Implantable Biomedical Recording System

Abstract

Implanting a tiny biomedical integrated circuit transceiver inside the human brain is quite challenging, especially when it is powered wirelessly to transmit the neural signal outside the brain. The power, area and noise constraints are quite difficult to achieve especially when considering the biomedical safety constraints. This thesis investigates the design of a low power, low noise and small size neural channel to support high density neural implant system. Various neural recording systems were studied prior to the beginning of this work. A classification of the recording systems is presented along with the advantages and disadvantages of each. The main contribution of this thesis is the design of a very small size neural channel that can be integrated with high density. A novel technique in autozeroing is presented to reduce the overall channel size. In addition, a digital system is presented to recover the neural signals that are affected by the autozeroing technique. To lower the noise of the neural amplifier, a chopper stabilized technique is considered. However, this technique causes the degradation of the input impedance. Accordingly, this work proposed a novel digitally calibrated impedance boosting circuit that is robust against process variations. The achieved neural recording channel area is 0.004 mm2; each consuming 0.7 μW from a 1V DC supply. The proposed system integrates 86,956 neural recording channels on a 2020 mm2. All simulation results were done by Cadence Virtuoso and Mentor Graphics Eldo-RF using UMC 130 nm CMOS technology.