Achieving High Efficiency Under Micro-Watt Loads with Switching Buck DC–DC Converters

The present-day and potential benefits of highly integrated miniaturized applications like wireless micro-sensors and biomedical implants in military, space, medical, and commercial markets fuel the demand for self-sustaining micro-electronic systems. Extending operational life to practical levels in such volume-constrained environments is difficult because space limits energy and power. Although switching dc–dc converters have been frequently used in power systems to supply and condition power efficiently, their quiescent and switching losses often render them inefficient at lighter loads, where micro-scale applications reside. Arbitrarily decreasing switching frequency with reductions in load, unfortunately, does not guarantee maximum efficiency because, for example, doing so in discontinuous-conduction mode also increases conduction losses. This paper therefore explores how power losses in switching dc–dc converters relate to load, switching frequency, and other design variables under extreme light loading conditions and ascertains how to manage them to achieve the highest possible efficiency across a micro-watt load. To that end, the paper discusses, analyzes, verifies, and graphically illustrates when and how each of the power-consuming mechanisms dominate efficiency performance. The results show that, after mode-hopping from continuous to discontinuous conduction when the load decreases below half the inductor ripple current, the switching frequency (and quiescent current) should decrease linearly with load at an optimal (derived) rate to balance the losses (and only use just enough quiescent current to sustain the needed bandwidth) and yield maximum efficiency results (e.g., 85–95%) across the entire micropower range. Simulations show that a constant peak-current control converter (as would a hysteretic converter) with the optimal frequency-load ratio achieves over 86% efficiency across a 50–500 A load range.