An Adaptive On-time Multiple-output Boost Converter with Maximum Power Point Tracking, Self-start-up Circuit and Winner-take-all High Voltage Selector for Thermoelectric Energy Harvesting System

• Main Authors: Yang-Jing Huang, 黃揚景
• Other Authors: Sheng-Yu Peng
• Format: Others
• Language: zh-TW
• Published: 2017
• Online Access: http://ndltd.ncl.edu.tw/handle/39nfuc

碩士 === 國立臺灣科技大學 === 電機工程系 === 106 === A power-efficient boost converter for a thermoelectric energy harvesting system (EHS) is proposed in this thesis. To improve overall system power efficiency, a multiple output architecture is adopted. To provide suitable cut-off voltage to power transistors in different conditions of output voltage levels, a novel high-voltage-selector based on a winner-take-all circuit is proposed. Besides, power transistor dimensions are optimized to balance the conduction and the switching losses for minimizing the loss of the power stage. Floatinggate technologies are employed in voltage comparators and voltage detectors to avoid the power consumption of a conventional bandgap reference circuit and to provide reconfigurability for output voltage levels. Floating-gate transistors are also employed in a novel cold-start-up circuit to reduce the threshold voltage of transistors in a charge pump circuit so that the start-up voltage can be as low as 380 mV. The cold-start-up circuit exploits the same inductor as the main boost converter without extra external component. To maximize the energy harvesting efficiency, the boost converter adopts an adaptive on-time and peak current control scheme and a fractional open-circuit voltage approach to achieve maximum power point tracking. To reduce the tracking time, a binary and incremental search algorithm is exploited with small input voltage ripples. A prototyped chip is designed and fabricated in a 0:35um CMOS process. The measurement results show that the cold-start-up circuit can start-up the system when input voltage equals to 380 mV and the start-up time is about 1:4 seconds. Furthermore, the simulation result shows the peak efficiency is 95:85% when the input power equals to 988 W.