OT?: Replacing CMOS battery on motherboard

[Frank]

I think I have correctly diagnosed a problem with my PC. It now
completely refuses to start. I don't hear any beep shortly after
powering up - POST has not completed? Nothing appears on screen. So the
CMOS battery needs replacing??

My system is old (>10yrs) - ABIT KG7-Raid motherboard.
Fortunately I recently had the same problem but the system recovered so
I made a note of all the CMOS settings.

Assuming a correct diagnosis, I have 2 questions:

1> Battery: I have never replaced a battery before and the system is
more than a decade old. Is this lifespan normal? My system is up and
running for just a few hours a day. Are the batteries rechargeable?
(CR2032).

2> BIOS: At this point powering up does not activate anything on screen.
Is that normal, i.e. will replacing the battery get me back to where I
can get into the BIOS and re-fill the CMOS?

Thanks
Frank

 

[Paul]

I think I have correctly diagnosed a problem with my PC. It now
completely refuses to start. I don't hear any beep shortly after
powering up - POST has not completed? Nothing appears on screen. So the
CMOS battery needs replacing??

My system is old (>10yrs) - ABIT KG7-Raid motherboard.
Fortunately I recently had the same problem but the system recovered so
I made a note of all the CMOS settings.

Assuming a correct diagnosis, I have 2 questions:

1> Battery: I have never replaced a battery before and the system is
more than a decade old. Is this lifespan normal? My system is up and
running for just a few hours a day. Are the batteries rechargeable?
(CR2032).

2> BIOS: At this point powering up does not activate anything on screen.
Is that normal, i.e. will replacing the battery get me back to where I
can get into the BIOS and re-fill the CMOS?

Thanks
Frank

There is a failure here, with bad caps (bulging or leaking capacitors).

http://www.badcaps.net/forum/showthread.php?t=1286

I wouldn't jump to any conclusions just yet.

Yes, a flat CR2032 can prevent *some* motherboard designs from starting.
And ten years of life for your CR2032 is pretty good.

If the +5VSB is available for the entire ten years, then no current
is drawn from the CR2032. The shelf life of the battery under no
load, might approach ten years. The motherboard is not allowed
to charge those batteries, because the battery would explode
if you did that. Really old motherboards used NiCd batteries,
which would leak and corrode the motherboard - but those actually
charged from the supply. CR2032's have to last all on their own,
no charging allowed. The max charge current allowed is 1 microamp,
so they're serious about allowing current to flow into them.
Current may only flow out.

If the operator turns off all power at the end of each day's eight
hour session, you might get three years from the battery before it is
flat.

If the fans spin when you turn it on, the power supply is making +12V.
But if the motherboard doesn't use the ATX12V 2x2 power connector, the
processor may actually be running off +5V.

With a multimeter you could check the power supply voltages,
just to get some idea. Power supplies are pretty unreliable,
so it could be the power supply which has the leaking caps.
I lost a couple Antec supplies that way.

Changing the CR2032 is certainly cheap, and they are readily
available. Don't buy them from sources that have known poor
turnover. (Like that battery store at the mall, that eventually
left the mall.) A big box store, might have a decent
turnover, so you get a fresh one, and not some old piece
of crap.

So try your idea out, with no guarantees.

1) Could be CR2032 is flat. Only a few motherboard designs
seem to stop dead when that happens. Many others, you can
set the clock in the BIOS (as it loses the correct time),
and boot with a flat battery. The power path there is dual
source, and the battery is diode ORed with a +5VSB derived
source of 3VSB. On my motherboards here (of which many will
have flat batteries), I just enter the BIOS, correct the
settings and they boot. They lose the time and all settings
at the end of the day, but the next day I can turn on,
press <Del>, enter the BIOS and set the BIOS up again, and
away we go. But a few motherboards will black screen like
yours.

2) Visually inspect motherboard for bulging/leaking tops of
capacitors. Especially on the VCore capacitors around the
CPU socket area.

3) If still no relief, pull the power supply, taking note of
where all the wires go, the polarity (which pin took yellow
+12V etc). Make sure AC is removed from the supply. Unscrew
the four top screws on the cover, and *just look*. If you
see rust colored stains on the top of the caps, put the
cover back on the PSU and replace it. For older systems, with
heavy 3.3 or 5V loading, you want a supply from an older era,
with "beef" on those rails. So if the supply does not use
the ATX12V type 2x2 power connector (two yellow, two black
wires), then look for a supply with perhaps 5V @ 25A.

Let's look at a supply, to see the potential problems.

http://images17.newegg.com/is/image/newegg/17-159-043-S04?$S640W$

Yes, 3.3V @ 30A and 5V @ 28A. I could run my Nforce2 motherboard
from that. Mine used to draw just over 20A on +5V, so that supply
appears to have just enough headroom. But be aware of the number
just below that. "Max combined power in watts on +3.3V and +5V
at any time = 130W". If the 5V load goes to max, there is zero
power rating left, to supply any 3.3V! Maybe that thing would
work, but I might want to look at some others.

Now if we look at my old fav, the B204

http://images17.newegg.com/is/image/newegg/17-103-013-S03?$S640W$

3.3V @ 30A, 5V @ 28A, combined limit 150W on both rails.

That leaves some room for both rails to get power.
The video card might draw 6A from 3.3V. And the processor
could be using the 5V.

Some other ones I looked at, are 3.3V @ 20A and 5V @ 20A,
and some of those might be marginal. Especially the 80+ Gold
supplies, where 3.3V and 5V were never intended for heavy
loads, and they actually use a separate regulator board
to convert the main 12V rail, into 3.3V and 5V. (That's
called two stage regulation, and the PSU designer intended
for those to be powering systems that mainly run off +12V.)

The Sparkle is not the worlds best power supply. It tends
to make my UPS beep when first powered. So it has a bit
of inrush current. But at least it's "headed in the right
direction". And this has relatively good reviews, compared
to some.

http://www.newegg.com/Product/Product.aspx?Item=N82E16817103013

*Always* check the ratings in the advert, even if you think
its the B204. As power supply companies can change nameplate
ratings while the same SKU is still being manufactured. For
example, the supply you bought last week, with lots of
Molex connectors, you could buy one today and find nothing
but SATA hard drive connectors on it. It is definitely
a caveat emptor situation. Prepare to be dissatisfied.

Have fun,
Paul

 

[Tim Slattery]

1> Battery: I have never replaced a battery before and the system is
more than a decade old. Is this lifespan normal? My system is up and
running for just a few hours a day. Are the batteries rechargeable?
(CR2032).

Ten years is long enough that I wouldn't be surprised if the CMOS
battery has died. They aren't rechargeable.

2> BIOS: At this point powering up does not activate anything on screen.
Is that normal, i.e. will replacing the battery get me back to where I
can get into the BIOS and re-fill the CMOS?

If the battery had died you would still get something on the screen
when you started up. I think your problem is something other than the
battery.

 

[philo]

I think I have correctly diagnosed a problem with my PC. It now
completely refuses to start. I don't hear any beep shortly after
powering up - POST has not completed? Nothing appears on screen. So the
CMOS battery needs replacing??

My system is old (>10yrs) - ABIT KG7-Raid motherboard.
Fortunately I recently had the same problem but the system recovered so
I made a note of all the CMOS settings.

Assuming a correct diagnosis, I have 2 questions:

1> Battery: I have never replaced a battery before and the system is
more than a decade old. Is this lifespan normal? My system is up and
running for just a few hours a day. Are the batteries rechargeable?
(CR2032).

2> BIOS: At this point powering up does not activate anything on screen.
Is that normal, i.e. will replacing the battery get me back to where I
can get into the BIOS and re-fill the CMOS?

Thanks
Frank

Though not impossible...a dead battery usually does not prevent a
machine from booting...simply the cmos settings are lost.
It cannot hurt to try a new battery though....and also resetting the
cmos by using the jumper

As mentioned elsewhere...if the capacitors on the mobo are bad
it's not worth going any further.
Good chance it's just the power supply though

 

[Ken Blake, MVP]

Ten years is long enough that I wouldn't be surprised if the CMOS
battery has died. They aren't rechargeable.

I'm going back about 25 years, but I once had a rechargeable battery
in my computer. It automatically recharged itself from the computer's
power supply.

But I haven't seen anything like that since then.


Ken Blake, Microsoft MVP

 

[Paul]

I'm going back about 25 years, but I once had a rechargeable battery
in my computer. It automatically recharged itself from the computer's
power supply.

But I haven't seen anything like that since then.


Ken Blake, Microsoft MVP

It's the electrolyte leakage that makes the idea of
nickel cadmium so unpalatable. The CR2032 by comparison,
isn't going to corrode all the copper in the neighborhood,
and is a better overall solution, even if it costs you $3-$5 every
three years.

Paul

 

[Chris S.]

It's the electrolyte leakage that makes the idea of
nickel cadmium so unpalatable. The CR2032 by comparison,
isn't going to corrode all the copper in the neighborhood,
and is a better overall solution, even if it costs you $3-$5 every
three years.

Paul

Just received 5ea CR2032 by Sony for $0.99 postpaid from China.
Took 40 days, but I was in no hurry. Local Safeway has them for $3.98 EACH.

Chris

 

[Bob F]

Yes, a flat CR2032 can prevent *some* motherboard designs from
starting. And ten years of life for your CR2032 is pretty good.

I have had one motherboard like this where replacing the battery fixed it.

 

[telsar]

Just received 5ea CR2032 by Sony for $0.99 postpaid from China.
Took 40 days, but I was in no hurry. Local Safeway has them for $3.98 EACH.

Chris

King Dollar has them for 1.09 for a card of six.

 

[Linea Recta]

I think I have correctly diagnosed a problem with my PC. It now completely
refuses to start. I don't hear any beep shortly after powering up - POST
has not completed? Nothing appears on screen. So the CMOS battery needs
replacing??

My system is old (>10yrs) - ABIT KG7-Raid motherboard.
Fortunately I recently had the same problem but the system recovered so I
made a note of all the CMOS settings.

Assuming a correct diagnosis, I have 2 questions:

1> Battery: I have never replaced a battery before and the system is more
than a decade old. Is this lifespan normal? My system is up and running
for just a few hours a day. Are the batteries rechargeable? (CR2032).

2> BIOS: At this point powering up does not activate anything on screen.
Is that normal, i.e. will replacing the battery get me back to where I can
get into the BIOS and re-fill the CMOS?

I have replaced the battery of my old Asus few years ago. This was very
easy, just push the old one away from under a ridge of the holder and insert
the new one. Mind the right polarity!



--
regards,

|\ /|
| \/ |@rk
\../
\/os

 

[Greegor]

[...] CR2032's have to last all on their own,
no charging allowed. The max charge current allowed is 1 microamp,
so they're serious about allowing current to flow into them.
Current may only flow out.

A 3+ Megohm resistor would work then.

That tiny of a trickle charge might make a BIG difference.

I've seen SEVERAL mothballed systems with CR2032
batteries where resetting the clock, shutting them down
and then checking them later indicated their RTC was
running slow when the system was off.
Leaving them on for 4 hours or so SEEMED to
charge the CR2032 because the RTC began to
keep time accurately when systems were off.

Is it possible they actually implemented the
1 microamp trickle charge you implied?

I've actually seen this trick work several times.

I've also seen systems where the BIOS settings
were "forgotten" due to low battery and running
it for a day caused the battery to hold the settings.

Even if they used a 6 Megohm resistor to
charge the CR2032 battery at half a microamp I would
expect that would be enough to make a big
difference to that tiny battery.

Is there a reason that flash hasn't
replaced CMOS for BIOS settings?

 

[Paul]

[...] CR2032's have to last all on their own,
no charging allowed. The max charge current allowed is 1 microamp,
so they're serious about allowing current to flow into them.
Current may only flow out.

A 3+ Megohm resistor would work then.

That tiny of a trickle charge might make a BIG difference.

I've seen SEVERAL mothballed systems with CR2032
batteries where resetting the clock, shutting them down
and then checking them later indicated their RTC was
running slow when the system was off.
Leaving them on for 4 hours or so SEEMED to
charge the CR2032 because the RTC began to
keep time accurately when systems were off.

Is it possible they actually implemented the
1 microamp trickle charge you implied?

I've actually seen this trick work several times.

I've also seen systems where the BIOS settings
were "forgotten" due to low battery and running
it for a day caused the battery to hold the settings.

Even if they used a 6 Megohm resistor to
charge the CR2032 battery at half a microamp I would
expect that would be enough to make a big
difference to that tiny battery.

Is there a reason that flash hasn't
replaced CMOS for BIOS settings?

http://web.archive.org/web/20011027...EM/Primary/Lithium/lithium_manganese_prod.asp

"(Click the Product Specification Summary link.)

Reverse Charging Current
DL2032 8 microamps *

* Li/MnO2 cells should not be charged.
UL has established these values as the maximum current
the cell will tolerate in the event of diode failure."

1 microamp is a rough figure for reverse leakage allocation
on the ORing diodes next to the CR2032 battery. The spec
sheet I was looking at, that occurs on a BAT54 diode at
around 50C ambient, which is a rough max ambient for
a typical good quality PC.

You shouldn't intentionally try to charge them.
The CR2032 is not intended for charging. The diodes
used around that battery (present in fact, to prevent
charging), have a leakage current associated with
them, and any charging going on, is due to the non-ideal
nature of the diodes (Schottky barrier, low Vf).

There are battery chemistries that allow charging,
but they'd require redesigning that section of
motherboard circuitry. And I'm not talking about
"just sticking a 6 megohm resistor" either. Other
changes would be needed. The Southbridge likely
has some upper limit on well voltage (VBAT), and you
don't want to blow the Southbridge while trying
to save $3.

Paul

 

[Paul]

In message

Is there a reason that flash hasn't
replaced CMOS for BIOS settings?

I can think of three reasons: (a) in cost-conscious cases,
battery-plus-CMOS is probably still cheaper; (b) I think it _has_ in
some cases; (c) you have to have a cell for the clock/calendar anyway,
so [especially in (a) cases] you might as well keep to what's known and
works.

The time for which a system is expected to run also varies widely,
especially between manufacturers (who want a good "refresh" cycle!) and
users. The cell running down can be seen as another trigger for the
consideration of system replacement, at least by the makers.

The CMOS RAM and RTC clock might be thought of as a
common block. The question would be, whether there
is any hope of removing the RTC powering requirement.
It could be done (only have clock running when PC is
powered), but it wouldn't be very convenient. If all
PCs could be guaranteed to be networked, and have access
to NTP protocol, you could always resync the PC when
the OS is running. Otherwise, a standalone PC which is
unplugged every night, would need to have the time typed
into the BIOS every morning.

The BIOS flash chip, is currently written on a regular basis anyway.
DMI and ESCD sections of the BIOS chip are dynamic. Changing
the PC hardware, results in the BIOS updating that information.
So that's an example of non-volatile storage, using the BIOS
flash chip. The CMOS RAM 256 bytes could be stored that way as
well - as long as there is no runtime requirement for the
system to be able to write the RAM at any arbitrary time.
I suppose a BIOS call could support that (some system software
calls the BIOS write routine). But then the issue would be,
whether any usage patterns exist, which would wear out the
BIOS flash. The BIOS flash does not have "wear leveling",
so it's life is limited by how many times a frequently
used block was rewritten.

Paul

 

[joey899244]

I was looking at, that occurs on a BAT54S diode at around 50C ambient, which is a rough max ambient for a typical good quality PC. http://www.hqew.net/product-data/BAT54S