Re: [SLUG] lm_sensors

From: Eben King (eben01@verizon.net)
Date: Fri Jul 13 2007 - 23:55:39 EDT


On Fri, 13 Jul 2007, Ian C. Blenke wrote:

> Levi Bard wrote:
>
>> If you google for your motherboard's model (or post it to the list),
>> it's very likely that somebody has already created a sensors.conf
>> specifically for it. In fact, Tyan distributes (or did for the S2875
>> at least) sensors.conf files that /they've/ created for their
>> motherboards.
>
>
> What you will find is that not only do the actual temperature sensors change
> from motherboard to motherboard, but also from revision to revision.
> Manufacturers do this to cut cost (and boost profit) depending on the
> supplier that has the best price for them.
>
> Also, no two sensors are the same. Each sensor is manufactured within a
> certain tolerance range, but the readings WILL DIFFER from sensor to sensor.
> This means that you WILL need to alter any sensor.conf file you find with
> adjustments that fit your specific sensors.

Unless you know what the _actual_ fan speed/temperature/voltage is, you have
no hope of calibrating the sensor. And you need two points if it's not off
by an offset. And more, if it might not be linear.

So I said "screw all that", normalized the sensor readings to what it
_should_ be, then watch for deviations from that. I wrote some cron / shell
/ gnuplot scripts that make 1-week graphs of related readings, so I can spot
trends easier. True, when it says "40C" it might not actually be reading
40C, but that's not as important as "is it rising / falling / twitching"
which the graphs show accurately.

It's pretty cool. I can see that parts of the machine get hotter when I
open the side cover (mainboard, I think); and that all temps vary in
lockstep when the house switches from cooling to heating (for a few days the
temperature "floats").

-- 
-eben                                      QebWenE01R@vTerYizUonI.nOetP
1101000 1110100 1110100 1110000 0111010 0101111 0101111 1110010 1101111
1111001 1100001 1101100 1110100 1111001 0101110 1101101 1101001 1101110
1100101 0101110 1101110 1110101 0111010 0111000 0110001 0101111
-----------------------------------------------------------------------
This list is provided as an unmoderated internet service by Networked
Knowledge Systems (NKS).  Views and opinions expressed in messages
posted are those of the author and do not necessarily reflect the
official policy or position of NKS or any of its employees.



This archive was generated by hypermail 2.1.3 : Fri Aug 01 2014 - 17:59:22 EDT