half-sine

Well, the FT of a Gaussian is another Gaussian, so you don't need too many sinusoids to do a good job. Depends on the rep rate you want, of course, but you can always cut one pulse out with switches.

Another approach would be to use the LTspice switch with negative hysteresis, which makes a nice smooth differentiable curve. Apply a constant voltage across the switch, stick a triangular pulse into the control input, and subtract off a little of the triangle to cancel the cusp in the output. It won't be perfect, of course, but then neither will the laser beam (especially if it's from a diode).

Cheers

Phil Hobbs

Half_Sine.jpg

Uh -- use the equation for a Gaussian in a behavioral source? Or do you need to launch the pulses at arbitrary times?

Version 4 SHEET 1 880 680 WIRE 160 160 160 96 WIRE 160 272 160 240 FLAG 160 272 0 FLAG 160 96 Vpulse SYMBOL bv 160 144 R0 SYMATTR InstName B1 SYMATTR Value V=exp(-((200 * (time - 0.02))**2)) TEXT 126 296 Left 2 !.tran 0 50m 0 10u

Sine squared, or better, cosine squared might be a better starting point.

You mean a pulse that looks like a Gaussian distribution... the bell curve. Or something with Gaussian filter.

Hmm I guess the two are the same...

Well if the pulse response of a Gaussian filter is the Gaussian distribution, Then hit your filter with a (narrow) pulse and not a 1/2 sine.

George H.

Well, that's approximately perfect. Thanks. I don't use Spice enough to use all the tricks.

Better yet: instead of a half sine, use a full sine.

Offset the sine by +1, start it from -90 degrees and run it to 270 degrees phase, that will likely need less filtering.

Dimitrij

See...

and "RealPulse.pdf" on the Simulation Tools & Macros Page of my website. ...Jim Thompson

You could use Python or Matlab to generate the Gaussian and write it to a .wav file and use that as input...

Tim's B source that depends on the TIME variable is perfect for the simple Gaussian. The .wav thing would be good for something more complex.

Is LTSpice sophisticated enough to convolve two sources with each other? Repeatedly convolve a square pulse with itself enough times and you get a Gaussian...for a square it actually converges really quickly.

Oops nevermind, looks like you can plug the equation right in! Cool.

Someone should use that to run audio through an LTSpice simulation of a tube preamp, and also a solid state preamp. See if audiophiles can tell the difference.

Once computing power gets fast enough someone is going to make a lot of money selling "component level real-time digital tube preamp simulations" I think.

Audiophools will still "hear" the difference.

Dip it in liquid oxygen and charge $12k for it and it will be the best thing they've ever heard.

Make sure that it doesn't have any graphite foam components. Charcoal soaked in liquid oxygen used to be an industrial explosive.

Liquid nitrogen is much more widely available because it is safer.

I've already done that. I had built a small vacuum tube guitar amp and didn't like the tone, so I recorded my guitar into a wav file, threw the schematic into LTSpice, and started changing component values and topologies until I got something that behaved a little better. Then I pulled out the soldering iron.