Fancontrol limit pwm decrease rate patch

[vincent.stehle@…]

Hi,

Here is a proposed patch against fancontrol.pl to limit the pwm decrease rate. This, to limit the user noticing the noise change, while keeping good response to temp. rise. IMHO it does reduce the annoyance at full load.

Thanks for the good work on lm_sensors and fancontrol.

Best regards,

-- Vincent Stehlé, a happy user of lm_sensors/fancontrol.

[khali]

First of all, I'd like to get rid of fancontrol.pl. The bash variant of the script will be much improved in lm-sensors 2.10.4, so there is little reason in doubling the maintenance effort. If you want to work on something, please work on the latest version of fancontrol (the bash script).

Why are you changing the code responsible for restarting the fan? This shouldn't be needed. This functionality is not related to the standard fan speed computation and shouldn't be affected by the regulation policy.

In the leading comment of you patch, you write: "temperature values reported by the sensors can be very noisy". Are they? Mine are quite stable and reliable. If the temperature is not reported properly, then using it to control the fan speed automatically is probably not a good idea to start with.

You also write: "pwm values follow a quadratic reaction". This is true, but I am wondering why it is so. The documentation states: "Currently the speed increases quadratically with rising temperature. This way you won't hear your fans most of the time at best." This doesn't sound like a good reason. If the goal is to not hear the fan most of the time, then a thermostatic (or near-thermostatic) approach is better, and the new version of fancontrol makes this possible. On my system, I've changed the formula to be linear instead of quadratic, and it tends to work better. So if it helps you too, I'd rather just do that than making the script more complex. I'll attach a patch.

[khali]

Use a linear interpolation formula instead of quadratic

[khali]

Patch applied to SVN (changeset r4588).