Design and Analysis of a General Purpose Biosignal Acquisition System

Abstract

Electrophysiological signals from biological sources have been measured and used in modern medicine as tools to address abnormal or unusual behaviors of muscles and organs. Their usage is of an increasing importance in modern medicine, providing a vital aid to the understanding of the body’s activity by mapping the activation of tissue or neurons to electrical signals. Nowadays, there are two clear approaches to the monitoring of these signals. In
a clinical environment, where the signals have to be acquired accurately and reliably, the patient is attached to a machine, via electrodes with long wires, greatly reducing the patient’s ability to move. The devices used for these recordings are big, bulky, and use the mains electricity as the power source. On the other hand, wearable devices focus on measuring non life-threatening parameters, such as heart rate, temperature, oxygen saturation, electrocardiogram (ECG), and in some cases even electroencephalogram (EEG), by reducing both the power and the accuracy used by the signal conditioning circuits, such as amplifiers and analog-to-digital converters.
This thesis aims to bring closer both platforms by proposing an ultra-low power
signal acquisition platform for remote biosensing. The platform senses and accommodates the electrophysiological signals using voltage-to-time conversion and time- mode signal processing which will be later transmitted wirelessly using a low-power communication scheme. A chain of analog-to-time converters combined with a low-power single-pulse harmonic modulation transmitter have been used to implement the platform in an application specific integrated circuit.
The integrated circuit has been designed in a 40 nm TSMC process, using 0.017 mm2 of the chip area. The resulting post-layout simulated energy consumption when powered from a supply voltage of 500 mV is 24.75pJ, with a power consumption of 6.336 nW at a sampling and communication rate of 256 Hz. Dynamic characterization of the circuit predicts a signal to noise and distortion ratio (SINAD) of 35.80 dB for a 1.915 kHz bandwidth and an oversampling ratio of 256. This work achieves unprecedentedly low power consumption compared to current solutions, showcasing the potential of time-
mode signal processing for energy efficient signal acquisition platforms.