A7V333: Odd shutdowns, as if no power -- clock/timer problem?

  • Thread starter Thread starter Chris Metzler
  • Start date Start date
C

Chris Metzler

I'm experiencing odd problems with my machine.

Symnptoms:

1. The hardware clock stores the time correctly, but doesn't
advance. I use Linux, and the normal procedure for Linux is that
the system (software) clock is loaded from the hardware clock at
boot; time is kept using the system clock; and then the hardware
clock is set to the current system clock value at shutdown. This
all happens just fine. But the next time the machine is booted,
the value loaded from the hardware clock is the time the machine
was shutdown. It's as if the hardware clock can hold the time
OK, and the hardware clock can be successfully accessed and set;
but it does not advance on its own.


2. Left up long enough, the machine will die. The video signal
stops (and the monitor shows a "please check signal cable" message,
and eventually switches to power-save), network and disk LEDs
show no activity (although the disks and the motherboard do seem
to still have power). If I power cycle, the machine will not even
POST; the disks spin up, but there's no beeps or talking; the mobo
sits like a lump. If I wait a couple of days and try again to
power up, it'll send me automatically to BIOS Setup, telling me
that the last time it tried to boot, there was something wrong
with the processor speed in BIOS, and I need to check the values
listed and make sure they were OK (they are). Once checked, the
machine boots OK.

If I shut down the machine myself, it will reboot later with no
problem. But if I allow it to die on its own as above, it won't
reboot until a day or more has passed, and will behave as above
when it boots. Once booted, the amount of time I can have the
machine up before it dies on its own as above is on the order of
a few hours.

Any suggestions/advice greatly appreciated. This is very, very
annoying.

Thanks,

-c
 
Chris said:
I'm experiencing odd problems with my machine.

Symnptoms:

1. The hardware clock stores the time correctly, but doesn't
advance. I use Linux, and the normal procedure for Linux is that
the system (software) clock is loaded from the hardware clock at
boot; time is kept using the system clock; and then the hardware
clock is set to the current system clock value at shutdown. This
all happens just fine. But the next time the machine is booted,
the value loaded from the hardware clock is the time the machine
was shutdown. It's as if the hardware clock can hold the time
OK, and the hardware clock can be successfully accessed and set;
but it does not advance on its own.


2. Left up long enough, the machine will die. The video signal
stops (and the monitor shows a "please check signal cable" message,
and eventually switches to power-save), network and disk LEDs
show no activity (although the disks and the motherboard do seem
to still have power). If I power cycle, the machine will not even
POST; the disks spin up, but there's no beeps or talking; the mobo
sits like a lump. If I wait a couple of days and try again to
power up, it'll send me automatically to BIOS Setup, telling me
that the last time it tried to boot, there was something wrong
with the processor speed in BIOS, and I need to check the values
listed and make sure they were OK (they are). Once checked, the
machine boots OK.

If I shut down the machine myself, it will reboot later with no
problem. But if I allow it to die on its own as above, it won't
reboot until a day or more has passed, and will behave as above
when it boots. Once booted, the amount of time I can have the
machine up before it dies on its own as above is on the order of
a few hours.

Any suggestions/advice greatly appreciated. This is very, very
annoying.

Thanks,

-c
Click to expand...

I can see two possible hardware problems with the machine.

You could have a bad CMOS battery. The battery normally runs
at about 3V or a little higher. It powers the CMOS RAM
and the real time clock, both of which are in the Southbridge.
The battery is not used all the time, as the +5VSB supply is
wired-OR with the CMOS battery, before feeding the Southbridge.
When the computer is unplugged, or the switch on the back of
the computer is off, the CMOS runs from the battery.

If you have ever done a "clear CMOS" operation, and left the
computer plugged in, you could have burned the ORing diode.
There are a couple of possible results - virtually all power
to the CMOS is lost, or the battery is being used all the time
to run the CMOS clock. The battery will run down faster if the
+5VSB powered source is no longer working. (The diode is a
small three terminal device, similar to the package used for
small transistors. It contains two diodes. It is normally
located near to the battery. A typical diode type is BAT54C.)

In terms of battery voltage, typical Southbridges require a
minimum of 2V to run the CMOS. The ORing diode drops 0.4 volts
(as it is a Schottky diode), meaning the battery is valid from
3V to as low as 2.4V. But if you haven't flipped the power switch
on the back of the computer, the clock should still advance,
as it will be powered from the +5VSB derived source.

The real time clock uses a 32768 Hz crystal. It is a metal
cylinder located next to the Southbridge. If it is cracked or
damaged, that will stop the clock as well. A visual examination
will not help much, unless the damage is real bad.

There are some documents referenced in this thread, if you
want to see some pictures.

http://groups.google.ca/group/alt.c..._frm/thread/d557c7fe3eac7705/b74fe4b5d51daefb

For your other problem, it could be power supply or any number
of other motherboard related failures. Watch the power supply
voltages with Asus Probe, and also watch the listed temperatures.
Maybe your CPU is overheating, because the thermal paste
has dried out. Maybe you have a broken clip on your heatsink,
so there is no longer good contact to the heatsink. The 3.3,
5.0, and 12V should not drop to less than 95% of their nominal
value, to still be valid.

Run a copy of Prime95 (mersenne.org) in torture test mode. Watch
the parameters in Asus Probe (lmsensors or mbmon in Linux).
See if the stress of Prime95 causes the computer to freeze
sooner. See if the temps shoot above 60C, or the voltages
drop below acceptable values, when Prime95 is running.

Your computer likely uses +5V to run the processor, so
expect that power supply output to have the most load on
it.

The first thing I would replace is the power supply, but
try to prove it is defective first, before running out
and purchasing another one. If you have a spare available,
it would be worth a speculative swap.

Paul
 
Back
Top