Pile-up reconstruction algorithm for high count rate gamma-ray spectrometry

In high count rate gamma-ray spectrometry, the pile-up phenomenon turns out to be an important problem with respect to energy resolution and detection efficiency. Pile-up effects occur when two events are detected so close in time that instrumentation cannot properly extract information from both of them. Because this kind of data is incorrect and marginally useful, such data had to be rejected in traditional pulse processors. In times of digital pulse processing however, one can reconstruct piled-up pulse amplitudes by special algebraic approaches. In fully digital signal acquisition, the moving window deconvolution (MWD) method is commonly used. This method requires two parameters to be carefully set, namely the flattop time (dictated by the maximum rise time of the signal) and the shaping time, to accomplish the best possible energy resolution. In this way, the maximum energy resolution is accomplished, but a lot of piled-up events are rejected, reducing detection efficiency. We propose a method that restores some of the pile-up events, using a parallel block MWD implementation where the shaping time parameter differs for every MWD block. Careful detection of as many true events as possible, as well as determining their exact occurrence in time (their respective timestamps) is the key in getting the most out of the measured signal. With proper analysis logic we get more experimental information through reduced dead time, at the cost of controlled and selectively worsened energy resolution, on an event-by-event basis, achieving better overall detection efficiency. This method was tested on real experimental data where the detection efficiency of our method is higher, by a factor of 4.4(9), than the efficiency of a standard method with pile-up rejection at 500 kcps count rate.