Jim:
Thanks for the explanation. No need to do more research on my
account, as you have more than adequately addressed my original
question.
I *knew* if I asked it on this forum, I'd get a knowledgeable
response!
Jim Elwell
> -----Original Message-----
> From: [EMAIL PROTECTED]
> [mailto:[EMAIL PROTECTED]]On
> Behalf Of James R. Frysinger
> Sent: Friday, July 13, 2001 9:20 AM
> To: U.S. Metric Association
> Subject: [USMA:14360] Re: More precise atomic clock
>
>
> I haven't looked at the article in Science yet, but I
> can mention a
> thing or two that might help see what they are getting at.
>
> The problem being tackled is the precision in the
> measurement of the
> frequency of the signal, in this case a photon in or
> near the visible
> region. (I'll look for more details later on this.) I
> would imagine that
> the basic technique for measuring this photon's
> frequency is to use it
> to "lock in" the frequency of a counting circuit's oscillator.
>
> Here it gets a bit more technical, though I am almost certainly
> oversimplifying what is done. Once an oscillator has
> been "frequency
> locked", it's output signal can be shaped to produce a
> square wave (as
> opposed to sine wave) and this can be fed to a
> counting circuit--more
> likely, a cascade or chain of circuits that step down
> the frequency at
> precise ratios and then into a counting circuit. Much
> of the research in
> the last decade or so has dealt with devising these
> stepdown chains (aka
> "frequency dividers") because electronic circuits just
> cannot count fast
> enough to keep up with the frequency of light. By the
> way, these
> oscillators nowadays are all lasers or masers, I
> believe, and are thus
> quite stable themselves.
>
> The precision of the measurement might then be seen as
> being the period
> of oscillation (in the original photon). Say I have a
> source that
> produces 1 kHz signals. My counting precision is then
> � 1 ms since I can
> count to the nearest cycle, which takes 1 ms (= 1/f =
> 1/(1000 Hz)). If I
> move up to a 1 MHz source, my counting precision then
> becomes � 1 �s
> (=1/(1 000 000 Hz)).
>
> Another factor that comes into play is the sensitivity
> of the phase-lock
> loop or whatever that "frequency locks" the oscillator. Higher
> frequencies lead to less wobbling (in terms of the
> amount of time
> deviation) about the base frequency since the waves'
> instantaneous
> magnitudes rise and fall faster.
>
> You're right, Jim. The accuracy and stability of the
> source must be
> considered. That has been worked in parallel with the
> development of
> chain-counters. The latest improvement I've seen
> mentioned in public is
> NIST's "fountain" clock. If this uses a single atom of
> mercury, it
> sounds like the fountain clock is involved and I think
> we've mentioned
> and described that in the past. USMA's Metric Today
> had a article on
> that as I recall. Here we are taking advantage of
> quantum mechanical
> effects related to the energy of the atom. "Cooler"
> atoms produce more
> stable emission lines than "hotter" atoms. Single
> atoms produce more
> stable emission lines (albeit at the expense of signal
> intensity) than
> many atoms acting as a canonical ensemble (technical
> term used in
> statistical mechanics for "a whole herd of 'em"). The
> latter bump into
> each other and the atoms are not all at the same
> energy level, so the
> emission lines are broadened, producing a spectrum of
> many frequencies
> centered about the main one.
>
> That's probably a whole lot more than you wanted to
> know. But I'll check
> up on this with Science Online just in case you were
> expecting more....
> ;-)
>
> Jim
