thru zero VCO questions (4th attempt)

[Haible Juergen]

	>I wan't to develop a thru zero capable VCO, and I have some
questions
	>regarding it. I can see from "prior art" that one uses integrators
that can
	>go up and down, depending on the polarity of the FM input signal.
In the
	>ones I've seen the FM input is rectified, fed into the reference
input of
	>the expo convertor, and the whole oscillator is switched between
forward
	>and backward. Why is it done that way? I think all one would have
to do is
	>to replace the schmitt-trigger with a window comparator which
toggles a
	>D-flipflop between forward and backward when the thresholds of the
	>comparators are met. The output of the flipflop would be switching
the
	>polaritiy of the FM input sum. Or am I missing something here? 
	>My brain already hurts, so any comments highly welcome.
	>
	>Bye
	> René  

I think there was in fact a circuit like this in Electronotes. If memory
serves,
the S and R inputs and comparators were used for the VCO's internal
loop, while the clock input was used to toggle between up and down
"from the outside". The zero crossing of the modulation signal was detected,
then short pulses were shaped with exor gates (one short spike for each,
positive or negative zero crossing), and this was used to toggle the
flipflop. I don't have the original circuit at hand, but you can get an idea
of
the pulse shaping part at
	http://www.synthfool.com/diy/hj2vco.gif
where this method is adapeted for a CEM3340 (don't use these for DIY
projects
anymore !).

For more recent thru zero VCOs I have abandoned this concept, however.
The major problem is that you might loose a pulse when your modulation
signal just touches zero and only goes slightly negative. This can add some
"dirt" to the signal at small modulation depth. Not always unpleasant, but
not very precise either. I have not made direct comparisons with VCOs
of the other method (up / down switching with the sign of modulation
voltage) - I've built them, but for different application.
But think of it: What you really need is a *switch* function to change the
VCO spinning direction, an what you have is a *sign* function of the
modulation signal, so that's exactly what you need. Why make a double
pulse from the sign signal, and restore the switch function by toggling
a flipflop with this double pulse ? (The 3340 *needs* this, but a discrete
solution doesn't.)

JH.