Astron. Astrophys. 359, 337-346 (2000)

2. The CANGAROO 3.8m telescope

The CANGAROO 3.8 m imaging telescope is located near Woomera, Australia (E, S, 160m a.s.l.). The 3.8 metre diameter mirror, of focal length 3.8 metres, is used to image the erenkov emission from gamma-ray and cosmic-ray induced extensive air showers (EAS) onto a high resolution multi-phototube camera. The camera consisted of 224 photomultiplier tubes in 1994. In April 1995 an extra 32 tubes were added to the corners of the camera bringing the total to 256 tubes in a 1616 square grid arrangement. However the extra corner tubes have not been used in the analysis of the 1995 W28 dataset in order to retain consistent imaging properties with the 1994 dataset. Each camera tube is a Hamamatsu R2248 with a photocathode size 0.12^o 0.12^o on a side, and the total field of view of the camera is 2.9^o. An event trigger is registered when the summed output of triggered tubes exceeds a preset threshold, denoted HSUM. Images with a minimum of between 35 tubes, depending on the image's compactness, trigger the telescope. A vertical event rate due to cosmic rays of 2 Hz is achieved. Monte Carlo simulations of the telescope indicate an gamma-ray energy threshold of 1.5 TeV at the vertical (Roberts et al. 1998), where the energy threshold is defined as that representing the half-maximum of the differential distribution of triggered energies. Tracking calibration is performed by monitoring the paths of bright (visual mag. 3-6) stars in the field of view, providing an absolute tracking accuracy of (Yoshikoshi 1996). A more detailed technical description of the telescope appears in Hara et al. (1993).

Data are recorded on clear moonless nights. An ON source run is generally followed by an OFF source run displaced in right ascension to provide a background run of matching zenith and azimuth angle distributions. However, since small sections of data from both observation seasons were removed due to cloud effects, a normalisation, described later, was used in estimating the statistical significance of any ON source excess. Pulse charge (ADC) and timing (TDC) information for each tube is recorded for each event. Calibration of the ADCs and TDCs is achieved by recording events triggered with a blue LED flasher before each observation. Tube signals are accepted as part of an image if they meet a number of criteria:

1. The TDC value of a tube must lie between ns, referenced against the event trigger time. The event trigger time is registered when the HSUM threshold is met.

2. The ADC value must be greater than one standard deviation above the RMS noise (comprising skynoise and electronic noise) for that tube.

3. The tube must not be isolated. An isolated tube is one which is not adjacent to any other accepted tube.

4. The tube must not have an outlying relative gain value. Tubes with relative gains outside the range 1.00.3 contribute significantly to trigger differences across the camera face. This factor is particularly important when comparing regions over the entire camera face.

The telescope is an altitude-azimuth type, which introduces a rotation of the camera relative to the sky about the tracking position during data collection. A `de-rotation' is applied to the tube positions to account for this effect, and is necessary when considering off-axis sources. The images are parameterised according to the moment-based method of Hillas (1985).

Pre-processing steps designed to minimise the effects of electronic interference are described below. The camera is divided into groups of eight tubes which share common high voltage and other circuitry, and a special cut, box (Yoshikoshi 1996), is designed to remove images arising from electronic contamination, and are concentrated in only one or two tube boxes, This box cut, in combination with a total ADC sum (adc ) cut rejecting events with fewer than 200 ADC counts, is very effective at removing such artifacts. Monte Carlo simulations show that the box cut does not reduce the power of the image cuts, and that the optimum adc cut lies at 200 ADC counts.

Mirror degradation resulted in an event rate drop by about a factor of two from 1994 to 1995, indicating a higher energy threshold for the 1995 dataset. The results quoted in this work are normalised to a 1.5 TeV threshold for gamma-rays, using different raw triggering efficiencies for gamma-rays (Sect. 3), which take into account the increase in trigger threshold between 1994 and 1995. In addition, only events with width 0.01 and with the number of triggered tubes, ntubes are accepted. The cuts described above are termed noise cuts.

The ON-OFF statistical significance is calculated following Li & Ma (1983), before and/or after application of all image cuts.

and is used to assess the likelihood of a gamma-ray signal. In order to account for the mismatch of observation times between ON and OFF source data (and hence zenith angle-dependent event rates), and trigger rate differences due to subtle changes in weather conditions and/or telescope response during observation runs, a normalisation is applied to the ON-OFF statistical significance. This normalisation, is defined as the ratio of the events available for image parametrisation, i.e. after noise cuts. A final systematic check on the ON-OFF statistics after application of all cuts, performed on a run-by-run basis, is explained in Sect. 4.

© European Southern Observatory (ESO) 2000

Online publication: June 30, 2000
helpdesk.link@springer.de