* Re: [lm-sensors] pwmconfig/fancontrol error (version 3.3.1)
2011-12-03 13:30 [lm-sensors] pwmconfig/fancontrol error (version 3.3.1) Charles
@ 2011-12-03 18:00 ` Jean Delvare
2011-12-03 18:15 ` Guenter Roeck
` (3 subsequent siblings)
4 siblings, 0 replies; 6+ messages in thread
From: Jean Delvare @ 2011-12-03 18:00 UTC (permalink / raw)
To: lm-sensors
Hi Charles,
On Sat, 03 Dec 2011 18:48:25 +0530, Charles wrote:
> A lmsensors3.conf generated by pwmconfig resulted in fancontrol message
> "Error: file hwmon0/device/fan4_input+hwmon0/device/fan1_input doesn't
> exist".
>
> The motherboard was an ASUS P8H67-V fitted with an S-Flex SFF21E case
> exhaust fan (3 wire) connected to CHA_FAN2 and a stock Intel Core i3
> processor cooler fan (4 wire) connected to CHA_FAN.
>
> pwmconfig detected that the Nuvoton NCT6776F controlled both CHA_FAN and
> CHA_FAN2 via pwm1 (pwm2 controls CPU_FAN; pwm3 seems not to be used).
> The sensors3.conf generated by pwmconfig included this line:
>
> FCFANS= hwmon0/device/pwm1=hwmon0/device/fan4_input+hwmon0/device/fan1_input
>
> The space after FCFANS= is irregular but did not seem to cause any
> problems for fancontrol.
This is a known issue. pwmconfig can detect if your board has several
fans connected to the same fan control output, but fancontrol can't
handle that yet. There's even this comment in pwmconfig:
else
fanactive="$j+${fanactive}" #not supported yet by fancontrol
fanactive_min="$S+${fanactive_min}"
fi
Workaround for the time being is to manually remove one of the two
fanX_input from the configuration statement. Until fancontrol knows
about the syntax. If you are familiar with bash shell, you could even
give it a try. Beware that fancontrol isn't the most friendly piece of
code out there though.
>
> Both hwmon0/device/fan4_input and hwmon0/device/fan1_input do exist:
>
> root@CW8:~# ls -l /sys/devices/platform/w83627ehf.656/fan[14]_input
> -r--r--r-- 1 root root 4096 2011-12-03 18:44
> /sys/devices/platform/w83627ehf.656/fan1_input
> -r--r--r-- 1 root root 4096 2011-12-03 18:26
> /sys/devices/platform/w83627ehf.656/fan4_inpu
>
> Further information: the processor cooler fan was connected to CHA_FAN2
> instead of CPU_FAN because pwmconfig could not slow it below 1,100 RPM
> when it was on CPU_FAN.
This probably explains the behaviour. It is common from board makers to
use a single PWM output to control all chassis fans. As you use one of
the chassis fan headers for the CPU fan, I'm not surprised that you end
up with both fans controlled by the some PWM output. A workaround would
be to connect the case fan to the CPU fan header.
--
Jean Delvare
_______________________________________________
lm-sensors mailing list
lm-sensors@lm-sensors.org
http://lists.lm-sensors.org/mailman/listinfo/lm-sensors
^ permalink raw reply [flat|nested] 6+ messages in thread* Re: [lm-sensors] pwmconfig/fancontrol error (version 3.3.1)
2011-12-03 13:30 [lm-sensors] pwmconfig/fancontrol error (version 3.3.1) Charles
2011-12-03 18:00 ` Jean Delvare
@ 2011-12-03 18:15 ` Guenter Roeck
2011-12-03 20:14 ` Jean Delvare
` (2 subsequent siblings)
4 siblings, 0 replies; 6+ messages in thread
From: Guenter Roeck @ 2011-12-03 18:15 UTC (permalink / raw)
To: lm-sensors
Hi,
On Sat, Dec 03, 2011 at 08:18:25AM -0500, Charles wrote:
> Hello :-)
>
> A lmsensors3.conf generated by pwmconfig resulted in fancontrol message
> "Error: file hwmon0/device/fan4_input+hwmon0/device/fan1_input doesn't
> exist".
>
Looking into pwmconfig, it seems the above is supposed to mean that pwm1 controls
both fan1_input and fan4_input. There is also a comment in pwmconfig suggesting
that fancontrol does not support this configuration.
> The motherboard was an ASUS P8H67-V fitted with an S-Flex SFF21E case
> exhaust fan (3 wire) connected to CHA_FAN2 and a stock Intel Core i3
> processor cooler fan (4 wire) connected to CHA_FAN.
>
> pwmconfig detected that the Nuvoton NCT6776F controlled both CHA_FAN and
> CHA_FAN2 via pwm1 (pwm2 controls CPU_FAN; pwm3 seems not to be used).
> The sensors3.conf generated by pwmconfig included this line:
>
> FCFANS= hwmon0/device/pwm1=hwmon0/device/fan4_input+hwmon0/device/fan1_input
Try replacing this with
FCFANS="hwmon0/device/pwm1=hwmon0/device/fan1_input"
or
FCFANS="hwmon0/device/pwm1=hwmon0/device/fan4_input"
Guenter
>
> The space after FCFANS= is irregular but did not seem to cause any
> problems for fancontrol.
>
> Both hwmon0/device/fan4_input and hwmon0/device/fan1_input do exist:
>
> root@CW8:~# ls -l /sys/devices/platform/w83627ehf.656/fan[14]_input
> -r--r--r-- 1 root root 4096 2011-12-03 18:44
> /sys/devices/platform/w83627ehf.656/fan1_input
> -r--r--r-- 1 root root 4096 2011-12-03 18:26
> /sys/devices/platform/w83627ehf.656/fan4_inpu
>
> Further information: the processor cooler fan was connected to CHA_FAN2
> instead of CPU_FAN because pwmconfig could not slow it below 1,100 RPM
> when it was on CPU_FAN. Here's the pwmconfig output with the processor
> cooler fan connected to CPU_FAN:
>
> PWM 255 FAN 2284
> PWM 240 FAN 2112
> PWM 225 FAN 2002
> PWM 210 FAN 1939
> PWM 195 FAN 1872
> PWM 180 FAN 1795
> PWM 165 FAN 1730
> PWM 150 FAN 1674
> PWM 135 FAN 1603
> PWM 120 FAN 1523
> PWM 105 FAN 1456
> PWM 90 FAN 1390
> PWM 75 FAN 1317
> PWM 60 FAN 1231
> PWM 45 FAN 1161
> PWM 30 FAN 1105
> PWM 28 FAN 1110
> PWM 26 FAN 1103
> PWM 24 FAN 1102
> PWM 22 FAN 1104
> PWM 20 FAN 1103
> PWM 18 FAN 1104
> PWM 16 FAN 1102
> PWM 14 FAN 1108
> PWM 12 FAN 1110
> PWM 10 FAN 1111
> PWM 8 FAN 1110
> PWM 6 FAN 1110
> PWM 4 FAN 1111
> PWM 2 FAN 1112
> PWM 0 FAN 1112
>
> Best
>
> Charles
>
> _______________________________________________
> lm-sensors mailing list
> lm-sensors@lm-sensors.org
> http://lists.lm-sensors.org/mailman/listinfo/lm-sensors
_______________________________________________
lm-sensors mailing list
lm-sensors@lm-sensors.org
http://lists.lm-sensors.org/mailman/listinfo/lm-sensors
^ permalink raw reply [flat|nested] 6+ messages in thread* Re: [lm-sensors] pwmconfig/fancontrol error (version 3.3.1)
2011-12-03 13:30 [lm-sensors] pwmconfig/fancontrol error (version 3.3.1) Charles
2011-12-03 18:00 ` Jean Delvare
2011-12-03 18:15 ` Guenter Roeck
@ 2011-12-03 20:14 ` Jean Delvare
2011-12-04 10:03 ` Jean Delvare
2011-12-04 12:55 ` Charles
4 siblings, 0 replies; 6+ messages in thread
From: Jean Delvare @ 2011-12-03 20:14 UTC (permalink / raw)
To: lm-sensors
On Sat, 3 Dec 2011 19:00:58 +0100, Jean Delvare wrote:
> Hi Charles,
>
> On Sat, 03 Dec 2011 18:48:25 +0530, Charles wrote:
> > A lmsensors3.conf generated by pwmconfig resulted in fancontrol message
> > "Error: file hwmon0/device/fan4_input+hwmon0/device/fan1_input doesn't
> > exist".
> >
> > The motherboard was an ASUS P8H67-V fitted with an S-Flex SFF21E case
> > exhaust fan (3 wire) connected to CHA_FAN2 and a stock Intel Core i3
> > processor cooler fan (4 wire) connected to CHA_FAN.
> >
> > pwmconfig detected that the Nuvoton NCT6776F controlled both CHA_FAN and
> > CHA_FAN2 via pwm1 (pwm2 controls CPU_FAN; pwm3 seems not to be used).
> > The sensors3.conf generated by pwmconfig included this line:
> >
> > FCFANS= hwmon0/device/pwm1=hwmon0/device/fan4_input+hwmon0/device/fan1_input
> >
> > The space after FCFANS= is irregular but did not seem to cause any
> > problems for fancontrol.
>
> This is a known issue. pwmconfig can detect if your board has several
> fans connected to the same fan control output, but fancontrol can't
> handle that yet. There's even this comment in pwmconfig:
>
> else
> fanactive="$j+${fanactive}" #not supported yet by fancontrol
> fanactive_min="$S+${fanactive_min}"
> fi
>
> Workaround for the time being is to manually remove one of the two
> fanX_input from the configuration statement. Until fancontrol knows
> about the syntax. If you are familiar with bash shell, you could even
> give it a try. Beware that fancontrol isn't the most friendly piece of
> code out there though.
I finally gave it a try myself. Please fetch:
http://khali.linux-fr.org/devel/misc/fancontrol
and give it a try, it should work OK with multiple fans controlled by
one PWM output.
--
Jean Delvare
_______________________________________________
lm-sensors mailing list
lm-sensors@lm-sensors.org
http://lists.lm-sensors.org/mailman/listinfo/lm-sensors
^ permalink raw reply [flat|nested] 6+ messages in thread* Re: [lm-sensors] pwmconfig/fancontrol error (version 3.3.1)
2011-12-03 13:30 [lm-sensors] pwmconfig/fancontrol error (version 3.3.1) Charles
` (2 preceding siblings ...)
2011-12-03 20:14 ` Jean Delvare
@ 2011-12-04 10:03 ` Jean Delvare
2011-12-04 12:55 ` Charles
4 siblings, 0 replies; 6+ messages in thread
From: Jean Delvare @ 2011-12-04 10:03 UTC (permalink / raw)
To: lm-sensors
Hi Charles,
Please keep the list CC'd.
On Sun, 04 Dec 2011 12:29:34 +0530, Charles wrote:
> On 04/12/11 01:44, Jean Delvare wrote:
> > I finally gave it a try myself. Please fetch:
> > http://khali.linux-fr.org/devel/misc/fancontrol
> > and give it a try, it should work OK with multiple fans controlled by
> > one PWM output.
>
> Thanks Jean ::)
>
> Tested; no error messages and GKrellM shows sane processor core
> temperatures and fan speeds.
>
> Sorry for the incomplete initial report:
> * I had applied the suggested workaround (thanks Jean and Guenter).
Hmm. Please test the updated version of the script _without_ the
workaround applied. That's the whole point of the update.
> * Connecting the case exahust fan to CPU_FAN is not an option because
> pwmconfig does not control it then, although it can control the stock
> Intel fan on the same header (subject to the ~1100 RPM minimum).
Most likely because the CPU fan header is a 4-pin header, and the Intel
CPU fan has 4 wires, while the case fan headers are 3-pin with a 3-wire
fan.
On 4-pin headers, the PWM signal is separated from the power. So you
can't control a 3-wire fan plugged in a 4-pin header (the PWM signal is
on the unconnected pin [1].) This also explains the minimum speed of
4-wire PWM-controlled fans. This is per design. If the minimum speed
doesn't please you, you have to either buy a different fan with a lower
minimum, or lower the fan voltage using a serial resistor (Noctua
amongst several other fan vendors includes such resistors in many of
their product packages.)
[1] Some boards have jumpers that will let you turn a 4-pin fan header
into a 3-pin one. You loose the benefits of 4-wire fans (such as linear
fan speed control and reliable readings at low speed) but are no longer
affected by the arbitrary fan minimum then.
> * PWR_FAN is not speed-controlled.
>
> The present situation is not satisfactory (not an lm_sensors issue,
> reporting in case someone in a similar situation finds this) because the
> processor and case fans have very different signal/speed characteristics
> so, for example, the processor fan runs at 1400 while the case runs at
> 700 and the case exhaust air is warmer than ideal.
Indeed, it is only recommended to connect similar fans to the same
control output. There's a reason why board makers often connect the
case fans together, but usually have a separate fan control for the
CPU(s).
--
Jean Delvare
_______________________________________________
lm-sensors mailing list
lm-sensors@lm-sensors.org
http://lists.lm-sensors.org/mailman/listinfo/lm-sensors
^ permalink raw reply [flat|nested] 6+ messages in thread* Re: [lm-sensors] pwmconfig/fancontrol error (version 3.3.1)
2011-12-03 13:30 [lm-sensors] pwmconfig/fancontrol error (version 3.3.1) Charles
` (3 preceding siblings ...)
2011-12-04 10:03 ` Jean Delvare
@ 2011-12-04 12:55 ` Charles
4 siblings, 0 replies; 6+ messages in thread
From: Charles @ 2011-12-04 12:55 UTC (permalink / raw)
To: lm-sensors
On 04/12/11 16:30, lm-sensors-request@lm-sensors.org wrote:
> Hi Charles,
>
> Please keep the list CC'd.
Oops! Sorry.
>> > Tested; no error messages and GKrellM shows sane processor core
>> > temperatures and fan speeds.
>> >
>> > Sorry for the incomplete initial report:
>> > * I had applied the suggested workaround (thanks Jean and Guenter).
> Hmm. Please test the updated version of the script _without_ the
> workaround applied. That's the whole point of the update.
That's how I did test the update. Should have made clear.
> Most likely because the CPU fan header is a 4-pin header, and the Intel
> CPU fan has 4 wires, while the case fan headers are 3-pin with a 3-wire
> fan.
>
> On 4-pin headers, the PWM signal is separated from the power. So you
> can't control a 3-wire fan plugged in a 4-pin header (the PWM signal is
> on the unconnected pin [1].) This also explains the minimum speed of
> 4-wire PWM-controlled fans. This is per design. If the minimum speed
> doesn't please you, you have to either buy a different fan with a lower
> minimum, or lower the fan voltage using a serial resistor (Noctua
> amongst several other fan vendors includes such resistors in many of
> their product packages.)
>
> [1] Some boards have jumpers that will let you turn a 4-pin fan header
> into a 3-pin one. You loose the benefits of 4-wire fans (such as linear
> fan speed control and reliable readings at low speed) but are no longer
> affected by the arbitrary fan minimum then.
>
> Indeed, it is only recommended to connect similar fans to the same
> control output. There's a reason why board makers often connect the
> case fans together, but usually have a separate fan control for the
> CPU(s).
Thanks for explanations, Jean :)
_______________________________________________
lm-sensors mailing list
lm-sensors@lm-sensors.org
http://lists.lm-sensors.org/mailman/listinfo/lm-sensors
^ permalink raw reply [flat|nested] 6+ messages in thread