基于现场可编程门阵列的质谱信号滤波器的设计与应用

Design and Application of Mass Spectrometer Signal Filter Based on Field Programmable Gate Array

低通滤波法是四极杆质谱仪信号去噪的常见方法,为了提高滤波器性能,引入一维离散平稳小波进行阈值法滤波,其不仅能够几乎完全抑制噪声,而且可以很好地保留原始信号的特征尖峰点,缩短与成熟商业化产品差距。同时针对四极杆质谱控制系统运算能力不足的问题,本研究提出了基于Zynq芯片的应用方案,大大降低了仪器得到高信噪比(Signal-To-Noise,SNR)和低均方误差(Mean-Square-Error,MSE)质谱数据的时间。基于本团队研制的四极杆质谱仪,测试使用赛默飞世尔科技(Thermo Fisher Scientific)公司的调谐液样品。结果显示,小波变换法噪声衰减25 dB以上,对比240阶低通滤波法MSE整体降低20%以上,有效部分降低8%以上,并且通过可配置现场可编程逻辑门阵列(Field Programmable Gate Array,FPGA)处理,使有效信号相较低通滤波法下降15%以上,同时处理延迟小于1μs。为优化质谱数据降噪和控制系统集成化提供了设计参考。

Low-pass filtering is a common method for signal denoising in quadrupole mass spectrometers.In order to improve filter performance,a one-dimensional discrete stationary wavelet is introduced for threshold filtering,which can not only suppress noise almost completely,but also preserve the original signal,and characteristic peak points,shortening the gap with mature commercial products.At the same time,in view of the insufficient computing power of the quadrupole mass spectrometer control system,this study proposes an application scheme based on Zynq chip,which greatly reduces the time for the instrument to obtain high signal-to-noise ratio(Signal-To-Noise,SNR)and low mean-square-error(Mean-Square-Error,MSE)mass spectral data.Based on the quadrupole mass spectrometer developed by our team,the tuner samples from Thermo Fisher Scientific were used for testing.The noise attenuation of the wavelet transform method is more than 25 dB.Compared with the 240 order low-pass filtering method,the MSE is reduced by more than 20%overall,and the effective part is reduced by more than 8%.The effective signal is reduced by more than 15%compared with the low-pass filtering method through the configurable field programmable gate array(FPGA)processing,and the processing delay is less than 1μs.It provides a design reference for optimizing mass spectrum data noise reduction and control system integration.