[lm-sensors] Testing LM-Sensor Support of SCH5127 in Acer easyStore H340
jrickman at myamigos.us
Mon Dec 13 05:02:13 CET 2010
> Hi Jeff,
> On Sat, 11 Dec 2010 01:02:02 -0600 (CST), Jeff Rickman wrote:
>> > Attached is a modified driver with support for Vtrip monitoring (1.5V,
>> > in7). Could you give it a try and let me know how it goes? The new
>> > driver should expose the following new sysfs files: in7_input,
>> > in7_min, in7_max, in7_alarm. No user-level scaling is required. The
>> > nominal voltage is 1.5V.
>> I successfully compiled the module. Then I removed the old module with
>> "modprobe -r dme1737". Next I copied the new "dme1737.ko" module into
>> proper location in "/lib/modules/`uname -r`/kernel/drivers/hwmon" and
>> the proper permissions with "chmod"; FC14 uses permissions of '744'. Now
>> installed the kernel module with "modprobe -v dme1737" and the Console
>> printed the following:
>> So far so good.....
>> Now I ran "sensors" without any changes to my
>> [root at anas-01 ~]# sensors
>> Adapter: ISA adapter
>> in0: +1.78 V (min = +0.00 V, max = +3.32 V)
>> Vccp: +1.19 V (min = +1.08 V, max = +1.32 V)
>> VCC: +3.31 V (min = +2.97 V, max = +3.63 V)
>> in3: +1.14 V (min = +0.00 V, max = +1.49 V)
>> in4: +1.48 V (min = +0.00 V, max = +1.49 V)
>> VTR: +3.33 V (min = +2.97 V, max = +3.63 V)
>> VBAT: +3.24 V (min = +2.70 V, max = +3.30 V)
>> in7: +0.72 V (min = +1.99 V, max = +0.25 V)
>> Side Fan: 1884 RPM (min = 700 RPM)
>> MCH Fan: 5027 RPM (min = 4500 RPM)
>> CPU Temp: +40.1°C (low = +20.0°C, high = +60.0°C)
>> SYS Temp: +32.9°C (low = +20.0°C, high = +60.0°C)
>> For the record I have "blacklisted" the 'coretemp' module on this
>> as it reports strange temp values (anything from ~9C to ~15C and those
>> too cold) for an Intel Atom 230 CPU (per "/proc/cpuinfo": CPU family 6,
>> model 28, stepping 2) compared to the "CPU Temp" value reported by
>> "in7" is now reported but it will require a "compute (@ * 2), (@ / 2)"
>> line in my "/etc/sensors.d/local.conf" to provide a reasonable value of
>> 1.44: 0.72 * 2 = 1.44, which is close enough to the BIOS reported +1.5V
> Close enough? I wouldn't say that. Which value does the BIOS report
During 1 test run I kept the system "in the BIOS" on the "Hardware
Monitor" screen for almost an hour. The voltages never changed. I suspect
the vendor isn't reporting any measured voltages here; could be "static"
voltage values. As for the "V+1.5" value in the BIOS, it never changed
from "1.50". Never. Only the System Fan Speed (LM_Sensors 'Fan1'), CPU
Temp, and SYS Temp values ever change on this screen.
> BTW, you never clearly reported all exact voltage values reported by the
> BIOS. This would help a lot.
As displayed "in order" from the "Hardware Monitor" screen:
V+1.5 = 1.50 V
5VTR = 4.90 V
VBAT = 3.20 V
V+5 = 5 V
Vccp = 1.20 V
VCC = 3.30 V
VTR = 3.30 V
> If you stay in the BIOS long enough to spot more than one value for
> each voltage, this can help deduce scaling factors.
As I write this email now, none of the voltages ever change, not even by
1/100 (0.01) of a Volt while sitting in the BIOS for almost an hour. I
hardly believe any power supply or system power rails are that stable,
based on experiences with other PC-style computers.
>> Both "in3" and "in4" are "unknowns" to me. I know they will be external
>> voltage inputs to the SCH5127, but it's a guess as to what is being
>> measured. I located a "service guide" for the Acer H340, but it is
>> in this regard. The same judgement applies to "in0". the remaining
>> are marked per the kernel documentation notes written by Juerg.
> I believe that in4 isn't used. It's too close to the ADC's high limit
> to be useful. And the dme1737 driver now shows has one more voltage
> value than your BIOS does, right?
>> Checking the BIOS screen for this machine shows the "V+1.5" reading out
>> steady ~1.5V.
> 1.5 V is quite different from 1.44 V, so either it's not in7, or the
> scaling factor of 2 isn't correct.
1.44V is within +- 10 percent of 1.5V, but I agree that improvement is
needed; changing scaling values in LM_Sensors is a "math problem".
> Anyway, it beats me why the manufacturer would scale down a voltage
> which it could monitor directly. Very odd.
This entire Acer system is "very odd".
Why would a vendor set the system fan to ~760 rpm when HDD temperatures at
that level almost reach 50C and before any serious system load is applied?
50C is a dangerous threshold for most HDD. Thus I use "fancontrol" to
override that automatic setting and push fan speed to "full" to bring HDD
temps down to <=40C. If there is a better method than "fancontrol" that I
can call from "rc.local" (like "echo 'value' > 'someplace in system'"), I
would appreciate it since there is no BIOS control for fan speed at all:
it's based on a "Phoenix TrustedCore" BIOS / Intel Shelton Reference BIOS.
>> Also, a comment for Jean. I pulled the CMOS battery and booted the
>> machine...before making this kernel module change. I did not see any
>> change in the BIOS reported value for "VBAT" and "sensors" did not see
>> reported change; BIOS reported +3.20V and "sensors" reported +3.24V.
>> Perhaps Acer makes that measurement ahead of the CMOS battery on the "+"
>> voltage line? I wonder why. Or perhaps I don't understand what "VBAT" is
>> supposed to measure.
> Vbat is measuring the voltage at the Vbat pin of the chip, which is
> supposed to be connected to the battery.
> Now it is possible that the board designer decided to also connect
> +3.3V or 3VSB there, so that the battery is only used when other power
> sources aren't available, to extend the battery's life. This makes
> sense, but makes monitoring of Vbat useless, of course. This would
> explain the relatively high Vbat value. CR2032 batteries are rated at
> 3.0V, not 3.3V.
>> Finally, for good measure I ran "depmod -a" and rebooted the machine.
>> 'updated' dme1737.ko module is correctly loaded on reboot.
>> Dec 11 00:59:00 anas-01 kernel: [ 24.137243] dme1737 dme1737.2160:
>> a SCH5127 chip at 0x0870
> Strange. The address is different from the one sensors-detect reported
>> Dec 11 00:59:00 anas-01 kernel: [ 24.167411] dme1737 dme1737.2160:
>> Optional features: pwm3=yes, pwm5=no, pwm6=no, fan3=yes, fan4=no,
> Jean Delvare
More information about the lm-sensors