Greetings denizens of Theremin World!
Last year I attempted to build my own version of the PAiA Theremax 9505 by swapping out the oscillators with an oscillator design of my preference. After learning a lot about how not to circuit card a Theremin, I discovered Theremin World. Reading through many interesting discussions about theory and construction on this site encouraged me to attempt another Theremin build. I chose to build a version of the EM Theremin by Robert Moog. I incorporated recommendations provided my members on this forum along with some modifications of my own. Below, I would like to submit for your consideration a few brief summaries of my experiences in tweaking the design.
Disclaimer: My work-in-progress prototype build is assembled on a breadboard. I know this brings its own complications of added capacitance and noise. Once I have finished tinkering, I plan to move to a circuit card design.
Power:
First, I want to power the Theremin using a standard 12Vdc wall wart as opposed to the AC transformer. To accomplish this, I replaced the MC78 / MC79 supplies with a Murata Power Solutions NMH1212SC power module. The module does include a switching supply for the -12V, but the switching frequency is around 90kHz, so there is no noticeable interference with the oscillators. I initially tested the Theremin build with power supplied by an Agilent linear benchtop supply and then swapped to the power module. Sound quality and electronic noise levels are comparable for the two supplies.
RF Chokes:
Like many before me, I discovered that the RF Chokes (J.W. Miller #6306 / #6304 / #6302) are generally no longer available. Since I wasn't particularly interested in winding my own inductors, I went searching for possible replacements. Although I was not able to find any chokes with the cross-winding patterns exhibited in the Miller chokes, I did find a few models that I thought were worth a shot based on similar electrical specifications. I settled on the Fastron inductor series (such as 09P-272J-50). I seem to be having reasonable success with these on the pitch side, judging from the sound produced by my build.
I'm still tinkering with the volume side. Because the chokes I've selected do not have values that exactly match the original design, I'm having to tweak the component values in the volume oscillator. Through LTSpice and trial & error, I've almost selected final modifications for this circuit, but my volume antenna seems to be throwing things out of whack. I'll need to retrace my steps here and quantify the effects each modification has had.
Pitch Oscillators:
I replaced the resistor long tail on each differential transistor pair with a transistor constant current source. It's probably an unnecessary frill, but I thought it might help with noise immunity through the negative supply. What I discovered is that once I added the current source, oscillator drift diminished. With the resistor tail, I found that the oscillators would drift out of tune within a few hours. After adding the constant current sources, I found the oscillators remained stable for longer periods of time. In a couple of instances, I was able to leave the instrument powered overnight and return in the morning to an instrument with the same tuning.
The current sources consist of another 2N3904 connected to the emitters of the differential pair. A ~1240 Ohm resistor runs from the transistor's emitter to the -12V supply. Two resistors run from the base, with the first (~2.9k) running to ground and the second (~15k) running to the -12V supply.
Mixer Circuit - work in progress:
I'm attempting to substitute the diode mixer with a dual MOSFET mixer. I'm trying a MOSFET mixer for two reasons. 1 - The diode mixer produces some asymmetry in the negative sweeps of the sine wave, and I'm curious if I can get a more symmetric waveform with a different mixer. 2 - I've seen analog engineers with far more experience than me add or remove distortion from a waveform using a similar topology. I'm hoping at a future point to use the MOSFET mixer to distort the wave and add sound color. So far, I had what I thought was a solid preliminary design laid out in LTSpice, but it doesn't seem to be panning out in implementation. More to come on this hopefully...
Lingering Problem - the VCA Processor:
My only real hang-up appears to be in the VCA Processor side of the LM1370. No matter what I try, the pin13 input wants to sit at +560mV, regardless of whether the volume oscillator is connected through R14 or not. I attempted to drive a voltage at the pin using a function generator. I found that at 0V at pin13, pin12 goes to -11.3v. With pin13 at -0.1V, pin12 shoots to +10V, with pin13 at -0.1V, pin12 goes to +11.3V. I haven't been able to figure this one out yet.
Audio Amplifier:
Based on earlier suggestions in this forum, I added a LM386 audio amplifier to the audio output of the LM1370. The amp is connected through a 10k pot, and I moved C26 and R34 after the pot to help with noise. Actually, I have found that the potentiometers I have used in my breadboard design are one of the largest contributors to acoustic noise.