Failed CMOS battery? Again?

M

MrTsquare

OK, I have totally failed Gravity search. Amyway, my goal was to post
this under my last post on the subject seems like November/December2014.
At the time I was asking why my system clock seemed to stay where it was
the last time the system was turned off rather than keep up. Seems to
have happened again. Paul was explaining how the 2032 cmos battery
should last up to 10 years if they were indeed fresh and the motherboard
didn't have some kind of heavy load. Getting same same symptoms some 2
to 3 months after changing the battery last. Any reasonable
troubleshooting before replacing the battery again? System was
initially assembled in August 2014 with the initial 2032 failure about
Nov/Dec2014

Z-97Pro/4790K/770 system

T2
 
P

Paul

MrTsquare said:
OK, I have totally failed Gravity search. Amyway, my goal was to post
this under my last post on the subject seems like November/December2014.
At the time I was asking why my system clock seemed to stay where it was
the last time the system was turned off rather than keep up. Seems to
have happened again. Paul was explaining how the 2032 cmos battery
should last up to 10 years if they were indeed fresh and the motherboard
didn't have some kind of heavy load. Getting same same symptoms some 2
to 3 months after changing the battery last. Any reasonable
troubleshooting before replacing the battery again? System was
initially assembled in August 2014 with the initial 2032 failure about
Nov/Dec2014

Z-97Pro/4790K/770 system

T2

Well, something is draining the battery. What could it be ?

(Detailed section of two Intel reference schematics, showing
a 1K ohm resistor just after the battery.)

http://i61.tinypic.com/2vsnfqs.gif

You would measure the voltage across the resistor, to get an
idea of current flow through it. The multimeter would have
a 1 megohm or 10 megohm input impedance, which is relatively
large compared to the 1K ohm resistor you'd be putting it
across.

With all PC power off (PC unplugged), 10uA comes from the
battery. The 10uA, flowing through the 1K ohm resistor gives

V = RI = 1000 * 0.000010 = 0.010V = 10mV

You could probably just barely see this on the multimeter,
on the 2V scale.

If the current draw is higher than the suggested 10 microamp
figure, the voltage across the resistor would be higher.

That's the easiest way to verify the load.

And if the load is above 10uA, there are a number of possible
culprits. The SuperI/O might be connected to VBAT (not shown
on the reference schematics, as the Intel schematics are
"not all that real"). The Southbridge could suck more than
the normal current. I don't know if the leakage or load spec
is verified and chips with high leakage are discarded.

It's also possible some transmission gate between
the CMOS well and the rest of the Southbridge isn't
working right, and phantom power flows into the rest
of the chip.

A "clear CMOS" jumper (CLR_RTC or similar), if left in
the wrong position, could suck power. But then it probably
wouldn't last 3 months while doing so, and you'd know as
the BIOS settings would be constantly reset.

*******

If you keep the supply connected and switched on at the
back of the computer, the other diode path from the
motherboard should have a higher voltage. By doing so,
it "cuts off" the battery side diode, and prevents any current
from coming out of the battery. And since the 3P3_standby
or 3VSB source is "strong", it doesn't take no for an answer.
Unlike the battery, with the 1K current limiting resistor in
the way, the 3VSB source has no current limit. If there is
a short there, then "something smokes" :) Well, not really
smoke, but what I'm saying is the 3VSB (with power supply
switched on at the back), is a "stiff" source, and it can
easily power a leaky load.

So just keep the PC powered at the back, as your workaround.
Yes, it costs a few dollars per year for electricity. Compare
the cost (and time wasted) changing CMOS CR2032, versus
the wasted electricity keeping the PC power supply on at
the back. No more switching off at the power strip...
Leave the standby voltage running on the PC at all times.
(On an Asus motherboard, the green LED would be always lit.)
That should stop the CMOS cell from discharging. If it
did not, then maybe the dual diode itself, needs to be
replaced.

There might even be solder blobs, unintentional loads
on the copper tracks. Which might be difficult to spot.

I remember having a board in the lab that was "acting weird".
I spent quite a while debugging, and stuck a scope on a
particular track. And I could see a capacitor discharge
curve (instead of normal square waves). I look at the board
carefully with my magnifying glass, and the factory had stuffed
the leg of a capacitor into the wrong hole! This caused the
circuit to behave in unexpected ways (because now the circuit
is no longer as I designed it). That's an example of "weird shit"
happening when the pick and place machine slips and shoves a wire
into the wrong hole. So even if you think you know how
a circuit works, the staff at the factory can throw
curve balls at you.

There are a ton of ways that poor CMOS cell could have
the charge leak away...

Paul
 
R

Rodney Pont

OK, I have totally failed Gravity search. Amyway, my goal was to post
this under my last post on the subject seems like November/December2014.
At the time I was asking why my system clock seemed to stay where it was
the last time the system was turned off rather than keep up. Seems to
have happened again. Paul was explaining how the 2032 cmos battery
should last up to 10 years if they were indeed fresh and the motherboard
didn't have some kind of heavy load. Getting same same symptoms some 2
to 3 months after changing the battery last. Any reasonable
troubleshooting before replacing the battery again? System was
initially assembled in August 2014 with the initial 2032 failure about
Nov/Dec2014

Z-97Pro/4790K/770 system

I'd contact the motherboard supplier and tell them that batteries are
only lasting 3 months.
 
F

Flasherly

OK, I have totally failed Gravity search. Amyway, my goal was to post
this under my last post on the subject seems like November/December2014.
At the time I was asking why my system clock seemed to stay where it was
the last time the system was turned off rather than keep up. Seems to
have happened again. Paul was explaining how the 2032 cmos battery
should last up to 10 years if they were indeed fresh and the motherboard
didn't have some kind of heavy load. Getting same same symptoms some 2
to 3 months after changing the battery last. Any reasonable
troubleshooting before replacing the battery again? System was
initially assembled in August 2014 with the initial 2032 failure about
Nov/Dec2014

Z-97Pro/4790K/770 system

T2

Computers as clocks are traditionally bad timekeepers. Not keeping
time is less a problem than losing enduser-set CMOS BIOS. A utility
to autoconnect every few minutes to an atomic clock relay is an easy
solution. (That's a bit radical, staying where the MB was last turned
off. A system clock off a few minutes a month is more usual to what
I've expect. Still, I wouldn't care as long as the BIOS settings
haven't changed, that the battery at least ensures they stay that way
for another 10 years.)
 
M

MrTsquare

OK, turned it on tonite after leaving the power connected to the PS all
day since last nite. No change to symptoms. Still shows last night's
date and time. Guess I need to just feed it a 2032 every few months.

T2
 
M

MrTsquare

Well, something is draining the battery. What could it be ?

(Detailed section of two Intel reference schematics, showing
a 1K ohm resistor just after the battery.)

http://i61.tinypic.com/2vsnfqs.gif

You would measure the voltage across the resistor, to get an
idea of current flow through it. The multimeter would have
a 1 megohm or 10 megohm input impedance, which is relatively
large compared to the 1K ohm resistor you'd be putting it
across.

With all PC power off (PC unplugged), 10uA comes from the
battery. The 10uA, flowing through the 1K ohm resistor gives

V = RI = 1000 * 0.000010 = 0.010V = 10mV

You could probably just barely see this on the multimeter,
on the 2V scale.

If the current draw is higher than the suggested 10 microamp
figure, the voltage across the resistor would be higher.

That's the easiest way to verify the load.

And if the load is above 10uA, there are a number of possible
culprits. The SuperI/O might be connected to VBAT (not shown
on the reference schematics, as the Intel schematics are
"not all that real"). The Southbridge could suck more than
the normal current. I don't know if the leakage or load spec
is verified and chips with high leakage are discarded.

Thanks, Paul. Not having appropriate test equipment and detailed
knowledge of the board layout, I don't think I'll try for and voltage
measurements. Also looks like that C15 0.1 microF cap if leaking a
little might cause something similar. I'll send something to ASUS as
well just to see what they say.

I think I'll leave the power strip on for a day or so and see if that
doesn't at least hide the symptoms. And then get a few more 2032
batteries. Totally O/T - I bought at an after Christmas sale 4 cute
little electric candles - powered by 2032 batteries. tried 1 and it
lasted for about a month when on for about 5 hours a day - timer built
into it somehow.

T2
 
M

MrTsquare

I'd contact the motherboard supplier and tell them that batteries are
only lasting 3 months.

Good thought, Rodney. Will be interesting to see what they say - if
anything. I'm not about to RMA the board over this.

T2
 
M

MrTsquare

Computers as clocks are traditionally bad timekeepers. Not keeping
time is less a problem than losing enduser-set CMOS BIOS. A utility
to autoconnect every few minutes to an atomic clock relay is an easy
solution. (That's a bit radical, staying where the MB was last turned
off. A system clock off a few minutes a month is more usual to what
I've expect. Still, I wouldn't care as long as the BIOS settings
haven't changed, that the battery at least ensures they stay that way
for another 10 years.)

Never had this kind of issue before, Flash. The rest of the CMOS
setting seem ok, just that the time stops when the system is not
powered. Nope, wouldn't worry if I just had to adjust the time every
few months, but this one just stops dead and then starts where is left
off when the system is restarted.

T2
 
F

Flasherly

Nope, wouldn't worry if I just had to adjust the time every
few months, but this one just stops dead and then starts where is left
off when the system is restarted.

Me, neither - never heard of the likes. One of a kind.
 
P

Paul

MrTsquare said:
OK, turned it on tonite after leaving the power connected to the PS all
day since last nite. No change to symptoms. Still shows last night's
date and time. Guess I need to just feed it a 2032 every few months.

T2

You could connect a much larger battery to it.

But that's soldering iron / hardware hacker type stuff.

You need to use a battery with the correct voltage.
(Voltage can't be too high, or the battery ends up
"running the PC" all day :) )

You'd locate something like this, check out the maH
rating (i.e. to work out how many months it'll work),
double-check the voltage is no more than CR2032
working voltage (must be more than 2.3V min). Buy a
battery holder, a couple of wires. And be damn
careful not to short something. Solder battery holder
in parallel with CR2032 socket (access solder points on
back of motherboard.) That particular battery is
for running digital cameras with high drain, so has
a relatively good peak current rating. If you short
a battery like that, who knows, it might begin to swell.
I would probably add a small series resistor, like 100
ohms, right on the battery terminal output, so a
downstream short couldn't pop anything.

http://www.all-battery.com/cr123_indvidual.aspx

When I was a kid, and playing with batteries in the
basement, I managed to develop enough current flow
in dry cell batteries, to make the positive
electrode "swell" and push out of the battery
case. (Try putting about 30 batteries in series...)
You don't want to be doing stuff like that. The
reason nothing bad happened to me, is they weren't
Lithium.

I would think putting a fresh CR2032 in there,
and leaving the power switch on the back of
the computer in the ON position, should lengthen
the time it lasts. If it does not, that tells you the
fault is very close to the battery itself (socket,
1K resistor, or problem at or near the three-legged
ORing diode BAT45C or similar). If the fault is
further downstream, the ATX PSU takes care of it.

The only thing that really stands out in circuits
like that, is the SuperI/O hardware monitors that
have a VBAT pin, for monitoring the CR2032. That's
a possible place for draining the battery to happen.
The CLR_CMOS jumper is another candidate location.
Inspect for dirt or solder splashes or tin whiskers.
The rest of it, the CR2032 socket, 1K resistor,
ORing diode, probably not as likely.

Have you ever used the CLR_CMOS jumper before ?
If so, did you leave the system powered when doing
so ? On some systems, this damaged (burns) the three-legged
ORing diode. And your leak could be there...

Paul
 
M

MrTsquare

You could connect a much larger battery to it.

But that's soldering iron / hardware hacker type stuff.

You need to use a battery with the correct voltage.
(Voltage can't be too high, or the battery ends up
"running the PC" all day :) )

You'd locate something like this, check out the maH
rating (i.e. to work out how many months it'll work),
double-check the voltage is no more than CR2032
working voltage (must be more than 2.3V min). Buy a
battery holder, a couple of wires. And be damn
careful not to short something. Solder battery holder
in parallel with CR2032 socket (access solder points on
back of motherboard.) That particular battery is
for running digital cameras with high drain, so has
a relatively good peak current rating. If you short
a battery like that, who knows, it might begin to swell.
I would probably add a small series resistor, like 100
ohms, right on the battery terminal output, so a
downstream short couldn't pop anything.

http://www.all-battery.com/cr123_indvidual.aspx

When I was a kid, and playing with batteries in the
basement, I managed to develop enough current flow
in dry cell batteries, to make the positive
electrode "swell" and push out of the battery
case. (Try putting about 30 batteries in series...)
You don't want to be doing stuff like that. The
reason nothing bad happened to me, is they weren't
Lithium.

I would think putting a fresh CR2032 in there,
and leaving the power switch on the back of
the computer in the ON position, should lengthen
the time it lasts. If it does not, that tells you the
fault is very close to the battery itself (socket,
1K resistor, or problem at or near the three-legged

Gosh, Paul, its a wonder that you lived... ;<)

Sorry, not that much of a hardware hacker to try the add-in battery
trick. Never did use the clear-cmos jumper, just put it together and
ran it. Will look more closely at the battery area for whiskers etc
when I get around to replacing it (bought 4 more today. Any preference
for brand names for the 2032 or are they all pretty much the same?

T2
 
P

Paul

MrTsquare said:
Gosh, Paul, its a wonder that you lived... ;<)

Sorry, not that much of a hardware hacker to try the add-in battery
trick. Never did use the clear-cmos jumper, just put it together and
ran it. Will look more closely at the battery area for whiskers etc
when I get around to replacing it (bought 4 more today. Any preference
for brand names for the 2032 or are they all pretty much the same?

T2

Even with branded batteries, you don't know who
really made them.

Try to find a review for a named product, before
you buy.

I've bought batteries at the "Battery Store" in the
mall, and they were flat. There wasn't any charge
left in them a day later. So you can't even depend
on purpose-built suppliers to stock good stuff.

My recipe for digital watch batteries, was to
go to a department store with high levels of product
turnover. And I always seemed to get good batteries
there.

But as for brands, I don't really know what's out
there. I know some people swear by the batteries they
got off Ebay or the like. But I really don't know
a way to judge products like that. There is even less
control over who really made them. (You could have a
Panasonic branded battery, as a knockoff from China.
An Ebay seller wouldn't really care where they got
the battery from.)

Paul
 
M

MrTsquare

Just replaced the cmos battery. Examined the area of the "Clear RTC"
jumper with a flash light. Removed what looked like a piece of lint
going from the jumper to the board. We shall see what we shall see.

T2
 
M

MrTsquare

Even with branded batteries, you don't know who
really made them.

Try to find a review for a named product, before
you buy.

I've bought batteries at the "Battery Store" in the
mall, and they were flat. There wasn't any charge
left in them a day later. So you can't even depend
on purpose-built suppliers to stock good stuff.

My recipe for digital watch batteries, was to
go to a department store with high levels of product
turnover. And I always seemed to get good batteries
there.

But as for brands, I don't really know what's out
there. I know some people swear by the batteries they
got off Ebay or the like. But I really don't know
a way to judge products like that. There is even less
control over who really made them. (You could have a
Panasonic branded battery, as a knockoff from China.
An Ebay seller wouldn't really care where they got
the battery from.)

Paul

Would think that a 2032 is a 2032... standards and all. Guess not.

T2
 
F

Flasherly

My recipe for digital watch batteries, was to
go to a department store with high levels of product
turnover. And I always seemed to get good batteries
there.

But as for brands, I don't really know what's out
there. I know some people swear by the batteries they
got off Ebay or the like. But I really don't know
a way to judge products like that. There is even less
control over who really made them. (You could have a
Panasonic branded battery, as a knockoff from China.
An Ebay seller wouldn't really care where they got
the battery from.)

Wasn't Ebay, for sure, but awhile back -after checking out some local
"mark-ups," labeled for batteries, to my mind- I went online looking
for alternatives;- Likely Wallymart, though they do include installing
the battery, in case one feels like potluck with an irreplaceable, 1st
production Casio atomic-clock radio-reception wristwatch.

Ittybitty tinseywinsey watch batteries. Probably got six (5-yr.
projected life) in a blister pack for a local price (even called
around, they were all expensive).

Anyway it was a pretty big WEB outfit specializing in only batteries,
everything and everyone. I'm going back one of these days to tear
apart, with a Dremel, a disposable sealed toothbrush. Now, those are
expensive -- priced like Garmin's batteries.

Some battery stuff, pretty much don't work any other way. Need the
web and damn near a catalogue for cross-referencing all the different
ways and nomenclature they're labeled by.
 

Ask a Question

Want to reply to this thread or ask your own question?

You'll need to choose a username for the site, which only take a couple of moments. After that, you can post your question and our members will help you out.

Ask a Question

Top