A 0.6V 1.07 μW/Channel neural interface IC using level-shifted feedback

This paper presents a 16-channel low-power neural interface IC for in-vivo neural recording applications. Each recording channel consists of two AC-coupled close-loop amplifiers: a low noise amplifier (LNA) amplifies the weak neural signal by 26 dB, and a programmable gain amplifier (PGA) provides an additional gain of 8 dB–26 dB. The LNA employs a current-reusing telescopic topology to reduce noise for achieving a better noise efficiency factor (NEF). To further reduce the power consumption, we propose a level-shifted feedback (LSFB) technique to lower the operational supply voltage. Theoretic analysis shows that the proposed amplifier with LSFB halves the minimum required supply voltage compared to conventional AC-coupled close-loop counterparts without degenerating the noise performance. The prototype chip is fabricated in 65 nm CMOS process. Operating under a single 0.6 V supply, each recording channel consumes 1.07 μW. The input-referred noise integrating from 10 Hz to 10 kHz is 5.18 μV, and the NEF/PEF is 2.94/5.19.