Observing overheads

The NOTCam observations have significant overheads. Both the telescope dither overhead and the data acquisition related overheads are well understood but the overheads in readout, on the other hand, are not well understood. It should be noted that the current readout overhead can add up to 40% overhead to NOTCam observations, and the amount is variable with time.

The cause of the readout overheads is believed to be due to an aging ISA PC-board. However, testing a ``spare'' ISA PC board gave no improvement. Installing the NOTCam SEQUENCER in an ``old type'' computer did give quite dramatic difference with the readout overheads being as small as at the very start. A drawback is the longer time needed to store files.

During the remote summer school in June 2010, it was discovered that ``talker'' program that provides the observing system log caused a lot of the overheads and a new talker was written. In practice, killing the talker when observing also improves efficiency in local mode, and it was realized that all other usage of the data acquisition computer (such as the electronic obslog) can influence the data acquisition rate. It was decided to move the BIAS program to its own machine, since it often needs a high percentage of the CPU. This computer has to be of a given old type known to have less problems reading the ISA PC board. We are in the process of obtaining such a computer.

In principle, the new controller should solve our current problems with long readout time and large overheads. In the mean time a special exposure command for NOTCam was suggested in which the dither overhead can be folded into the readout and file storage overhead. Depending on the type of NOTCam observations being made, this can save a very significant amount of time per night.