I may also have slightly cracked the acrylic when riveting. Using a hand riveter instead of a pneumatic one didn't change anything either. I think it had something to do with the holes not being aligned properly, but even when I redrilled it nothing changed. Not sure what's going on.
We couldn't find a DC barrel jack of the right size to power the USB hub, so we had to rip it apart and desolder the jack and solder on some wires in its place. It took a few tries and I learned more about soldering, 3 more to go! Calling it a night here.
Also wired up all of the fans together. Since we couldn't find any jumpers the size that we wanted, we made our own out of some wire. They work well; all of the fans were powered from the power supply and running smoothly. It looks great with all of the LEDs running as well. We can't wait to shove a strip of RGB leds in there too.
Day 3 summary: Finished the case, it works very well and the racks slide in nice and smoothly. We attached the raspberry pi model b's to the racks, 2 per rack. Since we only had 7 of them with FTDI drivers, we're leaving one of the racks unpopulated for now. After spending much longer than we probably should have on it, we discovered why we couldn't see the raspberry pis in /dev (didn't press enter). With that, we knew our software infrastructure was mostly done.
Day 3 summary continued: After figuring out the software, we thought about how to power the raspberry pis. Our original design assumed a power of 170mA per device, which is what the zeros draw. These raspis will draw significantly more current, so we have to revise our design to power each of the usb hubs. To interface them with our power supply we needed to get wires out of each of the hubs, so I spent a while taking one apart and fixing it.
Day 3 final: We should be ready to finish the electrical tomorrow. We can plug everything into the power supply, maybe with a breadboard as well for the many different wires we're going to need to supply.
We can send files from the master node to any of the other nodes now! It's a miracle! Tomorrow I'll rewrite the system in C to avoid the probably significant overhead of using python as just a command line utility. After that, we need to figure out how to get output back from the nodes, and we can start looking at a final demo.
Working on getting a demo working. Since so many people thought we were doing a bitcoin machine, that's what we ended up doing (though it's not connected to the internet, so it can't actually mine bitcoins).
Finally got the demo working! We ran out of time and didn't get all of the dependencies working on the raspi master, so we're running it off of my laptop for the demonstration. For some reason the login script doesn't quite work, it might have something to do with the proper screen dependencies not being on there. In any case, it works on my laptop and makes for a pretty solid demo.
Done with day 3! We're making good progress, but couldn't have done it without the help of our friends Connor Willison and Shubhangi Bhotika. We're going to start laying out and testing the final circuit tomorrow around 5 in roboclub, please come help if you can!