Re: [Help-gnucap] Re: using a pipe to make gnucap accept input from other program

[al davis]

On Thursday 11 January 2007 02:33, Hugo Coolens wrote:
> al davis wrote:
> > The reason Spice does it the way it does goes back to the
> > small computers of the 70's, where you needed overlays to
> > fit a large program into small memory.  It is a multi-pass
> > system. The "raw file" is the interface between the
> > simulator engine and a post-processor.  Computers are
> > bigger now, so we don't need to do it that way anymore.
>
> I understand what you mean but wouldn't it be nicer to have
> gnucap do itself the task of choosing the right statement
> order when using batch-mode?

No.

It would be very easy to do, but I prefer it the way it is.

You might also ask another question, .... why is "op" required 
before "ac", or why does "ac" not give the same results every 
time?

If you don't do an "op" immediately before "ac" you will get the 
operating point of whatever the most recent transient, dc, or 
op finished with.   This also is deliberate.  It is this choice 
that allows you to evaluate the ac perfomance with a variety of 
operating conditions.  Think about a class-B amplifier with 
negative feedback, that has a stability problem that shows as a 
glitch in the waveform at a particular level.  Only gnucap lets 
you home in on that point and do a detailed AC analysis there, 
at an instant in time.  Run the transient analysis to the 
glitch, stop it there, then run AC.  You can run several 
sweeps, move the probes, and run again, and see lots of detail 
at that point.

But back to the original comment.  I might want to do an overall 
analysis,see the results, then home in on some detail.  Let's 
take that same class-B amplifier.  I want to see a plot of gm 
of the driver stage, in time, with signal, or the nonlinear 
input capacitance, which does strange things as the output 
stage goes through its crossover region.  This kind of detail 
is only available when you have fine control that gnucap 
provides.

Even in batch, maybe I want to try several variants of the 
circuit.  First, a basic circuit, then maybe add an emitter 
follower, and see how that works, then change it to a 
complementary pair and run again, all in one batch file.  Can 
Spice do that?  Can Spice show me, in one batch file, how the 
AC performance for positive signals which use the NPN 
transistor is different from the AC performance for negative 
signals that use the PNP? in and near the crossover region?

There is also the possibility that "batch" mode isn't really 
batch, but an interaction with another program, perhaps an 
optimizer, perhaps it is doing a more complex analysis like 
periodic steady state, distortion of an oscillator.

All this is only available when you, the user, have detailed 
control.  This is lost when there is any automated rearranging.