A Low-Power Front-end Module Dedicated to Wireless Cortical Data Recording

Abstract

This paper reports on the design and the implementation of a fully implantable cortical signals acquisition system. The pre-processing stage, one of the main modules, has been designed in CMOS 0.18um process and sent for fabrication. It includes a low-noise multichannel front-end and a new prototype analog wavelet processor intended for on-line neural signal detection. Ultra low power consumption is achieved for both modules with usage of CMOS weakly inverted transistors. Special attention has been paid on recording quality in the front-end design by using Chopper modulation technique. The proposed front-end achieves an input referred noise of less than 30nV/√Hz and its power consumption is below 20µW per channel. Custom object oriented software has also been implemented for neural data visualisation, storage and analysis, and for system configuration.