PTB´s primary clock CS3: uncertainty evaluation and operational experiences

Summary form only given. The primary clock CS3 is now used for routine time-keeping and for the realization of the SI second in PTB´s time unit laboratory. In this contribution we present the final result of a Type B evaluation and some details about the methods involved. For the case of the following systematic effects the initial design goals have been achieved: second-order Doppler effect, end-to-end cavity phase difference, cavity pulling, Rabi-pulling, servo electronics, gravity, blackbody radiation. Their evaluation practically does not rely on frequency measurements. In other cases, systematic studies have been hampered by the short-term frequency instability which is 9·10 ^-12/√(τ/s), because the selection of slow atoms resulted in a smaller fraction of atoms contributing to the beam signal than naively expected. A combined standard uncertainty of 14·10 ^-15 was obtained. The largest components in the list of uncertainty contributions are connected with the quadratic Zeeman effect and the end-to-end phased difference. In the long term, a mean frequency difference between CS3 and CS2 of 15·10^-15 was found during a one year period