The Designer's Guide Community Forum

The Designer's Guide Community Forum
https://designers-guide.org/forum/YaBB.pl
Simulators >> Circuit Simulators >> Simulation results change as a function of simulation length
https://designers-guide.org/forum/YaBB.pl?num=1219769042

Message started by Noam on Aug 26^th, 2008, 9:44am

Title: Simulation results change as a function of simulation length
Post by Noam on Aug 26^th, 2008, 9:44am

I see very strange behavior of Spectre simulator and I hope I could get some help.

I am simulating switched capacitor circuit and I get different results as a function of the simulation length (transient analysis). For example, when the simulation stop time is 100ns I see different wave shapes (between 0 to 100ns) comparing to the exact same simulation when it runs for 400ns (again looking in the range of 0-100ns).

I ran it under moderate accuracy, tried to change to conservative with no change, try to reduce the maxstep (by several orders of magnitude) and no change. Does anyone saw such behavior? Is there any other parameter which depends on the simulation length (besides maxstep)?

Thanks.

Title: Re: Simulation results change as a function of simulation length
Post by loose-electron on Aug 26^th, 2008, 1:42pm

A suggested reading - chapter 4 of "The Designers Guide to Spice & Spectre" (by Ken Kundert) will give a lot of insight into transient simulation errors.

Certain types of circuits (and switch cap circuits are amon them) will give different results every time you run the simulation due to truncation errors.

Spice/Spectre is not infallible.

Title: Re: Simulation results change as a function of simulation length
Post by Ken Kundert on Aug 28^th, 2008, 3:35am

Default rise and fall time for sources are a function of the simulation interval. Explicitly specify them to eliminate this dependency on the total simulation time.

-Ken

Oh, one more thing. Do not ask simulation questions in the design boards. Please use the simulation boards instead. Thanks!