Slashdot Mirror


Power Grid Change May Disrupt Clocks

hawguy writes with an AP story about upcoming tests of greater allowed variation in the frequency of the current carried on the U.S. electric grid: "A yearlong experiment with the nation's electric grid could mess up traffic lights, security systems and some computers — and make plug-in clocks and appliances like programmable coffeemakers run up to 20 minutes fast."

22 of 439 comments (clear)

  1. here's the scale by Flyerman · · Score: 4, Informative

    20 minutes fast over the course of a year.

    1. Re:here's the scale by toastar · · Score: 3, Insightful

      3 seconds a day. But twice a year people manually change the time due to summer time.

      Wait... You check the accuracy of the minutes when daylight savings comes rather then just hitting the +1 hour button?

    2. Re:here's the scale by jd · · Score: 3, Informative

      The PET 3032 (back in 1978), free-running and unsynchronised, was capable of 30-seconds-per-year accuracy on a decent, clean power supply. That was, admittedly, about the absolute limit, but you could do it. A modern computer runs around 4 billion times as many cycles per second. More if you supercool then overclock it. A modern computer also has up to 16 cores per node and fairly typical clusters can have 64 nodes.

      As for analog watches, the high-end mechanical watches you can buy off-the-shelf have a drift of around 1 second per day (30 times better than your estimate and 3 times better than any computer is capable of doing if the power supply will induce 3 seconds a day error). For free-running digital devices, a typical Casio quartz digital watch is around six nines accuracy (0.1 seconds drift a day), no synchronization required. Which means you can actually buy a cheap wristwatch that's 30x more accurate on timing than the best home computer you can get.

      Sorry if I find the incompetence of hardware engineers a little hard to accept, I just prefer standards that, y'know, improve over time, not regress. 3 seconds a day drift is what vintage Swiss watches could do. I prefer modern technology to do better than the stuff that Huygens could do, not merely equal it.

      --
      It's a small world and it smells funny; I'd buy another if it wasn't for the money; Take back what I paid (SoM)
    3. Re:here's the scale by Achra · · Score: 4, Informative

      WWV is weak on the West Coast. My watch sync attempts to sync to WWV every night but it is only successful about one day in three. This was initially frustrating because the device is supposed to adjust for DST but would fail to make the switch because it could not receive the signal on that day. Now the changeover day is different from what is programmed in so it doesn't work anyway.

      Whoever designed that watch did a crummy job. There is a bit in the WWVB packet that tells the clock the current DST status: http://en.wikipedia.org/wiki/WWVB

      I've designed a WWVB nixie clock before, there are definitely some serious design constraints. For starters, the signal is 60khz which is "Longwave". You need a lot of antenna to pull in a longwave signal with any real success. I used a very long loopstick antenna, and even then the antenna is directional, so the direction the clock is oriented _matters_. Additionally, lots of things generate noise (QRM) in these frequencies, so watch out for CRT televisions and computers and (in the case of nixie clocks: high voltage switching supplies and multiplexed nixie tubes ionizing and de-ionizing neon hundreds of times a second). I built this thing because I am a HAM and also into nixie tubes.. but the truth is that WWVB is obsolete. Nowadays, the best way to get accurate time is via GPS. You can buy a GPS module to pull the time from for about $25-$35 to build into your clock.

      But I am sad about this line frequency change.. In the United States, one of the most accurate clock signals is the 60hz power. It's accurate to within about a minute a year at present. That is a LOT more accurate than a standard crystal. TCXO's (calibrated crystals that have temperature sensors in them that dynamically recalibrate for temperature) get to about a few seconds a year when they are brand new and then degrade from there with age. So, the long and the short of it is that if this change happens, and if it is a pretty noticeable hit to my clocks' accuracy, I'll be bodging in little TCXO controlled 60hz sinewave generators into all of my clocks.. :(

      --
      Each processor would proceed sequentially as if it had been better for them not to rise against Saul.
  2. Electric clocks by JohannesJ · · Score: 4, Informative

    Most clocks are not electric .Most Run on DC provided by a Crystal oscillator, the line frequency provided by the AC line to run them is irrelevant. only electromechanical electric clocks might be in error

  3. Re:Nevermind cheapo clocks by heypete · · Score: 3, Informative

    If your laptop power supply is anything like all the ones I've owned, it won't care. According to the label (and testing done while I travel), mine works just fine on nominally 50-60Hz mains power. I imagine it wouldn't really care if you went from 45-65Hz, though I suspect it might get a bit annoyed if you were to go to 400Hz or something extreme.

  4. Re:Wow... by blair1q · · Score: 3, Insightful

    Any clock with the word "quartz" associated with it is using a crystal timebase to determine how long a second, minute, hour, day, etc. are.

    http://en.wikipedia.org/wiki/Quartz_clock#Accuracy

    Half a second per day, even if the power goes out and you're running off a battery, no matter if the mains frequency wobbles (which it's designed to do anyway).

    http://en.wikipedia.org/wiki/Utility_frequency#Stability

  5. The real question by thegarbz · · Score: 3, Insightful

    The real question is why do devices add the additional circuitry to count pulses off the mains grid rather than add additional circuitry to actually keep time?

    A highly accurate crystal costs in the order for $1 for single quantities. A RTC $1-10 depending on feature set. If you already have a microcontroller you don't need the RTC either. Why are clocks etc reliant on an external signal to keep time? How do they keep time when they run on the battery which is a common backup for every $5 alarm you get?

    As for streetlights ... Really? How is this not a system which gets timing from some other central authority. I don't know much about street lights, but is this something that will only affect old small town streetlights, or do the shiny new modern LED powered ones in the city act independently enough that they aren't capable of contacting an NTP server?

    1. Re:The real question by Ungrounded+Lightning · · Score: 3, Informative

      A highly accurate crystal costs in the order for $1 for single quantities.

      And gains or loses perhaps a minute per year - while the grid has been good for a fraction of a second (adjusted when the powerhouse clocks drift more than that from the national standard committee of atomic clocks).

      So that's why line-powered clocks use the line for the primary reference and the crystal oscillator to avoid having to reset it after a power failure (and to insure you get your wake-up alarm). And why most appliances don't bother with a crystal at all. (Why spend extra to make them LESS accurate?)

      Keeping accurate time is HARD. Distributing it by the power grid is EASY.

      --
      Bantam Dominique roosters crow a four-note song. Once you've heard it as "Happy BIRTHday" you can't NOT hear it that way
  6. Re:"Clocks" by hawguy · · Score: 3, Informative

    Clearly, whomever thought this was a Pretty Neat Idea hasn't read this:

    http://yarchive.net/car/rv/generator_synchronization.html

    and doesn't understand what happens when you're even a bunch of *degrees* out of sync, much less a few decihertz. We don't have *near* enough HVDC intertie to make this not matter, and I can't imaging how they think this is gonna work -- nothing at all on NERC's website to say what's *really* gonna happen, either.

    Love all the warning, too.

    I think the organization that's responsible for the reliability of the entire USA power grid has some idea of the need for frequency stabilization when connecting new power sources to the grid. Not that it's relevant for what they are proposing - power plants already know how to sync up their generators to the grid and they don't care if it's 60.001 Hz or 60.002 Hz, they'll take that into account.

    The magnitude of this frequency deviation is tiny, 20 minutes/year is about .003% - the power grid can fluctuate much more than than on a daily basis, but until now, it's always been corrected to keep the overall frequency at 60 Hz.

  7. Re:"Clocks" by John+Hasler · · Score: 4, Interesting

    Believe it or not, the engineers that operate the network actually know what they are doing.

    The flow of power between tied AC networks is determined by phase, not voltage. To adjust the phase between your generator and that of a neighbor to whom you wish to send power you must run faster than he for long enough to accumulate the desired phase difference. Such adjustments are going on constantly throughout the network and conflict with the requirement to keep the average frequency at exactly 60Hz. Relaxing the latter requirement will make network operations easier and more reliable.

    --
    Warning: this article may contain humor, sarcasm, parody, and perhaps even irony. Read at your own risk.
  8. Re:"Clocks" by gottabeme · · Score: 3, Funny

    The word you're looking for is "nitpick".

    --
    "Those who consume the bulk of goods are those who make them. We must never forget this secret of our prosperity."
  9. You're SO wrong. by Ungrounded+Lightning · · Score: 3, Insightful

    So untrue. Maybe it was done that way in 1930. But not anymore.

    You're wrong.

    Line powered clocks with crystal oscillators generally use the line for the reference and the crystal for a backup during power failures.

    The line has been far more accurate than a cheap crystal - at least until these goons implement their harebrained scheme.

    --
    Bantam Dominique roosters crow a four-note song. Once you've heard it as "Happy BIRTHday" you can't NOT hear it that way
  10. Re:"Clocks" by jpmorgan · · Score: 3, Informative

    The US operates three separate power grids: the east interconnect, the west interconnect and ERCOT. They're only connected by DC links and are not phase locked.

  11. Re:"Clocks" by jpmorgan · · Score: 5, Funny

    You're telling me that an Internet denizen who's only been exposed to this idea for 5 minutes doesn't know more than the engineers who have been running the system for decades?

    That goes against everything /. stands for!

  12. Re:WHY? by jpmorgan · · Score: 4, Informative

    Load on the grid shows up as mechanical resistance to the big spinning generators that control the frequency. If there is more load than generated supply, the generators slow and the frequency drops; more supply than load and the turbines spin the generators faster. Maintaining a balance of power is done by keeping the frequency at 60Hz.

    That was easy enough when all power came from big generators, with predictable loads. But if you mandate photovoltaics and wind and other forms of power which vary in output, then things are a lot harder. The wind dies and a major wind farm drops a few hundred megawatts? The big generators can't respond quickly enough to keep frequency within its regulated range, so power companies have to install very expensive systems that can react faster.

    Utilities are often legally mandated to buy power from renewable sources, but those renewable sources aren't held to any of the grid stability requirements. This ends up shifting an enormous burden of cost onto the utilities, who aren't happy with it. Loosening the grid frequency requirements is a way to make renewable but unreliable power less expensive.

  13. Using the mains as a timebase by Anonymous Coward · · Score: 5, Interesting

    In the 70's I developed a system to control the of the light sensitive coating onto 35mm rolls of film. This ran on a PDP-11 that used the mains cycle to keep time (20ms interupts with the UK's 50Hz supply) and measured the coating by the amount of x-rays reflected by the silver halide in the coating each second.... there were coninual errors in the accuracy of the coating as the time approached midnight.

    It turns out that the National Grid was legally required to maintain a 50Hz average from midnight to midnight and would add or subtract cycles in the last minutes of the day in order to meet this requirement.

    Five or so years later I was working in the National Grid Control Centre and saw the 2 clocks, one with an independent time source and one running from the mains frequency. The aim of the controllers each night was to adjust the mains frequency to bring the two clocks in sync at midnight.

  14. Re:"Clocks" by russotto · · Score: 3, Insightful

    and doesn't understand what happens when you're even a bunch of *degrees* out of sync, much less a few decihertz.

    They understand very well. This isn't about allowing generators to drift out of sync with each other in the short term. It's about not correcting the long-term variations in the grid as a whole.

    Household clocks and coffeemakers seem unlikely to be a problem. Most of them nowadays aren't synced to house current, but use quartz oscillators. More likely problems would be old systems which have never been replaced because they've never needed to be; traffic light controllers are a reasonable example.

  15. Not too many analog or power people here I see. by loose+electron · · Score: 4, Informative

    This is no big deal. What they are talking about here is the additive cycles in a day and not worrying about the compensation process for that.

    Some basics:

    Anything connected to the 60Hz power is at 60HZ, You can not connect a 61Hz generator to the grid.
    In addition, when you connect a generator to the grid, you have to adjust its phase, as you bring it on line.
    If the phase angle does not line up you get you get into a "tug of war" between multiple generation sources and that doesn't work.

    The sine wave coming out of one generator has to line up with the other sine waves from the other sine waves from the other generators.

    60 cycles/sec X 60 sec/min X 60 mins/hour X 24 hours/day = 5.184E6 cycle/day

      What the article is talking about is the adjustment of the generating stations on the grid so that at the end of the day you get that exact number of cycles across the grid, not one more not one less. It is "really close" without tweaking but not exact.

    It costs money to do those tweaks, to get the numbers on the money. That tweak right now really doesn't serve much purpose anymore.

    Noting exciting, or interesting here, this is not Y2K nonsense, move along...

    --
    www.effectiveelectrons.com "chips that work" Analog, RF, Mixed Signal
  16. Music by soundguy · · Score: 4, Informative

    They don't specify how much of a frequency swing they are talking about, but I can think of a few legacy items still in use in the music industry that are affected by line frequency.

    1) - The mainstay of every old piano tuner's toolbox is the Conn Strobe Tuner.

    2) - There are still thousands of working Hammond B/C series electric organs in use.

    3) - Lastly let's not forget the audiophiles and their vinyl record turntables.

    In fact anything with a shaded pole induction motor is speed-locked to the line frequency.

    --
    Nothing worthwhile ever happens before noon
  17. Re:Is timekeeping really that difficult to solve? by hjf · · Score: 4, Informative

    No. If you take apart a clock radio with 4x7-segment LED display, chances are that it has a 9v battery compartment at the bottom, and an LM8560 inside. It's an IC that's been used for over 30 years and still in production. No matter how cool, modern looking, flashy blue LED display it is, it has the same IC a brown 1980s clock with red LEDs had. It could be a clone or have a different name, but it is that chip.

    Guess what: it takes voltage from the transformer, before rectification, into one of the pins. It also has another pin to set 50/60Hz operation. And a SHITTY RC circuit for running off battery (useless, it's off several minutes every hour).

    Another thing: most electric things CAN'T be plugged anywhere now. My grey-market XBOX 360 has a 120V power brick (I live in a 220V country). If you live in the USA, take a look at how many electronic stuff at your house doesn't even have a 220/110V switch. The only things you can pretty much plug in anywhere are chargers. Most other stuff either can't, either by design (things with motors or appliances you don't carry around), or by cost (most electronic stuff without a 110/220V switch).

  18. Re:Nevermind cheapo clocks by hjf · · Score: 3, Informative

    LED alarm clocks still use an LM8560. Go buy one, take it apart, and find the ic with the weird pin spacing (not standard 0.1"). That's the same IC that's been in use for over 30 years. And it still runs on mains frequency (it has a pin to select 50/60hz operation).