The Naked Scientists

The Naked Scientists Forum

Author Topic: Against what is the accuracy of the most accurate clock measured?  (Read 298 times)

Offline thedoc

  • Forum Admin
  • Administrator
  • Sr. Member
  • *****
  • Posts: 508
  • Thanked: 11 times
    • View Profile
ohn Savage asked the Naked Scientists:
   Atomic clocks

I read in two online articles (in Nature, and The Verge) that caesium clocks are being superseded by strontium clocks.  Both articles state that the latter achieve an accuracy of one second in 15 billion years.  But, with nothing more accurate to measure them against, how can anyone tell how accurate these clocks are?  Might it not just as easily be one second in 5 billion years ... or a trillion years?  In other words, are we not just plucking figures out of the air?

In fact, does "true" time actually exist - something against which clocks constructed by intelligent beings anywhere in the universe can ultimately be measured?
What do you think?
« Last Edit: 21/07/2016 15:23:01 by _system »


Offline Tim the Plumber

  • Sr. Member
  • ****
  • Posts: 240
  • Thanked: 10 times
    • View Profile
You build ten of each.

When all ten say the same time, to the millionth of a second or finer, after a year, you have a very good set of clocks.

Offline alancalverd

  • Global Moderator
  • Neilep Level Member
  • *****
  • Posts: 4505
  • Thanked: 139 times
  • life is too short to drink instant coffee
    • View Profile
We begin with the presumption that the energy of the fundamental mechanism, e.g. the ground state spin-spin interaction of electrons and the nucleus in cesium, does not change. It's a good mechanism because there's no reason why it should, or means by which it could change. So the question is one of the precision with which you can measure it.

On the one hand, the higher the fundamental frequency, the easier it is to determine accurately the end of a given pulse train. Suppose I have an oscillator running at 50 Hz and delivering 100 volt pulses, and I can detect when the output falls below 1 volt. This means I can find the beginning and end of the pulse train to within 1% of one pulse, so if I define the second as the time taken for 50 cycles, I can only measure it to +/- (1.4/50) % random error. But if the primary oscillator is running at 9000000000  Hz, my first estimate of accuracy is  +/-(1.4/9000000000)% .

On the other hand, counting very rapid pulses is more difficult than counting slow ones, so we may introduce more random and systematic errors by increasing the base frequency.

We make progress by improving the stability and reliability of the counting electronics, and by finding fundamental processes with ever-narrower bandwidths. Currently, there is some interest in reverting to rubidium (higher fundamental than cesium) thanks to recent improvements in counting electronics, but it may be overtaken by strontium engineering.
The following users thanked this post: hamdani yusuf

The Naked Scientists Forum


SMF 2.0.10 | SMF © 2015, Simple Machines
SMFAds for Free Forums