Build Project: Dewster's D-Lev Digital Theremin

Very nice!  I assume U4 is a 3.3V regulator.

Yes.  This is the same circuit that was used on the AFE boards.

Is the extra line of thru-holes above the display board connector to transfer connectivity from one side to the other?

Also yes.  I thought I could get away with a single-side board, but when I decided to use the rear-mount straight header instead of a front right-angle header, I had to put the pads on the front.  One of the downsides of these homemade boards is that you can only have opposite-layer pad connections for parts where you can't access the component side for soldering.  This requires extra vias and it eats up space around the components.  Fortunately the machined-pin DIP sockets allow enough space for top-soldering.  The other type of sockets are much more difficult to solder.

I ended up making a main board to support the FPGA module and LCD display and at the same time take care of all of the interconnects (except to the encoder boards on the sides - these will still be jumpers).  For a while I was just going to do what Dewster did with the Dupont interconnects for all FPGA connections, but the EEprom and toslink transmitter needed a board anyway, so I just went for it.  I seriously doubt that this board will become operational without at least some cut lines and blue-wire mods, but hopefully there isn't a major problem with flipped orientations or something like that.  I was confused quite a few times in the layout process.  

This board is 4" x 6" and I was a little worried about the layer registration using the faster toner-transfer process, so I had to make it using transparencies and photopolymer resist film.  Toner transfer works well for small 2-layer boards, but despite  my best efforts in pre-drying the transfer paper base, I still get unpredictable shrinkage and mis-registration over larger areas when heat bonding to the copper-clad material.

This is a picture of the exposed board and the artwork used to make it:

And here is the etched and drilled board:  ON EDIT - Be warned that this board layout has a few errors.  These will be noted later on.

This main board will hold the LCD display on standoffs and will itself be suspended on brackets by the two encoder boards which mount to the acrylic panel.  This way the entire assembly with encoders may be removed from the panel and bench tested.

The back of the board has only two active components other than the FPGA module.  It's all just interconnects.  The header banks in the four corners  connect to the headers on the encoder boards.  If I didn't screw anything up, the order should match.  If not I'll have crossed wires, but that's still not a problem.

And this is what the board stack looks like:

Next up:  Making the tuner enclosure and main board mounts.

Lookin' sharp Roger!

"... blue-wire mods..."

Where I worked it was "white-wire" and I never gave it any thought - until a new guy said it was "blue-wire" where he came from.

Well done! Roger, though i rather won't label my boards i am interested in your technique. It looks like silkscreen. You are the one for climbing steep hills for sure … but silkscreening single boards … ??

Where I worked it was "white-wire" and I never gave it any thought - until a new guy said it was "blue-wire" where he came from. -Dewster

I think it depends on what your official corporate repair wire color happened to be.  We had a large spool of blue wire-wrap wire, which everyone knows makes far better repairs than the white type .

Well done! Roger, though i rather won't label my boards i am interested in your technique. It looks like silkscreen. You are the one for climbing steep hills for sure … but silkscreening single boards … ?? - Dominik

The labeling is done by the same toner-transfer etch-resist process that I use for etching smaller boards when not using photopolymer.  I have an old stock of toner-transfer paper similar to the type sold at pcbfx.com, which is really just a laser-printable paper that has a water soluble coating.  You print the silkscreen layer from your board layout software (positive/mirrored for the top silk) onto the special paper with a laser (not inkjet!) printer.  The paper is then aligned with the board and run through a hot laminator, and the toner from the print melts onto and adheres to the board. The board and paper are submerged in water for a couple minutes, and the paper floats off, leaving the black toner print on the board.

The only difference between printing a graphics layer this way and printing traces to act as etching resist is that an extra step is included where a thin plastic foil is run through the heat laminator to bond to the surface of the traces, providing a more effective seal against etchant than the toner alone can provide.  My results in the past were always unreliable until I bought a decent thermal pouch laminator and started using the foil.  The traces are more ragged under the microscope when compared to those done with the photo process, but they are still very good and the process is extremely fast.

There is also a white foil for bonding to the board labels if you want your screening to look white.  I find that the black on G10 substrate is nearly as readable.  All of the foils are quite cheap, and the paper is about 1USD per sheet.  YouTube has lots of videos showing the process.

But since you brought it up - yes, I have done single board and panel artwork silk screening (genuine) at home in past years, and I'm afraid I'm going to have to do it again for the artwork logo for the Moog Melodia front panel.  I still have my screens, but I need to get some fresh photo film (it's different from pc board film).  I'm hoping that the recent experience with photo boards will make it less painful and messy.

Feel free to send me an email if I can be of more help.

Thanks! I wasn't aware that toner transfer is working that good.

Doing a little work on the final assembly of the main enclosure...

I bent some strips of 1/8" x 3/8" steel to support an internal acrylic top that would provide additional mounting surface for the toslink audio converter and USB blaster, and whatever else might come along.  I didn't want to have these extras mounted to the wood cabinet or the removable acrylic front panel.  Now by removing only the front panel nuts the entire electronics assembly can be separated from the outer acrylic shell to allow access to all sides of the circuit boards.

This whole assembly then mounts inside the outer shell with only the encoder nuts holding it in place:

It is getting dangerously close to showtime, and with that the reality of trying to make this thing work.  Nothing has been pretested, and the laws of probability say that there will be at least some wiring errors on the boards.  These are the last beauty shots that I will have before programming the FPGA and the inevitable cutting and modifying begins.

Wait.. I do have to make a tuner enclosure.  This will help me procrastinate a bit more.  

Wow Roger, super nice!  Nothing says "real equipment" like internal metal chassis stuff going on.

You probably won't need the USB Blaster in there much, but it's good that it has a spot in case you do.

I'm still not sure what I'll do with the SPDIF box (in or out of the enclosure).  Electrically connected audio is more universal, gives you a convenient way to ground the Theremin, chassis jacks are easier to mount, and it contains / concentrates all of the optical and powering mess inside.  But optical is just so clean. Could do both I suppose by driving a second TOSLINK SPDIF box off of another FPGA pin.

(BTW, is anyone else getting that heart stopping "an error occurred" message when posting to TW?  I've gotten one for each of my last 4 or so posts.)

Wow Roger, super nice!  Nothing says "real equipment" like internal metal chassis stuff going on.
Thank you! 

You probably won't need the USB Blaster in there much, but it's good that it has a spot in case you do.
It's mounted on 3M Command Strip velcros, so it can come and go at will.

I'm still not sure what I'll do with the SPDIF box (in or out of the enclosure).  Electrically connected audio is more universal, gives you a convenient way to ground the Theremin, chassis jacks are easier to mount, and it contains / concentrates all of the optical and powering mess inside.  But optical is just so clean. Could do both I suppose by driving a second TOSLINK SPDIF box off of another FPGA pin.

Personally I always prefer equipment that has an internal power supply with a standard IEC detachable power cord, and I would try to implement that for any final version of a homebuilt theremin as well (not this prototype, though).  This would provide the grounding without the use of a less-mainstream wall wart (as with the Etherwave) and would still allow the simplicity and noise immunity of an optical output.  A production theremin would be a different matter however; the added complexity (read:cost) of UL/ETL or CE certification with an internal power supply makes the use of pre-approved wall warts more attractive.

As a sidenote, since I play seated I have experimented with placing a foil "butt plate" under my seat that is then connected by a short wire to the theremin ground.  This provides a more well-defined ground return than that provided by a mains or audio cable ground, and the positive effect is most obvious when noting the much-reduced frequency shift when touching any metal on the theremin body itself (unlike the Etherwave, the Subscope has exposed metal switches). The volume and pitch sensitivities change slightly as well.  Clearly this is not for everyone, particularly standing players, unless you resort to conductive (chain-mail?) undergarments with a wrist-strap style snap connector. 

(BTW, is anyone else getting that heart stopping "an error occurred" message when posting to TW?  I've gotten one for each of my last 4 or so posts.)

Yes, every time recently.  And I still forget to grab and copy the text before hitting the "Post" button.  Sometimes I just write everything as a notepad file first and then copy/paste to the editor.  But I even forget to do this most of the time.

In the two weeks since the last update, my prototype of Dewster's D-Lev design has been completed (except for a tuner enclosure) and I've actually been playing it for over a week (yay!).   I've been in contact with Dewster on a daily basis to get up and running, and he has written some very detailed procedures that I need to extract from our many emails and edit for posting.  I'll come back and cover some of the detailed steps that took place to get to this point, but I wanted to show just how quickly this theremin came together.

I'm not going to gush too much until I can back up my enthusiasm with sound bites (or you can go listen to many in Dewster's main thread), but as it turns out it actually isn't a digital engineer's "R2D2 on a kazoo".  In fact I would say it's pretty friggin' awesome, and my bottom-side is telling me that I've been sitting and playing too long again today. 

Below are pictures of the current hardware configuration, which now has a conventional rod antenna instead of the original plate.  The pitch plate was incredibly sensitive and very linear, but having come from playing a standard rod/loop design, I immediately felt a difference with intervals and the way that I land on notes.  I did see this coming as a potential issue, but thankfully the design is so forgiving that I was able to simply replace the plate and adjustable mount with a randomly long 3/8" aluminum rod, and it continued to work with absolutely no tuning.  I need to go back and look at my pitch antenna waveforms and maybe do a little optimizing, but the nature of the digital PLL design allows this amazing tolerance.

Here's the overall view with pitch rod replacing the plate.  I did also tuck a small ferrite into the pitch inductor to compensate for the reduced surface area and capacitance of the rod.  The PITCH System Menu allows for a wide range of linearity settings, and spatial span for octave intervals can be adjusted too.  The pitch field can be huge - there's no cramping around the antenna for the high notes here.  It's very comfortable to play with the arm held more vertically instead of having to stretch out. I would call the pitch control awesome too, but I've already overused that word... 

The volume plate extended to a comfortable playing position.  Once I figure out my ideal playing positions I will eliminate the swivels on the real cabinet:

Main menu screen.  "load" is the preset number;  "stor" is the memory location for storing a preset;  "pnul" behaves nearly like a conventional pitch adjustment on an analog theremin; "acal" initiates the autocal for initializing settings to the player's position; "vol" is master volume; and "oct" is for register shifting as with the Etherwave Pro.  Menus are well-organized and easily navigated with each of the eight encoder knobs dedicated to the respective screen item.  Some major test equipment manufacturers could do well to take this approach and put more knobs back on their panels.

The conventional rod pitch antenna, threaded and screwed into the tapped hole in the ball.  In the future I will just eliminate the swivel on this end.

I milled a rectangular hole on the left for an RJ45 coupler (ivory plastic) for the external tuner connection. Internally a short cable running between the PC board and the coupler in the wood base allows for easy detachment when the acrylic cover and electronics assembly must be removed.  I managed to crack the acrylic top during this little machining exercise, but this isn't a very practical design for a permanent cabinet anyway. The USB programming cables simply hang through a hole so that I wouldn't have to deal with bulkhead panel connectors. Aside from the barrel power connector there is one audio jack and room for a second, probably for pitch preview (coming). Grounding is currently accomplished through the audio connection, but later I may opt to put a power supply inside the cabinet, allowing the detachable power cord to supply ground (I don't like wall-wart supplies with flimsy and short cords). This is powered by a single 5VDC power adapter, and it can be a switcher, too.

Next up:  Dewster's FPGA and EEPROM programming procedures with some of my commentary.