[sdiy] More VCO fun

[David G. Dixon]

(Warning: very long posting)

I've spent the last week or so re-thinking my 2164 Expo VCO.  I wasn't
completely happy with the previous version, for two reasons:

First, the sync was upsidedown, in the sense that it required a falling edge
to trigger it.  This is OK in audio mode (albeit pretty inconvenient,
because a saw won't work as the master) and is definitely not OK in very low
or sub audio mode, when one wants to gate the sync input for a crisp bass
attack or for uniform LFO effects, since gates are invariably rising edges.
I didn't want to add inverter chiclets to the sync attenuator pot, so I had
no choice but to add an opamp to the circuit board.

Second, the Linear FM was also upsidedown.  I had grown used to this with my
Thomas Henry VCO-1's, since the Linear FM is also upsidedown in that design,
and would require an inverter to change it.  However, changing it in the
2164 Expo VCO did not require an inverter, and if I hadn't been a bear of so
little brain, I'd have wired it up correctly in the first instance.

Also, since I now required one more opamp to invert the sync, I decided to
give in to Ian and use an actual comparator for the saw core (LM311).

Now I've got a more or less perfect 2164 Expo Sawcore VCO.  However, I had
fits trying to test and debug the circuit, because two weird things were
happening: the frequency was wavering randomly and showing a dependence on
the setting of the sync attenuator knob (even though nothing was connected
to it), and the frequency was very sensitive to the temperature of the JFET
switch.

At first, I thought that the sync problem might have been caused by
layout-related capacitance issues, since the sync output trace ran directly
underneath a jumper carrying the comparator output, but I did a calculation
and figured that the capacitances involved were on the order of 1pF --
hardly enough to cause any problems in the audio range.  Also, testing a
22pF capacitor between those same two points on a previous circuit caused no
problems.  I swapped chips, tested caps, did all sorts of things, and the
problem persisted.

Concerning the JFET temperature sensitivity (which caused the pitch to
increase by several semitones upon touching the JFET with my fingertip),
this really had me stymied, since my other circuits don't show this
sensitivity.

Anyway, to make a long story short, I had used an insufficiently large
resistor between the integrator and the comparator, thus giving too short of
an RC time constant to fully reset the integrator cap.  I knew this to be
the case, but I didn't think that it would cause any other problems.
However, when I finally increased the resistor value to bring the reset
point to within 100mV of ground, all of the other problems disappeared as
well!  The frequency is now totally insensitive to JFET temperature, and the
sync circuit is having no effect on the frequency.

I find all this to be very interesting (now that I'm no longer pulling what
little hair I have left out trying to fix it), and I'm going to have to
analyze it sometime.  Has anyone else here ever observed these problems with
JFET reset time constants in sawcore VCOs?