Re: A marginal problem

Norm Why wrote:

> 
> Thanks Paul,
> 
> These voltages are from HWMonitor:
> 
> Hardware Monitors
> -------------------------------------------------------------------------
> 
> Hardware monitor ITE IT87
> Voltage 0 1.07 Volts [0x43] (CPU VCORE)
> Voltage 1 1.81 Volts [0x71] (DDR)
> Voltage 2 2.98 Volts [0xBA] (+3.3V)
> Voltage 3 4.70 Volts [0xAF] (+5V)
> Voltage 7 12.54 Volts [0xC4] (+12V)
> Voltage 8 3.09 Volts [0xC1] (VBAT)
> 
> All voltages are below expectations except for DDR (12 GB RAM works good) 
> and +12V (there is just one HDD).
> Is there an explanation? Is there mitigation? 

I've had reason to check these myself.

The voltages read low in Windows. Using Speedfan.

    VCore  1.10V
   +12V   11.25V   <=== Likely to be 12V1 and 12V2
   AVCC    3.25V
   Vcore   1.10V
   +3.3V   3.25V
   +5.0V   4.92V
   +12V   11.93V

The voltages read low in Linux. LM-Sensors package, "sensors" command

mint@mint:~$ sensors

atk0110-acpi-0
Adapter: ACPI interface
Vcore Voltage:        1.07 V  (min =  +0.85 V, max =  +1.60 V)
  +3.3 Voltage:        3.25 V  (min =  +2.97 V, max =  +3.63 V)
  +5 Voltage:          4.92 V  (min =  +4.50 V, max =  +5.50 V)
  +12 Voltage:        11.93 V  (min = +10.20 V, max = +13.80 V)
CPU FAN Speed:      1406 RPM  (min =  600 RPM, max = 7200 RPM)
CHASSIS1 FAN Speed:    0 RPM  (min =  600 RPM, max = 7200 RPM)
CHASSIS2 FAN Speed:    0 RPM  (min =  600 RPM, max = 7200 RPM)
POWER FAN Speed:       0 RPM  (min =  600 RPM, max = 7200 RPM)
CPU Temperature:     +42.0°C  (high = +35.0°C, crit = +90.0°C)
MB Temperature:      +30.0°C  (high = +30.0°C, crit = +90.0°C)

A check with a multimeter (3.5 digit) gives

12.04mint@mint:~$ sensors
coretemp-isa-0000
Adapter: ISA adapter
Core 0:       +42.0°C  (high = +78.0°C, crit = +100.0°C)
Core 1:       +35.0°C  (high = +78.0°C, crit = +100.0°C)

nouveau-pci-0200
Adapter: PCI adapter
GPU core:      1.20 V  (min =  +1.20 V, max =  +1.20 V)
temp1:        +42.0°C  (high = +95.0°C, hyst =  +3.0°C)
                        (crit = +130.0°C, hyst =  +2.0°C)
                        (emerg = +135.0°C, hyst =  +5.0°C)

atk0110-acpi-0
Adapter: ACPI interface
Vcore Voltage:        1.07 V  (min =  +0.85 V, max =  +1.60 V)
  +3.3 Voltage:        3.25 V  (min =  +2.97 V, max =  +3.63 V)
  +5 Voltage:          4.92 V  (min =  +4.50 V, max =  +5.50 V)
  +12 Voltage:        11.93 V  (min = +10.20 V, max = +13.80 V)
CPU FAN Speed:      1406 RPM  (min =  600 RPM, max = 7200 RPM)
CHASSIS1 FAN Speed:    0 RPM  (min =  600 RPM, max = 7200 RPM)
CHASSIS2 FAN Speed:    0 RPM  (min =  600 RPM, max = 7200 RPM)
POWER FAN Speed:       0 RPM  (min =  600 RPM, max = 7200 RPM)
CPU Temperature:     +42.0°C  (high = +35.0°C, crit = +90.0°C)
MB Temperature:      +30.0°C  (high = +30.0°C, crit = +90.0°C)

Multimeter, 3.5 digit, Molex disk drive cable

    12.04V  5.06V

And those are very close to the correct values.

Remember that the ADC on the SuperIO making the voltage
measurements, may be an 8 bit ADC with 16mv step size.
That is 256 codes time 16mv or 4.096V max input voltage.
The SuperIO uses scaling resistors, to take the 12V, measure
it as a 16V full scale signal, and get about 3V or so at
the ADC input terminals. The ACPI table in the BIOS is set
up by Asus engineers, to represent the choice of the two
resistors in the voltage divider for 12V.

At one time, people used to check the scaling resistors themselves,
and report this to the author of MBM5 or similar. Then later, the
process was automated. If an Asus engineer makes a mistake, then
the voltage you read out could be permanently low-ball. That's one
of the dangers of the factory doing it. The resistors could even have
a 5% tolerance, for all we know. Who would be buying half-percent
(very accurate) resistors for such a task ?

It would be great if there was a "calibration procedure" during the
two-minute functional test, but that stretches credulity. There is
a lot to do in two minutes, such as make sure the PCI Express
slots work, and little time for frippery. At one time, the Asus
factory used to make five million motherboards per month. It's
less now. That means, every day, a shitload of motherboards have
to go through the two-minute functional test.

    Paul