Improvement to the National Physical Laboratory Atomic Clock

Abstract

THE second of time is now defined in terms of the hyperfine line of caesium-133, and is that interval which makes the frequency of the spectral line 9192631770 Hz (ref. 1). The second model of the National Physical Laboratory (NPL) standard described briefly in 1959 (ref. 2) has operated since then at a value estimated to be within one part in 1011 of the spectral line frequency, or approximately 1 µs per day. It can be used with a precision ten times better than this, and the main limitation to the use of the full precision is the possibility of a frequency shift due to a small phase difference between the two ends of the cavity resonant structure. The symmetry of the resonance shows that an error due to this cause does not exceed one part in 1011. Another way of checking the phase error is to reverse the cavity physically, thus reversing the effect. This method was used with the first NPL standard3 and also at the National Bureau of Standards (NBS)4, the National Research Council of Canada5 and the Laboratoire Suisse de Recherches Horlogeres6. Its drawbacks are that it involves the physical handling, movement and refixing of the cavity structure, and it occupies a time of possibly several days, during which the comparison standard must remain constant. The NBS overcame the handling problem by reversing the positions of the oven and detector, but although they then obtained more consistent results they were left with a residual uncertainty of ±3 parts in 1012. The residual uncertainty at the Laboratoire Suisse de Recherches Horlogeres was ±5 parts in 1012, although at the National Research Council, Ottawa, a frequency difference of 5 parts in 1012 due to this effect was reproduced in two measurements to 4 parts in 1013, a hydrogen maser being used as the comparison standard.