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

We are convinced that in the "normal" Theremin playing, 100 Hz is enough. The "quick"  and "abrupt" changes are not part of a normal theremin performance. Moreover, a little smoothing of that "abrupt" changes, could produce even more musical results. 

--------------- 

Inductor temperature dependance is not part of the proposed simulations.

Our inductors tests where done with two CapSensors side by side, in the same ambient.
We then logged the data, cyclically altering temperature with a hair dryer.
We then maintained the best inductor, changing the worst with a different kind.
This process was repeated several times (for more than six months and with at least 30 different inductors) to select the best one (at least for our Colpitts oscillators)

It is not only a direct temperature dependance problem. When temperature changes, large inductors tend to accumulate mechanical stress and to release it periodically, causing glitches. This doesn't happen on little inductors. 

"The "quick"  and "abrupt" changes are not part of a normal theremin performance. Moreover, a little smoothing of that "abrupt" changes, could produce even more musical results."  - livio

I think if the smoothing is at all obvious to the player, then the playing experience deviates from that of a "real" Theremin.  Whether this is desirable from a musical standpoint is I suppose in the eye of the beholder.

"Inductor temperature dependance is not part of the proposed simulations."

To post simulations here comparing your circuit to the Etherwave, you really should include inductor temperature dependence, as well as drift tracking of the heterodyning oscillator.

Frequency Sweeping

When designing / building / working with digital Theremins, the pitch may be quantized or "sticky" which can obviously be troublesome because pitch change on a traditional analog Theremin is generally quite smooth.

One way to check this is with the spectral frequency view in Adobe Audition 3.0:

The above is a portion of the audio track from my YouTube digital Theremin video that I posted over a year (!) ago:

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

At the end of the video (around 0:40) my hand is quite far from the pitch antenna.  I sweep it in to the antenna and pull it away.  This is the broad "lump" on the left above.  I then do the sweep again, but much more quickly, and this is the second narrow "lump" on the right above.  The pitch change is quite smooth, with no obvious sticky points or flat spots.

Anyway, if anyone suspects their digital Theremin is acting up in the pitch department but can't quite put their finger on what is going on, this type of analysis can be quite revealing.

You are right on everything.

Probably a digital Theremin responds differently from an analog one.

I also apologize for having misunderstood your objections to my simulations, which do not take into account the temperature dependence of the inductors. I correct my post on the comparison, already. For the other issues, it will take more time, but I will try to reply to everything. Please be patient with me, English is not my language.

Your graph made with "Audition" is very interesting and really helpful.

As soon as possible we will add this type of visualization to our software "Digital Audio Analyzer" and "Audio Examples":
www.theremino.com/en/downloads/uncategorized/#daa
www.theremino.com/en/downloads/multimedia/#audioexamples

"Your graph made with "Audition" is very interesting and really helpful."  - livio

Audition is quite nice for recording and analysis.  It started out as shareware (CoolEdit), Adobe bought it and improved it (and of course put a huge price sticker on it) - then they gutted it like a fish.  Luckily, Adobe released the best version into the wild when they retired a license server:

http://www.techspot.com/downloads/5733-adobe-audition.html

The MD5 of the file located there matches the MD5 of the file I downloaded directly from Adobe, so I'm pretty sure it is clean.

The spectral frequency, spectral phase, and spectral pan views are quite amazing and quite revealing.  You seem to be doing similar things in your nice looking software, perhaps it may inspire you further or give you more ideas.  I've also used Audition to generate warble tones for audio work, I've done some audio restoration with the filters, some audio mixdown, and I use the spectral views extensively for my digital piano reviews over at Piano World.

Thank you for editing your post!  Inductors are exceedingly problematic in so many ways.

Someone like you who has designed EKGs and similar "body electric" medical equipment could add significantly to Theremin development.  Also, have you taken a look at the Open.Theremin project?  The frequency (period) measurement is performed in a very interesting manner that increases resolution dramatically.

Many thanks,
I have downloaded Audition and tomorrow I will explore it. 

And also thanks about the Open Theremin,
can you help me to locate schematics and technical informations?

"And also thanks about the Open Theremin, can you help me to locate schematics and technical informations?"  - livio

http://www.gaudi.ch/OpenTheremin/

You want the .UNO page.  Schematics, BOM, code, etc. are there.

FredM and I have been discussing some of its features in this thread lately. 

Urs Gaudenz showed up in a thread Jason started here at TW a little while ago. 

I found the schematics, thank you.

The Open Theremin RC is not digital, we shouldn't take it in consideration.

Regarding the Open Theremin One, I will tell you better tomorrow, after doing some simulations, but it is certainly a rather complicated and slow way to digitalize...

"The Open Theremin RC is not digital, we shouldn't take it in consideration."  - livio

The RC is an old design I believe.

"Regarding the Open Theremin One, I will tell you better tomorrow, after doing some simulations, but it is certainly a rather complicated and slow way to digitalize..."

FredM and I have been discussing this in this thread lately.  If you go back a page you will see some of my Excel simulations regarding linearity and period measurement rather than frequency measurement, and digital heterodyning filters.

As you pointed out earlier, a certain amount of digital resolution is necessary for musical purposes.  How this resolution is distributed is also important.  Heterodyning (not to zero but to maybe 1kHz or higher minimum beat frequency) and then measuring the period gives better resolution in the far field where the information is getting sparse.  This is what the Open.Theremin.Uno seems to be doing.  This method also allows a fairly slow processor to be used.

Block Diagram - 2014-02-09

So this is the direction I'm currently heading in, showing the hardware / software division:

Everything has either been implemented in the past, or has been simulated to one degree or another.  Though coding it up and sticking it all together will be new.  Since the operating point won't be coming from the DPLL loop filter, I have a simpler construct for that waiting in the wings.

"offset" into the period measurement offsets and negates the effect of the period, so higher frequencies (smaller periods) give larger numbers.