Cannot modify pwm that controls a loud case fan ... what's the trick?

graysky · 2015-12-21 18:07:27

I have a case fan that runs at full speed despite efforts to slow it in the BIOS. The controller is on the motherboard and should be adjustable. Perhaps a buggy BIOS? I have disabled speed control in the BIOS thinking I can tweak it from Linux. What I don't understand why the fan speed doesn't change when I change the pwm.

From the kernel docs on the nct6775, I need to have the pwm1_enable set to manual mode (1) which I did, but despite me modulating the pwm value, the fan speed does not change. Any thoughts are welcomed.

Here is the freshly booted state:

pwm=/sys/devices/platform/nct6775.2576/hwmon/hwmon2/pwm1
speed=/sys/devices/platform/nct6775.2576/hwmon/hwmon2/fan1_input
ctl=/sys/devices/platform/nct6775.2576/hwmon/hwmon2/pwm1_enable

cat $pwm $speed $ctl
255
1142
5

Enable manual mode and try a slower speed but no change as you see:

echo 1 > $ctl
echo 30 > $pwm

<< I waited 30 sec here to allow the fan to adjust>>

cat $pwm $speed $ctl
30
1142
1

mich41 · 2015-12-21 19:00:03

Are setting made by BIOS visible in Linux and looking sane? This would pretty much rule out software issues, meaning that the fan is simply wired to other channel or maybe directly to power rail.

I guess you can also bother vendor's support a bit. Even if they don't care about Linux, BIOS settings should work.

graysky · 2015-12-21 19:03:21

The controller doesn't respond in the BIOS either... it's full speed 100% of the time. So that means either:
1) The BIOS is buggy.
2) Something is wrong with the connector on the motherboard (to your point).

I don't think #2 is true because the CPU fan does modulate its speed either by the BIOS or by me as described above... the catch is that if I tell it to stop (setting of 0), it doesn't but I can make it go faster than the idle speed.

mich41 · 2015-12-21 19:16:09

Why would CPU fan be relevant? Usually each fan connector is wired to a different channel on the SuperIO. Are you using Y-splitter to power both fans from the same connector?

How did you determine you need to tweak pwm1 and not pwm7, btw?

Last edited by mich41 (2015-12-21 19:29:21)

paulkerry · 2015-12-21 19:27:38

graysky wrote:

The controller doesn't respond in the BIOS either... it's full speed 100% of the time. So that means either:
1) The BIOS is buggy.

That could be it as I have an Asus Q87M-E board which by default in BIOS has the rear case fan on 100% even in BIOS mode and no matter what load is on the system when it is running. I managed to change it to a lower rpm by tinkering with the manual fan settings in BIOS, but still can only get it down to ~40%. Very annoying.

graysky · 2015-12-21 19:37:27

@paulkerry - That's lame indeed. Have you tried setting it to manual as I describe in my first post? Are you able to get it to move this way?

graysky · 2015-12-21 19:42:07

mich41 wrote:

Why would CPU fan be relevant? Usually each fan connector is wired to a different channel on the SuperIO. Are you using Y-splitter to power both fans from the same connector?

No splitter. I only mention the CPU not slowing below 1000 RPM as maybe a sign of a buggy BIOS. What I don't understand is why I cannot override the BIOS by using manual mode as I described. I asked upstream.

mich41 wrote:

How did you determine you need to tweak pwm1 and not pwm7, btw?

The BIOS shows the CPU idle at about 1,000 RPM and the case idle around 1,150 RPM. When I query the full set of pwm files I get similar readings so I concluded as such.

cat /sys/devices/platform/nct6775.2576/hwmon/hwmon2/fan{1..5}_input 
1146
1022
0
0
0

I only have 1-5, nothing higher.

Last edited by graysky (2015-12-21 19:43:10)

mich41 · 2015-12-21 19:58:36

Are the fans 4-pin (PWM)? These have low RPM limit below which they don't go, although 1000RPM would be unusually high for such limit, unless it's some very fast fan.

Or maybe the case fan is a 3-pin connected to 4-pin connector? This doesn't work.

Last edited by mich41 (2015-12-21 20:01:47)

graysky · 2015-12-21 20:59:54

No, it is a 3-pin. I have used these before on other MBs and found that they do in fact alter their RPMs as I change the pwm values. I guess the experiment is to purchase a 120mm PWM (with native 4-pin) fan and see if I can get it to work.

mich41 · 2015-12-21 21:15:08

And the header is 4-pin? Then it has no reason to work. You need to get a 4-pin fan or add some transistor to switch the +12V line together with PWM line or connect +12 to PWM output but only if you find the chip's datasheet and it says this doesn't exceed PWM output's current rating(forget this, no SuperIO I've seen had access to +12 power to drive fans directly).

Last edited by mich41 (2015-12-21 21:28:22)

graysky · 2015-12-21 22:02:34

I will get a 4-pin fan to test it out. Thanks to all for the discussion. I'll update the thread when I get one and verify.

Arch Linux

#1 2015-12-21 18:07:27

Cannot modify pwm that controls a loud case fan ... what's the trick?

#2 2015-12-21 19:00:03

Re: Cannot modify pwm that controls a loud case fan ... what's the trick?

#3 2015-12-21 19:03:21

Re: Cannot modify pwm that controls a loud case fan ... what's the trick?

#4 2015-12-21 19:16:09

Re: Cannot modify pwm that controls a loud case fan ... what's the trick?

#5 2015-12-21 19:27:38

Re: Cannot modify pwm that controls a loud case fan ... what's the trick?

#6 2015-12-21 19:37:27

Re: Cannot modify pwm that controls a loud case fan ... what's the trick?

#7 2015-12-21 19:42:07

Re: Cannot modify pwm that controls a loud case fan ... what's the trick?

#8 2015-12-21 19:58:36

Re: Cannot modify pwm that controls a loud case fan ... what's the trick?

#9 2015-12-21 20:59:54

Re: Cannot modify pwm that controls a loud case fan ... what's the trick?

#10 2015-12-21 21:15:08

Re: Cannot modify pwm that controls a loud case fan ... what's the trick?

#11 2015-12-21 22:02:34

Re: Cannot modify pwm that controls a loud case fan ... what's the trick?

Board footer