thru zero VCO questions

[Rene Schmitz]

At 00:40 13.09.99 -0700, Don Tillman wrote:

>I am completely fascinated by the problem though.  And I had a chance
>to think about some of the issues over the weekend.

I did some thinking as well, and I now see that the problem is more serious
than I thought. I did some simulations and the circuit crashed in some cases.
My simulation circuit was just the comparitor+FF arrangement, and a opamp
integrator, modelling the modulation/switches with multipliers.

>I think the current approach, the integrator/flop/invert-the-signal
>setup, is doomed to have a region of instablity or a region where the
>oscillator just dies.  Small offset voltages will ruin the performance
>of the circuit in several ways and the best workaround appears to be
>spewing pulses to the flop, as if to say "Will this polarity work?
>No?  Try this one.  No?  Okay let's go back...".  

Thats how I see it now.

But I now look at the problem differently. There are two independant, and
unsyncronized, mechanisms which can change the direction. First is the sign
change on the FM input, the second is the thresholds are crossed. 
Now when both occur "simultaneously" we get the trouble. Instead of
applying extra pulses, the trial and error method, one should sort the
extra pulse out which is the one unneccessarily produced by the comparitor.
I think when the input is near zero, and the cap voltage slightly below the
threshold one should block the comparitors pulse. This would be done by a
zero crossing detector which goes high for a small voltage region arround
zero, and a second set of comparitors slightly *below* the thresholds. Now
when the voltage is near the threshold, and the FM input is near zero,
nothing should happen.
Two things can happen, the FM input can wander into the small window, and
can leave it with either the opposite polarity, then the sign change
occured, and the direction is reversed by that. Or the input can suddenly
reverse direction, and wanders out of the window with the same polarity as
before. The comparitor would delay the toggle then until the FM input has
wandered out of the window again. This delay would cause only some little
overshoot, which is probably not a problem. Maybe the extra set of
comparitors below the thresholds are not even needed. The only drawback I
see is that it is impossible now to modulate with voltages (dc/ac) smaller
than the window. Especially the offset could cause a problem, when no input
is supplied.

>I think an *elegant* thru-zero VCO design is going to have to work in
>two dimensions.  By "elegant" I mean a VCO that works just as smoothly
>around the peak voltages as it does over the rest of its range.  And
>by "two dimensions" I mean you'll need two capacitors working in a
>quadrature arrangement.  I don't have a schematic to offer yet, but
>I'm thinking of having two integrators running through four states:
>
>   state 1: integrator 1 rises, integrator 2 holds at -Vref
>   state 2: integrator 1 holds at +Vref, integrator 2 rises
>   state 3: integrator 1 falls, integrator 2 holds at +Vref
>   state 4: integrator 1 holds at -Vref, integrator 2 falls
>
>This way you can run the oscillator forward and backward, however
>gradually, without hitting any funny states.

Well, this sounds interesting. But I'm not sure I got the idea. By
quadrature you mean that one integrator crosses zero while the other one
hits the reference? 
And the above states are just for the moments where the references are hit?!

Bye
 René


error:asciiartisnotfully  \ uzs159 at uni-bonn.de
supportedwithyourcurrent   \ http://www.uni-bonn.de/~uzs159
displaydriversettings...    \ http://members.xoom.com/Rene_Schmitz