[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: powerbook fan behaviour, therm_adt7467

> I've got a question about the fan behaviour on powerbooks using the
> therm_adt7467 module: I've noticed that the fan seems the start correctly
> (when the cpu temperature reaches the given limit), however, it starts to
> spin pretty quickly, causing a lot of noise.

That's adjustable via /sys/devices/temperatures/specified_fan_speed (or
using a module parameter upon loading the module).

> Under MacOS X, the fan spins a lot slower and is much more quiet in general.
> Since the quietness of these powerbooks was one of the things that turned me
> on in the first place, I'm interested in wether somebody managed to
> approximate the fan behaviour under OS X in linux, using the therm_adt7467
> module.

The module simply switches the fan to specified_speed when the threshold
is exceeded, and goes full tilt when the threshold is exceeded by more
than 8 K.

> I'm a bit scared about experimenting with the module parameters because I
> don't want to cause any heat-induced damage by having the fan work to
> inefficiently, yet I need to use the laptop in places were low noise levels
> are mandatory and I'm currently forced to rely on OS X there...

As long as you make sure there's some code to switch the fans to full
speed at some threshold you should be safe. The module needs some more
tweaking anyway - it's written for iBook where there's a CPU and GPU fan
but the PowerBook seems to have just two exhaust fans, and the sensors
don't sit on the CPU and GPU, exactly (the 'GPU' sensor really measures
the power supply temp). For me, the 'GPU' limit needs to be reduced or
both fans need to be synced to identical speed to keep the machine from
crashing too often. Depending on your sensor and fan locations, you might
benefit from similar tweaks (list the fan node in the OF device tree,
mine's at uni-north/i2c/fan).

Making the fan start out slower and gradually increasing the speed would
be just a minor add-on :-)


Reply to: