### OBSDATE ### 20120107 ### LOGDATE ### 120108 ### OBSERVERS ### MJI ### WEATHER ### Clear, moderately high humidity (70%), 26 Celsius. ### LOCATION ### SUSI ### PROGRAM ### Get SUSI working again! ### SIDEROSTATS ### n1 s1 ### TARGETS ### Canopus ### QUALITY ### Quite bad (no/almost no useful data ### PROBLEMS ### Fringes! It has been a while... It was after midnight before fringes were attempted, because neither argus nor gareth could use the DAC card. Data set 1: No alignment whatsoever. Data set 2: attempted to align S. The best S focus was at about 70000 and small tweaks were needed. *** Clearly the PAVO fringe algorithm is way wrong, because the fringes were very bright by eye, but had low S/N in the display and had a dispersion solution that was very wrong (-5m of air for straight fringes) *** N also was very obviously out of focus, filling the whole green box with an occasional dark central spot. Screwing in 1mm made the dark central spot go away somewhat, wich an average flux of 2.8e4 (t/t) and 2.5e4 (fringes). Screwing in another 0.5mm clearly made things worse... 2.6e4 (tip/tilt), 2.0e4( fringes). Back out 1mm (in 1mm from the start) wasmaybe better... 3.2e4 (tip/tilt) and 2.6e4 (fringes). Data set 3: with this (slightly) adjusted N focus. NB 1) Typical OPD errors were less than 0.1 microns all this time] 2) The previous night (with no tip/tilt), S1 would not track. Clearly an AV68K issue. 3) Fringes were at -900 microns, and the sid offsets were 414,18 for N and 143,-48 for S. Time spent on a pointing solution would clearly be time well spent. Next, tried to change the only thing I know how to... the pavo_default_params file. Distortion was zet to zero (even though I'm almost certain that this isn't the problem) and I initially tried to set the "mask hole size" slightly smaller to 1.8. This gave a horrible bias when there were no fringes. Next I tried to set it a little larger to 2.6. TheDISP number was still setting on -5.0, but it was possible that S/N was higher (often above 30). The next step is definitely playing back real data and testing systematically... something for the TODO. Very strange that his dispersion issue is happening now, when there has been no substantial pavo code changes in 18 months.