Thermoelectric Battery-Charging Voltage Regulators for IoT Microsensors
Loading...
Author(s)
Chang, Tianyu
Advisor(s)
Editor(s)
Collections
Supplementary to:
Permanent Link
Abstract
The objective of this presented Ph.D. research is to study, theorize, develop, test, and assess CMOS energy-harvesting battery-charging voltage regulators that draw ambient thermal energy from tiny CMOS thermoelectric generators to supply Internet-of-Things (IoT) microsensors, replenish batteries, and draw battery assistance when needed. One fundamental challenge is low input voltage. Because when the battery gets completely depleted after a prolonged energy drought, the thermoelectric energy-harvesting charging regulator must wake with only 40–350 mV that the CMOS thermal source outputs. The second difficulty is high efficiency under nanoWatt power budget. Because when the charging regulator is awake, it must draw energy efficiently from a millivolt mega-Ohm CMOS thermal source that only outputs nanoWatts. The third challenge is to control the charging regulator so it can respond quickly and stably under fast load dumps caused by heavily duty-cycled IoT microsensors. To tackle these challenges, this research develops the first-ever theory on the minimum input voltage of a switched-inductor boost both when it is waking and awake. A 1.6-μm CMOS prototype validates the theory and achieves the lowest possible input voltage, which is 1.5× lower compared to state-of-the-art switched-inductor boosts. This research then develops the first-ever theory on the highest-efficiency design of nanoWatt CMOS charging regulators. A 180-nm CMOS prototype validates the theory and can operate with up to 5.6× lower input voltage, 6.3× higher source resistance, 40× lower power, and achieving 3.3× higher efficiency compared to prior arts. Third, this research presents a fast and stable control architecture for nanoWatt charging regulators, together with its related design theory. With the proposed fast control architecture, the 180-nm CMOS prototype can respond to load dumps within 9.6 μs, which is 10–260× faster compared to prior arts. This way, tiny IoT sensors can operate with lower input voltages, higher efficiency, and across longer periods.
Sponsor
Date
2022-10-04
Extent
Resource Type
Text
Resource Subtype
Dissertation