Re: gEDA-user: How to?

[Al Davis <ad82@xxxxxxxxxxxxxxxx>]

On Tuesday 12 July 2005 11:05 am, Harold D. Skank wrote:
> I have created a state variable BP filter.  I can run the ac
> analysis using ngspice, and that seems to work very well.  It
> has allowed me to specifiy the circuit values to get the band
> shape that I need.  Now though, I would like to see the
> circuit response to a 5-cycle truncated sine wave input,
> since this is the "real" signal that the filter will see in
> operation.  How do I specify this as a voltage input and how
> do I actually run the simulation?

Why don't you try gnucap?

Gnucap has a "generator" that lets you set up things like this.

.generator  freq=1k delay=.001 width=.005
vin (1 0) generator

On Tuesday 12 July 2005 11:18 am, Stuart Brorson wrote:
> 1.  Model the truncated sine wave as a piecewise linear
> signal, use a PWL voltage source as the excitation, and then
> use a .tran analysis to examine the result.  This will give
> you a bunch of spurious harmonics whose magnitude and
> frequency will depend upon how finely you discretize the
> signal, but perhaps you don't need to worry about them in
> your analysis.

That's really rough unless you use lots of points.  Gnucap adds 
a "FIT" that is like "PWL" except that you can specify the 
order of interpolation.  Specifying an order of 3 makes it fit 
with cubic splines.  You can also specify the boundary 
conditions.  You can specify a knot by giving the same point 
twice.  Still, I don't recommend this for what you are doing.

On Tuesday 12 July 2005 11:18 am, Stuart Brorson wrote:
> 2.  Use a voltage controlled switch to connect an AC voltage
> source to your filter for the desired time, and then
> disconnect it afterwards.......
>.......  I believe that ngspice will support 
> voltage controlled switches too.  However, I don't have
> first-hand experience with this.

Of course it does.  It starts with "S". Spice 3 does, so as far 
as I know all Spice derivatives except some really old ones do.

Another possibility is  to use a poly or a "B" device as a 
multiplier.

On Tuesday 12 July 2005 11:29 am, Gene Heskett wrote:
> Just off thet op of my head, Harold, a state variable BP
> filter is usually going to have too high a Q factor to
> adequately respond to a 5 cycle burst. ............

Tone burst tests are often used to evaluate the performance of 
filters.  A state variable filter can have any Q you want, even 
very low.  I might use a cheaper design for low Q, but nothing 
beats the 3-op-amp design for ease of tuning.