摘要

This paper explores sizing and topology reconfiguration strategies for charging and discharging multiple supercapacitors as energy storage in sub-Watt-scale energy harvesters. To maximize energy efficiency for storage (eta(Estorage)), total leakage power is kept low by selecting the supercapacitors to charge sequentially, alternately, or in series based on their voltages. To maximize energy efficiency for driving load (eta(Edriving)), residual energy is minimized by switching to series composition of the supercapacitors in order to raise the voltage above the minimum input voltage of the load-side dc-dc converter. Due to the nonlinear, stateful, load-dependent behavior of such harvesters, parameter sweeping is used for system-level optimization. A topology-reconfigurable structure consisting of two larger, symmetric "reservoir" supercapacitors and one voltage-raising "bootcap" proves to be efficient and practical when considering combinations of symmetric and asymmetric capacitance values. Experimental results on an actual implementation show that for charging, series topology is best when input power is high due to lower leakage, and individual topology is best when input power is low due to lower voltage; for discharging, series topology is effective in reducing residual energy while individual topology with alternating discharging effectively minimizes leakage.