Goals of this project:
Create an instrument with two button decks using a single brain, but still allow for the portability that having separate devices allows.
Design a platform with interchangeable parts that can be modified or augmented without reprinting the entire case.
Build in a way that allows construction without access to (or the desire to use) a large volume 3D printer for parts. To that end:
The case walls, decks and panels are fastened together with wood screws rather than being printed in place.
Case walls share the same width/height dimensions and could easily be swapped with 1/2" x 3" (13mm x 64mm) hobby board.
Button decks and bottom panels could be constructed using flat 1/8" (3mm) wood project panels/acrylic sheets and a laser cutter/CNC machine (or some careful drill work with a step bit).
What was learned from this prototype:
I felt confident enough in the design to solder directly to the button lugs this time, and this may be my preferred method going forward vs. using spade quick-connectors. It's cheaper, the connection is more reliable, it takes about the same amount of time to assemble the wiring matrix regardless, and ironically I think it's actually easier on the buttons.
Magnets are not an ideal solution for holding case parts together. The pull is both too weak to hold together for transportation, and too strong to allow for easy separation. Future redesigns will likely use a locking pin driven through from the case top or bottom if needed.
Using DB25 connectors as an intermediary between the button decks and the micro-controller works amazingly well, and may be my preferred method going forward, even when not passing signal through a case wall. Not needing to worry about damaging an expensive microcontroller by re-soldering pins is a relief.
After months of testing, I was confident enough that Wicki-Hayden is my preferred button layout, but I was dissatisfied with the clumsy control layout (and general poor quality) of the Melodicade Prototype #3, and decided to try designing something that could be assembled and reconfigured from smaller separate pieces, rather than printed in place and then disposed of when changes need to be made.
My original intention was to build this using wood and laser cut acrylic, but the idea of printing smaller parts flat had me curious to know if the annoyances of warping and 36 hour long print failures could be resolved this way. Printing many smaller pieces is definitely easier. Warping issues persist however, and something more durable than PLA may be required for a fully finished product.
I was never entirely happy with the control module layout due to the compromises required to save space. Also, the lack of any visual feedback for settings changes was becoming a problem. Future designs will incorporate LCD/OLED displays, or at least LED indicators.