About Paia's Theremax...

[Tom May]

"Matti E. Leino" <mle at nic.fi> writes:

> Hi !
> 
> I have followed this now for few months, and seems like there's
> quite smart people there. Now, my question is, has anyone of you
> built this Paia's Theremax theremin...?

I did and it sucks.  I never got mine to work very well, although I
tweaked every subcircuit except the oscillators because pretty much
everything needed fixing.  (And yes, I had PAIA's mod for better
volume control or whatever it was.)  In many cases the decription of
how the circuit works does not relate to reality -- it would be better
if the circuit worked as described, not as it was built.  I gave up
working on mine over a year ago, but off the top of my head the
following things were wrong or strange with it:

1. The input and/or output from the ring modulator were attached to
   the wrong points of the diode ring.  There was huge feedthrough of
   the oscillator frequencies, and substantial filtering was thrown at
   the result in an attempt to leave just the difference frequency.  I
   rearranged mine to get much better oscillator rejection and backed
   off on the filtering.  I was careful to keep minimal interaction
   between the oscillators when I changed the ring modulator
   attachment points by using strategic isolation resistors.

2. The design analysis (http://www.paia.com/theresch.htm) says "The
   output of the ring modulators is boosted in level by discrete
   transistor amps which also provide a second pole of low pass
   filtering because of the the feedback capacitors from collector to
   base."  In reality, these filters don't create a second pole, they
   just shift the first pole (due to C22) to a lower frequency: C27
   causes a Miller effect capacitance from the base of Q8 to ground
   which is in parallel with C22.  One pole, not two.  In my redesign,
   only one pole of filtering is necessary, but the original design
   sure could use two poles.

3. The CV outputs (and this includes the CV that is controlling the
   built-in VCA that determines the volume) use a bizarre frequency to
   voltage / filtering / buffering circuit that sort of works, but not
   at all like the circuit description.  There is still a lot of audio
   frequency feed-through in the CV outputs.  The design analysis
   says: "The fully filtered signal is converted to a square wave by a
   Schmitt trigger, then differentiated to produce narrow pulses which
   are integrated to produce control voltages."  The circuit actually
   transfers charge with each pulse into C24 through C28 in a totally
   non-linear way.

4. I also did some tweaking on the Schmitt triggers . . . for some
   mysterious reason, the pitch and volume Schmitt triggers use
   differing amounts of feedback.  Maybe to cause enough hysteresis to
   ignore the oscillator frequencies leaking through . . .

5. It was difficult to get much range from the pitch antenna, and/or
   the pitch vs. distance curve didn't seem very usable.  But your
   mileage may vary.

> Because, there's four
> oscillator coils, 796 kHz, and it seems very difficult to find them
> here in Finland,

Just get the kit from PAIA if you really want to build one.

fTom.