is not supported yet by ADMS..
#
# #E Error Loading Design
# #E Simulation failed
# #E

Given that it is so blatantly declared that it is not supported, I
would give up, but then it works fine and generates correct MC results
if I use a simpler testcase. I've been assured by someone at Mentor's
support that it should work fine in the latest version 2006.1 which I
am using, in spite of the error message. They say there must be another
problem in the design that causes the misleading error message to
appear. The way I see it is that it sometimes works and sometimes not
for larger designs. The same designs work fine if I take out the .MC
statement. Anyone encountered this and know what's going on?

 

[microx]

Additional info:

I have been trying to see what the difference is between the way the MC
parameters are specified in the 2 designs, a simple one that works, and
the more complex one that fails. In both cases, the parameter is
declared in a .param statement at the top level of the design. The only
difference I could see is that, in the simple testcase, we were using
it on a .model directly in the top level, while in the failing case,
the parameter is used inside a .subckt to produce another intermediate
value which is then used as an instance parameter for several resistor
and capacitor instances inside the subckt, with no .model involved.
Could this be the problem?