Theremin Pseudo Science

I am pedantic about physics correctness, (my mode of PC, LOL!) and find loads of utter nonsence being promoted on sites related to Theremins and adverts by Theremin manufacturers.. And example of this is:

"Analog Theremins will have oscillator coils"

Now, if it was stated that "Inductive (LC)oscillators give better linearity" then there would possibly be some truth.. BUT to equate "coils" to "analogue" is just utter nonsense - I can get rid of coils and still produce an LC oscillator by using a gyrator circuit to effectively simulate an inductor by using a capacitor's transfer function... I wouldnt do this, because the circuit is (much) more expensive than an inductor - but it would be an entirely analogue circuit which did not have any inductors.

The term "RF" is bandied about - Sorry.. but when one talks about RF these days, one (as an engineer) does not think about a 250kHz sine wave! There is more RF radiated from the PC you are using than you will ever get from a well designed Theremin! - And even with a Theremin based on a 555 timer will only 'radiate' a triangle wave at say 250kHz.. The energy of the highest frequency components of this waveform are hardly 'RF' by any standards, and are at such high impedence as to be utterly trivial.

It is quite easy to construct an entirely analogue classic theremin, with sine reference and pitch oscillators, and not use a single inductor (coil).. The frequency -> capacitance ratio is almost identical to an LC oscillator, and there are techniques to make it identical (or even better) to a LC oscillator in terms of linearity.

So - When you see ignorant assertions that "analogue = coils" or that "antennas made from Latinum improve performance (LOL)" BEWARE! - Best case, the advertiser is talking rubbish to woo you and put you off a competitor who uses different circuit topology.. Worst case, the manufacturer doesn't understand the basics of electronics, and does not know the difference between digital and analogue!

Its nice to see the influence of postings in this forum!

I notice that one particular advert which got me angry, has been changed by the advertiser..
In truth, I think I was probably a bit "heavy" in my critisism.. Lots of 'ignorant' things get said in sales blurb - it is the nature of advertising.. And, in truth, if a (Theremin)circuit has coils then it is almost certainly an analogue design.. and most (Theremin) circuits which do not have coils are inferior, because they use simple RC relaxation oscillators (which includes circuits based on 555 timers or logic gates)which tend to also be far lower cost than LC oscillators.

When is a circuit "Digital" and when is it "Analogue"? - Well, use of logic devices or 555 timers does NOT always mean the circuit is "digital".. If it did, then every circuit which used any form of comparator (and this would include almost every type of analogue waveform generator or voltage controlled oscillator) would be "digital"..

Even the crudest Theremin circuit which uses logic gates .. see http://www.therasynth.com/html/a4093cct.html is not really a "digital" circuit, as it acts by charging its capacitors (an entirely "analogue" action), discharging its capacitors (an entirely "analogue" action), giving variable capacitance controlled frequency out (an entirely "analogue" action)..

Past this stage, the first critical (and significant) departure from the classic Theremin "Analogue" circuit (and the first instance where the digital vs analogue argument becomes valid) is that the outputs from the reference and pitch oscillators are square waves, and these are 'mixed' DIGITALLY through a NAND gate - this is the first AND ONLY Digital action undertaken by the circuit, and technically the circuit COULD thereform be described as hybrid digital/analogue.. but if the waveforms were instead mixed in an analogue ring modulator, the circuit could not be described as in any way "digital".

Once passed into the filter (R3,C3) the circuit returns to purely analogue - the digital components are all above 100kHz, and the filter gets rid of these completely.

So, even this cheapest, crudest Theremin circuit possible could not in any sense be described as digital - And this circuit will not work any worse than a poorly designed "fully analogue" circuit using expensive coils..

What DOES make a BIG difference to the sound is the shape and quality of (high frequency)waveforms fed into the hetrodyning mixer, and the quality of this mixer.. And here the "Analogue" argument really holds up..

With square waves from the reference and pitch oscillators, there is only a 1 and 0 state from each (even if these waveforms are fed into an analogue mixer) - so difference frequency components due to change of waveform are not 'reflected' on the output waveform.

The most audibly noticable effect of this is when the pitch is changed rapidly - as this causes ("good") distortion of the HF pitch waveform which will be heard more if the HF oscillators are each producing a sine or sine-like waveform and being mixed in an analogue ring modulator.

I wouldn't worry about it too much, Fred. It's part of a larger phenomenon I call [i]Everybody knows except the experts[/i].

Like this:

Everybody knows what a tree is, but ask three botanists and the clouds will roll past, the sun will rise and set, seasons will come and go, civilisations will rise and fall and they will still be arguing about it.

Everybody knows what music is, but ask three musicologists and the clouds will roll past, the sun will rise and set, seasons will come and go, civilisations will rise and fall and they will still be arguing about it.

and so on...

You are so right....

...but: I think that an individual detail of a theremin design will not decide about the theremin's overall quality. It is the "play together" of all components like in a mechanical watch.

Similar:
There are bad tube amps on the market and very fine Class D digital power stages and vice versa.

Out of that, back to theremins, there is surely an important thing for marketing: The vintage factor. I observed that theremins whose design is near the original Leo Theremin or RCA design have more attraction on me than a CMOS single chip based one.

Still one word on FredM's RF remarks: When I was young and started with electronics, that means at the end of the sixties, 250 kHz were really considered as RF and the circuit design rules were quite different from those of an audio circuit. So, still today, it is a little wonder for me to have 18 Mbit ADSL over my unshielded 2 wire telephone cable...

FredM, please forgive us elderly people our different view on some aspects of modern electronics. Thank you! ;-)

"FredM, please forgive us elderly people our different view on some aspects of modern electronics. Thank you! ;-)"

LOL! Thierry, you look about my age, or younger! - Yes, I am definately one of the "old folks" - When I was a kid, transistors where experimental objects in labs, and when I built my first synthesiser at 14, the cost of the semiconductors (741 op-amps and transistors) took me a year of work (building crystal radios for local wireless store) to earn!

And you are right.. back in those days 200kHz was "RF" !

Well you could build "analog" oscillators quite easily with Wien Bridge or similar circuits ... no coils at all!

I resemble those old man remarks! We were taught how to design vacuum tube flip/flops in one of my college classes in the 1970s! (I thought at the time "what, are you nuts!", but that knowledge actually did come in handy in more recent times).

I have to admit that I was amazed at how easy it was to make a Hartley osc. work right off compared to trying to get a Wien bridge oscillator to cooperate.

Don

I think this all boils down to the great battle between analog and digital. I like to think that analog is simple in its complexity and digital is complex in its simplicity.

Thierry >> "I observed that theremins whose design is near the original Leo Theremin or RCA design have more attraction on me than a CMOS single chip based one."

They do for me as well! If it wasn't for the power supply requirements, size, heat, fragility and danger of 'tubes' I would MUCH prefer to use them for all my audio designs.. But, alas, they are a luxury few can afford - I want a polyphonic valve synthesiser (even a monophonic would do as a start) but must compromise by having a valve pre-amp driven from my synth.. 2 tubes is all I can manage!

Using CMOS gates for Theremin design is the other absolute extreme of the spectrum - they work, but are only of use really for experimental 'instruments' not designed for serious playing.. the gate thresholds are too variable, and there are other 'nasties' .. But they do work, are cheap, and simple... I made a Theremin birthday card for a friend using a SMD 40106 and it was quite cute.. Silent when the card was closed, and almost playable when the card was open... The sort of application one could not do with valves! LOL!

The TS555C (fast CMOS 555) is a big improvement on CMOS gates, as the thresholds are accurate, and the CV input allows tailoring of the response.. These oscillators can compete with the much more expensive LC oscillators IF extra bufferring and waveshaping is added..

But if I was buying a Theremin and price was not an issue, I too would go for something with coils (if price was NO object, I would go for a valve RCA) - even though I know the performance of circuits without coils can be just as good (or in the case of the original TherAsynth design, better).. So I have given in.. I realise that I will have a big problem selling instruments as "High quality" if there are no inductors!

.. Funny old world!

FredM, I think that you may use coils (for the eyes of a eventual buyer) in a digital circuit, i.e. in series with the power supply for Vcc decoupling. Wind 130 turns of green coated copper wire on a 4cm diameter, add a blue LED to illuminate it, and you are done... *just kidding*

Back to topic: I would not like to have square wave signals at the mixer inputs, even if it would be a linear 4q mixer, you'd get out not only f2-f1, but also 3*(f2-f1), 5*(f2-f1) and so on - a square wave like output. A simple low pass filter with fixed cut-off frequency would not fix the problem. Sine waves from the oscillators would give a smoother sound, wave shaping (or adding some distortion) is easy to do after, but removing unwanted harmonics will be much more difficult.

"I would not like to have square wave signals at the mixer inputs, even if it would be a linear 4q mixer, you'd get out not only f2-f1, but also 3*(f2-f1), 5*(f2-f1) and so on - a square wave like output. A simple low pass filter with fixed cut-off frequency would not fix the problem. Sine waves from the oscillators would give a smoother sound, wave shaping (or adding some distortion) is easy to do after, but removing unwanted harmonics will be much more difficult."

In theory, everything you state above is true.. However, in practice, I have not been able to hear any difference between digitally generated HF and analogue generated HF pre-mixer signals - and spectrum analysis of the mixer output also shows no (measurable) difference..

I think the reason is the pre-mixer filtering I use.. A 5V 200 - 220 kHz square wave fed into a simple two pole RC circuit, reducing its amplitude to 100mV P-P maximum, gives a reasonable sine wave.. The amplitude variation as a result of the 20kHz change on the variable oscillator frequency (10%) does not make any noticable difference to output amplitude posat-mixer.

It is essential to to get the waveshapes of the signals going into the mixer (analogue 4Q modulator) clean and 'pleasant' as you say.. Harmonics generated as a result of mixing can sound awful, and cannot be removed easily. Digital "mixing" in fact sounds better than digital (square wave) signals mixed by an analogue 4Q modulator, as a simple digital (logic) mixer can be designed so that not all the product terms are generated, and the output can quite easily be filtered to give a reasonably pleasant wave.

With all the above, it comes down to cost.. It is probably easier to generate sines with inductive oscillators, and mix these in an analogue 4Q modulator, than it is to get a suitable digital equivalent.. But analogue components are more expensive. The 4Q modulator (mixer) is an area where severe cost cutting is performed even in some of the 'best' analogue Theremins.. A REAL 4Q analogue multiplier with low noise, like the AD633, is expensive .. but, in my opinion, the sound from a fully implemented good quality 4Q multiplier is FAR 'nicer' than that from a low cost mixer which uses semiconductor (PN junction) non-linearity to implement heterodyning.

My plan is to produce encapsulated modules - A digital module and an analogue module - and provide these to those evaluating the first Theremin units I produce.. They will be able to swap the modules, and select the one they like best.. They will not know which module is digital or which is analogue... I will then get feedback and see which (if any) is the more popular!