⚡ 本页包含 AI 生成的分析内容,仅供参考
本文提出了一种采用DR折叠技术的功率高效神经记录架构,实现了0.45V超低电压操作和每通道亚微瓦级功耗,解决了传统神经记录系统中噪声效率与功耗之间的权衡问题。该芯片集成了100个记录通道,在0.18μm CMOS工艺中制造。
Conventional neural-recording systems face limitations in simultaneously achieving a good NEF and low power consumption [1-4]. This is because the input amplifier current consumption is dictated by an input-referred noise requirement that determines the system sensitivity, while the supply voltage is determined by a DR requirement at the analog recording chain output that limits the maximum achievable resolution of the A-to-D conversion. In this paper, a power-efficient neural-recording architecture using a DR-folding technique is presented to enable low-voltage operation without compromising the DR performance. The proposed architecture can operate with only half of the typically required supply voltage, which results in about 50% power reduction.
Dong Han1,2, Yuanjin Zheng1, Ramamoorthy Rajkumar3, Gavin Dawe3,
Minkyu Je2,3 Nanyang Technological University, Singapore, Singapore Institute of Microelectronics, Singapore, Singapore 3 National University of Singapore, Singapore, Singapore 1 2