An activity-dependent brain microstimulation SoC with integrated 23nV/rtHz neural recording front-end and 750nW spike discrimination processor

This paper describes an activity-dependent intracortical microstimulation system-on-chip (SoC) that can convert extracellular neural signals recorded from one brain region to electrical stimuli delivered to another brain region in real-time. The system integrates an analog recording front-end with input noise voltage of 2.6μV
_rms
in 10.5kHz bandwidth, 5.5μW 10b SAR ADC, 750nW digital spike discrimination processor, and a charge-balanced constant-current stimulating back-end that can deliver up to 94.5μA with 6b resolution when triggered by neural activity. Electrical performance characterization and biological measurement results from a prototype chip fabricated in 0.35μm 2P/4M CMOS are presented.