exponential VCO problems

[Eric Barbour]

Paul Perry wrote:
> 
> At 11:21 AM 14/07/97 -0700, Eric Barbour wrote--
> >requires a special tube made by Raytheon in the 1950s, a
> >square-law tube called a QK329. Very rare today.)
>  
> .......and at no time did he succeed in making a useful analog synthesiser,
> if there is no high Q resonant filter.

You are welcome to your opinion. I have yet to see a really specific
definition of the term "music synthesizer", so such a cut-and-dried
statement is not really supported generally in the field.
Useful? Well, Greg Rule of KEYBOARD magazine liked the Phattytron
enough that he used it on his latest album. It was a crude prototype
monophonic instrument, prepatched, with 2 VCOs.

> And if he can make an exponential convertor with only one QK329 I would be
> amazed...a diode break point function generator would be more to the point
> if one is being thermionic.

I thought of that. Regrettably, vacuum diodes vary considerably in their
cutoff characteristics. They would have to be individually selected,
much like the thyratrons. The QK329 was an amazing device--typical
square-law error was 0.5% or less. Obviously you need more than one.
Boy are they scarce today!! I know serious tube collectors who have
never seen one, or even heard of it....

I have a circuit from a 1957 magazine to convert a voltage from linear
to exponential--it uses two 329s and four Philbrick tube op-amps.
10 tubes total.

> And if anyone wants to really build an analog synth  using valves, they
> could do worse than consult the works of Graninio Korn on analog computation
> ways and means. Of course it is possible to make a resonant filter from
> valves, playing around with voltage controlled transconductances......

It is more difficult than you think! The major problem is with
"motorboating"
oscillation, due to a low-frequency peak caused by the interstage
coupling
capacitors. The only way to do it RELIABLY is with a direct-coupled
circuit, probably
using tube op-amps. A basic state-variable filter would require a
minimum of
8 tubes, more if tubes are used as voltage-variable transconductances.
This
is workable, although you (again) have to select devices individually to
match each other or conform to a specific curve. (Same is true of
transistors.....
in some ways, it's amazing we can get any kind of analog circuits to
track each
other reasonably well.) My filter can be made to resonate, though it
won't track the
keyboard when oscillating. It uses only 4 tubes.....

So, would you care to share with us your tube synthesizer circuits?
Assuming you have actually tried to build any.......

-- 
Eric Barbour
Svetlana Electron Devices
Portola Valley CA USA