A Multi-Segmented LED Driver with LUT-Based Control using Background Calibration

Abstract

Light-emitting diode (LED) lighting is becoming a major illumination method with supports from global energy institutions such as department of energy (DOE) in the United States. It is because LED replacement can save a lot of global energy consumption due to its superior efficacy over conventional light sources. However, for a rapid adoption of LED lighting, lower cost and reliable driver is required. This thesis presents an AC-powered, boost-converter-based multi-segmented LED driver that maintains a high-power factor (PF) and accurate input current level, using a novel look-up table (LUT)-based digital control with a simple predictive duty-cycle control scheme and background calibration. The presented multi-segmented LED driver aims to reduce the costs of high-voltage capacitors and high-inductance inductor by segmenting the load LEDs into multiple strings and switching the number of LED strings which are connected in series for line regulation and power-factor correction. However, the increased number of power switches calls for a sophisticated control scheme that responds quickly to the switching mode changes and adapts properly to the operating condition changes. The presented LED driver addresses this by employing a LUT that can provide a pre-programmed, arbitrarily fast response to the periodic 220-V AC input, supplemented by a periodic background calibration and optimal switching mode selection to track any unexpected changes in the input, load, and environment conditions. A prototype 220-V AC LED driver with 6-segment LED strings demonstrates a 98.4% power factor (PF), 91% conversion efficiency, 26 mArms input current error, and 67% reduction of discrete component costs while delivering 38.3-W to the load.