Awesome! I cheated and bought a Moog Theremin Plus, mainly for the CV outs to control the modular. I would love to build my own though. Did you design it yourself?
Yes. This documents the basic build:
http://seventransistorlabs.com/Theremin/I built a few of those, with minor changes between them. The most developed of which I keep on hand, and modify periodically.
Right now, it includes:
- Discrete transistors (approx. 50 total, mostly 2N3904 and 2N3906)
- Pitch, Volume, and Local Oscillators: Hartley, ~2MHz, Buffered, Adjustable (via varactors)
- Pitch and Volume Mixers: single balanced (single ended output coupling)
- Level Control Mixer: single balanced (balanced current mirror output coupling)
- Volume IF ~10kHz; slope detector
- Riser board: all important signals and control voltages are available on an expansion connector
- Audio output: max. 0.1W power amplifier, 1/4" phono jack output
- Power supply: 9V wall wart (primary source), onboard converter supplying regulated 6.3 and 120V DC (9 transistors, discrete except for the TL431, an acceptable sacrifice)
Currently, the riser board includes a timbre circuit (primarily a single common-emitter amplifier stage with variable input and bias levels to vary distortion and clipping), and a PLL which implements an auto-tune-and-hold feature, allowing instantaneous adjustment (no fiddling with trimmer capacitors or antenna position!).
Future plans: the switching converter was designed to provide the perfect voltages for vacuum tubes, with enough current capacity to power one or two 9-pin miniature types. These will be added to a new riser board, or a second level stacked on top, to provide "That Tube Sound". It should also be possible to add a PLL to the volume control section, for automatic tuning of that section as well.
(this was mostly copy-and-pasted from another document)
The PLL I don't think I've heard mention of before, which is odd because tuning these things is a bitch. Any slight change in the environment (imagine setting it up on a metal music stand or something) or temperature (I haven't matched tempcos yet, so it drifts quite a bit) makes for a lot of tedious adjustment.
The circuit, I've posted before:
The mixer output (as mentioned above) has a DC coupled resistive load, so when the beat is zero frequency, it's zero as in DC, so it's a fine phase detector. Comparing that with a fixed threshold (depends on mixer bias, but since that's set with resistor dividers, a fixed resistor divider should be fine here as well) gets a phase error signal, which goes right into the pitch oscillator's varactor. The only funny part about this, I think, is the amp itself serves as the sample-and-hold: the 2N3904/6s have low enough leakage (and the MOSFET follower, basically none), that the 0.01uF compensation capacitor stays quite nicely at whatever operating point it stabilized to. No need for suboptimal building blocks (namely, an op-amp, analog switch and JFET op-amp follower), which would consume loads more current (not that power is in short supply here) and suffer from capture transient problems (while disconnected, an always-on error amp will wander off into la-la land and saturate, guaranteeing a worst-case frequency spike every time the analog switch is closed -- it might not even re-lock phase anymore).
I made a point of not buying any production modules...
Built myself, mostly through hole, but a few SMD's here and there. Some boards are home etched, others were blank manufactured double sided PCBs. The panels in the top (5U) section are all home drilled and handwritten, in the lower (3U) section, they are factory printed. There's a few interesting parts inside, like vactrol opto's for nice bouncy CV control & tempco's to stabilise the oscillators. Not all discrete components like your Theremin though, there's quite a few CMOS chips & op-amps lurking inside...
An insane amount of work for a few years, in my opinion! I'm not very musically obsessed though, which would surely count for a lot.
Doing stuff like OTAs, S&H, VCOs, timers -- not terrible to do discrete, but for the most part, yeah, even I would be piling the chips on, no problem there.
Add in things like switched capacitor filters, sequencers, muxes and so forth, and you're counting a lot of CMOS alone, that would've been just a royal pain back in the day (early 60s or earlier, I guess?).
Tim