Rabi pedestal shifts as a diagnostic tool in primary frequency standards

Some of the systematic errors in primary, atomic-beam frequency standards are much larger for the Rabi pedestal part of the lineshape than for the Ramsey fringe part. We have used a digital servo system to measure the frequency offset between the Rabi pedestal and the Ramsey fringe for all seven Zeeman components of the cesium hyperfine transition. The dependence of these shifts on magnetic field, modulation amplitude, and microwave power enables us to separate three distinct causes: Rabi pulling, cavity pulling, and magnetic field inhomogeneity. From the measured pedestal shifts, we infer the corresponding shifts for the clock transition with uncertainties of the order of one part in 10 ¹⁵ or less