ondary frequency standard by the Working Group on Primary and Secondary Frequency Standards (WG-PSFS), instituted by BIPM’s Consultative Committee for Time and Frequency. e results of multiple calibration measure-ments of the international time scale TAI have since been published in Circular T. As shown in Fig.7, they are in good agreement with the reports of other standards.By taking advantage of the inherent maser stability, intermittent clock operation allows monthly evaluations with statistical uncertainties of less than 3×10 (limited by the intercontinental link between NICT and BIPM) to be performed while only operating the optical lattice clock during four days. Although the contribution to the steering of TAI currently remains limited by the uncertainty =4×10 assigned to 87Sr as a sec on dary represen-tation of the second, the calibration data may already be relevant whenever an insucient number of primary fre-quency standards contribute.e calibration data collected in the circular T can also be interpreted in the context of remote comparisons be-tween the contributing standards [29], with TAI itself acting as a y wheel oscillator where measurement intervals are not com pletely aligned. Such an analysis has recently been used to determine the absolute frequency of a 171Yb optical frequency standard to a fractional uncertainty of 2.1×10 [11]. Ongoing contributions by dierent sec-ondary standards will rapidly accumulate such data and may prompt a re-evaluation of . In itself, reliable contribution of optical frequency stan-dards to TAI is also an important milestone towards a new de nition of the SI second [30], which may occur as early as 2030.AppendixConversion between Hadamard variance and power spectral densitye Hadamard variance relates to the power spectral density () through [26]()=()⋅|()| ,(14) where |()|=sin is the magnitude-squared transfer function aer normalization according to the denition of the Hadamard variance. In the conventions of [27], this results in the conversion factors given in Table 2.For a measurement with sampling interval , the cuto frequency is typically considered to be ) . e use of the Hadamard deviation makes it possible to include icker-walk FM noise in the analysis, for which the Allan deviation does not converge. Here, γ≈0.577 is the Euler–Mascheroni constant.FiF6Binning 10 d of data shows no statistically significant correlation of the frequency deviation with the time of day. Solid line shows a running mean over 6 hours. Orange shaded region indicates the interval where typical one-day mea sure ments take place, to which we assign a systematic uncertainty of _⁄=7.95×10 .FiF7Deviation of TAI scale interval from the SI second The measurements of NICT-Sr1 (red circles with heavy black outline) are in good agreement with the results of other primary and secondary standards (markers, as indicated) and the value of reported in Circular T (black line, with orange shaded region indicating uncertainty). The indicated error bars for the secondary stan dards NICT-Sr1, SYRTE Sr2 and SYRTE SrB do not include .98 情報通信研究機構研究報告 Vol. 65 No. 2 （2019）4 原⼦周波数標準
元のページ ../index.html#104