Nice write-up. Thank you for sharing!
Had to run some errands earlier and was inspired by your post to also capture my PSU noise intervals:
Tried some post-processing to bring it out a bit more. It’s hard to recreate the subjective real-life experience. Silence sounds really noisy in the recording which it does not in real life - whereas the noisy fan intervals are actually much more annoying in real life compared to the recording.
Audio recording duration ~38 min (low quality, highly compressed mp3)
- 5 noise cycles total
- first time PSU fan kicks in at ~ 5 seconds
- never timed the intervals and had guessed its a minute or two every 10-15 minutes
- was surprised to find out in reality it’s 4 to 5 minutes of noise every 4 to 5 minutes
As I said - was running errands and had maybe Spotify and Vivaldi open but nothing playing and nobody actively using the desktop during the whole duration.
Room temperature around 22 °C on a wooden table top. Side panel open to expose the noctua fan. Central European power grid.
It does not seem to be abnormal noise - the noise IMO is expected. It’s the very short intervals and long durations. Somehow the PSU seems to run hot (not ultra hot, but enough for the fan to kick in) even with hardly any load. Noctua fan kicks in too sometimes, but is more silent than the PSU fan and shorter duration than the PSU fan. Easy to observe it visually due to it sitting right in front of me on the table and the open chassis.
Hope this helps others to compare and get some insights into what we are moaning about 
Edit: Forgot to say - also no ramp-up / ramp-down period. Straight full on or full off. Saw other posts mentioning a deep-dive article that Framework published that the PSU fan should ramp up slowly.