Question windows time keeps falling behind

[pmv]

The time reported by Windows keeps falling behind. I changed the CMOS battery (the original battery was over 10 years old so that seemed a likely cause of the problem) but it hasn't made any difference.

What puzzles me is I never seem to catch the clock running slow - I set it from the internet (via the settings) and every time I watch it subsequently it correctly keeps the time. Till I stop watching it and paying attention to it, and then when I look at it again it's a couple of hours slow. Not sure if it's actually jumping backwards, or if it runs correctly for a while then goes slow. Haven't managed to catch it jumping backwards, indeed when I monitor it closely it seems to be keeping perfect time, it only seems to go wrong when I'm not looking!

Does the windows clock run on it's own recognizance, and just periodically sync itself with the bios clock? Is that how it works? That might explain it, if the bios clock is losing time, so the windows clock will only go wrong when it syncs with that bios clock. All I can think of trying is changing the CMOS battery yet again.

 

[Jimminy]

Check to be sure you didn't install the battery backwards ... easy to do. But I would be surprised the bios didn't demand you set the time whenever you boot it.

 

[WilliamM2]

Does the windows clock run on it's own recognizance, and just periodically sync itself with the bios clock? Is that how it works? That might explain it, if the bios clock is losing time, so the windows clock will only go wrong when it syncs with that bios clock. All I can think of trying is changing the CMOS battery yet again.

No, it syncs with the windows time server you set it to. It only relies on the bios when the computer is off, and even if it was off, it would re-sync once booted. Unless you turned sync automatically off.

Are you sure it's set to the right time zone? Check control panel, and then time and region.

 

[pmv]

Check to be sure you didn't install the battery backwards ... easy to do.

Hmm, I guess that's possible. Need to check that thing again. Just puzzled how inconsistent the wonky clock seems to be. Can't figure out what are the circumstances in which it goes wrong. It's as if it knows when I'm watching it and checking up on it, so only jumps back (or stops?) when I'm not watching it!

Time zone/region seems to be correct.

I notice in the event log some messages that say

"The time provider 'VMICTimeProvider' has indicated that the current hardware and operating environment is not supported and has stopped. This behavior is expected for VMICTimeProvider on non-HyperV-guest environments. This may be the expected behavior for the current provider in the current operating environment as well."

 

[In2Photos]

What version of Windows? Can you post a screen shot of your time settings?

 

[pmv]

Oh, Windows 10, sorry.

I've found what happens is:

I set the time from the web, using the settings adjust date/time and clicking 'sync now' (with time.windows.com).

Then it seems to keep time correctly for quite a while (doesn't run slow or anything like that), till it suddenly jumps back, to the exact time it was when I synced it with the time server. The event log says so, stating it changed the time and "Change Reason: System time synchronized with the hardware clock."

Then, if I boot into the BIOS it shows the current BIOS clock time...which seems to be stuck at the time I last did a clock sync within windows. Is this consistent with a failed CMOS battery? I changed the battery, but I suppose it's possible the replacement battery is bad as well. Seems strange that a purely digital clock can "stop", almost like a mechanical one.

 

[mikeymikec]

@pmv

My recent experience on this topic:

CMOS batteries (e.g. CR2032) in PCs

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...

forums.anandtech.com

My take-away is that faulty batteries can cause some really screwy issues. I think it's made more complicated by the fact that modern operating systems don't always update the hardware clock because of time zones and daylight savings etc. It's made more complicated by dual-booting too because both operating systems try to be 'helpful'.

I'd replace the battery if I were you and check the connections.

 

[pmv]

@pmv

My recent experience on this topic:

CMOS batteries (e.g. CR2032) in PCs

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...

forums.anandtech.com

My take-away is that faulty batteries can cause some really screwy issues. I think it's made more complicated by the fact that modern operating systems don't always update the hardware clock because of time zones and daylight savings etc. It's made more complicated by dual-booting too because both operating systems try to be 'helpful'.

I'd replace the battery if I were you and check the connections.

Yeah, thanks, that thread seems relevent/interesting. To be honest I groan a the thought of getting at that CMOS battery again - means moving everything off a crowded desk, and the damn thing is quite tricky to get at without knocking the CPU heatsink, and I'm not sure I've figured out how the battery catch clip thing works. But I guess that's the only way forward. Got replacement cr2032s from the pound-shop, so they might be bad as well.

Just find it really odd that rather than 'running slow' or something, the time in windows just periodically gets reset so as to jump backwards, apparently to match the 'hardware clock' - and, as far as I have been able to tell, that 'hardware clock' doesn't so much run slowly as just get stuck unchanged at whatever time it gets set to when I sync the windows clock with the internet time source. I hope it's the battery and not a weird motherboard fault (could the 'hardware clock' just simply stop working?).

 

[Red Squirrel]

This does seem like a weird one, I would just change the cmos battery to see if it makes a difference. I had a PC that used to randomly reset it's clock to a very specific date, like, a recent date, not the unix epoch or anything like that, but like a date that was within the year, and it was always the same one. That PC was not connected to the internet. I decided to switch the battery out just to see and it solved the issue.

For fun could also setup a time lapse camera pointed at the computer clock and a separate real clock, try to see if it does it suddenly or slowly, and if it's always at same time. Could help rule out some kind of script that's running or something.

 

[tcsenter]

I buy 20 pack of Japanese or South Korean brand CR2032 for like 50~60 cents each. If I acquire used mobo or PC that is older than two years I always toss a fresh one in for the heck of it.

 

[pmv]

Finally seem to have fixed it - seems as if every one of the CR2032 batteries I got from the poundshop were no good. Finally tried one I've had lying around here for ages that I got with a batch off of Ebay years ago, and the problem seems to have gone away.

Seems that a failing real-time clock causes a lot of strange problems (e.g. web based emails getting truncated or disappearing, not to mention this site not working).

 

[pmv]

Heh, spoke too soon. Having changed the CMOS battery three times, it still isn't working. Seems it's not a windows issue at all, but a motherboard one. Basically the hardware clock on the motherboard, ie in the bios, appears to be 'stuck'. Not running slow but just not running at all and just frozen at whatever time I set it to from within windows. Hence windows keeps time OK...except every now and then it decides to 'sync with the hardware clock' at which point it jumps back to the time that clock was last set to when I last synced with the windows on-line time server, because the hardware clock stays stuck at that time.

I found several threads of people with the exact same issue with this mother board, though all of those threads seem to date back to when it was a new board circa 2014. For some reason the problem has, for me, only appeared after a decade. One comment said something about the crystal (was that the word? I forget) used by the clock on the motherboard having failed. Frustrating, can't think of a way to fix it without replacing the motherboard entirely.

 

[tcsenter]

Could it be a resume from S3 ~ S5 problem. Is this an AMD processor? Maybe HPET issue?

 

[pmv]

No, it's an intel cpu. There seem to be various old threads talking about the z87 "clock freeze issue" as if it's a known problem (or was, as this is rather an old motherboard now). The only suggested solutions I found were to reflash the bios and clear the Cmos, but I ran into further problems when I looked at reflashing the bios (it said I first needed to update Intel manager something-or-other, and then when I tried to update that it told me some component was already newer than the 'new' version I downloaded from ASUS's site, and then after I reinstalled all the chipset drivers it decided to instead complain that the hardware was "incompatible", and then I ran out of energy to deal with it - currently I just manually correct the clock again every time it jumps back in time, but it's quite annoying).

e.g. (I'm not overclocking, this started after a hard disk crash and replacing that drive...really struggling to find the energy to deal with it though).

Asus Z87 MOBO Frozen Bios Clock Issue - Windows 7 Help Forums

Can someone please inform me is this an issue that occurs only when any kind of overclocking is done. This includes enabling XMP profiles of RAM that has XMP feature. In other words is there anybody o

www.sevenforums.com

 

[pmv]

Hmm, seem to have fixed it, finally. But can't really understand why or how. What seemed to fix it was going to "services", noticing that "windows time service" was "stopped" and set to "manual trigger" and telling it to start and changing it to "automatic trigger". Then the windows clock seemed to be fixed for a while... till suddenly the time jumped back again (and the event log, as always, demonstrated the causal event was windows deciding to 'synchronise system time with hardware clock').

But then after a windows shutdown and restart, it seemed to maintain the correct time indefinitely without that happening again.
Unless the problem comes back again the next time I 'hibernate' it.

I don't understand this stuff - the "windows time service" says it's purpose is to "synchronise the time between all clients and servers on the network", but _what_ network? What servers? This is a stand alone PC with just a router and an internet connection, so what does that mean in that context? Does a windows PC count as its own client and server?

Is it possible the hardware clock on the motherboard has always been broken (i.e. a motherboard fault - I see some claims on-line that this ASUS motherboard can have a faulty hardware clock problem from new) and I've never noticed before because the windows time service kept syncing it with some other time source?

It's possible I might have stopped that windows time service myself when I first encountered the wonky time issue and was floundering around trying to find a fix. But the time first went awry _before_ I did that. The wonky time problem started after a drive failure and a bit of disk corruption on the other drives.

Nothing about this makes any sense, neither the original problem or it eventually going away.

 

[mikeymikec]

Hmm, seem to have fixed it, finally. But can't really understand why or how. What seemed to fix it was going to "services", noticing that "windows time service" was "stopped" and set to "manual trigger" and telling it to start and changing it to "automatic trigger". Then the windows clock seemed to be fixed for a while... till suddenly the time jumped back again (and the event log, as always, demonstrated the causal event was windows deciding to 'synchronise system time with hardware clock').

But then after a windows shutdown and restart, it seemed to maintain the correct time indefinitely without that happening again.
Unless the problem comes back again the next time I 'hibernate' it.

I don't understand this stuff - the "windows time service" says it's purpose is to "synchronise the time between all clients and servers on the network", but _what_ network? What servers? This is a stand alone PC with just a router and an internet connection, so what does that mean in that context? Does a windows PC count as its own client and server?

Is it possible the hardware clock on the motherboard has always been broken (i.e. a motherboard fault - I see some claims on-line that this ASUS motherboard can have a faulty hardware clock problem from new) and I've never noticed before because the windows time service kept syncing it with some other time source?

It's possible I might have stopped that windows time service myself when I first encountered the wonky time issue and was floundering around trying to find a fix. But the time first went awry _before_ I did that. The wonky time problem started after a drive failure and a bit of disk corruption on the other drives.

Nothing about this makes any sense, neither the original problem or it eventually going away.

NTP servers. In Windows's case by default I think it's time.windows.com. You can set it to whatever server you like via the registry.

A Windows PC is a client; to my knowledge Windows workstations do not try to share times with each other. A Windows domain controller usually shares time with its clients because otherwise a number of domainy protocols get upset without the correct time.

 

[pmv]

NTP servers. In Windows's case by default I think it's time.windows.com. You can set it to whatever server you like via the registry.

A Windows PC is a client; to my knowledge Windows workstations do not try to share times with each other. A Windows domain controller usually shares time with its clients because otherwise a number of domainy protocols get upset without the correct time.

So does that mean the "hardware clock" on the motherboard is indeed non-functional? And the thing has only ever kept time correctly due to resynching with time.windows.com?

 

[mikeymikec]

So does that mean the "hardware clock" on the motherboard is indeed non-functional? And the thing has only ever kept time correctly due to resynching with time.windows.com?

I wouldn't like to say, but it's a reasonable hypothesis that I can't poke any evidence-based holes in. You've read my screwy experience with the clock

If you're intending to continue running this hardware for the foreseeable future, it might be worth your while to read up about adjusting how often the time service does a check and tweak it to be far more aggressive. Alternatively, just schedule a command to happen during login that immediately syncs the clock?

I realise you've changed the battery a number of times, but have you tried using the CMOS reset jumper? It seems like clutching at straws to me though.

In the BIOS, does the click actually tick forward in seconds?

 

[pmv]

I wouldn't like to say, but it's a reasonable hypothesis that I can't poke any evidence-based holes in. You've read my screwy experience with the clock

If you're intending to continue running this hardware for the foreseeable future, it might be worth your while to read up about adjusting how often the time service does a check and tweak it to be far more aggressive.

I realise you've changed the battery a number of times, but have you tried using the CMOS reset jumper? It seems like clutching at straws to me though.

Yeah, that's a good point about the CMOS reset - read that suggestion a few times but haven't been able to face trying it. For some reason the suggestion I kept seeing was to reflash the BIOS (Or is it more correct to say UEFI?) and then clear the CMOS, but couldn't do the reflash as it seems as well as ASUS's flashing utility it also needs some Intel utility installed (MEI - manager-something-or-other) and I couldn't find the version of that that supports this particular generation of CPU, and got nervous of ending up making things worse and ending up bricking the whole thing.

Does seem, as far as I can make out from various threads, that it's possible there was a problem with this ASUS motherboard from the start, just I'd never noticed it before because windows was reynching itself with the internet and masking the problem.

Still not sure why the clock problem started in the first place, as I'm sure that "time service" only got turned off _after_ the problem began and I was floundering around trying solutions. It's all rather baffling. Currently it seems to be working normally again.

So far, this machine (10 years old) has suffered a dying video card (possibly in turn caused by a dying PSU - I replaced both) a failed hard drive (not the C drive) and now this clock problem. Can't afford to replace the whole thing though, especially as a new motherboard would mean rebuying Windows (hence the likelihood next time I'll go Linux).

 

[mikeymikec]

It might be worth having a read about how say Windows or Linux goes about time synchronization.

Here's what the output of the timedatectl command produces on Linux:

mike@mikepc:~$ timedatectl
               Local time: Tue 2024-02-06 14:05:48 GMT
           Universal time: Tue 2024-02-06 14:05:48 UTC
                 RTC time: Tue 2024-02-06 14:05:52
                Time zone: Europe/London (GMT, +0000)
System clock synchronized: yes
              NTP service: active
          RTC in local TZ: yes

RTC is real time clock so I'm pretty certain that's the time according to my computer's internal hardware clock, but the OS's local time and the computer clock are not being kept in sync. Reading up about it might clue you in on how the process really works and thereby shed more light on your situation.