My gaming machine runs very hot, but the case has few fan mounts and poor airflow, meaning I can't push it past about half load before the CPU starts to overheat. I can't afford watercooling or a new case, so I'm solving the problem in true hacker fashion: using salvaged parts and a case mod.
Broken Micromax RF Centrifuge
A refrigerated benchtop centrifuge I found on the scrap pile. It didn't work and the lid latch was stuck shut, so I dismantled it.
A 240V relay I pulled out of the centrifuge.
NMB 5915PC-12T-B20 Fan
A large fan that was used to push air past the centrifuge's radiator. Runs on 115V AC.
My initial design premise has the fan mounted on the outside of the case (which does not have enough space inside to mount something this large), with some basic control circuitry mounted in a box (also attached to the side of the case).
To mount the fan to the case, I'm going to cut the flanges off the fan's original mounting plate, drill holes in the corners, and bolt that to the side of the case, with the fan blowing inwards (to generate positive pressure inside).
The datasheet for the fan lists it as being able to push 166 CFM (which is more than the entire remainder of my cooling setup), but with the caveat that it generates 50 dBa while doing so (and I live in an environment with extremely noise-intolerant people). While using a potentiometer to throttle it is the most logical option, potentiometers capable of generating sufficient resistance on mains AC voltage without exploding are rather on the wrong side of expensive. Toggle switches, on the other hand, are cheap (and look really cool), and I already have a relay that I can use to switch the fan on and off (also salvaged from the centrifuge). I figure that I'll only need to turn this on when I'm REALLY pushing the CPU, so throttling it isn't really necessary.
To ensure the fan can only run when the computer is on, the switch side of the circuit will pull its power from the computer's power supply (either the motherboard fan headers or one of the 4-pin molexes on the PSU). The diagram shows the basic design of the circuitry (there really isn't much to it).
Occasionally, I find interesting bits of scientific apparatus on the junk pile - recently, a laboratory centrifuge turned up there, and I felt compelled to dismantle it (if nothing else, to see what was inside). After verifying that it didn't work, I cracked it open and found a large refrigeration apparatus, and, more alarmingly, some samples left in the chamber (which I delivered to the university's environmental health department to be disposed of properly). Taking the centrifuge apart actually took far longer than I expected, but eventually left me with a very large brushless motor (which powered the centrifuge element), a 240V relay, and a large and powerful case fan that runs on mains-voltage AC. I debated taking the refrigeration loop (or at least parts of it), but decided against it because I wasn't sure of whether it had been purged and didn't want to accidentally poison myself.
Removing the fan gave me the idea for this project - my gaming machine often overheats during intense gaming sessions, and I've been meaning to upgrade the cooling system for quite some time. It's attached to this metal plate, which was originally mounted to the centrifuge's radiator unit. I kept the plate because I feel like it might be useful in mounting it to the case.