
>"or more precisely A1+"
Ironically less precise owing to typo!
US researchers recently named the atomic clock at the UK's National Physical Laboratory (NPL) in London the most accurate atomic timepiece on the planet. Experts from the University of Pennsylvania found that NPL-CSF2 loses just one nanosecond every two months. It's a badge of honour for NPL, the birthplace of the first …
No circular reference; all clocks work by having something happen "many times a second", the "something" (like a pendulum swing for example) will have a known frequency and the clock mechanism just counts how many cycles have occurred.
The tricky bit is working out the frequency of the thing you're using to "measure time" (most "real world" clocks are calibrated against a known accurate time source).
I'm fairly sure that Hz is defined as cycles per second, and as these atomic clocks are "more accurate" than any other clock, they don't have a "known accurate time source" to measure against, thus the comment about them having a circular reference. IMO this will become increasingly true if they ditch UTC, as they won't even have the 24/60/60 breakdown of the solar day.
Wouldn't it be better for them to define a second as x number of y event that occurs when z quantity of material has been heated up to temperature t?
as the duration of 9 192 631 770 periods of the radiation corresponding to the transition between the two hyperfine levels of the ground state of the caesium 133 atom.
So, you just have to count cycles, until you get enough to make up a second.
The issue is how to count and average the radiation. It's the error in generating and detecting the radiation that's key.
You compare your time with the average of a bunch of other clocks around the world.
Some individual clocks drift faster or slower - if yours stays at the average you are the winner.
You have to love the NPL attitude, leap seconds will become unworkable in only 1700years and if a massive comet hit the earth everyone would have to adjust their watches.
NIST (the US National Institute of Standards and Technology) has a pdf booklet named "From Sundials to Atomic Clocks: Understanding Time and Frequency" available at http://www.boulder.nist.gov/timefreq/general/pdf/1796.pdf
NIST has a good description of NIST-F1 which is a cesium fountain atomic clock at http://www.nist.gov/pml/div688/grp50/primary-frequency-standards.cfm
Nice article, but it's a pity that despite there being so many accurate time sources out there, nobody bothers to use them.
My VoIP desk phone, which presumably gets its time from an NTP server somewhere, is always showing a different time to my laptop connected to the same network...
Well, their time might be accurate, but the height of a foot (at El Reg!) is way off.
"NPL is at an elevation of 79 feet above sea level and suffers more from the effects of gravity on its results than Fort Collins, which is more than 3,315 ft, above sea level in the Rocky Mountains in Boulder Colorado, where the effect of gravity is less."
Wow. Fort Collins is in Boulder?? at 3315 feet?? REALLY?!?
(Is somebody, used to sea level, getting dizzy from the altitude?)
Denver, Colorado is famous as "The Mile-High City" (statue mile, of course, so 5280 ft.).
Boulder Colorado is just a bit higher - 5430 ft above sea level (at City Hall). It has the Univ of Colorado - and the NIST Time and Frequency Division. Don't know the elevation at their clocks, but it should be close.
( http://www.nist.gov/pml/div688/staff.cfm )
Fort Collins, Colorado, is NOT in Boulder, Colorado. It's not even in the same county. It's about 32 miles further north than Boulder, and is the home of Colorado State Univ. -- as well as NIST "time" radio stations WWV & WWVB, and an Internet time server. It's also lower -- 5003 ft at City Hall.
( the 10 MHz antenna is at: 40° 40' 47.8" N 105° 02' 25.1" W - to within 10 feet or so )