The old method to calculate the guide corrections used the intensities in the 4 quadrants of the guide star box. This only gives a rough measure of any displacement, not a specific distance of the star to the center of the guide box. A new method was implemented that calculates the position of the star in fractions of a pixel and a gain is applied to this distance to calculate the correction. This gives considerably faster centering of the guide star and follows the precise movements of the guide star slightly better (but at a level that does not affect the image quality). An advantage of this method is that the width (FWHM) of the star image is also computed.
There were some problems with the TCS apparently loosing contact with the guide star server causing a time-out. No specific errors were found in the logs, but it is also not clear how much time the guide star server takes. At least in one case the telescope was pointing in the galactic plane fairly close to the galactic center where there are many targets but also many targets that need to be excluded because they are too bright which might cause the program simply to take a long time to react. The response by the guide star server will be monitored so we can see what it is doing when a time-out occurs in the TCS when querying the guide star server.
The TCS occasionally looses count of the number of turns the guide-probe motors make and to minimise this problem the speed of the probe is limited. To not overload the TCS with this task a hardware solution has been designed to do the turn counting. It has been made such that it can operate in parallel with the existing hardware so the TCS still has access to the encoder counts. A board has been designed and preliminary software written, but manufacturing, testing and installation still has to be done.
Thomas Augusteijn 2013-05-10