Theremin Development - (more) questions for thereminists.

Many thanks for taking the time to look over my blurb, Roy! ;-)

Yeh, I love knobs - loads of them, everywhere! But (apart from in the analogue synth world) they seem to be regarded as clumsy - I think it may be a British thing (perhaps - although Bob wasnt British..) We were slapping 4 knobs for channel EQ on mixing consoles when everyone else had at best a simple tone control.. Funny to see the "British EQ" of the 60's being touted on modern mixers from Germany and America!

I HATE almost every DUI I have ever had to use - The main reason I originally got the BCR-2000 was to have knobs for my soft-synths etc, and I have probably spent more on buying slider based (Fu**ing useless!) controllers for my digital synths (Roland D110 etc) than on the synths themselves..

BUT..

In terms of cost and space, digital user interfaces win hands down - Particularly if they are crap with multiple menu levels and tiny display! - I would though still love to have a DX7 with every operator taken to a pot - One massive wall full of knobs! ;-)

Being realistic.. The theremin (unlike a synth) isnt a beast one tweeks during performance - you would only use the DUI to program the sounds into presets, and to access these presets during performance.

And in a way, this is the double whammy - its true for digital synths as well... You spend a day getting "used to" using the crap DUI and programming the presets to sound like you want - and all is fine thereafter, as you just switch between these presets... In this phase, having a simple knob free panel with one encoder you use to change presets is fine (unless like me you always want to be fiddling with knobs in real time)..

Problem comes when in 8 months you dont like a few of your presets, want to edit them, but have forgotten how (particularly if youve been going through the whole process using other equipment with different awful DUI's) and it takes 1/2 a day to edit 3 presets!

So me thinks..

A "simple" theremin with perhaps maximum 24 knobs in total, some of these being small "trimmer" pots (functions like formant's with which one would set the "character" of your instrument but rarely change)  and no preset memory, no DUI, no display, no MIDI.

Then an advanced theremin with only knobs for tuning/linearity/span/register and perhaps audio output level and preview, everything else resides in a moderately implemented DUI which usually just displays the preset number (and name?) and has an encoder normally assigned to this - but has a few buttons to allow you to select any parameter and edit it - this DUI also receives and transmits MIDI so knob twiddlers can buy a BCR-2000... Option 2 without any form of "DUI" isnt actually available I realized .. It needs to store the presets / edited values and enable to instrument to select a preset - this means that a minimum "UI" with encoder and display must be fitted - And if fitted, may as well add the ability to get to the parameters for those who want this - its not something I would use though, I would just leave it permanently on preset selection mode.

Looking at the BCR-2000 I wonder how, even with the huge sales, Behringer can put them out at anything close to the price! - People are hacking them to build step sequencers and lighting controllers and all sorts of other goodies because they are a tiny fraction of the price you would pay to buy the parts!

I just hope they dont get discontinued.

Fred.

"Perhaps I dont need to - perhaps some interface like this:.... If the user has problems with drivers etc, they can deal with the supplier" -- Fred

Completely agree. It is the strategy that allows to quickly start. In my case, it was the direct connection
GamePort<-->uC,  even without any optoisolation. 

"it was never that good in the first place (low bandwidth, not inherently bidirectional)" -- dewster

Below is an example of  session when the theremin send a digitized waveform from the volume resonance circuit (~300 kHz) in real time.

The current refresh rate is about 10 fps (because of operating algorithm, not of MIDI limits).

"Below is an example of  session when the theremin send a digitized waveform from the volume resonance circuit (~300 kHz) in real time."  - ILYA

I'm not exactly sure what you're getting at.  MIDI baud rate is 1MHz / 32 = 31.250 kb/s.  Even getting raw 7 bit samples at this rate (+1 command bit, +1 start bit, +1 stop bit = 10 bits) would give 31.250 kb/s / 10 b/sample = 3.125 ksamples/s.  Nyquist here is 1.5625kHz.

Your dialog box looks nice!  What is it coded in and what is the windowing (widget) manager?

Yeah, that does look nice, Ilya!

Ok - lets forget MIDI ;-) - It was fantastic back in the days when I had a ZX-Spectrum home computer and a Sequential Six-track synth (pre 1990?), it has past its "sell by" date really, but its still fine for solo keyboards, for slow (non-real-time) bulk transfers... Its no good for theremins and only just perhaps barely good enough IMO for EWI's... I would never use it for anything other than non-real-time theremin functions, and the only reason to use it for these is that one can get ready-to-go MIDI controllers.

But as I shake this all down, and start really doing an options / time / cost evaluation, I am being forced to change the design - I wanted to be able to produce every reasonable possible sound from the basic unit, but I think my plans are OTT, at least for the first instrument/s -

My 'plans' now - These are the present state, may change.. Quite technical, but it always seems to go like this.. I think I frighten musicians away - which is a shame, because they are the ones whose ideas I really need... The following probably isnt for them..

Main board: (P401_WC1)

Essentially, I am now looking at a core board with the PSoC4 and oscillators and everything required  for linearity and span adjustment and register switching - everything for pitch and volume, and a comprehensive mixed-signal audio generator will also be in the core PSoC.. these essential aspects will not come under the economizing 'knife' -

This board will output mixed signal (effectively PWM) audio waveforms, will output RF signals for reference and variable oscillators scaled by the register switching factor, and will output volume CV and other signals (some multiplied RF frequencies and SPI, PWMs etc to be used by other boards - it will also have a serial interface for communication with other boards, but for most builds this wont be in use. No digital functions will be used to control anything on this board, all controls (tuning etc) will be potentiometers.

Audio Board Level 1 (AM01_WC1):

With an Alpha board, this would create a complete theremin - All controls will be analogue (potentiometers) there will be no digital control, and therefore no ability to store or recall patches or presets. This board will have two analogue mixers (one diode based 'EM clone' voice, one multiplier based 'sine' voice) but I will not be fitting the comprehensive equalization / morphing controls I had planned - these require 6 potentiometers and a lot of extra circuitry, so they are best included in a theremin with digital control.

The board will also process one (of the available 4) mixed signal waveforms from the Alpha board, this waveform (the ramp) will have dynamic wave shaping as a function of pitch, and a comprehensive pitch dependent filter - all these relationships being under user control.. It is my belief that this one 'engine' can achieve most of the dynamic changing of the waveform as a function of pitch, that the more complex filters / morphing of my prior design did.... Simulations sound good, but I will only know for sure when ive built it... Anyway, this board will have everything else; VCA. out mix, preview, FM fb - and 16 user potentiometers rather than the 24 - 32 in my prior 'minimum' design.

Audio Formant board ("LEVitator" ):

Real pain at throwing the formant functions off the Beta board, but it consumes four dual potentiometers (so these cant be put on finger adjustable presets and four other potentiometers, - this board is far easier / cheaper to implement with digital potentiometers, and will be included as standard on digitally controlled boards.. Its optional - But IMO, its possibly the most important board for capturing classic theremin - making it a stand-alone manual potentiometer based board does open the possibility to use it with other theremins..

Audio Board Level 2 / DUI (DCA01/UI01_WC1):

This is to be a board set comprising a digital user interface and digitally controlled audio board replacing AM01_WC1 and "LEVitator" boards and hugely increasing audio functions, would work with the main board, take control of audio functions on this board (and download patches / edits to it), deal with MIDI control data (CC) exchange with external equipment, facilitate preset selection / editing / all user functions.

Status:

The RF oscillators are all based on a single design - I have not seen a design like this before, but its a series resonant topology with ~160V P-P on the "antenna" node, but also has a CV input node so any oscillator can be controlled either by capacitance (normal theremin VFO behaviour) or by voltage (VCO) or by a combination of both. All "antenna" oscillators are phase locked - Pitch to the pitch reference oscillator, volume to a fixed PSoC generated frequency, Reference is also locked to a PSoC clock, and the PLL error voltage is used for thermal correction of the one free-phase oscillator (the pitch VFO).

These oscillators and the system have been simulated, but only a single oscillator has been built and tested.

The PSoC4 "heart" has been configured and tested as far as is possible at this stage.

Audio circuits have been simulated, some to .wav files so I can get some idea of how they might sound.Some of the audio / wave-shaping stuff has been tested or tested before in past projects.

Now at the point of putting the main board final design together (wont do the software or config for the PSoC communications / SPI now, as I wont need these till I get to level 2 board) and im designing the level 1 board, and after simulation, its PCB time ;-)

TONS to do ;-)

If there's anything you want that I have left out, or anything important - please let me know soon!

I believe I have taken care of the following:

Pitch 'antenna' linearity and span (both adjustable), Volume antenna linearity and field size / location, Register switching (at present going for rotary switch, 4 octaves) and wave crafting, audio output and preview...

Have one pin left on my PSoC - I could implement automatic muting for wrong-side-of-null, and if nothing more essential turns up that I have forgotten, this will be done - I do actually have a couple of pins that could be used at the last step, because if I use them I cannot use my debugger..

Fred.

I am also thinking about selling the main board (with full TRM) and / or programmed PSoC4 boards to those interested in developing their own "upside down" (PLL topology) based theremins.. No promises, but if you are really interested in this possibility, email me.

phiu, so many technical terms, i  don't understand,but i don't need to, really. for expl. what's TRM? so the following might sound a bit or more than that stupid : CV- output? for the synthguys? or would this be superabundant, because your tmin features all possibilitys already?  so, sorry for bringing that pre-midi-era stuff back.

But that behringer knob board looks enormously! the price is probably so low, because behringer has his own facility in china?  and a slight slap on the backhead of anybody who calls you, well wathever.;-) keep the good work going on.

(mixture - quite technical at times)

Hello Dani !

CV - Yeh... The master PSoC has the required outputs to drive a separate pitch to CV board, this board will require another PSoC4 (because I do things a bit differently - I dont think that pitch-voltage which has huge latency at the bass end and falls over below ~100Hz as Bob's design does, is really that much use.. Yes, I could just slap in a P-V which takes the audio, as everyone else does.. But one of my best ideas, IMO, was a way to take the RF frequencies and from these derive the pitch CV far faster than the audio update rate and track right down to 8Hz)

What I do is to multiply both the pitch and reference oscillator frequencies by 16, then heterodyne these - this gives me an audio frequency 16* the audio frequency being played... This scheme only works for a theremin - it cannot be used with any instrument that doesn't employ heterodyning! - And this is probably why nobody else thought of doing it - I think I probably thought of it when I realized how Bob did the EWP register switching by dividing down both oscillators - and I thought about multiplying them up with PLL - and then the light turned on! .... Or the idea may have come when pondering the whole issue of CV latency and discussing it here on TW - I remember Thierry saying that the idea was nonsense and wouldnt work, and this inspired me more than anything else to build it! LOL ;-)

Its a teeny bit more complicated with a register switching theremin though - I struggled for a long time on my last prototype of this (at that time I was trying to use CV as the primary control route) because the PLLs are fine at tracking small changes in the oscillator frequencies, and these always change <10% on a theremin - but they wont track octave changes, which is what you get from register switching..

So I was messing about with switching capacitors on the PLL VCO's and filters when the register switch was changed, and all sorts of nonsense like that, and getting nowhere slowly - until it dawned on me that all I needed to do was to multiply the master frequencies (before register division), get the CV from this, do the required log conversion to get 1V/Octave, then subtract 1V from this for every octave lower the register switched to...

It is, however, quite a complex fast circuit required to obtain the CV - the "audio" into it spans about 128Hz to 128kHz (to cover 8Hz to 8kHz of "real" audio) and I am hoping to put it all inside a PSoC - implement the PLL multipliers and VCO's in the PSoC, do the heterodyning mostly on-chip, do automatic scaling on the difference frequency to improve resolution at the high frequency end, get the period value using counters if I can rather than analogue ramp (although there are analogue components on-chip that could do this if it works out better, which it might.. ) and then hopefully end up doing the log conversion and register scaling numerically before outputting the CV through some form of DAC.. (I would also like to include a quantizer, so I could implement on-key-emphasis or Gordon's "ping" and output this quantized CV on a separate channel - Ideally having scale lookup tables and interface for the digital user interface on instruments with this theres an Arm Cortex M0 in the PSoC, so all sorts of bells and whistles should be possible..)

As you will see from the above, CV has been an extremely high priority for me - but its going to have to wait.. But its a board that can be fitted to any build, and I am thinking that I would like to make it compatible with other theremins as well.. Including other register switching theremins ;-)

As for "sorry for bringing that pre-midi-era stuff back." - Dont be! IMO, CV is just perfect for theremins - its the only control scheme for which an abundance of compatible "sound engines" exist... IMO, CV and theremins were made for each other.

"what's TRM? " .. Technical Reference Manual - Its the thing that gives all the details and specifications, the document a builder / developer can refer to, and contains everything they need to know about the chip / board / instrument, and in the old days gave a load of helpful non-essential information as well... Writing these is something I did for about a year when the technical author resigned and I was the only person in the company with enough understanding and good enough English to do the job - apart from which, most engineers would rather be boiled in oil than take the job ;-) .... I loved it! -

"and a slight slap on the backhead of anybody who calls you, well wathever.;-)"

LOL ;-) ... Ok, I gave myself a slap! - Because im probably the most guilty! ;-)

Fred.

Dani -

Thinking and talking about CV..

All the complexity is because I am thinking in terms of having a theremin voice and CV tracking this..

But actually, a "CV Only" instrument should be real easy to do - the main board with some changes to the linearizing circuit and quite a simple small output board to replace the audio board, and I believe I could have a CV control "theremin" with all the features (Adjustable linearity and span, register switching, adjustable volume lin / span) which just put out pitch and volume CV's and gate... But no audio, and no possibility of audio from this instrument (unless this audio was generated by a CV synth engine, or a *CV theremin..).

Its been my obsession with "genuine" and "heterodyning" which has kept me from going this route - I nearly did - Its a LOT simpler than the route of deriving a CV from the difference frequency..

*But to use this CV to generate a "true" heterodyning theremin topology, one needs a theremin "voice" which accepts 1V/Octave control signals.. I have prototyped such a voice - its essentially the whole process of generating multiplied audio by PLL, deriving a CV from this, but instead of outputting this CV, using it to compare against the incoming CV, and using the error voltage to lock the 'voice' to the pitch defined by the incoming CV... Effectively a complete theremin tied into a phase-locked-loop and containing the whole PV board..

It has amazing potential - one could detune multiple theremin voices against each other, and play sequenced parts and.... I may revisit this idea once I have a PSoC doing the pitch-voltage, because thats the major chunk of the circuit.

Fred.

fred, as you mentioned, true: it's easy to getting slowly nowhere while even more possibilities and demands on a instrument show up. on any artistic project maybe. even more today. with all these tiny controllers, i'm completely overwhelmed by those and also overpowered. and somewhat aware that this will reduce the theremin to a gestural controller for any purpose of sound production. daw,synths,etc..maybe even poly. so, if the cv-min would be so easy to do, keep it as a seeyalater-project, maybe? as for me, i'll love the idea of it. but go on with your: the voice is important instrument. it sounds just too cool, in my imaginations i get, while reading almost all that stuff all you ee's etc.. write here. even without understanding or long wiki searches for explanations for things, i love it. yes i did read the horowitz-hill bible,unfortunately it belongs to the local library, but was a good read anyway. so understandable written.  i already can hear a distant scream: RTFTRM !,?#*+ LOL. so a good, understandable interface would be very important.

many people like that, to have a little lady or a lovely baritone singing in a box, so formant filters implementation sounds great. to have a talking machine inside already?  (i opened the pedal version of those, had a look and closed it back right away.).but i'll probably rehouse it anyway for a tabletop version laterly to do any no good with it...sometimes i could do simply with more cv-outs... :-)  have a nice day everybody.

"to have a talking machine inside already?  " - Dani

No, its not a TM.. I wouldn't want a TM in my theremin ;-) ... Oh, I was chasing that kind of idea intensely a while back, but the TM actually convinced me that this was folly.. IMO, one doesn't want a human voice from your theremin to get a "classic" sound - you just want enough of a "hint".. IMO, its possibly another case where with the theremin "less is more".

My focus was on "vocal" formants - but I was missing the obvious IMO - All acoustic instruments have formants, and each instrument, every violin, cello - every instrument with an acoustic resonator, has its own unique formant signature .. oh, these can be extremely close to other "identical" instruments, but its unique nonetheless. Some formants can be at the same frequencies as human formants, and such instruments will have some "vocal" qualities..

But the thing about the human voice is that the formants change dynamically and radically as the person talks and sings and these formants are fully under their control - this does not happen in the resonators of most acoustic instruments - their formants are fixed.. so if these are on vocal formant frequencies, one will get an "aah" or an "ooh" or whatever vowel sound permanently, only the excitation waveform, amplitude and pitch will change.

And this is where I believe the "just a hint" becomes important - because, IMO, a strong vocal signature, even if superbly "engineered" and having some formant movement (as the TM) becomes irritating after a while - whereas a theremin with a hint of vocal and string formants doesn't become irritating. THE TM is too good!

I think the theremin is probably the ideal interface for a true "Singing Machine" - I was working on a 3d loop "antenna" where hand movement on the X and Y axis controlled (changed) the formants so one could actually articulate and control the volume, while controlling pitch with the other hand - but right now I am not thinking "singing machine" I am thinking theremin.

"if the cv-min would be so easy to do, keep it as a seeyalater-project, maybe?"

Actually, your mentioning CV may have been the most helpful 'prod' I have had on this thread.. I have two "requirements" - one is to survive, the other is to make some great theremins.. I need to start earning money a year ago ;-) ..

Every instrument I build using this new topology will need the main board, so this is priority one.. My plan was to build this and a level-1 audio board first..

But actually, the CV board (and modification to the linearizing) is loads simpler than the level-1 audio board - as in, really simple - a few opamps... The CV synth stuff is simple tried and tested, so producing a CV only "controller theremin" onto the market first would be no deviation, almost everything for that needs to be done anyway.. Adding a CV synth voice so it was also a "theremin" that produced audio would be a doddle... And the whole synthesis market is there waiting for a product like this even if its spurned by the theremin community..

Realistically / commercially, this is by far the sanest first product - It would have everything (linearity, low latency, span control, register switching) that my "real" theremin will have, all it wouldn't have is a "true" theremin voice.... This bothers me - but it wont bother the theremini buying crowd, and it doesn't really bother precision thereminist who play the Moog 91 series theremins which have a nearly identical topology, and it would be simpler / lower cost than my true level-1 theremin..

And it doesn't require any major change in direction - A CV theremin with or without built-in synth for those more into synthesis than theremins, and a true theremin without CV at this time for those more into theremins than synthesis.. The two would be mutually exclusive (one couldn't change the one build into another, or add CV to the "theremin voice" build or a "theremin voice" to the CV build until I have produced the CV board and/or the CV-Controlled theremin voice, which may never happen)

All I need to do (the only changes to my present design) is to make the main board so the changes to the linearity correction can be implemented without a board-bodge, and to check that everything is in place so that I can just drop in either an audio or CV board.. I may actually put the linearization stuff on a small module so that I can keep the main board as a standard build and just plug in the appropriate EQ module - but these are all just trivial details.

Thank you, Dani.

Fred.

All I care about it linearity, responsiveness, a very nice selection vocal colors, minimalistic design that looks great on stage (easy to put together and dismantle) but sturdy enough to withstand travel bumps, Pitch preview, mute option ~ and of course made to also accomodate lefties.