Pseudo-Differential Neural-Recording Front-End Design Using High-Pass Cutoff Frequency Programmable Single-Ended Pixel

Abstract

This paper presents a pseudo-differential neural-recording front-end design for a high-density recording channel. For differential signal processing, the proposed front-end comprises the pseudo-differential stage (PDS) and fully differential stage (FDS). The PDS consists of the AC-coupled single-ended pixels. Among the multiple pixels in the PDS, two pixels are selected and connected to the input of the FDS for differential processing. By employing a narrow-band buffer in the feedback loop of the PDS and FDS, the front-end can flexibly control high- and low-pass cutoff frequencies with low power consumption. For implementing the area- and current-efficient pixel without gain error, a single-stage amplifier with a cur-rent-reuse input pair and a cascode output stage is employed for each pixel. The proposed design, implemented in a 180-nm CMOS process, achieves an input-referred noise of 3.56 μVrms, 5.23 μVrms, and 7.01 μVrms over low-pass, wideband, and high-pass recording modes, respectively.