← 返回 JSSC 论文列表JSSC 2024第8期Digital CircuitsNeural Network Accelerator
A 64-Channel Inverter-Based Neural Signal Recording Amplifier With a Novel Diffe
提出一种基于逆变器的64通道低功耗低噪声神经信号记录放大器，采用新型差分OTA设计。
NEF 0.84, 每通道平均电流422nA, 每通道面积0.044mm²
逆变器神经信号记录低功耗低噪声多通道
▸创新点1：新型差分OTA设计（电路创新） - 采用不对称分支结构，反相分支比非反相分支并联更多逆变器，优化电流比例使NEF逼近理想值√2/2，显著降低噪声并提高能效。
▸创新点2：虚拟低阻抗轨设计（电路创新） - 在差分OTA中设计极低阻抗虚拟电源轨，有效抑制通道间串扰，实测64通道下NEF低至0.84，通道电流仅422nA。
▸创新点3：多通道集成优化（系统创新） - 通过通道数增加实现面积/功耗协同优化，64通道时单通道面积仅0.044mm²，NEF和电流随通道数增加呈超线性下降趋势。
▸创新点4：可扩展架构创新（方法创新） - 提出通道数依赖的性能优化模型，4/16/64通道实测验证架构可扩展性，功率-面积效率比传统方案提升3倍以上。
Abstract
This article presents an inverter-based multichannel
low-power low-noise neural signal recording amplifier with a
novel differential-like operational transconductance amplifier
(OTA). The differential-like OTA consists of two asymmetric
branches. The inverting branch used for multichannel inputs has
more inverters in parallel than that of the noninverting branch
used for reference. Two virtual rails with very low impedance
are designed in the differential-like OTA to effectively reduce the
noise