When I try to run a Spectre dc and trans simulation with my 8-bit-adder circuit, have an internal error stopping the simulation. I get the following messages in the spectre log window:
1) the usual "unable to compile ahdlcmi module library" message (which has not stopped my simulations in the past),
2) a warning from spectre during initial setup about having PMOS and NMOS from the driver circuit running at 1 A/exceeding imelt = 10 mA (the test driver/load circuit provided by our course instructor)
3) "Internal error found in spectre during topology check, during initial setup." --> "Assertion failed in file 'topology.c' at line 1008"
Any ideas how to interpret error #3?? I've done the Virtuoso save and check with no error for the 8-bit-adder or any of the lower schematics in the hierarchy.
To explain how I'm implementing an 8-bit-adder: The design is four 2-bit-adders linked together. Each 2-bit-adder is made of 2 individual adders. Individual adders have Carry Out implemented with CMOS and Sum implemented with pass-logic transistor gates.
Thank you for your time.
Spectre 5.1.41 USR1 it says.
You're using a version of spectre built 6.5 years ago. That doesn't make sense, given that there have been 7 major releases of the analog simulators since then. In 2004 we split off the simulators from the IC stream, and have released MMSIM60, MMSIM61, MMSIM62, MMSIM70, MMSIM71, MMSIM72 and most recently MMSIM101. The version of spectre in the IC5141 has not been maintained for some time (although since you're using USR1 which is from December 2004, that's irrelevant).
I would suggest you try installing a recent MMSIM version; I would expect the problem to go away as there have been many advances in the last 7 years.
Much agreed. However, this version is provided by our university, who I imagine have no incentive to update unfortunately.
I did solve my internal failure issue by taking my wires which were previously sharing common names, but not connected, and hard-wired them together in the schematic. I'm sure the circuit logic is correct, but somehow the provided driver circuit and my adder circuit were not simulating well together. Thanks again for the help.
I can't really see how that would help, since presumably the resulting netlist would be the same - spectre has no visibility of the schematic. I can only assume that something subtle changed which allowed it to proceed.
Most universities would be using more recent versions than that - even if the IC version is old. But I agree, things may be more constrained there.