PC losing date-time after power outage

[Michael T.]

Every time there is a power outage my Dell 4600C displays an old date.

The rather obvious solution seemed to be that CMOS battery needed replaced.

So I replaced the battery (3-V CR2032 coin cell).

Many months went by until another power outage occurred today. Well lo &
behold when the power came back on the same problem reappeared - or perhaps
I should say "reared its ugly head."

Any ides short of calling DELL, who has not been particularly helpful in the
past.

Michael

Footnote:
I was very careful to use the identical battery and to carefully follow the
steps including those pertaining to grounding. I have added memory, modem
cards and network cards for a variety of PCs. So while we can all make
mistakes I am certainly not a novice at this.

 

[Paul]

Every time there is a power outage my Dell 4600C displays an old date.

The rather obvious solution seemed to be that CMOS battery needed replaced.

So I replaced the battery (3-V CR2032 coin cell).

Many months went by until another power outage occurred today. Well lo &
behold when the power came back on the same problem reappeared - or perhaps
I should say "reared its ugly head."

Any ides short of calling DELL, who has not been particularly helpful in the
past.

Michael

Footnote:
I was very careful to use the identical battery and to carefully follow the
steps including those pertaining to grounding. I have added memory, modem
cards and network cards for a variety of PCs. So while we can all make
mistakes I am certainly not a novice at this.

The way it is supposed to work, is the RTC has two power sources

ATX_+5VSB_related_source ----->----+
|
+---------- 32KHz_digital_watch_circuit
| RTC (real time clock)
CR2032_battery_source -------->----+

The "->-" is meant to indicate a diode. It prevents reverse current flow,
so that the battery is not accidentally charged. The higher of the two
voltages "wins", and as long as the +5VSB is present, no current flows
out of the battery (its diode is reverse biased).

That means, if you put the computer "to sleep" in S3 Suspend To RAM,
the CR2032 battery is not being used. Under those circumstances, you
should get close to the 10 year shelf life.

The other way of doing things, is to use the computer for 8 hours a
day. Then switch off at the back at the end of the day. In some
countries, this is a more popular practice than in others. In some
countries, the 10-20W of wasted electricity in S3 sleep is a concern,
so the people switch off at the back.

The end result, is the RTC runs on battery for 16 hours a day, and
from the ATX power supply for eight hours. A slightly worse situation,
is storing the PC without AC power being present, in which case the
battery is used 24 hours a day.

In either of those two cases, you may see a battery life of three years
or so. I've had a couple failures on stored PCs and that is a rough
guess as to how long they lasted.

I have read an account of at least one motherboard design, where it
seems they just run the RTC off the battery all the time. The battery
life there is a lot shorter than it has to be, all for the cost of a
few pennies for the dual diode used to logic OR together the two
power sources.

Fortunately, CR2032's are relatively cheap (compared to some other
battery form factors), and the main expense is the user time wasted
fitting it.

I've bought some silver oxide batteries, from a battery shop in the
mall, that were just about flat when I got them. Only lasted a matter
of a week or two. So it is possible to get bad products.

Paul

 

[Michael T.]

The way it is supposed to work, is the RTC has two power sources

ATX_+5VSB_related_source ----->----+
|
+---------- 32KHz_digital_watch_circuit
| RTC (real time clock)
CR2032_battery_source -------->----+

The "->-" is meant to indicate a diode. It prevents reverse current flow,
so that the battery is not accidentally charged. The higher of the two
voltages "wins", and as long as the +5VSB is present, no current flows
out of the battery (its diode is reverse biased).

Paul

Thanks Paul. Although you came dangerously close to going "over the head" of
this old crusty software engineer with 30 years experience.

What I didn't realize is that 'it appears' that this Dell PC (it is actually
at my sister's condo) only loses time during a power outage. This makes me
wonder if the battery is only used when the power cord is disconnected. No
problems when she powers down her PC overnight or for a few days.

I did not call Dell, but I did logon to their chat help where I have had
mixed results. Their solution is to "flash the BIOS" after downloading the
file.

Note: With a 2004 date stamp it appears to be the same BIOS that was flashed
at the factory.

MIGHT THIS HELP? Or am I justified in being skeptical?

If so, I'll drive the 30 miles and try it.

 

[Michael T.]

Every time there is a power outage my Dell 4600C displays an old date.

The rather obvious solution seemed to be that CMOS battery needed
replaced.

So I replaced the battery (3-V CR2032 coin cell).

Many months went by until another power outage occurred today. Well lo &
behold when the power came back on the same problem reappeared - or
perhaps I should say "reared its ugly head."

One more thing that seems peculiar.

The owner of this computer is my sister and unfortunately she lives in a
part of San Diego that had to be evacuated during the recent fire storms.
(Fortunately no damage to her condo.)

Anyway, the power company turned the power off for 2-3 days back in
November. Her computer was *not* on (running) when she lost power. No
problems with the clock when she returned to her home.

But this morning she was using her computer when the power went off. Then
about five hours later power was restored and her PC clock was back to the
old factory setting - not real-time.

Perhaps this will offer a clue to someone.

 

[Paul]

Thanks Paul. Although you came dangerously close to going "over the head" of
this old crusty software engineer with 30 years experience.

What I didn't realize is that 'it appears' that this Dell PC (it is actually
at my sister's condo) only loses time during a power outage. This makes me
wonder if the battery is only used when the power cord is disconnected. No
problems when she powers down her PC overnight or for a few days.

I did not call Dell, but I did logon to their chat help where I have had
mixed results. Their solution is to "flash the BIOS" after downloading the
file.

Note: With a 2004 date stamp it appears to be the same BIOS that was flashed
at the factory.

MIGHT THIS HELP? Or am I justified in being skeptical?

If so, I'll drive the 30 miles and try it.

Flashing the BIOS has nothing to do with the battery state, or
the ability to remember settings. The Dell folks are reading
from a chart, and not using their heads.

The battery is only used, when the other source of power is not present.
Switching off at the back, or unplugging the computer, are two examples
where the battery instantly comes into play. The two diodes I
drew in my picture, make that logic possible - as soon as the 5VSB
side drops too low, current starts to flow through the diode on
the CR2032 battery side. And that is because at that point, its
diode is forward biased and in the conducting state.

The battery is called on to deliver a small number of microamps
of current, when things are running on battery. I think I'd have
trouble here, doing an accurate measurement of that current flow,
even if there was some way I could put a multimeter in series with the
battery socket. (It means not being able to seat the battery, to
make a measurement of the current like that.)

If you're going to drive the 30 miles, slap another battery in it.

Some people have proposed the use of a larger battery type, to
"solve" the problem. It all depends on the shelf life of said
battery scheme, as to whether it lasts for enough years to make
it a practical solution. Naturally, the shape of the battery
socket, makes anything other than slapping a new battery in there,
difficult.

Paul

 

[Ghostrider]

One more thing that seems peculiar.

The owner of this computer is my sister and unfortunately she lives in a
part of San Diego that had to be evacuated during the recent fire storms.
(Fortunately no damage to her condo.)

Anyway, the power company turned the power off for 2-3 days back in
November. Her computer was *not* on (running) when she lost power. No
problems with the clock when she returned to her home.

But this morning she was using her computer when the power went off. Then
about five hours later power was restored and her PC clock was back to the
old factory setting - not real-time.

Perhaps this will offer a clue to someone.

Same answer...possibly CMOS battery. However, check that the CMOS clearing
jumper (or reset) is not set.

 

[Michael T.]

Same answer...possibly CMOS battery. However, check that the CMOS clearing
jumper (or reset) is not set.

Forgive me, but what is the clearing jumper?

I have a new theory that something may be wrong with the circuitry for the
battery and/or the battery socket and that it may be the case that replacing
the battery (last year) had no effect.

It is my understanding that once Windows XP boots and connects to her DSL
connection the PC synchronizes the correct time and no longer needs the
battery.

Before I go out and buy another battery the next time I am over there I will
perform a test:
1) Power down
2) Disconnect the power cord
3) Wait a bit then reconnect the power
4) Unplug the DSL line
5) Boot w/o the Internet connection
6) Check the PC clock

If the clock is the old default date-time, I'll purchase and replace the
battery (again).

Then I will repeat the six steps above as it is important for me to see if
it is the battery as you and Paul have suggested. Obviously if there are
still problems it has to be something other than the battery. What are the
odds of purchasing two defect batteries. For that matter, what are the odds
that a desktop computer less than two years old would have a bad battery. I
have two other DELL computers - a 5 yr desktop and a 2 yr laptop - with no
date-time problems ever. So I am puzzled why this DELL 4600C has had battery
problems (if that is the case) twice since 2004. Oh well I am sure I'll
solve it eventually.

 

[Jay B]

if she has the computer on a power strip and turns it off for long
periods. that could be draining it faster than normal.

where did you get the last battery from?
if you got it from a small store, perhaps they dont sell a lot, and the
battery was on the shelf for a long time, so the battery may not have
been fresh.

 

[Paul]

Forgive me, but what is the clearing jumper?

I have a new theory that something may be wrong with the circuitry for the
battery and/or the battery socket and that it may be the case that replacing
the battery (last year) had no effect.

It is my understanding that once Windows XP boots and connects to her DSL
connection the PC synchronizes the correct time and no longer needs the
battery.

Before I go out and buy another battery the next time I am over there I will
perform a test:
1) Power down
2) Disconnect the power cord
3) Wait a bit then reconnect the power
4) Unplug the DSL line
5) Boot w/o the Internet connection
6) Check the PC clock

If the clock is the old default date-time, I'll purchase and replace the
battery (again).

Then I will repeat the six steps above as it is important for me to see if
it is the battery as you and Paul have suggested. Obviously if there are
still problems it has to be something other than the battery. What are the
odds of purchasing two defect batteries. For that matter, what are the odds
that a desktop computer less than two years old would have a bad battery. I
have two other DELL computers - a 5 yr desktop and a 2 yr laptop - with no
date-time problems ever. So I am puzzled why this DELL 4600C has had battery
problems (if that is the case) twice since 2004. Oh well I am sure I'll
solve it eventually.

The example I see here ("PWD1"), looks like it might only include a subset of
CMOS contents. There is no mention that this procedure clears the
RTC or the entire contents of the CMOS battery backed RAM in the
Southbridge. (The Southbridge is the big chip without a heatsink
on it, in the lower right corner of the motherboard. It handles
lower speed I/O, like for the disks. It is also home of the
256 bytes of CMOS RAM, where BIOS settings are stored. The RTC
clock is also inside that big chip. The CMOS battery only powers
a tiny portion of the chip, so the whole chip doesn't get powered
by the battery. The portion of circuit powered by the battery,
is referred to as sitting in the "CMOS well", isolated from other
Southbridge logic.)

PWD1 is an example here, in "Clearing Forgotten Passwords".
http://support.dell.com/support/edocs/systems/dim4600C/sm/syssetup.htm

It could be, that removing the battery is how they expect you to
clear the entire contents of CMOS.

http://support.dell.com/support/edocs/systems/dim4600C/sm/battery.htm

On some motherboards, a jumper labeled "CLR_RTC" or "CLR_CMOS" will
reset the BIOS settings and the real time clock. On some of those,
placing the jumper in the wrong position results in draining the
battery, or even damage to the ORing diodes. That is one reason,
that as a general principle, any RTC clearing description should
begin with an instruction to unplug the computer. You never know
for sure, exactly how a "clear CMOS" setup is designed, and at
least some of them are not designed defensively.

Some retail motherboards, have been shipped with the CLR_RTC jumper in
the wrong position, leading to premature battery failure. But I
don't think I see that kind of jumper on this motherboard. I
cannot be sure - this is the best picture of the 4600C motherboard I
can find so far.

http://i16.ebayimg.com/03/i/000/be/c5/cbe3_1.JPG

Paul

 

[CBFalconer]

.... snip ...

Then I will repeat the six steps above as it is important for me
to see if it is the battery as you and Paul have suggested.
Obviously if there are still problems it has to be something other
than the battery. What are the odds of purchasing two defect
batteries. For that matter, what are the odds that a desktop
computer less than two years old would have a bad battery. ...

The odds are fairly small, but a bad isolation (anti discharge)
diode in the battery isolation circuit could cause discharge or
destructin with power on.

 

[CBFalconer]

Jay B wrote: *** and top-posted - fixed ***

if she has the computer on a power strip and turns it off for
long periods. that could be draining it faster than normal.

where did you get the last battery from? if you got it from a
small store, perhaps they dont sell a lot, and the battery was
on the shelf for a long time, so the battery may not have been
fresh.

Please do not top-post. Your answer belongs after (or intermixed
with) the quoted material to which you reply, after snipping all
irrelevant material. I fixed this one. See the following links:

--
<http://www.catb.org/~esr/faqs/smart-questions.html>
<http://www.caliburn.nl/topposting.html>
<http://www.netmeister.org/news/learn2quote.html>
<http://cfaj.freeshell.org/google/> (taming google)
<http://members.fortunecity.com/nnqweb/> (newusers)

 

[wm_walsh]

Hi!

Every time there is a power outage my Dell 4600C displays
an old date.

The rather obvious solution seemed to be that CMOS
battery needed replaced.

Well lo & behold when the power came back on the same
problem reappeared

That is an interesting problem. While I think there is a faint
possibility that perhaps you got a "dud" battery, the circumstances
seem suspicious. I would start by cleaning the system unit, paying
special attention to the Super I/O controller on the motherboard
(usually a chip from SMSC, Winbond or ITE) because many of them
integrate the clock and CMOS area.

I have found that sometimes some dust or other gunk will get on the
pins of the chip and cause strange behavior, including battery
depletion. Cleaning it off (using some form of compressed air) solved
these problems.

William

 

[wm_walsh]

Hi!

Forgive me, but what is the clearing jumper?

The CMOS clear (or "password reset") jumper is a small block of
plastic covering a set (usually two or three) pins on the motherboard.
Some designs use two pins and others have you switch the jumper
between three pins--e.g. you'd have the jumper on pins 1 and 2
normally and change it to 2 and 3 to clear the password.

I have found computers with these jumpers set in the wrong position,
and I can say that in some cases it will drain the battery.

You may find that your Dell computer has a legend showing where many
of the jumpers and connectors are on the motherboard. The "CMOS clear"
or "password reset" jumper will be somewhere near the battery in all
likelihood.

It is my understanding that once Windows XP boots and
connects to her DSL connection the PC synchronizes the
correct time and no longer needs the battery.

That is correct, at least to a point. If the time is far enough off,
Windows XP won't synchronize it. I'm not quite sure why Microsoft did
it that way, but they did.

So I am puzzled why this DELL 4600C has had battery
problems (if that is the case) twice since 2004.

I think you might have figured the problem out in one of your posts
above. If I understood you correctly, you said that when the system is
properly shut down and then disconnected from power, the clock keeps
working fine. But if the power fails while the system is running, the
clock is messed up.

That really sounds like some kind of odd hardware defect to me. As
long as the clock works otherwise, the cheapest and easiest solution
would be to simply reset the clock after the power fails.

William

 

[DevilsPGD]

Flashing the BIOS has nothing to do with the battery state, or
the ability to remember settings. The Dell folks are reading
from a chart, and not using their heads.

It's remotely possible that Dell knows something we don't here, so I'd
at least try it before going too far.

I had an old old Gigabyte mobo that would lose the date but keep time
under certain circumstances (it depended on how you powered off, IIRC),
which a BIOS upgrade fixed.

Admittedly, power off being okay, and unplugged resulting in lost time
definitely points to battery related issues, but it's equally possible
that the BIOS stores the time in two places (one for AC, and one in
battery only) and fails to read the battery-only one in some
circumstances.

 

[Michael T.]

Some retail motherboards, have been shipped with the CLR_RTC jumper in
the wrong position, leading to premature battery failure. But I
don't think I see that kind of jumper on this motherboard. I
cannot be sure - this is the best picture of the 4600C motherboard I
can find so far.

http://i16.ebayimg.com/03/i/000/be/c5/cbe3_1.JPG

Paul

That appears to be the correct picture for the Dell Dimension 4600C
motherboard.

Also, here is a diagram/picture of the "System Board" from the Owner's
Manual.
http://tinyurl.com/2fze8s

I'll see if I can find out if there is a CLR_RTC jumper.

Thanks for your help.

 

[Michael T.]

It's remotely possible that Dell knows something we don't here, so I'd
at least try it before going too far.

I had an old old Gigabyte mobo that would lose the date but keep time
under certain circumstances (it depended on how you powered off, IIRC),
which a BIOS upgrade fixed.

Admittedly, power off being okay, and unplugged resulting in lost time
definitely points to battery related issues, but it's equally possible
that the BIOS stores the time in two places (one for AC, and one in
battery only) and fails to read the battery-only one in some
circumstances.

OK. I will flash the BIOS then. That's an easy one - or should be <g>.

 

[Michael T.]

if she has the computer on a power strip and turns it off for long
periods. that could be draining it faster than normal.

where did you get the last battery from?
if you got it from a small store, perhaps they don't sell a lot, and the
battery was on the shelf for a long time, so the battery may not have
been fresh.

The power strip check is a good idea, as that seems consistent with
depleting a CMOS battery twice in three years. Unfortunately I think the
power strip is on the floor behind the desk and difficult to access.
Therefore I think it is likely she only turns off this desktop from the
front panel pushbutton.

As to which store, I believe I purchased the battery at a local drug store
(Sav-On or CVS).

I wonder if I take in the old one they can test it. Or perhaps a Radio Shack
instead of a drug store.

 

[Michael T.]

Good stuff William.

 

[wm_walsh]

Hi!

Good stuff William.

Thanks. I hope it might be helpful.

"If the time is far enough off, Windows XP won't synchronize
it."

It seems that this might be behavior that is specific to Windows XP
Service Pack 2 only. I honestly don't recall whether prior versions of
Windows XP would complain in this way or not. However, I don't think
they do. If you go into the Time/Date settings (double click the time
on the taskbar if it is displayed there or use date/time in the
control panel) and the clock is off far enough, clicking the "Update
Now" button under the "Internet Time" tab will result in a message
stating that "For security reasons, Windows can not synchronize with
the server because your date does not match. Please fix the date and
try again."

I'm not sure what kind of security benefit there is to be had here.
You could always point the time/date settings applet to a malicious
time server, but that would require that the machine be exploited in
some other manner or physical access to it. Then again, maybe I use
this feature the wrong way...I usually use it on computers that have
far off time settings for some reason or another when a clock isn't
handy.

But that's a little off the topic...good luck figuring this out.

William

 

[Michael T.]

Internet Time (tab) for Date and Time dialog

The "Update Now" button under the "Internet Time" tab will result in a
message
stating that "For security reasons, Windows can not synchronize with
the server because your date does not match. Please fix the date and
try again."

That's exactly what we saw (also) when my sister called about this problem
last night.