Detector controller

The ALFOSC CCD was moved to the modified dewar which has the correct mountings and connector for the new controller. The new controller and the modified dewar for the ALFOSC CCD were mounted on the telescope on the 23rd of June. The first nighttime test were made on the 25th of June. During the installation period, the controller was adjusted for optimal noise performance, and the new system was characterised in order to compare its performance to the one of the camera system using the old CCD controller.

A report was made that describes the properties of CCD with the new controller; see:

http://www.not.iac.es/instruments/development/ALFOSC_Controller_Report.pdf

The report details the performance of the controller both in relation to its properties in such measures as, e.g., read-out noise and linearity with the ALFOSC CCD, and in its operations through software. The main advantage of the new controller compared to the old controller is the increased readout speed (at the default readout speed of 200 kpix/sec 24.5 sec for full frame readout). In general, the performance of the CCD comply with the specifications, but there are various issues that need to be resolved before the new system can be fully accepted. The report is intended as basis for a comparison with the user requirements to define what is needed for final acceptance of the detector controller and associated software for use with the ALFOSC CCD. The details of what should be done and when it should be done is currently being discussed with people in Copenhagen.

Table 4: Performance CCD with new controller.

Speed	Read out	Read out noise
kpix/sec	time (sec)	ADU	e-
113	41.5	10.5	3.4
200	24.5	12.2	3.9
400	13.5	17.2	5.5
800	9.5	30.6	9.9

A serious issue has been the fairly frequent problems with the horizontal charge-transfer-efficiency which requires a hard reset of the controller to cure. The main problem is that the bias image is significantly affected directly following a reset, showing a strong offset and slope that only disappears slowly with time. It appears that these problems are generated by the use of large ($>$4) binning factors in the serial (X-) direction, and avoiding the use of such binning basically has removed the need to reset the controller regularly.

The bias level has been found to show some sudden change in average level, but at all times the over-scan provides a proper estimate of the average level. An advantage with the new controller is that we now can provide over-scan in both directions on the CCD. With the old controller we could only provide over-scan in what is normally the spatial direction (along the slit) for spectroscopy. To reduce readout times often people would window the CCD in the spatial direction, excluding the over-scan. With the over-scan provided in the dispersion direction this is no longer an issue.