A Multi-Channel 1.52 μVrms Front End with Orthogonal Frequency Chopping for Neural Recording Applications

This paper presents a multi-channel neural recording front end (FE) with orthogonal frequency chopping for neural recording applications by using a standard 0.18 mu m CMOS process. Considering the dc electrode offset (DEO) and the limited input impedance, dc-servo loop with duty cycle resistor and positive feedback loop with T-bridge capacitors are employed, which achieves high pass corner frequency of 0.1 Hz and input impedance of 850 M Omega, respectively. The proposed SAR-ADC with dynamic element matching (DEM) algorithm enhances the FE linearity. The total input referred noise (IRN) is 1.52 V-rms with 4 times oversampling, which is the lowest compared to prior-art of works, while exhibiting comparable power consumption.