Re: [Ngspice-users] control of maximum step size during transient analysis

[Attila K. <at...@ki...>]

On Tue, 30 Aug 2016 13:23:10 +0200
mh...@ia... wrote:

> > I'm doing some transient simulations that require very small step sizes
> > during some critical parts (like <0.1fs) but there are some stretches of
> > time, where a cruder step size (~1ps) would be good enough. As the
> > simulation
> > is taking a lot of time (even a simple test circuit takes half an hour)
> > I would like to have a possibility to change the step size to speed up
> > the simulation. Unfortunately, I couldn't find anything in the manual
> > and google didn't return any usefull results. Is there a way to achieve
> > this?
> 
> SPICE does not support fixed step integration.
> The simulation time steps are adjusted based on
> accuracy you request with RELTOL, ABSTOL etc..
> The best you can do is *limit* the maximum time
> step that is taken, but that doesn't help you here.

Actually it does. What I am doing is analysis of the
metastable point of latches. If the step size is too
large, then the result of the simulation is off by quite
a bit (I've seen a factor of two in metastability resolution time
depending on the set maximum step size). Restricting the maximum
step size helps to keep these errors small.

> A lower requested accuracy does not necessarily
> lead to faster simulation when there are convergence
> issues.

I do not want a lower accuracy, but a faster calculation.
I know that the first 50-100ps are uninteresting, because
nothing happens during that period. This first period is
only there to ensure that any "inaccuracies" from the DC
analysis settle such that the transient analysis starts
from a steady state. 

 
> The 0.1fs you mention is so small that it should violate
> the engine heuristics. I am amazed that you can get
> that low, but maybe you use special .OPTION settings.

Well, the output looks correct and I don't get any warnings.
So i guess ngspice does the right thing :-)


			Attila Kinali

PS: Why did my mail take 20 days to get through the moderation queue?
Is that normal? I already thought my mail was rejected.


-- 
It is upon moral qualities that a society is ultimately founded. All 
the prosperity and technological sophistication in the world is of no 
use without that foundation.
                 -- Miss Matheson, The Diamond Age, Neil Stephenson