Tell us about your experience with Open.Theremin

I hope no one thinks I'm sniping for its own sake here, or using this thread to puff me or my designs up at the expense of others, but I'd very much like the opportunity to comment on RoyP's very interesting experiences above.

That the heterodyning minimum is set to 600Hz makes sense because this keeps the counters away from zero beat and gives a theoretical minimum response time of ~2ms (before averaging).  It isn't too surprising that the analog calibration is a bit tricky, and this the major reason to have the logic in charge of everything if possible as it can get you around all of this.  (One can also digitally dither the oscillators to remove any sticking points in the detection, something I encountered early on in development).  The addition of a knob or buttons to trim the null after the auto process does it's thing would be good (YMMV, it usually needs a bit of touching up immediately after the auto process on my prototype).

The warm-up stability period is likely due to the choice of inductor, which is a Bourns CM453232-102JL.  1mH, with good SRF (2.5MHz) but rather low Q (30) and low current (30mA max, which means the onset of saturation is considerably below this).  I don't know for sure, but I assume the ferrite core is the closed type (no air gap) which means it could be quite temperature sensitive.  Inductor selection is SO critical - outside of hand wound air cores, slug tuned IF type transformers and the Bourns 6300 series (both of which have an air gap), I haven't found anything else remotely suitable after probably months of frustratingly fruitless searching.

Back when I was using the Bourns 8250 series I noticed lots of temperature effects, the worst was null point movement after grasping and letting go of the pitch antenna.  Is anything like this evident on the Open.Theremin?

And I guess I've come to the conclusion that the higher the sensitivity (octaves of range) the more difficult the Theremin is to play.  Is there any adjustment for this on the Open.Theremin?  The ability to alter sensitivity, ideally setting it to whatever the player feels most comfortable with, is one of the huge benefits of the digital (or in this case mixed signal) approach.

dewster: "They are heterodyning before detection, which likely limits low end frequency capture resolution."

After brief exploring of the code:

1. author uses a capture function of timer, so the period of input signal is measured (instead of counting of pulses per fixed interval). This gives the maximum resolution on low end.

2. "Zero point" does not match the null beats from heterodying. It  matches around 600 Hz. So the measurement rate is 600 values per second too. Plus some kind of averaging ("accumulation with forgetting"). It gives an extra resolution. 

More:

3. DAC is 12 bit -- this is enough for a stationary signal, but not enough for wide volume dynamics.

4.  "the theremin is auto calibrated for zero beat" - RoyP.  Realy this is a position of pitch "95 Hz" (maybe I am wrong!). 700 Hz from heterodying  corresponds to 1004 Hz of pitch.

5. Ferrite coils and not-identical oscillators (crystal vs LC) both are mean the significant thermal drift. 

PS. oops! overlapped with dewster' post.

"3. DAC is 12 bit -- this is enough for a stationary signal, but not enough for wide volume dynamics."  - ILYA

At 12 bits * 6dB/bit = 72dB SNR it's almost good enough.  But there is no low pass reconstruction filter (other than the output ~7kHz RC) to deal with aliasing (this isn't a DAC that was designed for audio use).  Stored waveforms could be low pass filtered before storage, but any dynamically generated ones could alias.

I have an ancient Yamaha TX16W sampler - its 12 bit throughout - But I love the sound from it and have never had a problem with its dynamics.. In fact, of all the samplers / wavetable synths etc I have played with, the TX16W sounds the most "analogue" to me..

I do not know how the digital filters or EG's / DCA's in this machine operate, the filters are implemented digitally using ancient 6809 CPUs running, as far as I can remember, at 2MHZ and operating in real time! The main CPU is a 68000 running at 8MHz I think, and they all feed a custom Yamaha IC which somehow manages to produce 16 voices outputting stereo at about 33ksps or mono at about 50ksps (?).

Now it may be that I am completely biased and not hearing bad stuff - I got this sampler oh, many years ago, it was dysfunctional and I spent many weeks with a logic analiser and 68000 and 6809 development kits (all borrowed from work) 'hacking' it untill I found the faults (and, in truth, the faults werent found by the fancy kit, they just gave vague clues - I was way out of my depth!)

But I am inclined now to think that 12 bits can sound great - The pianos have good dynamics, (as do all the sounds to my ears) - I think it probably comes down to some mixed-signal magic in the custom IC... I suspect there is some log conversion going on with the dynamics (DCA) operations and that the processed wave-tables are having their levels altered in a quasi analogue way (as in, processed wavetable converted to analogue and fed as the input to a log MDAC controlled by the dynamics data, or something like that - (But I may just be thinking this because its what I did with an early 8 bit wavetable synth I designed - I fed an analogue EG to a linear MDAC which was converting the wavetable - effectively giving analogue<->digital multiplication rather than digital<->digital computation ).

But I have no evidence for the above, and may well be talking nonsense! - and its well more than 20 years since I had it open, so may be remembering some details wrongly.. All I know for sure is that its 12 bits and sounds great to me.

Fred.

see http://www.thereminworld.com/Forums/T/29168/digital-dynamics for more details on picture below:

Here is a good quickie on DACs:  http://homepage.ntlworld.com/thorin92/mit/modules/electronic_musical_instrument_design_I/05-Digital_to_Analogue_Conversion.pdf

A somewhat meatier paper: http://www.nanophon.com/audio/antialia.pdf

DAC aliasing (the amelioration of which requires an anti-imaging or reconstruction filter) produces images at integer multiples of the sample rate.  Ideal ADC anti-aliasing and DAC reconstruction filtering requirements are the same: 100% transmission in the passband, 0% transmission above 1/2 the sample rate.  For oversampled audio CODECs much of this filtering is done digitally and coupled to a fairly simple analog filter.

12 bits could be enough, particularly for a single tone output, but the lack of DAC filtering is an issue because the unwanted images can easily mix with other signals in the design and end up down into the audio band.

From the Open.Theremin schematic, it appears there is an unused LM324 op-amp - not sure why this wasn't used to make a third or fourth order LPF at the DAC output (there are two third order LPFs post heterodyning) instead of the simple RC.

Think I will change my RS to Real Sensible,

I remember when I was young someone told me here in the States that if I did not Vote I had no right to gripe. I wonder if that transfers to theremins, if one of my students has never posted audio of their playing or a sound byte of their construction do they have the right to publically critique someone else's work. A sound byte is the only way I can evaluate the creditability of my many students from over the years.

The Open.Theremin is open, cost effective and the sound when viewed in Audacity has a nice sine tone and great volume shading. These fundamentals are the foundation of all theremins and it is from here we can improve upon our designs.

All of my active students know how to generate this natural theremin sine wave.mp3 How easy is it to do?

When I began my theremin study years ago, that sound was the "brass ring" as all other theremin sound develops from it.

I had a brief email exchange in the past with the author of open.source, I never asked if he was one of my students?  I would be proud to brag about him. (-'

Christopher

Woow, thanks for all your thoughts and comments and suggestions and tests on the open.theremin. Thanks Jason for opening this thread. Yes, open.theremin is an open source hard and software development and you are welcome to contribute, improve, write ( and document, thanks ILYA :) software, build your own cases, change coils and even download the design files, modify them and distribute it in any way you like. My intention was to start an open platform for people who want to experiment around the stunning theremin instrument and the mysterious heterodyne oscillators that are the basis of many of Léon Theremins inventions. I hope it creates more innovations and already I see many extensions to it in the community like CV out (thanks Roy), new wave tables (thanks Christopher), wave table switching, volume knob, improved stability through temperature compensation (thanks Yasuski), Midi interface, Pure Data interface, laser cut cases, polyphonic theremin... To encourage this I produce the open.theremin.UNO boards in small batches and already shipped almost 100 board to people all around the world. I am having a hard time keeping up with the project so any help is welcome. I plan to present the project at the N/O/D/E Theremin Festival, 24-25 January 2014 in Lausanne, Switzerland where you can test the open.theremin and where we can discuss further development.

Urs

PS: I wish I could play the theremin a bit better :-)

And here is the Open Polyphonic Theremin. First seen on Theremin World:

http://www.youtube.com/watch?v=lHeZW5QNamY&feature=share&list=UUze-aBdbS0B_oJMPvCs85dA

Urs

Urs / gaudi, your poly video is quite interesting!  I wish you great luck with this project! 

Thanks Dewster. I like your suggestion to use the free opamp for better filtering of the DAC as in fact there is still some digital noise on the audio signal as Roy pointed out. Maybe we could collect ideas for improvement here and make a Theremin World version of the open.theremin. (as Jason sugested in an earlier post). The current batch is soon used up and I would be open to it. I saw you are working on an FPGA project that is cool too.

Urs