模拟与混合信号集成电路中基于系统辨识理论的高精度频谱测试方法研究

After several decades of development in integrated circuits, the performance in electronic circuits and systems today has been improved by several orders of magnitude in terms of accuracy, speed and consumption. In some high-accuracy or high-sensitivity applications, such as the test of the high precision analog and mixed-signal (AMS) integrated circuits, the accuracy of the tested signals has even exceeded the measurement range of the best available measurement instruments, which leads to the spectrum evaluation becoming extremely challenging with existing methodologies. This project aims at attacking the following three grand challenges in high precision analog and mixed-signal integrated circuit tests, i.e., the stringent requirements on the high-accuracy signal sources and high-accuracy measurement devices, the coherent sampling and the ultra-low clock jitter. Different from the conventional methods that always make great efforts to meet the above stringent requirements, this project attempts to recast the problems in AMS integrated circuits test as a system identification problem. Exploiting the property that signal processing algorithms can be easily implemented in the digital domain, this project attempts to develop a high-accuracy and low-cost spectral testing method. The expected method has the ability to achieve accurate spectral testing results even if none of the above three requirements is satisfied. As a result, the high-accuracy spectrum can be achieved without precision instruments, and hence the test cost can be reduced significantly. Furthermore, the developed method will provide practical test solutions for on-chip spectral test and the state-of-the-art circuits whose performance exceeds the test capabilities of the best available commercial test equipment.

集成电路历经数十年的发展，如今的电子电路与系统在精度、速度、功耗等方面的性能已获得几个数量级的提升。在一些高精度或高灵敏度的应用领域，如高精度模拟与混合信号集成电路测试领域，测量信号的精度有时超出现有测试仪器的测量范围，采用传统的方法已无法对其进行精确的频谱测试。本项目旨在解决高精度模拟与混合信号集成电路频谱测试所面临的三大挑战，即测试中对高精度信号源和高精度测量器件、相干采样，以及超低的时钟抖动的严格要求。不同于传统的竭力满足上述苛刻要求的做法，本项目拟从系统的角度将模拟与混合信号集成电路测试难题转化为系统辨识问题，并借助较易实现的数字域信号处理算法，探索出一种高精度低成本的频谱测试方法。预期的方法在上述三个要求均不满足的情况下，仍能获得精确的频谱测试结果，从而降低高精度频谱测试对精密仪器的依赖，大幅降低测试成本，并为高性能集成电路频谱测试和片上频谱测试提供切实可行的解决方案。

随着集成电路的持续发展，模拟与混合信号集成电路的性能日益提升，其频谱性能测试变得更加艰难。在一些高精度或高灵敏度的应用领域，如高精度模拟与混合信号集成电路测试领域，测量信号的精度有时超出现有测试仪器的测量范围，采用传统的方法已无法对其进行精确的频谱测试。因此，探索一些低成本、高精度、且易于片上实现的新的测试方法具有重要的意义。.本项目围绕模拟与混合信号集成电路频谱测试和线性度测试面临的三大难题：高精度信号源和高精度测量器件难以获得、输入信号与采样时钟之间的非相干采样问题，以及传统直方图测试法中固有的归一化误差问题开展了深入研究。为降低模拟与混合信号集成电路频谱测试对信号源和测量器件的精度要求，本项目提出了一种基于多路滤波技术的放大器频谱测试方法；为消除多音频谱测试中因非相干采样导致的频谱泄漏问题，本项目提出了一种无需相干采样的多音识别与替代方法；为避免传统直方图方法中固有的归一化误差，提高直方图方法的测试精度，本项目提出了一种无归一化误差的高精度直方图方法。上述方法均已通过严格的理论推导和仿真验证，部分方法已通过实验验证。本项目所提出的方法解决了当前高精度模拟与混合信号集成电路测试中的三大难题，降低了高精度频谱测试对精密仪器的依赖，从而大幅降低了产品测试成本，并为高性能集成电路频谱测试和片上频谱测试提供切实可行的解决方案。.本项目发表论文12篇，其中SCI期刊论文5篇，国际会议论文2篇，申请发明专利4项，获省部级一等奖2项，另有3篇SCI期刊论文在投。本项目培养博士生1人，硕士生12人。