"Paradox MX" - heterodyning theremin with built-in pitch display

Glad to present "Paradox MX" , a second generation of "Paradox" theremin
(and maybe a second generation of "Skywave H1", in some kind ).

Features:
- CY8C27443-24 and TDA7056A ICs,
- 5V power input,
- 3W speaker output,
- visual and audio pitch preview,
- dual purpose pitch display (for easy diagnostic/adjustment),
- miniature air core coils (minimum of ferrite is used for tuning),
- EQ coil free topology,
- 12 voices and 4 selectable pitch ranges,
- super decoupled oscillators (beat freq is down to 1 Hz and lower).

This is not an "ideal theremin", but
just an attempt to implement what Fred considered impossible to do (hi quality analog processing inside PSoC).
Unfortunatelly some kind of noise is still audible, especially at bass notes.

Schematics:

Internal configuration of PSoC for heterodyning and audio processing:

Full description and audio samples can be find here  (Russian)
and here  (Google translating).

Presets with formant filters are great!

I enjoyed this build. A little bit of different worlds and it works! Not daily posting of non-sense year after year.

When I use to hire people I prefered the quiet type. Those that talked too much could never close the deal.

Christopher

Another fantastic build and fantastic article ILYA!  Thank you for sharing it with us and the world!  If only I could have read your articles when I was first starting out.  Even in Google translate it is quite clear, it must be even clearer in Russian.  And it's a cover story!  Who is the woman playing it?

It was very nice of you to dedicate it to FredM and his Theremin efforts with the PSoC, he was so free with his knowledge and time, and was a really great guy.  And the Peter Pringle quote at the end is a nice touch!

The inside looks really neat and spare, and this seems to be the case often when concentrating functionality in one big chip.  I remember looking at the PSoC when they first came out (I think a Cypress rep pointed me to them) but couldn't think of a use for them at the time.  You've managed to stuff axis processing, tuner, pitch preview, formants, and even presets in there, which is pretty amazing!  (Presets open up a whole new world, and become more and more mandatory as the audio synthesis path grows beyond simple waveform shaping and filtering.)

As you state in the article, FredM was using the PSoC mainly to detect and suppress the pitch region beyond null (dealing with this when everything is digitized is super trivial: zero out negative differences), and it's rather surprising he didn't use it for other ancillary functionality.

It's interesting (but perhaps not surprising in retrospect) that supply coupled internal interference is stronger than electrical activity happening in the volume and pitch fields, and I believe this has been my experience too.  Which is a good thing, because one can deal with it easier than "hiding" things from the fields via shielding.

Strange how the editors injected their own views into your article, that must be rather frustrating.  Particularly with the topic of coil self capacitance!  I'm not sure I would be "correcting" an author who builds Theremins, who has thoroughly researched and quantified antenna-hand C as you have, and who almost certainly has a more intuitive grasp of these things.

I may have missed it in the article, but how exactly are you doing the heterodyning?  Is it XOR or D-flop or something else?

Thank you also for providing the MP3 sound clips!  You must get a lot of correspondence...

so great!

"but how exactly are you doing the heterodyning?" - dewster

The PSoC User Module has an input sampling circuit and two nonoverlapped phases Ф1 and Ф2 (derived from clock freq) to it operation:

"Who is the woman playing it" - dewster

Just a person from my regular job who approximately looks like Clara.  The magazine's editor called her  a "thereminist"  (in caption).
Her son, who engaged in music, jokes:
"Mom, there are only four famous thereminists in the world.  Now you are the fifth!"

"The PSoC User Module has an input sampling circuit and two nonoverlapped phases Ф1 and Ф2 (derived from clock freq) to it operation:"  - ILYA

Wow, I'm not very familiar with switched capacitor topology but that's quite clever!  There are switched capacitor mixers for RF, something to look into, thanks!  It seems that there are all sorts of new (>1920) technologies that might be applied to the Theremin.  The comparison to its super advanced cousin, the metal detector, is like night and day.

"Just a person from my regular job who approximately looks like Clara."

She does look remarkably like Clara in that photo!  I had to glance down and see that she was playing your Paradox MX to know it couldn't be her!

Thank you again ILYA for all the effort you must put into conveying that information so clearly and practically.  I particularly appreciate your precision C measurements and encapsulation of them in an interactive PC program, as well as your pushback to my overly simplistic 1/d linearization scheme.  Technical workers learn to have a keen interest in when they are wrong, as it is a time waster, and one may spew embarrassing idiocies until corrected (I speak for myself here) - this unfortunately runs counter to human nature, so it is a difficult lesson.

[EDIT] So what are your thoughts on having a tuner to look at when you're playing?  Does it have obvious / significant delay?  And are there any resources left in the PSoC to do PWM on the tuner LEDs?  (You might use simple counter roll-over to do PWM.)  Do you find the volume LEDs to be useful?  I barely pay any attention to the octave display (though if I were to play on stage I might more) so I'm wondering what you think of that as well.

Thank you for this superb technical solution and insight in the development. The description seems a scientific based work with high quality. The Clara double is very impressive! 

In one of the mp3s (demo03.mp3) a slightly flicker and noise problem is visible in a waterfall diagramm. Beacons near zero, below 300 Hz,  range by 3000 Hz. Could it be due to a LED flickering? 

JPascal -- no, LED scanning cycle has a rate about 73.2 Hz. Five of time intervals ("slots") are used here so the maximum frequency of switching  is 366 Hz.
Your question needs in additional researching.

"what are your thoughts on having a tuner to look at when you're playing?  Does it have obvious / significant delay?  And are there any resources left in the PSoC to do PWM on the tuner LEDs?" - dewster

My thought isn't original and coincides with opinions of professional thereminists:  constantly looking at the tuner is not only useless, but also harmful for playing in tune. So the pitch display is just a cheap show-off.

Nevereless it is still useful for:
1. Adjusting after production.
2. Soundless tuning before performance.
3. First note search.
4. Finger position trainings.
5. All kinds of checks (linearity, thermal drift an so on).

6.(Maybe) ear training.

The latency due to given measurement time (0.1 s) is not felt (in my feelings). If desired, it can be reduced (because of overresolution 0.04 Hz obtained).

In early design (using other microcontrollers) I tried PWM, but found  easy algorithm "hit/missed" is better (where "hit" event = "target interval of +-25 cents"). At given imlementation the PWM is tricky.

(Сlarification. The current algorithm provides:
two adjacent LEDs are on -- miss;
just one LED is on -- hit into +-25 cent interval.)

Thanks ILYA!  I replied over on my D-Lev Thread [LINK] as I was using it to highlight your various excellent points for comparison purposes.

I found your comments in the article regarding absence of drift tracking (LC & LC - vs. - LC & XTAL) very interesting (as this is the case for the D-Lev as well, and something to perhaps consider in the design process).  It's been my unscientific observation that, when the L is a stable air-core, and the oscillator supply voltage is well regulated, then drift is likely caused more by air temperature effects (relative humidity, etc.) surrounding the antenna than by thermal changes in the oscillator.  The "matched oscillator drift cancellation" thing is probably something you would turn to more when significant ferrite is present and / or voltage regulation is problematic (tube designs?).