The ability of a PET system to reproduce the source distribution accur
ately depends, in the first place, on proper calibration and efficient
quality control of the detectors and processing electronics. These pr
ocedures are simplified with individual detectors such as BGO/avalanch
e photodiode detectors, but the large number of channels and the sever
al interdependent parameters that must be adjusted make the process de
licate and lengthy. This paper describes the procedures and algorithms
being developed to adjust and verify the detector bias, the noise and
energy thresholds, the signal references and the timing delays for th
e Sherbrooke PET tomograph. All settings are adjusted using a sequence
of iterative tuning algorithms whose convergence is optimized for eac
h parameter. Preliminary results show that the complete calibration ca
n be carried out automatically and reliably overnight and that specifi
c checks can be performed in a few tens of seconds as needed.