256光电管阵列四象限CMOS光电传感器研究

【作者】 周鑫；

【导师】 朱大中；

【作者基本信息】 浙江大学 ， 微电子与固体电子学， 2004， 硕士

【摘要】 基于硅光电传感的象限传感器广泛应用于激光的瞄准、制导、跟踪，搜索装置，精密测量，如激光微定位、位移监控、精密机床的光电控制等领域。传统的硅象限传感器主要采用四象限，八象限的结构，采用焊接的方式将分别制造的感光象限和电路结合在一起，不能实现感光象限与信号处理电路的单芯片集成，难以实现器件的微型化和系统优化；同时，传统象限传感器的感光象限单元数量较少，获取的目标信息有限，难以实现对目标的坐标位置获取、形状识别等功能。 在本研究中，将传统的象限传感器与当前迅速发展的CMOS图象传感器相结合，提出了使用有源传感阵列感光的256单元光电管阵列四象限CMOS光电传感器。该传感器的感光单元采用了CMOS图象传感器中使用的有源像素传感器(Active Pixel Sensor，APS)设计，在感光单元内部由光电信号预处理电路直接将传感产生的光电信号转化为幅度较大的电信号输出，避免了对微弱信号的处理，降低了噪声的影响。传感器应用阵列采集光信号，可以直接确定目标光源的坐标位置并实现一步到位的快速调整。传感器使用标准CMOS工艺制造，将传感阵列与信号处理电路集成在同一芯片上，可以实现传感器的SOC集成和智能化(Smart Sensor)设计。针对CMOS制造工艺中MOS器件固定模式噪声(Fixed Pattern Noise，FPN)较大的不足，研究中采用了相关二次采样(Correlated Double Sampling，CDS)电路降低固定模式噪声，从而提高了传感器的信噪比(Signal Noise Ratio，SNR)。针对CMOS光电传感器感光动态范围不大的不足，提出了改变复位信号频率两次扫描的传感器工作方式，有效提高了传感器的感光动念范围。 本研究中的256光电管阵列四象限光电传感器包含有16×16有源光电管阵列，相关二次采样降噪电路，输出缓冲放大电路和数字控制电路四个主要模块。传感阵列中各列感光单元的传感信号并行输出，分别由对应的相关二次采样电路进行降噪处理，去除固定模式噪声后的信号通过输出缓冲放大电路进行输出。整个传感系统在数字控制部分电路产生的控制信号的作用下工作。 本研究中的256光电管阵列四象限光电传感器采用了上华0.6μm两层金属两层多晶硅CMOS标准工艺制造。传感器感光单元的面积为60μm×60μm，感 ．一浙江大学硕士学位论文 摘要光面积百分比（Fill Factor）为 64．5％，传感阵列中光电管为 16 XI6排列，芯片面积为 2．smm X 2石mm，电源电压为 SV。复位信号为 SV时的单帧感光动态范围为60dB，采用改变复位信号频率的二次扫描方式可将传感器的总的感光动态范围扩大到 84dB，可对 0石＊～10，000lX光照强度的信号进行传感。 在对感光单元进行器件物理结构优化的研究中，通过采用深结深光电管结构，提高了传感器的感光响应，其中 P＋M阶”衬底结构的传感器面积为 100 umX 100urn，感光面积百分比为77石％，可对0刀＊～98，000lx照度的目标信号进行传感，感光灵敏度大于 3 SV／IX·S，采用了变频两次扫描后，动态范围可达139．sdB，在己经报道的传感器中处于较高的水平。 本文的第一章主要介绍了图象传感器和象限传感器的发展，特点和比较。其中对传统的马赛克结构图象传感器和新出现的 Fovea X3 结构的图象传感器结构，CMOS传感器和 CCD传感器的比较都进行了详细的介绍。第二章主要介绍 传感器的各种感光单元结构，包括无源感光单元和有源感光单元的结构，并且对 传感器中的各种噪声进行了分析。第三章主要介绍了本设计中的256光电管阵列 四象限光电传感器感光单元模型建立，各个模块的电路设计和仿真，同时也对电’路的具体工作原理和方式进行了详细的介绍。第四章介绍了本设计中传感器的版 图设计以及各种不同传感单元结构在器件物理结构上的优化设计。第五章主要介 绍了通过数字电路设计方法（Verilog语言描述，Synopsys软件综合，Cadence SE 自动布局布线）进行数字控制电路设计的方法以及传感器感光动态范围调整的设 计考虑。第六章介绍了传感器的测试方法和测试结果以及对测试结果的分析，提 出在新的设计中应该注意的问题。第七章对整个设计工作进行了总结，并提出了 下一步的设计工作中可以进一步发展深入的方向。 0 五二更多还原

【Abstract】 Quadrants sensor based on silicon are now widely used in many fields, such as laser collimation, homing, tracking and searching devices, precision measurement et al. The conventional quadrants sensors mainly have 4 or 8 quadrants, which are not integrated with the signal processing circuits in one chip but are soldered with the discrete signal processing circuits. In this way, it’s difficult to realize the micromation and system optimization of the devices. In addition, there are few number of photoelectric sensor parts in conventional quadrants sensors, which limits the targets information got from the conventional quadrants sensors. Thus, it’s hard to get the coordinate and shape information of targets.In this design, conventional quadrants sensors and CMOS image sensors which are now developing very quickly are jointed together. A 256 photodiodes arrayed 4 quadrants sensors which uses active pixel sensor array is designed. The active pixel sensor structure widely used in image sensor is used here to convert the micro photoelectric signal generated in pixel to large electric signal to facilitate the signal processing and reduce the noise. The arrayed sensor structure 昳s used to obtain the coordinate and shape information of targets. CMOS process is used to realize the one chip integration of the sensor array and processing circuits, which accords with the development direction of SoC(System on a Chip) and smart sensor. The Correlated Double Sampling(CDS) circuits structure is used to reduce the Fixed Pattern Noise(FPN) of the CMOS sensors, and improve the signal-noise ratio(SNR).The 256 photodiodes arrayed 4 quadrants sensor consists 16X16 active photodiodes array. correlated double sampling processing circuits ,output buffer amplifiers and digital control part. Every column in sensor array work in parallel and have their own CDS noise reducing circuit. The signals after FPN reducing are output from the output buffer amplifiers. All the sensor system is operated under the control of the digital control circuits of the sensor.The 256 photodiodes arrayed 4 quadrants sensor is designed and fabricated in 0.6 u m double metal double poly standard CMOS process. The area of each active photodiode is 60 u m X60 u m and the fill factor is 64.5%. The chip area is 2.8mmX 2.6mm . The dynamic range(DR) of the sensor can be enlarged with the changeable reset frequency, as a result, from 60dB to 84dB DR. It tan operate in the illuminance range between 0.61x and 10,0001x. The sensor operation speed can be 64ms/frame ~2ms/frame.In the research of photoelectric cell, device physics structure of pixels have been optimized. Deep junction depth photodiodes, such as P+/N-well/P-sub structure, have been used and the photo-response of the sensor has been greatly enhanced. The pixel size of P+/N-well/P-sub structure is 100 umX 100 n m, fill factor is 77.6%. It can obtain target information with illuminance intensity in the range of 0.011x~98,0001x, and the sensor photoelectric sensitivity is 35V/lx ?s. When the method of changeable reset frequency double scanning is used, the photoelectric dynamic range can be 139.8dB, which is high in the 0.6 um level CMOS image sensors already reported.In the first chapter, development, characters and comparison of the image sensors and quadrants sensors are introduced. Moveover, conventional mosaic structure and new fovea x3 structure image sensors and comparison of the CMOS and CCD image sensors are also introduced. In chapter 2, different pixel structures, including passive and active pixel structure, and all kinds ’of noise in image sensors are introduced. In chapter 3, circuits operating principle, design and’simulation of the sensor are introduced. In chapter 4, layout design of the sensor and pixel structure optimization in device physics are introduced. In chapter 5, design methods of the digital control circuits are introduced. Further more, sensor dynamic range adjustment methods are also introduced. In chapter 6, measurement and results are introduced and analysis a更多还原