Please help me understand the problem with my theremin design

Hello,

I am a 4th year university student of EE, but I don't specialize in radio and analog stuff, so I only know some basics in this branch.

I started building a theremin a couple months ago with the help of a channel called "eleneasy". This is the guy's design of his oscillators: https://eleneasy.com/2019/08/20/theremin-v-1-pitch-reference-oscillator-analysis/

And here is his heterodyne mixer
https://eleneasy.com/2019/01/30/more-on-the-theremin-the-heterodyne-mixer/

I left the mixer as is and made a few changes to the oscillators, so I ended up with this:

The reference oscillator is the same as the variable one, but instead of the antenna, it has trimmer capacitors.
The transistors I am using are somewhat older Czechoslovak Tesla KF509. But I don't think there is any issue with them.
Building the circuit on a breadboard proved to be very inconsistent, so I made my own little PCBs. The theremin did produce quite good sound, but it had two problems:
1. the PCBs were hard to work with and didn't allow for further changes to the circuit
2. The pitch sensitivity was terrible. I think this is due to me increasing the capacity of C203 and decreasing L201.

This led me to start over on a protoboard, with some changes to the circuit. I tried to maximize the inductance and minimaze capacity, which should result in maximising the change of frequency to a change in capacity. These are the values I went with:

Both oscillators proved to oscillate well. The reference oscillator had a lower frequency than the variable oscillator. As expected, that resulted in the output audio frequency decreasing with proximity to the antenna. But it proved the mixer was working as well.
So I fine tuned the frequency, but to my horror, there was no audio output with the correct setting - reference frequency slightly higher than variable. I still could measure a sine wave on both of the oscillators. But mind I only have a soviet analog oscilloscope from the 80s.

I tried to increase R305 all the way to 2M2, but it had no positive effect.
So I went with less extreme values for the LC tank:

But still no effect. The mixer still doesn't provide any reasonable audio output. I can try to further increase the capacity, but with that I'm losing sensitivity.
Here is the reference oscillator, but as I said, it is mostly the same as the variable oscillator:

Does anybody have any idea what is wrong with the circuit?
Does one of the oscillators try to pull the other one onto its own frequency? If so, how can I stop that?
Or could there be a problem with using a protoboard for this kind of circuit?
I will be very grateful for any sort of answer. Thank you.

Brave soldier Shweik
are you ready for heroic deeds on the theremin‑building front?

“Does anybody have any idea what is wrong with the circuit?” – Shweik .

In the original circuit the coil L302 has an indecently small inductance. I assume it’s a linearizing coil, and in that case the resonant frequency of the antenna circuit, consisting of this inductance and the antenna capacitance (usually about 5…8 pF), should be equal to the resonant frequency of the transistor tank circuit (that’s the first approximation; in reality it needs some careful tweaking).

“Does one of the oscillators try to pull the other one onto its own frequency?”

Possibly, but that needs further confirmation.

“If so, how can I stop that?”

A common way is to add buffer stages (followers) to the oscillators. Less often, you can increase the division ratio at the summing point (i.e. either, as you suggest, by increasing the resistor values, or by lowering the input impedance of the mixer stage), but then you may need an additional gain stage.

“Or could there be a problem with using a protoboard for this kind of circuit?”

Most likely not, since you’re not in the microwave range. The increased parasitic capacitances of the wiring will have some influence, but it’s not fatal.

"2. The pitch sensitivity was terrible. I think this is due to me increasing the capacity of C203 and decreasing L201."  - Shweik

You're playing around with a double resonant design, and I'd probably recommend a single resonant design as a first project.  As ILYA points out, the interaction between the tank LC and the antenna C and series L needs to be carefully accounted for.  The tank has a huge capacitance compared to the hand capacitance delta (~1pF) which the series L magnifies.  The series L also gives you high antenna voltage, which is good.  Pro instruments are usually double resonant, but even under the best of circumstances they can be difficult to align.

Theremin design is where you find out that inductors of equal value aren't necessarily the same.  The antenna series inductor L302 in particular needs to have good Q at the operating frequency, and this means the use of RF chokes like in the Moog Etherwave, or large hand wound air core solenoids.  Modern component miniaturization is kinda the enemy here.  I'm kind of surprised the series coils the guy in the video picked worked at all, I haven't had any luck with single wound ferrite bobbin types, though I haven't experimented with them a ton either.  Coil Q is actually pretty easy to measure with a function generator and scope if you are interested.

ILYA designed some very handy software that helps you pick critical component values for various topologies.  If I were designing an analog Theremin that's where I'd start.

Welcome to the theremin community. 

Lets have look at your idea. The 4 mH is connected to a rod antenna. In series resonance with the antenna capacitance you need round the same frequency as the parallel 600 uH to 330 pF = 360 kHz.  Results in 50 pF antenna capacitance. That dependes on geometry: length an thickness. In practice the rod would be made for 35 - 40 pf and a little variable capacitor to ground has to be adjusted for the rest. Assumed, you have that proper rod, the next problem is the selfcapacitance of the 4 mH inductor coil. This dumpes the series resonance and can lead to no oscillation of the whole oscillator.  

I assume, the pitch oscillator works without antenna? Then prove at first the quality of the 4 mH coil. Remove the antenna and connect a variable capacitor (f.e. 10-100 pF) in series to the 4mH coil to ground. Control via oscilloscope the voltage over the capacitor (10:1 probe because of input capacitance). This should be in resonance 30 Vpp and higher. 

Hope that can help bit.