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

Oh... covid is still there to bother people ... I wish you a quick recovery.

The big C, still a problem. The whole family got it the second time as christmas present last year, my youngest daughter still fights against the follow ups.. 
Hope you will recover fast dewster, 
crossing fingers
Chris

Thanks much for the well wishes Vincent and Chris!  That must be really tough for your daughter Chris, so sorry to hear that.  We know several folks with long Covid, it can be quite disruptive.

I'm mostly back to normal (whatever that is!) but it's taken much longer than anything like a cold or flu, particularly mentally.

Reverb

Very nice post on Hacker News yesterday: https://signalsmith-audio.co.uk/writing/2021/lets-write-a-reverb/

I've read a bunch of reverb papers.  The way they generally go is the author picks / develops a topology, and then (if you are really lucky) magic numbers from a billion man-hours of manual tuning sessions are provided.  Or an algorithmic approach is presented to find the magic numbers, often with disappointing results.  What chance does a noob have if these guys are struggling?  None of the papers say "do this and this and this, and then - BAM! - you've got yourself a high quality reverb".  So that link above is quite refreshing and encouraging.

Diffusion does seem to be the thing, and making it entirely feed-forward with three types of all-pass constructs jumbling things up makes sense (almost reminds me of encryption algorithms).  The following feedback delays are similarly mixed together.

I think there is often too much attention given to minimizing the delay memory and necessary calculations, when we're awash in a sea of RAM and MIPS - most of this is historical I suppose.  It's amazing what one could do with what little they had in the way of computers many decades ago, but frugality can be its own trap.

"For somewhat better ergonomics the plate spiral turns could be rounded squares instead of a circular, perhaps printed windings on a PCB." - Dewster 23/07/2024.

And why not printing the solenoid as well. Probably not mature enough technology but but it sparked my curiosity.

"And why not printing the solenoid as well."  - Mr_Dham

Would love to do that and leave the winding biz behind!  Unfortunately "electromagnet" and "resonant coil" occupy very different design spaces.  One could probably do something like a flat spiral / solenoid hybrid via many PCBs:

It might (or might not) be OK to print spirals on both sides of each PCB, halving the required number.  But then you need to electrically connect them all together somehow.  And the result might not function as well as a solenoid - I can't really predict this with my coil software, which would make for a lot of expensive and laborious trial and error, with perhaps nothing to show for it at the end.  But it is an intriguing construct.

Accordion Solenoid?

For high Q:
1. Adjacent windings (e.g. windings 5 and 6) can be more or less next to each other, though some extra spacing can usually help, but only up to a point.
2. Increasing the winding spacing lowers the magnetic field concentration, so more wire is needed, which directly increases DCR and lowers Q.
3. But lowering the magnetic field concentration lowers eddy currents and skin effect, so there is a tradeoff between ACR and DCR going on here.
4. The less adjacent the windings are, the higher the phase difference, and therefore the voltage difference, and therefore the "Miller" capacitance between them.
5. Due to #4 windings should be spaced proportionally to their adjacency (which also results in the first and last windings being farthest apart).

By the above rules, a single layer solenoid, possibly with two coat magnet wire (or perhaps even better Litz wire) is fairly ideal.  Could the solenoid windings be somehow scrunched together without violating the rules?

Here is a blow-molded plastic squeaker toy:

Now imagine using something like that - though of course more rigid and with more sections - as a coil form.  Windings would still be spaced according to their adjacency, but the overall thing would be shorter for a given mH.  Winding it would be a nightmare though, particularly on the down sides of the form, you would need adhesive or something.

Underwater Basket Weaving

Because the AM broadcast band lies just above the average Theremin operating points, there is a lot for Theremin designers to learn from hams and other radio hobbyists.  Basket / spider web coils were fashionable before the introduction of ferrite core materials, particularly in crystal radio sets, where high Q is needed for good received signal strength and selectivity.  

It looks like there are two basic forms of basket coils: solenoid and spiral.  When more wire is added the solenoid form becomes longer; conversely more wire causes the spiral form to become larger in diameter.  Here is someone doing rather crude basic experiments, comparing solenoids to solenoid baskets: http://www.vk2zay.net/article/188.  It isn't surprising to me that they don't test terribly differently because the overall dimensions are similar, and adjacent windings can be pretty close without causing too much in the way of Q degradation.  Someone in the comments recommends DIY coil dope via Styrofoam dissolved in lacquer thinner, which I found interesting.  Someone else says to microwave the coil form, and if it heats up don't use it.  Another comment says coils should be in metal boxes to reduce radiative losses?

Here is someone winding a spiral basket inductor on a coil former made from a discarded CD: https://crystalradio.net/spiderweb/index.shtml.  No mention of the final mH value nor Q, which is a rather glaring omission.  And, like multi-donut scramble wound RF chokes, no mention anywhere of how to best design these things.  Is it all just seat of the pants / trial and error?

The coils used in AM radios are smaller in value than Theremin series inductors due to higher capacitive padding in the radio RF tanks, which is generally on the order of hundreds of pF.  Whereas Theremin series coils are loaded with 10pF to 20pF of the antenna intrinsic, so they operate closer to free air resonance.

[EDIT] One thing not really stressed in these articles / experiments is the necessity of physical stability of the windings for long-term stability of the inductance.  To me this is where the single layer solenoid really has the advantage, because tight windings against a substantial cylindrical former can provide this.

Unfortunately "electromagnet" and "resonant coil occupy very different design spaces."- Dewster
Yes, I see... (my only experience of weaving wire was arround an industrial homopolar toroid then I am more from the electromagnet world)

It might (or might not) be OK to print spirals on both sides of each PCB, halving the required number. - Dewster
In fact my initial research was about seeing if a cylindrical PCB technology would exist... It seems not, unfortunatelly.

Underwater Basket Weaving...

What strikes at first glance is that these topologies of coiling, basket weaving, etc. have been around since prehistoric times when it comes to everyday objects. So, we can imagine that people know manual or industrial technics to achieve them.

Main difference is that you have some current in the wire. I understand that you don't want to trade too much Q for ease of construction. So, at the end, you will always need to come back to electronic consideration.

But did you ever consider closer looking into weaving world (first without electronic consideration) to search some ideas about how to achieve a given coil topology ?

"my only experience of weaving wire was arround an industrial homopolar toroid then I am more from the electromagnet world"  - Mr_Dham

This is the first time I've heard of a homopolar toroid.  Was that for a current sensor (e.g. Rogowski coil)?  This patent is interesting:

https://patents.google.com/patent/US6414475

"In fact my initial research was about seeing if a cylindrical PCB technology would exist... It seems not, unfortunately."

Lines on a flex PCB might work, but they would all need connecting, and the conductor geometry isn't ideal (you would probably want taller than wider), nor is the physical integrity.

"But did you ever consider closer looking into weaving world (first without electronic consideration) to search some ideas about how to achieve a given coil topology?"

Good suggestion!  In fact a woven basket that gets larger towards the opening is a blend of solenoid and spiral windings.  It would seem to me that the need for physical stability of the windings pretty much precludes any sort of weaving in the air without supports.  And the need for large values (10 to 20mH) pretty much precludes anything with a lot of air in it.