I just replaced my stock CPU cooler with a Noctua NH-D15 and I've noticed some strange behavior. On booting up, both of the CPU fans (I'm using the Y adapter and the low-noise adapters) spin. They continue to spin after signing into an Antergos Linux session. When I check back later after using the computer for a while, the outer fan has stopped spinning (have observed this every time I've checked so far) and once I even saw that both fans had stopped. Putting a load on the CPU to bring the temperatures up did not start the fans (even with both fans off, this cooler keeps my temps at around 30 celsius under light load.)
Does anyone have some recommendations for how to diagnose whether this is a hardware or software problem? I've got xsensors set up, and it reports 4 fans with 2 of them at 0 RPM (I'm not sure which fan is which, but right now one of my CPU fans and both of my case fans are running.)
Yeah get rid of either the low noise or y adapters. Fans when the PC boots run full speed and thus you have a little momentum to keep them going at idle but they are constantly slowing down bc they don't have enough voltage. So after awhile the fans stop and the voltage isnt enough to overcome the friction. let me guess the one that isnt running is twitching right? if you spin it with your finger i bet it run for awhile
I tried moving the stopped one with my finger and it did nothing. The one that's running seems to be smooth. I'll try changing the cables around later if I can get everything to reach (the fan plugs are in slightly awkward places.)
Hmm, yeah, maybe when the intended RPM is very low, it's not enough to power both and the Y adapter favors one fan somehow? I know very little about electric motors etc. from a physics standpoint. I'll definitely try removing the Y adapter next boot-up.
I amuse these are 3 pin fans? if so go in to the BIOS and set the fam control mode to voltage from auto. Then ajust the fan slope to keep them from stopping. and dont use the low-noise adapters.
So, I tried to remove the Y adapter and connecting both fans to the motherboard directly, but the cables are nowhere near long enough (the only other 4-pin connector on the board is on the opposite corner of the board, so I would need something like a 1-foot extension to reach it.) It looks like every fan is connected to the only header it will reach. I'll continue to poke around with pwmconfig and see if I can learn anything from that angle.
So when the pc boots it spins the fans at 100% full speed but then backs off. (I've noticed the same thing with my D15 and fans) At around 700 rpm one of the fans on my y-spliter doesn't have enough power, and it stops and will twitch slightly. The other fan will continue to spin. I just bumped the pwm percentile to 40% (no speed reducers) and that results in both fans spinning in the low to mid 700's. Go any lower for me and one fan turns off. If I turn the speed to 35% I see the exact same behavior your describing, fan spins up when the pc starts and then turns off when the fan control kicks in and lowers the speed down to 35% or lower.
The second fan at 0 rpm is probably just an unoccupied fan header :)
I would suggest trying to do fan control within Linux. I have the updated version of that motherboard, the 990FXA-UD3 Rev. 4, and it has terrible fan control from within the bios. Your going to have to find some kind of fan control software from within linux, and frankly good luck with that. I looked into it a bit a while ago and it seemed like it was a nightmare to find something that is equivalent of SpeedFan in windows, but in Linux.
I saw other people talking about having the same problem specifically with this cooler and Gigabyte motherboards. I've removed the low noise adapters, but retained the y-adapter, and everything seems fine for now. I'll let it run for a while.
You're right. There's one unused header which none of my cables can reach. That must be the one :)
I hope I don't have to do that, after removing the low noise adapters (even with those removed, the Noctua fans are the quietest ones in my case.) I played around with fancontrol/pwmconfig but I'm pretty sure I don't know what I'm doing. I was able to identify that one of the controllers was for my GPU (but the software didn't seem to be able to correlate it with an RPM response.) The other two controllers that the software tested both controlled the CPU fans (not sure why.) I tried saving a config, but wound up with some fans going 100% - if I have trouble with the motherboard fan controls, I'll look back into an OS software solution.
Unless the motherboard fan control has been updated, your not going to want to use. I have my 990FXA-UD3 updated to the newest bios and all the fan control I have found is to stick the fans at one fixed speed, which I personally find worse than setting up a fan speed curve to ramp up and down with the system temperatures. There are a couple presets, one of them being a silent mode, but all of them are pretty loud if I remember right. I currently have a 212 evo on that system with 2 low noise adapters installed to keep it reasonable, albeit still the loudest system I run.
Surely power is too low for those fans that's why when the system is idling one fan is not spinning. I have an NH-D15 and even with the fans 100% I can't tell if the system is stressing the CPU or not. Go into the UEFI and bump up the lowest setting for those fans untill the system when idling has both fans running. Mine are idling around 600rpms, if that's something that might help.