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

If anyone is interested in my "Hive" processor core, I posted a design document for it today over at opencores.org:

http://opencores.org/usercontent,doc,1371986749

Here is the project page:

http://opencores.org/project,hive

I'll have the verilog up as soon as I'm done verifying it (within the week I hope, no bugs so far).  If the core doesn't produce too much heat I plan on using it in this digital Theremin project.

[EDIT 2012-06-24] The verilog is now up as well.

.dewster,

I wish I was more technical in understanding the theremin you are developing. What would be your features list that it would demonstrate that sets it apart from currently available theremin designs?

I imagine the goal would be to dampen interest in the E-pro. Have you recorded any type of audio or visual demonstration over the past year of this thread of what you are working on?

What would be the hoped for retail price if you offered it publicly? The big question is……should I wait for your first marketable version or is this more your hobby with no commercial intentions?

T

"What would be your features list that it would demonstrate that sets it apart from currently available theremin designs?" - Touchless

It will be mostly digital, which means (unlike most analog Theremins) the distance sensing and voice generation functions will be decoupled, allowing for more control, variety, and variation in both.  I'm hoping to implement at least 2 axes of control for the left hand.  There will be presets, pitch preview, a fast LED "tuner", and the whole thing should be easier to break down and reassemble than an EWS.

At long last, I believe I've finally hit the point where I can really start prototyping, trying out my new oscillator coils, topologies, and linearization methods.

Not sure on price or availability yet, even if I don't get dragged down with too much house repair stuff it will be many months.  What I'd really like to do is formalize the design, work out the bugs, and have some kind of field upgrade option for the software, which would allow me to continue to add features post sale.  Certainly the digital route, if done right, can really facilitate manufacturing by putting a lot of the sensitive stuff in digital logic, where it doesn't need tweaking post assembly.  I'm hoping to have a design that doesn't take me more than a day or two of work to assemble and ship out.

"the distance sensing and voice generation functions will be decoupled, allowing for more control, variety, and variation in both." - Dewster

I agree with this philosophy - Whether analogue or digital, I believe this is the best route for future high-end theremins. (I believe this is the topology employed in the Moog 91 series and at least partly employed in the Moog E-Vox)

The above allows the distance data (Whether digital or voltage or current) to be scaled allowing playing field "density" (number of notes / octaves, distance between these, etc) to suite the players requirements / ability, and also allows things like register switching and linearity correction to be implemented without messing at the high frequency front end.

With the "old way" where the sound generation and distance sensing were combined into a single circuit, change to anything caused change to everything. "decoupling" the functions allows changes to be made with zero impact on other 'modules' ..

The developer is then free to implement the "voice" in whatever way they choose - whether this be with analogue synthesiser technology (as per 91 series / E-Vox) or digital technology (As Dewster is doing), or voltage controlled analogue heterodyning technology, or combining heterodyning with analogue synthesis (as I am doing).

The "old way" is by far the simplest and cheapest to implement, and was the only practical way to build a theremin in the 30's - but since about the 60's "decoupled" modular design has been possible. I believe there is a place for both topologies - but for high-end theremins or where a wider range of voice / tone charactaristics are required, I believe the "decoupled" route is superior.

"even if I don't get dragged down with too much house repair stuff ".. Perhaps your greater hazard will be interest in Hive!  - It would not surprise me at all if demand for / interest in this (IMO) brilliant core became your primary distraction! 

"I imagine the goal would be to dampen interest in the E-pro" - Touchless

Apart from one hobbyist "developer" I have seen, who has such gross delusions of grandeur that he advises E-Pro owners to sell their instruments because his toy will cause them to de-value, I doubt that it is the goal of any developer to "dampen interest in the E-pro" - The E-Pro is not in production, and is no "competition" (or if it is "competition" then there simply is no market! - if one is competing against 2nd hand E-Pro's, and demand is limited to perhaps a few dozen potential purchasers of these.. ) - I do not think that the E-Pro's status will ever be affected by any future developments - it is iconic, its the last production theremin Bob created, and it is a great theremin...

I think developers are (or should be) aiming to equal or better the E-Pro (or better the E-Vox which is far superior to the E-Pro I think, from my cursory examination, and from the sounds I have heard from them) - But I have an ancient Moog Prodigy, and it is still worth more than the new low-cost Moogs being produced today, even though these new Moogs are far more capable instruments - And I would not sell it to buy a new Moog!  One gets attached to ones instruments - even to their "imperfections"!

Fred.

Hello, I have a popular tourist shop with several respectable theremins and the E-pro is one of them. When I mentioned features the first one that comes to mind is prevention of the drifting tone higher or lower over the course of the day. When I first open up in the morning the room temperature can be as low as 45 degrees, then warms up to 90 degrees with the outside temperature over 100.  Some of the inexpensive models were ridiculous so I keep them turned off, but then it is hard to sell them. The RCA runs warm but seems the most stable; otherwise I walk around every hour or more and readjust each one. I think I am doing a nice service for the theremin as visitors usually spend more time exploring the theremin room than anywhere else. They are spaced about 10 feet apart with knickknack wood shelves in between. On an occasion I do resell the used models. There is no profit selling something for $100, so dewster give your theremin respect with the proper retail market value.

PS: Whether a theremin is huge or tiny they still require the same floor space. I would rather sell one a month and make $500 with minimal inventory space used than sell five a month and make $100 each. This is a bit exaggerated but gets the point across. People on vacation are looking for a novelty to take home. I get quite a few of the Hollywood names passing through.

T

"The "old way" is by far the simplest and cheapest to implement, and was the only practical way to build a theremin in the 30's - but since about the 60's "decoupled" modular design has been possible. I believe there is a place for both topologies - but for high-end theremins or where a wider range of voice / tone charactaristics are required, I believe the "decoupled" route is superior." - FredM

This neatly sums up my feelings as well. 

And Fred, I want to publicly (and personally) thank you for your openness with your excellent engineering ideas, your constant kind words and encouragement, and for helping me sort out how to handle the Hive release - I was feeling kind of alone with that.

"When I mentioned features the first one that comes to mind is prevention of the drifting tone higher or lower over the course of the day." - Touchless

Very interesting to hear your observations regarding the thermal stability of the Theremins you sell.  This is an extremely important issue and I should have put it high on my feature list.  Some instability is unavoidable due to the changing electrical properties of the air surrounding the antennas.  The remaining instability can be largely cancelled if the fixed and variable oscillators are designed to "drift together" and are thermally coupled to one another.  But I suspect there are many Theremins out there that don't play nice with changing temps (I know because some of my prototypes were worse in this way than others).  Much of this could be due to the use of ferrites in the tank and EQ coils, something that is making me strongly consider the use of air core coils (I've simulated this quite a bit, but haven't actually built an air coil oscillator yet).

"There is no profit selling something for $100, so dewster give your theremin respect with the proper retail market value."  - Touchless

Thank you and I do hear what you are saying.  Unless the prototype gets crazy, I'm thinking $500 or so ballpark.

Dewster, I am glad to have been of help - and thank you for the acknowledgment. The most frustrating aspect of my life is lonelyness - Having someone to share ideas with (be these crazy or brilliant) who can understand what I am talking about, and having someone with ideas they share with me - well, its probably the most important interaction I get.. So thank you!

Touchless -

Thank you for your comments on thermal stability - it is really interesting to hear your observations.. The " I walk around every hour or more and readjust each one." is a really useful reference. (and actually fits my expierience with my H1 theremins - they needed re-tuning a few times each day - they were not designed to a high standard regarding drift, and its surprising they were as good as they were.. I expect that with the temperature change of 25C you are subjecting your theremins to, they would have needed re-tuning within an hour) I suppose that a few hours of operation without re-tuning is probably acceptable in most situations -

I had wondered if I was being pedantic in my efforts to avoid thermal drift - as an engineer, drift (to me) is a sign of incompetent design.. LOL ;-)  .... However, the truth is that sometimes increasing the cost of an instrument just to produce a good specification, is perhaps more "incompetent" - particularly if this "improvement" doesnt confer any actual benefit.

My method of obtaining thermal (on real theremins I an developing, not low - end stuff like the H1) stability is perhaps massive overkill - A quite inellegant "sledge hammer" approach.. I encase temperature sensitive components in polyurithane foam casting, and have a temperature sensor and heater (resistors) in close proximity to these components, thermally coupled to them... An "oven" circuit heats the assembly to above the maximum temperature they would ever attain due to the environment (I usually set this to 50C), and keeps the assembly regulated within 1 degree of this temperature - once the temperature is attained (a few minutes) there is no source of drift - except external factors such as expansion of the antenna, which is extremely minor*. (The main reason for the thermal insulation is to reduce the power consumed by the heater)

But the question I have is whether the extra expense and difficulty is worth it - Also, once assembled and potted, repair or modification to these components is difficult or impossible.. If all theremins other than the RCA require adjustment hourly, then perhaps a drift time of an hour or a bit more is acceptable.. It may well be that the reason the RCA is so stable is because it is running hot and virtually self-regulating its temperature... That, and the fact that there are no ferrites.

I have recently been playing with simply wrapping the inductors and temperature sensitive components in thermal insulator, and not having a heater - performance is no where near as good as implementing an oven, but it is far better than just having these components open, and FAR cheaper and easier.

" Whether a theremin is huge or tiny they still require the same floor space."

Yes, this has been the case - The solution IMO is directional antennas - if the antenna is completely insensitive to the field behind it, one halves the required "clear" area - if one focusses the forward field, one can reduce the required area even further. The "decoupled" topology allows this to be done quite easily.. I had been playing with active shielding of antennas, but with the "decoupled" topology one can use passive shielding, and compensate for the loss of sensitivity by increasing the gain of the sensor circuitry - something you cannot do with the old topology.

Fred.

*There are some external components which are a headache - tuning controls for example - these cannot be encased, and need to be compensated by other more complex means - variable capacitors and potentiometers are the worst - and adjustable inductors need to be in the oven with access to their slug as they are a major cause of drift - its the damn ferrite inductors, fixed or adjustable,  which are the biggest pain.

This is what I like about voltage control - one can generate a temperature dependant correction voltage to compensate for variation in the voltage from a tuning control (potentiometer) - and you can adjust this compensation manually under controlled conditions - no need to mess about adding NTC capacitors or whatever.

But how far does one go ???

Hive is a barrel processor!  I knew the idea wasn't original with me, but I had no idea the concept was so old (1964 Cray designed CDC 6000 series peripheral processors) and that it has been implemented many times since:

http://en.wikipedia.org/wiki/Barrel_processor

Thanks to Thomas over on comp.arch.fpga for letting me in on this!

Very little if anything new under the sun it seems.

"Very little if anything new under the sun it seems" - Dewster

Yeah - And that "matters" if one is trying to make money out of ones ideas - but its less important if one having ideas for other reasons...  Reading the Wiki page, I suspect that what you have implemented is a variation on the barrel processor (warning - I REALLY dont know what im talking about here ;-) and is still "new" in terms of its implementation - seems the originals had a lot of registers...

Whatever - coming up with an idea that Cray used for his super-computers and implementing this on an FPGA, without any reference to Crays design - well.. Nothing there to be embarrased about or to devalue your design in any way!

Fred.

"Whatever - coming up with an idea that Cray used for his super-computers and implementing this on an FPGA, without any reference to Crays design - well.. Nothing there to be embarrased about or to devalue your design in any way!"  - FredM

I wish I had come up with that independently!  I examined high level documentation for a barrel processor several years ago but had never encountered them before so I assumed the concept was new - so much for that!  Just as well, if the concept has been in use since the 60's I assume anyone can use it at this point (cue the lawsuits).  It's a pretty nifty construct, not sure why it isn't used more often.  I suppose programmers are more familiar with single threaded coding, but those days are kind of past with multi-threaded multi-cores in all of our PCs.

AFAIK my hybrid register/stack has some newness associated with it - I've never encountered an explicit pop bit anyway.  And I haven't seen shift distance treated as signed in any opcode listing I've had the pleasure to peruse (yet).

[EDIT] Glen Herrmannsfeldt over at comp.arch.fpga kindly pointed out to me that the PDP 10 uses signed shift distances.