In 2006 and early 2007, major modifications and upgrades have been done at the Plateau de Bure Interferometer. A number of them had a possible impact on the performance of the PdBI as a phased array VLBI station:
The new system had proven itself in local interferometric observations, and high expectations had been met and surpassed. However, VLBI is more demanding in terms of signal stability than local observations, and it is quite difficult to test a system locally to the required precision. There was a possibility that a flaw or even a fundamental characteristic of the new system would make VLBI observations impossible.
The refinement of the data reduction and calibration for local interferometer observations with the new system delayed the work on the VLBI part. Finally local VLBI software tests were successful, but while a first VLBI test between Bure and Pico Veleta produced powerful autocorrelations, cross-correlation fringes remained elusive. A second Bure - Pico Veleta test on the brightest quasars and masers in the sky produced very weak fringes, with a S/N of just 30 on 3C273 over 30 seconds. Although observing conditions had been excellent, the data suffered from a strong phase noise.
With just three weeks left to the Global session (10-15 May), major hardware modifications were excluded. The fact that the system worked well in local interferometry mode allowed to eliminate a number of causes for the instability. A thorough verification with a spectral analyzer by J.-Y. Mayvial from the IRAM backend group allowed to spot a partly broken cable, and to improve the quality of the 5 MHz reference distribution, which is the ``heartbeat'' of the interferometer from which all local oscillator frequencies are derived. But it remained doubtful if this had been indeed the problem which perturbed the VLBI.
In order to allow a test on several baselines connecting to the PdBI, it was agreed to start the Global session on a strong astronomical source and transfer the data by network to the MPIfR correlator in Bonn. Unfortunately, the Bonn correlator suffered a technical problem shortly before the Global session, and could not be repaired in time to perform the fringe test. On Friday evening after 19h of Bure VLBI participation, the IRAM direction decided that the risk of a week-end of unusable observations was too high to be taken with the present proposal pressure. PdBI therefore returned to local observing mode, promising to join the Global session again as soon as fringes were found. Pico Veleta continued with the VLBI observations. Meanwhile the colleagues of MPIfR Bonn worked on the fringe search with the Swinburne software correlator in Australia, which runs on a parallel supercomputer (300 CPU Beowulf cluster).
On Monday, fringes were found: 30 seconds on 3C273 yielded a S/N of 462 on the Bure-Pico Veleta baseline, with an excellent phase stability (Fig. 1). Bure stepped back into the Global session in time for the observation of the Galactic center, and continued until the end of the schedule.
One week later, the PdBI phased array joined with Pico Veleta in the second VLBI observation of the Galactic center.
A first analysis of the VLBI performance of the Plateau de Bure shows that the phased array is now better than ever before. The new hardware provides excellent stability (including the new IF processor developed by IRAM and the new EFOS-38 maser by T4Science), and the new dual polarisation receivers allow now simultaneous LCP/RCP observations with an improved system temperature on all antennas (before, Bure could only observe in LCP).
Many thanks to all who have contributed to this success.