AW: About NATIONAL APP NOTE #299 VCO

Haible_Juergen#Tel2743 HJ2743 at denbgm3xm.scnn1.msmgate.m30x.nbg.scn.de
Fri Apr 26 01:55:00 CEST 1996


> IMO, the
> essence of AN-299 is the 3046 temperature regulating scheme. The VCO
> itself is not what I'd use for serious work.

Joachim is absolutely right here, IMO.


> The most important modification, though, is to place a large
> capacitance across the LM329 reference.

Honestly, I wouldn't use the LM329 reference at all. IMO, this was just
a self-advertising of another NS component. All you need is a fairly
constant voltage in comparison to the temperature-dependant
be voltage to do a proper temperature regulation.
I built 2 VCO's of this kind some years ago (for JH-2 modular, which
was never fully completed). I also used a completely different
VCO engine (window comparator), and a slightly different transistor
array (some TCA ??? thing). But I used the heating scheme of
AN299, only that I derived the reference voltage directly from the
15V rails. And yes, I bypassed the resistor divider with a large
capacitor. To keep ripples out is much more important than a
very precise reference.

JH.

PS.:
I've seen some Moog schematics, where there is a trimpot for the
reference voltage. In this case, you can even *choose* the temperature
you want to run the array with!














The current drawn through
the 600 ohm resistor is relatively large, and also dependent upon the
output waveform. If you monitor the voltage across the LM329 reference
without decoupling, you'll see saw-shaped fluctuation. What we want
is a stable, ripple-free 6.9V reference, so decoupling here is essential.
I connected a 100uF/16V electrolytic in parallel with the LM329, and
this made the circuit behave properly.

Another thing, I found large positive spikes at the sawtooth retrace
phase. These might have been caused by bad circuit layout, didn't have
time to experiment more. The phenomenon might also be due to the fast
response of the retrace transistors. Retrace time was found to be approx.
1 microsecond. To get rid of the spikes, I connected a 1N4148 diode from
the inverting input of the op-amp to ground (cathode to ground). This
also improved high frequency stability a lot. The VCO now oscillated well
above 40kHz. The drawback to adding a diode is that it might affect
temperature stablilty.

Also, while it might not be obvious from the drawings, the op-amp(s) have
to be powered by a dual supply for the circuit to oscillate at all.

 -  -  -

 -joachim



More information about the Synth-diy mailing list