Figure 4a. The defocus in arcseconds of blur on the image plane is shown as a function of time for the night of August 14. Arrows mark the times at which the telescope focus was manually changed. The labels show the magnitude of the changes made. The squares show measurements of the defocus just prior to and after manual focus changes. The first focus change actually decreased the telescope image quality while the second made a significant improvement. Below I will show clear indications of errors in the secondary motions which are probably contributing significantly to focus errrors in the telescope.

These measured image blurs can be converted into secondary focus motions required to place the telescope into optimal focus. The following graph shows these data with this conversion applied. In addition, in the following graph I have also taken the manual focus changes into account by subtracting the focus position as determined from the TCC log from each measurement. All of the data are therefore shown as if the telescope had been kept at a focus position of 0 during the entire period.

Figure 4b. In this graph the defocus blurs shown in Figure 4a have been converted to secondary focus corrections required to optimize the telescope focus. As discussed above, all of the defocus corrections have been adjusted to correspond to a fixed telescope focus of 0. As in Figure 4a, the circles show the measurements taken during the period in which no focus adjustments were made. The squares correspond to the measurements shown as squares in Figure 4a.

The defocus shift between the circles and squares is an indication of errors in the focus motions of the secondary. In this example, these errors appear to be on the order of 20 µm, which corresponds to 20,000 µ-steps in the secondary encoder readings. It has been known for some time now that we problems this large in the secondary actuators. Because of these errors, we do not include the squares when fitting the defocus data.

The inset shows the results of a least squares fit to the circles. The form of the fit is chosen to match the corresponding TCC secondary piston coefficients. On this night, only small changes of focus (10 µm) are seen as a function of telescope altitude. A linear fit to these same data gives a slope of 0.175 µm/degree.