Astron. Astrophys. 343, L1-L4 (1999)

4. The VLT test camera

The VLT Test Camera was not meant to be a scientific instrument, it was specifically designed for the commissioning of the telescope. Therefore, its characteristics have been determined almost entirely by technical rather than scientific requirements. The Test Camera has two modules, an imaging module and an analysis module. The latter provides a pupil imager and a wavefront sensor using a Shack-Hartmann analyzer. During SV the imaging module was used. This is a Offner optical system which re-images the telescope image plane onto a Tektronix 2k2k CCD at unit magnification. A wheel located ahead of the re-imaging optics provides 7 positions for different filters.

Re-imaging is accomplished by four reflections off three mirrors which are coated with high-reflectivity dielectric coatings yielding a total throughput of larger than 70% above 350nm and more than 90% above 550nm out to 1µm. The pixel size of the CCD being 24µ, the actual pixel scale was then /pixel. All SV frames, however, were obtained in a binned (22) mode, hence with a pixel size of .

A set of regular broad-band UBVRI filters in the system of Bessell (1990) was available together with two special narrow-band filters; a filter isolating the Ly emission line of QSO J2233-606 (), the QSO at the center of the HDF-S STIS field, and a narrow-band filter tuned to the Ly emission line of the Einstein Ring ER0047-2808 at .

The Test Camera CCD is not a science-grade device and is affected by major, extended blemishes. A large region near the center of the chip shows a markedly different sensitivity with a very strong color dependence. Regular bias frames were obtained to check for any drift in bias level. No significant drift was detected. Dark frames were also acquired on four different nights equally spaced throughout the SV observing period. No corrections for dark current were applied as the CCD was measured to have about 7 /pixel/hour, which is negligible compared to the average sky levels reached. The gain and the readout noise of the CCD were measured in the laboratory and confirmed at the telescope to be 2.5 /ADU and 7.9 RMS, respectively. The shutter accuracy was determined by a series of short exposures with the dome lights on, which where then compared to long exposures. The shutter map shows the typical structure of an iris shutter. The variation between the center and the edge of the field for a 1 second exposure is about 4%. The uniformity for a 5 second exposure is better than 1% across the whole field with a small area close to the center enhanced by 0.8%.

Flat-fielding of the images proved to be very difficult. Twilight flats were obtained whenever the conditions allowed us, but they did not produce good results on the science data. The best results were achieved by combining a number of science exposures in a median stack, thereby removing signal from any astronomical sources. Two sets of such flatfields for each broad-band filter were produced and applied to the science frames. The large CCD blemishes could be corrected to better than about 1%, but dust features, which moved between nights could not be eliminated from some of the frames. For the narrow-band observations, the twilight flats were used. The accuracy here is better than 2% for the 392/7 filter. The 557/6 filter was vignetting the field considerably and the combined image shows a strong radial gradient.

A bad pixel map was generated from a series of low-level dome flatfields which was compared to a series of high-level flats. There are only very few pixels which deviate by more than 5 from the mean. These are located in small sections (10-15 pixels) of five columns.

© European Southern Observatory (ESO) 1999

Online publication: March 1, 1999
helpdesk.link@springer.de