[time-nuts] Primary Standards...

[Magnus Danielson]

Poul-Henning Kamp wrote:
> In message <20100223214204.EAE711174BE at hamburg.alientech.net>, Mike S writes:
>> renamed, since the discussion has shifted.
> 
>> "In the time and frequency field, the term primary standard is 
>> sometimes used to refer to any cesium oscillator, [...]
> 
> That rhymes with and Karls and my perception of the term:
> 
> A Cs clock is primary because when you turn it on, it latches onto
> the physical phenomenon of a known and invariant frequency subject
> to no systematic errors.

No, that's not quite accurate. The key is that the systematic errors is 
predictable and that you can compensate for them such that residue 
systematic errors becomes very small. The C-field frequency shift is one 
such systematic factor.

The phase-error of the Ramsey interregation cavities is another, which 
is first-degree compensated for in some laboratory standards, but which 
can be reduced by carefull design and compensated for if it can be 
controlled to be repeateable and stable.

Even for Caesiums there exists numerous shifts. Caesium fountains is one 
approach to address some of these issues, alongside that of thermal noise.

> The reason the small Rb's do not qualify as primary is that each
> unit has a slightly different frequency, due to vapour pressure,
> isotopemix and other physical details, and thus you cannot know the
> frequency of a particular unit, until you have measured it relative
> a primary clock.

There are many factors for gas standards which makes them have 
unpredictable systematic errors. They also show long term drift factors.
The motivation for them is much lower price, volume and power 
consumption. Rubidium has proven a good choice for a gas standard.

> In other words, Primary and Secondary has nothing to do with which
> atoms, but depends a lot on the interogations mechanism used.

Agreed.

The reason Caesium is chosen is that when they where choosing, they 
higher frequency of Thallium was considered problematic as it would 
become harder to achieve the wanted repeatability from a technological 
point of view at that time. Thallium showed however a lower sensitivity 
to magnetic field than Caesium, so technically it is a better standard 
and it was known at the time. Thus, the aspect of gentlemens agreement.

By todays knowledge, Rubidium fointain outperforms Caesium, so with that 
technology scope Rubidium would be chosen and Caesium beams would be 
handy secondary standards....

> So the tiny 1cm^3 Cs standards are secondary, because they are also
> subject to all sorts of pulls and offsets.

Agreed.

> The "experimental" clocks based on lonely ions and quantum embraces
> are very likely primary, once somebody has measured their intrinsic
> frequency relative to Cs once.

Various clocks have been made for this purpose. The field is being 
investigated. The Aluminium-ion clock that was reported on recently is 
one among many different projects.

> The way to find out if your new invention has a chance to become a
> primary clock, is to build N of them, turn them on, and see if they
> all find the same frequency once they are locked, if they do,
> you're on your way to become famous.
> 

That helps. However, unless someone is able to independently build 
clocks and get equivalent levels of accuracy and stability, you still 
have a problem to show the actual performance.

It is challenging not to remake the same systematic design mistake. For 
hydrogen clocks use of different sized glas bulbs to investigate the 
wall-shifts and be able to cancel the effect, which has been important 
when measuring the unperturbed hydrogen frequency.

It is not an easy science, but it is a science which excels in 
increasingly improved methods.

Cheers,
Magnus