So the electronics for this project has ended up being much more of a nightmare than I anticipated
Those familiar with the Arduino will be aware that it's stock PWM frequencies are 490Hz and 970Hz. Running the fans off PWM directly turned out to be extremely noisy, basically to the point where you'd have to raise your voice to have a conversation in the same room. The only other clock prescalers gave ~32Khz and ~62Khz PWM frequencies, which probably due to the motor's inductance, didn't do anything at all.
I decided to build up a couple of beefy RC filters to try and smooth out the PWM - they worked, but because of the resistors they did waste a bit of power, but most of all, they were extremely hard on the power supply (and the caps in the filter itself too). The only thing limiting inrush to the cap was the overall resistance of the wires/FET's etc, and the RC filter resistors. I calculated something like 50A peaks into the caps. One of the caps later decided to vent. Now the screeching noise had moved from the fans to the power supply.. which can't be good for them.
I decided to scrap the RC filters and attempt some LC filters instead. Essentially I was creating a crude PWM-based buck converter. I ripped a few random inductors out of a computer power supply, along with a few fast recovery diodes. In order to get away with a smaller inductor though, I also had to up the PWM frequency to ~32Khz, so I did this for one of the non-critical timers. Surprisingly, without any calculations or anything, it worked
The fans were now quiet, and the PSU too.
However, I now had an EMI-spewing mess, which would constantly reset and crash the microcontroller - bah! I knew at this point a complete rebuild of the controller would be necessary, as the previous design was a total fail from an EMI standpoint.
I decided I'll put all the high current, high speed switching stuff and LC filters in a separate metal box, instead of trying to put it all in the same box as the Arduino.
Also after having to remove the controller from the machine countless times to work on it/program it, I decided to use pluggable screw terminals this time.
I think it came out pretty well, and best of all, it works perfectly, with very minimal heating of the inductors and kickback diode, easily taken care of by a small fan. The FET's now run totally cool
I think it came out pretty well.
The switch/LC filter box:
New controller + driver box connected for a test:
Basically now I'm just waiting on a larger SMPS to install permanently, then I can mount all the electronic and wire it up neatly.
It also turns out shining lasers through it is the most awesome looking thing ever .. it looks so much better IRL than I could get a pic of.