Europe Loses Six Minutes Due to Sagging Frequency and International Politics

You might be reading this six minutes early. Assuming that the Hackaday editors have done their job, this article should have appeared in your feed right on the half-hour. We have a set schedule to keep you supplied with the tastiest of hardware hacks and news. For some of you though perhaps there has been a treat, you’ve seen it and all the other stories six minutes early.

Have you perfected time travel? Sadly not unless there’s something you’d like to send to our tips line last week, but the culprit is equally fascinating. A dispute between Serbia and Kosovo has caused the frequency of the interconnected continental European electricity grid to drift below its usual 50 Hz figure for a sustained period, and as a result all clocks that use the mains frequency as a time reference have been getting ever slower.

How Can a Continental Power Grid Dip?

Think for a minute of a modern car on a hot day. When you turn on the air conditioning you will hear a slight dip in the engine revs as it accommodates the extra load. So it is with an alternating current power grid; a simple example is a power station supplying a city. In periods such as cold nights when the demands of the city go up, the result would be that the power station needs to work harder to satisfy it, and until that happens there would be a slight dip in its line frequency. Power grids compensate for this by increasing and decreasing the available generating capacity in real time, maintaining a mean frequency such that the “grid time” of a clock controlled by it matches an atomic clock as closely as possible over time.

Screenshot of Swissgrid live frequency data

In the case of continental Europe, grids across multiple countries are connected (PDF), and the task of maintaining that mean frequency falls upon Swissgrid which currently shows a sagging 49.976 Hz frequency. The time deviation of -346 seconds puts this close to a six minute loss and clocks running on this frequency will make you late to your next appointment.

It is at this point we leave the realm of electrical engineering and enter that of international politics, normally something far removed from Hackaday’s remit. It is fair to say that the history between Serbia and Kosovo is extremely delicate, and to understand some of the context of this story you should read about the war at the end of the 1990s. After the conflict the Serbian-majority region of what is now Kosovo refused to pay the Kosovan utility for its electricity, eventually leading to the Kosovans refusing to pay for that region’s share of the power received by Kosovo from Serbia. The resulting imbalance between demand and supply was enough to drag the supply frequency down across the whole continent, and though a short-term agreement has been reached the problem still remains on the grid.

Clocks and Mains Frequency

So if you are a continental European and you find yourself six minutes behind your British or American friends, don’t worry. We know that among our readers are people with significant experience in the power generation world, perhaps some of you would like to use your six minutes to give us a bit of insight in the comments. Meanwhile here at Hackaday we maintain an interest in the mechanics of power distribution even if some might say that it is Not A Hack. We’ve taken a look at utility poles, and examined how power grids are synchronised.

As for those slow clocks, the use of mains frequency to keep accurate time is quite brilliant and has been used reliably for decades. Tightly regulating grid frequency means that any clock plugged into an outlet can have the same dead-on accuracy for the cost of a few diodes. These clocks count the zero crossing of the alternating current. There may be moment to moment drifts but the power utility injects or removes cycles over the long term so the sum of crossings is dead on over the course of the day. It’s an interesting phenomenon to experiment with and that’s why we see it in microcontroller projects from time to time.

69 thoughts on “Europe Loses Six Minutes Due to Sagging Frequency and International Politics

  1. We’ve discussed this yesterday in the office and concluded that:

    1. Grid was a good for keeping clocks in sync 50 years ago, when clocks were mechanical and driven by synchronous motors.

    2. It was good for electronic clocks too, until 32768 Hz resonators became dirt cheap (~40 years ago?).

    3. Today syncing to the grid requires additional circuitry because transformers, which enabled access to 50 Hz at low voltage, were replaced by SMPSs in most electronic appliances.

    4. On a normal grid (EN 50160) a clock can be off 5 minutes a day.

    The mains frequency may be a bit off in Europe, but 0.024 Hz seems well within limits. Crystal oscillators provide better accuracy than mains so the news that many clocks are 6 minutes behind is just a clickbait.

    1. You make good points but I disagree with your dismissal that this is just clickbait. I have at least 5 clocks in my house which are tied to mains frequency. If I had to keep resetting them over the course of two months I would want to know why. To me the most obvious problem would be the clocks themselves and not the grid frequency. This is news.

      1. I am sorry, I wasn’t precise enough. This article is not a clickbait because it explains a phenomenon which occurs from time to time. I referred to the original news, that has circulated for the last couple of days in more regular news outlets, that emphasized the fact that clocks are off. My point is that 0.024 Hz isn’t anything special. 0 Hz off is much more special and people shouldn’t get used to it.

        Thank you Mike and all the Hackaday crew for publshing a lot of (at least) interesting information.

      2. Even my Panasonic smps uWave oven counts line cycles. In my off-grid world, that’s usually a crystal controlled inverter supplying the AC, and while not as good for clocks as the grid was, pretty good. Now, when I run a backup generator, it can be 5hz off either way – pretty nasty – and usually slow, as I don’t run it unless there’s a serious load. And…well, the bread oven needed its own built in filter added to the PIC pin or it saw transients as real zero crossings and had odd results…
        All these little things you learn when it’s all on you to get it right…

    2. NERC seems to agree with your points as they’ve been floating the idea of relaxing the mains time error correction requirements in the US for a few years. Back in 2011, they’d planned a year-long study to assess the impact but never followed through with it. See:

      It seems like there’s a more recent (2016) proposal for the same thing detailed here:

    3. Where I live, our mains clocks are a couple orders of magnitude better than our crystal clocks (including brand name alarm clocks, and vehicle clocks). Of course it doesn’t have to be that way, but that’s what on the shelves.

    4. If I can give a European (UK) perspective, mains synchronised clocks are still common here.

      For me though the clock angle on this story is incidental. I learned stuff I didn’t know about power grids and how they are run.

  2. Most clocks which are running behind are clocks from equipment connected to mains like alarm-clocks and magnetron etc.

    I am in the utilty-industry. Before this news went public, several customers contacted us that their clocks where not accurate.

    1. As long as the power grid is properly stabilized, they’re the way to go because of zero long term drift. Traditionally, in the US, the short term error is a few seconds, at most, and is corrected every day.
      Those days are coming to an end, as there’s talk of ending the tight frequency control in the US.

      1. DCF77/WWV/MSF/RWM etc. have been the “way to got” for decades. Not just because their accuracy is on a completely different level, but also because the device is able to auto-synchronize the time (and date) after a power failure.

    2. There’s one big advantage of mains-referenced vs quartz: All your mains-referenced clocks will have exactly the same time, even if it’s slightly off.

      Imagine clocks in a school — you don’t care if they are off by 1 minute, but you definitely don’t want them to ring at anything but exactly the same time…

    1. So they say… but I’m not buying it. Somebody just forgot about it or else the alarms would have gone off at noon plus six minutes. But they did not go off at all and that doesn’t compute.

      1. If there’s one thing I’ve learnt over the years it’s that subtle changes in time expose a whole world of bugs you never thought could exist. It’s amazing that the world didn’t implode due to Y2K. There’s too many people who think time is something absolute that you can use to make a decision.

  3. Just as many grocery items, e.g. candy bars, canned good, hide price increases by slowly shrinking the product over time.
    It is probably just the power companies in Europe doing the same thing.

    1. Many people care, including yourself, even if they aren’t aware of it.

      Highly precise time counting signals keep the telecommunications grid working, very likely including your Internet connection. Without that your computer wouldn’t be able to read or post comments to hackaday for example (or anywhere else)
      Slightly less precise time counting is critical to some forms of encryption as well, including Windows domain networks and unix LDAP authentication, not to mention the more obvious HOTP devices for two-factor authentication. Without that you likely wouldn’t be able to login to your computer at work, and perhaps be unable to login to your online banking, and quite possibly not be able to sign in to play WoW.
      Cellular networks would not be able to function, GPS would fail spectacularly, and depending where you live and how far behind your banks are both credit card and debit transactions would fail to authenticate.

      It’s a pretty big deal. But thankfully mains AC cycles are not the only option, and those of us keeping the world around you running have already long ago cared, worried, planned ahead, and solved this problem so you can spend your time worrying and caring about other things instead.

      1. I think you’ve missed the GP’s point. What time it is does not matter. Not to the internet, not to GPS. I mean if I look at my GPS receiver the time it gets is off by 18 seconds, and that changes over time too.

        No one cares what time it is. No device cares what time it is. The only thing people or systems care about is what another system or person thinks the time is in relation to your own reference.

        That isn’t quite true though. There are devices that do care what time it is. They are the kind of devices that fail spectacularly when something happens to the clock (leap year, leap second, y2k, unix time rollover) because some developer was silly enough to consider time as something absolute.

        1. I guess we both missed the point, as that phrase was a song. Or at least I was certainly got.

          About your GPS receiver however, are you sure that time skew is all the time? As in not just when powered on before it gets a satellite lock and sync?

          GPS kinda needs sub-second accuracy to math out where you are, that’s sorta the whole idea in measuring distance.
          18 seconds is insane! Even 1 second should report your position completely incorrectly, as one light second off is a 186000 mile difference. Even if the display looked correct down to the second, a tenth of a second difference could put the position hundreds of miles away.

          That just seems like the entire receiver is so broken as to not function at all, and I’m curious to what’s actually going on.

        2. The 18 seconds are leap seconds.
          GPS time does not account for leap seconds due to the potential of timing issues caused by the jump in time.
          Some receivers include additional information to adjust the time, and some may just output with the adjusted time.
          The leap second adjustment is only send out periodically, so it may take up to 15 minutes to get the adjusted time.

        3. A good reading:
          Splitting The Second: The Story of Atomic Time
          For a lot of devices they don’t need to have an ‘absolute’ time (which has no meaning) all they need is to have the same time which is not the same for example if ALL the GPS satellites are 10s more or 5s less this is not a problem. I am quite sure that they don’t re-synchronise when for example we add a second to the civil time at the end of a year.
          There is so many ‘times’ Atomic, Civil, Astronomical,….

  4. A motor generator set capable of delivering a few hundred amps. Maybe even add a large flywheel to smooth over mini power cuts or spikes. It may be more cost effective than a huge switchmode supply of similar capacity.
    Smooths out the line frequency. Well balanced and dampened it could me made whisper quite. Peheaps even delivering large currents for projects that would usually trip the mains beaker. Granted it would add a little onto the electric bill. Perhaps shutting it off and bypassing it when you’re at work.
    Maybe a little one in the back of a clock, just for exercises and amusements sake.
    [ plus in thoery should stop the government listening to and controlling your electronics over the line frequency of course that doesn’t stop them listening in with TEMPEST style apparatus or people at the other end of the phone line but it’s a start. ]

    1. Don’t use a motor generator, the efficiency would be terrible.

      Use a motor with a flywheel, and let the flywheel backdrive the motor. It won’t help at all with frequency issues, but can help with transients.

      1. There’s one huge transient with the motor switching to a generator, because the torque is determined by the slip between the rotor and field frequencies.

        In other words, such a flywheel would exhibit hysteresis. It needs to spin a couple % faster than the grid to push energy out, so it would still allow the grid frequency to dip significantly before the flywheel kicks in.

  5. Why does the article leaves off the interesting part? The one where we could actually learn something specific and technical? How are the regional or national grids interconnected and synchronised such that a drift in one network can cause other networks, supposedly managed by a central entity in Switzerland, to drift too? What if another region had their mains frequency too high at the same time, would it compensate for the loss in Kosovo?

    1. Thank you.

      There’s nearly 50 comments here until somebody finally points out what a lazy article this is. Claiming that Serbia and Kosovo stiffing each other on their electric bills creates and inbalance in supply and demand which results in a drop in the power frequency is a bunch of gobble good and at best leaves out many noon obvious steps.

    2. It’s a fair point. Hackaday publishes two types of story, the short form news ones like this one, and long form in-depth articles like Brian’s recent BadgeLife piece. The former are reportage, the latter, in-depth research of the type you mention. This is a news story, therefore it came out as a short form.

      As to your question about frequency, simply increasing the frequency would not work, by definition a mains grid has to have a common frequency. Increasing the power produced would fix it, in effect supplying shortfall created by the non-paying region.

      1. Am I the only one that is thinking of simply shutting off the non-paying customers as a potential solution? Or is that not an available option in those areas?
        Here in the US if you don’t pay your bill the power company shows up at your door a couple of months later to remove your power meter, cutting your home off from the grid and preventing you from stealing any more of their electricity (though a clever and somewhat reckless person can work around this).

        1. This. The mains frequency regulation was discussed at length in past articles and above; what I’d like to know (as someone not familiar with the regional politics) is why supplying free power for going on 2 decades at the cost of grid regulation was seen as a better option than throwing the big off switch for nonpayment, and why the rest of the customers paid an explicit surcharge to cover them without revolting.

          1. Removing that amount of load isn’t as simple as “throwing a big switch”. Just cutting it off is going to put massive imbalance on the remaining part of the grid and would wreak havoc on the control systems. To shut of a load like that it needs to be done gradually and basically in cooperation with whomever is getting cut off. And something makes me think they might not have been amenable to being cut off.

  6. I moved to the Philippines 10 years ago. The mains here are 220VAC / approximately 60Hz. I say ‘approximately’ because all of my mains-synced clocks run fast here, either syncro-motor driven or electronics-synced. I measured the mains when I first arrived and found it was not exactly 60Hz, but closer to 61Hz – and it would fluctuate up and down quite a bit.

    Just one more thing that makes living in the Philippines FUN. *yikes*

  7. A few years ago I got fed up of seeing different times on my clocks, and having to adjust them twice a year for that godawful DST.
    So I turned off the clocks on my microwave and oven (check your manuals), hung a radio controlled station clock in the kitchen (which amazingly runs for 2.5 years on a single alkaline AAA), and replaced the cheap (free actually) projection clock I had in the bedroom with a good radio controlled Oregon Scientific.

    Now all clocks in the house, except for the built-in one in my wife’s car, are either radio controlled (be it RDS or atomic clock), or ntp controlled.

    Getting rid of small frustrations like these is half the reason I’m a “hacker”…

  8. I wonder why appliance designers used mains as a clock source. I searched web little bit and discovered 32768Hz XTALs are dirt cheap. Digi-key sells them for 17 cents, some Chinese vendors sell them for just 4 cents.

    IMO, they’re negligible cost compared to the whole price of appliances BoM. Why they used mains as clock source? Any ideas?

    1. I think this is not about cost, but about convenience. This grid-synchronous clocks usually (except for the current situation) run very stable and don’t need to be reset except after power-loss.
      Cheap crystall-based clocks can easily accumulate an error of a few minutes every year when not synchronised to some external clock source (like radio signals or ntp servers).

    2. Xtals are accurate enough over a day, but what about a week? A month? Years? The amount of drift on a xtal is going to cause all devices to slowly shift time, all at different rates, making it a frustrating experience to try to keep all clocks pointing at the same time. If all clocks are driven from the SAME oscillator source (the grid) then they all gain/lose time at the same rate and always show the same value.

  9. The grid frequency in all electric grids around the world is constantly changing so the cited frequency is nothing more than a snapshot. If the author waited for 20s the graph on the swissgrid website would have shown a different frequency.
    Secondly, swissgrid is not responsible for maintaining the frequency, all network operators are in an combined efford. They just provide the nicest graph of the grid frequency.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s