Low power programmable front-end for a multichannel neural recording interface

We propose in this paper a programmable gain preamplification front-end for a fully implantable multichannel data acquisition system (IMDAS) that is dedicated for chronic neural signal recording. This application calls for very low power and low voltage circuit techniques. To satisfy these constraints, the preamplifier is designed in 0.18mum CMOS technology and all employed transistors operate in weak inversion. To maximize the dynamic range of the recorded signals, the preamplifier's gain is set by a 4-bit digital-to-analog converter (DAC). This DAC is used to tune the bias currents of a variable transconductor cell in order to vary its DC gain. The simulation of the whole proposed module gives a maximum power consumption of 530nW at a supply voltage of 0.9V. The circuit provides a maximum gain of 47dB, has a cutoff frequency of 2.65kHz and presents an input-referred noise of 7.65muVrms which is sufficiently low to meet the required precision. In addition, the phase margin is higher than 50degrees on the entire gain range.