Abstract :
Since the end of 1992, the quality of the timing data received at the BIPM has rapidly evolved dues to the extensive replacement of older designs of commercial Cs clocks. Consequently, the stability of the reference time scales has improved significantly: it is characterized by an Allan deviation of 2.6×10-15 for averaging times τ=40 d. However, further improvement in stability is still possible without threatening reliability. This was tested by running modified algorithms over the real clock data collected at the BIPM. Results of different studies are shown here; in particular the implementation of an upper relative contribution, chosen equal to 1.37% for any contributing clock, leads to σy(τ=40 d)=1.8×10-15 . The accuracy of TAI is estimated by the difference between the duration of the TAI scale interval and the SI second as produced on the rotating geoid by primary frequency standards. In this paper, TAI accuracy is evaluated from six primary frequency standards LPTF-FO1, PTB CS1, PTB CS2, PTB CS3, NIST-7 and SU MCsR 102 all corrected in a consistent manner for the gravitational shift and the black-body radiation shift. This led to a mean departure of the TAI scale interval of 1.8×10-14 s over 1995, known with a relative uncertainty of 0.5×10-14 (1 σ)
Keywords :
atomic clocks; 1.8E-14 s; Allan deviation; International Atomic Time; SI second; TAI reference time scale; algorithm reliability; atomic clocks; black-body radiation shift; deferred-time; gravitational shift; long-term stability; optimised algorithms; primary frequency standards; relative uncertainty; rotating geoid; time dissemination; time scale accuracy; upper relative contribution;
