[...]
The problem is that this board ever lost the date. Sometimes it
looses
the
setup. Mainly in summer (below 30 °C however)
Cleaning the battery contacts and replacing it generally cures the
issue
for
some weeks (months in winter).
The C-Mos battery voltage doesn't drops below 3 Vdc. [...]
I see in the manual, that the board has provision for AT power supply
or ATX power supply. The ATX supply has +5VSB to run the clock, so
even when Windows is shut down, the +5VSB can power the clock. (Of
course, if the ATX power supply is switched off at the back, or
if the power supply is unplugged, then you are running on the
battery again.)
My guess is, you are using an AT supply ? Try using an ATX instead.
It could be that the Southbridge has become "leaky" and is drawing
more power than when it was new. By using an ATX supply and leaving
the ATX supply powered on, even while Windows is shut down, will
supply power to the clock when you aren't using the PC. The ATX
supply will draw around 3 watts from the wall while doing this.
HTH,
Paul
Hello Paul,
1- Yes this is an ATX PSU (I used my Athlon's one when I replaced it for a
465W Enermax last year).
2- No, I'm paranoid. Mains is open as well as the phone line (via a relay)
when I'm not at home !
3- Good idea, first step: for some days, I'll let the PC sleeping and see if
it looses anything.
Some remarks:
Although I'm keeping a lill' bit puzzled: monitoring the battery with two
different multimeters shows that the voltage is around 3.08 / 3.11 V,
whatever the computer status is (On or Off). I didn't verify the battery
voltage with a scope, assuming that a short drop (less than some seconds)
shouldn't reset the memory. This voltage drop, if any, should probably be
visible at boot or at switch off. No appearant reason it occurs randomly. In
consequence, since I was staring at it, I obviously should have seen it. I
didn't. The voltage looks to be correct and very stable.
I'm also wandering if replacing the CR2032 by anything else (two 1.5V AA
type batteries in series for instance) could help. I don't know the what the
circuits specs are and wouldn't harm them. On another hand, since there's no
appearant voltage drop, a higher capacity battery won't bring anything else.
I'd like trying to monitor the voltage deeper in the circuits. The question
is that I don't know where to probe ! Moreover I don't know the schematic.
A question: What do you intend by "the southbridge may have become leaky" ?
Usually, electrolytic capacitors are focussed on first but it looks there
aren't around the battery and due to their technology this is quite normal.
I know that junctions may leak, the current doubling every 6°C step. In my
case, the problem is more or less random, with a tendency to occur more
often in summer. But now the temperature is 10°C below summer's one and it
happened again.
The problem is that the voltage is a bit greater than 3 Volts !
Thanks for your patience.
I thought maybe your battery was constantly running down.
If the battery is good, and still the problem happens, then
you have to look elsewhere.
For a schematic, this schematic uses the same Southbridge as your
board. See page 18, lower left hand corner.
http://www.intel.com/design/chipsets/designex/BXDPDG10.PDF
Try measuring the voltage, not by touching the battery, but by
touching the metal of the CLRTC jumper plug, while it is in place.
The metal bridge of the jumper plug, should be at the same potential
as pin L16 on the 82371EB. (First, try this measurement with the
ATX power supply powered, and then try with the computer unplugged
from the wall. If the voltage is low, like getting close to 2.0V,
when the computer plug is pulled, then something is interfering
with the connection from the battery, to pin L16.) I would
have suggested measuring right at pin L16, but that would
mean pulling the motherboard to get at the pin.
It could be something as simple as corrosion on the jumper
plug itself. Find a tin plated jumper plug, and replace the
copper one, if the current jumper is just copper.
The PIIX4 datasheet is here. Strangely, I cannot find a spec
for how low the RTC battery voltage can drop, and the chip
guarantees the RTC info content. Most Southbridge chips
will have a spec for that.
ftp://download.intel.com/design/intarch/datashts/29056201.pdf
OK, this doc contains the battery voltage spec. The voltage
at L16 should be between 2.0V and 3.6V, from 0-85C case temp.
The Schottky diode in the battery path drops 0.4V, so the 2.0V
Intel spec translates to a minimum 2.4V battery voltage to
maintain RTC operation.
ftp://download.intel.com/design/chipsets/datashts/29054801.pdf
HTH,
Paul
