M/B was new in 2011 and battery died in 2017 so I replaced with a Duracell one that said 'made in Indonesia'.
It has now failed, voltage down to 1.3 V, so I have put the other one of the packet pair in.
Shouldn't they last longer than this ?. I might just have will connected up an external battery holder with a pair of flying leads to make it easier to change without losing settings.
what was the shelf-life date on it?
Does the PC spend most of its time turned on, "soft" turned off but plugged-in, unplugged or physically turned off by a hardware switch
I'd expect the 2017 battery to have lasted as long as the original one. BUT, the life of the CMOS battery may well depend on how much time the computer is powered from the mains. Depending on the M/B design it may well not draw any power from the battery when powered up. So, possibly, you kept the PC powered up 24/7 to begin with, but later on changed that?
Or possibly Duracell from Indonesia are to be avoided? I think we probably need much more information. :)
best before 2026, and the other one, just installed was in the freezer with my battery and E6/B&W film stash.
PC is typically on from about mid morning to late evening then hard off (switched off at wall socket). Only rarely will it be off for more than a day or two.
I think I will get some ?Renata 2032's just in case, and some silver oxide ones for my trusty Pentax LX. Renata are a Swiss company with a good reputation, from what I gather.
I hope it isn't a Win 10 issue ??
No, I haven't changed my usage significantly over the last 10 years or so.
I always though they were made in Belgium. I'll pay more attention to the small print in future.
The O/S hasn't got anything to do with it, if the PC is on, the power to the CMOS/RTC should be from the PSU (probably the 5VSB rail) same if it's sleeping, or "soft off", but when it's hard off the power has to come from the battery.
I've seen suggestions that modern motherboards don't really store stuff in CMOS any more, they just pretend to, but actually dump it into flash memory, then reload it from flash into CMOS at power on, so the battery could just be powering the RTC.
It's a Gigabyte 880-GM-UDH2 Rev 1.4 with an Athlon 1055 6-core
Most of the BIOS settings didn't change after the battery change. Only the date/time and some of the thermal/fan settings needed resetting.
Pressing DEL at bootup shows options to save the BIOS to disk. One of its features is to keep a bios copy which it can reload if it thinks it has been attacked.
I have never changed any battery in my PC. I bought it soon after Windows 7 was introduced - probably 2010. Is there anything I should be checking or testing?
Unplug it at the wall for a couple of hours, when you turn it back on, if it says "CMOS error, press F1 to continue" then your battery should have been replaced earlier :-)
I bought my PC in 2010 (that's the date the Speccy gives as the OS installation date) and it's still on its original CMOS battery. But my PC is permanently connected to mains - either on 24/7 or else put to sleep (not hibernate). It says "sleep" but there's a pause of about 90 seconds between initiating so-called sleep and the power going off, so I wonder if it's saving the system state to disc (which is hibernation) rather than just keeping power to the RAM (sleep).
So my CMOS battery doesn't get much usage if the mains is supplying CMOS power.
Come to think of it, probably the only times the power has been removed was power cuts, moving house and when the PSU failed during a thunderstorm: I had a spare and I had the PC running again within about 10 mins including extricating the PC from under the desk, taking the side off, removing a couple of drives that were blocking access to the CPU power sockets, replacing the PSU and then refitting everything. Once I'd proved that the PC worked (ie that it was *only* the PSU that had failed) I unplugged it until the storm had gone over. I'm glad I did because shortly afterwards there was a lightning strike on a pylon just down the road and that fried several of my neighbours PCs and blew a couple of our CFL light bulbs.
Sounds like hybrid sleep/hibernate.
It saves memory to disc, but then just goes to sleep, so long as power stays ok it just wakes up when you ask it to and continues, but if power is lost while it's asleep then when you power it back on, it's as though its coming back from hibernation ... either way nothing should be lost.
I'm not sure how much they enjoy -18C. That might be enough to part freeze the electrolyte and damage the seal. How long they last is often a function of how carefully you handle them. If you bridge the insulation between + and - with your fingerprint they die much sooner.
They keep OK at lowish room temperature cool and dry.
Switched off at the wall will stress the battery more. Most decent PCs these days have a very low <1W standby so it isn't worth the hassle.
Many PC's will boot with a completely dead CMOS battery these days. You might have to go round the loop twice and let it autoidentify the HD but it isn't necessarily a show stopper any more. Internet time sync and autodetect of hardware coupled with many BIOS's using EEPROM and copy back into CMOS has made the RTC the only thing a flat battery really affects. That can be fixed by doing an internet time sync.
The CR2032 battery has an ampere-hours rating. You take that number, and divide by the 10uA the CMOS RTC 32768 Hertz clock draws. The end result, is a number that is slightly less than 3 years.
Now, the question is, how will we "spend" our three year runtime.
Let's take an example. I unplug the PC, put it in the closet for 3.5 years. I take it out and uhoh, the battery is completely flat. OK, so I used up my three years of maintaining that digital watch circuit inside the PC, and it's flattened.
Now, let's take the case where the user selects "shutdown" at the end of every day, yet that user leaves the switch on the back of the PC in the "ON" position. The ATX PSU has a +5VSB regulator that operates without the benefit of fan cooling. Voltage from that rail, preferentially powers the RTC circuit. The CR2032 is not used. How long does the battery last now ? It's closer to the shelf life of the battery, 10 years. By rights, it should last longer than 10 years.
Some PC users (me) are inveterate fiddlers. We end a PC session, and turn off the switch on the back of the PC ("to prevent Windows 10 from waking up at 4AM and messing around").
But, by operating the equipment that way, all of the time spent with the switch "OFF", is time that comes out of our three year runtime on the CR2032 battery.
It really depends on your "ON" versus "OFF" habits, as to whether it takes 6 years to use up the 3 years runtime, or only 3 years to use it up. The battery must last for 3 years, under any circumstance, as the watch circuit near the Southbridge only draws 10uA and cannot drain the battery any faster than that. The circuit uses a ripple divider to minimize the complexity of the divider chain, and the "seconds" counts that engage only once a second, don't use that much power. The ripple divider chain, only the first DFF runs at 32KHz, the next runs at 16KHz and so on.
The 10uA figure for this implementation, is about 5x higher than an actual digital watch. But it must be remembered that this circuit is in the middle of a VLSI (Southbridge, many I/O on the bottom), and the leakage characteristics of the logic gates in devices like this, may not be as "optimal" as your digital watch. With the digital watch case, I can construct the chip with 3 micron geometry silicon, and reduce the leakage currents to zero. The Southbridge is a much smaller geometry (65nm, 32nm, or smaller), and more of a pig.
The RTC sits in the "CMOS well", separated from the other parts of the chip with transmission gates, and these are intended to reduce leakage along I/O paths when the rest of the Southbridge is de-powered. At shutdown, most of the SB power is cut off (four rails or so), but VBAT or similar runs to the CMOS well, and all the transmission gates go open-circuit, so the "well" remains a "well" and is isolated and not wasting battery life.
At one time, PCs relied on a separate chip for timekeeping, a chip that had a battery strapped to its back. And that was an opportunity to make a more optimal solution. Until you wanted to change the battery... and you couldn't. Back in those days, it was Dremel time :-)
In those days, you would see an infinite number of homebrew battery repair articles, to replace the flattened Dallas chip. I completely missed this era. No machines here with these in it ...
Paul
They should last a lot longer and there should be no reason that a good battery should last any less than the original.
I've not had any problems with the Duracell 2032 batteries that I've purchased in the past but have had problems with a batch of Maxwell "branded" (probably fake) 2032 batteries. In Christmas LED T lights the maxwell batteries lasted 6 hours while the Duracell lasted around 100 hours.
The Amstrad PC's used a pair of AA batteries you could access just by lifting off the monitor (and if quick enough, without losing the settings, assuming they weren't already lost and why you were changing the batteries of course). ;-)
Cheers, T i m
They have a shelf life of 10 years and last 5 to 10 years in circuit. Dont fit one with less than 3 to 5 years shelf life unless its your only one and you plan to replace it in a year or 2 with a fresh one. Only keep the real-time clock running in a modern motherboard so failure is no longer a biggie.
Yup - I was surprised to find the CMOS battery flat on a desktop. Then realised it had been unplugged for quite a long time.
A few systems won't start if that battery is flat. That's when you'll begin to care about it. None of my systems here have exhibited that behavior (as just about every computer here has run the CR2032 battery flat at some point).
VBAT is sometimes an input on the SuperI/O, but I've never been able to find a reason why. Or whether it's monitored at all. (It gives the appearance that perhaps an ADC measures the value of that voltage, but I couldn't find any details in the documentation. It never shows up in the BIOS Hardware Monitor as a table entry. The SuperI/O has one ADC with a mux in front, able to measure one of nine voltages at any point in time. Some BIOS tables only list three monitored values.)
There aren't supposed to be a lot of electrical connections to that, as a means to prevent electrical leakage.
Paul
I suppose it depend on spec. If you have more than one HD, the CMOS won't remember which one to boot from. But doesn't exactly take long to reset the date and time, and tell it which one is needed.
HomeOwnersHub website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.