UCLA

Neural signal recording systems are vital for understanding the working of the brain. With the growing number of recording channels, it is imperative that their power consumption be kept as low as possible. The front-end amplifier used in these recording systems consumes a significant portion of the system power. This work presents a fundamentally different approach in the design of amplifiers by using concepts from signal processing to improve the achievable performance. Frequency division multiplexing is used to share the most power-hungry blocks of the amplifier, while maintaining all performance characteristics. Simulation results using a 65- nm CMOS technology show that the power consumed is lower than the theoretical limit for a single-channel amplifier by a factor of 2. The input referred noise over a bandwidth of 0.2 Hz to 6 kHz is 4 �Vrms, while burning 500 nA per channel from a 1.2 V supply. The power consumed is a factor of 3.6 lower than the best designs published so far. This shows that the design presented in this work, when fabricated, could be the best design to date for a neural amplifier.