Forgot your password?
typodupeerror
Hardware Hacking Build Science

Playing With Atomic Clocks At Home 167

Posted by kdawson
from the as-i-was-walking-down-the-street-one-day dept.
Wired is running a profile of the Time Nuts, a small group of people who buy surplus precision time equipment — cesium clocks for example — on eBay and keep really accurate time, because they can. The article quotes Tom Van Baak, who has outfitted a time lab superior to those of many small countries: "If you have one clock... you are peaceful and have no worries. If you have two clocks... you start asking, 'What time is it, really?'"
This discussion has been archived. No new comments can be posted.

Playing With Atomic Clocks At Home

Comments Filter:
  • by brunes69 (86786) <slashdot@keirst[ ].org ['ead' in gap]> on Tuesday December 11, 2007 @10:20AM (#21655023) Homepage
    You do know that they proved this like 60 years ago right, when the first Atomic clocks were produced.... In addition there is an atomic clock on the shuttle. The time difference between it and it's perfectly synchronized counterpart on earth is very visible.
  • by FridayBob (619244) on Tuesday December 11, 2007 @10:37AM (#21655205) Homepage
    NTPD isn't good enough for me -- bad weather on the Internet has caused my server to loose synchronization one too many times, which can be mighty irritating when comparing your log files with those of other systems. On the other hand, acquiring an atomic clock seems a bit over the top to me. So, I figured a good compromise solution would be to connect a GPS receiver to my serial port and synchronize NTPD to that. I've ordered a Garmin GPS 18 OEM LVC [garmin.com] that I will receive later this month (hopefully). According to these instructions [qnan.org] it's not that difficult to set up, while the result is microsecond precision on Linux 2.6 and nanosecond precision on BSD -- good enough for me. All you need to do is to make sure that your GPS device has a reasonable view of the sky.
  • by Relic of the Future (118669) <dales@@@digitalfreaks...org> on Tuesday December 11, 2007 @11:11AM (#21655595)
    They got the saying all wrong. It goes "A man who wears one watch always knows what time it is; a man who wears two watches is never sure."
  • Re:The real reason (Score:5, Informative)

    by timeOday (582209) on Tuesday December 11, 2007 @11:31AM (#21655897)

    Isn't it obvious? It wasn't relativity, the family lived an extra 22 milliseconds because they drove up a mountain
    After noticing your comment I read the whole article just because 22 ms sounded like an impossibly large relativistic effect for a car. It was actually 22 nanoseconds. You're off by a factor of a million.
  • by Device666 (901563) on Tuesday December 11, 2007 @11:57AM (#21656319)
    Despite the amount of confusing clocks you might have, a benevolent dictator of time becomes handy: Temps Atomique International (french) abbreviated: TAI. For us mere mortals who use time for civil needs, another timescale is dissiminated, Coordinated Universal Time (UTC). UTC is derived from TAI, but synchronized using leap seconds to UT1, which is based on actual rotations of the earth with respect to the mean sun.

    International Atomic Time (TAI, from the French name ) is a high-precision atomic time standard that tracks proper time on Earth's geoid. It is the principal realisation of Terrestrial Time. As of 2007 TAI is exactly 33 seconds ahead of UTC: 10 seconds' initial difference at the start of 1972, plus 23 leap seconds in UTC since 1972. TAI in this form was synchronised with Universal Time at the beginning of 1958, and the two have drifted apart ever since.

    Accurate time is very important for computer systems/networks. The best way to keep track of time is to install a local timeserver which synchronizes against a reliable public timeserver like pool.ntp.org. The local time server can be used to synchronize other computers you might have.
  • by russotto (537200) on Tuesday December 11, 2007 @12:16PM (#21656699) Journal
    With a GPS time source you get the time value (in ASCII) through a serial port, but the synchronization is done through a pulse per second interrupt. So the latency on the serial bus doesn't matter as long as it is significantly less than one second.
  • by Anonymous Coward on Tuesday December 11, 2007 @12:17PM (#21656711)

    NTPD isn't good enough for me -- bad weather on the Internet has caused my server to loose synchronization one too many times, which can be mighty irritating when comparing your log files with those of other systems.

    How many upstream servers do you sync to? "Bad internet weather" shouldn't desync your NTP server given enough peers to talk to. My local NTP server syncs up to five remote servers spread over three continents, and the only event that my NTP server process noticed was my DSL modem melting down.

    My long-term NTP traffic rate is maybe 1 packet every 5 minutes, so it takes a hella big network disruption to make NTP take notice.

    I've ordered a Garmin GPS 18 OEM LVC that I will receive later this month (hopefully). According to these instructions it's not that difficult to set up, while the result is microsecond precision on Linux 2.6 and nanosecond precision on BSD -- good enough for me. All you need to do is to make sure that your GPS device has a reasonable view of the sky.

    ... and that you have a way of using the GPS18's PPS output to set your clock, because the *ONLY* way you are going to get sub-second accuracy out of your setup is to set your clock as soon as possible after the PPS output goes active. The GPS isn't going to give you any other help, since its NMEA stream only provides time accuracy to one second (or 0.1 second if you get the GPS18 version that gives you five PPS transitions per second). The PPS output is the only authoritative source as to when that second occurs.

  • Re:I don't get it. (Score:4, Informative)

    by ChrisA90278 (905188) on Tuesday December 11, 2007 @02:24PM (#21659271)
    What's the fascination with uber-accuracy at home?

    They are calling these "clocks" only because that is what the typical reader understands. A better term is "frequency standard". There are many uses for a stable frequency, the most common one is running a microwave transmitter. This is the major source of the surplus devices too, from cell towers. As the phone companies modernize equipment these "clocks" find their way to eBay and then into people's houses.

Are you having fun yet?

Working...