"Tanks" For Nothing!

Randy Rote: "I think there is one very significant reason why a theremin greatly benefits from a linear playing interaction: to make the instrument easier to play. And I mean universally easier... this means a person with a moderate facility to play the instrument will have a better experience, and the person who wishes to play with virtuosity on the theremin can also reach higher levels of virtuosity...."

Far too much time must be spent by a newcomer to the theremin learning to play on pitch, even when that newcomer possesses all the required talents and skills (muscle memory, hand/ear coordination, etc.). No other instrument demands that kind of investment of energy and hours just to develop the skill to do something so fundamental that doesn't even have to be thought about by players of most other instruments (with the notable exception of the family of viols). 

When young people learn to play a traditional instrument and take lessons from an experienced teacher, they not only learn the technical tricks of the trade, they also learn MUSIC. By "music" I do not just mean notation, I mean phrasing, expression, and a personal style and taste. This, in my not so humble, is often what is missing in theremin performances - even performances from people who seem to have mastered the skill of playing on pitch - because almost all thereminists are self-taught. Clara Rockmore said of the study of the theremin, "It takes hard work, sensitivity, sensibility....attention to detail. The music comes from the heart, the mind, and YEARS AND YEARS AND YEARS of the study of music." 

The notion has been floating around for a long time that if some engineer could come up with a dependable theremin, with perfect linearity, the problem of the difficulty of the instrument would be solved. While such an improvement would be very nice, I don't think it would do much to make the instrument significantly easier to play because the reason it is hard to play in the first place has nothing to do with the fact that theremins are not perfectly linear. 

The theremin is hard to play because of its lack of any visual or tactile reference. Perfect linearity is not a fix for this basic problem. In fact, I doubt that there is a fix for the problem. Paul Tanner tried to find one back in the 1950's and built his "tannerin" (electro-theremin) but unfortunately his solution to the problem of theremin uncertainty also destroyed the unique and enchanting theremin "soul".

One other thing, perfect theremin linearity will depend on the thereminist remaining absolutely still. When I play, I use a Tai Chi stance and move my body position all over the place, significantly changing the pitch control "threshold" and the configuration.

I do not reach into the electrostatic field. I swim in it. 

"When I play, I use a Tai Chi stance..."  - coalport

When I try to tune my EWS I'm strongly tempted to take a Karate stance and chop the thing in two!  I remain almost as mystified by the thing as I was the day I bought it.

The Post In Which dewster Possibly Eats Some Crow

Question: When an EWS is properly tuned, is the tank resonance above or below the EQ / antenna resonance?  The reason I ask is I'm seeing some linearity improvement in my Excel Theremin simulator when the tank resonance is set a bit (~20 kHz) above the EQ / antenna resonance:

But the effect seems to be fairly touchy and highly dependent upon the tuning.  I'm trying to find this effect on my EWS but there are so many interacting adjustments that it's difficult to tell if I'm getting warm or cold.  With the local oscillator set for proper null I haven't found any setting on my EWS that causes a pitch drop as my hand nears the antenna.  Though I believe I've encountered settings that kill the oscillations, which is a bit disconcerting.

Dewster, you are about to get it! As I explained in the other topic (link somewhere above in this topic), the serial resonance (EQ/antenna) has to be slightly below the parallel resonance. That makes that the eq/antenna circuit is excited above its own resonance and will thus behave itself as a small inductance. This additional inductance is to be seen in parallel with the parallel tank circuit and makes it thus initially go up(!) in frequency. If you add now some hand capacitance, the series resonance will go down which makes the total inductance of the EQ/antenna circuit increase. Thus the parallel tank "sees" more inductance in parallel which makes it go down in overall oscillation frequency. Since the effective inductance is calculated as the parallel circuit of the initial inductance Lp of the parallel tank and the virtual inductance of the series tank 1/Ltot = 1/Lp + 1/Lvirtual, you will find that the variation of the overall resonant frequency is the less sensitive, the bigger Lvirtual compared to Lp. That means that, since more hand capacitance makes Lvirtual grow, the circuit becomes less sensitive to the hand capacitance, the bigger the latter becomes. That is linearization.

And yes, the range within that phenomenon does not only happen, but give musically useful results is very small, so that the tuning of the L5 in an Etherwave has to be done very carefully, within +/- 150Hz. It is effectively as you found highly tuning dependent.

Conclusion: there are no tanks for nothing. In a classic theremin design, be it by L. Theremin himself or by R. Moog, it is the subtle interaction between both tanks, the initial oscillator tank (be it parallel or serial) on one side, and the EQ/antenna circuit on the other side, shape the pitch response: playing range, sensitivity and linearity. The latter depends mainly on two factors: the ratio between the eq and the tank inductance, and the (difficult to set up) gap between the resonant frequencies of both tank circuits.

That's one of the reasons why I tell people whose Etherwave has gone out of tune (cannot longer set up the zero point with the pitch knob) to correct that only on the fixed oscillator (L6) without touching at L5 since the latter has much more influence on the pitch response and is thus much more tricky to set up. I have tuned more than 100 Etherwaves in my life and although I have acquired this little experience, I still need about one hour to tune it for optimal pitch response...

@Christopher: you might theoretically be right with all your thoughts and theories, at least nobody has proven the contrary for the moment. But that doesn't mean that your behavior, throwing your stuff into almost every discussion about oscillators and linearity, is acceptable. No, it's trolling and this prevents people from seriously considering your ideas.

This topic is about the classic serial linearization invented by L. theremin and widely used by R. Moog and others. Your schematics have absolutely nothing to do with that very special phenomenon and are thus simply off topic. You are cordially invited to start your own discussion thread(s) about linearization without EQ circuit.

Thierry, thanks so much for that clear explanation and your comments re. EWS tuning.  Taking one hour to set up and EWS is something I find very easy to believe!

Here is a view of the "Bode" worksheet of that sim:

By inverting the start and stop frequencies (grape arrows on left) one can force this sim to deal with the highest resonance encountered (green arrow on graph) rather than the lowest (purple arrows on graph).  The "Linearity" worksheet resonances can then be confined to the desired region by judicious setting of the start frequency.  The higher resonance point gives fairly low sensitivity (green arrow on left).

I can see how this effect might be used to marginally increase linearity, and thus playability, near the antenna. 

The following shouldn't be construed as an attack on anyone, nor as me puffing up my designs at the expense of others:

The process of good design employs cost / benefit analysis for the inclusion / removal of any significant feature.  The question then is this: is this fussy triple tuning of a parallel tank (EQ / antenna; LC tank; fixed "beat" oscillator) worth a minor increase in linearity near the antenna?  Different designers can legitimately have different answers to this. 

My feeling is "no" - I'd rather just let the EQ / antenna resonance fall where it may and then adjust the "beat" oscillator (or its digital equivalent) for null and call it a day.  Uncorrected linearity in the mid field is really pretty decent, but the opportunity is always there to post process any non-linearities in the log 2 domain via a quadradic - we can do these things while also altering note spacing and offset to taste.  Along these lines Fred's earlier comment in this thread is particularly germane:

"Personally, I can deal with non linear signals from a "transducer" which has a known response, and convert this signal to any required response - this is "bread and butter" EE stuff.. Rarely (if ever) do we try to modify the natural law of the sensor - oh, in some rare cases, where a simple mechanical change can produce the required law (for example with a float, where a shaped actuator can linearize a response) we may opt for this route - but usually we opt for processing the transducers raw data by either analogue or digital computation.."  FredM

I also want to minimize any factory tuning as much as possible, spending an extra hour of labor on each one would cut into profitability, and doesn't sound like much fun either. ;-)  Tuning means mistuning - from drift or curious fingers - and mistuned units out in the field are poor ambassadors for a design.  Just my 2 cents.

Dewster, in my eyes you are wrong talking about only a minor increase in linearity near the antenna. If well designed and well tuned, there is a huge difference between an EQed and an un-EQed oscillator. When you go to the extreme, grasping the pitch antenna and thus adding a hand capacitance of >=100pF, an un-EQed oscillator will go down by 30kHz or even more while the deviation of a well EQed oscillator will be limited to 5 or 6kHz. That's a huge difference!

In terms of cost / benefit analysis there might exist cheaper ways nowadays where almost everything can be digitized, numerically processed and then, if necessary, be converted back to analogue. But using such technologies in a theremin would move it more far away from the rather classic music instrument approach which Leon Theremin had in mind, but making it just another cheap and easily to mass-produce DSP device. The idea of almost zero factory tuning kills totally the artisanal aspect of traditional music instrument building. Am I right that you never took the time to move the chevalet of a violin tenths of millimeters up and downwards, experimenting the minimal but often important changes in the tonal character? 

Building (and optimizing the sound of) traditional music instruments is (among others) the meaningful use of the multiple and natural full and partial resonances of every single part. A similar approach, the resonances of the multiple resonant circuits and also the impulse response of the interstage transformers in Leon Theremin's tube designs gives them their unique tonal character and playability. I still doubt strongly that a similar quality can be achieved with normalized control functions.

Thierry said: @Christopher: you might theoretically be right with all your thoughts and theories, at least nobody has proven the contrary for the moment. But that doesn't mean that your behavior, throwing your stuff into almost every discussion about oscillators and linearity, is acceptable. No, it's trolling and this prevents people from seriously considering your ideas.

The thread is about tank circuits and achieving ideal linearity. I gave one opinion and have not posted in months, what exactly are you up to?

You need to stop disqualifying what I post. Get your finger off the delete button or your hall pass will be taken away.

Thank goodness for the roll over.

Christopher

"Anyway, I can't quantitatively check simulation linearity without proper resonance detection, and the above is something of a can of worms." - Dewster

Just a quick interjection here (Moving out is taking longer than anticipated - damn estate agents just keep dragging the process and demanding more non-returnable fees to do meaningless credit checks etc)..

I dont know about the EW (although I see no reason why this shouldnt work with the EW) but I have used a method to crudely determine resonance, which you may find useful..

I have a seperate big (>= 10uF) and parallel low ESR/ESL decoupling/filtering/smoothing capacitor across the supply to my variable oscillators, and an accessible link upstream (I use a 0.1" "handbag") to connect the supply to the oscillator.

By replacing the link with a low Z mA meter, I can monitor current drawn by the oscillator - As the majority of this current is determined by the antenna current, and as this current is primarily determined by proximity to resonanace, this has been a useful tuning and diagnostic aid.

The advantage of this scheme over others, IMO, is that one gets a change in DC current (integrated by the supply capacitors) and there is nothing in the HF antenna path to distort or change the resonant frequency etc.

I have actually used the above scheme in some designs to facilitate linearization, by providing a seperate control voltage derived from this current as frequency control feedback.

Fred.

ps.. @Thierry .. I think the "art" comes down to the playability and quality of the tone. I do not believe that it has anything to do with how fiddly or difficult the instrument is to set up.

With electronic musical instruments, the playability comes down to (1) the quality of the player / instrument interface IMO.. Simple example - the quality of the keyboard has a HUGE effect on playability - the same instrument equipped with a good weighted keyboard is hugely superior to one fitted with a crappy plastic keyboard.. (2) the quality of the tone, regardless of how this is produced.

I therefore do not agree with

"The idea of almost zero factory tuning kills totally the artisanal aspect of traditional music instrument building."

I would however agree that having loads of internal presets or whatever to adjust will make the likelyhood of producing nearly identical instruments far lower, and therefore each instrument is more likely to be "unique" ... But is this a good thing? Also, is "traditional" really a word that can be used with any "merit" when it comes to electronic musical instruments?.. "Tradition" spans many generations, there is no person alive today who witnessed creation of the first (original) violin, or even a great grandchild alive today of anyone who witnessed this (in fact, I think the 'original' violin probably goes back more than 1000 years)

- but we have musicians alive today who met (and were taught by) the inventor of the theremin, and nothing like it existed 100 years ago! IMO, we cannot talk about "traditional" electronic musical instruments - not until a few hundred years more have elapsed!

I do not believe that "luthier" or even perhaps "artisan" are really the correct words for creators of electronic musical instruments - oh, I agree that they are nice words, and create an aura of something special - but I do not really think they are justified or accurate - not even for Lev or Bob! - I think the better word would be pioneers.. They (and us) are just at the start, these instruments will evolve (and some perhaps go extinct as is the way of natural selection) - perhaps this is what makes it exciting, and why perhaps a disproportionate number of engineers,inventors,developers and "explorers" are drawn to this field...

IMO, we are not traditionalists, we are innovators and pioneers.. Traditionalists work on creating and refining traditional acoustic instruments using traditional methods and not introducing massive changes or innovations, because these have, well, a tradition! The electric guitar is, IMO, an example of a "traditional" instrument where the applied innovation moved the instrument outside the remit of "tradition" - too much innovation and change, and one moves into a murky realm on the fringe of tradition - IMO,  No such "problem" exists with purely electronic instruments - they are new, and have no ties to tradition - even if they try to have (Theremin Cello, for example).

IMO, if one has a load of adjustable controls, these should be accessible to the player, so that they can be adjusted to the players preference - and they should not be critical - as in, the instrument should always be playable regardless of how these controls are set.. Or, at least, adjustment should never endanger the functioning of the instrument - so if one did adjust such that the instrument ceased to play, it could easily be brought back to operation. I do not see any point in making the instruments more difficult or costly to manufacture or calibrate than is absolutely essential, or that designing for minimum (or zero) manual adjustment at manufacture in any way "kills" anything!

I think this was one of Bob Moog's great insights - He produced instruments which gave musicians huge control over the sounds and interface / control.. In the main, his instruments did not impose "presets" with his or someone elses idea of a sound or even mode of operation - He let the musician be the "artisan".. I believe that he made a big mistake when he failed to extend this principle to the theremin, where things like the volume response curve was invariable, and control of tone was limited.

Bob's designs (and analogue electronic musical instruments in general) have required a lot of manual adjustment at manufacture, but this is being reduced in recent designs - with analogue, one is often forced to trim circuits, but this is not by choice - and 'new' programmable resistors and potentiometers (and other mixed signal components) are making automated calibration easier to implement, with huge beneficial results.