基于ZYNQ的ICP-MS数据采集处理系统设计与实现

数据采集系统是电感耦合等离子体质谱仪(ICP-MS)的重要组成部分。本研究针对ICP-MS需要同步采集脉冲和模拟信号，且信号信噪比低的问题，提出了一种基于ZYNQ芯片的数据采集处理解决方案。利用ZYNQ集处理器和可编程逻辑于一体的特性，在ARM端搭载了用于流程控制及以太网通讯的FreeRTOS实时操作系统，在FPGA端基于FIR compiler和MCDMA IP核实现了数据的采集、缓存、动态滤波及传输。脱机测试结果表明，脉冲计数误差率小于0.000 5%,模拟计数线性度高达0.999 98,数字滤波幅频特性曲线与理论基本一致。在本团队自行研制的ICP-MS仪器中，利用本系统测试标准调谐液TUNE D,数据采集满足仪器需求，配置64阶低通滤波后，模拟计数中¹¹⁵In谱峰信噪比提高了49.59%。该系统FPGA端各资源占用率均小于50%,片上总功耗仅为2.28 W,有利于推动质谱仪小型化、智能化发展。

Design and Implementation of ICP-MS Data Acquisition and Processing System Based on ZYNQ

Data acquisition system is an important part of inductively coupled plasma mass spectrometer (ICP-MS). Aiming at the problem that ICP MS pulse and analog signal need to be collected synchronously and the signal-to-noise ratio is low, a data acquisition and processing solution based on ZYNQ chip were proposed. Using the characteristics of ZYNQ integrating processor and programmable logic, the FreeRTOS real time operating system for process control and Ethernet communication was equipped on the ARM side. Three modules were designed on the FPGA side to realize data acquisition, cache, dynamic filtering and transmission, including counting module for synchronous acquisition of ICP MS pulse signal and analog signal, FIR filter module supporting online configuration based on FIR compiler IP core, and dual channel data transmission module based on MCDMA IP core. The dynamic configuration of FIR filter module was controlled by ARM, and the filter coefficients were generated by ARM calculation to improve the universality of the system. Moreover, the communication between ARM and FPGA and between modules were all based on AXI4 bus. The Modbus/TCP transport protocol was used for Ethernet communication to ensure the stability and universality of the system. The system test was divided into two steps of offline test and online test. The offline test verified the function and performance of each module in the system. The performance of the pulse count module was tested by using the pulse signal generated by the signal generator AFG320 to simulate ICP MS. After eliminating the error caused by the crystal vibration deviation, the relative error of the pulse count was less than 0000 5%. The ADC in the analog count was calibrated by using DP832 DC regulator power supply and 34401A six bit half digital multimeter. The result showed that the fitting linearity is as high as 0999 98. The digital filter performance was tested by inputting sinusoidal signals of different frequencies, and the results showed that the amplitude frequency characteristic curve is basically consistent with the theory. A standard tuning solution, tune D (Thermo Fisher Scientific), was tested in our team's self study ICP-MS instrument in-house using this system. It was found that the pulse count had little noise, while the analog count had a large amount of noise thereby affecting the subsequent spectral peak feature extraction. The noise spectrum distribution in the data was analyzed by FFT analysis of the analog count results, and the appropriate parameters of FIR digital filter were selected according to the characteristics of Gaussian peaks. After filtering, the analog count spectrogram is improved significantly, the main noise intensity attenuation is greater than 40 dB, and the signal to noise ratio of 115In peak is increased by 4959%. The filtered spectrum fully meets the requirements of subsequent peak feature extraction. The resource occupancy rate of the FPGA end of the system is less than 50%, and the total power consumption on the chip is only 2.28 W, which is conducive to promoting the miniaturization and intellectualization of mass spectrometry. In the future, algorithms can be designed at the ARM side or upper computer end to automatically adjust the filter parameters and further improve the degree of system automation.