FACT -- Operation of the First G-APD Cherenkov Telescope

TL;DR

The FACT telescope successfully operates using G-APD photon counters, demonstrating real-time calibration, high data throughput, and effective data compression, advancing Cherenkov astronomy technology and monitoring capabilities.

Contribution

This paper presents the first operational implementation of a G-APD based Cherenkov telescope with real-time calibration, high data rate handling, and efficient data processing, marking a significant technological advancement.

Findings

Stable G-APD operation under varying environmental conditions.

Real-time data compression outperforms gzip in speed and ratio.

Effective temperature and current feedback systems maintain detector stability.

Abstract

Since more than two years, the First G-APD Cherenkov Telescope (FACT) is operating successfully at the Canary Island of La Palma. Apart from its purpose to serve as a monitoring facility for the brightest TeV blazars, it was built as a major step to establish solid state photon counters as detectors in Cherenkov astronomy. The camera of the First G-APD Cherenkov Telesope comprises 1440 Geiger-mode avalanche photo diodes (G-APD aka. MPPC or SiPM) for photon detection. Since properties as the gain of G-APDs depend on temperature and the applied voltage, a real-time feedback system has been developed and implemented. To correct for the change introduced by temperature, several sensors have been placed close to the photon detectors. Their read out is used to calculate a corresponding voltage offset. In addition to temperature changes, changing current introduces a voltage drop in the supporting resistor network. To correct changes in the voltage drop introduced by varying photon flux from the night-sky background, the current is measured and the voltage drop calculated. To check the stability of the G-APD properties, dark count spectra with high statistics have been taken under different environmental conditions and been evaluated. The maximum data rate delivered by the camera is about 240 MB/s. The recorded data, which can exceed 1 TB in a moonless night, is compressed in real-time with a proprietary loss-less algorithm. The performance is better than gzip by almost a factor of two in compression ratio and speed. In total, two to three CPU cores are needed for data taking. In parallel, a quick-look analysis of the recently recorded data is executed on a second machine. Its result is publicly available within a few minutes after the data were taken. [...]