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

It is not a counting limit but a so great frequency change that we exit from the 2 to 3 MHz band.

And (when using very large antennas) a hand can induce a so high capacitance, that the oscillator stops to work. With a limiting capacitor, this never happens.

And also, we have demonstrated that with a great capacitor, the 50 Hz induced from the mains wires, can be a significative noise font (if for example a 220 Volt lamp is very near to the antenna)

Found a mother-lode of tube books here:

http://www.tubebooks.org/technical_books_online.htm

Sigh.  Another thing to spend my time investigating.

"And (when using very large antennas) it is also possible to create so high capacity with the hand that the oscillator stops to work. With a limiting capacitor this never happens."

Aha!  Oscillator stalling!  Why didn't you say so?  I don't think this is due to high capacitance, I believe it is due to lowered Q of the tank (energy removed by conduction).  This is the problem IMO with oscillators that depend on adding "just enough" energy to overcome Q losses.  They can crap out.

I've never measured the Q of an antenna in free space, but it's very interesting that the Q is large enough to have this capacitance be the only capacitance that resonates with a fairly high Q inductor.  I've seen this on my bench.  Grabbing the insulated antenna with my hand very noticeably increases tank losses.

"And also, we have demonstrated that with a great capacitor, the 50 Hz induced from the mains wires, can be a significative noise font (if for example a 220 lamp is very near to the antenna)"

Yes, Fred brought this up a while back as a good reason to have a capacitor in series with the antenna.

Yes! Oscillator stops for the removed energy. Our oscillator is only 2 mW and can not send power to the antenna, max is about 100 uW after it stops to work. 

Yes it stalls! :)

But do not worry, with a series capacitor of 18pF or less it does not stops also with one square meter antennas and the whole body on the antenna. And works without stalling also with large antennas with any ambient temperature. Tested in the freezer and with the hair drier.

-----------

I have found an old 4069-cap-meter. 4069 contains logic gates that are not designed for low noise.

This video shows the noise differences between it and a new CapSensors with the FET BF862.

http://www.youtube.com/watch?v=N6IW2Q0ZRKA

The hand distance is about 80 cm and the antenna is only 130 square centimeters, these are difficult conditions also for the new CapSensors.

I know im going to be hated for saying this - But this is the  "Let's Design and Build a (mostly) Digital Theremin!" thread and in this context, not specifically in the context of the Theremino, I feel compelled to go back to basics..

The front end for a digital  theremin needs IMO the following charactaristics:

1.) Protection from destructive event possibility - IE, adequate ESD protection.

2.) Limited thermal drift.

3.) Maximum immunity to signals induced by electrical wiring and appliances.

4.) Maximum "raw" resolution so that adequate digital processing can be applied to the data.

And some of the routes to the above seem clear to me:

1.) A shunting ESD protection device between antennas and ground.

2.) careful design and selection of oscillator/s and components AND/OR active thermal correction by having onboard / at antenna thermal sensors.

3+4:) Use heterodyning and a small series capacitor or other high-pass filter between the antenna and the oscillator.

Heterodyning is, to me, the key.. If one gets a heterodyned output varying between say 1kHz and 10kHz, one has 9/10 loss of total resolution, if one has a direct oscillator variation of 1MHz to 2MHz (impossible to achieve) one still has 5/10 or 50% loss of resolution - By comparison, all other things being equal, heterodyning gives at least 5 times the resolution (within a given sample period) that any other method I know of can.

Dewsters charts show the advantage clearly, but perhaps the diadram below may aid visualisation:

In every complete cycle above, the green area is time where useful data can be captured, and the other area is wasted time. The direct methods results, regardless of any divisions, will result in a useful data capture time equal to the percentage that the oscillator frequency varies as a result of changing capacitance, the heterodyning method gives a useful data capture time equal to the period of the minimum difference frequency minus that of the maximum difference frequency.

For example, a 1MHz oscillator varying by 100kHz will have a useful capture time of 10% of the period (period varies from 1us to 1.1us or after division say from 1ms to 1.1ms), but the same oscillator with a reference oscillator giving (after heterodyning) a minimum frequency of 1kHz will have about 99% usable capture time (period varies from 10us to 1ms).

It is (and has been since my earliest attempts where I was developing a digital theremin which didnt use heterodyning) my opinion and expierience, that heterodyning simply clears away all major  problems related to resolution, and can more than compensate for any losses incurred due to rejecting LF noise by decoupling the antenna - both for analogue and digital designs.

In this regard (and I have not evaluated the oscillators of fine details of the circuitry) I believe the open.theremin is going the best way.

Fred.

ps - Dewster, that is one motherload! .. I found a stack like that a while ago and lost several months! ;-)

OK, I spent some time with LTSpice and livio's oscillator:

http://www.mediafire.com/view/2sobw16xxxzwtjl/livio_osc_2014-03-06.asc

Here's the transistor model I got from the manufacturer:

.model BF862 NJF(Beta=47.800E-3 Betatce=-.5 Rd=.8 Rs=7.5000 Lambda=37.300E-3 Vto=-.57093
+ Vtotc=-2.0000E-3 Is=424.60E-12 Isr=2.995p N=1 Nr=2 Xti=3 Alpha=-1.0000E-3
+ Vk=59.97 Cgd=7.4002E-12 M=.6015 Pb=.5 Fc=.5 Cgs=8.2890E-12 Kf=87.5E-18
+ Af=1)

You have to add the model text (without the '+' and in a single line) to the "standard.jft" file.

Here is a picture:

I guess I should have seen right off that C3 in series with C1 and C2 is yet another capacitance adder, which further reduces sensitivity. 

If I'm simulating this correctly, with a 10pF antenna I get 2.50591MHz, with 1pF additional (11pF antenna) I get 2.48222, which gives a %delta of 0.945%.  (Read the frequency directly via the "Spice error log" CTRL-L.)

Theoretically, with ONLY a 10pF antenna and 1pF adder, one can get a 4.65%delta frequency change with a simple LC tank.  So that's the best one can do without heterodyning.  It seems livio's circuit is then giving only 0.945/4.65 = ~1/5 of the available sensitivity (if I didn't screw up the simulation).

Certainly stability is important, but it isn't everything as are many other characteristics that can make an oscillator unsuitable for Theremin work.  But this sensitivity is pretty low, I think too low for me to consider using in my stuff.

The thing with sensitivity (or, digitally, "resolution") is, if you are forced to amplify it via heterodyning, then you will likely be amplifying thermal drift as well.  So my goal is to get the most stable sensitivity I can, and perhaps augment it with a bit of heterodyning if necessary.

When I first started looking at oscillators I ran across a lot of circuits meant for VFO RF work, and most looked very similar to livio's circuit.  If constructed correctly they can be very stable for single transistor circuits, and with the addition of suitable variable capacitors and output buffers may be used in critical applications.  But they aren't usually designed for absolute maximum sensitivity to capacitance change.  I believe they are designed more for stability and spectral purity, though of course pullability via a voltage variable capacitor is likely important too.

===========

I also want to say that it's my feeling the deleterious effect of noise of the amplifier in an oscillator is very dependent on how that gain is employed (oscillator topology), how the output of the oscillator is used, and the nature of the noise itself.  We generally think of noise as bad, and all things being equal (they almost never are) we want less of it.  But our efforts to reduce noise could be at the expense of other desirable properties, so we as engineers can't make sweeping value statements about it.

"...they aren't usually designed for absolute maximum sensitivity to capacitance change..."

This is true!
But, until now, we have not found better alternatives. 

Last weeks we tried to follow your ideas and to build some heterodyning oscillators at 300 KHz and we were forced to change direction by instabilities and trimming problems. Having to use an inductor for over 10 mH, its parasitic capacitance, its mechanical instability, the total oscillator PCB dimensions, the sensitivity losses and the need to add trimmer capacitors to be calibrated in the factory are for us unthinkable things.

Please suggest a better solution (with frequencies and component values) and we will test it as soon as possible. 

(but please, a simple and workable solution, not a theoretical block diagram, perhaps impossible to achieve and ever made by anyone) 

(and please, not the OpenTheremin oscillator, with capacitive trimmers and oscillator sensitivity reduced to almost nothing, by parallel capacitors for a total of nearly 150 pF)

LTSPICE LIBRARIES

LTSpice simulator is faster, more efficient, and well done for all the ones we tried till now. In years past we used them almost all, from suffering to another, but since we use PSpice everything is simplified and after the first difficulty has become a really nice friend. In addition LTSpice is freeware.

LTSpice, kindly distributed by Linear Technology apparently only supports chips from them, has very few generic components in its libraries and is a bit strange editor (but just learn to use it well)

Finally, the components of LTSpice are disproportionate, resistors and transistors are very large and therefore the appearance of the patterns is not serious ...

Solution: 
1) Download latest version from LTSpiceIV: http://www.linear.com 
2) Install it normally 
3) Go to the folder "C: \ Program Files \ LTC \ LTspiceIV" and delete the folder "lib" 
(Who prefers to be less bad can rename the "lib-old") 
4) Download this ZIP: LTSpiceIV_Libraries_for_Theremino_simulations_V1.6 
5) Remove the "lib" folder and put it in "C: \ Program Files \ LTC \ LTspiceIV" instead of the original lib

Once this is done you can open all our simulations without any problems and you benefit of thousands of transistors and diodes. Although all operational amplifiers will be available by selecting them with the most used and finally OpAmp5. schemes are no longer strange with huge resistors and "bizarre" transistors.

This is better explained in our download pages, and more precisely here:

www.theremino.com/en/downloads/uncategorized/#ltspice

"if you are forced to amplify it via heterodyning, then you will likely be amplifying thermal drift as well. " - Dewster

I dont see this - Apart from if one had two oscillators with different thermal behaviours, otherwise IMO heterodyning can even act to reduce or compensate for thermal drift.

Any form of straight "amplification" (Whether this be heterodyning or extending the period through division) will amplify signal and noise equally, so the SNR wont change, and AFAICS this also applies to thermal error.

"Please suggest a better solution (with frequencies and component values) and we will test it as soon as possible

(but please, a simple and workable solution, not a theoretical block diagram, perhaps impossible to achieve and ever made by anyone)

(and please, not the OpenTheremin oscillator, with capacitive trimmers and oscillator sensitivity reduced to almost nothing, by parallel capacitors for a total of nearly 150 pF)"

 - livio

I dont like this..

This IMO should be an exploration, not a contest, and when asking someone to provide designs, tying their hands behind their back aint helpful - Also, your ideas of "simple" and someone elses are not likely to be the same - to me "simple" is fitting a damn discharge tube, to you its fitting a 20kV capacitor!

Apart from which, Dewster has presented several oscillator designs here which to me look far better suited to digital application than your oscillator - but he is going the route of air coils - and as I understand it you want ready-to-buy Ferrite inductor - This is one of the things which annoys me about "newcomer" theremin designers (and I was the same!) They look at things like air coils and think "i can use a ferrite" but dont ask the simple question "Why did a competent designer use an air coil if a ferrite is just as good".

and please, not the OpenTheremin oscillator, with capacitive trimmers and oscillator sensitivity reduced to almost nothing, by parallel capacitors for a total of nearly 150 pF

The 150pF is there by design! It doesnt "reduce sensitivity" if one is comparing it to the theremino design, because the open.theremin employs heterodyning, and in fact the RESULT of this is GREATER SENSITIVITY than the theremino!

I really dont believe that you are in any position to be critical of anyone elses design because, quite frankly, there is nothing particularly novel or good or clever about the theremino design - the only thing I have seen from you which does look really good is your software, but I havent explored that deeply. Your claims for theremino are what gives it the edge over the open.theremin, but some of these claims (resolution for example) have been shown here to be completely untrue.

I certainly think the open.theremin is capable of FAR higher resolution due to its topology, and MUCH lower latency, than your present theremino can even dream of.

I do however agree with you about capacitive trimmers, and I think the open.theremin oscillator could / should be improved, and that probably a better oscillator for this application could be designed to replace both the theremino and the open design.

As you are both "open" projects, why not collaborate? Get the best of both of your designs and software - I really think that with your software, open.theremins heterodyning topology, and perhaps an oscillator improvement / redesign, some really good digital modular theremin stuff could get to market.

One "problem" is that you are looking at short to-market time scales - This is great in that you actually do get to market.. But to do a GOOD theremin takes much more time than you are giving it - in the case of Dewster and I, years have been devoted - In my case, I could have put many products on the market which were as good as other stuff there but I went too far the other way in seeking perfection -

In order to impact the market, you need something special - You wont get that by rushing - but you can get a bigger impact if you believe things about your product which arent true, and convince others to believe this as well.

If, like me, you arent a salesman - and if you pull your products apart more mercilessly than any competitor or critic ever could, then you aint going to sell anything even if they do get to market - unless you really have something exceptional and believe in it and can convey this.. But getting there takes a long time!

Fred.

"Please suggest a better solution (with frequencies and component values) and we will test it as soon as possible. 

(but please, a simple and workable solution, not a theoretical block diagram, perhaps impossible to achieve and ever made by anyone) 

(and please, not the OpenTheremin oscillator, with capacitive trimmers and oscillator sensitivity reduced to almost nothing, by parallel capacitors for a total of nearly 150 pF)"  - livio

Oscillator choice IMO is at the heart of Theremin design, and it's been dogging me since the beginnings of my investigations.  I'm currently trying to simulate the Open.Theremin oscillator, which clearly is a design adapted from a crystal oscillator, and coincidentally was what I started with, but you are right livio in that it almost certainly has a very low basic sensitivity due to the high parallel capacitance.  As Fred points out this is boosted via heterodyning. 

I think there is likely a happy medium between stability and overall sensitivity (which includes heterodyning if it is used) to be struck in a successful Theremin design, and quantifying the location of the peak of that happiness seems the thing to do, though I feel rather far away from being able to do that.  And of course there are other things like stalling to consider.

Due to the LC 90 degree shift at resonance, I don't think there are many simple (e.g. one transistor or single inverter) solutions.  livio, your FET oscillator employs slightly delayed in-phase low gain drive with a parallel tank.  The Open.Theremin uses 180 degree inverter delay with 90 degree RC delay, high gain drive, and a mostly series tank.  The Smirnov designs are similar to the Open.Theremin but are more sophisticated and involve a second series inductor going to the antenna.  The EW uses in-phase parallel tank drive with a series tank hanging off of it.  My designs use phase lock techniques to maintain 90 degrees at resonance with high gain drive and a series tank.

Even if you have the luxury of doing phase lock stuff in high speed logic, the problem of sensing without loading down the tank or introducing a phase shift in the sense portion can be an issue.

I believe simple parallel tank in-phase drive must be low gain in order to not lock up.  Simple series tank quadrature drive must generally be high gain in order to sustain oscillation but otherwise will not lock up due to the phase offset between sense and drive.  Phase lock techniques, depending on how the phase detector is implemented, can lock onto harmonics rather than the fundamental.

"As you are both "open" projects, why not collaborate? Get the best of both of your designs and software - I really think that with your software, open.theremins heterodyning topology, and perhaps an oscillator improvement / redesign, some really good digital modular theremin stuff could get to market."  - FredM

This is what I was thinking too livio.  Help the Open.Theremin project by improving the oscillator (I don't know, but I suspect your oscillator is better), reap the sensitivity benefits from their heterodyning method, and use this to drive your PC software.  Were I to get involved that's what I would do.  I'd also consider using an NCO rather than a simple divided xtal for the heterodyning source, which would let you remove the variable capacitor, and would allow software to adjust the null point automatically (or semi-automatically with some user input via software).

I believe that Thierry is somewhat involved with the open.theremin project now, and he has probably had more exposure to real theremin oscillator circuits (including the EW-Pro,Henk and Tvox oscillators which are reputed for their stability) than anyone else - With his input regarding oscillators and his unique insights aquired on the bench rather than in a simulation environment.

Add to this input from people like Dewster who has done a great deal of work on simulating many aspects which were never evaluated so completely before he did this, and livio and others who have played with and produced oscillators, and Gaudi's topology, and Dewsters digital skills, and livio's software, and the other collective skills of all or some of these people, there is possibly the most capable team ever assembled to make a really good mudular digital theremin a reality.

I really think that IF people truly want a good "open" project (and are not just using the word "open" as a disguise for promoting their products) there is enough skill and knowledge "out there" for this to be realized NOW.

IMO, the only thing preventing this at present is the competitive aspect of human nature - Even "open" projects fall foul of this, and we all (and I certainly include myself here ;-) want some of the "glory" - whether this is justified or not, we all seem to want the spotlight to fall on us, or for our ideas to be central to any collaborative project,  even if money isnt what we are after.

If it wasnt for this, and if we were more genuinely collaborative, humanity would probably have collonised our solar system and be on our way to the stars by now! ;-)

Fred.

"The 150pF is there by design! It doesnt "reduce sensitivity" if one is comparing it to the theremino design, because the open.theremin employs heterodyning, and in fact the RESULT of this is GREATER SENSITIVITY than the theremino!"

More the oscillator sensitivity is low and more it must be "amplified" in some way. Heterodyning is a good method to amplify little frequency changes of the oscillator but, doing that, you amplify also the oscillator instabilities and the temperature induced frequency changes.

Here you can see a PDF with the sensivity of the most known oscillator configurations:
www.theremino.com/files/Theremins_Sensitivity.pdf

And here a test of the CapSensor sensitivity:
CapSensor oscillator sensitivity - YouTube Video

And the following video are some tests of the CapSensor working at long distances:
Theremino Theremin, chords played at great distance - YouTube video

Theremino Theremin, eight octaves range with steady notes - YouTube video

A large area antenna for CapSensor tests - YouTube video