[Noisebridge-discuss] Velocity of Time. De-syncing the electric grid

[walter park]

Time.
To see the constant variation in powerline frequencies, here's a real-time
U.S. map maintained by U. Tennessee:

http://fnetpublic.utk.edu/gradientmap.html

F wobbles 0.04 Hz over just a few seconds.  If you look around the site,
you'll see how they correlate those changes in F and voltage to 'events'
like power stations going off-line.

To see how stable a really stable oscillator can (cesium, not quartz!),
here's a guy who took his kids and three Cesium clocks 1300 meters up Mt.
Rainer to measure the difference in the velocity of time at higher
altitude.  In a two-day trip he observed a 22 ns speed-up v. an Einsetin
prediction of 23 ns!!

Why three cesium clocks.  Well, because he can, of course.  He also says, *A
person with one clock knows what time it is. One with two clocks is never
sure*. But I would add further: *A person with three clocks is more sure
than a man with two clocks*. And so my clock collection started...


http://www.leapsecond.com/great2005/index.htm




On Sat, Jun 25, 2011 at 20:57, Jonathan Foote <jtfoote at ieee.org> wrote:

> Interesting!
>
> Actually, quartz crystal frequency can vary in the tens of ppm at
> constant temperature due to aging and other factors, and a rule of
> thumb is ~1 ppm change per degree C. Since 12 ppm = 1 second per day,
> the AC powerline frequency -- set to precisely 24 X 60 X 60 cycles per
> day -- is typically more accurate.
>
> (In the old days the operators would change the AC frequency slightly
> near midnight to add or subtract a few cycles so they came out to the
> right number; these days it's continuously controlled by a feedback
> loop -- at least until they bork it.)
>
> The beauty of the system is that if you use a synchronous motor on
> your clock, it just works. If you want to make a microcontroller-based
> clock, you'd do better to count AC cycles than listen to your own
> crystal! If you want to run off DC, use a temperature-compensated RTC
> module like Garrett's Chronodot.
>
> There was at least one proposal to fingerprint remote computers by the
> distinctive clock skew on their TCP timestamps.
>
>
> On Fri, Jun 24, 2011 at 7:44 PM, Taylor Alexander <tlalexander at gmail.com>
> wrote:
> > "Since 1930, electric clocks have kept time based on the rate of the
> > electrical current that powers them."
> > Uh, really? I know that its possible to do, but I thought ever since
> > we figured out how nice quartz crystals worked we stopped making
> > clocks like that.
> >
> > Is this method of timekeeping really all that common anymore?
> > -Taylor
> >
> > On Fri, Jun 24, 2011 at 7:27 PM, walter park <waltsfo at gmail.com> wrote:
> >>
> >> AP story:
> >>
> >> Experiment with power grid may throw off your electric clock
> >>
> >> By SETH BORENSTEIN
> >> AP Science Writer
> >>
> >> 06-24-2011 15:57
> >>
> >> WASHINGTON (AP) _ Our power supply has been so precise that we've set
> our clocks by it. But time may be running out on that idea.
> >>
> >> A yearlong experiment with the electric grid may make plug-in clocks and
> devices like coffeemakers with programmable timers run up to 20 minutes
> fast.
> >>
> >> The group that oversees the U.S. power grid is proposing a change that
> has the potential to disrupt electric clocks in schools, hospitals and other
> institutions, according to a company presentation obtained by The Associated
> Press. It may also mess with the timing of traffic lights, security systems,
> sprinklers and some personal computer software and hardware.
> >>
> >> Since 1930, electric clocks have kept time based on the rate of the
> electrical current that powers them. If the current slips off its usual
> rate, clocks run a little fast or slow. Power companies now take steps to
> correct it and keep the frequency of the current _ and the time _ as precise
> as possible.
> >>
> >> The experiment would allow more frequency variation than it does now
> without corrections. Officials say they want to try this to make the power
> supply more reliable, save money and reduce what may be needless efforts.
> The test is tentatively set to start in mid-July, but that could change.
> >>
> >> Tweaking the power grid's frequency is expensive and takes a lot of
> effort, said Joe McClelland, head of electric reliability for the Federal
> Energy Regulatory Commission.
> >>
> >> "Is anyone using the grid to keep track of time?"
> >>
> >> McClelland said. "Let's see if anyone complains if we
> >> eliminate it."
> >>
> >> They will and they should, timekeepers say.
> >>
> >> "A lot of people are going to have things break and they're not going to
> know why," said Demetrios Matsakis, head of the time service department at
> the U.S. Naval Observatory, one of two official timekeeping agencies in the
> federal government.
> >>
> >> The changes, however, are out of the hands of timekeepers and in control
> of officials who supply the electrical power.
> >>
> >> No one is quite sure what will be affected. This won't change the clocks
> in cellphones, GPS systems or even on computers, and it won't have anything
> to do with official U.S. time or Internet time.
> >>
> >> But wall clocks and those on ovens and coffeemakers _ anything that
> flashes "12:00" when it loses power _ may be just a bit off every second,
> and that error can grow with time.
> >>
> >> It's not easy figuring what will run fast and what won't. For example,
> VCRs or DVRs that get their time from cable systems or the Internet probably
> won't be affected, but those with clocks tied to the electric current will
> be off a bit, Matsakis said.
> >>
> >> This will be an interesting experiment to see how dependent our
> timekeeping is on the power grid, Matsakis said.
> >>
> >> The North American Electric Reliability Corp. runs the nation's
> interlocking web of transmission lines and power plants. A June 14 company
> presentation spelled out the potential effects of the change: East Coast
> clocks may run as much as 20 minutes fast over a year, but West Coast clocks
> are only likely to be off by 8 minutes. In Texas, it's only an expected
> speed-up of 2 minutes.
> >>
> >> Some parts of the grid, like in the East, tend to run faster than
> others. Errors add up. If the grid averages just over 60 cycles a second,
> clocks that rely on the grid will gain 14 seconds per day, according to the
> company's presentation.
> >>
> >> Spokeswoman Kimberly Mielcarek said the company is still discussing the
> test and gauging reactions to its proposal, and may delay the experiment a
> bit.
> >>
> >> Mielcarek said in an email that the change is about making the grid more
> reliable and that correcting the frequency for time deviations can cause
> other unnecessary problems for the grid. She wrote that any problems from
> the test are only possibilities.
> >>
> >> In the future, more use of renewable energy from the sun and wind will
> mean more variations in frequency on the grid, McClelland said. Solar and
> wind power can drop off the grid with momentary changes in weather.
> Correcting those deviations is expensive and requires instant back-up power
> to be always at the ready, he said.
> >>
> >> The test makes sense and should not cause too much of a hassle for
> people, said Jay Apt, a business professor and director of the Electricity
> Industry Center at Carnegie Mellon University.
> >>
> >> But Tim O'Brian, who heads the time and frequency division at the
> National Institute of Standards and Technology,
> >>
> >> expects widespread effects.
> >>
> >> He said there are alternatives if people have problems from the test:
> The federal government provides the official time by telephone and on the
> Internet.
> >>
> >> Official U.S. government time:
> >>
> >> http://time.gov or call 202-762-1401
> >>
> >> _______________________________________________
> >> Noisebridge-discuss mailing list
> >> Noisebridge-discuss at lists.noisebridge.net
> >> https://www.noisebridge.net/mailman/listinfo/noisebridge-discuss
> >>
> > _______________________________________________
> > Noisebridge-discuss mailing list
> > Noisebridge-discuss at lists.noisebridge.net
> > https://www.noisebridge.net/mailman/listinfo/noisebridge-discuss
> >
> _______________________________________________
> Noisebridge-discuss mailing list
> Noisebridge-discuss at lists.noisebridge.net
> https://www.noisebridge.net/mailman/listinfo/noisebridge-discuss
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.noisebridge.net/pipermail/noisebridge-discuss/attachments/20110626/a0ef8cdf/attachment-0002.html>