700mV low power low noise implantable neural recording system design

Abstract

This dissertation presents the work for design and implementation of a low power, low noise neural recording system consisting of Bandpass Amplifier and Pipelined Analog to Digital Converter (ADC) for recording neural signal activities. A low power, low noise two stage neural amplifier for use in an intelligent Radio-Frequency Identification (RFID) based on folded cascode Operational Transconductance Amplifier (OTA) is utilized to amplify the neural signals. The optimization of the number of amplifier stages is discussed to achieve the minimum power and area consumption. The amplifier power supply is 0.7V. The midband gain of amplifier is 58.4dB with a 3dB bandwidth from 0.71 to 8.26 kHz. Measured input-referred noise and total power consumption are 20.7 μVrms and 1.90 μW respectively. The measured result shows that the optimizing the number of stages can achieve lower power consumption and demonstrates the neural amplifier's suitability for instu neutral activity recording. The advantage of power consumption of Pipelined ADC over Successive Approximation Register (SAR) ADC and Delta-Sigma ADC is discussed. An 8 bit fully differential (FD) Pipeline ADC for use in a smart RFID is presented in this dissertation. The Multiplying Digital to Analog Converter (MDAC) utilizes a novel offset cancellation technique robust to device leakage to reduce the input drift voltage. Simulation results of static and dynamic performance show this low power Pipeline ADC is suitable for multi-channel neural recording applications. The performance of all proposed building blocks is verified through test chips fabricated in IBM 180nm CMOS process. Both bench-top and real animal test results demonstrate the system's capability of recording neural signals for neural spike detection.