Ask Hackaday: Is Your Clock Tied To Mains Frequency?

Earlier in March we heard about a quirk of the interconnected continental European electricity grid which caused clocks to lose about six minutes so far this year. This was due to a slight dip in the mains frequency. That dip didn’t put anything out of commission, but clocks that are designed to accumulate the total zero-crossings of the power grid frequency of 50 Hz don’t keep accurate time when that frequency is, say 49.985 Hz for an extended period of time.

An interesting set of conversations popped up from that topic. There were several claims that modern alarm clocks, and most devices connected to mains, no longer get their clock timing from mains frequency. I’ve looked into this a bit which I’ll go into below. But what we really want to know is: are your alarm clocks and other devices keeping time with the grid or with something else?

“The” Alarm Clock Chip

I’m always interested in learning a lot about something I have gotten wrong in my head. In this particular example, I put a picture of this ~30 year old alarm clock I use everyday on Twitter as an example of the mains-frequency clocked device (one of at least five I have in my house). Someone replied that it’s a quartz clock because it’s digital and therefore not getting its timing from mains.

I certainly could have been wrong. I only assumed it was tied to mains frequency but I couldn’t say why I thought that other than having heard that most wall-powered clocks are. In thinking about this I wondered how the clock could have a battery-backup if it is tied to mains? That’s actually a pretty interesting question. Turns out I was not wrong about how this clock works, but I only know that now because I took the thing apart to investigate.

Looking inside the clock there are two distinct parts: the clock module and the radio module. I ignored the radio and looked closely at the clock module for a 32.768 kHz clock crystal which usually show up in the “tuning fork” form factor. I couldn’t find any.

The next thing I looked at is what chip is driving the timekeeping board. It’s marked LM8562 and a quick datasheet search results in a “Bingo”! This part is listed as a “Digital Alarm Clock”. With a bit more internet searching, I think this is the chip that is in the majority of alarm clocks. If you set time and alarms by holding one button and using the fast-up or slow-down buttons to change it I’d wager it’s based around an LM8562.

Ew. 30 years of yuck.

The chip drives the display and has the user input buttons I just mentioned. The key feature I was looking for is listed as follows: “12/24-hour mode, 50/60 Hz selectable (provided that it is impossible to select the combination of 24-hour mode and 60 Hz)”. This chip is indeed tied to mains and the next feature touted in the datasheet answers my question about battery backup: “On-chip CR oscillator for backup use at the time of power failure”. So there’s an RC oscillator built in to run off battery when the power is out. I should do some testing to see how accurate this is. It’s quite likely your time goes off very quickly running from an RC clock source but as long as it doesn’t completely reset the clock it’ll still wake you up in the morning if the power comes back on.

Clocks That Aren’t Tied to Mains

The cheapest built-in microwave I could buy certainly doesn’t have special time keeping.

I’m now operating under a rule of thumb for identifying mains-tied clocks. If it plugs into the wall and doesn’t set itself, it’s likely clocked from grid frequency. Think especially about appliances like microwaves, VCRs, traditional ovens, and non-internet connect timers that have a clock feature like programmable thermostats and light switches. But of course there are exceptions to every rule. If the clock is using a wall wart that converts AC to DC then it is not tied to mains since the DC doesn’t have a zero crossing to count. It’s possible you have a wall-wart that converts AC-AC but unlikely.

You’ll also note that I mentioned the clock shouldn’t set itself. There are some clocks that when powered on (from a plug or batteries) will set their own time and date. In pitching this article to Elliot Williams he mentioned that DCF77 modules are like 1€ apiece and likely clocking most things in Germany. These work by listening for time data from a radio station. There’s a similar standard in the USA called WWVB. My assertion is that if a clock has one of these modules they will set themselves which is an easy way to tell the device is not using mains frequency. But these devices are probably limited to the slightly higher-end: one Euro added to the cost of a competitive consumer good might not be worth it when the same stable timekeeping job is available without it.

Finally, it’s possible you have a clock that uses a very accurate time standard. My absolute favorite example is the DS3232 (you may know it as the Chronodot), which is a temperature controlled oscillator (TCXO) built right into the chip. These things have incredible precision for a single-chip that can run on a coin cell for years. But they’re kind of pricey for consumer goods so I believe you rarely see devices that use them. The Bulbdial clock I have in my living room has one — it’s an add on, but as I said this is my fav time chip.

Tell Us About Your Clocks

So we ask you: how do the clocks around you get their time? We’d love to hear if your clocks are using the LM8562, pulling from a more advanced source like radio signal, or something not covered here at all. Where do your appliances get their time standard?

162 thoughts on “Ask Hackaday: Is Your Clock Tied To Mains Frequency?

    1. The whole incident about clocks drifting is more about ovens and other time keeping appliance, which are obviously running on mains. This post is about a 30y old US clock, so its not really representative of current EU alarm clocks. Actually most alarm clocks that I see in the EU are both battery and mains powered, whichever you want, or battery backed is you use both. In my opinion a mains-only clock is a cheap one. But really, it’s for old people, everyone use their smartphone nowadays.

      1. Sometimes using your phone for unnecessary things like alarms just wears on the phone and feels like putting all your eggs in one basket. I mean really if you lost YOUR phone you’d be out an alarm clock, a wallet, your only computer (lol pathetic) and phone. If I lost my phone, no big deal. And I can do other stuff like navigate with paper maps, use phone books, read real news papers, run 10 miles and fist fight more than any 5 snowflakes.

    2. You say that as if it’s some outdated concept. However, it basically comes for free with the power grid. By virtue of that being carefully regulated, you get incredibly accurate yet cheap timekeeping and clocks. No extra infrastructure or accurate components needed.

      Of course, it does presume a reasonably well built and maintained power grid.

      1. It’s *kind of* broken in that it’s using three freakin’ watts just to show the time and beep. That’s half the power of my Celeron N3150 passively-cooled industrial PC with 32 GB SSD, dual gigabit NICs, 4 GB of DDR4 RAM, 802.11ac WiFi, Bluetooth, and… it can get the time from NTP servers.

        I guess it doesn’t have a 7-segment display though.

        1. 3 Watts over 10 years costs you about $40

          If you buy a new clock radio that costs $40 it would have to use almost zero power and still it would take over 10 years for the investment to pay back.

          And I was talking about things like classroom clocks and other public timepieces that have a small AC synchronous motor to turn the hands.

          1. 3W over 10 years works out to 192kg of CO2, or 24million tonnes if every house in America had one over 10 years.

            No one cares about your cheap electricity cost when talking about efficiency.

          2. @garbz: Um… efficiency is (by the most common definition) about financial effort. Effectiveness is what you appear to be worried about. Maybe research the issue a bit better (lookup research on “CO2 sensitivity”) and you might feel a little less worried…

          3. Simon, as you were too lazy to give us a link I’ve found one for others who might be interested in this CO2 sensitivity (it’s not like an allergy at all):

            “Some global warming ‘skeptics’ argue that the Earth’s climate sensitivity is so low that a doubling of atmospheric CO2 will result in a surface temperature change on the order of 1°C or less, and that therefore global warming is nothing to worry about. However, values this low are inconsistent with numerous studies using a wide variety of methods, including (i) paleoclimate data, (ii) recent empirical data, and (iii) generally accepted climate models.”


        2. Get time from NTP servers… what, even without an Internet connection?

          That 3W clock gets told by the user what the time is, and then does its own time keeping without an Internet connection.

          I’ll bet your industrial PC will use a lot more than 3W if you hook a display up to actually see what the time is, unless you use the PC speaker to beep out the time in Morse. (Or use the PWM channel for crude speech.)

    3. In (western) Europe the mains frequency was held very stable for the last decades and power outages are rare. That went that far, that there was a DC link station for energy transfer in times of the soviet/eastern-europe block, where the mains were not that reliable and frequency stable. But after the fall of the iron curtain this was not be deemed necessary any more and the grids were tied together.
      I am not sure about the micrawave oven, but my other clocks are quartz or DCF77 controlled.

  1. I have a battery backed up mains clock and it is pretty accurate if the power is power is out a short time, but if its out for 3 or 4 hours it usually losing about 15 minutes.

          1. >That sounds about right.

            Due to tolerances in components (+-5%) you can’t design the oscillator to run at exactly the right speed unless you also design compensating circuits, which costs money. If you specify the oscillator frequency exactly right, half your devices will run slow and half will run fast, so to make things consistent you have to specify the oscillator to run fast by the amount of your component tolerance to make sure it never runs slow.

            For example, if you specify precision components (1%) and you have an RC circuit, the worst case error you have is +-2% while the most likely error is +-1.4% (square root of the maximum error). Hence, you must make the oscillator run 1.4% fast (20 minutes a day) to make sure most of your customers won’t miss their alarms.

          2. It could be possible that the tolerance spec for the backup oscillator is something like -0% and +3%. Similar to car speedometers which all show a higher than actual speed, so nobody has the excuse for speeding “my speedometer showed less”.

    1. I’ve noticed this too. I have a clock almost identical to the one in the article, and if the power’s been off for more than a few minutes, I can tell. The clock loses multiple minutes per hour, so a couple of hours power outage during the night means waking up ten minutes or more before I expect to. As someone who’s spent a lifetime waking to alarm clocks, I’m almost always waking up on my own at the exact time the alarm goes off, so ten minutes early is really jarring and obvious. I’ve wondered if this was a design feature, but I try not to attribute genius to something that would also be the result of cheapness or incompetence, which are so much more common.

  2. My old-tech nixie clock gets its time from a GPS module I bought off amazon intended I think to plug into a car dash cam. It outputs the time in ASCII using serial but doesn’t know about DST or time zones.

          1. Apparently some idiots still think that messing with the clocks makes the sun stay up longer. Maybe businesses could just have summer hours instead of screwing up everyone’s biological clock shifting the time forward and back.

          1. The southerners argue that we should have DST in Queensland because it would align us with the southern states for business reasons.

            My usual come-back is to point out that if the whole country adopted the UTC+8 timezone like Perth, we’d all be in sync with China, one of Australia’s biggest trading partners.

            Few seem to like that idea.

          2. Fine with many, but not mommies who send kiddies to schools in tbe dark only to have them get out at 12 or 3. We could work 7.5 hrs in winter and 8.5 in summer. Or at least break the hr down into 15 min changes ea month.. Airlines study this stuff and say it takres 3 WEEKS to get used to a 1 Hr Time Change. Maybe just vary the mains frequency. Our clocks will adjust. Like Digital TV… all the new-old clocks required will create a New Industry! I never did find a $40 convertor. I have no TV. Don’t miss it much.

          3. Local Standard Time: Nahh!!
            DST: Nahh!!
            Switichg Back and Fourth: **NAHH!!**

            UTC: Much Better!! and it eliminates transglobal confusion once everyone gets to ISO standard of YYYY/MM/DD HH:MM:SS

            Unix Time: Also usable for some. Its 1522506258 right now. :-)

      1. I spent a nontrivial amount of time attempting to live in a decimal time system derived from Julian day numbers (a count of days where 0.0 is noon UTC on 24 November, 4714 BC in the proleptic Gregorian calendar, to be excessively precise); the changeover from one day to the next being noon UTC kept most events here in US/Pacific time from beginning in one day and ending in the next.
        (One of the projects I’ve long meant to get to is putting together an analog 24-hour clock so the dial behind the ‘hours’ hand can be this .0 through .9 display.)

  3. It’s a bummer your clock’s LM8562 chip excludes 24 hour/60 Hz, that would be neat to be able to convert it to 24 hour.
    I’m going to check out some of the other chips mentioned here for building my hexadecimal LED clock.

    1. There was a chip LM…, perhaps LM538x (?) with 40 pins and non muxed display but it’s more than 30 years ago that I used it, where you had separate inputs for 50/60 and 12/24. I used crystal control with a 3276800MHz crystal, or was it 4915400MHZ ?

    2. 50Hz folks speak “24Hr” but 60Hz folks speak AM/PM… so 60Hz folks don’t “have to select” 12 or 24 Hr, and that is a “feature” if you are selling a chip to cheap clock makers. My Aerstar was funny like this. MPH stopped updating past 85 but I read metric also so swithing to kph my speed would again be indicated.

      1. That was because of a ridiculous law in the US for a few years that mandated speedometers could only go up to 85MPH. You had high end sports cars with 85MPH speedos, it was dumb. All because some twit of a politician had it in her head that this would save lives. As if anyone would just think their car couldn’t go any faster than the speedometer could read. My friend used to amuse himself by burying the needle and then letting off the gas and seeing how long it took to drop down off the peg.

  4. “That’s a bingo!”
    “You just say bingo.”

    Can’t tell if that was an intentional reference or not. Interesting article, I also assumed that most clocks weren’t mains-driven anymore. I like how the chronodot’s PCB is stylized to look like a watch face. Pretty handy little chip.

    Interestingly, two of my clocks are powered by a DC wall wart, yet they set their time the way you describe for the LM8562. They definitely don’t use mains frequency, so I don’t know if the method of setting the time is a reliable indicator.

    I’ve never seen a clock that automatically set its time upon powering up except those connected to computers or cell phones getting that info over the internet or cellular network. The old-fashioned radio method must not be in common use in my country.

    1. The AC adapter might put out only AC voltage, so inside the clock is a rectifier and filter capacitor. Thus the clock has access to AC. Or maybe it counts on a very rough rectified AC, so it has the needed AC for clocking the clock.

      The Commodore 64, as someone reminded us in a recent comment, needed some AC for clocking something, so you couldn’t use any old DC supply, and running it off a car battery meant extra circuitry. The power supply brought some AC out from the transformer.


    2. It isn’t. The chip I used 30yrs ago had the same time setting and even supported mains timing. But I decided to include a crystal oscillator. So the only reliable way is to look for a crystal.

  5. Mike,

    CR May stand for ceramic resonator- it’s a component that is more accurate than a r-c network, but not as accurate as a crystal resonator.

    I am a bit surprised that the EU has such lax control over the frequency of the power grid. In the U.S. there may be slight variations of the 60Hz frequency, but over the long term the frequency averages out to 60Hz, so the clocks tied to it have excellent long term accuracy.

    The U.S. has three time and frequency broadcast stations- WWV, WWVB, and WWVH. WWV and WWVB are collocates in Boulder, Colorado, while WWVH is in Honolulu, Hawaii. WWVB broadcasts timecode at 60KHz longwave frequency. The other two stations broadcast on numerous HF (High Frequency) bands. There are some clocks, mostly kits, that used WWV and WWVH. The only consumer clocks I have at home are in my stove and microwave, and those are power grid synchronized clocks, as you stated. I have two mechanical clocks in the house that are very accurately tuned to provide correct time to within a few seconds drift over multiple winding sessions. I am very pleased with their function. On the opposite side of this scale, I have a network time clock using a GPS disciplined clock that provides network time to milliseconds accuracy for my telephones and computer network.

    1. The EU threw a small cou try a bone, and it runs the frequency for Europe. But a yet smaller country which separated in a feud from its previous controlling mother country, uses power but refuses to pay. The load burden is costly and drags the freq down. Negociations succeeded, sort if, but the contract cannot be valudated due to minor detail of rrefusal of the smallest country to use certain ” required terms of legal language,” read, verbal subserviance.” Politics and ego has slowed, time, almost literally.

      1. Are you saying Monaco is stealing electricity? Luxembourg has their electric supply bypassed with an enormous metal coat hanger? Or what? For people who don’t already know what you’re talking about, surely the group who’d find the information most useful, your post is a mystery.

    2. The EU network is generally as accurate as the US one and as suitable for timekeeping. It’s supposed to even out in the same way. However, there’s some political shenanigans going on, which means not enough electricity is produced, causing the frequency slump.

    3. It’s a bit of a special situation, as the frequency drop was caused by two countries that recently went to war with each other and had to settle their differences in power supply, the effects of which then spread to the whole synchronized grid as one of the parties refused to replace the energy they had due.

      The lag or advance in grid frequency is effectively a measure of the power system’s state of balance. Someone has to push more energy into the grid to shift the frequency back up, and if nobody’s willing to pay for other countries’ power debts then the frequency anomaly remains.

  6. In Western Europe the mains infrastructure is very reliable, so running your alarmclock from mains power is preferred instead of having to replace batteries every 6-12 months.
    In Malaysia or even in the USA, the infrastructure is known have to outages during storms, cold or whatever. Running off of batteries will be more reliable.
    The ‘new’ alarmclocks are our smartphones – connected to fitbits or smartwatches, to buzz us awake in the morning.
    What I miss in a smartphone is the ability to tell me the time at night without having to reach over and press some button.

    1. I have a Google Home Mini next to my bed and I use it primarily as an alarm, turning my smartbulb on and off (not having to leave my bed to control it, or use my phone is beyond awesome!) and last but definitely not least, I can ask it to tell me the time without even having to open my eyes for those lazy mornings in bed :D

        1. Same here I really don’t trust these always listen connected devices and it’s not because I’m clueless about how they work but because I do fully under stand how they work.
          Most of the heavy lifting for voice recognition is done on remote servers which means it’s sending back actual audio recordings.

    2. It’s the same synchronized grid, so disturbances in the east or south, or north, will affect your grid frequency as well.

      Or have we forgotten the 2006 EU-wide blackout that started from wind farms in northern Germany and spread to France, Italy and Portugal?

    3. I noticed watching the Tour de France, that France didn’t have any power lines above ground. Most of US’s is still old fashioned. I know this has nothing todo about line frequency, but it sure has less problems with trees falling on overhead lines

  7. In parts of the world where the mains frequency is still maintained for zero long term drift, using it as the reference for a clock makes a lot of sense even though the cost of crystal oscillators is low. It’s nice having a clock that keeps perfect time.

  8. I have a self built Nixie Clock that gets its timekeeping from the mains. At the other end of the spectrum I have a clock and a watch that gets time from the UKs NPL beacon.
    Everything else is battery powered and quartz timed.
    The mains Nixie is more accurate than my automaitic watches and I only set it after power failures and summertime changes. I’m suprised mains powered alarm clocks are still popular – though I still have my eye open for a reasonably priced “groundhog day” clock.

  9. I live close to a cell tower and every time the wind damages the antennas I can tell that a couple of my clocks run fast while the cell signal is swamping my house.
    plays hell with bleed through on my landline phone and “old school” audio equipment too.

    Used to be able to find those generic little clocks in the shape of a small cigar box. roughly 2″ tall 5″ wide and bout 4″ deep.
    Red led display.
    piezo disc chirper for the alarm.
    9 volt backup battery.
    The good ones would still sound the alarm while running on the battery.
    Usually ran the battery flat within one cycle, but at least you weren’t late for work that day.
    Wish I could find a few more of them again.

    1. My RadioShack LED Clock/Timer uses a 9 volt backup battery. IIRC, the manual said the battery will die withing 9 minutes if the alarm is triggered during a power outage. (sigh!)

  10. I have a LED alarm clock that uses mains frequency, but I have to add that the thing has been in the family for more than 40 years.
    It’s as accurate as can be and way more accurate than most all other clocks, apart from the one I have that syncs on a GPS signal.
    Even computer clocks are awful in accuracy if you don’t let them access a NTP server every day.

        1. Fortunate in both the cord being very long as well as a window not far away. Some GPS (my other one) was even designed to sit outside. The first is both USB (long cord) as well as BT, other very long cord.

    1. There are two “clocks” in your computer. The hardware one (RTC) which holds the time even if it is powered down. The other one is kept by the operating system and is driven by interrupts. It is set up from the RTC upon boot and wake up. Software clock can slow down significantly if system loses interrupts, e.g. do to really heavy I/O.
      Depending on the system in charge there is various synchronization between RTC, system and NTP time.

    1. A digital clock is just lot of counters. They just count seconds, and the readout follows. They’d probably never have an oscillator at 1Hz, too big components and hard to adjust. So the oscillator runs at a higher frequency, and is divided down to 1Hz to be counted by the main counters.

      Using the AC line, they square it up and divide it down to 1Hz.

      When people made clocks with discrete logic ICs, the schematics were obvious. Endless articles in the hobby magazines. Then later, but in retrospect fairly soon, LSI came along, just about everything in one large IC. That caused endless kits being offered, with a transformer, the readouts, the clock IC and the few external components and a circuit board. It was a big wave. But all the details in the one IC. A few years later, there were clock modules for about the same price, just needing an external transformer, but lots of options via jumpers.

      Some people built a crystal oscillator and divider chain to get the 1Hz pulses, some putting a lot of effort into making it very stable and exactly on frequency. Ten National, I think, came out with an 8pin CMOS IC that took a color subcarrier crystal, cheap and available, and divided it down to 1Hz, with some error but “good enough” and that would work better than an RC oscillator when AC power was unavailable. But I’m not sure any clock radios were that fancy.


      1. Then there’s always the old synchronous motor, analog clock.
        I have one from early 1960’s, it was still keeping proper time when I took it down to paint the room about four years ago.
        I really should dig it out and rehang it.

        1. Sounds like a Telechron :) I have one from 1930, and with a bit of cleaning I was able to get it going again. As long as there are no power outages, it keeps time like an atomic clock.

          1. This strange construction not even uses the mains frequency as timing reference, but an ordinary 555. Although this part was called a precision timer, it is not precise enough for a useful clock.

        1. This guy did not build it, it just ripped out the guts of a commercial clockradio and put it into a new housing. It’s arguable, if he made it deliberately suspicious looking.

  11. I have a Sony clock radio that is plugged into the wall but it seems not to use the mains frequency, as it runs fast. No other clock in the house does that, but I really should find a way to check the frequency of my mains supply to be sure.

    1. I have a Sony Dream Machine clock radio that drove me nuts, was always 10 minutes fast. Discovered that i mistook the Auto Time Set feature to mean Atomic Time setting. They are set at the factory, so consumer doesn’t have to the first time. So it probably does use mains freq, except when powered by CR2023 backup.

      1. I found one of those in a pile of junk. I got excited at the “auto time” bit. But either looking at it or a web search revealed that it only meant that the clock was set at the factory, and kept on ticking. The clock was soon parts.


  12. Are alarm clocks still a thing? Y generation like myself have only heard of them down here in Brazil.. specially after circa 2007 with the widespread of cellphones. Now when it comes to DVD’s, Blueray and microwaves, I believe they all rely on the mains as well

      1. Thanks for mentioning: I can’t seem to figure out how to predict wether a set alarm or calendar event in my Android smartphone will effectively be acoustically hearable or not!
        It happens now and then that I find my smartphone displaying the alarm animation, but nothing is hearable at all nor vibrating even tho the settings are “right” for it.
        I already missed some happenings due to this unreliability of the droid!!!

    1. I still keep an alarm clock on my nightstand. It’s nice to just be able to glance over and see the time and it is a good back up for the phone if I happen to forget to plug it in, to charge or fall asleep reading something on the phone and it ends up jammed under pillows where I won’t hear it.

      I also tend to be a deep sleeper, often the phone alarms will not even wake me up, but the alarm on the night stand will wake me. It’s also easier to blindly smack the snooze button on the thing while you’re still all groggy and blury eyed than to snooze a phone alarm

  13. My 1930s telechron clocks keep perfect time using the US grid frequency. I have to replace the wires every 50 to 60 years and apply a lightweight oil every couple of decades. Totally not kidding.

    I have fans from that era, too, that still work just fine. The brushless AC synchronous motor is a very wonderful thing.

    In 1917 Henry Warren invented the Telechron clock, and accidentally invented the modern power grid as a side effect of wanting to keep good time.

    You see, if grid frequency is not precisely controlled, power is wasted where generating stations connect. A power grid that does not have reliable enough frequency control to run clocks is an wasteful grid, a poorly administrated grid – a broken grid, in my opinion.

    Unfortunately, after the US 2003 blackout, Congress instituted laws that placed large fines on power companies for voluntarily trying to make a clean grid and failing. So, power companies being what they are, they no longer wish to voluntarily provide this service, instead preferring to waste power making a b0rked product and passing the costs along to the consumer.

    More information here:

    1. > A power grid that does not have reliable enough frequency control to run clocks is an wasteful grid, a poorly administrated grid – a broken grid, in my opinion.

      Welcome to the renewable future, where randomly inserting power into the grid is dictated by law and every other operator has to just deal with it.

      1. Haha! Dream on.

        Here in the unrenewed present, this is a solved problem. You literally cannot legally purchase a grid-tie inverter that doesn’t sync to power company frequency. Even if you do guerilla solar, the hardware will enforce frequency and non-islanding regulations.

        Once again reality intrudes on political talking points…

        1. The problem isn’t if the inverter is grid tied or not. Infact it won’t matter as far as effect on the grid frequencies. If you put a non grid tied inverter on the mains it would likely blow the inverter transformer or mosfets as soon as you attached the mains since they would be out of phase.

          What the real problem with all the renewable energy sources is them removing load off the large generating plants. If the wind suddenly picks up, or the sun appears from behind clouds in your area, you could suddenly have megawatts of extra power being dumped into the local grid depending on how much renewable energy is deployed. When this happens load is taken off the large generating plants. When you have thousands of horse power going into a generator whose load suddenly is lowered, it is going to spin up faster affecting the mains frequency. This is no different to how an electric motor or gas engine will spin up faster once it’s load is removed

          1. Vanadium redox batteries could solve the grid balancing problem with alternative energy since they last much longer than lithium ion and don’t catch on fire as easily.
            Just have large sites store at least 24 hours of capacity.

          2. Grid supply and demand problems aren’t changed due to the sources of energy being input into the grid. A properly engineered grid deals with differences in supply and demand in real time – that’s literally what it’s for. Yet another solved problem.

  14. I got the exact same clock shown in the article(that probably also needs a cleaning just as bad,need to find the right bit that’s long enough to reach the screws.) I thought it wasn’t tied to the mains frequency as it has a battery backup. Go figure.

  15. I have two clocks in my home that run off mains. Both of them can be adjusted for summer-winter time without touching the minutes part. I made a mistake with one (I thought it hadn’t started running again after I’d adjusted the hours, but it was running just fine).

    Anyway… This article is a bit late. Now about 2 weeks ago, they flipped a switch and started to catch up quite quickly. The officially published target frequency is: 50.01 Hz when the accumulated lag is more than 20s. But they are catching up much quicker than that. I suspect they agreed on “target frequency is 50.025 until the lag is < 20s". The offset is now below 2.5 minutes. With about half a minute a day in catching up, everything will be back to normal by next wednesday.

      1. I hear this “low frequency” a lot.

        The number of periods in an hour is related to the average frequency. The frequency bounces around quite lively, but it’s the average that counts for those clocks. If after losing 6 minutes worth of cycles, things would magically suddenly run at exactly 50Hz, no “low frequency” happens, but still the clocks stay 6 minutes slow.

        They are still correcting for about half a minute worth of cycles per day. So things will be back to normal in about three days. (offset is around 1.5 minutes now).

  16. Yes, there are AC-AC wall warts – they have a transformer. If a digital clock has uses a AC output wall wart, it almost certainly uses power line frequency for time keeping.

    1. My first dance with AC-AC wall warts was a guitar analog delay pedal… I put it on my pedal board using the DC power distribution system I had, and it made horrible whale sounds. It took me FOREVER to figure out that it wanted AC @ 50Hz (we don’t even use 50HZ in the US) and that without that, everything fell apart.

      1. I am quite sure, this analog circuitry used the AC to make symmetrical voltage (+/-12V) for the OpAmps inside with capacitive voltage doubling. So I can not really think of an influence of the mains frequency. For audio delay you need higher clock frequencies than 50 or 60Hz. I would not expect a mains synchronized PLL in this device.
        Did it really not work with an AC voltage of 60Hz?

  17. Motor driven clocks use the line frequency. In Japan you know a clock is timed with line frequency if it has a 50/60 Hz switch.
    So do other devices that either use the frequency for some timing purpose or otherwise have a power supply that can’t handle the difference automatically.

    HaD should do a retro tech article about how Japan’s divided electric supply with two frequencies came from the USA and UK sharing the job of rebuilding and extending their electrical infrastructure after WW2.

  18. By the old ghods and the new: Clean that board and replace those damnable electrolytics!
    …oh…I’m sorry. I work QA doing repair evals and slipped into work mode for a second…
    (Shot and a beer hasn’t kicked in yet. Only excuse i gots…) :|

  19. Very interesting piece Mike. I have 2 battery powered clocks (AA battery) and one mains alarm clock that probably has that same chip as yours as they are about the same vintage and use the slide for minutes/hours. I really enjoyed reading this and learned a few things.

  20. Have this problem here. Incidentally many clock radios can be “hacked” to run without mains simply using a basic oscillator that uses multistep current/voltage to get an approximation of the required 50Hz but with atomic precision.
    It syncs to mains using one of those tiny 5V SMPS’s (got one here) and oscillator charges a capacitor bank so that if mains goes out the only effect is that brightness drops a bit.

    1. You do not need to approximate something with “multistep current voltage”. You just need a clock signal and you do not need “approximately” 50Hz, you need _precisely_ 50Hz for timekeeping

  21. TCXO = Temperature *Compensated* Crystal Oscillator. Essentially it measures the temperature and has a good idea how far off the crystal oscillator will be at the temperature and compensates for it.

  22. What surprised me is that my parents kitchen radio from BOSE (quite expensive) was running just as late as the microwave oven and other, very older clocks.
    I didn’t expect a BOSE device to be mains frequency timed *shakehead*

      1. Dyson is another company that falls into this category. Some of their motor stuff is impressive, but in the end the things are too damned expensive for what they do. My $60 shop vac will out suck any Dyson

  23. Well, it turns out that 2 clocks in my house were running from mains. One was a clockradio alarm clock, the other was my microwave. And I noticed that the time drifted on both.

    All my other clocks are either run from batteries and drift a few seconds per month, or are DCF77 clocks and only need batteries sometimes, or are connected to the internet and use NTP, or use GPS for time synchronisation.

    Ah well, I also have a really old Nicolet logic analyser (running on CP/M!), for which I have to enter the time every time I switch it on. ;)

  24. Yes, I have a 25+ year old Radio Shack/ Micronta 3.5″ LED clock that takes power from an 11V AC wall wart. It keeps near perfect time here in the USA, except when the power is out. When the mains power is off, the display is also, but the clock keeps running. When the mains power returns, the display comes on again and the “correct” time is shown again. The interesting thing is that the clock runs a bit fast when the mains is out, gaining as much as 1 minute per hour. (a very crude estimate) There are definately two different timing schemes going on between mains on and off. :-)

  25. NOW that you mention it.. My KitchenAide Stove loses time proportionate to power outages. My fridge knows no time, but alerts me to look. The microwave also seems to lose time then. I DO like to know how long an outage was!!! This could spawn a new clock-need/feature.

    I do bate DLST but worse us when dark eves come all too soon and then BLAM! The gubbamnt makes it 1 hr WORSE! It was explained that mommies did not want their children to go to school in the dark. Well, they used to carry a kerosine lantern to help w pre-breakfast chores, way pre-dawn. This gonna make them too tired for Nintiendo later??? Sheesh. I had it easy. I woke at 4AM to do math that I refused to do the night before. And on the weekend to hunt with dad. My option. In 1970s German companies had a mandatory 3 start times an hour apart to spread traffic out.

  26. Strict adherence to to 60 Hz power generation is mandated by law in the US; generators are adjusted once per day to ensure a 24-hour average of exactly 60 Hz. TEC (Time Error Correction) is mandatory, and results in a frequency accuracy measured in parts-per-million. Any electrical/electronic clock in the US, which uses ‘the mains’ for time-keeping purpose, is inherently more accurate than any (except ridiculously expensive) stand-alone electronic clock.

    Electronic clocks which ‘keep time’ during a power outage revert to a low-cost oscillator as a time source during an outage–usually an R-C oscillator, but, if the clock is very expensive, a ceramic resonator (all things considered, it’s almost a toss-up as to which solution results in lower production costs). The only purpose of a back-up oscillator is to allow the clock to keep approximate time–not accurate time.

      1. …Not certain what you are advocating; if the single-piece, individual-consumer price is $1.50, it’s conceivable that the high-volume price for use in mass-production would be one-tenth that amount or less–perhaps 10¢ or 15¢; far higher than the current current cost of the solution to power-outage timekeeping (which only needs to be ‘close’ in order to be acceptable).
        TEC short-term accuracy is on the order of parts-per-million; its long-term accuracy is perfect, due to corrections being made to the generators every twenty-four hours. This is the reason that motor-driven clocks have been known to keep perfect time while being run for years. Electronic clocks with AC-mains-derived timing would show perfect time-keeping–over years–also, but they tend to be discarded after a relatively short time in favor of ‘trendier’ devices; or their semiconductors or other electronic parts fail.
        The major, over-riding reason that well-designed AC motors fail after…well, forever, is that the bearings fail; all lubrication has disappeared. Well-designed AC motors, without commutators, are counted among our more perfect industrial-age machines.

  27. About home invertors and power islands… if no mains power is present, the invertor detects this and ahuts down. The reason is to avoid feeding power back into lines, which can overload your equipment, but most importantly, kill a line repairman who believes all power is gone. But illegal? No. However, you must have an automatic equivalent to positively remove your home from the grid.

  28. I agree: DS3232 (and DS3231) are great chips. They can be soldered manually due to the few pins needed and the good layout of the pins, despite the SOIC package. My ‘Time Line Clock’ uses the DS3231 as a battery backup clock and as a second’s pulse source together with a DCF77 receiver and a central Atmel AVR to be a clock without any button: it just shows the time (and it does it incredibly self-sufficiently and precisely).

    I was horrified to find my new expensive stove/oven to use mains as a clock source. It may indeed use the LM chip you identified here. But consumers do not seem to care (probably do not know how cheap a good clock is) and the extra 1 EUR is just too much for the company to produce something less annoying. That clock is on my replacement list.

    1. What do mean with:”…soldered manually…despite the SOIC package…”? SOIC packages are huge and always easily soldered manually. TSSOP and 0402 is a little more difficult, DFNs and QFNs are tricky, or 0805 common mode choke with 4 pins.

  29. I was quite irritated that my 800€ oven was affected by this. You really can’t argue with me that in a product of that price range you can’t spare 50 cents for an accurate crystal or even better one of those DCF77 recievers. Especially given the fact that the oven has quite extensive timing features like automatically starting it at a given time and and running for x minutes and then turning off again. So its not just a stupid clock that has no real use.

  30. We (mostly yours truly) designed a clock synchronization subsystem based off of NTP sources at a former workplace. Based on the required periodic correction, it would adjust the frequency of the clock so that further corrections would be reduced. We achieved the theoretically minimal clock drifts for the OS, like ± 50 ms per day. We’d often be able to detect if servers were plugged in or removed locally by detecting a slightly larger adjustment on a particular day.

  31. most of the actual “clocks” in my house are still mains analogue clocks that rely on the 60hz for the motor to run – always fun when the generator kicks on and we loose about 10 minutes a day from the sloppy power

    1. If I had mains disruptions that often I would connect the clocks to a small UPS with crystal controlled frequency. That is probably easier than controlling the speed of the generator really precisely.

  32. How about running an a 240 vac undersea power cable (3-conductor; wire size of 4/0; standard U.S. 200 amp Service Entrance cable–ought to do nicely) from Con-Ed in New York, to the other side. Then we can provide backward GB and Europe with accurate timekeeping. Forget the 60 Hz issue; a minor detail; a mere bag of shells, as Inspector Clouseau would say.
    Oh, wait…”Governor Andrew M. Cuomo today directed the New York State Department of Public Service to conduct an immediate investigation into Con Edison’s substation failure on Sunday afternoon that led to power failures across Brooklyn…the status quo can’t continue…”

  33. The frequency of the transmission system was, is and will be never stable at exactly 50Hz for a reasonable amount of time, either leading or lagging depending on the generation, loading and other conditions of operation of the transmission system.The following regulations apply : Short term +3%/–5% frequency variation range and normal ±2% frequency variation range as defined by
    IEEE C50.13 and IEC 60034-3 . Therefore the claim is nonsense

  34. Personal experience from about 25 years ago. On a consumer grade (aka “VERY inexpensive”) range timer, we literally brought the 110v 60hz line directly into an NMI interrupt pin on a 4-bit CPU (Yep.. really 4 bit) via an R/C divider. count 60 interrupts and increment the seconds by one. Optionally you could set for 50 or 60 hz independently of the 12/24 selection.

  35. I’ve clock alarm that started running faster (something like 1.5x times faster than normal).
    I’m not an electronic expert, but I’ve measured the pin 25 of the LM integrated circuit with my multimeter, and i’ve got a frequency of 49.99 Hz, does this mean that the IC have gone bad after almost 25 years of usage? Or there may be a change that some other component isn’t working?
    I’ve noticed that also the two dots : blink also faster

  36. About once a week my clock just flashes the electric hasn’t went out nothing no other fox is doing it just the one in my room I’ve been trying to figure it out for a few weeks it just started a couple months ago

Leave a Reply

Please be kind and respectful to help make the comments section excellent. (Comment Policy)

This site uses Akismet to reduce spam. Learn how your comment data is processed.