September 25, 2026
Do you remember how I spent a year building a development board for a cool display I found on DigiKey? It was awesome.
You might have noticed that I then fell into the pit trap of retro computing, and while my time down there was reasonably fruitful, retro computing is ultimately a dead end. This took me a long time to realize, because for years I would get extremely hyped over any given Macintosh. They're cool! But they're obsolete for a reason and unless you have a really cool idea that ties the old machine to the new world, they're also a dead end for projects. This is a problem! I really like having side projects simmering in the background, as a destination for unused brain cycles. Once I'd restored the PowerBook, I couldn't invent any projects with it that I found actually motivating.
I don't have the space, interest, or money to be a collector, so the PowerBook sits on my shelf, as a toy to show off and an excuse to head over to the local retro meetup every few months. In other words, the gray plastic box on my shelf cost me $250+ and sometimes I get a "huh, neat" out of it.
Money well spent. At least I know how to install a tantalum capacitor now.
In any event, one fun fact about me is that I've conceptualized at least five cyberdecks (Google that if the word is unfamiliar) over the last five-ish years:
These were all intended to be a long-term and extremely captivating side project. These also all died in the planning stages, save for rakko which got a blog post, and the 4th one, which got a mostly complete schematic. I don't like making projects that feel derivative, but I simultaneously hold myself to way too high of a standard for what I decide is "creative" or "novel". This is why the only cool and technical thing I've done in the last six months was shove a 1993 LCD controller into a logic analyzer and discover four mostly irrelevant nuggets of information. Was it rewarding? Yes. Is that really enough to satisfy my passion for creation and understanding? Not really.
Philosophy aside, let's start with a conclusion I've drawn about cyberdecks. If you're building a deck, you have two options:
Neither choice is necessarily better or worse than the other. The fun part of a project like a cyberdeck is how you apply your imagination to the creation, not the raw materials and parts you build it from. If you're happy with a spray-painted Pelican case and a bunch of off-the-shelf parts, more power to you---but I don't find that appealing and I know I wouldn't find it satisfying.
Obviously there's a technical tradeoff here as well. A Pi is power-hungry and can't go into low-power sleep, but it can run all the Linux software out there and can perform tasks that require real computing power. On the other side, a microcontroller can sleep and runs at lower power anyway, but you pretty much have to program it yourself, so it's limited on the software side by the creator. No answer is perfect, but some are cooler.
This brings us to today's project. I was chatting with a friend about a month ago, and our conversation pointed me toward my Next Great Idea: a cyberdeck built on the Baochip-1x open source microcontroller. I call it Kirdos, or maybe kirdos, which I suppose establishes fake Greek words as recurring theme with me.
Kirdos is another cyberdeck, but I'm a lot more confident in this one than any of the previous ones, except rakko, but that was 2+ years ago and much less interesting to start with. (You might reasonably read this as "but this one's different!" but I think it's actually different.) Kirdos is a little different from other cyberdecks:
std environment
including process management. This isn't like FreeRTOS, it's
much closer to a real OS that you can compile against. A
functional, available operating system solves the biggest
issue with microcontroller cyberdecks: you can't run Firefox,
but you don't have to reinvent the same primitives everyone
else has already reinvented.This list is paired with the fact that I'm currently a better engineer than I have been at any previous point in time. If I stick to these principles and do good work, Kirdos can justify its own existence by being something new and cool and challenging that doesn't completely reinvent the wheel.
I have decided that Kirdos needs to have a reasonably well-established mechanical form before I commit to electronic design. You can get a basic idea of what I'm planning from this image:
This is heavily inspired by some of those funky laptops from the 1980s, like the Epson PX-8 Geneva:
By Electrickery - Own work, CC BY-SA 3.0, https://commons.wikimedia.org/w/index.php?curid=7391164
Kirdos's keyboard is a small ortholinear built for Kailh Choc keys, mounted at an angle that rises up to the bottom of the display panel (which I call the "lid" even though it's not a laptop lid in the traditional sense). In the display panel is the display mixer, a PCB that connects the big horizontal display to the same pins as the Sharpie display, which itself mounts on top of the display mixer. The space underneath and behind the lid is for the battery, electronics, and two large expansion module areas.
I decided to start with the lid hinge, because that is necessarily a moving part and those are never easy. My first concept was a ratchet:
This design could work, but it would have a lot of slop and probably wouldn't come off the printer nearly as well as I'd want it to. The total width of that sprocket is maybe 15mm and the sprocket teeth are 2mm tall. The previously mentioned friend (who's a genuine mechanical engineer instead of whatever I am) saw this and suggested a much wiser design: bearing balls to make some kind of clicky spring-loaded mechanism. What you see here is my implementation of that concept.
Ignore the fact that everything is mated face-to-face, and imagine some space between the green part and the blue part to the left occupied by bearings.
On the inside face of a Kirdos side panel is a circular cutout that holds a square post and four short pillars.
This coupleswith the bearing mount, a circle with offset hemispheres cut into one surface and some features on the back that match the side panel. The opposite side of this surface has a protruding square channel that mates onto the square post, and four similar pillars. When the bearing mount is mated onto the side panel, the square channel sits over the square post, locking the rotational position of the bearing mount. The clearance on this mate is sufficiently large that the bearing coupler is still able to move horizontally, and the pillars hold springs that push the bearing mount away from the side panel.
The bearing mount holds 3mm bearing balls that are superglued in place.
The opposite side of these bearings couples against the bearing coupler, which has similar offset hemispherical cutouts and arms that mount to the lid. An M3-sized hole runs through the side panel, bearing mount, and bearing coupler, and the screw that passes through these holes mates with a nut that fits in a cutout in the bottom of the bearing handler. This keeps the hinge assembly together and holds the springs under tension. When rotational force is put on the hinge, the bearing mount moves away from the bearing coupler on the springs, and the bearing coupler snaps into the next position that aligns its cutouts with the fixed bearing array.
I can host videos now!
This works really well. I designed the hinge to only support 12 snapped positions, which translates to 4 usable positions in the actual system. In the final assembly, the tensioner/assembly screw needs a drop of Loctite so that rotations of the hinge don't work it out of the nut in the bearing coupler.
There is a lot left to do. I won't list it all here, but there will be future blog posts that slowly transition from mechanical to electronic as Kirdos hatches and grows. (There really will be!) Stay tuned.