Research On High Precision Readout System Of FGMOS Image Sensors

Image sensors are widely used in automotive electronics,security,portable medical,mobile devices and other fields.With continuously increasing requirement for image quality of electronic products,high-performance image sensors need to be developed.Among the various types of image sensors,CMOS image sensors have gained popularity due to their advantages such as low cost,low power consumption and high dynamic range.The floating gate MOS(FGMOS)image sensors are also attracting more and more attention owing to their multi-signal control of pixel devices and variable threshold control.The imaging mode of FGMOS image sensors can be divided into the threshold reading and the current reading.This thesis presents the design of a high-precision peripheral analog readout circuit for image sensors based on the current reading method of FGMOS-type pixel devices.Two temperature correction methods using comparators and analog-to-digital converters(ADC)are proposed for the overall readout circuit architecture and the FGMOS pixel array.In addition,this thesis conducts some research and exploration on the new sense-memory-computing integrated image sensor,and designs two novel data processing circuits based on FGMOS-type devices for the compute-in-memory of NOR Flash memory cells.The main research contents of this thesis are summarized as follows.(1)Firstly,the main working mechanism of the FGMOS-type image sensor is introduced,including the working principle of FGMOS pixel devices,the current reading principle and the imaging principle.Then a high-precision peripheral readout circuit of image sensor is designed for the current readout imaging mode of FGMOS devices,including clamp circuit,current mirror copy circuit,current-voltage(I-V)conversion circuit,comparator circuit,and ramp voltage circuit.Since the current readout mode of FGMOS pixel devices is similar to that of ordinary MOS devices,the drain terminal voltage of FGMOS is fixed by the clamp circuit,allowing it to work in the linear region like ordinary MOS and generate a fixed current value.The current mirror circuit can copy the current generated by the pixel device with high precision.The I-V conversion circuit converts current into voltage through capacitor integration,and the comparator circuit compares the voltage of the I-V conversion to the ramp voltage and outputs a digital value for the counter to count.In order to fully evaluate the performance of the circuit,this thesis proposes several key performance indices for the readout circuit,which can serve as its implementation standard.(2)According to the above circuit analysis,the detailed circuit design and optimization design for each sub-module are carried out.The simulation results indicate that,within the0～10μA input current range,the clamp voltage of the clamp circuit is approximately 200 m V with a fluctuation of 0.3 m V,which satisfies the high precision clamp requirements and can realize the high precision copy of the output current of FGMOS devices.The maximum error of the copy accuracy of the current mirror is 4 n A,while the minimum comparison voltage of the comparator is 550μV.Subsequently,the effective number of bits and signal-to-noise ratio analysis of the whole analog readout circuit are then conducted,which are 9.72 d B and 60.75d B,respectively,meeting the high precision design requirements of the readout circuit.As the current reference providing bias for operational amplifier is susceptible to temperature effects,a new current reference with low temperature coefficient is proposed,which is mainly composed of bandgap reference,operational amplifier and current mirror.After completing simulations of the sub-module circuits and the whole circuit,the layout design and the post-simulation verification of the circuits are carried out.The post-simulation results are found consistent with the pre-simulation results,meeting the requirements of the preset indicators.Moreover,based on the study of the temperature drift characteristics of the electronic mobility of pixel devices,two temperature correction methods based on the comparator and ADC are proposed.The reference voltage of the comparator and ADC is generated through the output of image sensor pixel array,so that the above two voltage values have the same temperature variation trend,achieving the aim of temperature correction.(3)The improved current reference is taped out in a 55 nm CMOS process and tested at ten percent fluctuation of power supply voltage.Results show that,the reference output current changes from 15.34μA to 15.76μA,with an overall deviation of 0.5μA when compared to the post-simulation results from 14.91μA to 15.00μA.Under the condition of temperature change,the reference output current changed to 0.4μA,and the temperature drift coefficient is slightly higher than the post-simulation result.But for the three-branch current reference,this work shows a great improvement.(4)After completing the above readout circuit design,temptative exploration of sense-memory-computing integrated image sensor is made.Two data processing circuits for FGMOS-type sense-memory-computing integrated image sensor,namely the absolute value circuit and the current weight accumulation circuit,are designed for the NOR Flash compute-in-memory architecture.The current weight accumulation circuit utilizes a combination of the current mirror and the sampling switch to achieve cyclic accumulation of current.Simulation results show that,the current accumulation error of weight accumulation circuit can be as low as 1.4%.The absolute value circuit compares the input voltage with the 0potential through the comparator and determines whether the input voltage is outpiut according to output result.The inverted proportional inverse circuit is used to convert the negative input voltage into a positive voltage,which is then transmitted through the switch to achieve the design of an absolute value arithmetic circuit with a discrete output function.Simulation results show that,at a power supply voltage of 1.2 V,the absolute value circuit achieves an error of less than 1%,with a zero amplification area of less than 400μV,power consumption of 670μW,and a layout area of 4447μm~2,realizing a high precision arithmetic circuit design.This work provides a useful reference for the comute-in-memory architecture design in the sense-memory-computing integrated image sensor.