Digital signal processing for CdTe detectors based on a waveform clustering algorithm

A tournament clustering algorithm for treating pulse-height variation of detector signals related to the waveform change has been introduced. The algorithm does not need prior modeling of signals. The algorithm was applied to CdTe/CdZnTe detectors of 2 mm thick. Each classified spectrum was summed up to form a combined pulse-height spectrum using normalization factors obtained during a test run. The obtained energy resolution for 137Cs was improved to 13.1 keV FWHM. This method can be applied to any type of semiconductor detectors that exhibits great variation in individual signal waveforms. The classification results contain information on the signal collection process and can improve the effective performance of the detector.