Noctua case fans on Xeon-SP Tower Server

comicnerd

New Member
Oct 4, 2020
3
0
1
Is your problem with the thresholds restting themselves back to default or do your noctua fans actually read below the threshold (or even 0 rpm) at any point?
So the thresholds resetting themselves back to default even I have set thresholds to 500, 800 and 1100 ...
 

Falloutboy

Member
Oct 23, 2011
221
22
18
I don't know what you chaps think of this idea but for many years people have been adding a resistor into the fan wiring to slow them down slightly, the system as far as I am aware is absolutely oblivious to this provided it is not monitoring the RPM's, could this possibly be a solution in this case?
 

Frank Bello

Member
Nov 14, 2018
35
10
8
I don't know what you chaps think of this idea but for many years people have been adding a resistor into the fan wiring to slow them down slightly, the system as far as I am aware is absolutely oblivious to this provided it is not monitoring the RPM's, could this possibly be a solution in this case?
I had this problem and ended up using the Noctua fan controller. NA-FC1
For me, it really was the no-hassle solution. I tried the IPMI approach but every 5 minutes, the Supermicro motherboard put the old fan settings back (fans to 100%). With the Noctua, I was able to get practically silent operation and no fan alarms. Best £18 I spent on my server. It's more expensive than the resistor but it you need to turn your fans up for the summer, it takes less than 60 seconds, you can do it with the server switched on and no messing with soldering irons.
 
  • Like
Reactions: dswartz

EffrafaxOfWug

Radioactive Member
Feb 12, 2015
1,395
501
113
I don't know what you chaps think of this idea but for many years people have been adding a resistor into the fan wiring to slow them down slightly, the system as far as I am aware is absolutely oblivious to this provided it is not monitoring the RPM's, could this possibly be a solution in this case?
Assuming PWM fans, an inline resistor would merely exacerbate the problems. The issue with the Supermicro IPMI is one or two things:
a) fan rpms fall below the IPMI's configured threshold and sets off an assert (which throws all fans to 100%)
b) IPMI forgets the custom thresholds you set, resulting in a) happening

Merely reducing the speed of the fans makes the low rpm situation a) more likely to occur. I've never had b) myself so I don't know how to fix it.

If you use 3-pin fans or non-Noctua fans (which don't seem to drop rpms nearly as much) then the issue probably won't occur, just very annoying for people who have/like Noctua fans and considered a non-issue by Supermicro.
 

dswartz

Active Member
Jul 14, 2011
527
55
28
I had this problem and ended up using the Noctua fan controller. NA-FC1
For me, it really was the no-hassle solution. I tried the IPMI approach but every 5 minutes, the Supermicro motherboard put the old fan settings back (fans to 100%). With the Noctua, I was able to get practically silent operation and no fan alarms. Best £18 I spent on my server. It's more expensive than the resistor but it you need to turn your fans up for the summer, it takes less than 60 seconds, you can do it with the server switched on and no messing with soldering irons.
Same here for a Supermicro JBOD. I ended up plugging all 3 midwall fans into it, so the JBOD only reports 1 fan present :)