Building a Scotto9 Pico

I too got sucked into the custom keyboard space. It was a combination of learning that there is significant ergonomic gains to be had and they seemed like fun and useful electronics projects.

Thus I built a low profile Kyria rev2.1 last winter as my first custom keyboard, coming from a stock 60% Ducky one 2 mini. I’m quite happy with it and glad I took the plunge (these things aren’t exactly cheap).
Researching and assembling the Kyria taught me that keyboards are, in fact, fairly simple devices. They’re just a bunch of switches (that part wasn’t revelatory) assembled in a matrix hooked up to a microcontroller, preferably through some diodes. Exactly because of this simplicity, people have been handwiring keyboards, omitting the need for PCBs. Combined with the increasing availability of 3D Printers for manufacturing custom cases for said handwired builds, there’s some pretty cool things possible as the forms a keyboard like that can take are neigh endless.

Scotto9 Pico

So I wanted to get into it aswell. And to do so I decided to buy a 3D Printer, learn some CAD and build a Scotto9 modified to use a Raspberry Pi Pico for it’s MCU instead of a Pro Micro and with hotswappable switches.

(Cute right?)

Instead of giving a build guide, I wanted to talk about my modifications and challenges.

Build Process

Soldering the Kailh Hotswap sockets was quite difficult. Because they’re designed to go onto PCBs, their contacts don’t lend themselves super well to just tacking a wire onto them. Worse if (like suggested by people in the handwiring space) you use thick copper wire, since heating up both the thick wire and the unwiedly pad was hard. But obviously nothing you can’t do with a little patience.
On that note, I wouldn’t try to use enamelled copper wire again. Mine said “solderable” but the enamel wouldn’t burn off, likely due to the heat issues mentioned just now. I went with “normal” .75mm² multistranded wire in the end, removing a couple strands to make it easier to heat. It works just fine and still gives enough structural integrity to the construction which is why the solid copper wire is usually recommended.

The other deviation from the stock Scotto9 was using a Raspberry Pi Pico MCU. Reasoning here is quite simple: The Pico costs a third of a pro micro and is far more powerful. Only drawback is it’s bigger footprint, which is why I had to modify the case and keyplate by enlarging the slot for the MCU and moving two conflicting screwholes. (Even though I printed the plate before noticing these modifications were necessary so I had to drill two holes. One of them even lined up in the end!)

Here you can see the full wiring of the build: (Absolutely flawless solder job with zero room for improvement if I do say so myself /s)

Finally while trying to flash my QMK firmware I ran into the issue that when I programmed it to use the second row, the keyboard wouldn’t even be recognized.
Advice: Check for shorts. I had a single strand of the jumperwire touch the neighboring GPIO on the underside of the Pico.

Only gripe I have with the design still is that it’s quite hard to plug a cable into the Pico through the case. I do wanna try fixing that design. Once I do I will provide the STEP files here should anyone wanna follow this route.

Until next time!

Edit 2023-09-02: As promised I fixed the design. However, instead of narrowing the wall with the cablehole, I made sure the Pico sat more snugly in its slot. This way the cable wouldn’t press it into the case and thus make good contact. A dab of hotglue should do the trick in case the Pico is still pushed back with this new design.

As promised here are the STEP files for the Case and the Plate.