My Mirrored Drive Door PowerMac G4 was pretty damn noisy. In fact, back in they they were called “Wind Tunnel G4s” due to the amount of noise they made! They used a huge, heavy duty 120x40mm fan in the case and dual 60x40mm fans (the type you see in 1U servers) in the power supply. At full tilt it sounded like a vacuum cleaner!
Eventually Apple offered a replacement fan and power supply for free, which I guess I ordered but never got around to installing. I recently found it in storage and installed it in my MDD G4; soooo much quieter!
My 386 system has gotten pretty loud, too. It’s a Lunchbox Portable (about the size of a Compaq Portable III) that uses a single fan for cooling, installed in the power supply. By itself it’s not that loud, however I put a 10kRPM Cheetah in for the HDD and decided it really needed extra cooling in such a cramped space, so I picked up a nice Antec Hard Drive Cooler off eBay. It sits in a 5.25” bay and wraps the hard drive in an aluminum heatsink; dual 40mm fans suck air in from faceplate and force air over the heatsink fins. It’s well machined and overall a nice design. There’s even a dual 7-segment display on the faceplate that shows the temperature of the hard drive and another location inside the case (the unit has two temperature probes, one short one that you tape to the bottom of the HDD and a longer one that you can place anywhere).
There’s just one fatal flaw to the design... The HDD temperature probe controls the fan speed. Now, you’d think this would be an advantage, right? The fans stay slow and quite when the drive isn’t very active. That’s what I thought too, and the reason I bought this unit over a slightly cheaper “dumb” drive cooler. The problem is there’s absolutely *no* hysteresis to the speed change and it’s not a smooth, linear speed ramp. Below 30c the fans run at 3000 RPM. At 35c they change to 4500RPM. At 40c they go to 6000RPM. Now, I could live with that, but the lack of hysteresis or any sort of PID algorithm kills it.
What do I mean by that? Well, say you’re sitting at 34.9c and the fans are running at 3000RPM; the temperature jumps up 0.1c to 35c and the fan speed kicks up to 4500RPM, which after a few seconds causes the temperature to go back down to 34.9c and the fans to slow down again. So you end up in a situation where the fans are going like this: Quiet...WHIRRRRRR...quiet....WHIRRRRR..quiet.....WHIRRRRRR...
I understand PID algorithms can be difficult to program into an MCU, but simply adding hysteresis isn’t hard at all! Just make it so that if the fans speed up at 35c they don’t slow down until 34.0c. All it takes is a second lookup table and a few more lines of code. Whoever designed the firmware from this should seriously have their fingers cut off so they can no longer code.
It’s so annoying that I’m designing a replacement PCB to control the fans myself. (And I’m replacing the 40mm fans with quieter 4-wire Sunon MagLev Bearing units, so I can actually control the speed.)
Any sufficiently advanced technology is indistinguishable from magic. (E.g., Cheez Whiz, RF, Hot Dogs)