Let's Design and Build a (mostly) Digital Theremin!

"The next post will explain why your theoretical graphs regarding the area of the antennas, do not correspond to reality (they take in consideration an antenna isolated in space and forget to consider the noise and the static capacitance of the measuring and driving electronic components)" - Livio

I hope it does, because at this time I can see almost no relevance regarding the noise of the oscillator transistor - the whole question of even moderate noise in the oscillatory circuit is, as far as I can see, of quite low importance.. (with capacitive sensors noise becomes a critical factor at the comparator and post-sensor stage, particularly if one is trying to obtain a period measurement of a high frequency - but even here, noise contribution from the oscillator is trivial compared to noise contribution from reference voltages or comparator / active components following it))

As an analogous example - you have a swing (pendulum) being pushed with a regular pulse, this is the resonant circuit - a little wind blowing on the strings on the axis of thrust of the swing aint going to change the oscillation frequency one iota! .. in fact, one could get the resonance even if one was to drive the oscillator with bursts of noise, and the frequency would still be defined by the LC as far as I can see.. Only sideways variations of wind could significantly affect the frequency, and this would be equivalent to capacitive "noise" which is entirely different to amplitude noise as I see it.

Of course, heterodyning followed by a good filter gets rid of all but the most extreme (as in, really badly designed or faulty oscillator) noise problems.

" (a radial capacitor with 20 KV isolation)."

I am interested to know where you get a 10pF 20kV NOP/COG capacitor - because at such a low value capacitance, the dielectric is important - any capacitance change due to temperature will significantly affect pitch (the N4700 dielectric common in HV capacitors has about 0.5% / degree C variation in capacitance!) .. If you were to use a couple of larger 10kV capacitors (say 1n or minimum 470pF each) in series then themperature variation will be less significant and you could probably get away with almost any dielectric (the larger the series capacitance, the less significant any changes to the value of this capacitance due to thermal effects becomes, because the antenna capacitance increasingly becomes the more significant contributor to the final sum).

20kV capacitors I have sometimes been forced to use for other projects have been horrendously expensive, and usually its far more economical to use 10kV's in series.. For theremin circuits its far more economical to put a discharge tube in the circuit (between antenna and ground), and this also gives better protection.. you can then safely use good low cost  100V NOP / COG capacitors and eliminate thermal drift from this source - this allows you to use smaller series capacitance if you need this to reduce sensitivity close to the antenna.

Once again, I dont understand the aversion to discharge tubes or other shunting type ESD protection! Trying to block an ESD spike with a series isolating component is, IMO, just insane! The cost and problems with such a component, and the extra care required in layout etc to insure that this component actually has any real use, makes this a complexity that IMO is utterly pointless!

A 12kV ESD spike can jump tracks and go routes you never expect - The only way to ensure safety from these events is to kill them at source!

Fred.

PENDULUM

You example is good to explain the integration of the hi frequency noises (100 Hz and up) but the noises I meant, are in the range 1 to 10 Hz and they modify the oscillator frequency with the following mechanism:

1) An oscillator is never completely insulated in the empty space, it is always connected to sensing components and driving components.

2) The comparation voltage of the sensing components and the driving current, are slightly increased or decreased by the instantaneous component of the low frequency noise.

3) This changes slightly the phase and the oscillation frequency.

Frequency changes produced by te noise are very little, but also frequency changes produced by a hand at 50 cm that moves by 1 mm are incredibly little!

Frequency changes produced by the noise, are not different by those produced by the hand movements, they are all in the 1 to 10 Hz range, so no filtering can attenuate the noise, without attenuating the useful signal. 

About six years in the past, we used 4069 oscillators like OpenThereminUNO and the frequency was very noised, compared with our recent BF862 oscillators. With 4069 the noise was becoming to disturb at distances more than 30 cm. Now I can hear stady notes also at 1 meter and more.

CAPACITORS

Please, stop with those temperature dependance problems, we do not use Heterodining so we do not amplify oscillator defects, so we do not have temperature problems.

And also, as already explained, musicians are frequently tuning because of ambiental changes, instrument movements, people that moves around... These detuning sources (with our solutions) are preponderant. Frequency changes caused by the temperature are so slow and minimal that the musician correct them with millimetric distance changes, without notice them at all.   

- - - - -

With our last Kapton antennas super high voltage capacitors are not needed so now we use super stable COG/NP0 6KV models like this:

FARNELL UK - COG NP0 10pf 6kv rad

- - - - - -

Precedently we used the following 20KV models. But all the capacitors below 100 pF are very stable, specified or not. It is so easy to build a 10 pF capacitor that there is no reason to use hi capacitance dielectrics. I can do a perfect 50KV capacitor twisting two HiVoltage wires in some second. But naturally we prefer Kapton antennas.

Application :X-ray machine, medical equipment ,electrostatic spray equipment ,welding machine ,HF equipment ,etc.

Ceramic dielectric properties

Please notice that 520 ppm is better than EW bobbins...

"Please notice that 520 ppm is better than EW bobbins..."

Sorry livio, I am not trying to be critical, just trying to be helpful. I do not understand how you "do not have temperature problems" if you have a 10pF 520 ppm capacitor in series with your 10pF antenna.. This makes absolutely no sense to me.

[Ok  - Just did quick calculation and 520ppm isnt too bad - But bad enough!  I need to check those capacitors out as they are better than any other HV capacitor I have found before - but the principles in my post still remain valid.]

But I aint going to argue - you say theres no problem.

Regarding noise - I really dont understand anything you are saying about that, I dont know where the 1Hz-10Hz is coming from - this is a tiny part of the noise spectrum and should be only a tiniest fraction of the overall noise component, so cannot see why you had problems in this area, I have never had or observed such problems.. Problems up in the 50Hz range due to magnetic and electrical field interference yes - but thats a different subject.

Also, heterodyning doesnt amplify noise or errors, it in fact dramatically improves the signal to noise ratio.

Fred.

"You should take into account, that in series with C1, there is the antenna capacitance, which in this scheme is 10 pF, but that in our 100 to 200 square centimeters standard antennas, is much less."  - livio

A 100cm^2 disc antenna has self-capacitance of ~4pF.  And there are stray capacitances that are likely on the order of this as well, so say the antenna has 8pF.  Connecting this to a voltage divider that throws away ~half of the signal seems kind of crazy to me, particularly when dealing with such small capacitance changes in a noisy environment.

"In the CapSensor we already put a​ very large one (18 pF), to allow maximum sensitivity, even with very large antennas and while working very closely. But in a Theremin would be good to reduce it, to increase the linearity, to avoid too large frequency variations, when the whole hand is in contact with the antenna and to increase the resistance to ESD events."

I understand how this capacitor can increase linearity near the antenna by decreasing nearfield relative sensitivity, but you are already doing post processing, so why not correct for non-linearity with a few multiplications?  This would give you the the best of both worlds.

On one hand you are doing all you can think of to increase sensitivity (paddle shaped antenna, distance the antenna far from capacitive metal and noisy electronics, super low noise oscillator transistor, etc.) but with this tiny series capacitor you likely wipe out much of the gain you've worked so hard for.

About the 10pF capacitor please read with more attention my post. It looses some sensitivity (about 30%) but this is OK (not for linearizing but to limit the max frequency change) 

We do not use a single cycle. We measure multi cycles and integrate for a time about 1 to 20 mS (dynamic resolution) but noone can integrate for more than some ten of mS, without reducing also useful signals. The useful signals (hand signals) are in the same frequency range of the 1 to 10 Hz noise, so it can not be filtered out.

To really get the maximum distance you have to reduce the noise as much as possible. The noise reduction from a digital component to a FET is not a 30% (like the sensivity lost by our 10 pF capacitor)... it is a 100 times rapport! Digital gates and comparators are not designed to make a ultra low noise front end.

livio,

I think we (or certainly I) are trying to understand why you are doing the things you are doing!

I can understand having a small series capacitor to reduce close-to-antenna note compression on an analogue theremin, particularly if one has a small tank capacitance and more sensitivity than one needs, so can afford the reduction in sensitivity..

But I dont understand it for a digital theremin where one want to maximise frequency (period) variation, and where one is just getting data at as high resolution as possible to process in a digital "engine".

As I see it, the whole reason for leaving the analogue implementation is to seperate the sensing from the processing and sound production.

How one implements the capture of data and processeng thereof, and whether this be from single or multiple cycles is an irrelevance - One can only capture data you havent thrown away, and I agree with Dewster here, it looks like you are throwing away a reasonable percentage (and even 10% of resolution being needlessly thrown away would IMO be silly) of your resolution away.

In fact, I really dont understand any of the things you are doing! You are full of energy and enthusiasm, but argue about every, even simple, suggestion made to you! - Why is increasing this capacitor any big deal? Why is adding a discharge tube a big deal? What is the point of being here if you are going to stick with your ideas even if they are shown to have simple-to-implement improvements ?

There is no single theremin design that is better in every way than anyone elses - there are designs which may have some better implementations than yours, and your design may have some better implementations than others. I also think there is a HUGE difference between "theremins" designed for an exibition type environment or for casual entertainment, and those for profesional or serious thereminists, and you may be making a big mistake by thinking the technology from one can be simply ported to the other, no matter how "cool" they look.

Fred.

The sensivity loosed by the capacitor is about 30%, but without it the frequency changes are not limited at all.

The capacitance of a hand completely surrounding the antenna is very high and our oscillator is so sensitive to the capacitance variations the we prefer to limit the max frequency changes, also at the expenses of a little decrease in the sensitivity.

The decrease in sensivity is 30% (or even less depending on the case) but the limitation of the max frequency change is huge.

And also, we have demonstrated that with a great capacitor, the 50 Hz coming from the mains wires, becomes to induce a significative noise font.

"The capacitance of a hand completely surrounding the antenna is very high and our oscillator is so sensitive to the capacitance variations the we prefer to limit the max frequency changes, also at the expenses of a little decrease in the sensitivity."  - livio

OK.  If you have sufficient sensitivity and things work out so that decreasing it solves other problems and gives you a happy medium, I understand that.

I'm curious what the inherent noise is in a tube, particularly at low frequencies?  A stable Theremin can obviously be built with tube-based oscillators.

Tubes were very similar to the FETS and some tubes where very low noise components. Also transistors can be low noise but only low noise voltage. Transistor noise current is 1000 times more than the FET nouise current. FET are component with an incredibly high gate impedence and about zero current.

"The decrease in sensivity is 30% (or even less depending on the case) but the limitation of the max frequency change is huge." - livio

Ok, I understand - its a way of maintaining a period width in which usable counts can be obtained when close to the antenna, and reducing sensitivity in order to achieve that.

This is perhaps one of the problems caused by having a high oscillator frequency combined with a not-so-fast counting clock.

Problems which go away in a blink if you use heterodyning!  ;-)

Fred