Vacuum Tube Theremin

Latest schematic update revision 2.8

Text updated 2020-08-20

Update: I was having issues getting the Volume Controlled Amplifier to drive to cutoff.  I made several incremental changes to the wiring in order to  improve the cutoff voltage.  None except the last change (driving both #1 & #3 grids with variable bias) worked.  However, having made a number of changes, I decided not to back up through the changes.  As a result the schematic is more complicated than necessary.  The astute designer may wish to simplify it.

1  Mark’s Fourth Theremin of Five – First Vacuum Tube Version

1.1    Intro

As a classical musician (pianist/organist), I was taken by Canadian author Sean Michaels’ wonderful novel “Us Conductors” about Lev Termen, published in 2014.  By ten pages into the book I knew I had to build Lev Termen’s theremins.  I decided to embark on a journey to build five theremins, starting with solid state analogue, a format with which I was comfortable, move to a digital experimental form, then build two vacuum tube theremins, the first using ‘modern’ tubes, and the last using the same tubes, if possible, found in Lev’s creations.

This is my first theremin designed from first principles.  My first two are solid state analogue devices, Hartley and Colpitts based, using well known designs readily available on the ‘net, work well, and the third was/is a digital design which is sitting on the shelf at the moment.  I want to see how far I can move the signal along the system before converting to analogue for the output.

My research into ‘modern’ vacuum tube theremins led me to John D. Polstra’s wonderful site  Having not built using vacuum tube technology for some 50 years, I lacked the tools needed to breadboard my designs – a variable HV supply, heater sources, a coil winding device, and an actual breadboard.  I unashamedly copied John’s tools.  To him I say “Thank you.”  I also used his basic design concept of 12AU7s for the oscillators and 6L7s for the Mixer and Voltage Controlled Amplifier.  Great tubes.  “Thanks again.”  That is where the similarity ends.

1.2    Schematic, breadboard, and chassis

I have a spatial relations disability so I have written the schematic with the modules in the order that they occur from left to right on the actual breadboard and chassis.  The volume antenna circuitry is on the left and the pitch circuitry is on the right.  If you are left handed, the typical theremin schematic may be more to your liking.  

Update:   I discovered something.  Maybe all these schematics with the pitch antenna and circuitry on the left actually make sense – at least for building with valves.  When I built up my breadboard, all the components were in the correct location and order with respect to the schematic. However, when I built up the chassis, everything was upside down and backward.  My spatial relations problem caused me to mis-wire several times.  It’s much better now.  

The final schematic shows minor adjustments due to parasitic capacitive and inductive effects in situ.  The Pitch Antenna inductor was changed from 20mH to 17.5mH, Then I tried bending the antenna output wire against the chassis.  I had to change the loading coils back to 20mH total.  the Voltage Control Oscillator tank circuit is now back to my original size, using two 4700pF caps, and the Volume Antenna inductor is now 12.5mH, up from 10mH.

The original filament supply was floating at 6.3VAC.  This seems to cause 120Hz hum that varies with the output volume control.  Here is a picture of the hum taken at the audio output jack.   It shows spikes that I assume are diode charging pulses to the reservoir capacitor in the HV DC supply and not the AC filament supply. The harmonic distortion of the waveform is over 77%, covering the whole audio spectrum through 24kHz!  

120Hz spikes-2

However, here is the change in the hum type noise by adding a 100Ω resistor in each leg of the filament lines to ground as shown on the schematic.  The hum level has not reduced significantly (ignore the difference in measured amplitude as I used a different gain setting), but the spikes and resulting distortion noise are gone and no longer heard.  Hum level is 60dB down from the max signal. 



Please note that a 5814A vacuum tube is, in fact, a 12AU7. I would not recommend it for high quality audio work due to not undergoing testing for audio limitations, such as microphonics, but for use as an oscillator, it is excellent, sturdy, readily available, and cheap.

Here is the final breadboarded design. I have lifted the breadboard from the wood surface at the suggestion of ‘dewster’ from Theremin World in order to limit the capacitive effects of the wood. It made a big difference. The wooden case in the background is a variac-based, unregulated, variable high voltage supply I built. It provides up to 450 volts DC at about 60mA, 6.3VAC or 12.6VAC for filaments, and a -0.5vdc to – 25vdc bias supply.  Cheap LCD voltmeters running on independent 9 volt batteries provide actual supply load voltage and current data. 


Here is the chassis build in test mode.

Chassis with antennae

The 10 Watt and 5 Watt resistors showing below will be relocated to the top of the chassis for heat dissipation efficiency.  I located them temporarily inside for ease of adjustment to the correct values.

Chassis Inside

Chassis Template

I laid out my design on a centimeter square grid paper and, using sealing tape, overlaid it on a Hammond 12″x8″ aluminum chassis. I used to own, when I was 12, a complete set of chassis punch tools for all the common vacuum tube socket sizes. Sadly it disappeared, as did most of my vacuum tube electronic gear, much of which I could have used for this project. To drill the holes I am using a unibit step-drill.  I have never used one before, so I will learn on a spare piece of aluminum.

Update:  The unibit drill worked great. Much, much cheaper than punches.  I highly recommend it for light chassis work.

Chassis template

Here is the Vacuum Tube Theremin in it’s final form.

I decided not to build a case for this theremin as I plan to build a close copy of the early Termin theremins, including the case, for my fifth and final theremin design. The bottom plate of the chassis is attached to the wooden structure. The chassis is attached to the bottom plate with four quick release hand-tightened bolts, allowing easy removal for experimentation.



With the large resistors on the top of the chassis, the total heat dissipated inside the chassis is under one watt distributed under 96 square inches of aluminum.   As a result, the stability of the shunt-fed oscillators is amazing.  As there is no DC current running through any of the coils to heat them up, the system is stable within 11 seconds (tube filament warm-up design parameter) with no noticeable drift over a test period of two hours.  That is to say, no adjustment of the pitch null or hand/antenna linearity was required during that time.  Total drift was on the order of a few 10s of Hertz, but all in the same direction, so the delta between the oscillators was basically constant. 

1.3    Design Considerations

ANTENNAE – The antennae, of course, are not antennae.  They are one plate of a capacitor, the player’s body and hands forming the other plate of an air-spaced variable capacitor.  They are based on Termen’s specifications, except for material and diameter.  I used ¼ inch solid brass, because, with annealing, it is easier to bend than the hollow stuff. Each antenna is mated to a set of linearizing coils.  Linearizing is simply allowing the player to adjust how much distance the hand must travel toward an ‘antenna’ to achieve the amount of desired change in frequency or dynamic.  Linearizing coils are interesting.  It is assumed by many designers that smaller coils (i.e., lower inductance) have lower parasitic capacitance.  This is not the case. The Hammond series Radio Frequency Chokes (RFCs) ranging from 2.5mH to 10mH all have parasitic capacitances in the 6pF to 5pF range.  It is far better to use four 2.5mH coils at 6pF each to make up 10mH than to use one 10mH coil at 5pF.  Why, you say? Because capacitors in series produce a capacitance of 1/(1/C1+1/C2+1/C3+1/C4), or, this case, 1.5pF. 

SHUNT-FED COLPITTS OSCILLATORS are used throughout the design. Shunt feeding isolates the tank coil from DC current and the resulting heating effects on frequency stability.  I originally designed the oscillators for operation at 270kHz using equal size capacitors which are quite large (10nF) when the coil is only 72uH or so.  I could not get the voltage swing I desired for distortion control so, in this iteration, I have chosen to increase the frequency of operation to 370kHz, which allows me to use slightly smaller, asymmetrical capacitors to control the oscillation feedback voltage and the output voltage swing, while still maintaining a small coil, which is critical for the ratio of the linearization inductance to oscillator inductance, which, according to Thierry at Theremin World, should exceed 150.   I chose 5814A tubes for the oscillators.  They are 12AU7s, but were really designed for digital computer use, that is, operating saturated or in cutoff mode.  They are plentiful and cheap and several datasheets for the 5814A and 12AU7  are found on the ‘Resources’ page on this site.

I originally wanted to use distortion in the oscillators to create the tone, but now I find the intermodulation products to be undesirable, so I have kept the changes to the oscillators’ drive parameters (grid bias and coupling), but am taking the oscillator output from the grid, not the plate. The grid has a nearly perfect sine wave output and a lower amplitude ( ≈30%) than the plate.  The 6L7 mixer (OK, I’m an ‘old guy’ – I should say Product Detector.) turns out to be the best place to modify the tone using varying amounts of coupling of the two oscillators.  This mimics Termin’s own investigations.

BUFFER/AMPLIFIER – The second section of each 5814A is used to amplify and buffer the associated oscillator output.  The amplitude of the second section output is determined by the oscillator plate coupling capacitor, the cathode resistor, the plate load resistor, the output coupling capacitor, and the next stage grid leak resistor.  The amplitude of each stage was chosen for its critical effect on the harmonic distortion of the waveform. The output now has a second harmonic about 20dB down, which is low, but plenty to work with when varying the coupling of the two pitch oscillators.  The third harmonic and larger are all greater than 50dB down.

Most of the bypass and coupling capacitors, and the grid leak resistors (with one exception – the 6L7 mixer grid 3 resistor which should not exceed 50K in mixer mode) are arbitrary values and can usually be varied by 200% or more and not adversely affect the outcome.  I used what I had lying around.  As I closed in on a final design, the value of the components became more critical to the desired outcome.

An aside regarding manufacturer datasheets and max values for grid resistors. If the recommended value is exceeded, under certain conditions, the tube can go into thermal runaway. I know this to be the case, as I have experienced the phenomenon by watching the plate current unexpectedly and slowly start rising exponentially.  I was able to reproduce this phenomenon at will, and lowering the value of the grid leak eliminated the problem.  Adding a cathode bias resistor helps, allowing for a higher value of grid leak, but does not eliminate the problem.

Update:  I have changed the grid bias in both the oscillator and buffer sections of the Variable Pitch Oscillator (VPO), Fixed Pitch Oscillator (FPO) and Voltage Control Oscillator (VCO) to drive the tubes harder. The total oscillator/buffer current load for all three is about 9.15mA.

Update:  I have moved the Anode supply of the VCO/AMP/BUF from regulated 107v to ~250 unregulated.  This increased the anodes’ current load from 1.72mA to 2.95mA.  The purpose is to increase the voltage swing at the buffer output from about 19v peak to 27.6v peak.  I will speak about this further on.

The 6L7 is a real find. It is a pentagrid Mixer tube, but is also a potent Automatic Gain Controlled device.  They are plentiful and cheap as well. Mine are from the Lend-Lease Act with hand painted serial numbers and dates (just a few months before Dec. 7th, 1941).  A 17 page document from KEN-RAD on the 6L7, published in 1936, is available on the ‘Resources’ page and gives exhaustive detail on using the tube.

The 6L7 is happiest with about -3 volts on the first grid and about -10 volts on the third grid, 100 volts on the screen grids, and anything from 150 to 500 volts on the plate.  With the voltages as stated above, the plate current varies less than 0.5mA over the entire range of plate voltage from 150 to 500 volts.  The ‘dynamic’ plate resistance is one meg ohm in mixer mode and 800,000 ohms in amplifier mode.

With equal bias on grids 1 and 3, the 6L7’s transconductance goes to 5 µmhos with only -15 volts bias.  With -2.6 volts on Grid 1, the Voltage Controlled Amplifier (VCA) drives Grid 3 from about -6 to -21 volts, causing an plate current change from 3.3mA to less than 0.01mA or about a 50dB volume range.  This allows enough ‘bleed’ to faintly hear a pitch before starting a piece.

Update: I am now driving both grids (grid #1 through a 1MΩ resistor) from the VCO and the max DC bias is now -27.6v.

An interesting thing about dynamic range.  I read about theremin designers trying to get 70dB or greater dynamic range – to what point though?  The max practicable dynamic range of a violin is 35dB, a tuba is 30dB, and a clarinet or horn is 40dB. There is no need for a greater range than this, and the sensitivity is a more important issue. 

Update:  That being said, you still need to be able to silence the Theremin when desired.

 6L7 Mixer plate current is 4.6mA.

6L7 screen currents vary a bit, but run about 5.5mA in the Mixer, and is variable from 6.7mA to 8.6mA, depending on the grid #1 & #3 bias, in the VCA.

All positive voltages, except the 6L7 plates and VCO/BUF are regulated to ≈107 volts using a 0C3 gas discharge tube.  The 6L7 plates are at about 197 volts unregulated for the Mixer and 241.3 volts for the VCA.  The total current variation of Vbb from full volume to null volume and max pitch to null pitch is about ±1.5mA, so really, the regulation is about evening out the mains supply variations.

When measuring voltages, keep in mind that Vbb and Vb are not the same thing.  Vbb is the supply voltage to the circuit.  Vb is the voltage measured between the plate and the cathode directly – no resistors in the way.  Tube datasheets are not consistent with regard to terminology and symbols.  When looking at a chart (not a table), the Anode Voltage is the voltage measured from Cathode to Anode.  When looking at a table, you may see the words ‘Anode Supply Voltage’, meaning the voltage measured from system ground to the power supply side of the Plate Load resistor.

1.4    Module descriptions

VOLUME OSCILLATOR – The Volume Oscillator is fixed at about 425 kHz and does not vary.  This number is arbitrary, though keeping it in the range of 150-700 kHz is useful for parts values and desired hand/antenna distance response.  The rf is extremely well shielded, the ‘antennae’ produce no measurable output, and, try as I might, using a long wire antenna connected to my FFT display with the Theremin located 1 metre away, could not detect any fundamental or harmonics above the noise level.

The buffer/amplifier section provides about  28vpk-pk (peak to peak) output.  The Volume Antenna is a series resonant circuit at about 11.3pF in situ and with about 12.5 mH total linearizing coils resonating at or slightly below the Volume Oscillator frequency, within a few hundred Hertz, when the hand is furthest away from the antenna (far field), and is coupled via a human protecting capacitor to the buffer’s plate. Moving the hand close to the antenna lowers the resonant frequency to about 380kHz.  The Volume Tuning Capacitor trimmer changes adjusts the resonant frequency of the Volume Control Oscillator to bring the Antenna’s resonant frequency closer to or further away from the VCO.

In the far field, with the Volume Antenna circuit at resonance, the high Q of the series resonant circuit causes it to present a low impedance to the plate output, resulting in a low output voltage and minimal bias to the next stage Volume Control Amplifier which results in high volume output.  Moving the hand closer to the Antenna increases the capacitance of the circuit, moving the series resonant frequency further away (lower) from the frequency of the oscillator.  This raises the plate output impedance relative to the oscillator frequency, raising the output voltage from about 3v pk-pk to about 28 pk-pk.

Update:  On the breadboard, I found -21v was enough to bias the 6L7 to cutoff.  In situ, due to a loss of Q because of coupling effects of the VCO coil to the chassis, the VCO maximum output is about 19v which is not quite enough to bias the 6L7 VCA to cutoff. I need more than 21v.  I will change the output of the Buffer/Amp to compensate.

Update:  The bias is now -27.6v and drives both grids. The dynamic range is now 54.6dB.

The plate output is coupled via capacitor to a simple rectifier/filter unit to yield a variable negative DC control voltage.  The 0.01µF filter cap from the diode output can be changed in value to provide faster or slower response to hand position changes.  This voltage is fed to the #1 and #3 grids of the 6L7 VCA and provides a variable DC voltage from about -3v to -28v or so.

VOLUME CONTROLLED AMPLIFIER – The 6L7 is configured with about +2.6 volts on the cathode using a cathode resistor.  The Plate resistor is chosen to provide sufficient peak AC output to drive the next stage.   The basic Grid #1 & Grid #3 bias is obviously -2.6v relative to the cathode, but the VCO/Buffer provides an additional -3 to -28 volts.   The screens run on +107v regulated passed through a 510 ohm resistor to drop the voltage to about 100v and the current varies from 0.01mA at nil volume to about 3.3mA at full volume. The plate current varies roughly inversely to the screen current.

So here is how it all works.  The Audio Frequency Output from the MIXER is applied to Grid #1.  The Volume Oscillator’s DC output is fed to both Grid #1 & Grid #3.  With the hand furthest from the Volume Antenna, the Grids’ voltage is -6v or so.  Basically LOUD. As your hand gets closer to the antenna, the Grids’ voltage gets increasingly more negative, driving the transconductance down.  Very close to the Volume Antenna, the voltage reaches <-28v, and the transconductance is below 5 µmhos, resulting in nil or barely perceptible volume out.  The AUDIO OUTPUT of the theremin is taken through a 0.1µF cap to a 250k Audio Taper pot to an external amplifier.  Again, the output cap and pot values depend on your next audio stage input.  In my case the output at the capacitor (not the potentiometer) is about 15v pk-pk max – easily controlled with a pot.

FIXED PITCH OSCILLATOR/BUFFER/AMP – Frequency is set to about 370 kHz {1×10^6/((√LμHxCpF)x2π)} using a parallel tank circuit of 72uH and 10000pF and 3300pF caps in series plus trimmers. The series caps total 2480pF (C1xC2)/(C1+C2) and the oscillator is ultimately tuned, using a trimmer capacitor, to the VPO/Antenna/performer far field frequency.  The oscillator grid is coupled through an 82pF capacitor to the buffer amp grid.  A variable air capacitor with the frame grounded (hand proximity does not affect the frequency) is used to fine tune the frequency to cause mode locking at a desired low frequency.  I have found great stability and excellent mode locking.  I can easily play (not hear obviously) notes at one Hertz, the sound being made up of harmonic distortion.

MIXER – The 6L7 MIXER is configured almost identically to the VCA 6L7. The FPO is fed to Grid #3, which is biased the same as Grid #1  from the cathode resistor. The screens are held close to 100 volts. The 100 ohm screen resistor is there to facilitate current measurement. The Variable Pitch Oscillator is fed to Grid 1.  The plate output radio frequency energy is filtered out by a 1nF capacitor/10k plate resistor combo, and the resulting audio frequency energy is fed to the VCA.

VARIABLE PITCH OSCILLATOR – The VPO is nearly identical to the FPO.  It’s base frequency, without the antenna coupled to it, is about 366kHz.  The antenna, linearizing coils (≅9pF and ≅20mH), and inductive self capacitance (≅1.5pF) have a far field resonance at about 350kHz. When coupled to the VPO, the resultant frequency rises to about 370kHz, and the hand motion gives a range of 0Hz to about 2.5kHz or greater.  The VPO output amplitude varies from about 10v to 8.5v pk-pk over its active range, a standard feature of most oscillators, causing a 30-26v output range from the Buffer.  This is actually convenient as it inversely mimics our natural hearing response. We are less sensitive to low frequencies and increasingly sensitive to high frequencies, peaking at about 4kHz.  A trimmer capacitor is used to match the frequency of the FPO, but, more importantly, to find the optimum desired linearity. 

The Pitch Antenna is coupled directly to the plate side of the tank circuit.  The circuit is already isolated from High Voltage DC by capacitors.   As the hand moves closer to the Pitch Antenna, the capacitance increases and the resonant frequency of the entire tank circuit lowers as well, increasing the frequency difference between the FPO and the VPO.

HAND WOUND COILS – These are direct copies from John D. Polstra’s instructions.  I used a spreadsheet calculator (see the ‘Resources’ page) to confirm the data.  The height to width ratio is important to achieve the lowest distributed capacitance.  I used ABS since it is commonly available.  A single turn of 26AWG copper on a 1.5” nominal form is over 2µH, or about 5 kHz at 360 kHz, so accurate counting of turns is necessary.

POWER SUPPLY – The power supply is very straight forward.  A capacitive input filter is cheap, but puts high peak load current pulses on the rectifiers (up to 10 times the load current, depending on the capacitor size and load), which must therefore be appropriately rated. Multiple sections and types of filters can also be cascaded. That being said, if you decide to use a choke in the filter, there is a minimum critical inductance that ensures the choke will act like a choke (which exhibits good inherent voltage output regulation at about 0.9 input voltage) and not like a capacitor. The formula is L(crit.)=E(output load voltage)/I(load current in mA).  In my original choke input filter design (150volts out at 35mA), I only needed 4.3H.  I found a bunch of no-name 4.5H chokes (146 ohms, so probably good for 100mA), so I designed a choke input supply using stuff I had in the box, except for the transformer.  That is the one that John Polstra used, from Hammond. (Funny, I grew up in New Hampshire buying and using Hammond transformers and chassis’, and now I live in Guelph, Ontario where Hammond was founded and this particular transformer was made, along with the chassis I am using.  The world is a small place.)

A 5 Watt bleeder resistor across the final filter capacitor provides additional load to the choke and is a safety feature as it bleeds the lethal charge from the capacitors after shutdown.

Update:  In the chassis build, it appeared that the choke input filter did not provide adequate filtering.  I reluctantly added a 50μF capacitor  ahead of the choke.  This dropped the initial ripple to about 0.9% which was then further dropped to about 0.001% by the choke input filter.  As a result, the output voltage rose to 250 volts due to the relatively low load current.  The load current is now about 44mA.  The power supply provides 250 volts directly to the 6L7 plate load resistors and is fed through a 4.7KΩ 10 Watt limiter resistor to the 0C3 regulator.  This is fed directly to the 6L7s.  The 10K plate load resistors on the Mixer and VCA 6L7s are used, along with 0.022µF caps, to decouple the power supply from the generated RF. This could also be done on all the tubes by using fairly large RFCs (at 360kHz, you would need about 100mH).

Update to the Update:  Adding the input (reservoir) capacitor did not eliminate the hum and noise.  It is possible that the choke input filter was just fine.  I found I needed to add an artificial ground to the filament supply.  This seems to have significantly reduced the hum and noise.  Too late to change my design now. 

GAS DISCHARGE TUBE REGULATORS – This is an interesting subject.  Maybe you all know this stuff, but I think a lot of people, including me, thought that the current limit on a gas discharge regulator was a limit on the power supply current.  It’s not.  It’s only the limit that the tube can regulate.  Confused?  I was for a long time.  My 0C3 datasheet says no more than 40mA can flow through the tube, and no less than 5mA if you want good regulation.  What this means is that you can have 40 mA flowing in the circuit or 500mA flowing in the circuit.  It doesn’t matter. What matters is that the CHANGE in the flow of current in the circuit can NOT be more than 35mA (40mA MAX minus 5mA Minimum) if you want the tube to live long and prosper.  Thus, the limiting resistor is the stuff of life. Because my +107v regulated circuit current only varies by a couple of mA, I only need to make sure that at least 5mA is flowing through the tube all the time.  I decided to run about 10mA through the tube.  I found that, even though the total power supply current remains almost constant (because the 6L7 screen current varies inversely with the anode current), too low a current through the regulator caused it to flutter, causing a loss of regulation.   About 19mA flows through the tubes and 10mA through the 0C3.  If the power supply voltage is 250v, then the drop through the resistor must be 250-107v=143 volts.  With a total current flow of 19mA through the tubes and 10mA minimum through the 0C3, the resistor needs to be 143v/0.029A=4900 ohms, so I used a 4.7K 10W resistor to push the regulator tube current a little higher. 

If the regulator tube fails, it is possible, in a circuit where the total current load exceeds the maximum change in current load allowed by the regulator, for the maximum power supply voltage to be applied to the circuit, causing damage to the components.  For this reason, most builders limit the circuit current to below the maximum regulator current load.

1.5    Basic principles

Two key principles come out of Lev Termen’s investigations – heterodyning of two radio frequencies as a means of producing audio frequencies, and magnifying the effect of hand capacitance by using very large loading coils.

The MIXER multiplies (hence Product Detector) the VPO and the FPO frequencies.  The result is a waveform containing the sum and differences of the frequencies.  Since the waveforms are not pure sinewaves, the number of sums and differences frequencies is effectively infinite.  By using a bypass capacitor (bypassing energy to ground) that presents a low impedance to the radio frequencies you are generating (0.001µF is about 400ohms at 370kHz, 0.01µF is 40 ohms, and so on) relative to the output load impedance of the tube, only the audio frequencies are effectively coupled to the next stage.

The mixer is critical in determining the tone of the theremin.  I have added, à la Termin, a variable air tuned coupling capacitor of about 3-30pF directly attached to grids #1 and #3 of the mixer.  The results of this coupling are audible in the sound samples below.  A different size capacitor would yield different results.

Update:  I discovered that the Tone Control coupling capacitor has a grounded shaft which was the source of unpleasant tone I was experiencing.  I was sure that I had checked the capacitor for an insulated shaft. All my trimmers have insulated shafts. Anyway, I lifted the shaft off chassis ground and all my issues went away.

LOADING (linearizing) COILS – The capacitance of the hand/antenna interface is very small, from 0 to about 1pf, maybe a little more.  It is wise to physically align the loading coils with the antennae.  That is, the coils, either miniature Hammond coils or a hand wound super large coil, should be directly underneath and axially aligned with the Pitch antenna, and the Volume coil horizontal and axially aligned with the Volume antenna.  You can see pictures of some of Lev’s theremins where this is the case.  It enhances the capacitive effect of your hand/arm as they near the combined coil/antenna.  The Q of the coil/antenna/hand combination is very high, meaning that a small increase in the capacitance causes the influence of the inductive reactance on the decreasing resonant frequency of the circuit to be enhanced.

PITCH LINEARITY – A search on the word “linearity” in the Theremin World Forums area yields a treasure trove of data on this fascinating topic.  I won’t go into detail here about it.  However, my theremin design has three controls which affect pitch linearity – an FPO trim capacitor, an FPO PITCH NULL control air-gap capacitor, and a VPO trim capacitor. The adjustment and relationship among these three capacitors has an overwhelming effect on pitch control – from no pitch control with hand movement at all to full pitch control with a movement range near the antenna of about 2cm.

For the initial setup, a scope or frequency meter is essential. You need to measure the FPO frequency with the PITCH NULL control set in the middle of its range, then set the VPO frequency with the antenna removed to about 4-10kHz lower than the FPO, then attach the antenna and recheck the VPO far field (hand not near the antenna) frequency. It should have risen to match the FPO frequency. If not, iterative adjustments of all three controls must made until it does.  This may include changing the total inductance of the loading coils.

I find there is not just one optimal pitch range for playing.  I like the lower ‘cello register range for some music and adjust the theremin to comfortably play in that range (C2-C6). For piano music, I adjust it to a larger range (C1-C7) and for soprano range music I prefer C4-F6, a narrow range with good linearity.  All these ranges are adjusted to occur over distances of about 55-60cm for the large ranges and about 40cm for the small range.

1.6    Sound Samples

These monophonic sound samples are from the final chassis build revision 2.7.  They are ‘dry’ samples. That is, they are recorded directly to digital from the audio output potentiometer so that the sound is not distorted by anything in the recording chain (other than the ADC preamp).  I did not remove the DC offset.  When heard through a good pair of headphones or speakers that reproduce significant energy in the sub 30Hz range, one hears the richness and fullness of the harmonics.

Here is an example of pitch range:

Here are graphs of high and low pitches: Note the change in harmonic content. Please forgive the redundant display.  I forgot to go to mono mode.



Here is a tonal variation:

Here are graphs at high and low volume of the same pitch:  Note how, when using some tonal variations, that the harmonic content is varied, not only by changes in frequency, but also changes in volume as in this case.  This is not the case for all tonal types.



The tonal range is remarkable for just a single variable coupling capacitor.

Here is a graph of the (I think) full tonal range waveforms of the theremin.  I am not sure since I don’t have markings on the Tone Control Knob so don’t know if it is full range or not.

Theremin test 20170508 full tonal range

Here is the matching .wav file with no editing.  You don’t have to listen to the whole thing, but you will hear me, starting at the fifth ‘blob’, change the tone control ‘on the fly’.  Please note that no adjustments were made during or after the recording.  You are seeing and hearing precisely what I was hearing in my headphones during the recording.  I limited the pitch range from C#5 (I think) to C#1.  The pitches are done from memory from beginning to end, no reference notes, so there will be errors in the actual pitches.