I currently have reduced the electrical part of these Dell AA23300 PSUs to this quick & dirty hack; it turns the PSU on and sets the output voltage to a nice stable 13.85-13.90V at idle. These will do up to ~14.1V before they trip OVP; this gives a little leeway to ensure stability.
I've tried multiple different attacks to slow down the idle RPM, but anything that makes any appreciable difference in the noise level causes it to go into shutdown or start doing a POST loop.
Currently exploring mechanical means of noise reduction.
mnem
I haven't seen one of these units before... is the fan just plain noisy or is it being driven at higher than necessary RPM?
My first idea was to swap the fan for a quieter one. Google says it has a proprietary connector, but in theory it should not be hard to bodge the connector onto another fan. Of course, that would go right out the window if the fan control is also proprietary.
Do you think the PSU is shutting down or POST looping since it is failing BIT with the slow fan speed, or could it actually be going over-temp with not enough air flow? If the PSU is being over-protective and failing BIT, then how about scaling the RPM signal with a counter/divider? We did something similar once for a rally car. Guy bought a rally computer which expected 4 or 5 pulses per wheel revolution. On his particular car, the wheel sensor he finally rigged up would only pulse once per revolution. So, the signal was out of the range of calibration. A 4x multiplier was used to get enough pulses to make the computer happy. Accuracy was only equivalent to wheel circumference anyways which was still good enough
normally (but, that is another story). Google says the fan RPM is 3.3V signal, so that might be possible.