The Designer's Guide Community Forum

The Designer's Guide Community Forum
https://designers-guide.org/forum/YaBB.pl
Simulators >> Circuit Simulators >> correlation between pss and transient simulation in spectre
https://designers-guide.org/forum/YaBB.pl?num=1294651667

Message started by subgold on Jan 10^th, 2011, 1:27am

Title: correlation between pss and transient simulation in spectre
Post by subgold on Jan 10^th, 2011, 1:27am

i have a circuit which is perfectly stable (no ringing/oscillation observed, and the output waveform is a perfect periodic signal) in transient simulation. however, running the same circuit with pss can never get convergence, even with a much loose tolerance (>100) than the transient. the largest current devation always occurs at ideal DC sources, which are used for supply voltage or some biasing for the stimulus.

could somebody kindly give some hints about what the problem is, and how to get the pss simulation to work? thanks a lot.

Title: Re: correlation between pss and transient simulation in spectre
Post by ssahl on Jan 10^th, 2011, 3:54am

Hi, sounds strange...

Have you set any tstab? Maybe your circuit has some metastability inside it that pss tries to find a steady-state around. Running a couple of periods with a transient can help that. When having tstab>0 you can also save the initial transient waveforms to have a look at them in case you pss does not converge.

Title: Re: correlation between pss and transient simulation in spectre
Post by ssahl on Jan 13^th, 2011, 11:19pm

Hi subgold!

Have you got any success in your pss simulation? Just interested to know what the problem was.

Title: Re: correlation between pss and transient simulation in spectre
Post by Geoffrey_Coram on Jan 14^th, 2011, 6:50am

subgold wrote on Jan 10^th, 2011, 1:27am:

the largest current devation always occurs at ideal DC sources

Have you looked at the waveforms for those DC sources? (You said only that the circuit output is nicely periodic.)

Also try shifting the pss by a partial period, in case there's some sort of transition right at the boundary where pss wraps around.

Title: Re: correlation between pss and transient simulation in spectre
Post by subgold on Feb 18^th, 2011, 1:34am

sorry for the delayed comments.
i dont have any success in this issue. shifting the pss period by tstab can temporarily solve the problem, but doesnt work for corner verification as different corners could possibly generate different starting point in pss.

i am just wondering if the tran and pss simulation have the same default integration method and accuracy parameters?

Title: Re: correlation between pss and transient simulation in spectre
Post by sheldon on Feb 18^th, 2011, 5:28am

Subgold,

You have not really provided enough information to debug the
problem. There are some other factors to consider, for example,
there are some circuits that can cause issues for PSS. In the past,
I have had trouble getting PLL to converge with the lock detect
present since it has a hidden state. Another example would be a
Pipeline ADC with a register to detect that the pipeline is full,
another hidden state. Also have you tried APS-RF, my experience
is that it has more robust convergence.

Best Regards,

Sheldon