Tube Challenge!!

media at media at
Sat Nov 20 20:27:46 CET 1999

>This would be far in advance of current technology. The current
>vacuum tube VCO are sawtooth relaxation-oscillators, not capable of
>Sines or Triangles. If you can only have one waveform, sawtooth is
>probably the one to have. To do a sine oscillator, several approaches
>could be taken - voltage variable capacitances, variable inductors etc..

Actually, I have seen schematics for tube relaxation oscillators that can
produce pulse and sine waves.  However, these designs might not be
musically useful.

The problem I see with tube oscillators is that they aren't easily voltage
controlled.  Afaik, even the commercially produced Phattytron uses digital
circuitry to switch between hand-selected resistor values for various MIDI

I would like to see a tube oscillator that can be used with existing CV
sequencers, as well as the various modulation sources typically found in
modular synths, to produce non-standard tunings, oscillator sweeps, and
portamento -- basic stuff you can do with the CV inputs on an SH-101,
Mini-Moog, or any other CV synth.  Extremely large CV ranges are
impractical, incompatible, and potentially dangerous.

> My time (and space) are super-limited right now, but this would be a
>good I/O project for an old computer.. See, we need a special controller
>for the Sawtooth VCO because it has a non-linear voltage response curve.
>We must build a 2-3 octave resistive step ladder and control each note
>with a gate (and step) of it's own.. This could be a simple key-contact
>device, or one under MIDI/computer control.

I'd like to get away from needing such a special controller for the above
reasons.  So, I have a couple of ideas.  I'm not an EE, so maybe they
aren't good ideas, and I welcome any criticism.

Let's look at a few tube oscillator designs.  First, we have a neon lamp
relaxation oscillator.  (Do they still make neon lamps??)  Friedrich
Trautwein used such a design in his Trautonium.  Tuning was controlled by a
resistive strip set up as a rheostat.  Closing the circuit on different
places along the strip produced different tones.  It could produce
approximations of a saw and pulse waveform which were mixed to create
various timbres.

Here is a sine neon tube oscillator:

100V ----/\/\/----+----@@@@@@--------+
         510K     |    L 100H        |
                  |                  |
                  +--( neon )---| |--+--/\/\/---| |---> ~100Hz
                               .005      5.1M   .05

The "problem" with most relaxation oscillators (as opposed to say a
transitron oscillator) is that they vary in frequency according to the
supply voltage.  Now, why couldn't a CV input be scaled and used to control
the supply voltage??  According to many tube fanatics it's OK to use solid
state as long as it's in the power supply :)

Most of the sweep generators in old tube oscilloscopes used thyratrons.  I
believe this is what is used in the Phattytron, but I haven't seen a
schematic.  There are also Hartley oscillators with many variations,
dynatrons, and transitrons.  They all use inductors which are hard to find,
bulky and expensive.  If they weren't, we'd all be building ourselves diy
Neve Eq's  :)  Wein-bridge, phase-shift, and thyratron oscillators are
built with only resistors and capacitors.  The problem remains that none of
these are voltage controllable (the resistors do not serve as voltage

The bugbear of solid-state analog VCO design is building a stable
exponential converter.  Can you build an exponential converter out of
tubes??  Although I haven't found anything, schematics for tube analog
computers (used in Soviet avionics, among other things) might be a source
of inspiration.  If you can, then perhaps a hybrid VCO could be built using
a tube exponential converter followed by a linear FET oscillator.

However, there are many perfectly good solid-state exponential converter
designs.  Perhaps one could be used to control a tube oscillator, or this
could be taken a step further and a solid-state voltage-to-frequency
converter could be used to drive a tube "waveshaper."

If a tube oscillator has a non-linear response to an incoming voltage,
perhaps a digital circuit could be used to translate standard control
voltages to values that would result in the correct pitch.

All things considered, I still have trouble launching lumber over the idea
of a tube VCO.  What I would want is a tube VCF!!  According to Metasonix
their VCF doesn't track the keyboard, and it has an extremely impractical
CV standard.  Is it possible to build a "triode ladder" filter???

Tube VCA's seem easy enough, but I think building a tube ADSR to control
that VCA would be next to impossible.  Building tube LFO's seems easier
than building tube VCO's because you can wire pots as rheostats, and they
don't have to be as stable.  Sample and hold is another matter entirely!
How about a tube noise source??  If you could build tube VCO's, VCA's,
VCF's, ADSR's, and LFO's, how would you mix modulation sources??  For
example, how would you send a mix of LFO and keyboard voltage to control
the VCF??  What's the best way to sum voltages using tubes??

So I'm thinking that many of the basic features found in subtractive
synthesizers (eg. ADSR, S&H, keyboard tracking, modulation mixers) would be
impossible, or impractical far beyond any desire for novelty, to build with
tubes.  I'd be very happy if someone came up with CV tube waveshaper, and a
resonant VCF that responded to 1V/Oct from 0-5V.


More information about the Synth-diy mailing list