Low-Power Neural Recording Amplifier Design Based on Advanced Noise Modeling

Abstract

Neural recording is an indispensable function required for the brain machine interface (BMI) and neuroscience research. In order to obtain high-quality neural signals with minimal noise, a low-noise performance needs to be provided by a front-end amplifier used for neural recording. At the same time, a lowpower operation must be achieved to avoid tissue damage. This paper presents the neural recording amplifier design optimized to achieve highest possible signal-to-noise ratio (SNR) based on advanced noise modeling which takes into account the effect of a finite source impedance determined by the characteristics of the tissue-electrode interface. The research on the finite source impedance was performed through animal experiments by using rats. The commercial products of NeuroNexus Technologies were implanted over the sensorimotor cortex. Then, the impedances were daily checked for 21 days by using TDT (Tucker-Davis Technologies) system. The front-end amplifier was designed for neural recording over the frequency range from 1 Hz to 10 kHz. The amplifier was implemented with TSMC 0.18-μm CMOS process. We also introduce the systematic procedure for optimizing the neural recording amplifier under the constraint of given power budget, while considering various electrode site areas and the change of the source impedance value over time. ⓒ 2016 DGIST