GRBM in-flight performances

The BeppoSAX GRBM started its in-flight operations one month after the satellite launch. The overall performances were successfully tested during the BeppoSAX Commissioning Phase, and during the Scientific Verification Phase the proper setting of all thresholds and trigger parameters, on the basis of the radiation environment in which the experiment had to operate, were identified ([Feroci et al. 1997a]).

At the beginning of the mission the default values for the energy thresholds were 1 (nominally corresponding to 24.67 keV, calibrated to 32.8 keV) for the LLT of the four shields and 6 (corresponding to a calibrated energy of 604 keV for the photo-peak) for the ULT. The ACT was set as a default at 0 (lowest value, nominal 100 keV). The trigger condition at that time was set with LIT=128 s, SIT=4 s and n=8. Considering also that the low efficiency of the GRBM detectors at the lower energies, due to the opacity of the materials in their field of view (chapter 3), it was decided to rise the LLT to its step 3, that is a calibrated value of 42.5 keV, for the four shields. This new threshold setting is therefore almost ineffective from the point of view of the source detection, but it largely suppresses the noise due to spikes and also reduces the background count rate (see next section). We note, however, that single spikes do not affect the trigger condition, that requires the time coincidence of the trigger on at least two shields, but a SIT of 4 s increases the probability of getting by chance a simultaneous trigger condition on two shields is enhanced by the fact that the temporal coincidence is extended to 4s. Furthermore, we note that the intensity of many spikes frequently exceeds the 8$\sigma$ level over the background, and therefore the n=8 condition cannot be considered an efficient discrimination method of the real GRB signals from the spikes.

Another important calibration and setting concerned therefore the triggering parameters. With a SIT value of 4 s and n=8 the trigger condition selects GRBs with high intensity and long duration (i.e. one second or more). Given the typical duration of GRBs the SIT was lowered to 1s in order to be able to catch also shorter/weaker events. The high value of n was no more needed after the GRBM count rate was reduced to a almost-Poissonian statistics with the suppression of most of the spikes. Taking into account that the trigger condition must be satisfied simultaneously on at least two shields in order to give a proper GRB trigger, then the value of 4 for n can be considered safe enough against false triggers. In fact, considering that in one orbit (97 minutes) we get about 4000 s of good data (due to the data acquisition interruption at the passage over the South Atlantic Geomagnetic Anomaly, SAGA), following a Gaussian statistics we expect that the count rate exceeds by 4$\sigma$ the average background count rate with a probability of 6x10^-5 each second in one shield. Since the non-source events must be uncorrelated on two shields the probability of having a GRB trigger by chance in Gaussian statistics is about 2x10^-8 each second, that is of the order of 8x10^-5 each orbit. Nevertheless, as we still discuss later, the in-orbit background is not Gaussian, and this causes the number of false triggers to be about 1 per orbit.