一种植入式神经元记录系统信号处理电路

A Conditioning Circuit for Implantable Neural Recording Systems

提出了一种植入式神经元记录系统信号处理电路,由一个带通放大器和流水线模数转换器(ADC)构成。带通放大器采用具有共模反馈的跨导运算放大器(OTA)来放大神经元信号,采用最优的2级放大器级数,减小了功耗和面积。流水线ADC采用全差分结构和CDS技术,减小了非线性失真,其中,MDAC采用一种新的消除技术,降低了输入漂移电压。采用0.18μm CMOS工艺进行设计与仿真,仿真结果表明,带通放大器的带宽为0.71Hz^8.26kHz,中频增益为58.4dB,输入参考噪声(rms)为20.7μV,功耗为1.90μW;采样频率为16kHz时,ADC的有效位数为8位。经动物实验验证,该神经记录系统能够用于神经元峰电位的检测。

A conditioning circuit for implantable neural recording systems was proposed. It consisted of bandpass amplifiers and pipelined analog to digital converters （ADC）. The bandpass amplifier based on operational transconductance amplifier （OTA） with novel common mode feedback （CMFB） was utilized to amplify the neural signals. The minimum power and area consumption could be achieved by two stage amplifiers. Full differential （FD） structure and correlated double sampling （CDS） used in pipelined ADC could reduce the nonlinear distortion. The multiplying digital to analog converter （MDAC） utilized a novel cancellation technique to alleviate the input drift voltage. The performances of all proposed circuits were verified through test chips fabricated in 0.18μm CMOS process. The simulated results showed that the midband gain of amplifier was 58.4 dB with a 3 dB bandwidth from 0.71 to 8.26 kHz. Measured input referred noise（rms） and power consumption were 20.7 μV and 1.90 μW respectively. The ADC＇s effective number of bits （ENOB） was 8 bit at 16 kHz. The real animal test results also demonstrated that the system was capable of recording neural signals for neural spike detection.