Acceptable mobo voltages?.

[Tom McCreadie]

Perhaps off topic for this NG: Below is a report by the utility 'PC Wizard' of
the motherboard voltages for my Asus desktop PC (P4C800-E del / Antec
TruePower480 PS) and my core 2 Duo laptop (Toshiba Satellite A110-323). This was
measured with no major apps running. The desktop PC was loaded with 2 internal
HD's, a Matrox P650 video card and 4 PCI cards.

The Asus desktop data seem solid enough but the low laptop values surprised me.
(Switching from mains to battery operation made no change). Should I be
concerned about those low laptop voltages? It works fine. though I don't have
any earlier measurement history to reveal any possible trend Or are such
voltages par for the course for laptops in general, and within tolerances
anyway?

Thanks for any insight,
Tom

Toshiba Asus P4C800-E del/ TruePower480
Voltage CPU 2.05 V 1.41 V
+3.3 Voltage 2.05 V 3.09 V
+5 Voltage 3.44 V 4.81 V
+12V Voltage 8.19 V 11.86 V

Battery #0 11.10 V n/a

 

[Paul]

Perhaps off topic for this NG: Below is a report by the utility 'PC Wizard' of
the motherboard voltages for my Asus desktop PC (P4C800-E del / Antec
TruePower480 PS) and my core 2 Duo laptop (Toshiba Satellite A110-323). This was
measured with no major apps running. The desktop PC was loaded with 2 internal
HD's, a Matrox P650 video card and 4 PCI cards.

The Asus desktop data seem solid enough but the low laptop values surprised me.
(Switching from mains to battery operation made no change). Should I be
concerned about those low laptop voltages? It works fine. though I don't have
any earlier measurement history to reveal any possible trend Or are such
voltages par for the course for laptops in general, and within tolerances
anyway?

Thanks for any insight,
Tom

Toshiba Asus P4C800-E del/ TruePower480
Voltage CPU 2.05 V 1.41 V
+3.3 Voltage 2.05 V 3.09 V
+5 Voltage 3.44 V 4.81 V
+12V Voltage 8.19 V 11.86 V

Battery #0 11.10 V n/a

Those measurements are made by the hardware monitor interface on
the SuperI/O chip.

There are a number of identical measurement channels. The channels
have a certain full scale value, such as 4.096 volts measured in
256 steps (8 bit ADC). By applying smoothing or averaging
functions to the stream of readout values, you end up with
values that are not a multiple of 0.016 volts.

Voltages higher than the full scale value, must be run through a
voltage divider, so that they fit within the measurement range.
The resistors used for that, tend to degrade the accuracy of
the measured result (to a point worse than a cheap multimeter).

Take the 12V rail as an example. If a voltage divider is set up,
to take 1/4 of the measurement lead, then a measured value of 3V
on that channel, would correspond to 12V. Somehow, the software
needs to find out, that it needs to multiply the measured value by 4
to get the right answer. The extra "headroom" in my example, allows
the hardware monitor to read a badly out of spec 15V value, and
stay within the dynamic range of the 4.096V ADC conversion (15/4
volts applied to the converter input).

If we take all your Toshiba values, they appear to be about 2/3rds
of the correct value. It looks like some scaling math isn't working
right.

As a double check, you might want to look up what CPU your Toshiba
uses, to see what measured value for VCore would make sense.

On your P4C800-E, it's hard to say what is happening there. My
recollection is, some of the Antec Truepower supplies, have
feedback wires on all three major rails, so the Truepower should
be able to better regulate to the correct value. (If your
Truepower was "sick", I suppose anything is possible.)

Any time you have doubts, a multimeter set to the volts range,
can be used to verify some of the power supply voltages. Knowing
where to probe, to get the CPU voltage, is a bit more difficult.
(The best place to measure, is likely two pins on the CPU designed
for this purpose. The pins monitor the supply voltage as seen
by the CPU silicon die itself, for the most accurate result. Those
pins may be what is connected to the hardware monitor, to give the 1.41V
reading.)

An early P4 would have been in the 1.5V range or so, so the
value shown isn't alarmingly different.

For the Toshiba battery voltage, you can try taking 11.1 and multiplying
it by the suspected missing scale factor of 1.5 . That gives 16.65 volts.
You can compare that to the nominal value printed on the battery pack.

My experience, of checking a few different motherboards, usually
shows one rail measurement is out to lunch, which probably means
the scaling math in the program displaying the results, is just wrong.
Checking with my multimeter, shows the power supply values to
be pretty close to the correct values, and the multimeter measurements
usually make the power supply look heroic, while the hardware monitor
readings usually make the power supply look like a slouch (out of
spec).

On occasion, someone will report they're getting a 0.0 reading
on a major rail, and you just know that can't be right The
computer would be "black screen", if it was really zero.

Paul

 

[Tom McCreadie]

Those measurements are made by the hardware monitor interface on
the SuperI/O chip.

There are a number of identical measurement channels. The channels
have a certain full scale value, such as 4.096 volts measured in
256 steps (8 bit ADC). By applying smoothing or averaging
functions to the stream of readout values, you end up with
values that are not a multiple of 0.016 volts.

Voltages higher than the full scale value, must be run through a
voltage divider, so that they fit within the measurement range.
The resistors used for that, tend to degrade the accuracy of
the measured result (to a point worse than a cheap multimeter).

Take the 12V rail as an example. If a voltage divider is set up,
to take 1/4 of the measurement lead, then a measured value of 3V
on that channel, would correspond to 12V. Somehow, the software
needs to find out, that it needs to multiply the measured value by 4
to get the right answer. The extra "headroom" in my example, allows
the hardware monitor to read a badly out of spec 15V value, and
stay within the dynamic range of the 4.096V ADC conversion (15/4
volts applied to the converter input).

If we take all your Toshiba values, they appear to be about 2/3rds
of the correct value. It looks like some scaling math isn't working
right.

As a double check, you might want to look up what CPU your Toshiba
uses, to see what measured value for VCore would make sense.

On your P4C800-E, it's hard to say what is happening there. My
recollection is, some of the Antec Truepower supplies, have
feedback wires on all three major rails, so the Truepower should
be able to better regulate to the correct value. (If your
Truepower was "sick", I suppose anything is possible.)

Any time you have doubts, a multimeter set to the volts range,
can be used to verify some of the power supply voltages. Knowing
where to probe, to get the CPU voltage, is a bit more difficult.
(The best place to measure, is likely two pins on the CPU designed
for this purpose. The pins monitor the supply voltage as seen
by the CPU silicon die itself, for the most accurate result. Those
pins may be what is connected to the hardware monitor, to give the 1.41V
reading.)

An early P4 would have been in the 1.5V range or so, so the
value shown isn't alarmingly different.

For the Toshiba battery voltage, you can try taking 11.1 and multiplying
it by the suspected missing scale factor of 1.5 . That gives 16.65 volts.
You can compare that to the nominal value printed on the battery pack.

My experience, of checking a few different motherboards, usually
shows one rail measurement is out to lunch, which probably means
the scaling math in the program displaying the results, is just wrong.
Checking with my multimeter, shows the power supply values to
be pretty close to the correct values, and the multimeter measurements
usually make the power supply look heroic, while the hardware monitor
readings usually make the power supply look like a slouch (out of
spec).

On occasion, someone will report they're getting a 0.0 reading
on a major rail, and you just know that can't be right The
computer would be "black screen", if it was really zero.

Thanks, Paul. for the considered and informative reply.
Very thorough...some German chromosomes back in your ancestral line? <g>

Further info, FWIW:
Asus:
- 3.2 GHz P4 Prescott.
- 'Hardware Monitoring', reported by PC Wizard, was Winbond W83627THF
Toshiba:
- Intel i945GM/GME chipset with 1.66 GHz Core 2 Duo Mobile T5500
- 'Hardware Monitoring' National LM79.
- AC adaptor PA-1650-02 output 19V / 3.32A
- rechargeable battrey options listed in the manual for their
Satellite model range:
- 4-cell Li ion pack , 14.4V / 2000mAh
- 6-cell Li ion pack , 10.8V / 4000mAh <= I had this one (model no
PA3465U-1BRS)
- 8-cell Li ion pack , 14.4V / 4300mAh

So maybe my laptop voltage values are being scaled by the factor
10.8/14.4 ?

Anyway, both machines seem to be working OK, so I'll probably just let
sleeping dogs lie for now <g>

Tom

 

[larry moe 'n curly]

Below is a report by the utility 'PC Wizard' of
the motherboard voltages for my Asus desktop PC (P4C800-E del / Antec
TruePower480 PS) and my core 2 Duo laptop (Toshiba Satellite A110-323). This was
measured with no major apps running. The desktop PC was loaded with 2 internal
HD's, a Matrox P650 video card and 4 PCI cards.

Toshiba Asus P4C800-E del/ TruePower480
Voltage CPU 2.05 V 1.41 V
+3.3 Voltage 2.05 V 3.09 V
+5 Voltage 3.44 V 4.81 V
+12V Voltage 8.19 V 11.86 V

Battery #0 11.10 V n/a

Rule of thumb: really wierd voltage readings have to be wrong because
if they were accurate, the computer wouldn't be able to run well
enough to do anything, including run PC Wizard. Hard drives need +5V
within about 5-10% of that value, and desktop HDs also need +12V
within the same tolerance.

 

[Paul]

Rule of thumb: really wierd voltage readings have to be wrong because
if they were accurate, the computer wouldn't be able to run well
enough to do anything, including run PC Wizard. Hard drives need +5V
within about 5-10% of that value, and desktop HDs also need +12V
within the same tolerance.

In some cases, you can get different values displayed on the computer,
by using Asus Probe, MBM5 (for older computers), Speedfan from almico.com
and so on. So an alternative to grabbing the multimeter, is to try some
other programs to do the readout. It's possible even a Linux LiveCD,
will have some utilities available for reading the hardware monitor.
Some other utility may give better values.

And I agree with the observation about hard drives. They're picky
about both the 5V and 12V, and slightly low values cause a desktop
hard drive to "spin down and spin up". I've seen that happen
here, when I had too much load on one of my Molex cables. It
doesn't take much to upset the drive.

Paul

 

[larry moe 'n curly]

In some cases, you can get different values displayed on the computer,
by using Asus Probe, MBM5 (for older computers), Speedfan from almico.com
and so on. So an alternative to grabbing the multimeter, is to try some
other programs to do the readout. It's possible even a Linux LiveCD,
will have some utilities available for reading the hardware monitor.
Some other utility may give better values.

I've usually been able to find at least one monitoring program that
gives fairly accurate readings, but I once had an ECS K7VTA3 ver. 8
mobo that didn't work with anything. SpeedFan said the +12V rail
alternated between 6V and 8V, 1-2 times a second, while MBMx read a
steady 10.62V, which I had earlier confirmed wouldn't let the very
same hard drive spin. A multimeter read 12.11V.

 

[Franc Zabkar]

In some cases, you can get different values displayed on the computer,
by using Asus Probe, MBM5 (for older computers)...

I like MBM5 because you can edit a configuration file to tune the
scale factors to your particular motherboard. It also allows you to
use your own names for voltages, fans, and temperatures, plus many
other settings.

Do any of the current hardware monitor programs have such features?

- Franc Zabkar

 

[Paul]

I like MBM5 because you can edit a configuration file to tune the
scale factors to your particular motherboard. It also allows you to
use your own names for voltages, fans, and temperatures, plus many
other settings.

Do any of the current hardware monitor programs have such features?

- Franc Zabkar

My approach is, if the reading isn't right, ignore it

I've always got the multimeter, if it's important to verify something.

Paul