Designer’s Guide Community

New Member

Offline

Posts: 8

Wrong simulation results given by Spectre Turbo
Sep 06^th, 2010, 1:09am

Hi all,

I found when running spectre turbo for DC analysis in some circumstance (ex. low supply voltage, low temperature or transistor corners), it gives different DC operating point compared to spectre w/o turbo enabled. I think the results calculated by spectre turbo is wrong because if I just increase the temperature or voltage a little bit (even 0.1 degree), the results between spectre with and without turbo enabled are the same.

The circuit is an OTA with CMFB simulated in closed-loop configuration. Wrong DC operating points usually show the output common voltages saturate to supply voltage.

I tried to tighten the reltol but there were no differences between results. Is there any possible reason causing these wrong simulation results? How to fix it? Thanks.

p.s. the simulator is MMSIM72

Back to top

IP Logged

Geoffrey_Coram

Senior Fellow

Offline

Posts: 1998
Massachusetts, USA

Re: Wrong simulation results given by Spectre Turbo
Reply #1 - Sep 7^th, 2010, 5:30am

Is it possible you have multiple dc operating points?

You could try taking the "wrong" solution from +turbo and giving that to plain spectre and seeing if it converges. If so, there are multiple operating points, and you need some sort of nodeset to push +turbo to find the right one. Hopefully, you can find a few nodes to set, rather than resorting to using plain spectre just for the dc op pt.

Back to top

If at first you do succeed, STOP, raise your standards, and stop wasting your time.

IP Logged

frozenduck

New Member

Offline

Posts: 8

Re: Wrong simulation results given by Spectre Turbo
Reply #2 - Sep 7^th, 2010, 8:04pm

Hi,

Thanks for your reply.
Taking the results and giving them to Spectre does it mean using some Vdc to force these nodes� voltage and checking if it converges?

Because Spectre turbo gives this possible DC op pt, it means my circuit may work on this unwanted point and I need to make sure (using aux ckt?) it will not happen, right?

Best Regards

Back to top

IP Logged

Geoffrey_Coram Senior Fellow Offline Posts: 1998 Massachusetts, USA	Re: Wrong simulation results given by Spectre Turbo Reply #3 - Sep 8^th, 2010, 6:11am frozenduck wrote on Sep 7^th, 2010, 8:04pm: Taking the results and giving them to Spectre does it mean using some Vdc to force these nodes� voltage and checking if it converges? No, not Vdc. Look at spectre -help dc for information about readns and nodesets.
Back to top	If at first you do succeed, STOP, raise your standards, and stop wasting your time. IP Logged