Keppinger Theremin Oscillators

Posted: 10/14/2010 12:12:36 PM

From: Richmond Hill, Georgia

Joined: 9/18/2005

When I was building my Keppinger theremin, I recall that it sounded wonderful. What was not wonderful was the issue that the oscillators would not keep oscillating. I could momentarily restore function by grasping the large coils. The suckers would work for a minute or so and then quit. I believe the X factor, be it the transconductance of the tubes or my own unique and extensive personal magnetic bio-field, somehow prevented my theremin from working correctly (perhaps an unrealistic expectation). Permanent dunking of the unit in the nearby Atlantic Ocean was almost realized due to engineering frustration on my part.

I have a WW2 era tome, Radio Engineer’s Handbook, by Frederick E. Termen, Sc.D. I wonder if he was related to Lev? Anyway, one of the items discussed in the oscillator chapter was that the time constant formed by the grid shunt bias resistor Rg and the coupling resistor Cg to the tank was important.

The text stated that if intermittent operation of a Hartley oscillator is experienced, then reducing the time constant Rg x Cg may be helpful. In essence, the tank is not getting enough “kick” to keep it running. The text also stated that the preferable method of accomplishing the reduction was via decreasing Cg, and NOT Rg. Reducing Rg would change the bias operating point of the tube.

The initial values were Rg = 100 Kohms and Cg = .001 uF. I had reduced Rg to around 80 Kohms and voila!, the sucker worked.

Now memory being what is, I seem to recall that the theremin sounded better when it was intermittent. I posit that my changing of the grid shunt resistance, not only shifted the bias point of the tube, but also changed the tone of the instrument. In retrospect, I perhaps should have reduced the value of Cg. My napkin calculations indicated values in the neighborhood of 750 pF with the original 100Kohm shunt. Obtaining a variable cap in that value is not practical. So, one would tack solder a range of values from 470 to 750 pF and see if the intermittent operation is cured and the sound is better. Mica caps in that size are cheap.

My primary complaint has been lack of character in the tone, i.e. the waveforms closely resemble a sine wave at some points. The high frequency harmonic issue has been largely cured by careful selection of the two oscillator frequencies so any mixing products are outside the human range of hearing (i.e greater than 20 KHz), and careful selection of the volume oscillator bias point. If the bias of the volume oscillator circuit is advanced too much, i.e. more volume for a given hand position, “fuzzing” of the tone at low frequencies (approximately 500 Hz and lower) is noticed and is objectionable, rather than what one would expect from pleasant overtones.

I am a quintessential Electrical Engineer. I read this text while on the throne at my day job.



Philip Neidlinger
Posted: 10/18/2010 4:56:51 PM

From: Colmar, France

Joined: 12/31/2007

Philip, here my 2 cents:

Decreasing the grid resistor will increase the anode current of the tube because the same grid current will give a smaller grid voltage. This will drive the tube in the more linear range and naturally change waveform of the oscillator to a more sine wave like waveform. What makes the tube sound so characteristic? It's the asymetrical soft clipping of the negative half of the sine wave which gives a very similar overtone spectrum as a sawtooth wave (=more string like sound). So my idea would be to come back to the 100kOhm grid resistor and changing the value of the grid capacitor. If the oscillator doesn't work constantly, that means that there is not enough drive at the grid, so you should increase the grid capacitor's value because its impedance decreases with increasing value. If this theremin was mine, I'd put the original 0.001uF capacitor back into its place and solder experimentally 100pF or 220pF parallel to it and seek a point with stability of oscillation and a good sound quality. Since the Keppinger theremin has an additive mixer, you should do this operation synchronously on both, the variable and the fixed pitch oscillators in order to judge the results.

Please accept my apologies for the linguistic form of this posting. It was written quickly without spell checker and online dictionnary.

P.S.: Speaking of the time constant of Rg/Cg makes no sense in my eyes. They form a high pass filter with a 3dB frequency of 1.6kHz which is far away from the oscillator's frequency and thus will not have any effect. I tend rather to see both components in an independent way, Rg for the DC bias and Cg for the decoupling of the tank circuit with a slight influence on the feedback ratio.
Posted: 10/19/2010 12:30:01 PM

From: Richmond Hill, Georgia

Joined: 9/18/2005

This makes more sense than anything I have heard or read. As to the time constant, that was almost a direct quote from the textbook. However, the text was geared towards communications electronics, and not musical instruments.

I will experiment each way, ie adding and subtracting capacitance. I just want that elusive sound back that greeted me intermittently when my theremin spoke for the first time on my test bench.


Gabby on the test bench. (

You must be logged in to post a reply. Please log in or register for a new account.