Self-Playing Piano, Part Eight

Bradley Gannon

2026-02-22

TL;DR: I partially built a second octave module and updated the host-side code to drive more than one octave at a time. To my surprise, everything kind of worked as I expected.

Quick test of both modules working together. I need to sand down some of the plungers and solenoid supports to get them to come back up more easily when released.

There isn’t much to say about the octave module build process because I’ve covered all the components individually in previous posts. Nothing really went wrong during assembly. It just takes a long time. I’ve listed the general procedure below.

Plungers
- Print (5) black plunger extensions, (7) white plunger extensions, and (12) plunger hats
- Cut, sand, and drill steel rods to act as plungers
- Glue plunger hats and extensions to plungers
- Sand down the extensions a little
- Glue silicone domes to plunger extensions
Solenoids
- Print (12) solenoid supports
- Sand down the inner walls of the supports
- Wind solenoids
- Solder extension wires on solenoid leads and insulate solder joints
- Twist extension wires together and crimp connectors
Driver circuit
- Cut PCB on CNC router
- Sand down PCB and manually fix any shorts between nodes
- Install and solder all components
- Flash firmware on Arduino Nano and install
Final assembly
- Print solenoid platform
- Bolt solenoids to platform
- Drill mounting holes and bolt platform to rails
- Connect solenoids to driver board
- Load plungers into solenoids
- Smoke test
- Chromatic scale test

I started writing that list thinking it wouldn’t be that long, but I guess there’s a lot to it. Each individual key’s hardware takes maybe an hour or two of attention from start to finish, but there are just so many keys that the time really adds up. This has made me a lot less excited about the project lately, but I intend to spend the next period or two working on making the system better at actually making music instead of expanding it further. Maybe that will help me reset.

A partially assembled octave module PCB held by a third hand clamp. Various components sit waiting for their turn to be soldered on.

Plungers in the process of being assembled. Drilled steel rods next to plastic extensions.

I was planning to build a complete second octave, but I ran out of diodes halfway through. I’ll need to order more, although if I get desperate then I can salvage some from previous boards. I also forgot about the sanding steps on both the plunger extensions and the solenoid supports, so I’ll need to do that too. (That’s part of why some of the keys are sticking in the video above.)

The host-side code also got some improvements this week. I’d previously just been using it to control a single octave module with a hardcoded index, but now it can handle all eight, at least in principle. I was happy to find that it worked with the second module right away with zero debugging. Rust boosters like to say that “if it compiles, it works”, and I suppose this is an example of that. In the future I might refactor the code to use a thread for each module rather than hitting them serially, but I kind of doubt that will even be necessary.

The two octave modules installed on the piano and viewed from the front such that the plunger extensions and keys are visible

The octave modules viewed from above, revealing the plunger hats and solenoid leads

I’m not looking forward to repeating what I did this week six more times—to say nothing of the fact that I already had a dozen solenoids already wound. In total I’m probably looking at 100 more hours or so of this kind of work just to build out the keyboard. I’ll have to be careful to spread that work out over a long period so I don’t resent it too much. Fortunately, I think now that I have more keys to work with, I can have some fun working on key dynamics to really dial in the driver logic.