[time-nuts] Primary Time Standards

[Magnus Danielson]

On 15/07/11 10:24, Tom Van Baak (lab/iPad) wrote:
> A primary frequency standard is one that faithfully implements the definition of the SI second. Thus primary standards are based on Cs. But not all Cs-based clocks are primary. CSAC, for example, is not a primary standard. Rubidium, hydrogen, quartz, or pendulum clocks are not primary.

There is an over-focus om the atom being used rather than the physical 
apparatus being used. Certain physical apparatuses (physical packages) 
has severe biases frequency so regardless of which atom used, you will 
have repeatability issues (build two devices, and they will tick 
differently).

However, for various reasons has different atoms (and isotopes) been 
chosen for various physical packages due to various reasons. For 
instance, rubidium has two isotopes which makes it feasable to create an 
optical pumping due to how the D1 and D2 lines of the isotopes is located.

Rubidium was investigated in beam configurations just as cesium.

In fact, Cesium wasn't even the most stable one when chosen. It was 
chosen for its lower frequency was deemed more practical to implement.
Still, the early beam devices had severe systematic biases in their 
frequency and much work has gone into prediciting them, reducing them 
and compensating them.

> The definition spells out zero magnetic field, zero temperature (zero velocity), and zero altitude on the earth's rotating geoid. There are many other practical physics details that need to be addressed. For a good example of what it takes to make a Cs clock a primary standard see:
>
> http://tf.nist.gov/general/pdf/1497.pdf
> http://tf.nist.gov/general/pdf/1846.pdf
>
> So strictly speaking no Cs clock actually runs at exactly 9192.631770 MHz since you need a certain amount of magnetic field to isolate the hyperfine transition, you can't run at absolute zero, no labs are actually at sea level, and atoms are not simple toys, etc.
>
> A lot of work is required to identify, predict, and quantify a host of factors. Again, please read or glance at those papers to appreciate the work that national metrology labs do to "make copies of the SI second" for their country.

The 9192631770 cycles per second thing is the easy poster statement, the 
unattainable goal so to speak. The real world is quite complex.

Today I would say that "digital cesium", i.e. the locking of C-field 
from the side-band Rabi responses is among the key technologies which 
needs to be in a cesium to compete for systematic error removals.

Cheers,
Magnus