[en] Signal processing in the digital domain is known to have better performance than analog processing for pulse shape discrimination of neutron and gamma ray pulses. Digital signal processing (DSP) can synthesize any filter response without the associated signal degradation which happens in the complex analog signal path. Neutron absorption cross-section measurement experiments derive the information of neutron energy from time of flight along a length of channel. Single channel multi-hit time marker approach gives required time resolution and dynamic range for the neutron energies of interest. The predominant gamma background affects the pulse count rates possible in such an experiment. The development of an FPGA based digital signal processing system has been undertaken for this application with counter based multi-hit time marker approach. The inherent parallel architecture of a FPGA implementation will result in better performance compared to DSP processor based implementation. A trigger input to the system indicates the start of the neutron beam which restarts a fast counter. The time of arrival (TOA) for each neutron/gamma ray generates a pulse from a detector. Each pulse is processed in two parallel paths, one for counting and one for pulse shape discrimination. The counting channel latches the output of the fast counter and transfers it to a TOA FIFO. This approach results in higher count rates compared to multi-channel scaling or time-to-analog conversion followed by MCA approach. The pulse shape discriminator is based on the fact that the detector pulses have different decaying tails for neutron and gamma rays. A longer tail is expected for a neutron pulse than for a gamma pulse. Each acquired pulse is passed through a chain of signal processing which compares the total energy with the pulse amplitude to differentiate neutron and gamma pulses. The filtered neutron events are transferred from the TOA FIFO to a Neutron TOA FIFO which is then used for further analysis. The counting channel is much faster than the shape-processing channel thus limiting the event rate. Implementing a number of signal processing channels in parallel improves the event rate. (author)