AW: Re: AW: Re: AW: 2-comparator sawtoot

[Haible_Juergen#Tel2743]

> True, true, but then the second comparator and FF are superfluous, the
> same result is obtained with just one comparator controlling the switch:
> As the capacitor voltage drops below the treshold the comparator switches
> state within 200ns. Then as the cap gets (dis)charged the comparator flips 

> again, but not before another 200ns have passed.

Right. If You are very sure about the different timings, You could build
it like that. But comparator delay times might likely be *guaranteed* times,
that means an actual device might be much faster. Personally, I wouldn't
rely on this. But I am quite sure it might work - it's only that I always 
want
to be way on the save side.


> I was aiming for a *very* short discharge time, making even the
> discharge time compensation unnecessary.

But then there still is this bulk resistance error from the exponentiator,
that normally can use the same trimpot to be compensated ... You might
not spare anything. But OTAH, Saw->Triangle conversion works
far better with short discharge slopes.


>Franco in his book
> shows a circuit with a cap over the comparator, which keeps the comparator
> in it's state for an amount of time, ca 0.7 microseconds. I think he never
> incorporated the 2x200ns response time of the comparator! I now think that
> cap is not necessary at all, but I'll have to experiment to really find 
out.

This (the cap) might be another way to be on the "save side" with varying
comparator delays. I think I have also seen this in some Moog circuits.


> The thyristor option looks interesting too. What properties does it have
> considering speed and 'switch resistance'?

I never calculated it. But I think both the Roland Jupiter-8 and the Korg
Polysix use thyristors - well, thyristor surrogates built of one NPN plus
one PNP.

JH.