A front-end circuit with active spike and LFP separation via a switched capacitor filter structure for neural recording applications

A fully differential front-end circuit for a neural recording system with spike and Local Field Potential (LFP) separation by means of switched capacitor (SC) filters is reported in this paper. In many neural recording chips the transmitter is the most power hungry element and defines the number of channels on a chip. With a separation of the neural signal, it is possible to reduce the data rate and so the power consumption of the transmitter without any reduction of resolution in the quantized signal. The system consists of a low-noise-amplifier (LNA), an anti-aliasing-filter with tunable gain and a corner frequency at 50kHz to enable SC circuits with a clock frequency of 100kHz. Two SC biquad filters with tunable gain and tunable corner frequency are implemented to split the spikes and the LFP from each other and to amplify the sub signals to the full swing. Both channels provide an input referred noise of 3.8μVrms in the passband and the whole front-end circuit has a power dissipation of 52μW. A prototype is fabricated in a standard 0.18μm CMOS process and tested successfully.