Tyan S2735-8M weirdness

[The Doctor]

Question, why would my BIOS report 0.0V on all the voltage readings in the BIOS?
--
Member - Liberal International This is doctor@@nl2k.ab.ca Ici doctor@@nl2k.ab.ca
God,Queen and country!Never Satan President Republic!Beware AntiChrist rising!
http://www.fullyfollow.me/rootnl2k Look at Psalms 14 and 53 on Atheism
Abuse a man unjustly, and you will make friends for him. -Edgar Watson Howe

[Paul]

The Doctor wrote:
> Question, why would my BIOS report 0.0V on all the voltage readings in the BIOS?

Well, you know logically that's not possible.
The logic on the motherboard, the voltages can
only deviate by a certain amount, before the thing
would not function properly any more.

Possible reasons:

1) ADC in hardware monitor is ratiometric. The reference
gets shorted to ground. All readings end up "0.0". I don't
think that's how they do the ADC on these things.

2) Modern hardware monitors, use external scaling resistors.
Full scale readout might be 4.096V. To measure a 12V signal,
use a resistive divider, that reduces the magnitude of the
voltage by a factor of four. Multiple all readings by 4.
If the ADC measures 3.000, times 4 gives 12V. If the input
voltage rises to 15V (abnormal, overvoltage about to be
triggered inside the PSU), the reading is still on-scale.

The scale values are now stored in a BIOS table. If a BIOS
update mis-calculated the values, or stored them in the wrong
place, the scaling values used might all be zero.

3) The BIOS code is defensively designed. If the hardware monitor
LPC bus is not responding, the BIOS does not "crash". It substitutes
a value of zero for every thing it cannot read. It doesn't put
NAN or INF or some other crazy stuff as a flag. Zero is good
enough on a hardware failure.

*******

Use Speedfan from almico.com (for Windows), or Linux has
lmsensors and some readout packages, so you can verify
the readings while an OS is running. This will tell you whether
the problem is isolated to just the BIOS readout page.
Speedfan in particular, has a little log of its "discovery"
process when the program starts. So you have some idea
what hardware it thinks it is detecting.

Some motherboards have two hardware monitors, because
one monitor doesn't have enough voltage channels. There
is more room for technical slip-ups, with configurations
like that.

I don't really like my answers so far, in terms of them
being convincing. Therefore, I recommend collecting
more data, in the OS of your choice. Also, review the
BIOS update history, when was the last time the board
was flash upgraded, have you reviewed the BIOS release
notes to see if something got changed. You can do a bit
of that kind of investigating too.

Paul

[The Doctor]

So far so good. Just to find a BSD version of what you are talking about.

[Paul]

The manual is a bit confusing. I can only see a general s2735
manual, but there are two models. I included both LMsensors
files, just for contrast.

ftp://ftp.tyan.com/manuals/m_s2735_100.pdf

ftp://ftp.tyan.com/software/lms/lms_s2735-8M.tgz
ftp://ftp.tyan.com/software/lms/lms_s2735.tgz

It's possible just one BIOS design is included for both
boards, and this leads to the GUI "doing the wrong thing"
for one of them.

For the 2735-8m, there are two apparent sensor chips.
The adm1027 may be replaced by W83782D, depending on
board model. In this case, the assumption is adm1027
is present. There is some program in lmsensors, a
command line program, which delineates what it finds
on LPC or I2C or whatever. There is a utility that
probes the bus, prints out the chips it finds, and so on.

adm1027
label in0 "DDR VCC"
label in1 "CPU Vcore"
label in2 "3VSB"
label in3 "+5V"
label in4 "+12V"
w83627hf
ignore in0
ignore in1
ignore in2
label in3 "+5V"
ignore in4
ignore in5
label in6 "-5V" <--- tentative identification, no wire on PSU,
will likely read zero, no reason to include this
label in7 "V5SB"
ignore in8
ignore vid <--- dual socket, two VIDs, no way to monitor
this that would make sense ? (Each socket
has its own VRM or VRD.)

If you had the S2735, the scheme would be different.

w83782d
label in0 "CPU Vcore"
label in1 "+1.2V"
label in2 "+3.3V"
label in3 "+5V"
label in4 "+12V"
label in5 "+2.5V"
label in6 "+1.8V"
ignore in7 <--- maybe 3VSB ?
label in8 "VBat"
w83627hf
ignore in0
ignore in1
ignore in2
ignore in3
ignore in4
ignore in5
ignore in6
ignore in7
ignore in8
ignore vid

Now the question is, are the BIOS designs for the
boards unique ? Do they have the same GUI ? Is
the GUI "one size fits all", with some channels
reading zero, because the lineup on the two
boards is completely different ?

Even if I was sitting in front of the LCD screen
and staring at the hardware monitor display,
I'd probably have trouble lining it up, against
the .conf definitions.

I got the LMSensors files from here.

http://www.tyan.com/archive/support/html/software_utilities.html#lms

Paul

[The Doctor]

There is a possiblity that I remove the said board due to
some heavy flunctuation that made it unrealible.