An ADAU1701 "Theremin"

Discussion created by KJBob on Jun 23, 2018

   Invented in 1928, the Theremin -- Theremin - Wikipedia was a milestone in early electronic music.  Ninety years later, you can demonstrate a Theremin-like device with just a ADAU1701MINIZ board, a speaker, two resistors and a pair of metal objects to serve as antennas.


   In this version, a fixed 44 KHz oscillator drives the antennas through 22 K ohm resistors. A nearby hand contributes capacitance to ground, adding phase lag,  The project detects this phase shift on each of its two antennas, driving the pitch via a VCO and the volume with a simple multiplier.

As one might expect, approaching the pitch antenna raises the pitch.  In the original Theremin the volume decreases as one moves toward the volume antenna.  In my version the volume increases with proximity since this helps insure silence when not being played.  This of course could easily be reworked to provide traditional operation.  As shown below the 22 K resistors are under the orange tape, and two cat food bowls serve as the antennas.  



   Adjust the fractional delay for a near-zero measurement at both readbacks.  An auto-cal arrangement brings volume and pitch to zero at startup, eliminating tedious fine-tuning (at the risk of stating the obvious, stay away from the antennas during this five-second startup delay).  Since the external circuit ties up both input and output jacks on the -1701 eval board, I used the board's class-D amplifier for the output audio.  The speaker I had handy is a 8" ceiling mount with a 70 V transformer -- this keeps the volume low.  Approach the volume antenna too closely and the thing overloads and screeches -- "That's not a fault, it's a feature!"  In any event, the fix is the punch line for the old joke, "Doctor, it hurts when I do this."  Sometimes I hear "bumps" in the pitch and volume, apparently caused by noise pickup at the antennas -- after all, the circuit by design must be very sensitive to what happens there.  Making a 1 KHz change in the operating frequency accompanied by a corresponding re-tuning of the fractional delay often cures this.  Powering the board (in self-boot) from a switching wall-wart caused hum modulation for me -- I suggest a 5V regulated linear supply.  If anyone can make this project more "bullet-proof" from these effects, I'd love to hear about it.


   Likely you might imagine additional features which would prove to be "ahead of what's possible" in the analog world -- multiple VCOs in selectable major/minor/7th harmony, for example.  Presently these are limited by the project's 192K sample rate with its corresponding instruction count.  You do need a high operating frequency to sense the pico-farad capacitance changes.  Even worse, the multiplying phase detectors double the 44 KHz working frequency to 88 KHz.  If despite this you can get the thing to work at a 96K rate, you'll have plenty of extra instructions to play with.


   The now-defunct kit company, Ramsey Electronics, used to print on their cartons, Electronics is fun!  --  where I work we often joke about this when things aren't going well, yet this project was fun for me!