A tunable biquad switched-capacitor amplifier-filter for neural recording.

Area of Science:

• Neuroengineering
• Integrated Circuit Design
• Biomedical Electronics

Background:

• Local field potentials (LFPs) are increasingly utilized as input signals for brain-machine interfaces (BMIs).
• Existing integrated circuits often lack the capability for simultaneous amplification of both neural spikes and LFPs.
• A need exists for efficient, high-performance hardware for multimodal neural signal acquisition.

Purpose of the Study:

• To develop and characterize a two-stage complementary metal-oxide semiconductor (CMOS) amplifier-filter for simultaneous recording of extracellular unit spikes and LFPs.
• To implement frequency tuning and 1/f noise reduction techniques for enhanced signal quality.
• To create a compact and power-efficient integrated circuit for neural recording applications.

Main Methods:

• A 0.18-μm CMOS technology was used to implement a two-stage amplifier-filter architecture.
• A switched-capacitor technique was employed for frequency tuning and reduction of 1/f noise.
• The filter bandwidth was designed to be reconfigurable by adjusting the sampling clock frequency.

Main Results:

• The prototype amplifier achieved gains of 19.1 dB (low-pass only) and 37.5 dB (cascaded filter).
• At a 100-kHz sampling frequency, the equivalent input noise spectral density was measured at 38.8 nV/√Hz.
• The circuit demonstrated low power consumption (69 μW at 1.6-V supply) and a small footprint (44 × 148 μm²).

Conclusions:

• The developed CMOS amplifier-filter enables simultaneous, multimodal recording of neural spikes and LFPs.
• The switched-capacitor technique effectively addresses frequency tuning and noise reduction requirements.
• This integrated circuit is a promising solution for next-generation brain-machine interfaces demanding efficient neural signal acquisition.