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

Is it a theremin? No, it is "Ave Maria"...

Maybe we still have to move from looking at our instrument in terms of performance to looking at our instrument in terms of musicality, pleasure to play or to listen...

"Yes, you should take credit! Both is ment."  - JPascal

As I believe Vincent is pointing out, the song itself should get some credit too!

"Maybe we still have to move from looking at our instrument in terms of performance to looking at our instrument in terms of musicality, pleasure to play or to listen..."   -  Mr_Dham

But I'm not playing jazz? 

"The singer holding vowel A is certainly surprising for an audience. But no artist would do so. I wonder if a timbre not clearly a voice or wind instrument is more credibel for the faszination of theremin. But this is only my thought, without any relevance. We all are digger in the wide field of timbres."  - JPascal

Yes I think about this a lot, is there an ideal timbre for the Theremin?  (Or for any synth for that matter?)  Should it closely mimic some acoustic source?  Or is it better to just suggest one?  Or best to be nothing at all like one?  After the experience of trying to model Lydia's tVox, I certainly have a new appreciation for the subtle nuances in the generation of that sound. 

The timbres I like to hear most coming out of Theremins are female vocal-ish and string-ish, which some analog Theremins can do sorta OK with their generally very limited tone shaping circuitry.  Is this a chicken / egg thing, that is, are they setting up an expectation and then fulfilling it, or would I prefer it nonetheless?  I believe it's the latter.

I very much enjoy playing human vocals on my Theremin, they're a good fit with the continuous pitch & volume controls, and both are kinda spooky.  Strings are fun too.  Plainer sounding timbres aren't nearly as interesting to me so I never play them, but they're sometimes fun to have in the preset bank for showing off and goofing around.

Read an interesting article on synthesis off Hacker News the other day:

https://www.syntorial.com/tutorials/ten-steps-to-becoming-a-synth-dynamo/

I concur with his advice to pick a subtractive synth to learn the basics.  It's too bad in a way that the canonical analog synth has jelled to essentially mean the miniMoog.  The wave choices of sine / triangle / ramp / square / pulse form a rather static & boring excitation source, followed by insufficient filters to pep it up.  The D-Lev synth uses very simple FM to provide a modulated all / odd harmonic source mix; the noise source can be AM modulated via the instantaneous oscillator phase; and these are followed by enough filtering to place it in the modal synthesis category, which I believe is where you want to be.  A small formant bank in series / parallel with a delay based resonator is quite flexible and manageable, and provides amazing pseudo stereo.

Not sure I catch your reference to jazz.

My comment was rather a reaction to

"I think a fair amount of the amazement people have for certain Thereminists is due to the exactness of their pitch during performance, and not their playing style so much" - Dewster

I am happy because I can recognise the song (Ave Maria here) and performance (this being played with a theremin which is a difficult instrument to play) becomes secundary.

The same for the timbre: the real question is not about realism but does it allow you to play what you want to play. (And potentially have a lot of fun). That said, I am very happy with my D-LEV in this aspect.

"Not sure I catch your reference to jazz."  - Mr_Dham

I was sorta kidding that jazz is more performative, and less musical / fun to listen to.

"I am happy because I can recognise the song (Ave Maria here) and performance (this being played with a theremin which is a difficult instrument to play) becomes secundary."

Ah, I see.  Sorry, I didn't mean to put words in your mouth.

"The same for the timbre: the real question is not about realism but does it allow you to play what you want to play. (And potentially have a lot of fun). That said, I am very happy with my D-LEV in this aspect."

Music to my ears! I haven't received much feedback, though I figure folks are probably busy trying to come up with physical configurations they can play, a downside to not providing a standard case with the guts.

"After the experience of trying to model Lydia's tVox, I certainly have a new appreciation for the subtle nuances in the generation of that sound." dewster

Because of tvox is an analog theremin my interest is instantly waked up. What I suppose in time signal is the typical stretched rectangular wave in the bass region followed by a band pass filter? But can this allone lead to that voice like frequency spectrum?

Plate Antenna Geometry

Theremin players are in a unique position to develop a very intuitive sense of capacitance, and can deepen that by experimenting with various plate geometries.  Augmenting this with a bit of finite element analysis can reveal the interaction between intrinsic and mutual C, which isn't otherwise easily detectable, but isn't super necessary either as the oscillator responds to the sum of the two (which is dynamic - surprisingly both change with hand position).

All series coil type oscillators (both analog and digital) require a minimum amount of capacitance to operate in a stable region, and intrinsic capacitance is directly proportional to surface area.  So why not make that surface area work for you by having it maximally interact with your hand too?

If one were designing a generic capacitive distance measuring device, one would likely start with two round or square plates, so it seems prudent to more or less start there when designing a Theremin (say, from total scratch) by making the antenna geometrically somewhat similar to the hand.  I think square plates (or at least the area presented to the hand in the case of U plates like I use) make the most sense because your hand can be off vertically and it doesn't impact horizontal targeting (the plate width is a constant), and vice versa (the plate height is also a constant).  One possible improvement here might be a flat pincushion distortion type plate, where the corners curve outward farther (b in the image)?  I dunno as I haven't experimented with this (I've been pretty happy with a simple square face).

After playing plates for a couple of years, it seems to me that horizontal and vertical targeting are roughly equally easy / difficult, hence the square geometry on my lab unit, though this is something to perhaps experiment with and fine tune a bit in one's own custom instrument, and the kit coincidentally comes with one square and one rectangular plate.  The size of the plate could be increased somewhat to make for an easier target, but then you start running into the plate "seeing" and interacting more with your entire body and the room, and the reduced (proportionally vs. intrinsic) mutual C in the near field if the hand is significantly smaller than the plate (neither of these issues are anywhere near the catastrophic category, though really big plates will significantly lower the Q and will require electrical adjustment in the AFE C-divider).

I don't believe it would be helpful to curve the plate edges towards the player (making them 3D) as the optimal curvature would depend on the non-constant arm / hand hinge point in space, the curved in edges might cause further targeting trouble, and any benefit curvature might provide would likely be firmly in the diminishing returns area, but I can't say that I've experimented with this either.

Another experiment I haven't tried is the use of a conductive mesh as the antenna.  This would be super easy to test with the kit by simply clipping a similarly dimensioned metal screen to the coil instead of the plate, and examining the operating frequency with a nearby counter or by querying the D-Lev Hive processor pitch register.  I'll try to do this soon and report the results.

At one point Roger was contemplating the use of a plastic conductive mirror as the pitch antenna.  He inquired about the difficulty of reversing the tuner display so he could observe it in the mirror while playing, but he ended up abandoning plates altogether before that idea went any further IIRC.

I say all of the above not to discourage anyone nor to derail their creative trains of thought, just relating some of my observations to-date.

[EDIT] Wanted to add that it's been my personal experience that various antenna shapes are more or less very similar feeling "blobs" out in space, and the feeling only really differentiates in the near field.  That is, the mid and far field "feel" of an antenna is much less dependent on geometry than I initially imagined it would be - and I'm not presenting this observation as an argument for nor against anything in particular.

Updated Librarian

With help file & video:

https://d-lev.com/support/d-lib_2022-01-24.zip
https://d-lev.com/support/D-Lib_Help_2022-01-24.pdf
https://www.youtube.com/watch?v=0AAHfw5ZG5A

Users can now take a snapshot of the presets and software (the entire state) and store it in a single *.eeprom file, and restore it later.  This is a feature I should have implemented long ago, simple and lots of general utility, even for me with the kits and stuff.

There's also a text file viewer, which is just an extension of the help file viewer, though I added a line numbering option.

I swear I've spent more time on the simulator and the librarian code than I have the actual D-Lev assembly!  Support tooling in a complex environment like a modern OS (or three!) is super time consuming.

Ground loop

I decided for the use of SPDIF for my Bolla-plate enclosure.

This lead me to have several ground loop situation if I use my computer as power supply for the audio converter, the D-LEV and my audio interface. 
I can resolve them by using specific power supply when applicable and some commonsense.

The most strange thing is that I have disturbance of the touch pad of my computer as soon as SPDIF is connected and my computer is too much close from the D-LEV (the mouse pointer jumps from one location to another on the screen).

I already ordered a Toslink panel feedthrough and an optical cable to resolve it but I was just wondering if you had an interpretation of this phenomenon.

"This lead me to have several ground loop situation if I use my computer as power supply for the audio converter, the D-LEV and my audio interface. 
I can resolve them by using specific power supply when applicable and some commonsense."  Mr_Dham

What are the ground loop symptoms exactly?  Audio hum, or pitch field stability?

Roger's P3 uses both SPDIF outputs, which go to two unboxed DAC boards inside.  Perhaps it avoids ground loops by having the SPDIF wiring kept quite short?  Connecting a secondary ground to my lab unit sometimes makes any interference slightly better, but usually it doesn't make any difference.  I've seen ground loops while hunting for interference with an analog Theremin oscillator in the upstairs of our house.

Using TOSLINK neatly sidesteps any ground loop issues.  I suppose that's one reason why properly terminated MIDI is optical.

"The most strange thing is that I have disturbance of the touch pad of my computer as soon as SPDIF is connected and my computer is too much close from the D-LEV (the mouse pointer jumps from one location to another on the screen)."

I haven't seen any interference with my laptop touchpad, but this sort of thing isn't too surprising, as the pad uses a grid of very low voltages to sense fingers, and the D-Lev fields are a couple hundred volts peak-to-peak.  Someone recently told me their cell phone touchscreen wouldn't work when near their Claravox and that it was quite repeatable, but then later couldn't get it to do it again.  Probably the Cvox was operating near some multiple or sub multiple of the touchscreen frequency, and Theremin frequency is quite variable.

Theremins in general are a problem looking for a problem.  It's amazing they work as well as they do most of the time.

Edited multiple time to get page layout accepted in the published version.

"What are the ground loop symptoms exactly?  Audio hum, or pitch field stability?" Dewster

My ground loop topology is the following one:

Edited: D-Lev -USB- PC -USB- PC's Sound board (Steinberg UR22)  -Audio Cable-   Sound amplifier -Audio Cable-  AD Converter  -SPDIF- D-Lev

It results in a very audible hum, field looks stable.
Anything that breaks this loop topology will resolve the problem. I need this topology if I use D-Lib and DAW at the same time (rare but probable use case)
 

Using TOSLINK neatly sidesteps any ground loop issues.  I suppose that's one reason why properly terminated MIDI is optical.
Yes, exactly. Hence toslink re-implementation in my enclosure.

I haven't seen any interference with my laptop touchpad, but this sort of thing isn't too surprising, as the pad uses a grid of very low voltages to sense fingers, and the D-Lev fields are a couple hundred volts peak-to-peak.  Someone recently told me their cell phone touchscreen wouldn't work when near their Claravox and that it was quite repeatable, but then later couldn't get it to do it again.  Probably the Cvox was operating near some multiple or sub multiple of the touchscreen frequency, and Theremin frequency is quite variable. [...]
Effectivelly, it is probably the same that I am experiencing and it is intermittent as well.

Theremins in general are a problem looking for a problem.  It's amazing they work as well as they do most of the time.
D-LEV is the king of my electric jungle. Other device will have to adapt !