Re: [Gnucap-devel] user input to a running simulation

[al davis]

On Wednesday 18 April 2012, Pawel Ludwikow wrote:
> Thanks for the idea. It works even on vanilla gnuspice.

gnucap is not spice ...

> However to my surprise it produces exactly the same data as
> my hack:
> 
> -----[part]
>  2.         5.         4.9672
> #Time       v(1)       v(2)
>  2.         1.         4.5227
> -----[part]
> 
> so it looks that my hack hasn't introduced this behavior.

What happens here is that the values at the start time are 
recalculated using the new value.

Remember .. in transient analysis, nothing can change 
instantaneously.  The best you can do is to have it ramp from 
the old value to the new value within one time step, and time 
steps are variable.

In this case, the previous time step is one-back from the 
previous run, or 1.9 seconds.  So, v(1) is 5 at 1.9 and now 1 at 
2.0.  The voltage v(2) reflects that, so by 2.0 seconds it is 
already starting to go down.

Since changes like this do not have proper time step control, 
you get different results with different "integration" methods. 
All are equally correct (or incorrect) because the actual 
waveform between points is unknown.

Probably the best way to do what you want is to write a new "bm" 
plugin, where tr_eval would get a new value from outside.