Abstract. Two liquid-helium-cooled cryogenic sapphire-resonator oscillators (CSOs), have been modified to operate using cryo-refrigerators and low-vibration cryostats. The Allan deviation of the first CSO was evaluated to be better than 2×10-15 for averaging times of 1 s to 30 000 s, which is better than that of the original liquid helium cooled CSO. The Allan deviation of the second CSO is better than 4×10 -15 from 1 s to 6 000 s averaging time.
IntroductionCryogenic sapphire oscillators (CSOs) are still the most frequency-stable oscillator operating in the microwave frequency region [1]. Their frequency stability, expressed as Allan standard deviation, reaches a few times 10 -16 over averaging times of 1 s to 1 000 s. This short-term stability cannot be obtained in any commercial device, for example, in BVA OCXOs or hydrogen masers.Recently, the cesium atomic-fountain primary frequency standards, when limited by a quantum projection noise, have a short-term frequency stability approaching a few times 10 -14 . In order to avoid the influence of the Dick effect (the degradation of frequency stability due to dead time), a more stable local oscillator than obtainable in commercially available devices is required. Thus CSOs have been successfully implemented as local oscillators for such frequency standards [2], [3].On the other hand, the development of optical atomic clocks is rapid and their frequency stability and uncertainty approaches 10 -18 . In near future, the redefinition of the unit of time, the "second," is expected [4]. One purpose of optical clocks is the improvement of the international timescale, TAI (International Atomic Time). The uncertainty of TAI has been determined by a balance of the frequency stabilities between primary frequency standards (currently Cs atomic fountains), time comparison methods (GPS or two-way satellite time transfer), and local oscillators (LOs, currently, hydrogen masers), to the current value of about 3×10 -16 . Because primary frequency standards (current Cs atomic fountains, and optical clocks in future) do not operate continuously over years, better LOs to replace the current hydrogen masers will be necessary. The LO should be stable enough to keep good time and phase, even when a primary standard stops operating and hence it cannot calibrate the LO during those periods. If stable LOs are successfully developed that can operate continuously without phase jumps over many years, they can be calibrated with the best optical clocks and can be