Algorithmic Study Of Digital Processing On Sampled Signal Of Nuclear Measurement

In the work, it pays more attention on the research of digital processing of high speed sampled data of nuclear signal, containing the filtering of discrete serials of nuclear signal and the affection of energy spectrum the filter filtered to discrete serials of nuclear signal, and the arithmetic study of pulse shape discrimination (PSD).I filtered the signal with the noise including different frequencies by the highpass filter with the function of Kaiser window, researched the result the signal filtered influenced to the nature of energy spectrum. In the different cutoff frequencies and different amplitudes of the signal noise, the resolving power of energy spectrum signal filtered was improved clearly, but when the signal noise rate (SNR) of signal amplitude had arrived at 20%, although the energy spectrum improved, the distribution of energy spectrum distinctly departed from the normal gauss distribution. Lifted up the cutoff frequency could upgrade the resolving power of energy spectrum, as well as the amplitude ratio of both the signal amplitude with disturbing noise and the amplitude the signal filtered was linear. However, the higher the cutoff frequency, the more the amplitude declined. So, we better get the balance in the practice work.Next, I studyed the characteristics of the self-adaptive filter used the signal resource the same as the highpass filter. The amplitude of signal filtered declined light not liked that the center position of energy spectrum moved to the left obviously used the highpass filter to filter. And the distribution of energy spectrum is symmetrical, the width of energy spectrum is narrower than that used by the highpass filter, all of above thing I said proved the self-adaptive filter was more superior than the highpass filter. So, in the digital processing of sampled signals, we should choice the self-adaptive filter first, but the performance of work is lower than the highpass filter as it's arithmetic is too complex.At last, I researched the pulse shape discrimination (PSD) originally, submitted three methods of PSD, such as Zero Crossing Method, Charge Integration Method, Mode Nature Extracting. Research found based on two styles signal, every method had some shortcoming in the different situation as the rising time, the amplitude of signals, the width of signals changed. Zero Crossing Method is good at the situation the amplitude is close but not the rising time. Charge Integration Method was well do the situation the rising time was close but not the amplitude. Mode Nature Extracting, as thought about these factors of the amplitudes, the rising time of signals, the width of signals, the accumulation of signals and so on, is higher performance, more extent applications than forward two method.Finally, all of above arithmetics was programmed by C language implanted completely on the CCS platform of TI company, supported the software design of core arithmetics of DSP. Hence, the results of above all I said is very important to finish the research of PCI acquiring card.