"Tanks" For Nothing!

After my latest experiences with coil winding and capless tanks, something dawned on me that I believe is at the very core of Theremin design.  I'm likely not the first person to think of this, and it is semi-embarrassingly quite simple, but here goes.

When working properly, the otherwise normal operation of the traditional Theremin LC tank is completely swamped by the resonance of the series EQ / antenna capacitance LC network.  Seeing the EQ inductor merely as some kind of performance enhancement device tacked on to a "real" LC tank (to increase sensitivity and/or improve linearity) is to completely miss the point.  Indeed the most sensitive designs stimulate the EQ / antenna LC at its natural resonance point, minimally load it when detecting its resonance, and help it form a high Q network.

So the large EQ (or "linearizing") inductor and antenna / hand capacitance is the real LC "tank" in a Theremin and all the action goes on there.  Circuitry we normally think of as the heart of a Theremin, the LC tank on the main PCB, exists primarily to stimulate and sense the resonance of this EQ / antenna LC network.  This "beating heart" can be almost anything that can perform this task and need not be LC at all.

Is there any advantage to stimulating the "real" EQ / antenna LC tank via another LC tank?  I don't believe so.  In fact, given the drive / sense requirements I think the drive / sense circuitry should by itself be non-resonant.

Can one take an LC circuit sans EQ coil and make a Theremin?  Of course, but it works best if the L is quite large and driven from the non-capacitance-sensing end by a low impedance (to preserve Q), and the C quite small (ideally existing purely as antenna capacitance) - which brings the designer right back to the EQ coil / antenna capacitance LC series tank as the real deal, and all else as something of a red herring.

Going forward my designs will not only be "capless" - they will be "tankless"!

Every component, the parallel resonant circuit and the series resonant circuit have their function. The real pitch response results from the interaction between both. See my "thesis" on this topic here: http://www.thereminworld.com/Forums/T/28530/antenna-tuning?Page=3 (scroll down).

You may naturally go to the extremes, make Lp and Cp tend towards zero, and get eventually a much more responsive pitch circuit, most probably waaaaaays too responsive for a classic heterodyning design. The question is, will this behavior feel good for the musician who plays the instrument, will he/she find a convenient tone spacing and a sufficient stability of such a pitch field?

Not every brilliant idea from "us technicians" will find favor in the eyes of the musician. That's why I always recommend that theremin builders should in a very first time not only take lessons and learn to play this instrument "the classic way", but also work close together with as many professional or semi-professional thereminists whose primary role in this game will be to bring you out of your ivory tower....

"one man's meat is another man's poison".

Therry,
I would agree with you if dewster was talking about traditional (heterodyning) design. But I assume he meant the digital theremin where the sound is synthesized using  measurement data. If so, he can feel free in handling of this information (linearize and adjust the responsivity as much as musicians want).

 
Too much sensitivity in HF part will help overcome the measurement dispersion and discreteness effects.

"Every component, the parallel resonant circuit and the series resonant circuit have their function. The real pitch response results from the interaction between both. See my "thesis" on this topic here: http://www.thereminworld.com/Forums/T/28530/antenna-tuning?Page=3 (scroll down).

You may naturally go to the extremes, make Lp and Cp tend towards zero, and get eventually a much more responsive pitch circuit, most probably waaaaaays too responsive for a classic heterodyning design. The question is, will this behavior feel good for the musician who plays the instrument, will he/she find a convenient tone spacing and a sufficient stability of such a pitch field?"  - Thierry

Some tunings are better than others, but my simulations for the EWS invariably produce situations where the pitch near the antenna is cramped.  I'm just not seeing any magic associated with the parallel tank + EQ, just an endless supply of mistunings.  I own an EWS, and my experience with it, qualitatively anyway, bears this out.  The tuning headaches alone have probably turned away droves from forming a positive view the Theremin.

When properly tuned, the EWS has sensitivity that is very similar to my capless / tankless designs (~4 %F/pF) and has very similar linearity, so these types of arguments against tankless (whether for digital or analog designs) don't really apply.

"

Some tunings are better than others, but my simulations for the EWS invariably produce situations where the pitch near the antenna is cramped.  I'm just not seeing any magic associated with the parallel tank + EQ, just an endless supply of mistunings.  I own an EWS, and my experience with it, qualitatively anyway, bears this out.  The tuning headaches alone have probably turned away droves from forming a positive view the Theremin.

When properly tuned, the EWS has sensitivity that is very similar to my capless / tankless designs (~4 %F/pF) and has very similar linearity, so these types of arguments against tankless (whether for digital or analog designs) don't really apply.

" - Dewster

I can see what you are saying - but I think there are several possible red herrings.. and in my expierience, this is the one area where simulation and reality, for some reason I dont fully understand (but I suspect its to do with the inductors) gives nearly meaningless results a lot of the time.

The idea of simply energising an antenna LC has been done - effectively using a large "EQ" inductance together with a small capacitance at the antenna (or only using the antenna capacitance) as effectively a series LC "tank".. And then there are the conventional parrallel LC tank circuits without and series "EQ" LC circuit, where the  the tank (paralell) capacitance is varied by player proximity..

As I see it, neither of the above has any "compensation" mechanism, there is none in theory, there is none in practice.

You say "I own an EWS, and my experience with it, qualitatively anyway, bears this out." - I suspect that there is a reason why you are not seeing any improvement in linearity - not sure what it is (wrong EQ inductance / tuning / whatever) but my expierience is that there is always SOME (albeit often marginal) improvement in linearity with a correctly tuned dual resonant (seperate tank and EQ resonators) topology.

My "visualisation" of operation is non-mathematical.. I am often stressing that theremins operate on capacitance change - and this is always true with regard to the player - antenna interface... But I actually see the operation at the tank side (on a dual resonator topology) as variable inductance - Effectively, it is change in the inductance of the tank (because the antenna resonator is acting like a variable parallel inductance across the tank inductance) this effective (or, as I have called it in the past, "virtual" inductance) is governed by the tuning of the EQ resonator, which is controlled by antenna-player capacitance.

There is a huge range of possible values for tank L and C for any given operating frequency - but for a given operating frequency the possible value for the EQ inductance is constrained by the available antenna capacitance.. It seems to me that best operation is obtained when the tank capacitance is large, and the tank inductance is small (for example, in the order of 2 to 3nF with tank inductance in the order of 100uH) rather than where the inductance is large and the capacitance is small (for example, C = 180pF, L = 680uH)

IMO, it is the selection of the values for these components (not just the values, but the type / quality - and also the physical components such as antennas connectors wire etc), and selection of operating frequency, which determines performance in terms of linearity and sensitivity.

IMO, some form of equalization is essential for a reasonably playable conventional analogue heterodyning theremin, where the audio output frequency is directly proportional to antenna - player capacitance.

And yes, poorly designed theremins with tuning problems will, I am sure, have put a lot of people off the instrument, and is probably the reason why the theremin (or at least playing them or developing them) only appeals to a certain "type" of person.

IMO, as long as one stays with the basic topology invented by Lev (as in, a fixed frequency beating with a players capacitively controlled variable oscillator) one needs to correct the entirely natural non-linear response one gets - the capacitive relationship is inverse square, the human pitch response is exponential - the two cannot naturally give an output frequency which sounds linear.. 

But we dont need to stay with Lev's topology.. Your capless AFE looks ideally suited to what you are doing - you can correct the data and provide a linear output through digital means. I am doing something similar by generating a voltage which I "equalise" through analogue means and then use to drive voltage controlled heterodyning (conventional)theremin modules..

Provided we can get repeatable data from the player-antenna proximity, and this data does not drift or distort, we can manipulate this data as we choose and use it in whatever way we want.. Analogue or digital, I believe this is the topology for future theremins.. It enables all the complex / impossible (for Lev's topology) tasks such as register switching or multiple simultaneous voices, to be implemented with ease.. Given a consistant signal corresponding to player-antenna distance, we can correct this signal to compensate for its natural law / non-linearities, and "map" this to the required law - whether this be an exponential output (to directly drive the pitch) or a linear output to drive pitch generators with an exponential law - and we can do this over the whole range - right from the antenna to the 60cm or wherever we set the 0 frequency point.

But, IMO, going back and trying to re-invent the original is probably folly - or it certainly has been an extremely expensive folly for me.. IMO, the most exciting discovery was how Lev's front-end really worked - But, while not entirely empty, even the promise this seemed to hold has evaporated a bit.

I think that Lev's topology will be with us for a long time to come (it is still by far the lowest cost and most elegant simple solution) but it has difficulties - I believe that you (and I, if I can be bothered anymore) have everything required now to overcome these difficulties using post-antenna (after the AFE) equalization / linearization.

Fred

Added -> There are other equalization methods employing feedback and / or oscillator coupling, and there are some theremins which I suspect employ shemes like this.. but I have never got my hands on a Tvox Tour or a Burns - its the Tvox Tour I am most interested in, because the designer assured me it did not use antenna equalization, and its damn good IMO!

BTW - this is just a brief visit - am moving home, getting divorced, trying to salvage my business and get work - and probably wont even have much internet access for the next month...

@FredM (just for info):

The linearity of the tVox is relative: It is fine in the mid range by carefully selected operating frequency and L/C ratio, it is fine in the low range thanks to adjustable coupling, and the field is progressively compressed in the two highest octaves due to the lack of further compensation.

Another example for a theremin without linearization coils is the Henk theremin. Here, basically the same applies as for the tVox, out of the fact that the coupling factor can not be adjusted. But the linearity in the high range is much better, perhaps because the antenna is longer and thinner, so that the ratio between dynamic and static antenna capacitance remains smaller when the player's hand approaches.

The tVox can also be somewhat more linear in the high register when a longer and thinner antenna is used (and the oscillator is re-tuned). But it never attempts Henk qualities, perhaps the large metal bottom plate plays a role or the different oscillator topology (tVox: difficult to understand without schematics, the coil is taped but it's no Hartley, Henk: Colpitts).

Both instruments have several things in common: 
- A relatively low pitch oscillator frequency (135kHz for the tVox and 150kHz for the Henk)
- Large inductances (tVox: 63mH, Henk: 50mH) and small capacitances in the parallel resonant circuit
- Large coupling capacitors between the tank circuit and the antenna (tVox: 1nF, Henk: 150pF) but I don't think that this plays an important role.

I think that the Henk has still a kind of "hidden secret" which I don't understand yet. But as soon as I get it, I'll tell about.

I'm working on my Excel simulation one last time (!) to make it simpler and more amenable to both tankless and parallel tank sims.  When it's done I promise to do a thorough testing of the EWS again.  As I've described previously in other threads, it has fairly high fidelity antenna and hand capacitance models, and all of the major inductance parasitics are taken into account.  Even with all this I've never seen any advantage to a parallel tank, just a bunch of downsides in terms of high currents and mistunings. 

I can imagine how the series EQ inductor / antenna LC can interact with a parallel tank LC in complex ways, but none of that plays out in simulation nor on my very physical EWS.  From this I can only think that parallel tank Theremins are like those rabbit ear TV antennas they (used to?) sell in stores with multi-pole switches connected to god knows what inside - most positions of the switch yielded poorer reception, but this was too easily misinterpreted by the lay user as degrees of improvement associated with the less worse settings.  (And by that please don't think I'm trying to insult anyone's intelligence here, this is complex stuff. But there is so much mud in the waters, much (most?) of it for prestige and profit.)

Has anyone here ever encountered a Theremin with a completely non-cramped - or better yet expanded - pitch range right next to the antenna?  If so I'm all ears, but I've never seen evidence of it in my Theremin journeys.  That might make me believe there is something going on that I haven't modeled / directly experienced.  Sans that, and particularly sans a clear explanation of that, I have to believe the parallel tank is existing mainly on smoke / mirrors / honest misinterpretation (again, not meant as an insult to anyone here).

(Personally, I'm kind of appalled it took my sorry brain soooo long to reach such a simple conclusion.  1.5 years into this "adventure" - and counting!  Ye gods!  I may have to hand over my genuine antique Theremin Shaman credentials to someone more on the ball.)

If I ever get around to actually marketing a Theremin, the last thing I want is some kind of crazy mystique associated with its operation - IMO there's too much of that currently going around.  I get trade secrets and all, but this field is just too small for it.  I mean, how do you trust the (perhaps) yo-yo that made your Theremin without some kind of mental paper trail or pedigree of the design process?  How do you trust them enough to fork over some serious change?

I'm not sure if the Excel way is the best way to study and to understand real theremin behavior. In reality, the zone where conventional linearization happens in way musically useful way is very small and it's only if the self resonant frequency of the series circuit is slightly below the self resonant frequency of the parallel circuit. But in this zone, even slightest changes in one of both self resonant frequencies can have a huge effect, if the Ls/Lp ratio is large enough. 

With both resonant frequencies approaching, the field becomes less and less compressed near the antenna thanks to an asymptotic behavior which will even limit the maximum frequency deviaton in a way that there remain only a few half tones between your hand at 1cm from the antenna and fully grasping it. That can be reached on the EPro in a simple way thanks to the small variable capacitor in the pitch arm and on the EWS by a) bending the antenna wire away from the aluminum foil to raise the series frequency and b) by tuning L5 to lower the parallel frequency. At the same time the maximum frequency deviation will be reached. It can be calculated to the half of the resonant frequency based on Ls and Cp (which should be low enough to make the linearization not happen only in the ultrasonic range but in the highest register by design...). 

BTW: the stretching of the high tone spacing is still better on the Epro because of the better Ls/Lp ratio (350 vs 320 on the EWS) which makes the asymptotic part of the curve still flatter.

In ever case one shouldn't bring both resonant frequencies too close one another because this will lead to the inverse effect, that everything jumps because it's suddenly the series circuit which becomes capacitive and the parallel circuit which becomes inductive.

I think that most of these are things can not be simulated in a satisfying way. To get the understanding of all these phenomenons and to get a feeling for what is a well shaped pitch field geometry, you will first have to study many different existing theremin designs, play music on these instruments, play with their tuning to see what has which effect on which part of the pitch field and what are the side effects. Then you'll have still to learn that the player's posture will play an important role: the same theremin which Carolina Eyck (almost fixed body-antenna distance, only stretching the arm towards the antenna) would consider as rather unlinear because of compressed tone spacing in the high register could be loved because of its "perfect" linearity by players like Lydia Kavina and Peter Pringle who approach their whole body to the antenna when playing in the high register which gives an additional field stretching.

My personal conclusion is that a good theremin designer (as any other conventional luthier) will have to study existing instruments in theory and practice, playing them and with them, and to study also different players and playing techniques in order to make an instrument which will not be appreciated mostly by engineers but by musicians.

A German proverb says: "Experience can't be replaced by anything"

@ Thierry :

"In reality, the zone where conventional linearization happens in way musically useful way is very small and it's only if the self resonant frequency of the series circuit is slightly below the self resonant frequency of the parallel circuit. But in this zone, even slightest changes in one of both self resonant frequencies can have a huge effect, if the Ls/Lp ratio is large enough. "

EDIT --- > I have just seen that I misread your statement WRT the circuits, and that you were not referring to SRF... So the following is redundant / bs..

Fascinating! .. Is this perhaps saying that a eq inductor with lower SRF may perform better than a "better" inductor with higher SRF under certain circumstances?

We would then need to add another layer to any simulations - one would effectively end up with parallel tank LC with a parallel antenna network across it - this network consisting of a series LC, except that the L would in fact be a parallel LC.. Rather than just juggling with the interaction of two resonant circuits (which with all the possible values that can be selected for operating frequency and components, is complex enough) one would be playing with an even more complex network.

But what you are saying does ring true.. and would (does) certainly give a possible explanation for some of the bizarre things I have seen.

All of which re-inforces my belief that Lev was either brilliant or incredibly lucky, or both - and also  re-inforces my belief that (certainly for me, who does not, even with advanced computational tools, have even a tiny fraction of the required mathematical ability or understanding required) splitting the sensing / equalization / audio generation into seperate "modules" makes the most sense.

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..

Many thanks for the info on the Tvox and Henk.. I believe (as it appears you do) that the long thin antenna will assist linearization, particularly close to the antenna.. I believe (as I have exponded before) that it is geometry which facilitates this.

Fred.

"If I ever get around to actually marketing a Theremin, the last thing I want is some kind of crazy mystique associated with its operation - IMO there's too much of that currently going around.  " - Dewster

LOL ;-) Dewster..

I think you have less risk of this because you are going the digital route! - However - at some future time, when the FPGA is an ancient art the way valves/tubes have become, I have little doubt that the theremin world will have found some way to mystify the operation of your theremin!.... I can just see it.... arguments over why pre-quantum-computing parts sound warmer and are more linear than post-quantum-computing parts.. Why circuits built using superconducting nano structures sound harsher than those built on old copper laminate FR4 boards......

Your problem, I suspect, will be at the opposite end of the spectrum for a decade or so.. You will suffer from the "I dont like it because it sounds digital" mystique.. Even Bob Moog, with his great reputation, didnt publish how his E-Pro produced its sound -

People seem to want "crazy mystique" and "mystery" - They (and I must include myself here) want to believe that they have some special "gift", that they can hear "qualities" that most other people cannot, that their "discernment" (be this with regard to linearity, or tone, or style, or whatever) is somehow more "refined" or "better" or "trained" than that of others..

Pander to this, and I suspect you will go further (in terms of sales and esteem) than if you are rational and logical and put all the facts on the table. Put the facts on the table, remove the bulls**t and (a) people wont believe what you are saying anyway and (b) most will be less interested even if they do believe you.

I think the above applies to everything, but seems to apply in excess to theremins and musical instruments - most people want a Shaman.

Fred.