Real-time implementation of a deconvolution filter for gamma ray spectroscopy

Abstract

In this paper, we propose a new approach for real-time detection of pileup in gamma-ray spectroscopy using finite length deconvolution filters; in particular, a 3-point deconvolution with Savitzky Golay prefilter is discussed. The approach was tested and proven to be able to resolve accurately pileup in 93% of the cases. Using the proposed approach, the number of pileup events can be reduced by eight folds. The setup was tested with both simulated data and random signals from a ¹³⁷Cs test source. More specifically, gamma pulses, from a 2 inch NaI(TI) scintillation detector, were captured as single and double pulses for the purpose of testing the proposed peak detection algorithm. The algorithms developed here were then implemented in real time using a high performance floating-point processor, the TMS320C6711. A number of optimisation levels were achieved using the Code Composer Studio profiler based on critical timing to satisfy the real-time constraints of gamma-ray spectroscopy systems. A comparison in complexity and real time computations among various deconvolution algorithms has been carried using the TMS320C6711 processor.