【作者】 高岑；

【导师】 姚素英；

【作者基本信息】 天津大学 ， 微电子学与固体电子学， 2012， 博士

【摘要】 时间延迟积分(Time-Delay-Integration, TDI)型图像传感器是一种特殊的传感器，它以面阵图像传感器的形式完成线阵扫描成像的功能，能够实现高速扫描和高分辨率影像，可获得比传统面阵图像传感器更高的分辨率和精度，在空间遥感、航空成像、工业监测控制和安全监控等领域具有极其重要的应用价值。随着CMOS工艺特征尺寸的缩小和CMOS处理电路可集成性的提高，基于标准CMOS工艺的TDI型图像传感器已经成为该领域的研究热点。因此，本文立足于混合域累加的系统架构和读出处理电路设计，开展TDI型CMOS图像传感器设计的关键技术研究。本文深入分析现有TDI型图像传感器的研究现状，提出一种可节省芯片面积并同时实现高速行频的基于混合域累加的系统研究方案；立足于CMOS工艺特点和发展趋势，设计了一种输出电流信号的有源像素结构；研究可用于实现32级混合域累加的电流型模拟累加器和数字累加器，并对其噪声特性进行分析；针对本文研究方案的列级读出架构，提出了两种模数转换器结构。一种为带有误差校准的单斜ADC，另一种为可直接量化电流信号的Cyclic ADC。本文通过理论分析、系统建模、电路设计仿真以及测试，完成TDI型CMOS图像传感器整体系统，包括像素与读出电路的设计，并采用GSMC混合信号工艺对系统中的关键模块进行投片验证。仿真与测试结果表明，所设计的电路能够满足系统要求。本文的创新性工作包括：1、基于32级TDI型CMOS图像传感器，提出一种新型的混合域累加的读出电路结构。该读出结构结合模拟域累加和数字域累加各自的优势，能够大幅节省累加器占用的芯片面积，在保证高速输出行频的基础上，放宽数模转换器对转换速率的要求。仿真与验证结果表明，该混合域读电路结构能够实现与单纯模拟域累加或单纯数字域累加相同的信噪比提升性能。2、在研究TDI型CMOS图像传感器工作原理和时间延迟积分技术的基础上，基于混合域累加的读出结构，分析推导出电流累加器的低噪声设计方法，并设计一种低噪声的新型电流信号累加电路。3、提出了一种量化电流信号的模数转换器电路结构，在保证转换精度和低芯片面积的前提下，提高了ADC的功耗效率。同时设计了一种用于该模数转换器的电流采样保持电路，该电路能够消除开关电荷注入中与信号相关的噪声，并抑制与信号不相关的噪声。更多还原

【Abstract】 Time-Delay-Integration (TDI) image sensor is a particular visual sensor, which cancapture image information in the form of two dimensions array working as thescanning mode of one line pixels. And the TDI image sensor achieves higher scanningspeed and higher resolution compared to traditional array image sensor, so it isespecially suitable for applications in space image, aerial photography, industrialdetection and security. As the scaling of CMOS standard process technology andincreasing of integration in processing circuit, the TDI image sensor based on CMOSprocess technology becomes the most popular issue in the field of imaging system.Therefore, this paper will focus on the mixed-domain-accumulation mode andprocessing circuit design, and research the key technology in the TDI CMOS imagesensor.Supporting by in-depth analysis of current research scheme in TDI image sensor, asystematic based on the mixed-domain-accumulation mode which can save the siliconarea and realize high speed readout is firstly proposed. Considering the feature and thedevelopment tendency, a linear current-mode pixel is designed. And a32stagesmixed-domain accumulator is designed based on the research of the readout circuit inTDI CMOS image sensor, and the noise characteristic is also analyzed. Finally, twotypes of column level analog-to-digital converters (ADCs) are proposed for the TDIreadout system. One is a single-slope ADC with error calibration, the other one is aCyclic ADC which can be used to digitalized current signal directly. In this paper, thecorresponding current-mode pixel and the readout circuit are discussed in detailthrough theoretical analysis, system modeling, circuit simulation and testing. Themain blocks have been fabricated and verified. The result of simulation and testingindicates that the proposed TDI CMOS image sensor works effectively.The creative work of the paper include:1. A structure of mixed-domain-accumulation is proposed based on the32stagesTDI CMOS image sensor. The new structure combines the advantages ofanalog-domain and digital domain, which can save the silicon area and relax therestrictions of the converting speed of ADC. The result of simulation indicates that thesignal-to-noise ratio can be improved as well as analog-domain or digital-domain.2. A low noise current accumulator is proposed based on the principle of TDI imagesensor and mixed-domain-accumulation mode structure. And the optimization designmethods of low noise current accumulator are derived.3. A current mode ADC which can be used to digitalized current signal is proposed.The power efficiency can be guaranteed without adding the silicon area or decreasing the precision. Meanwhile, a current-mode sample-and-hold circuit, which can be usedto remove the signal-dependent charge injection and suppress the signal-independentcharge injection, is adopted to improve the precision of the ADC.更多还原