王超

个人信息Personal Information

研究员(自然科学)   博士生导师   硕士生导师  

性别:男

在职信息:在职

所在单位:光学与电子信息学院

学历:研究生(博士)毕业

学位:工学博士学位

毕业院校:南洋理工大学

学科:微电子学与固体电子学
电路与系统

论文成果

当前位置: 中文主页 >> 科学研究 >> 论文成果

A 28-nm 0.23fJ/OP, 0.41 LSB Average Absolute Error Reconfigurable Merged-in-ADC Computing Circuit with Novel Parasitics-induced Error Elimination

点击次数:

第一作者:W. Zhu, Y. Zhao

通讯作者:C. Wang*

合写作者:L. Huang, Z. Zhang, Y. Yang, Y. Zheng

发表刊物:2025 IEEE Interregional NEWCAS Conference (NEWCAS 2025)

收录刊物:EI

学科门类:工学

一级学科:电子科学与技术

文献类型:C

摘要:High-energy-efficiency and high-linearity analog and mixed-signal domain computing is one of the cutting-edge research trends in intelligent edge sensors. This paper proposes a merged-in-ADC reconfigurable three-mode switched-capacitor computing circuit to achieve a good linearity by a novel top/bottom-plate alternating based input and charge sharing method. Multiplication is implemented by a new 4-phase switching scheme to eliminate the parasitic capacitance induced additive error, while also effectively reducing latency and power consumption. The prototype chip at 28-nm CMOS demonstrates a high energy efficiency of 0.23 fJ/OP with an accuracy of 0.41 LSB average absolute error, outperforming the state-of-the-art designs.