在能量色散谱仪(energy dispersive spectroscopy,EDS)中,特征X射线信号幅值提取是其进行元素分析的关键,数字成形算法则是现代X射线信号处理的核心所在.本文对硅漂移(silicon drift detector,SDD)探测器前置放大器输出信号进行了转化处理,针对调理后的双指数脉冲信号采用FPGA设计了一种高频率、高精度、参数可调的实时脉冲梯形成形算法.并使用理想双指数信号和实测的X射线特征谱信号验证了算法的可行性与实时性,与MATLAB双精度算法成形结果比较测试了算法精度.实验结果表明:该算法可在100 MHz频率下实时将双指数脉冲成形为梯形脉冲,成形效果稳定.其实际信号成形结果与双精度算法结果保持一致,平顶幅值相差在0.6％以内.

Energy resolution, considered to be one of the most important parameters, reflects the performance of Silicon Drift Detectors (SDDs). A constant area spectral peak reconstruction algorithm is proposed to improve the energy resolution of SDDs in this paper. Without changing the peak position and peak area, the peak width is reduced by spectral peak reconstruction. The results show that the energy resolution of the SDDs is improved by 28.57% using the algorithm. Moreover, with the increase of energy resolution, the peak amplitude increases, which in turn increases the peak-to-back ratio by 18.71%. For sum peaks, it is effective to decompose sum peaks by the least square fitting. The method is applied to fit the curve to solve the relevant parameters and the peak widths are changed to reconstruct the Gaussian peak.

BackgroundIn the X-ray energy spectrum measurement system, the time constant is an important factor affecting trapezoidal pulse shaping. Accurate measurement of the time constant of double exponential pulse signals is beneficial to obtaining a better trapezoidal pulse shaping.PurposeThis study aims to propose a fast fitting double exponential pulse algorithm for measuring the time constant in trapezoidal pulse shaping.MethodsFirst of all, the linear regression method based on the least square method was adopted to convert nonlinear fitting linear fitting. Then, functional expression of double exponential pulse signal was fitted by fast trapezoidal pulse shaping, and the feasibility of time constant measurement in trapezoidal forming by fast fitting double exponential algorithm was verified by MATLAB programming simulation. Finally, the practicability of the algorithm was evaluated by experimental pulse data of 55Fe radiation source received by X-ray silicon draft detector (SDD). Goodness-of-fit R2 of this algorithm was compared with that of cftool toolbox of MATLAB on both the simulation data and experimental data.ResultsThe goodness of fit R2 of proposed algorithm exceeds 0.95 for both the simulated signal with 5% standard deviation noise and the shaped pulse signal of experiments. When the goodness of fit exceeds 0.95, the running speed of this algorithm is 2 orders of magnitude faster than that in cftool toolbox of MATLAB.ConclusionsThe algorithm of this paper possesses high fitting degree, fast calculation time, hence suitable for real-time measurement of time constant of double exponential pulse signal in embedded system.