Effect of failing CMOS Battery?

[Juan Wei]

From a cold start, my Gateway desktop boots to the Gateway (BIOS) splash
screen and offers <DEL> to enter Setup and F12 to enter Boot Menu.

It then sits there for a minute or more before I get "Starting Windows".

Someone suggested a low CMOS battery. Could he be right?

Wouldn't there be other symptoms if the battery were low?

The battery is about 2 y.o. and is constantly being charged.

 

[VanguardLH]

From a cold start, my Gateway desktop boots to the Gateway (BIOS) splash
screen and offers <DEL> to enter Setup and F12 to enter Boot Menu.

It then sits there for a minute or more before I get "Starting Windows".

Someone suggested a low CMOS battery. Could he be right?

Wouldn't there be other symptoms if the battery were low?

The battery is about 2 y.o. and is constantly being charged.

The EEPROM is not the copy of the BIOS used to boot the OS or effect
settings. THe EEPROM defaults get copied into CMOS memory (hence why it
is called the CMOS battery). It is the CMOS memory copy that gets used.
When you make changes to the BIOS settings, you are changing those in
the CMOS memory copy. Some boards allow users to copy back the changes
in CMOS into the EEPROM to make those the defaults or provide a
user-config table to load by selection instead of having to make all the
selections again; however, most copy from EEPROM into CMOS and it is the
CMOS copy that you edit and that gets settings used from there.

So if the CMOS battery is dead then there is nothing to energize the
CMOS chip to retain its values. That means on every boot you will get
the defaults loaded from EEPROM. If the defaults prompt for user input
than that's why you're getting the prompt.

CMOS batteries typically last 5 years before needing replacement. That
is from the date of *manufacture*, not from when you received a computer
with a CMOS battery that was sitting in someone's inventory for a couple
years or you bought off the shelf (get those only if they specified an
expiration date, not how many years they last since you don't know how
old is the battery that is dangling on the shelf). It's good
maintenance practice to replace the CMOS battery about every 3 years to
ensure it is strong enough in voltage to retain the CMOS settings. You
typically have about 1, or maybe 2, minutes to put in a new battery
after removing the old one before the charge drains off the CMOS chip.

A dead CMOS battery is also the cause of why your computer won't retain
the *BIOS* time and date. Your OS might sync later with an NTP server
to get back on time but it first takes the date and time from the BIOS.
The RTC (real-time clock) chip is usually the same chip as where is the
CMOS memory to hold the changed settings of the EEPROM copy.

 

[Paul]

From a cold start, my Gateway desktop boots to the Gateway (BIOS) splash
screen and offers <DEL> to enter Setup and F12 to enter Boot Menu.

It then sits there for a minute or more before I get "Starting Windows".

Someone suggested a low CMOS battery. Could he be right?

Wouldn't there be other symptoms if the battery were low?

The battery is about 2 y.o. and is constantly being charged.

CR2032 coin cells, are not charged by the computer.

If you unplug a computer, and provide no alternative
sources of power, the CR2032 lasts approximately three years.
(By calculation, it's slightly less than three years, but
I like nice round numbers.)

If the computer has +5VSB available, that takes the place
of the CR2032. If you leave the computer plugged in all the
time, and leave the switch ON at the back, then the available
+5VSB means the CR2032 won't get used at all. With no load,
the CR2032 lasts around ten years.

On a laptop, the main battery pack may be diode ORed in
a similar way. If the battery pack is present, it could
effectively be running the CMOS clock, when the laptop is
not plugged in. In a laptop, the CR2032 would only be
needed, when the user unplugs the main battery pack, and
no adapter is connected.

You cannot charge a CR2032. The manufacturer limits charging
currents to 1uA (micro amp), and the only reason the level is
that high, is to allow motherboard makers to use Schottky diodes
for an ORing diode implementation. At elevated temperature,
some Schottky diodes leak that much current, and so the battery
rating takes that kind of thing into account. I don't think it
was an accident that particular number was chosen. The battery
manufacturer would choose to set that number to zero, but then
it would make using the battery very difficult.

There are some battery types, charging the battery could
burst it. And for such batteries, the manufacturer may state
what the most charging you could apply to it, without
pressure building up in it.

The two diodes in the ORing diode circuit, picks the "source with the
highest voltage". If the computer power supply is delivering
+5VSB, the voltage at that node is higher on the computer side
than the battery side. And then no current flows out of the battery.

If the computer is completely powered off and put in storage,
the ORing diode then switches to drawing power from the CR2032.

A brand new CR2032 can be close to flat. There were some
batches of Asus motherboards, that shipped with weak batteries.
There have also been instances of motherboards, with much higher
than normal loading characteristics. A normal motherboard is
rated at around 10uA of loading, to run the CMOS 32KHz clock
circuit, as well as power the CMOS RAM. But there have been
motherboards, where leakage in the circuit was near the milliamp
level, and that flattened batteries in no time at all. To put
the 10uA computer drain in perspective, a typical digital watch
draws around 2uA to run a 32KHz time piece. Digital watch
batteries are smaller, but the current drain is smaller
as well.

*******

A dead battery, can stop some computers from starting.

Other than that, there should be no "slowdown" caused by
the battery. The CR2032 battery doesn't run the hard drive
motor or anything

*******

If you want to see the circuit around the CR2032, there is an
example here. See PDF page 18, lower left hand corner.

http://www.intel.com/design/chipsets/designex/BXDPDG10.PDF

Diode D3 and diode D2 are the ORing diodes. They select the
source with the higher source of voltage. The battery has
a 1K ohm series resistor, to limit current flow. Since the
battery never provides current while the computer is operating,
there is no need for a "stiff low impedance" power source.
When the CMOS clock is being relied upon, the current
drain is minimal. And the C15 (0.1 microfarad) capacitor
provides hold-up to transient current flow (when the ripple
counter transitions). When the computer is running,
diode D2 provides current flow, and diode D2 has
no current limiter in it. It is a "stiff" source of power.
D2 runs the CMOS, when the computer is in a running state.

The jumper JP17 in the sample schematic, provides a "safe"
implementation of the Clear CMOS function. You can also
set up JP17, so that it shorts the power source. And if you
then used a "bad" implementation of JP17, then the diode D2
gets burned by the level of current flow. This is why I always
give a warning to "unplug the computer when clearing the CMOS".
That is to provide some measure of protection against
a badly designed JP17 wiring. In that Intel example,
you don't have to unplug the computer, to use JP17.
And no matter how many times companies like Intel
show the other companies how to do it, they still
build the "bad" ones.

On that schematic page, you can also see the 32.768KHz
quartz crystal, and the two loading caps. That's what
keeps time while the computer is unplugged. The 32KHz
signal is divided by a ripple counter, to make a once
a second pulse. Counting those pulses is where the
time-of-day comes from.

Paul

 

[VanguardLH]

A dead battery, can stop some computers from starting.

Other than that, there should be no "slowdown" caused by
the battery. The CR2032 battery doesn't run the hard drive
motor or anything

What if the OP changed the BIOS settings to, say, change from default
settings to an optimal config? Or he changed the memory specs different
than the SPF rating read from the memory modules, like for overclocking?
Or especially if he changed the CPU's multiplier or clock rate? A dead
CMOS battery would result in loss of settings from the CMOS memory which
means on power up the EEPROM defaults would have to be used which could
end up with slower clocks and lower multiplier. If the previously
overclocked CPU and/or memory are now at default settings, the computer
will be slower.

The OP says "From a cold start, my Gateway desktop boots to the Gateway
(BIOS) splash screen and offers <DEL> to enter Setup and F12 to enter
Boot Menu." That is the typical prompt after the POST screen appears so
users can get into the BIOS config. Sounds like normal behavior.

Then the OP says "It then sits there for a minute or more before I get
"Starting Windows"." Uh huh, again, so what? The OP doesn't mention
how long it took before the CMOS battery died or how long it takes when
doing a warm reboot to get to that point in loading Windows. "Gateway"
doesn't say if this is a new or old desktop host. Gateway is a company
name with lots of models produced since 1985. Last spring trashed my
old slow Pentium MMX 100Mhz to check how long it takes before getting
past the POST, loading the OS, and until the OS decided to paint
something on the monitor. The OP never mentioned which version of
Windows painted the "Starting Windows" image to know the release year
for that version of Windows to know the maximum age of his "Gateway".
For the lack of details, the OP could be asking about a 25+ year old PC
in which the CMOS battery has never been replaced.

It is unclear from the OP's post if he is asking why he now sees the
POST screen and message about how to go into the BIOS settings or about
the time to get to the "Starting Windows" point in loading Windows. For
the first, if he didn't see it before and is seeing it now, sounds like
the BIOS reverted to default settings which means hiding the BIOS splash
screen (probably the POST screen) is not the default and instead a
setting was enabled before to hide that stuff but is now disabled by
default. For the second, we don't know what used to be the normal time
from a cold start to get to that point in the bootup sequence before the
CMOS battery died or when the OP does a warm reboot.

 

[Paul]

What if the OP changed the BIOS settings to, say, change from default
settings to an optimal config? Or he changed the memory specs different
than the SPF rating read from the memory modules, like for overclocking?
Or especially if he changed the CPU's multiplier or clock rate? A dead
CMOS battery would result in loss of settings from the CMOS memory which
means on power up the EEPROM defaults would have to be used which could
end up with slower clocks and lower multiplier. If the previously
overclocked CPU and/or memory are now at default settings, the computer
will be slower.

The OP says "From a cold start, my Gateway desktop boots to the Gateway
(BIOS) splash screen and offers <DEL> to enter Setup and F12 to enter
Boot Menu." That is the typical prompt after the POST screen appears so
users can get into the BIOS config. Sounds like normal behavior.

Then the OP says "It then sits there for a minute or more before I get
"Starting Windows"." Uh huh, again, so what? The OP doesn't mention
how long it took before the CMOS battery died or how long it takes when
doing a warm reboot to get to that point in loading Windows. "Gateway"
doesn't say if this is a new or old desktop host. Gateway is a company
name with lots of models produced since 1985. Last spring trashed my
old slow Pentium MMX 100Mhz to check how long it takes before getting
past the POST, loading the OS, and until the OS decided to paint
something on the monitor. The OP never mentioned which version of
Windows painted the "Starting Windows" image to know the release year
for that version of Windows to know the maximum age of his "Gateway".
For the lack of details, the OP could be asking about a 25+ year old PC
in which the CMOS battery has never been replaced.

It is unclear from the OP's post if he is asking why he now sees the
POST screen and message about how to go into the BIOS settings or about
the time to get to the "Starting Windows" point in loading Windows. For
the first, if he didn't see it before and is seeing it now, sounds like
the BIOS reverted to default settings which means hiding the BIOS splash
screen (probably the POST screen) is not the default and instead a
setting was enabled before to hide that stuff but is now disabled by
default. For the second, we don't know what used to be the normal time
from a cold start to get to that point in the bootup sequence before the
CMOS battery died or when the OP does a warm reboot.

I agree there is much left out of the description.

Computers with BIOS splash screens, usually default to having the
splash screen enabled. And then the user has to disable it. If the
battery is dead, I would expect to he seeing the splash screen again.

A machine like a Gateway, the default is likely to be "useful out of the
box". Meaning, the boot order probably matches the shipped hardware
configuration, or the like.

I know on my laptop, there is only one hardware setting, and
there is precious little to foul up on it. All I can do is change
between IDE and AHCI, and that's it. And I'm willing to bet,
the BIOS default is AHCI, to match the Windows 7 installed on it.

Paul

 

[John McGaw]

From a cold start, my Gateway desktop boots to the Gateway (BIOS) splash
screen and offers <DEL> to enter Setup and F12 to enter Boot Menu.

It then sits there for a minute or more before I get "Starting Windows".

Someone suggested a low CMOS battery. Could he be right?

Wouldn't there be other symptoms if the battery were low?

The battery is about 2 y.o. and is constantly being charged.

A failing battery can cause a number of odd symptoms, loss of clock time
among them. Some batteries last a long time (I've seen 6+ years) while some
fail rather quickly (a matter of weeks) and the only way to tell which is
which is to wait until the failure occurs and then to to check the
calendar. The biggest question is: why not make the minor effort to change
a dirt-cheap battery to see what happens knowing that it won't make
anything worse but which will eliminate one potential trouble area if it
doesn't actually fix the overall problem?

 

[VanguardLH]

I agree there is much left out of the description.

Computers with BIOS splash screens, usually default to having the
splash screen enabled. And then the user has to disable it. If the
battery is dead, I would expect to he seeing the splash screen again.

What I've seen is pre-builts come with a BIOS where they hide the POST
screen and instead show their ad banner screen. You get to see "Dell"
and no POST screen. Well, if it works one way it could work the other
way, like the default being to show the POST screen and the user who
doesn't want to see that technical stuff could hide it by selecting a
different boot screen. by the OP's description, maybe he was used to
seeing one boot screen but saw something else which was the default in
the BIOS.

A machine like a Gateway, the default is likely to be "useful out of the
box". Meaning, the boot order probably matches the shipped hardware
configuration, or the like.

But does that mean the BIOS is necessarily, by default, using SPF to
determine the spec'ed settings for the memory modules?

If the OP had overclocked the CPU, like upping its multiplier (for
unlocked CPUs), then the default for BIOS would use the lower multiple,
the CPU would run slower, so the OS would take longer to load. For
example, I had an Athlon XP 2800+ with a Barton core that was known for
overclocking and it worked for 6 years when upped to the equivalent of
an Athlon XP 3200+ (until the PSU went bad and took out the mobo). One
day I was wondering why everything seemed significantly slower. The PC
was still quite usable but not as snappy as before. Then I found out
the CPU multiplier in the BIOS had somehow got reset to the default.
The POST screen verified this. Upped the CPU multiplier in the BIOS
and, voila, the PC was like before.

I know on my laptop, there is only one hardware setting, and
there is precious little to foul up on it. All I can do is change
between IDE and AHCI, and that's it. And I'm willing to bet,
the BIOS default is AHCI, to match the Windows 7 installed on it.

I usually find laptops have fewer settings in their BIOS. Some laptops
are dismal regardding user-configurable settings. However, the OP said
he has a desktop and those often have a more robust BIOS. Usually but
not always. My Asus mobo had so many settings that it took me a lot of
online research to figure out most of them but there still are a few
that I don't understand or aren't sure about. My Acer Aspire's BIOS is
weak. I kept going into its BIOS figuring to find what I consider basic
BIOS settings but they weren't there.

My experience is that if the CMOS copy of the BIOS settings (defaults
from EEPROM and with later changes by the user) is corrupt that more
severe problems arise, like the user gets prompted about the corrupt
settings. Can happen when the battery is too weak or completely dead.
The hash of all the current settings doesn't match the old value or a
valid value and the user has to clear CMOS using the onboard jumper.
Rebooting doesn't get rid of the problem. Putting in a new battery
doesn't fix the corruption. The CMOS settings are corrupt (invalid or
mismatched) and have to be wiped by clearing them to force a reload from
the EEPROM copy. With a corrupted CMOS copy of settings, who knows what
artifacts can be exhibited by the host.

Since it sounds like the OP let the CMOS battery die and it has probably
been that way for awhile, he should replace the CMOS battery and also
clear the CMOS copy to get a fresh copy of the defaults from EEPROM into
CMOS and then go tweak the settings in the CMOS copy.

 

[Paul]

What I've seen is pre-builts come with a BIOS where they hide the POST
screen and instead show their ad banner screen. You get to see "Dell"
and no POST screen. Well, if it works one way it could work the other
way, like the default being to show the POST screen and the user who
doesn't want to see that technical stuff could hide it by selecting a
different boot screen. by the OP's description, maybe he was used to
seeing one boot screen but saw something else which was the default in
the BIOS.


But does that mean the BIOS is necessarily, by default, using SPF to
determine the spec'ed settings for the memory modules?

If the OP had overclocked the CPU, like upping its multiplier (for
unlocked CPUs), then the default for BIOS would use the lower multiple,
the CPU would run slower, so the OS would take longer to load. For
example, I had an Athlon XP 2800+ with a Barton core that was known for
overclocking and it worked for 6 years when upped to the equivalent of
an Athlon XP 3200+ (until the PSU went bad and took out the mobo). One
day I was wondering why everything seemed significantly slower. The PC
was still quite usable but not as snappy as before. Then I found out
the CPU multiplier in the BIOS had somehow got reset to the default.
The POST screen verified this. Upped the CPU multiplier in the BIOS
and, voila, the PC was like before.


I usually find laptops have fewer settings in their BIOS. Some laptops
are dismal regardding user-configurable settings. However, the OP said
he has a desktop and those often have a more robust BIOS. Usually but
not always. My Asus mobo had so many settings that it took me a lot of
online research to figure out most of them but there still are a few
that I don't understand or aren't sure about. My Acer Aspire's BIOS is
weak. I kept going into its BIOS figuring to find what I consider basic
BIOS settings but they weren't there.

My experience is that if the CMOS copy of the BIOS settings (defaults
from EEPROM and with later changes by the user) is corrupt that more
severe problems arise, like the user gets prompted about the corrupt
settings. Can happen when the battery is too weak or completely dead.
The hash of all the current settings doesn't match the old value or a
valid value and the user has to clear CMOS using the onboard jumper.
Rebooting doesn't get rid of the problem. Putting in a new battery
doesn't fix the corruption. The CMOS settings are corrupt (invalid or
mismatched) and have to be wiped by clearing them to force a reload from
the EEPROM copy. With a corrupted CMOS copy of settings, who knows what
artifacts can be exhibited by the host.

Since it sounds like the OP let the CMOS battery die and it has probably
been that way for awhile, he should replace the CMOS battery and also
clear the CMOS copy to get a fresh copy of the defaults from EEPROM into
CMOS and then go tweak the settings in the CMOS copy.

The hash on CMOS is pretty weak. It's just a simple checksum.
There is a checksum over the password fields. And a separate
checksum over the rest of it. It's a format developed eons ago
(for legacy BIOS). I don't know if UEFI changes that
substantially or not.

A corrupted CMOS is useless - so the BIOS should be
refreshing it right away.

Paul

 

[VanguardLH]

A corrupted CMOS is useless - so the BIOS should be
refreshing it right away.

If the refresh were automatic then the user would never see an error
message about "corrupted BIOS" ... but they do! I've seen hosts that
refuse to boot (the POST screen never shows up) until the user wipes
(clears) the CMOS which *then* loads the defaults from BIOS EEPROM.
Refresh of the BIOS EEPROM copy into the CMOS table is not always
automatic. Is it ever automatic? Corruption is not just the hash. It
could also be values that are supposed to be non-blank (there are no
"none" or blank choices) are missing or out of range. It all depends
the self-check routine.

 

[Juan Wei]

VanguardLH has written on 11/30/2013 10:31 PM:

The EEPROM is not the copy of the BIOS used to boot the OS or effect
settings. THe EEPROM defaults get copied into CMOS memory (hence why it
is called the CMOS battery). It is the CMOS memory copy that gets used.
When you make changes to the BIOS settings, you are changing those in
the CMOS memory copy. Some boards allow users to copy back the changes
in CMOS into the EEPROM to make those the defaults or provide a
user-config table to load by selection instead of having to make all the
selections again; however, most copy from EEPROM into CMOS and it is the
CMOS copy that you edit and that gets settings used from there.

Not sure why that is germane but it is interesting.

So if the CMOS battery is dead then there is nothing to energize the
CMOS chip to retain its values. That means on every boot you will get
the defaults loaded from EEPROM. If the defaults prompt for user input
than that's why you're getting the prompt.

You misunderstand, I think. Every PC I've ever seen shows on a screen
during POST or just after a list of keys to press if you do not want an
ordinary startup. If you do not press any key, eventually an ordinary
startup begins.

A dead CMOS battery is also the cause of why your computer won't retain
the *BIOS* time and date.

When did we decide that my computer won't retain BIOS and date? I don't
think I said that.

Your OS might sync later with an NTP server
to get back on time but it first takes the date and time from the BIOS.
The RTC (real-time clock) chip is usually the same chip as where is the
CMOS memory to hold the changed settings of the EEPROM copy.

So if I shut down my computer for 5 minutes and pull the Ethernet cable,
and then reboot, the clock should be 5 minutes slow, right?

 

[Juan Wei]

VanguardLH has written on 12/1/2013 12:38 PM:

What if the OP changed the BIOS settings to, say, change from default
settings to an optimal config? Or he changed the memory specs different
than the SPF rating read from the memory modules, like for overclocking?
Or especially if he changed the CPU's multiplier or clock rate?

He didn't do any of those things.

The OP says "From a cold start, my Gateway desktop boots to the Gateway
(BIOS) splash screen and offers <DEL> to enter Setup and F12 to enter
Boot Menu." That is the typical prompt after the POST screen appears so
users can get into the BIOS config. Sounds like normal behavior.

Aside from the fact that I don't get a POST screen because I have
"quiet"enabled, I agree.

Then the OP says "It then sits there for a minute or more before I get
"Starting Windows"." Uh huh, again, so what? The OP doesn't mention
how long it took before the CMOS battery died

Why are we assuming that the battery died?

or how long it takes when doing a warm reboot to get to that point in loading Windows. "Gateway"
doesn't say if this is a new or old desktop host. Gateway is a company
name with lots of models produced since 1985. Last spring trashed my
old slow Pentium MMX 100Mhz to check how long it takes before getting
past the POST, loading the OS, and until the OS decided to paint
something on the monitor. The OP never mentioned which version of
Windows painted the "Starting Windows" image to know the release year
for that version of Windows to know the maximum age of his "Gateway".
For the lack of details, the OP could be asking about a 25+ year old PC
in which the CMOS battery has never been replaced.

Sorry. It's a 2 y.o. Gateway DX4860.

It is unclear from the OP's post if he is asking why he now sees the
POST screen and message about how to go into the BIOS settings or about
the time to get to the "Starting Windows" point in loading Windows. For
the first, if he didn't see it before and is seeing it now, sounds like
the BIOS reverted to default settings which means hiding the BIOS splash
screen (probably the POST screen) is not the default and instead a
setting was enabled before to hide that stuff but is now disabled by
default. For the second, we don't know what used to be the normal time
from a cold start to get to that point in the bootup sequence before the
CMOS battery died or when the OP does a warm reboot.

The OP is asking why doesn't the computer respond in good time when he
presses DEL or F12, the keys that the BIOS suggests?

 

[Juan Wei]

John McGaw has written on 12/1/2013 3:29 PM:

A failing battery can cause a number of odd symptoms, loss of clock time
among them. Some batteries last a long time (I've seen 6+ years) while some
fail rather quickly (a matter of weeks) and the only way to tell which is
which is to wait until the failure occurs and then to to check the
calendar. The biggest question is: why not make the minor effort to change
a dirt-cheap battery to see what happens knowing that it won't make
anything worse but which will eliminate one potential trouble area if it
doesn't actually fix the overall problem?

As soon as the one I ordered gets here. I cannot easily run out to Radio
Shack/Staples/whatever as my mobility is somewhat limited.

 

[Paul]

Sorry. It's a 2 y.o. Gateway DX4860.

The OP is asking why doesn't the computer respond in good time when he
presses DEL or F12, the keys that the BIOS suggests?

(A DX4860 review. 8GB of DDR3.)

http://www.pcmag.com/article2/0,2817,2387922,00.asp

On my machine, the BIOS memory test has to complete,
before the BIOS key or the popup boot would work.
The BIOS memory test can be set to "quick" to save
some time. Which matters a little bit at the 8GB level.

Also, disk detection time is included in that.
I've had cases when five hard drives are connected
to the machine, and the "drive with the bad spot"
is one of the attached drives, where startup takes
longer. There is a pregnant pause when the bad drive
is accessed. The feedback on the screen is a dead
giveaway (long pause, then bad drive prints its
details on the screen).

On that drive, the funny thing is, my favorite two
SMART parameters remain at zero ("good drive"). And
yet, the sectors where C: used to sit, seem to be
damaged. Access in that area is slow, and a benchmark
of the drive showed a slow patch where the OS was.
Seemingly, it's almost as if the area where the heads
float normally, has damage underneath. No data was lost,
no SMART statistics show imminent death, but the drive
was slow enough, I replaced it. The new drive was
supposed to have spiffy performance for a hard drive,
but all it seems to do as a replacement, is make more
noise Still, the extra capacity I got is welcome.
Even if I'm still grinding a hole in the same
70GB space I use for C: .

Recommendations:

1) Verify what RAM test is being done by the BIOS
at startup. Try setting it to "Quick". Verify
that the computer still detects all RAM. You
could be down to 4GB, if the memory test has
invalidated the other RAM stick.

2) Observe screen, to see if there is a pause just
before the details of the hard drive detection
are printed on the screen.

3) Not mentioned above, but do an inventory of all
USB storage on the computer. Even an HP printer with
an SD card stuffed in the front slot, can constitute
a "USB flash", and interfere with booting. The review
above says the machine has "multiformat card reader"
and you can try unplugging the internal cable from
that card reader, then retest the boot time.

Note: the word "Battery" does not appear in my answer
It's not the battery...

HTH,
Paul

 

[miso]

On 11/30/2013 9:13 PM, Paul wrote:

Did you notice they don't use the 5VSB but rather 3VSB on that
schematic? That would make the diode-or a bit of a contest.

You know my theory about CR2032. They are all made in China now. The
quality is crap.

The notebook case is a little different, depending on if they use a
bridging battery or not. Notebooks are going with captive batteries, so
bridging isn't needed.

http://www.gizmag.com/lenovo-ifa-thinkpad-business-ultrabooks/28909/

Bridging really isn't new. It is more like back to the future.

 

[Paul]

On 11/30/2013 9:13 PM, Paul wrote:

Did you notice they don't use the 5VSB but rather 3VSB on that
schematic? That would make the diode-or a bit of a contest.

You know my theory about CR2032. They are all made in China now. The
quality is crap.

The notebook case is a little different, depending on if they use a
bridging battery or not. Notebooks are going with captive batteries, so
bridging isn't needed.


Bridging really isn't new. It is more like back to the future.

Yes.

For the diode ORing to work, the leg on the PC ATX power
supply side, has to have a higher voltage than the 3.0V
of the CMOS battery. A value of 3.3V would likely do nicely.
The schematic labels it 3VSB, but I don't actually know
if it is 3.0V or 3.3V. It would work a bit better with a bit
more margin. To help keep the diode cut off, that isn't
supposed to be conducting.

There is a reference schematic here.

The 3VSB in this one has signal name V_3P3_STBY. PDF page 82
has the diode circuit in the upper left hand corner. The
regulator that makes the 3P3 (3.3V) level is on PDF page 85.
The regulator is dual footprint. I looked up the MC33269
and using the equation from the data sheet, the adjustment
resistor pair is setting the regulator to 3.31 volts.

http://www.intel.com/design/chipsets/schematics/252812.htm

Which means the ATX power supply side, is at least 0.3V more
positive than the CMOS battery. Enough so the Schottky on
the ATX side will conduct, while the CMOS battery diode
is close to zero bias. And as the battery discharges, the
diode on the battery is more cut off (goes to reverse bias).

This is the Schottky used on three of my motherboards.
It has "K45" printed on top, and is the BAS40W-05.
The characteristic looks a bit better than a BAT54C.
It looks like the leakage would be below 1uA at zero bias,
on the CR2032 side. By extrapolating the curve in Figure 1.

http://www.diodes.com/datasheets/ds30114.pdf

Paul

 

[VanguardLH]

Why are we assuming that the battery died?

Because of the subject of YOUR post. What did *you* mean by "failing"?
To us it means the battery is too weak to sustain a proper voltage or it
is completely drained both of which means the battery "died".

It's a 2 y.o. Gateway DX4860.

The OP is asking why doesn't the computer respond in good time when he
presses DEL or F12, the keys that the BIOS suggests?

Oh, and here the rest of us figured it was the time to start the cold
boot and get to the "Windows starting" message. So that mention has
nothing to do with any measure of slowness of starting the OS which
could be the result of drivers waiting for hardware to respond. Instead
you want to know why it takes longer (and presumably with a dead battery
and not under a premise of what might happen) to get the POST screen at
the bottom of which tells you to hit the Del or F12 keys.

Did you config the BIOS or is it a default setting to test the memory?
With a lot a memory, and despite it's a simplistic touch test, it can
take a long time to check the memory. Look at the top of the boot
screen (usually after the BIOS name and version) to see if a counter is
incrementing as the BIOS goes through the memory to test it.

You mentioned the slow boot (now measured to when the POST screen
finally paints) from a cold boot. Is it longer or the same time to do a
warm reboot?

How long after you cold or warm boot when you hear the single beep?

Other than the mouse and keyboard, what USB devices do you have
connected to the computer when you boot? Have you tried disconnecting
them and then time the boot?

 

[VanguardLH]

VanguardLH has written on 11/30/2013 10:31 PM:

Not sure why that is germane but it is interesting.

You asked about the CMOS battery. It doesn't do anything for the EEPROM
where the BIOS is stored. That battery is to retain the settings copied
from EEPROM and possibly altered by the user. You asked about the CMOS
battery so I mentioned what it is for.

You misunderstand, I think. Every PC I've ever seen shows on a screen
during POST or just after a list of keys to press if you do not want an
ordinary startup. If you do not press any key, eventually an ordinary
startup begins.

Which are you talking about from the list below?

- Keys mentioned at the bottom of the POST screen itself.
- Keys shown a bit later by other hardware, like a RAID controller,
after the POST screen, often on their own screen, and before the BIOS
even tries to read the MBR to decide where is the OS to load.
- The startup menu for Windows where you can select which OS to load.
The recovery menu doesn't appear unless there is a problem with the OS
or the user has pressed a hotkey, like F8, to bring up that recovery
menu.

When did we decide that my computer won't retain BIOS and date? I don't
think I said that.

It was extra info whether you noted it or not as to what happens with a
"failing" battery which others and I assumed meant the battery wasn't
usable anymore (too low on voltage or dead) and equates to a dead
battery. You mentioned one symptom. That doesn't preclude other
symptoms that you chose on which not to focus. If the battery is dead
as per your Subject (where failing = dead) then you'll have to check the
OS clock after you boot into it or rely on the Windows time service or
an NTP utility to sync the time for you. Paul already mentioned that
besides the CMOS battery, the RTC (real-time clock) chip can also get
power from the 5VSB (5 volt standby) line from the PSU as long as the
PSU still gets power).

So if I shut down my computer for 5 minutes and pull the Ethernet cable,
and then reboot, the clock should be 5 minutes slow, right?

Nope. If the RTC chip is still getting power then it still keeps time.
With a dead, too weak, or removed CMOS battery, the PSU's 5VSB can still
be used to keep the RTC chip tracking the time lapse. If there is no
5VSB from the PSU and the battery is dead, doesn't produce sufficient
voltage, or is removed (and after a few minutes to a few hours to let
the capacitor drain on the motherboard), the RTC will *reset*. With no
power, the RTC chip will not be tracking time. Your computer's hardware
clock will not be slow. The reset (on power up) means the date and time
will jump back to an initial or base value encoded in the BIOS. It
might, for example, jump back several years. Without power, the RTC
chip has nothing against which it can base its time value. It goes back
to scratch. It all depends on what the BIOS thinks is the starting
timestamp for a time lapse of zero ticks.

 

[VanguardLH]

As soon as the one I ordered gets here. I cannot easily run out to Radio
Shack/Staples/whatever as my mobility is somewhat limited.

Then, as Paul first mentioned, keep the power cord from the PSU plugged
into the wall (which should not be tied to a wall switch) and hope you
don't have a power outage or you trip the breaker. In an ATX PSU and as
long as it is connected to power (house wiring or UPS), it will provide
5V standby to the motherboard which can keep the RTC chip powered.

 

[Juan Wei]

Paul has written on 12/1/2013 8:27 PM:

(A DX4860 review. 8GB of DDR3.)

http://www.pcmag.com/article2/0,2817,2387922,00.asp

On my machine, the BIOS memory test has to complete,
before the BIOS key or the popup boot would work.
The BIOS memory test can be set to "quick" to save
some time. Which matters a little bit at the 8GB level.

Also, disk detection time is included in that.
I've had cases when five hard drives are connected
to the machine, and the "drive with the bad spot"
is one of the attached drives, where startup takes
longer. There is a pregnant pause when the bad drive
is accessed. The feedback on the screen is a dead
giveaway (long pause, then bad drive prints its
details on the screen).

I don't get any message about bad drives.

Recommendations:

1) Verify what RAM test is being done by the BIOS
at startup. Try setting it to "Quick". Verify
that the computer still detects all RAM. You
could be down to 4GB, if the memory test has
invalidated the other RAM stick.

2) Observe screen, to see if there is a pause just
before the details of the hard drive detection
are printed on the screen.

3) Not mentioned above, but do an inventory of all
USB storage on the computer. Even an HP printer with
an SD card stuffed in the front slot, can constitute
a "USB flash", and interfere with booting. The review
above says the machine has "multiformat card reader"
and you can try unplugging the internal cable from
that card reader, then retest the boot time.

Note: the word "Battery" does not appear in my answer
It's not the battery...

Thanks.

 

[Juan Wei]

VanguardLH has written on 12/2/2013 9:41 AM:

Which are you talking about from the list below?

- Keys mentioned at the bottom of the POST screen itself.
- Keys shown a bit later by other hardware, like a RAID controller,
after the POST screen, often on their own screen, and before the BIOS
even tries to read the MBR to decide where is the OS to load.
- The startup menu for Windows where you can select which OS to load.
The recovery menu doesn't appear unless there is a problem with the OS
or the user has pressed a hotkey, like F8, to bring up that recovery
menu.

"Keys mentioned at the bottom of the POST screen itself."

Nope. If the RTC chip is still getting power then it still keeps time.
With a dead, too weak, or removed CMOS battery, the PSU's 5VSB can still
be used to keep the RTC chip tracking the time lapse. If there is no
5VSB from the PSU and the battery is dead, doesn't produce sufficient
voltage, or is removed (and after a few minutes to a few hours to let
the capacitor drain on the motherboard), the RTC will *reset*. With no
power, the RTC chip will not be tracking time. Your computer's hardware
clock will not be slow. The reset (on power up) means the date and time
will jump back to an initial or base value encoded in the BIOS. It
might, for example, jump back several years. Without power, the RTC
chip has nothing against which it can base its time value. It goes back
to scratch. It all depends on what the BIOS thinks is the starting
timestamp for a time lapse of zero ticks.

IOW, if I have a dead battery and shut the computer down for 5 mins and
pull the Ethernet cable before I restart (so it won't get updated time
info from the Internet), then when I restart the displayed time/date
will be different from actual?