The new gain (using this source flux) is now 11 K/Jy flat out to 15 degrees and then starts dropping because of the za spill over (actually it should start dropping a bit higher in za). The absolute gain value is as certain as the flux and cals. The relative gain increase is about 10 % and the gain is no longer a function of za below za=15. Prior to the move the gain had a linear za dependence all the way down to za=0. This was also seen in the gain curve measurement done for lbw back in sep01 (-.1 K/Jy per deg za see figure 3). The linear dependence was probably being caused by the focus error in the rails (green lines). The 0 to 2" rail error vs za should not affect lbw too much. Adding 3.67" would make the 0 to 2" ramp an appreciable error (this sounds good, but our current model of pitch, roll, and focus does not place the rail ramp at the azimuth of this source !).
Fig 1 top is the gain [K/Jy] before (black) and after (red) the move. This includes rise and set.
Fig 1 bottom is the fractional gain change (new/old)-1.
Fig 2 shows the gain,Tsys,Sefd, and average beam widths before (black) and after (red). Fig 3 has the coma, first sidelobe height, Main beam efficiency, and main beam + 1st sidelobe efficiency. Fig 4 plots the pointing error before (black) and after (red).
The increase in gain must come from somewhere. Looking at fig 2, the average beam width actually increased a bit after the focus move. On figure 3, the coma parameter has decreased and the first sidelobes after the move have decreased.
The system temperature also looks like it has increased by a bit (although we probably need more measurements to verify this since the weather conditions were not identical on the two days).
The pointing error for azimuth and setting za have may have changed by a few arcseconds. The total pointing error is large (probably do to the changes that have been made on the structure.
This is the 3rd kildal feed to have been moved. You can find the results of the lband narrow move and the cband move.processing: x101/021120/doit.pro