高速高精度ADC频域特性测试方法研究

发布时间：2018-03-02 03:27

本文关键词： 模数转换器频域特性非相干采样频谱泄漏平均频谱法三谱线插值快速傅立叶变换算法　出处：《江南大学》2015年硕士论文　论文类型：学位论文

【摘要】：随着电子信息技术的迅猛发展,作为数字和模拟接口电路的模数转换器(ADC)在数据采集、精密工业测量和音视频应用等不同领域获得了广泛应用,对ADC频域特性的研究也获得了越来越多的关注。快速傅里叶变换(FFT)法是频域测试中应用最普遍的一种方法,但利用FFT法测试ADC的频域参数时,很难做到相干采样和整周期截断,即非相干采样存在必然性,由此造成的频谱泄漏和栅栏效应将影响频域参数测量的结果。本论文主要研究利用FFT法测试高速高精度ADC的重要频域特性参数,包括信噪比(SNR)、信噪失真比(SINAD)、有效位数(ENOB)、无杂散动态范围(SFDR)和总谐波失真(THD)等,并提出了采用平均频谱法和三谱线插值FFT法来抑制频谱泄漏和栅栏效应。论文的主要研究工作如下。首先介绍了ADC的基本结构和工作原理,强调流水线型是目前高速高精度ADC中应用最广泛的结构;给出了高速高精度ADC重要频域特性参数的定义及计算方法;详细描述了高速高精度ADC频域参数的测试方法,包括传统的数模转换器(DAC)测试法、码密度直方图法、正弦波拟合法及FFT法,并比较和归纳了它们的优缺点及适用范围。其次,利用Simulink对12 bits 200 MSPS和14 bits 160 MSPS两种高速高精度流水线ADC架构进行了系统建模;采用美国模拟器件公司(ADI)三款典型高速ADC产品AD9230、AD9246和AD9461的行为级动态模型构建了验证平台;并基于现场可编程门阵列(FPGA)、高性能信号源、滤波器等设计搭建了一套硬件测试系统,为提出的频域特性测试方法提供了仿真及实验平台。再次,对平均频谱法开展了较深入的研究。为了提高频谱的分析精度,降低信号中随机信号的影响,对ADC输出端的数字信号采样两次或两次以上,对这些数据分别进行加窗FFT运算,将得到的频谱图叠加求得平均频谱,然后基于平均频谱计算出频域参数。仿真及实测结果均表明,即使在非相干程度最大的情况下,与加窗FFT法相比,平均频谱法得到的频域参数误差明显减小,接近相干采样的结果,达到了ADI的测试标准。最后,为了降低频谱泄漏,构建了最大旁瓣衰减窗对信号进行处理;为了减小栅栏效应引起的误差,提出了三谱线插值FFT算法对加窗的结果进行修正。利用函数拟合得到了基于常用组合余弦窗及最大旁瓣衰减窗三谱线插值的幅值修正公式,并与平均频谱法的测试结果进行了对比。仿真及实测结果均表明,平均频谱法能准确测试SNR、ENOB等参数,而三谱线插值FFT法非常适合测试高速高精度ADC的SFDR参数。为了验证本论文所提方法的有效性和准确性,采用由两个仿真验证平台和一个实际测试系统所得的数据对算法进行了验证。结果表明,本论文所提方法通用性强,能有效测试高速高精度ADC的频域参数;将其应用到实际测试系统中时,不仅能降低系统构建难度,而且能显著减少测试成本。
[Abstract]:With the rapid development of electronic information technology, A / D converter (ADC), as a digital and analog interface circuit, has been widely used in different fields, such as data acquisition, precision industrial measurement, audio and video applications, etc. More and more attention has been paid to the study of frequency domain characteristics of ADC. Fast Fourier transform (FFT) method is the most popular method in frequency domain testing, but it is difficult to achieve coherent sampling and integer period truncation when using FFT method to test the frequency domain parameters of ADC. That is, incoherent sampling is inevitable, and the frequency leakage and fence effect will affect the results of frequency domain parameter measurement. In this paper, the important frequency domain characteristic parameters of high speed and high precision ADC are measured by FFT method. These include signal-to-noise ratio (SNR), signal-noise-distortion ratio (SNR), effective bit number (ENOB), no stray dynamic range (SFDR) and total harmonic distortion (THD), etc. The average spectrum method and three-line interpolation FFT method are proposed to suppress spectrum leakage and fence effect. The main work of this paper is as follows. Firstly, the basic structure and working principle of ADC are introduced. It is emphasized that pipeline type is the most widely used structure in high speed and high precision ADC at present, the definition and calculation method of important frequency domain characteristic parameters of high speed and high precision ADC are given, and the testing methods of high speed and high precision ADC frequency domain parameters are described in detail. Including the traditional DAC testing method, code density histogram method, sinusoidal wave fitting method and FFT method, and compared and summarized their advantages and disadvantages and applicable scope. Two kinds of high-speed and high-precision pipeline ADC architecture, 12 bits 200 MSPS and 14 bits 160 MSPS, are modeled by Simulink, and the behavioral level dynamic models of AD9230 AD9246 and AD9461 are used to build a verification platform. A hardware testing system based on FPGA FPGA, high performance signal source and filter is designed and built, which provides a simulation and experimental platform for the proposed testing method in frequency domain. In order to improve the accuracy of spectrum analysis and reduce the influence of random signal in the signal, the digital signal of ADC output is sampled twice or more times, and the data are windowed by FFT operation, respectively, in order to improve the accuracy of spectrum analysis and reduce the influence of random signal in the signal. The average spectrum is obtained by superposition of the obtained spectrum, and then the frequency domain parameters are calculated based on the average spectrum. The simulation and the measured results show that, even when the incoherence degree is the greatest, it is compared with the windowed FFT method. The error of frequency domain parameters obtained by the average spectrum method is obviously reduced, which is close to the result of coherent sampling and meets the test standard of ADI. Finally, in order to reduce the spectrum leakage, the maximum sidelobe attenuation window is constructed to process the signal. In order to reduce the error caused by the fence effect, the three-spectral line interpolation (FFT) algorithm is proposed to modify the result of window adding, and the amplitude correction formula based on the commonly used combination cosine window and the maximum sidelobe attenuation window is obtained by function fitting. The results of simulation and measurement show that the average spectrum method can accurately measure the parameters of SNRX ENOB. In order to verify the validity and accuracy of the proposed method, the three-spectral line interpolation FFT method is very suitable for measuring the SFDR parameters of high speed and high precision ADC. The algorithm is verified by the data obtained from two simulation verification platforms and an actual test system. The results show that the proposed method is universal and can effectively test the frequency domain parameters of ADC with high speed and high precision. When it is applied to the actual test system, it can not only reduce the difficulty of system construction, but also significantly reduce the test cost.
【学位授予单位】：江南大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：TN792

【参考文献】