C.Anderton's Pulse Width Multiplier

[Martin Czech]

> > One could even do freqency multiplication, with no pll lag, though the 
> > accuracy will get down at the high frequencys due to comparator slew rate,
> > I think. E.G.: multiplication by 7 is possible by scaling the input, so that only
> > 14 comparators are used, the odd ones with -1V output, the even with +1V
> > 
> I don't understand how you would be able to multiply a frequency with 
> this.
> Although, I think I see how you might produce square waves at X times 
> the input freqency.
> 
Yes, that's it ! You`ll get square waves at n times the input frequency,
and this is frequency multiplication. Other waveforms may also be approximated.
Of course, due to nonideal 
behavior you will never get rid of the fundamental, especially at
higher frequencys. But it is enough to get beautifull, very strong  partials,
as my experiments with wavetables did show. 

Example :
              /     /
             /     /
            /     /
           /     /
          /     /
         /     /
input   /     /     saw wave


window1 H_____H______  6 comparator windows equally distr.
window2 _H_____H_____  odd windows -1V
window3 __H_____H____  even windows +1V
window4 ___H_____H___
window5 ____H_____H__
window6 _____H_____H_

Output  _-_-_-_-_-_-_  square 3 times input frequency with some fundamental dirt.

> It sort of sounds like you mean to clock an analog sequencer at audio 
> rates using a VCO.

No, because this would divide the audio clock, not multiply.
I would need 16 clocks to step arround a 16 syep sequencer, and
even if each pot has opposite voltage value this means divide by 2.


clock  h l h l h l h l ...
output l l h h l l h h ...

My english is very bad...
I hope the idea is clear now.

m.c.