I have avoided using variable capacitors until this last theremin build as I always thought being mechanical they would also add to drift. Using the Var-5 variable capacitor 2 - 18pf I find it difficult to balance the oscillators for heterodyning as it sweeps by Zero-beat too fast. The advantage of using a variable cap is I can position the fixed ferrite coils perpendicular so they have no influence on one another. 

Amazon has a simple set of variable capacitors and a blue 5 pf variable could parallel the main Var-5 for fine tuning.  You could substitute the Var-5 on the Mouser list for the Amazon green 30 pf variable cap. Always make sure the rotor connection the screw driver touches is soldered to the ground side.

* In theremin design there are always trade-offs.*

I used the MPSA42 Transistors to test if the 300 volt junctions might have less PN junction influence. This still needs more study but stayed with them as they created a wave shape that mixed for a slightly better sound. I did try 1000 volt transistors but they would not oscillate with my favored oscillator circuit design.

* A test that needs to be done is find which is more stable? Vary the emitter resistance or use a low value variable panel mount capacitor and leave the PN junction voltage fixed.

The EWS drift resistance is no better than my approach.

click on picture to enlarge

The long wires on the red capacitor I use to connect 9 volts. Before power is attached you should measure 
about 1.2k ohms across this green 9 volt terminal. Regulated power supply best, a 9 volt battery will work.

- Finish stuffing the Oscillator Section as seen above -

The variable capacitor Var-5 is 2-18 pf or similar. The rotor part of any variable capacitor should be on the ground side so touching it with a screwdriver does not affect it. This can sweep past Zero Beat rather quickly. Not necessary but you could place a 5 pf variable cap (Amazon) in parallel at terminal pads C & D to the left of it. This would be Var-5b. Move the yellow wire in D to the pad to the right of it. Have the rotor side of the variable caps in the ground side pad. Use an ohm meter and touch the rotor or screwdriver section and then the pins. The pin of zero resistance must be the grounded side.

On Var-5 Datasheet     C17- 10pf is to complete the L1 LC circuit.

At the bottom of every webpage is a transparency of the parts against the copper traces so you can see the connections. The view is from the top of the board.

6.09.20 This is the first Phoenix test, the oscillator section. Pot-3 1k is set half way (blue on this board). No Pot-4- 1k needed in this test as R25- 220 ohms completes the circuit. R25 in parallel with Pot-4 allows a 1k or 5k as on the EWS panel to be used with a range of about 0 - 200 ohms for emitter PN junction tuning. This is in series with R8 & R20 the transistor emitter resistors.

Pot-3 balances the thermal drift difference between the two oscillators by shifting the voltage across the transistor PN junctions..

If the C4 10uf capacitor by the 555 is backward it will behave like a delayed short circuit.

Notice the L1 & L2 choke/coils are perpendicular to one another. This reduces inductive field cross talk.

L1 emitter signal is about 4 volts p-p. 
L4 is attached to ground

L1 emitter signal is about 5 volts p-p. 
L4 is not attached

L2 emitter signal is about 7 volts p-p. .

emitter dc voltage is 3.72v

L5- 330uh attached to 8" of wire

emitter dc voltage is 3.69v

Starting frequency of L1 & L2 is about 980 kHz which is ideal, a capacitor is then needed in the green terminal next to it to drop down to 900 kHz.

L1 sine amplitude is lower than L2 because R9- 470 ohm needs a .1uf bypass capacitor to match the L2 amplitude. I have been aware of this but never tested it.

For the Pitch Board less amplitude may be better for no low end early coupling and more amplitude for Volume Control to get the PWM wave shape pulling.

L4 & L5 do not drag the RF wave shape lower in amplitude on the Pitch Board of 900 kHz, The sine wave shapes look very nice.  On the Volume Board at 750 kHz   L4 should not be connected.

The goal is the ideal theremin sound wave shape that has eluded most for over 100 years. It is not having the ideal sine wave or low end coupling. The beautiful upper end of audio is a reflection of the lower end wave shape..

I forgot to mention that a 22pf variable capacitor in series with .1uf is still 22pf, the .1uf is transparent in the circuit.

Perfect
(more droop the better)

Excellent
(Vocal)

Poor 
(antenna feedback)

Over Coupling for PWM
(Bottom Signal Out 555)

In theremin design locking the oscillators at the Null Point is a flaw more than a benefit.

- To avoid audio wave shape distortion antenna orientation and shielding are important -

Use Ctrl-F5 to update this Image

Having C8- .1uf increases sine wave amplitude 33%, this is not always desired.

- Right click mouse on any picture, open in a new tab for larger view -

Notice in the picture to the left that I do not allow the board to sit against the wood which has natural moisture in it.

Seen in the picture above on the upper left corner that the L3 3300uh choke/coil is leaning over toward L1, the distance determines the RF signal mix strength. To close could pickup static noise in the sound, seen on the scope and also the same static heard over the AM Radio. L3 picks up the L1 & L2 magnetic fields which are mixed in the D1 diode for the heterodyned audio output. Your final frequency may need adjustment if a strong AM Radio Station is in the area.

The two green terminals off the right side of the oscillator sections allow adding extra capacitance when you want the Pitch board frequency to be lower for a Volume Board.

steel mesh for shielding

The Radio Frequency Sniffer Coil

In the first test we are only powering the two RF oscillators. You must have the L4 & L5 antenna coils setup and have a direct connection to Earth ground, this is a must.

For your first antenna use an alligator clip lead dangling below the Phoenix board off the table. This helps avoid signal distortion from unshielded antenna RF/capacitive feedback.

In the picture to the left is how you measure the frequency of L1 & L2 to avoid overloading the circuit. Clip the extra 330uh coil/choke on the Mouser List to the end of your frequency counter probes. 

This is used as a RF sniffer when placed next to the L1 & L2 to get an accurate measurement of the frequency of each oscillator. Ideally you want L1 to be at 900khz and by adding capacitance to the green terminal next to L2 you lower its frequency to match the frequency of L1.

The Q1/L1 side frequency should be from 850 kHz to 920 kHz. (900 kHz is ideal) Measuring the frequency any other way will load down the circuit and stop oscillation. 

If you do not have a frequency counter it is possible to use an "analog" AM radio tuned to around 900 kHz and listen to the blank spots in the background static while finger touching the L1 & L2 coils. If the Q2/L2 oscillator is out of range of matching the frequency of the Q1/L1 oscillator you want to add or remove capacitance from the Q2/L2 side so L1 remains at 900 kHz. One thing that can cause this mismatch is the circuit are variables like the 10% tolerance of the coils, etc.

Good Earth Ground

.5 v/div     1 ms/div

well grounded is the ideal clean audio wave shape

Poor Earth Ground

.5 v/div     1 ms/div

With no earth ground RF pollutes the audio signal, this also indicates a poor Pitch Field response.

My two RF oscillators do not need buffering as they have no component connection between them, just earth ground. Between the circuits L1 & L2 magnetic fields will interact with one another. This can be controlled by the amount of current driving the coil and the L1 & L2 coil orientation to one another.

This L3 inductive pickup approach is unique and original. It allows me to control the amplitude and mix of the magnetic fields to get the best heterodyned audio wave shape out of the D1 diode.

The transistor emitter current determines the strength of the oscillator RF inductive fields. Too much you could get over coupling between the oscillators which causes unwanted distortion or too strong of a heterodyned mix.

The D2 diode is a future experiment. This allows for a second channel using the audio section of another board and mixing it back into the final audio for possibly a much fuller or interesting sound.

Nice LC Frequency Calculator

I use the MPSA42-AP 300 volt transistors for a slightly better sound and thermal drift control. I changed the R8 & R20 4.7k resistor to 3.3k. If you must use a 2N3909 NPN you might want to keep the 4.7k for less range with Pot 4-1k. If you need even less swing when Nulling with Pot-4 1k the range can be reduced by adding a paralleled resistor, R25 with a 470 ohm. The fact that the MPSA42-AP reduces the Pot-4 range tells me the transistor will be less sensitive to thermal drift.

Thermal drift is reduced by the choice of transistor and matching the hFE values. Then balance the current through the two transistors using the Pot-3- 1k at the bottom of the board in the oscillator section. At first set it half way so it matches the R9- 470 ohm resistor value. With this method you can actually reverse the direction of drift were the Pitch can drift in the other direction.

- To avoid audio wave shape distortion antenna orientation and shielding are very important -

Seen below the board mounted vertical with a vertical pitch antenna, this is ideal

 Original build of Leon Theremin

Valery S in St Petersburg Russia, the original home of the Theremin family

I said 20 years ago that Clara's Voice was a gift to me, that is what kept me going. Jokingly I would say, ask Dorit Chrysler if she still has it. I am having fun, if ever you saw that movie "The Book of Eli", that is how I feel, I must finish the journey in the West and write down what twenty years revealed to me. Did something metaphysical happen along the way, I will leave that up to you to decide.