Be still, my quivering atoms: Here's a new way to count a second Atoms trapped by light and measured with a CCD camera have the potential to outdo today's most accurate atomic clocks, and although it's early days, a pair of linked optical lattice clocks have yielded accuracy of a second every 300 million years. In doing so, the group that performed the test, outlined here in Nature and …
A big application of GPS is the ns accuracy time pulses.
A lot of infrastructure uses these and things would get difficult if the DoD felt the need to pull the plug.
A chip version (and some people are working on such an idea) would make this a backup system at best.
IIRC with an accurate enough clock you can eliminate exterior signal systems altogether, although I'm not quite sure how this helps you cancel the gyroscope drift effects.
However generally thumbs up for moving forward the "Frontiers of Measurement."
You can't dispense with the exterior signals because it's only by measuring their time-of-flight that you can triangulate your position. Having a clock in the GPS receiver with that kind of accuracy just means you could get a 3D fix with 3 satellites minimum instead of 4, so unless you could make one for a few pence, there would be little point in having it.
The 1pps output from a GPS does have significant jitter. Most systems will use a local oscillator, either Rubidium or Cesium based, that is then disciplined by the GPS pulses. The local oscillator provides your short-term stability, with the discipline providing long term stability.
A device like this optical lattice is unlikely to be miniaturized very much, and you aren't going to see that being used for something like keeping a radio site on frequency.
Being a bit of a dunce, how do you measure the second more accurately?
If we currently know a second to be so many vibrations of an atom, and using light here we increase the frequency so yes it is more accurate.
But what about the fact that the atom originally used was not perfect (hence the need for more accuracy) meaning that what has been measured here is slightly longer or maybe shorter depending on the measurement.
Have they actually redefined the length of a second here?
The definition of a second did not change; however we might be using a different second than before because we are able to stick closer to what the definition is.
It's like the definition of π never changed; it always has been the ratio between the perimeter and the diameter of a circle. However, with centuries we have been able to know more precisely what the actual number is, from ~3 to 22/7 to billions of digits.
Yes but as the time before was measured as the radiation of a caesium atom, then this would be in waves yes? (correct me if I am wrong anywhere please)
We know that this is off by so many seconds in so many millions of years based on what? ephemeris time?
I am not a mathematician but is the ephemeris time formula the absolute time for a second? Is the new accuracy then measured against this formula rather than against the wave of radiation from caesium that we know is not accurate?
We know that this is off by so many seconds in so many millions of years based on what? ephemeris time?
No, based on known flaws in the measuring system. A caesium second is defined as 9,192,631,770 cycles of radiation corresponding to the transition between two energy levels of the caesium-133 atom.So to work flawlessly, the system would have to capture all of those cycles (or at least synchronize with them.) Signals from other atoms, temperature fluctuations, and a whole range of factors beyond my ken can all interfere with that. Scientists use those flaws to calculate the probability of a system missing a cycle or counting an extra one, and then adding up that probabilistic effect to see how long it would take to be off by as much as a second. In the case of the UK's NPL-CsF2, that's about 138 million years, or roughly one missed or extra cycle every 5.5 days (net; errors in opposite directions would offset.)
So what we're talking about here is actually expected inaccuracy, not empirical inaccuracy.
It's not a question of the clock being "fast" or "slow" vs. some other reference, but rather a question of uncertainty - if you had 2 of these clocks, and let them run for some period of time, they will be out of sync with each other to some degree. In this case, the expected amount of drift between them would be 1.5e-16 - if they ran for one second, they'd be at most 1.5e-16 second difference. Which one is "right"? Both, and neither - it's a question of probability, accuracy, and repeatability.
I remember that in D.H. Lawrence's The Plumed Serpent the clocks were destroyed and replaced with a guy in the clock tower who would announce "halfway up the hill of the morning" and "halfway down the hill of the afternoon," etc. In a larger sense, in Tracy Kidder's Soul of a New Machine, one of the engineers quit, leaving a resignation note which read, "I quit. I am going to a commune in Vermont and will deal with no unit of time shorter than a season."
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2025
