Ultra-low-power Neural Recording Microsystem for Implantable Brain Machine Interface

Abstract

We propose an implantable CMOS micro system for detection of neural spike signals from complex brain neural potentials which achieves the characteristics of ultra low-power and high-precision. The neural recording micro system consists of a low-noise bioamplifier, a neural spike detector based on nonlinear energy operator (NEO) and a precision hysteresis comparator. The DC offset in the bioamplifier is rejected by introducing a new active feedback configuration instead of the large capacitors, the NEO algorithm is implemented through simple analog circuits operating in sub-threshold region, and the hysteresis comparator is added to determine the location of neural spike. The current consumption of the recording system is 3.02 μA with 3.3 V supply. The proposed system has been implemented in 0.35 μm CMOS process and precisely detects neural spike signals from extra cellular recording.