Some results with Eric Barbour's TS-2 tube VCO

[Eric Barbour]

>1. Omit the 4.7K cathode resistor on the thyratron.  It doesn't seem
>to serve a useful purpose since the "self-bias" voltage it provides
>drops low enough near the end of each cycle for the 2D21s I am using
>to start drawing grid current by the time the tube fires (current
>starts at a grid voltage of about -1V wrt the cathode).

It's there to provide some degenerative
feedback and improve drift. Supposedly this resistor is only active
during the discharge phase of the sawtooth. A cathode resistance was
apparently considered an undesirable parasitic when thyratrons were
used as motor controllers and relay drivers--wasted energy.

(It's also there to limit anode current in the event of a problem,
such as a shorted anode resistance or a grid-anode fault.
A 2D21 can suck more than 0.5 amps of fault current.) It can be
left out if desired.
A 1/8 Amp fastblow fuse in the plate supply may be a good idea.

>2. To increase the range, the "diode" voltage, which controls the
>firing point, must be made more sensitive to the CV voltage.  This can
>be done by decreasing the 12K resistor between the CV cathode follower
>and the thyratron grid.  This drops more of the CV across the "diode"
>and less across the resistor (note that omitting the cathode resistor
>also increases the sensitivity because it was also dropping some of
>the CV voltage).  I changed the 12K to a 4.7K.  I didn't really want
>to go smaller because the thyratron grid spikes to 30V or so when it
>fires, and that resistor provides current limiting.

This series grid resistor is there to allow injecting a sync
pulse if desired, as well as for limiting grid current. It can be
decreased in value if the sync is applied to the cathode-follower
input. You may need to buffer/amplify the sync pulse.

>3. The cathode follower adds about 2.5 to 3.5V to the CV (with the
>12AT7s I tried).  The range can be increased if the shift isn't as
>large.  I replaced the 150K cathode resistor in the CF with a
>2.7K/150K divider to drop the voltage somewhat.  (This may be a bad
>move since the CF output now contains a fraction of the -150 supply
>ripple.)  Note that this 2.7K is in series with the 4.7K and so
>reduces the sensitivity; I may drop the 4.7K a bit to get it back.
>Also note that the value of 2.7K was hand picked for the particular
>12AT7 I am using.  Larger values drop too much voltage and prevent
>oscillation by making the thyratron's grid so negative it never fires.

An op-amp buffer would work even better as a thyratron grid driver.
This method is limited by the thyratron's grid sensitivity--many
gas tubes require a CV which exceeds the 36v power-supply range of
most op amps. Also, I was trying to do it all-tube, just to see
how well it worked with an all-retro design. The cathode follower
is a good synergy with the thyratron--no danger of damage if the 
thyratron fails or due to grid-current spikes, good power supply
rejection, simple, and an aging tube will tend to drift along with
an aging thyratron.

The 2.7k resistor (Which I assume is between the cathode and
the grid resistor) is not really needed. It's easy to provide
an offset voltage in your keyboard resistive divider....

>Instead of mod #3, I am considering trying to shift the range of the
>control grid values by tweaking the voltage on the shield.  Should be
>interesting.  The shield could also become a point for injecting a
>pitch modulation voltage.

That will work. Supposedly, the shield grid is less sensitive than
the control grid, so more voltage swing is required. Using it
separately will also probably decrease the CV sensitivity of the
oscillator.

>4. This has nothing to do with the range.  I changed the 7.5K resistor
>in the 0B2 regulator circuit to 3K to ensure the 0B2 stays within its
>5 - 30mA spec for all settings of the tune pot and for worst case
>supply and regulator voltages.  I'll probably increase it somewhat as
>the current draw seems a bit excessive, but I think 7.5K runs the tube
>out of spec.  Maybe that's not a problem though.  Eric?

As shown, the 0B2 isn't extinguished unless you REALLY drag the
thyratron down. As when, for example, setting the tune pot to
minimum and putting 0v into the CV input. This is a compromise
of tube performance, lifetime and control range. And again, I wanted
to see it done all-tube. A solid-state regulated +100v supply
would give much better performance (not necessarily better
sound--I suspect that 0B2 limitations have something to do with
the fat sound I'm getting). Besides, 0B2s are almost indestructible
and do not care about heat, the way a large shunt zener diode
would. You would probably need to clamp a zener to a heatsink.

Again, the tube is a good synergy with the other tubes.

>Weirdly enough, even with all its non-linearities, deionization
>delays, etc., Eric's circuit was surprisingly linear.  I was getting
>about 8 - 12Hz/V over a 60 to 150Hz range.  I haven't checked it out
>with my mods yet though.  And I also haven't listened to it yet
>(sheepish grin).

For shame! Still, thanks for giving it a try. You're the first person
on synth-diy to actually try it and report back.

Side note: I tried this circuit with a 5823 cold-cathode triode.
It worked, MUCH less stable, much less CV range. A considerable
positive "grid" offset is required to reach the oscillation
trigger point--40 to 90 volts. The 5823s also vary from
unit to unit enormously.

This tube is like an argon discharge lamp with an extra
electrode. It sure looks pretty when lit up, but the hot-cathode
thyratrons are superior in other ways.

Eric Barbour
Svetlana Electron Devices
Portola Valley CA USA