Slow clocks in mainland Europe - how much of a problem is it really?

[NY]

makes it sound as if it's a big problem that affects significant numbers of alarm clocks, central heating timers and oven clocks.

Is that really that case? I would have thought it would only affect mechanical clocks with synchronous motors, and that all electronic clocks would use a built-in quartz crystal as their time reference. Are there many mechanical clocks and central heating timers still in use? Or are there any electronic clocks/timers that sync to the mains rather than their own quartz crystal? Is it as important as it used to be for the mains frequency to be kept within such very tight tolerances and to average out at 50 Hz over any

24-hour period so there is no nett loss or gain?

My parents have one mains clock with a synchronous motor, but all their other clocks, and all of our clocks, are mains or battery-powered electronic (ie digital or else analogue-by-stepper-motor).

[Max Demian]

"So yes, you can manually adjust your clocks - but they will automatically stop losing time once the grid stabilises."

Or will they attempt to correct them by increasing the frequency? In which case resetting them will result in them being fast.

[Harry Bloomfield]

NY formulated the question :

I have one mains synchronous clock, which is a solar clock turning the drive lights on and off. I keep intending to replace it, but it works fine and does need to be that precise. The rest are radio controlled clocks, using Anthorn.

Crystal controlled clocks, like my laptops and desktops are much more variable, but they sync to internet time once per day. Sat receiver has a crystal controlled clock, but syncs to the satellite every time it is turned on for a couple of minutes. Likewise landline phones and mobile phones.

[NY]

So true mains-sync clocks (as opposed to free-running or else periodically syncing to a master clock) sound to be fairly rare, as mechanical ones with a synchronous motor are gradually replaced with crystal-controlled ones. I wonder how many people have actually been affected by it, if most clocks are free running (but accurate) or periodically synced rather than mains-synced.

Oh, and I think Windows PCs sync once a week rather than once a day. I'm not sure how often Mac and Linux sync their clocks to a time source. The maximum that my PC seems to get adrift is about two minutes per week, and I'm surprised it's even as much as that - evidently quality control on crystals isn't as good as it should be.

The only clock which keeps really poor time (consistently loses time) is the digital one on the cooker. If it was mains-synced, I'd expect that it would be correct averaged over a long period of time, even if it lost a bit of time during heavy usage periods during the day and gained a corresponding amount overnight. So it may have a very out-of-spec crystal...

[Harry Bloomfield]

It happens that NY formulated :

No - mine are configured to sync once per day, via a reg hack.

[NY]

Ah, OK. I'll have to look up that hack to reduce the chance of clock skew between different PCs on the network.

seems to describe what to do with SpecialPollInterval. I've always been surprised that the ability to alter that interval isn't provided on the Internet Time tab of the clock.

[Dave Plowman (News)]

Not so. I've got two large LED digital clocks here which are mains locked. Since they have to run off mains due to the power consumption of the LEDs.

Generally, they're closer to the real time when it comes to have to alter them when the hour changes than the average cheap quartz one.

[The Natural Philosopher]

Linux and probably Mac synchronise very very frequently

"These options specify the minimum and maximum poll intervals for NTP messages, in seconds as a power of two. The maximum poll interval defaults to 10 (1,024 s), but can be increased by the maxpoll option to an upper limit of 17 (36.4 h). The minimum poll interval defaults to 6 (64 s), but can be decreased by the minpoll option to a lower limit of 4 (16 s). These option are valid only with the server and peer commands."

[Brian Gaff]

I do recall the 3 day week when the minors strike was on and then the cps were often all over the place and in some places where supplies had to be maintained, generators were in use and these were far from stable, yes clocks were inaccurate, and old fashioned turntables and tape recorders were all over the place, but generally no real harm occurred. Brian

[NY]

Ah, I was mistaken in that case. I naively assumed that modern mains powered clocks and CH/cooker timers etc were simply a battery-driven quartz clock mechanism with a PSU in place of the battery. I didn't realise that modern clocks (as opposed to old ones with synchronous motors) still synced to the mains.

In that case, it *is* a problem.

[NY]

Yes that works perfectly:

Change

HKEY_LOCAL_MACHINE\SYSTEM\ControlSet\Services \W32Time\TimeProviders\NtpClient\SpecialPollInterval

from its default value of 604800 (0x093A80) [1 week]

to 86400 (0x015180) [1 day]

I'm not sure whether you need to restart the "Windows Time" service for it to take effect, but I did to be on the safe side.

I proved that it worked by temporarily setting the value to 60, and then deliberately setting the time manually to a few minutes fast or slow. Within a minute the clock changed automatically as it reesynced. Having set it back to 1 day, I can see in the clock's Internet Time tab it says "Next sync

10/03/2018 at 21:21. Last successfully synced on 09/03/2018 at 21:23" - in other words it wants to sync 1 day later than it last did.

[Dave Plowman (News)]

It's not been up until the last few weeks.

One of the large digit LED clocks I've got is my alarm clock. Being rather long sighted means I can read it in bed in the dark without needing specs, etc. I usually listen to the radio while dropping off to sleep and can check it against the pips. Generally, it was within a few seconds (up to say 30) of being correct, and sometimes fast, sometimes slow. But over the last couple of months have had to re-set it three times. And since I have another mains locked clock which is the same, not a fault on it.

[Martin Brown]

There is a fair proportion of clocks in traffic lights and domestic settings that rely on mains frequency being on a daily average right. I can't see that 5-10 minutes delay will cause all that much grief but it might be a nuisance where rush hour altered flow timings are affected.

There is a lot of legacy kit out there that uses mains frequency.

These days 32kHz clock crystal mechanisms and RTCs are dirt cheap but it was not always the case and there is a lot of legacy kit out there.

[Roger Hayter]

I imagine Microsoft felt a certain obligation to protect public NTP servers from overload. The proper way to do it if you have more than one or two computers is to have a local NTP server which only contacts the higher tier servers when it needs to (algorithmically) and sync all the local computers as often as you desire to the local server.

Linux is still a sufficiently minor part of the installed base of non-networked computers to be able to afford to be less public spirited, but people should still be encouraged not to generate totally unnecessary traffic.

[Johnny B Good]

Before I switched to using Linux Mint 17.1 KDE almost 3 years ago, I had win2k configured to re-sync every 2 hours and 59(or 55) minutes to hold the PC's clock to within a few seconds of UTC yet avoid on the hour traffic peaks. The clock settings were readily accessible from the system tray icon, unlike Linux Mint KDE which is devoid of any obvious means to set or even check the frequency of NTP updates.

I'm pretty certain that Linux Mint calibrates and applies a correction factor to the system time of day clock to cancel out errors due to the on- board RTC and the MoBo's System Bus Clock since it's always within a second of my Casio DB360 which by pure chance has been keeping to within a second of UTC for several months at a time over the last 5 or 6 years.

Anyhow, this reduces the frequency of resyncs with the NTP service. I've no idea how frequently my desktop machine syncs up to the NTP server but I should imagine a lot less frequently than the 3 hours I was using with win2k.

My IT Works weather station's radio controlled clock drifts several seconds a day between sync up events (4 seconds slow right now). It doesn't always pick up the radio signal (2 am seems to be its favoured 'check in' time to resync) due to RFI from my Asus 23 inch Full HD monitor - I initially suspected RFI from my almost always on desktop but it now seems the monitor is the main culprit in this case. As a result, my radio controlled clock can be nearly half a minute out at times, making the Casio watch my best option to provide a 'sanity check' on the desktop's time keeping.

[Michael Chare]

I have a number of mains powered clocks with LED and LCD displays. I would expect them to rely on mains frequency. My Casio watch uses a crystal but does gain about a second a day.

[NY]

Interesting. Until this problem in Europe it had never crossed my mind that any electronic clock (ie with a digital display or else analogue driven by stepper motor) would still be synchronised with the mains. I thought that mains sync technology went out in the 1970s with the gradual demise of "electric clocks" (ie those with synchronous motors) and that "electronic" more or less implied a quartz crystal. I confess to being utterly gobsmacked that the technology still lives on.

That's why I thought that the story about Europe was a great big fuss about nothing which would only affect a minority of customers.

How wrong I was.

[Dave Plowman (News)]

Think you'll find things like cookers and microwaves with digital clocks are still mains locked. My Neff ones - not *that* old - are.

A really accurate quartz clock isn't cheap to make. And when mains does stick to averaging out 50 Hz over 24 hours, ideal for something that has mains all the time. As it won't need adjusting until the hour changes.

[newshound]

[Martin Brown]

It was common until fairly recently. Plenty of older industrial control systems for heating are synchronous motor based - they often don't get replaced until they fail. Traffic lights are the other big one.

The mains was treated as a long term reliable 50Hz reference source.

OTOH there are not that many situations where being 5-10 minutes adrift from actual UTC can make a significant difference. It isn't that long ago that cheap mechanical watched used to drift significantly in the days before quartz crystal locked oscillators.