Crazy (?) theoretical / technical ideas

"But one thing I have pondered over for some years is whether there could be any way to "see" electrical / capacitive "fields" - I abandoned this thinking after going over all the possible mechanisms I could think of, and concluding there was no way with known science (or at least science known to me).."  - FredM

Maybe not the same thing, but maybe close, is capacitive tomography.  In one application there are electrodes placed at an equal distance around the center of a flame, forming a ring.  By measuring the capacitance between the various electrodes and feeding this to an algorithm, a 2D "image" of the flame is formed.  (I don't know why, but tomography always seems like borderline magic to me - output info somehow more than the input info.)

In the past, images of standing sound waves were formed by waving a stick with a microphone driving a light on the end (a visual SPL meter), with an open shutter camera taking a time-lapse photo of the whole thing.

"Maybe not the same thing, but maybe close, is capacitive tomography." - Dewster

Yeah - ive had minor involvement with some ECT stuff long ago - Its kind of theremin like.. one is using an array of capacitive sensors to provide data which is then processed, hopefully to produce a usable image.. (ECT was used on a large project several decades ago here in the UK - AVO was involved in building some of the equipment - not sure if its supposed to be secret, LOL, but they were searching for an ancient chariot that got sunk in quicksand on the coast - this chariot was carrying the kings treasures and collected taxes etc, loads of gold and other valuables - when the tide came in and the sand sucked it all down.. Electrical Capacitive Sensing and LF antennas sunk into the ground were used in an attempt to locate it - but they failed) (or at least thats what I heard ;-)

I suppose though you are right - it is a bit similar.. If one had a single sensor (antenna) and was able to move a shielded ground point around this changing its distance etc, and recorded the output from this sensor, then you could map the fields tomography... But to do this one would need seperation from all other grounded objects - and would get quite a boring and worthless 'map'..

I am thinking more on the lines of a thermal imager - point a "camera" at a theremin and player, and be able to see by the colour of the image the field strength - to observe the capacitive coupling to other grounded equipment and to the player.... But I think this is way in the star-trek future! ;-) .. And when it comes, the least likely use for it will be theremins!

Time to stop - been trawling through piles of incomprehensible physics and maths trying to see if electrical fields through an air dielectric will give resistance to motion of those particles in the field - and am inclined to think that what I read indicates they will - but dont have enough comprehensible data or understandable maths to be sure, let alone be able to quantify what any such force comes down to in real life.

One thing I did find was that th really high dielectric constant of H2O makes (as far as I can see) humidity a far greater potential cause of theremin drift than I had expected .. my sums are probably wrong at some fundamental level, because it looks to me like even a few % change in humidity will have far greater effect than even a massive change in temperature..

But time to put this all (and me) to bed, and leave it to fester for a month or two at least - none of this stuff is going to help me right now even if I got the answers to it all....

Fred.

"Interesting mixer circuit Fred!  You're making me want to stick my digital woes on the shelf for a while and build an analog Theremin or two!" - Dewster

IMO, unless one commits entirely to analogue, and follows this "Yellow Brick Road" in the search of heterodyning mysticism, my mixer is absolutely pointless..

Looking at the waveforms one can predictably generate using heterodyning, and comparing these to the rich pallette of controllable timbres one can generate with my mixed signal (logic gate) mixing circuits (ramp, triangle,pulse,square,multiplied and divided versions of these) - and the simplicity of mixed signal, the invariance of amplitudes concequential to audio frequency (allowing one to apply ones own frequency dependencies simply using filters) .. Well, IMO the mixed signal approach beats the analogue multiplier approach hands down.. Even for an analogue theremin..

So, why am I wasting time on this kind of mixer? - Simply because I want to have one conventional fully analogue heterodyning voice -  As I have committed to having a "conventional" mixer ( one reason also being sales / marketing as rightly or wrongly, use of logic gates for audio will attract derisory comments from some competitors - like those who sell a 2 transistor pitch only board as a true theremin, for >£90 and others who spout garbage to the theremin ignorami ) I want this mixer to be as exceptional as possible - to extract maximum "theremin-ness" from its available timbres - to emulate, as closely as possible, the behavior which produced the sounds from Lev's mixers.. Changing the signal levels into the Fets above does get quite close to what I would expect from the tube mixer in the RCA (The MOSFETS have lovely distortion similar to tube distortion when overdriven) and making these adjustable might approach the behaviour of the Claramin - but only if followed by specific audio processing (which I am working on).

I am not utterly convinced that conventional heterodyning has nothing "special" but I am inclined to think that its at least over-rated .. By having both, I am covering all possibilities - particularly if I can create an above average quality heterodyned voice.

Having a two voice theremin (one being "true" heterodyning, one being mixed signal, both being operable simultaniously) will give a full tone pallette (including sine etc from the pure analogue heterodyning voice) capable of keeping everyone happy - I hope!

But for low cost and simplicity, the far more versatile mixed signal voice is probably all one really needs IMO.. Whilst the CMOS Analogue heterodyning mixer looks simple and low cost, its price / complexity increases rapidly when one adds control elements to allow timbre to be altered, as one is dealing with HF signals.. With my mixed signal 'heterodyning' one gets audio frequency waveforms out that can be simply mixed together with low cost standard audio mixers etc - everything is clear of the horrible RF zone..

Fred.

"IMO, unless one commits entirely to analogue, and follows this "Yellow Brick Road" in the search of heterodyning mysticism, my mixer is absolutely pointless.."  - FredM

Crazy idea 0xACDC:

1. Use a standard LC oscillator for the pitch side.

2. From (1) somehow obtain a number that corresponds to hand location.

3. Run (2) through a mathematical function. 

4. Use the result of (3) to generate an analog waveform.

5. Heterodyne (1) and (4) together with an analog mixer to get audio out.

I think the idea is crazy because the player would likely audibly detect the various lags in the response of the various processes.  And it's not as straightforward as just proceeding digitally from (2).

Dewster:

1. Use a standard LC oscillator for the pitch side.

2. From (1) somehow obtain a number that corresponds to hand location.

3. Run (2) through a mathematical function. 

4. Use the result of (3) to generate an analog waveform.

5. Heterodyne (1) and (4) together with an analog mixer to get audio out.

Its obtaining the "number" and processing this which is the crunch - its dead easy when one has this with high enough resolution, to generate a HF "variable oscillator" waveform and a "reference oscillator" waveform out, and do analogue heterodyning on these.. If one has a digital processing scheme, one can simply output a reference square wave (say 250kHz) and generate a VFO square wave (say from 244kHz to 250kHz, or whatever span one chooses), put each of these through a LC filter tuned to say 247kHz, (or more elegantly, through a variable HF LPF/BPF to adjust the harmonic content) and get two sines out that can be analogue heterodyned.

As I see it, there is no need to heterodyne 1+4, doing so will (as I understand it) greatly complicate the maths required in (3) if (1) is variable frequency -  I would want (3) to process the linearity curve.. and this is only "easy" if one has a constant against which to apply the function (or at least its easier to me ;-).

Heterodyning 1+4 is effectively what I am doing with my "upside down" topology, but this only works because (1) is in a PLL so its locked to a constant reference frequency - the analogue computed maths takes the "number" (2) from the PLL and runs it through a mathematical (analogue) function (3) the result of which controls a HF oscillator (4) which is heterodyned with the same frequency as (1).

I have recently (almost) decided that I dont need a PLL locked antenna oscillator - that I can simply drive an antenna LC with constant reference frequency, derive a CV from this in a similar manner to volume antenna circuits, process this voltage (linearity, span etc) and use it to drive a HF VCO which is mixed with the reference oscillator - The maths function becomes a little more complex, but this is more than compensated by removal of the antenna-side oscillator and PLL..

Added ->

"I think the idea is crazy because the player would likely audibly detect the various lags in the response of the various processes.  And it's not as straightforward as just proceeding digitally from (2)." - Dewster

In essence, I dont see the idea as crazy at all - If one can get the 'numbers' fast enough and with high enough resolution, there is absolutely no reason not to produce HF signals based on this and perform analogue heterodyning on these.. No extra processing overhead would be required, and one could have a hybrid digital theremin outputing true heterodyned audio (as well as whatever other wavetable or "digital" output you choose)..

ok, disclosure time ;-) .. I am looking at FPGA to implement register switching and "mixed signal" heterodyning - I am looking at 3 voices - one being true, pure analogue heterodyning which isnt messed with by the FPGA in any way and doesnt switch registers - it stays at its natural pitch regardless of the register setting and is purely analogue derived directly from the reference and variable oscillators..

Then the FPGA processes two register switches which can be locked together - these control two seperate voices, one being my mixed signal voice, the other being another analogue heterodyning mixer driven from signals generated by the FPGA and filtered (external to the FPGA probably with LC - but register switching is giving some headaches here as this idea requires switching capacitors in these filters) to produce sines / modified sines from square waves.. I might dump this voice if it becomes too complex / costly.. But if I do dump it, I will probably dump the FPGA - It wouldnt be doing enough to justify its expense (unless I implement a DUI).. And I am balancing between FPGA and PSoC.. FPGA gives more possibility of other intervals, PSoC is home ground...

All voices would be locked to the "natural" voice but can be shifted in (as yet to be confirmed - certainly octaves up / down, but hopefully some other musically useful) intervals reletive to the natural voice. The tonal charactaristics and levels of each voice independently controllable.

Fred.

Crazy idea #lost count 1:

Thinking about your last, Dewster..

If your FPGA derived a number (whatever means) and you were to go for an external analogue heterodyning "engine", would / could this simplify the maths processing ?

You would have two HF outputs whose frequencies you could control - there would be no need to keep any one of them fixed at a constant frequency, all that would matter would be the difference frequency..

So if one was to decrease the reference frequency proportionally (or via some simple function) to increase in capacitance "seen" and perhaps apply a similar function to the "variable" oscillator output, one might (?) get a linearizing function more easily (??) - One could also apply different multiplicands to facilitate span control etc....

Oh, I know - its probably nonsense.. Its the kind of thing id mess with because the idea of doing anything more complex mathematically sends chills through me ;-)

Fred.

....... 

"If your FPGA derived a number (whatever means) and you were to go for an external analogue heterodyning "engine", would / could this simplify the maths processing ?"  - FredM

I haven't worked it through (and probably won't ;-) but I suppose the math processing might not be too different.  What I was aiming for with the crazy idea was to have digital correction and waveform generation in the loop, but still have analog heterodyning going on for the audio output.  Audio out from an FPGA is more difficult than it should be.

"...would it be possible to use a magnetic field?" - TheIMUman

Build it and (if it's crazy enough) they will come!

"Audio out from an FPGA is more difficult than it should be." - Dewster

Heterodyning makes it easier! I have implemented my mixed signal stuff in a simple PLD, one gets an output you can take straight to a RC filter, and get ramps, triangles, squares at the difference frequency - And theres no reason why the "reference" and "variable" frequencies shouldnt be generated inside the PLD using "hardware" counters one loads with the required divisor.. A few gates, one D-Latch, a couple of counters, all inside the PLD output to about 3 external R's and C's and one has audio!

Fred.

Crazy Idea 01010: HF Variable Sensitivity Range Mixed Signal Theremin [EDIT: for analog Theremin use]

High frequency oscillators get you away from the AM broadcast band and let you use small air core coils.  A potential problem in using them is that heterodyning generally makes them too sensitive.  An obvious way around this is to divide both the VFO and the local oscillator (running at 2 to 3 MHz) equally down into the 100's of kHz zone and heterodyne the result: [EDIT] After a bit more thought this clearly only changes the register of operation.

The VFO could be a Clapp adapted for Theremin use.  The local oscillator could be a more conventional Clapp, with smaller L and larger variable C to lower the voltage swing, minimize capacitive pickup, and to set null.  Or the local oscillator could be from an NCO.  One could use "tracking" powdered iron or ferrite L as well (but I probably wouldn't).

The divisor n might go from 2 to 32 or so, giving some nice steps to adjust sensitivity register.  Null would conveniently stay put while changing n.

The semi-crazy part here is the division, which would give square waves into the mixer - people might like something other than harsh square waves coming out.  Hmm, heterodyning two square waves via XOR gives a nice sine triangle wave.  And heterodyning via DFF gives square waves.  How to get even harmonics? 

Damn it Fred, you're turning me into an analog Theremin designer!