Build Project: Dewster's D-Lev Digital Theremin

I've decided to break this project into a separate topic to keep clutter out of Dewster's original thread.  If this ends up being successful it may also provide others who may be interested in building the D-Lev with some more curated information relevant to the build.  Dewster's years of effort on this project, his documentation of the evolutionary progress, even including the dead ends, and the open sharing of the ongoing results represents a remarkable effort that I think should be applauded. I hope this project will be good for both of us and others as well.

As I've said earlier, I'm totally at Dewster's mercy regarding the programming side of this.  I may pick up a thing or two about the mechanics of dealing with FPGAs, and maybe even learn a little about HDL, but I'm really only comfortable with the other hardware and the packaging aspects of the project.  And Dewster left precious little analog to play with.  Nevertheless, we move forward...

I'm taking a number of liberties compared to the way that Dewster built his prototype.  Although the most important task is to make a working prototype for evaluation, it seems worthwhile to try to make some incremental changes along the way.  I've voiced some concerns about pitch and volume plates requiring a different playing technique, but I'm going to give them and the visual tuner a good evaluation as-designed before going off on my own.  I don't think that anyone who follows Dewster's thread would accuse him of designing anything without putting a lot of thought into every detail, but some aspects of playability and aesthetics are very personal, and my main goal is to also put a lot of my own thought into making a theremin that feels comfortable to play and doesn't require too much technique adjustment from other theremins. Some of the fabrication methods that I use are not really mainstream, and the pictures aren't intended to represent a tutorial.  I'm just throwing ideas out there to add to the Internets of Knowledge, the same source that I research and borrow (steal) from every day.  

The pictures below will chronicle the progress on a prototype enclosure and the pc boards for a first whack at this.  Everything will change as this progresses.  This enclosure has no resemblance to how a final version would look - for one thing, I dislike using PVC pipe in any projects where it's obvious.  The conventional box-with-two-arms evolved from the way the two coils were made on PVC forms.  These were the first parts built, and the adjustable pipe arms and main box just sort of grew onto them.  I don't use a tuner, but I'm going to build this one anyway with the intent of making it detachable or otherwise viewable in the direction of pitch plate; this is why the main box that you see here has no tuner yet.  The whole wood/acrylic combo was an easy approach for a prototype that also provided an opportunity to experiment with simple graphics on the back of acrylic.  The sliding arms and fully adjustable plates are necessary for now to give as many setup options as possible, and hopefully these could be simplified later after learning more about the theremin.

For convenience, the links to the build information that Dewster compiled recently in his original thread are repeated here:

[1598] - PROTOTYPE: Parts List.
[1602] - PROTOTYPE: Tools List.
[1605] - PROTOTYPE: Pitch & Volume Antennas, AFE Board Schematic with PIX.
[1610] - PROTOTYPE: LED tuner with PIX.
[1613] - PROTOTYPE: SPI EEPROM, TOSLINK TX with PIX.
[1615] - PROTOTYPE: Rotary Encoders.
[1616] - PROTOTYPE: FPGA & LCD.
[1621] - PROTOTYPE: Fashion Shots (PIX).
[1625] - PROTOTYPE: Operating Frequencies.
[1634] - PROTOTYPE: Rotary Encoder hookup.

Photos to come....

PS: this will be my first attempt to link to photos hosted on Amazon Drive, and it's not working quite like I would expect. If I can't figure it out I'll put them elsewhere. EDIT: I had to host on Dropbox instead.  The images are larger than shown here.  Right Click>View Image will show the full 1200 x 800 size.

"I'm totally at Dewster's mercy regarding the programming side of this."

*Sinister laugh*

"... the adjustable pipe arms..."

I have a couple of CCTV aluminum tube camera mounts that have a nice flange on one end and a ball joint on the other.  The ball joint / thread mount are aluminum so it wouldn't take much punishment, but if the ball were replaced it probably could take more.  Otherwise they'd make passable articulated variable length antenna arms.  You could even run the CAT5 through the pipes.

"I've voiced some concerns about pitch and volume plates requiring a different playing technique, but I'm going to give them and the visual tuner a good evaluation as-designed before going off on my own."

I'm afraid I've pushed back too hard on your proposed changes, it's only because I was afraid that you might not do so, so thank you for the chance for an evaluation!  But in the end, the customer is always right, particularly if they are musicians or otherwise "intimate with" the product.

Roger, I'm really looking forward to your ideas, observations, and build details!

I have not been following the original thread all that closely since Dewster went all digital (ha), but I was still surprised to see how few parts are needed.  Not shown are the passive components and discrete semiconductors, but trust me, the parts count is low.

This is the heart of the theremin, the FPGA board and USB Blaster to program it:

LEDs and 7-segment for the tuner display, 4x20 LCD main display, rotary encoders, RJ45 jacks:

Toslink Digital to analog interface:

USB to TTL interface cable:

Next: air-core inductors for pitch and volume antennas

The coils were the first parts that I made, and even here I went off the rails a bit.  Dewster used two different PVC form diameters for the pitch and volume inductors, but knowing what was coming for my prototype cabinet design I wanted to make them the same diameter.  The parts list linked above has his specification for the coil build. 

My volume coil (the one with clear insulation) is 2.73" inches of close-wound #36 single insulated wire on a 1-1/4" thick-wall (schedule 40) PVC pipe with an OD of 1.9".  The pitch coil is the green one with 2.68" of #32 wire on the same type of form.  Sorry I don't have the turns count, but I was too lazy to set up a counter on my lathe, and I started to lose count after the first winding tangle hiccup.  I ended up just filling the form with wire and then removing turns while measuring the inductance, hence the measurement of the finished coil lengths rather than number of turns.

Dewster's approach for these coils is solid.  Winding is easy even if done manually and the clear heat shrink makes for a pretty and rugged end result. I routed the ends of the wires through holes in the PVC pipe and anchored them with hot-melt inside.  Leads were soldered to the magnet wire and again hot-melted to the inside of the pipe for strain relief.  I kept my inductors open because I would be routing the sense signal through the center.

Both coils came out very close to the 2mH (pitch) and 4mH (volume) targets.  In the picture below I was looking at the general behavior of the air-core coils including the self-resonant frequency and sensitivity to materials placed nearby:

Next: making a display and control panel for the main enclosure

"... I was still surprised to see how few parts are needed."  - pitts8rh

As I told FredM, my goal was to do as little as possible in analog, then run screaming back into the digital realm.  For a low-budgeter, I'm somewhat certain all of the actives on the AFE boards could be removed, and If there were enough 3.3V power on the FPGA board (there isn't because the regulators only have PCB copper as heatsinks) the LED tuner actives could likely be removed as well.

I'm not seeing your pictures in any of my various browsers?  [EDIT] Seeing your pix now, nice!

I wanted to use plastic for the main enclosure panel.  Metal would be more convenient but I like to avoid it as much as possible for sensitivity reasons.  Wood is too thick for a front panel in most cases, so acrylic was the choice.

For me it seemed easier to just make a 3-sided acrylic half-box to mate with a matching wood half-cabinet.  I do this with metal often for home projects - it's just easier than an all-metal or all-wood enclosure for accessibility, which is what was needed for a prototype.

While I was waiting for some cell cast acrylic sheet to come from Delvies Plastics, I went through a dry run of bending up a panel using hardware-store acrylic sheet, which is more difficult to work with because it melts when you saw or drill it.  Nevertheless, I ended up with something that looked quite good and I proceeded to build a wood cabinet to fit it.  I finished the backside of the clear acrylic with the painted graphics you will see below.  The next day every bend had fractures and the piece was garbage.  When the good material arrived I had the more difficult task of forming the new acrylic to fit the already-made wood cabinet, but it came out, even if it is a little tight.  The cell cast material is worth waiting for.

Anyway, I'm getting ahead of myself. 

These two pictures show the the sum total complexity of my hot-wire acrylic bender for this project.  It is a piece of nichrome wire held in tension at a uniform height through an aluminum channel.  The channel helps contain the heat to give sufficient heated material for the bend radius while keeping the bend sharp.  I used a 9v, 100VA transformer with alligator clips, and it was plugged into variac to heat the wire a dull red.  The acrylic sheet to be bent is set with the bend line centered in the channel.  When sufficiently heated after about 90 seconds, the acrylic is hand bent to the required angle.

There are dozens of videos on YouTube showing this procedure.

I don't have any pictures of the panel forming process,  but you'll see it all finished a few pictures below.

The acrylic was bent along four lines into the finished shape, drilled for the switch and rotary encoders, and then cleaned inside and out.  A protective film of sticky shelf paper was applied to the outside.  I wanted to put the D-Lev name on the front and back, and have a bordered LCD display, so I cut vinyl shapes in reverse and applied them to the interior to act as a paint mask:

The inside was then sprayed with several coats of white and allowed to dry a bit, but not overnight.  The masks for the lettering and the outer LCD display border were peeled off.  The inner LCD window mask was left in place.

The interior was now painted completely black.  After a couple hours of drying the last remaining mask over the LCD window was peeled off;

Next:  the reveal

Now we are beginning to see the acrylic portion of the enclosure.  Having the paint on the inside means that it cannot be scratched (except during work inside or assembly) plus it looks perfectly glossy.  This has a lot of potential, but I would switch to paints without ketones for the longevity of the acrylic.

I next built a wood enclosure to fit the acrylic panel, but I'll show that a little later.

I had been working on some layouts for the encoder and antenna drive (AFE) PC boards and I took a day to clean those up, head to the basement to make the boards, drill, and then populate them.  These are simple 2-layer layouts - there will still be plenty of jumper wire interconnects because I can't do board vias (except hand wired), so crossover tangles are best accomplished in the air with wires.

The encoder boards are simply lines running to edge header pins (there are pads for RC debounce filters on all connections).  These were made a little wide to allow for screw holes that will support a bracket for a central main board.  They were made a little too wide, and they conflict with the wood cabinet if both are left upright, so one is flipped over.  No problem.

The antenna driver boards were etched and plated at the same time as the encoder boards.  These were intended to slide into the 1-1/4" PVC pipes that would be used for the extension arms for the antennas.  But when I installed the RJ45 connector I realized that they should be centered a bit more on the board for clearance so that the top edge doesn't conflict with the pipe radius.  They fit, but I'll move the connectors if any more boards of this design are made.

The RJ45 connector brings 5v power to the board, a drive signal IN from the FPGA board, and the zero and quad signals OUT to the FPGA through a short ethernet cable. The green terminal block on the opposite end has an output to the inductor and an input directly from the antenna.  I put a ground in the center just for convenience. 

In this picture the 1pF/100pF capacitor divider (the two caps near the green terminal block) is used on-board.  I also have an extension board to allow the 1pF cap to be placed closer to the antenna, but for now it seems to work where it is.  The first cap to the left of the block is a high voltage 1pF@200v.  Note to D:  I did check the adjacent capacitance between terminals and decided that I didn't need an extra space.  Plus I didn't have any 4-section terminals...

Link to antenna driver board schematic.