The Computation Easel (perhaps to be known as the Oblique Easel? We’ll see where this goes) is something I’d like to get at least to being a functional prototype. So I’ve been doing the nitty gritty of building something that works right, even if it’s a little inelegant.
The first big challenge I had was heat. The front end of the entire machine is PlugData, a very serious update to the venerable, if hard to read, Pure Data graphical programming language.
PlugData is pretty resource intensive and it’s already running the Raspberry Pi 4B at 78°, which is within spec, but that’s without even doing anything intense.
Thermal pad that moves heat out of the CPU, RAM, and GPU.
A vapor-filled copper heat pipe. Next time, I should get one twice as long so it can make better contact with the heat sink.
Some thermal paste on the opposite end of the heat pipe. The pipe has to extend past all the ports.
Three standoffs to hold the RPi in place. There would be four, but the drill bit broke off in the fourth, leaving an extremely sharp shard that I cut myself on. Three will have to do.
The RPi bolted in place. Note the extremely inconvenient location of the GPIO pins.
The heat pipe means that, even overclocked to 2GHz, the RPi is only running at 50°C!
This heat sink is taken from a car amplifier I found on the side of the road.
But oh no! The ribbon connector is too tall! It’ll prevent the heat pipe from contacting the heat sink!
Gotta clip off the GPIO pins
Man, this low-lead solder is hard to work. Some of the pins I can’t heat up enough to get out.
I had to keep stopping to prevent the board from overheating. This was a really tedious process. Some pins just wouldn’t heat up enough, and some had solder or some other metal that simply would not flow out
So I had to drill them out. This is risky.
OK, all drilled out!
Pin headers now come out the back.
I tried putting them in the front, but there really wasn’t room.
PlugData is pretty resource intensive, but in principle at least it can do most of the experimental programming one might want to do on a low-power computer. It can talk with GPIO pins and I2C analog/digital converters, so with a little level shifting and Op-Amp buffering, it should be a really good digital platform.
It’s my hope that, by the end of this phase of the project, it can safely talk with experimental electronic circuits, in particular Eurorack synthesizers. Ideally, that breakout board will be something generally useful for digital synthesizer experimentors!
Modular systems are a function of industrial society. But do people of The Fifth World still know how to agree to standards? With their acute interest in efficiency, I think they might have carried that lesson forward!