An Inverter-Based Amplifier Structure for Neural Signal Recording with an NEF of 1.28 and Area-per-Channel of 0.06mm2

摘要

The Noise-Efficiency-Factor (NEF) and area efficiency of the neural signal amplifier has been continuously improved over the years to achieve prolonged recording lifetime and increased channel count. By far, the most energy efficient neural amplifiers are configured based on fully-differential inverter structures yielding double input transconductance and reduced NEF with a theoretical limit of $sqrt{2}$ [1– 5]. Recently, a stacking technique has been superimposed on conventional inverter-based structures to further push the NEF below 1 by reusing the bias currents through stacked input stages [1], [2]. Unfortunately, the output dynamic range, maximum allowable bias current, and thermal noise floor in a stacked structure are all compromised due to smaller voltage headroom. Moreover, large isolation capacitors are required at both input and output nodes of each stacked stage for DC voltage separation, so area expands significantly as more stages are stacked [1].