Implementation of a neural network for digital pulse shape analysis on a FPGA for on-line identification of heavy ions

Abstract

Pulse shape analysis techniques for the identification of heavy ions produced in nuclear reactions have been recently proposed as an alternative to energy loss and time of flight methods. However this technique requires a large amount of memory for storing the shapes of charge and current signals. We have implemented a hardware solution for fast on-line processing of the signals producing the relevant information needed for particle identification. Since the pulse shape analysis can be formulated in terms of a pattern recognition problem, a neural network has been implemented in a FPGA device. The design concept has been tested using 12,13C ions produced in heavy ion reactions. The actual latency of the system is about 20μs when using a clock frequency of 50MHz.