CMOS batteries (e.g. CR2032) in PCs

[mikeymikec]

This is my understanding:

CMOS information including date and time is stored with the aid of the CMOS battery. My understanding is that the CMOS battery gets used as the power source for storing CMOS information when there is no other power source (e.g. a laptop's normal battery, its mains supply, or a desktop PC's live mains connection). In my experience, CMOS batteries can last a decade or so in a desktop PC if the PC has a live mains connection, whereas if the owner is in the habit of shutting down the PC then switching off its mains connection, the battery tends to only last about 5-6 years.

Is this correct? I'm encountering a scenario with a customer's PC that challenges my belief, in that it developed the typical issue of losing time/date and giving CMOS errors on boot-up (and sure enough the customer had been disconnecting the mains after shut down), so I replaced the battery, but now the PC is apparently not keeping time after resuming from sleep mode or being shut down (despite the live mains connection).

 

[Paperdoc]

I believe your understanding is correct, with one small correction. The battery power is required to MAINTAIN the data in the BIOS PROM, not to store it there. Storage and retrieval of that BIOS data still requires an operating mobo with power.Battery power also is used (when other power is not available) to operate the clock and keep it correct when other power is not available.

What you describe is certainly odd. IF the "new" battery is a good one with proper output, BOTH the clock and the BIOS Setup options should be maintained when the system is DISconnected from the wall outlet. NOTE that IF the system is still plugged into a live outlet, even though the system is "Off" (or "sleeping"), there IS power from the PSU to that BIOS chip to maintain its functions, and the battery is not used. I am not sure, but if the system IS plugged into the wall BUT the user has switched OFF the main power switch in the back if the PSU, it may NOT be supplying power to the BIOS chip.

In addition to a problem with the clock, a failed battery also should cause loss of any custom settings in BIOS Setup. Such loss would cause BOTH the clock time AND those BIOS settings to be replaced with factory default values on the next boot. Note that failed battery power shuld result not in a clock being merely slow, but a complete reset to a very early default date / time. So, in this machine are there any non-default settings in BIOS Setup that can be checked to see if they are being lost when power is off? If not, I suggest you deliberately change a couple Setup options so you can test whether they are lost when power is off. If they are, (as well as losing proper time) then the new battery is not doing its job. Either it is too weak, or has a bad connection, or is installed backwards, OR the jumper on the mobo RESET pins is in the wrong position.

 

[mikeymikec]

Thanks for the response. I'm curious to see whether anyone has anything else to add. My first thought is that the replacement CMOS battery is defective in some way, but I would have thought that the live mains connection ought to offset that. I'm also curious to know what would happen if a system had no CMOS battery, just a live mains connection the entire time, whether it would maintain CMOS settings/time while shut down; though I imagine that there would be some BIOS code that does some kind of integrity check on the battery regardless of whether the CMOS checksum checks out.

Good idea with the custom setting though. As a side note, I'm kicking myself for using one of the customer's stash of batteries though they assure me it was new in all senses of the word. I tend to keep a stash with me too but I've never felt the need to wonder how old my stash is.

 

[VirtualLarry]

I'm kicking myself for using one of the customer's stash of batteries though they assure me it was new

You forgot the Golden Rule in Tech Service... NEVER believe the customer outright.

 

[Steltek]

Thanks for the response. I'm curious to see whether anyone has anything else to add. My first thought is that the replacement CMOS battery is defective in some way, but I would have thought that the live mains connection ought to offset that. I'm also curious to know what would happen if a system had no CMOS battery, just a live mains connection the entire time, whether it would maintain CMOS settings/time while shut down; though I imagine that there would be some BIOS code that does some kind of integrity check on the battery regardless of whether the CMOS checksum checks out.

Good idea with the custom setting though. As a side note, I'm kicking myself for using one of the customer's stash of batteries though they assure me it was new in all senses of the word. I tend to keep a stash with me too but I've never felt the need to wonder how old my stash is.

Definitely check the batteries with a multi-meter. I've bought several packs of different major brands last year and this year. Even though they all had distant future expiration dates, more than half of every pack was simply defective with no or very little charge.

I'm suspecting that someone has a major counterfeiting operation going somewhere for them.

 

[mindless1]

Any chance the clear CMOS jumper is in the wrong position? Otherwise, does seem like a bad battery.

 

[Steltek]

The only other thing I could think of is that someone has previously used the CMOS jumper while the computer was operating and damaged the CMOS memory on the motherboard.

I have seen this a few times in the past from owners that try this procedure without realizing you aren't supposed to do it with the machine powered up.

 

[mikeymikec]

I took another look today. As suggested, I changed a BIOS setting, switched off the computer, switched off the power, gave it a couple of minutes and switched it back on again. While it maintained the changed setting, the clock timer stayed at the same time as when it was switched off. The same thing happens in sleep mode, the clock keeps the time of when the computer was put into sleep mode, and the event log resume entry shows the sleep time and the wake time as identical.

The customer doesn't seem to be entirely sure if the clock is losing time while the PC is running. I suggested disabling auto sleep mode in Power Options, letting the computer run overnight and see whether there's any skew... though come to think of it, it's still set to auto sync. I can ring him to disable that setting

The customer didn't want me to switch the battery again, he says the problem doesn't bother him greatly. We will probably revisit this problem in the near future.

 

[mikeymikec]

I finally had another look at this problem today (the customer took a while to agree to another visit on this topic, I think they hoped it would go away by itself), swapped the battery out for one of mine (bought recently), apparently problem solved. The customer is going to keep an eye on it because they believe it's not 100% consistent in its presentation.

CMOS jumper - when I had the computer open today, I found the CMOS jumper, but as one might expect from HP, it can't be a simple 3-pin jumper but 6-pin because another jumper position is the 'password reset' jumper. As the pins weren't labelled in any meaningful fashion and because I couldn't find a quick result online to tell me which position is which (the 6 pins were in two rows of three), I decided to leave it alone.

When I booted the computer back up, the first thing it complained of was an invalid CMOS checksum which I'll take to be as good as a jumper pin change to effect it myself.

I did two shutdowns with the computer off for 5 minutes in between, it maintained the clock timing properly on both occasions.

 

[tcsenter]

Could be they mistakenly used CR 2025 instead of 2032. They are same voltage and diameter but 25 is thinner (and less capacity), might not have been maintaining good contact in the holder.

 

[mikeymikec]

Could be they mistakenly used CR 2025 instead of 2032. They are same voltage and diameter but 25 is thinner (and less capacity), might not have been maintaining good contact in the holder.

The battery that I replaced yesterday was labelled CR 2032.

 

[William Gaatjes]

Hi, i would like to share some humble tips learend from experience accompanied with tears and fears to rule out different possible scenarios :

Use an ESD wristwrap, make sure to touch the ground plane of the motherboard always first, ore the chassis. This way you and the system have the charge level and will prevent sparking.
The cmos jumper has pins that can be ESD sensitive and may cause damage to board wich can lead to unusual random faulty behavior or just constant issues. Always think of ESD when handling ESD sensitive material.

If it is possible, always measure the voltage of the battery, the old one and the new one.
When measuring the old one, you know for certain that the battery is low on charge or defective or just good. Measure it in place in the battery holder and when taken out of the batteryholder.
Measuring the new one ensures and validating that is has proper charge and output voltage.

Please check for corrosion or oxidization on the batterij pins of the battery holder. Use a glassfiberpencil to clean the contacts when it is the case. That works great.
Please check for loose solder connections between the battery holder and the motherboard.
This all gives certainy and therefore convidence.

Definitely try to find out what the jumper settings may be.
It wil not be the first time, some jumper has by accident been placed in the discharge position to clear the cmos ram by the customer or a friend or some random person being helped.
Sometimes it is forgotten. Like brainfog situations or stress, haste.

When you set the jumper to clear cmos, some scenarios are present :
you actually either disconnect the battery from the cmos ram storage chip power supply rail.
And you short the cmos ram storage chip power supply rail with the jumper. The jumper for example connects to some relatively low omhmic resistor in the 100 ohms range or that is placedin parallel with the cmos ram storage chip power supply rail. Therefore putting a load on the supply capacitors and discharging them, there is no more voltage and the cmos ram wil forget the data because all the bits get randomized.

What happens when the bits get randomized ?
You can be pretty sure that in the cmosram , there is also a calculated cheksum present that is used to verify and validate the coms ram data. If the checksum is wrong, it means the cmos data is corrupt. Then default settings stored in the bios rom wil be used and the computer will display a messga that there is a cmos error and the time will be reset too. This is al done when the computer boots up and runs program code from the bios before executng the bootloader and then the Operating system.

So to come back to the cmos jumper in the wrong position. This can be a viable suspect for the problems if everything else is ruled out.
Please check for how it the jumpers are related and connected.

Manual may help or call the motherboard vendor ? email or a question message on a support forum ?
Good luck.