The October 9-15 Global Millimeter VLBI session did not work well for both IRAM instruments. The 30-m lost 66% of the observing time due to bad weather conditions (winds beyond the operation limits and snow), while Plateau de Bure had good weather conditions, but encountered a phase stability problem in the VLBI system. This problem was diagnosed in a fringe test at the beginning of the session; as it became clear that it could not be resolved within the duration of the session, Bure went back to local observing mode (100% VLBI session loss).
Since then, the origin of the problem has been tracked back to the Bure Racal Dana frequency generator; this unit had been specially adapted for VLBI to allow an excellent phase stability, and was only connected during VLBI sessions. Unfortunately, absolute phase stability is not a design feature that is well established on the frequency generator market: The Racal Dana unit that failed was already more than 20 years old, but still better than many high-level modern units. As second-hand Racal Danas of the model in question become increasingly rare and fragile, the IRAM backend group started a test series to search for a suitable modern replacement while the MPIfR Bonn kindly agreed to loan us a frequency generator to restore the Bure VLBI capabilities independently of the result of this quest.
Fortunately, a likely candidate was found in the Rohde & Schwarz SMA100 B22 that passed the rigorous phase stability tests of the IRAM backend group, although its sensitivity to ambient temperature variations is a bit higher than we would have prefered. IRAM has bought a unit, and we plan to install it in a temperature controlled rack on Plateau de Bure. A small VLBI test with this new generator is foreseen between Plateau de Bure and Pico Veleta before the installation of EMIR begins.
In the IRAM Newsletter No. 71, we announced a tuning of the Plateau de Bure EFOS-38 maser that anticipated the very slight drift acceleration of the instrument. The monitoring of the maser against GPS since then (Fig. 7) indicates that this ballistic tuning was successful. The residual drift during the October VLBI session was below 0.4 nanosec/day, i.e. the relative stability over that week was better than . Our second-order estimate based on the monitoring since the last maser retuning is about 39 picosec/day, which is comparable to the previous value of 42 picosec/day.