[time-nuts] Primary Standards...
richard at karlquist.com
Wed Feb 24 00:46:34 UTC 2010
Mike S wrote:.
> There is no difference between Cs and Rb in that regard, except perhaps
> The frequency of a Cs is subject to gravitational and electromagnetic
> effects. Which is why the definition of the second was clarified in
> 1997 by the CIPM to refer "to a caesium atom at rest at a temperature
> of 0 K."
> Since absolute rest and absolute zero are impossible conditions for a
> real world clock, Cs clocks do not have a "known and invariant
> frequency." If they did, why would they have a C-Field knob to twiddle,
> and why would TAI be a weighted average of multiple Cs clocks? I'd
> guess that Cs clocks can also be thrown off by trace gasses in the
> tubes and numerous other effects, impossible to completely remove.
You covered many topics here. First, let me put the C-field knob
issue to rest. This was used to vary the frequency of older/cheaper
atomic standards, whether primary or secondary, before frequency
synthesizer technology allowed the C-field to be fixed at one value.
In a secondary standard such as Rb, the synthesizer is used to "set"
it on frequency against a reference. In a primary standard such
as a Cs beam standard, the synthesizer (if it is used at all) is used to
get a known offset from 10 MHz for various reasons unrelated to "setting"
it to a reference.
The TAI is a weighted average to improve short term stability and
to average out random frequency errors. The TAI is not a crutch
to make up for inferior clocks; rather, clocks have to earn their
way into TAI by being accurate. I haven't looked recently,
but traditionally something like 85% of the weight has been 5071A's.
Also, the "big" clocks that are better than the 5071A's are used more
as frequency standards and don't necessarily run continuously and
actually keep time as a clock. Most 5071A's run continuously,
because no one needs to baby sit them to keep them going.
Because the 5071's have non reversible beams, they are subject to a small
frequency error due to end to end cavity phase differences. Because
of various proprietary techniques used to machine the cavity, these
errors are very close to truly random and average to nearly zero
over a large group of 5071A's.
Unlike the buffer gas in an Rb standard, the inside of a cesium
beam tube is at a very high vacuum and is continuously pumped down.
Pressure shift due to trace gases is not a factor. Velocity effects
and doppler shift, etc in Cs standards have all been addressed
and a small upper bound has been established.
Rick Karlquist N6RK
More information about the time-nuts