VLA Test/Observation Coordination Meeting B.G. Clark May 15, 2003 1. eVLA A transition issue. Current VLA LO system is based on the final LO being supplied by commercial Fluke synthesizers. These synthesizers work internally on 10 MHz, and have dividers which divide that down to lower frequencies for the lesser digits. Since these dividers are never reset, when the synthesizer is commanded to a different frequency and then back to the original, it comes back with an arbitrary phase offset with respect to its original sine wave, unless the frequency is a multiple of 10 MHz, in which case the dividers are, effectively, not used. This does not matter for the VLA, since the same synthesizer is used for all antennas. However, eVLA antennas do not use the Fluke synthesizers. So spectral line observations using the Flukes will suffer an arbitrary phase jump between VLA and eVLA antennas whenever the frequency is changed. This can be handled by calibration protocol, making sure there is a phase calibrator observation before and after every frequency change, with a resultant loss of observing efficiency; or by observing with a fixed frequency and correcting post facto for changing doppler. E. Greisen points out that the latter is unlikely to be satisfactory, because it only works well if all lines in the spectrum are well resolved, and VLA correlator users are usually too starved for channels to arrange that. 2. Water vapor radiometers C. Chandler reports on the current state. Antennas 26 and 28 have had WVRs since January, and intensive testing has been on for the last month or so. The WVRs work on the principle that the signal in the water line is proportional, with a time dependent coefficient, to the extra phase path introduced by the water vapor in the troposphere. The WVRs have three frequency channels, one near the line center, two near the half power points, and the signal they work on is the difference between the center channel and the average of the two side channels. Variations in this amount to of order 30 mK. Therefore, measurement accuracies of much better than 10 mK are needed, about 0.0001 of system noise. Sources of noise are receiver thermal noise, digitization noise, variation of the Tcal values, and variation of receiver gain. In order to control the latter two sources of noise, the noise source and IF amplifiers are mounted on a special plate temperature stabilized to a few mC, which improves the stabilities of these items by a factor of order 100 over the equivalents in other antennas. In the lab, stability is that needed, about 0.00005, with a 5 minute integration (limited by thermal noise for shorter times, by 1/f noise for longer). On the telescope, in general, the interferometer phase and the WVR track well (with a variable scaling factor) on timescales of a few minutes, which is the appropriate timescale for correcting visibility data. On longer timescales they diverge, but this should be taken out by ordinary calibration methods. There is, however, a considerable variety of behaviors. Perhaps most interesting are times in which the WVR exhibits small variations but there is still a significant phase RMS (CC suggests these may be due to dry air or liquid water terms, C. Walker suggests that it could still be ionosphere). With the caveat that this limited experience has been with the fairly short baselines of the D configuration, CC ventured that applying the WVR correction would often result in a factor of two reduction of phase RMS. She feels that to really understand how well the WVRs are working, though, requires four, rather than two. 3. Bandpass stitching. C. Rodriguez-Rico reported on combining data taken on M 82 in the 7mm recombination line with three overlapping bands; a single LO setting was insufficient to cover the wide velocity range. He found significant problems dealing with the continuum. He found it necessary to independently determine a continuum image from line-free channels at every LO setting and to subtract that continuum before combining the line cubes. The three continuum images also proved very useful in maintaining the same calibration for the data from the three LO settings. C. Carilli remarked that he had good success with making separate, continuum mode observations off the line to determine the continuum image, which he then subtracted in the u,v plane (with much less complex images than that of CRR, though).