On Sat, Mar 14, 2009 at 02:33:50PM +0100, Stefan Jahn wrote:
> Am So, 8.03.2009, 17:22, schrieb Oswald Buddenhagen:
> > so i'm trying the next one: make a transient simulation (to average and
> > plot the results ultimately). to keep both data size and accuracy
> > reasonable, i want to use a constant number of periods and data points,
> > so i have to vary various parameters with frequency. that does not work
> > - the stop time is taken from the first value of the sweep; the time
> > axis is not dependant on the sweep parameter (attachments 3&4). ideas?
> > (yes, of course this would cause problems for the visualization ...)
> I see the "bug". Also I fixed in in CVS right now. Now these times
> can be steered by sweep variables correctly.
> But, as you already noted there is now the problem about the
> visualization. The "time" vector keeps now only representing the one
> from the first sweep. But, could you live with that?
even of better than expected - the time axis is actually stretched, so i
have a cool superimposition of the results. :)
to round it off, one could put two labels on the axis, noting the
minimal and maximal values.
the alternative presentation i though of originally: do not stretch the
axis, so the higher-frequency data sets would just end earlier. i
haven't considered actual use cases for that, but if some are found one
could make it switchable through a property of the diagram.
some more thoughts on the "ac simulation through transient simulation":
instead of having the user calculate a sensible number of cycles based
on some initial results, the simulator could have an "auto-stop" mode:
as soon as some named parameters are stable within some tolerance from
one period to the next one, the simulation stops.
of course, this all helps relatively little when the final summing
simply does not work as detailed in the other mail. oh, well.
did you give my patches some consideration?