Supporting the below United Nations Sustainable Development Goals:支持以下聯合國可持續發展目標：支持以下联合国可持续发展目标：
Examination Committee
Prof Philip MOK, ECE/HKUST (Chairperson)
Prof Chi Ying TSUI, ECE/HKUST (Thesis Supervisor)
Prof Wing Hung KI, ECE/HKUST (Thesis Co-supervisor)
Prof George YUAN, ECE/HKUST
Abstract
Energy Harvesting (EH) is an energy provision technology that collects small-amount but everlasting ambient energy to power up micro-Watt system such as Wireless Sensor Node (WSN). Usually these nodal systems are required to have tiny volume and near-perpetual operation period. As a result, bulky disposable batteries are not desirable and thus efficient energy harvesting and managing algorithms are required for the operation of such systems.
In this work we target energy harvesting applications for indoor-scenario where indoor-lighting, Wi-Fi radio wave and thermal gradient are the generally-available sources. Comparing with the other two, indoor-lighting is more important as it not only provides up to hundred-mW level of power, but also commonly exists in normal building environment. Thus we mainly focus on designing solar-energy harvesting systems for indoor lighting condition.
A Multi-Input-Multi-Output (MIMO) Buck-boost converter is developed to handle hybrid energy sources and multiple loadings. Pulse frequency modulation control scheme is developed to regulate the input and output nodes in a time-multiplexing manner. As the amount of the harvested energy is limited, efficient energy utilization is the highest priority. In this work, we perform efficiency enhancement at both light load and heavy load condition. Burst clock mode and duty-cycled band-gap are designed to reduce the power overhead of the controller and quiescent analog circuitry to nW-level during light load condition. During heavy load condition, the power loss is optimized by a novel switching transistor sizing to balance the switching loss and conduction loss for different loadings and a reconfigurable switch array is designed to achieve the optimal efficiency at different states. Simulation results show that a peak transfer efficiency of 95% at 30mW load with 20mV output voltage ripple is achieved.