[lm-sensors] sensors.conf for DFI LP MI P55-T36 (it8720-*)
lmsensors at lklundin.dk
Sun Jun 27 12:35:53 CEST 2010
On Fri, Jun 25, 2010 at 11:17:08AM +0200, Jean Delvare wrote:
> Can you please wait a bit in this screen and try to gather other sample
> values for +12V? If you can gather 2 or 3 different samples, we may be
> able to guess the scaling factor.
> 2.62V in sensors for 11.776V in the BIOS suggests a scaling factor of
> about 4.5. For something more accurate, we need more samples from both
> the BIOS (see above) and the it87 driver. "sensors" only displays 2
> decimal places, so better get the exact reading from sysfs directly:
> cat /sys/class/hwmon/hwmon*/device/in4_input
> If you can gather 2 or 3 samples of each, finding the scaling factor
> should be reasonably easy.
Good idea. I temporarily modified /usr/sbin/fancontrol to log the in4
to a file, so the values would be logged right before and after reboot,
i.e. close in time to my BIOS readout.
The BIOS +12V readout changes in increments of 64mV and the in4 in
increments of 16mV. The ratio between the values is exactly 4.
> > (I wonder if this could have the slightest relevance, but my board is
> > powered by a 200W PW-200-M from MINI-Box).
> Wow, this is cute :) Yes, this probably explains why your voltages are
> rather below the average.
Since my BIOS +12V was quite a bit below 12V, I took the opportunity to
increase the voltage of my external 12V PSU, since this caused the +12V
to increase. The new +12V and in4 voltages also have a ratio of exactly 4.
Unfortunately, the in4 readout seems to indicate that the voltage on the
+12V rail of my PW-200-M drops when the CPU goes from idle to fully
loaded. With a short, thick (load-speaker) cable between the PW-200-M
and the external 12V PSU this drop is 128mV, with a longer, thinner
cable this drop can exceed 600mV. :-(
But now I can at least monitor my +12V via LM sensors. :-)
> We can write the sensor chip drivers (and have done so for many
> already: LM63, LM99 etc.) but the I2C access to these chips is the
> graphics driver's job. Which are improving over time in this respect,
> so it should work someday. I'm not actively working on this though
> (probably due to the lack of nVidia card in my systems.)
OK, thanks for the info.
I have updated
with the complete voltage labeling + scaling.
Many thanks for helping with this,
PS. I have one or two other mainboards. If I happen to find some new
configuration info for them, what is the best way to provide that
to the LM sensors project?
More information about the lm-sensors