Etherwave: grounding, power, output and other considerations

Greetings to all the brilliant people of this forum. I've been reading through it for a while and have arrived at some questions, which I feel I am ready to ask.

Some background first. A little while ago I set out to build a Theremin. After somewhat extensive research and some minor experiments I arrived at the understanding that a handmade Etherwave would be optimal for me in terms of its capabilities, expenses and personal involvement. So what I have now is a EW schematic from respective Hotrodding manual and couple Thierry's posts dating back to 2009 which mention improvements made by Moog on the part of linearization coils (3 coils in H-formation instead of 4 coils in a zig-zag pattern, reducing their mutual induction), and the Hotrodding manual predates these improvements, so they are not included in the schematic.

What I want to know now:

1. Is there any more up-to-date EW schematic available than what's in 2003's Hotrodding Manual?

2. I am considering a different power supply scheme. Everything to the left of C21/C22 is replaced with a ±12VDC buck-boost converter (like this one), which is in turn fed +18VDC from two possible sources: either a 220VAC→18VDC wall adapter or two 9V batteries (and all that means I install a three-position power switch instead of a simpler one to be able to choose the source).
This leads to a series of questions:
2A. The buck-boost converter includes some inductors. They're not big, they're toroidal (which, as I understand, makes their magnetic field more or less self-contained), they work at 180 kHz which is far beyond audible frequency and far below operating frequencies of oscillators. How big of an issue this might be? Should I attempt to remove this converter as far from the oscillator and antenna circuits as possible? A separate, shielded enclosure on the outer side of the cabinet, maybe?
2B. The bigger the C21/C22 — the smoother the voltage supplied to the inner circuit, right? How about 1 uF instead of 0.1? Or the bigger the caps, the worse their electrical field interference, and 0.1 uF is just enough?
2C. The batteries themselves are wrapped in big rectangles of thin metal. Capacitance issues? Should I remove them to an outer enclosure, replace 9Vs with lithium accumulator or ditch the standalone power idea altogether?

Also, neither batteries nor AC→DC adapter have any grounding (one could possibly argue about some grounding through adapter's own capacitance, but that doesn't cover batteries anyway). Moreso, here in Russia we generally don't have ground in our power outlets at all anyway. So, that raises another question:
3. How about I craft a solid metal stand for the cabinet with a large metal plate on the bottom end and a screw connection with cabinet's bottom, and solder a wire between circuit's ground and the cabinet's mounting nut? Would that be enough of a path to ground? Well, I can even make it large enough to stand on it with one foot while playing, if the conductance of a wooden floor and whatnot is too low for the job.

4. Through my initial research I have several times come across mentions of wrapping the antennas (made out of copper tube) with some heat shrink: it was said to prevent unnecessary contact of player with the antenna and protect it against environments of falling, scratching and dirt. Any associated capacitance issues, any other counterarguments?

5. Those 47 uH variable inductors. How crucial is their build quality? If I wind my own some thirty loops of a thick enough copper wire on a threaded piece of tube and add a threaded iron bar which screws in and out of it, won't I ruin everything, or is there a good chance they end up being made with a low self-resonant frequency and therefore unusable?
Actually, this very question extends to antenna linearizations coils as well. Can coils of required inductance (up to 20 mH) be made at home with acceptable Q and a high enough self-resonant freq? If so, what nuances will require most attention?
5A. As I understand, ferrite-core coils are unacceptably sensitive to external fields, and so for antenna linearization only air-core ones should be used. Is that so?

6. Back in the days of EM Theremin guide when PCB printing wasn't publicly available, the guide impied assembling everything on simple prototyping boards with occasional wires here and there connecting distant circuit elements. Is this still an option, or does the PCB approach have some untradeable benefits?

7. Being new to analog electronics, I feel like I largely miss the concept of a ground plane. Is that an electrical abstraction of a common point of a circuit being connected to ground, or is it a real sheet of metal on the bottom of the cabinet, to which all common point connections are made as geometrically (and ohmically) short as possible?

8. And another thing where I run out of understanding and get really dumb. EW's output is said to be line level (circa 1 V RMS × single-digits mA). I remember seeing something about driving some low-impedance earphones from the same output jack, which would require installing a potentiometer somewhere around the LM13600 — I guess it goes in place of R38, and spans like 0—50 kΩ, right? Line out for further amplification at ~50k, larger current for earphones at zero?
8A. Playing with earphones on I'd better clip the cable to my clothes to keep it as far away from the antenna as possible, right?

I thank in advance everyone who comes to this thread and shares even a tiny bit of knowledge.

Hi nazro,

I'll try to answer some of your questions.

2. I would use linear regulators for the oscillators, you don't want high frequency ripple there.  I've seen designs were the 50/60Hz line transformer is inside the Theremin cabinet, not sure how bad that is in terms of interference as I don't want to even try it.

2c. Batteries in the enclosure should be fine.

3. You need a good ground.  I'm unfamiliar with Russian AC outlets, but is one of the conductors grounded back at the fuse /  breaker panel?  If so, maybe try to use that somehow (if the outlet is phased).  You can ground through a capacitance (perhaps as small as 0.01uF) to keep the possibility of shock to a minimum.

4. Other than looks, IMO there is no down-side to insulating the antennas, and would highly recommend doing so for any Theremin.

5. You can make all of the inductors if you wish, but you have to know what you're doing.  The variable tank inductors are probably the easiest, though I would use ferrite rather than iron for the moveable core.  Making the EQ inductors is more complex, as they must have a large inductance with very low self capacitance.  You can make a really large air-core (ala Theremin) or one or more ferrite core in series (ala Moog).  Using winding "donuts" on a plastic or phenolic former might work the best?  The inductors are generally quite critical to an EQ Theremin design such as the EW.

5a: If anything, ferrite core inductors are less susceptible to external fields.  Ferrite is used mainly for general miniaturization purposes, not performance.  RF designers don't want huge honkering coils in their radios and such, Theremin designers tend to not mind them or even actively desire larger inductors for low self capacitance reasons.

6. Pad-per-hole prototyping boards should work fine, though I would get the glass epoxy plated through type and avoid phenolic like the plague as the traces easily delaminate when soldering.  Plastic breadboards have a lot of capacitance between rows, but even these can be used for experimentation if you skip every other row, don't use the type with an aluminum backplane, and stick the the whole thing on top of a 2" or higher plastic box or spacers.

7. You generally don't want an actual ground plane inside a Theremin.  Grounding everything to one point I suppose can help subsonic oscillator interaction, but you don't have to go too crazy here.

8. Experiment with this.  I would try using an op-amp as the headphone driver.

8A. I play with headphones all the time and have never really had a problem with the cord interfering.  The cord usually comes out of the left earpiece, so it's far away from the pitch antenna.  And for my setup, the cord runs away over to my PC, so that's not an issue either.

I think a lot of your questions will be answered via experimentation and experience.  Do get a decent oscilloscope, otherwise you'll be flying blind.

One very simple experiment every budding Theremin designer should try is driving one end of a coil with a function generator, with the other end attached to an antenna or plate.  Observe the drive signal via one channel of the scope, observe the antenna signal via the second channel of the scope, with a series 1pF cap to somewhat isolate the measurement.  You can even just place the second channel scope probe near the antenna (perhaps with a couple of inches of wire attached to the probe end) and it should work.  Ground the probes and the generator, adjust the generator until you get resonance.  With this setup you can easily see phase shift at resonance, measure coil Q, and calculate the intrinsic capacitance of the antenna + self capacitance of the coil.  You may also immediately discover that many types of coils do not perform well in this scenario, and thus will not function well as EQ inductors in a Theremin.

I use the instrument cable as the ground, laying a generous portion on the floor around the theremin and if I feel extremely particular I will place a foot without the shoe on the cable.  You can also stick portion of the cable looped into the top of one's sock. I never worried about a three pronged ground plug for the power supply of the Etherwave. My current theremin is battery powered so there is no ground via an electrical outlet. No problem. The music instrument  cable on the floor or ground does the job.

@dewster

I'll try to answer some of your questions.

Thank you very much, Eric! That pretty much covers most of what I need for now.
Yet new questions emerge, if you don't mind:

2. I would use linear regulators for the oscillators, you don't want high frequency ripple there.

And that implies negative influence of said ripple through halvanic coupling (which can be mitigated with regulators) or through electromagnetic field coupling with nearby elements (which can not)?

3. You need a good ground. I'm unfamiliar with Russian AC outlets, but is one of the conductors grounded back at the fuse / breaker panel?

True, that might well be the case. I guess I can check the voltage of either of the holes against a central heating radiator, and then if one hole shows 220 V and another one shows zero — that's my grounded neutral, right? Shame I hadn't thought of this earlier.

5. You can make all of the inductors if you wish, but you have to know what you're doing. Making the EQ inductors is more complex, as they must have a large inductance with very low self capacitance.

If I only knew what I am doing But theoretical basis is clearly not enough for confidence in decisions.
The way I understand self capacitance of an inductor, it can be lowered by lowering wire surface area, which in turn calls for thinner wire and shorter wire. Shorter wire = less turns of winding = higher permeability core — and there go air cores, now we're talking at least iron; thinner wire = higher resistance = lower Q — and where should I stop with that? My gut tells me it'll be something in range of 25-30 AWG, how true is that? (I tried to filter some coils on Mouser for given L, f and Q, and look at them to see how it's done in the real world, but they have neither data on wire gauge nor even photos.)
And how exactly does low Q affect the performance of a Theremin?

Anyway, thank you once again for such an informative response. Take care!

@rupertchappelle

That's a pleasant thing to hear, I hope this is indeed a smaller problem than I imagine it now  Thank you too!

"And that implies negative influence of said ripple through halvanic coupling (which can be mitigated with regulators) or through electromagnetic field coupling with nearby elements (which can not)?"  - nazrow

Sorry, I was talking about voltage ripple on the oscillator supply.  Though constant average voltage is probably most important.  You could try using the switching supply directly to see if it is a problem.  Regardless, I would decouple the oscillators via separate RC filtering for each oscillator.

"If I only knew what I am doing But theoretical basis is clearly not enough for confidence in decisions.

The way I understand self capacitance of an inductor, it can be lowered by lowering wire surface area, which in turn calls for thinner wire and shorter wire. Shorter wire = less turns of winding = higher permeability core — and there go air cores, now we're talking at least iron; thinner wire = higher resistance = lower Q — and where should I stop with that? My gut tells me it'll be something in range of 25-30 AWG, how true is that? (I tried to filter some coils on Mouser for given L, f and Q, and look at them to see how it's done in the real world, but they have neither data on wire gauge nor even photos.)

And how exactly does low Q affect the performance of a Theremin?"

A single layer winding air core ("solenoid") with aspect ratio of 1:1 or higher (taller rather than squatter if the winding axis is vertical) will give minimal self-C.  Everything else is a compromise, though one you may be able to live with, it really depends on the oscillator topology, operating frequency, etc.  Ferrite will always introduce temperature drift and Q losses.

Less distance between the ends of a coil tends to correlate with higher self-C. 

Q is a direct factor in voltage swing at the antenna.  Drive a Q=60 coil with a 1Vp-p signal and you get 60Vp-p if the driving phase is perfect and if there are no RF or human body losses.  You want something over 30Vp-p so as to overcome environmental interference. 

The Q I'm referring to is as measured at the resonance frequency of the Theremin, not some arbitrary and usually quite a bit lower frequency as is normally given the datasheet (Q always drops with increasing frequency).

For my air core solenoids I've used anywhere from 32 to 40 AWG, it depends on the former dimensions and the target mH.  One would think that larger diameter wire with its lower DCR and lower RF skin losses would give dramatically higher Q, but that's not the case in this application, where RF and human body resistance losses dictate maximum Q of the arrangement.  38 and 40 AWG are a real bear to wind, not fun.  My designs tend to be higher frequency than traditional analog Theremins (>1MHz vs. ~300kHz), so the coil values are smaller, and the coils are physically quite a bit smaller, which is a big air core enabler.  You could also use digital dividers to work at higher frequencies, though that would substantially deviate from the EW.  And if you're going that route, you might as well go all digital IMO.

I can recommend the Inca program for simulating inductors and transformers: [LINK]

And I use my own spreadsheet to design single layer air core solenoids: [LINK]

@dewster

A single layer winding air core ("solenoid") with aspect ratio of 1:1 or higher (taller rather than squatter if the winding axis is vertical) will give minimal self-C.

Ah, I think I see now. I have initially thought of self capacitance as consisting of only elementary capacitances between any piece of copper and its respective closest pieces on neighbouring turns of winding, which is clearly wrong; what I forgot is capacitance between said piece and all other pieces of copper and other metal. So by making the coil taller we are bringing the diametrically opposite pieces of copper on the same turn closer to each other (and somewhat raise the self-capacitance of the closest group of turns), but the distant turns are moving further away along the coil axis and with greater speed, which outweighs the effect of reducing the diameter.

Anyway, even if I am again wrong and even without complete physical understanding, I feel like I can trust your expertise and go with that. Thanks again.

Less distance between the ends of a coil tends to correlate with higher self-C.

Seems to fit my understanding so far.

The Q I'm referring to is as measured at the resonance frequency of the Theremin, not some arbitrary and usually quite a bit lower frequency as is normally given the datasheet (Q always drops with increasing frequency).

In a couple datasheets I've seen graphs of Q(f), for f from zero to self resonance — and the curves were kind of bell-shaped with peaks at about 0.4×SRF (and Q indeed falling after that). In those datasheets the specified Q (Q[sub]min[/sub], to be exact) was given as the lowest value on that graph, usually either at 0 or at SRF, depending on the coil. So I took it to be the standard format of specification, and, if I understand correctly, I need to look for an inductor with Q[sub]min[/sub] of at least 30 and SRF at least 1.5 times the resonant frequency of the circuit. Is that correct?

38 and 40 AWG are a real bear to wind, not fun.

I can only start to imagine different reasons for why is that so

Guess I'll try to wind some aircores with 32-35 AWG to start and also get some manufactured ones for comparison and possibly for backup.

You could also use digital dividers to work at higher frequencies, though that would substantially deviate from the EW. And if you're going that route, you might as well go all digital IMO.

Something in me is strongly against that But I understand this path has many benefits both in design and final product usability and versatility. Maybe some couple decades later, when I will need something more from a Theremin than just cello-like sound output.

I can recommend the Inca program for simulating inductors and transformers:

Thanks! Exactly what I've started to look for, but only found Gecko Magnetics so far.

And I use my own spreadsheet to design single layer air core solenoids:

And thanks again!
By the way, what would you say if I made a user-friendly publicly available web version of that spreadsheet? The world needs more searchable online tools for rare and complex computations IMO.

"Ah, I think I see now. I have initially thought of self capacitance as consisting of only elementary capacitances between any piece of copper and its respective closest pieces on neighbouring turns of winding, which is clearly wrong; what I forgot is capacitance between said piece and all other pieces of copper and other metal. So by making the coil taller we are bringing the diametrically opposite pieces of copper on the same turn closer to each other (and somewhat raise the self-capacitance of the closest group of turns), but the distant turns are moving further away along the coil axis and with greater speed, which outweighs the effect of reducing the diameter."  - nazro

There's a phase effect across the windings at resonance, where the end windings have a 90 degree phase difference, and therefore the most possible capacitance between them, but they are also most distant from each other in a single layer solenoid.  Windings right next to each other have very little phase difference but are quite close.  It's apparently a very complex problem for which to produce an exact closed-form mathematical solution.

"In a couple datasheets I've seen graphs of Q(f), for f from zero to self resonance — and the curves were kind of bell-shaped with peaks at about 0.4×SRF (and Q indeed falling after that). In those datasheets the specified Q (Q[sub]min[/sub], to be exact) was given as the lowest value on that graph, usually either at 0 or at SRF, depending on the coil."

You're probably right, and I probably misspoke, I can't say I've ever seen a graph of Q vs. frequency.  The datasheets I've seen, and test equipment as well, specify Q at fixed frequencies like 10kHz, 100kHz, etc.

"So I took it to be the standard format of specification, and, if I understand correctly, I need to look for an inductor with Q[sub]min[/sub] of at least 30 and SRF at least 1.5 times the resonant frequency of the circuit. Is that correct?"

Yes, for Theremin EQ inductors the SRF must be substantially above the operating point, and the higher the Q the better.

"By the way, what would you say if I made a user-friendly publicly available web version of that spreadsheet? The world needs more searchable online tools for rare and complex computations IMO."

Feel free to do whatever you wish with the spreadsheet.

I feel like I'm steering you away from ferrite inductors for your EW clone, and that's not what I intended.  Air core EQ inductors can be quite large, and will behave differently than ferrite core, and the value of the EQ inductor is something you may need to "play with" in the design to get the response of the axis to your hand correct (hence the changes here in the design of the EW, which I can attest to are real improvements).  These RF chokes are becoming much more difficult to purchase since Bourns / Miller stopped producing them some time ago, but there may be alternatives out there.  And there are vintage coil winders that you might purchase if you really want to get into it.  Another possibility is a custom cylindrical wooden or plastic former with deep slots cut in it to hold the individual donuts of windings, a brilliant solution TW member pitts8rh came up with.

For your very first Theremin you might want to go the non-EQ coil route, which is much simpler to tune.  And you can tune things with small variable capacitors rather than tuning the coils, which might make things easier from a construction angle.  Even with all of my test equipment, tuning a perfectly functional factory made EW for the first time was a grueling experience for me, and subsequent tunings only somewhat less so.  I never really know if I got it "right" nor if it will experience possibly catastrophic non-oscillation in different playing locations / over time / etc.  My digital approach is infinitely simpler to calibrate and gives a much more linear playing field.

I feel like I'm steering you away from ferrite inductors for your EW clone,

No, not at all I will anyway fiddle with Inca for a while before I finalize my decision, it's just that aircores have some non-technical advantages — first of which is that I won't have to source a core from somewhere.

For your very first Theremin you might want to go the non-EQ coil route, which is much simpler to tune. And you can tune things with small variable capacitors rather than tuning the coils, which might make things easier from a construction angle. Even with all of my test equipment, tuning a perfectly functional factory made EW for the first time was a grueling experience for me, and subsequent tunings only somewhat less so.

There actually is a scheme I'm looking into as a second candidate [link], which is tuned by variable caps, though it still calls for some 4 × 10 mH inductors for antenna linearization — but given a way lower frequency (172 kHz) I guess it is less critical to coils' self-C (the central idea here is that self-C lowers SRF, right?). However ambitious I might be, I hear you — an "Etherwave from scratch" is probably really too much for someone with zero skill like me

Once again, I deeply thank you for all your help!

"... aircores have some non-technical advantages — first of which is that I won't have to source a core from somewhere."  - nazrow

That's very true.  If your design can accommodate air cores, you can fabricate better ones (higher Q and SRF) than you can buy anywhere.

"... but given a way lower frequency (172 kHz) I guess it is less critical to coils' self-C (the central idea here is that self-C lowers SRF, right?)."

I've found that it's oddly the case that SRF and mH needs tend to track a given operating frequency, hence the use of many chokes in series.  For that reason I predict that you'll never find a commercial choke that will function singly as a Theremin EQ coil.