[lm-sensors] [PATCH v4 00/10] hwmon: (w83627ehf) Add support for NCT6775F and NCT6776F

Phillip Susi psusi at cfl.rr.com
Wed Feb 23 06:00:57 CET 2011

On 02/22/2011 09:46 PM, Guenter Roeck wrote:

> How many fan counters (fanX_input) do you see on your system ?

The device reports 5 fan inputs and 3 pwm outputs, but the motherboard 
only has 4 fan headers: 2 4 pin and 2 3 pin.  From that I would expect 
pwm1 and pwm2 to be routed to the two 4 pin fan headers ( first one is 
cpu ), and the other two fan headers input to be connected to fan inputs 
3 and 4, with no control.

> I am a bit confused about the correlation of pwm to inputs and fan behavior.
> Are you saying that pwm1 affects three of the four fans ? That would be a bit odd.
> What are the values repoted by pwm2 and pwm3, and do you see anything change
> if you modify those values ?

Indeed, it is odd.  If I set pwm1 to a value of zero, all but one of the 
fans in the system stop.  Specifically the cpu fan, the front case fan, 
and the rear case fan stop, while the top case fan continues to spin.  I 
will need to open it back up to make sure, but my guess is that the top 
fan is connected to the second 4 pin fan header on the motherboard, but 
since the fan is only a 3 pin fan, it does not respond to the pwm output 
that is likely connected to it from pwm2.  The other two fan headers are 
only 3 pin headers, but since they seem to respond to changes to pwm1, I 
guess that the motherboard has on board FETs that respond to PWM1 to 
regulate the voltage on these two headers so their speed corresponds to 
the cpu fan speed.

> Would it be possible to run pwmconfig and let us know what it reports ?

It appears to agree with my assessment in that fan1 responds to pwm1.

More information about the lm-sensors mailing list