Date of Award
Doctor of Philosophy (PhD)
This thesis studies high frequency dc-dc power converters for automotive LED driver applications. A high-frequency zero voltage switching (ZVS) integrated-magnetics Ćuk converter is well-suited for automotive LED-driver applications. In this converter, the input and output filter inductors and the transformer are realized on a single magnetic structure, resulting in very low input and output current ripples, thus reducing electromagnetic interference (EMI) and minimizing the required input and output filter capacitances. Active-clamp snubbers are used to mitigate the effects of the transformer leakage inductance. A prototype 1.8~MHz Ćuk converter with integrated magnetics is designed, built and tested. The prototype converter supplies 0.5 A output current to a string of 1-10 LEDs, and achieves 89.6% peak power-stage efficiency.
The use of active-clamp snubbers introduces additional conduction and gate-drive losses. This thesis introduces a planar integrated magnetics structure that is designed to minimize the transformer leakage inductance and therefore eliminates the need for snubbers. The planar integrated magnetics structure is optimized using 3D finite element modeling (FEM) tools. Two 1.8 MHz-to-2.4 MHz Ćuk converter prototypes are constructed: one using Silicon MOSFETs and the other using GaN transistors. The former achieves a peak efficiency of 92.9%, while the latter achieves a peak efficiency of 93.5% and a wider ZVS range. Both prototypes maintain greater than 90% efficiency across their wide output voltage range.
A new control architecture for the ZVS integrated magnetics Ćuk converter is presented. A Spice-based averaged circuit model is employed to model the converter dynamics. The duty-cycle-to-output-inductor-current transfer function is obtained and an integral compensator is designed to precisely regulate the output inductor current (LED current) over the entire output voltage range of the converter (3 V-to-50 V). To achieve high-resolution PWM dimming, new turn-off and turn-on strategies are proposed. The proposed turn-off strategy reduces the fall time of the LED current by up to 83%, and the turn-on strategy reduces the rise time by up to 43%. The controller is implemented digitally and experimental results are presented.
This work also investigates resonant dc-dc converters as an alternative approach for automotive LED driver applications. The LLC resonant dc-dc converter is studied and is found that this converter suffers from high circulating currents, when designed to operate over a wide input and output voltage range. An LC3L resonant dc-dc converter is proposed. The converter exhibits minimal circulating currents. Furthermore, it is shown that when appropriately designed, the converter behaves like a current source, with its output current being independent of the output voltage. This property is particularly favorable for automotive LED driver applications. A 10 MHz LC3L resonant dc-dc converter is designed and simulated. This converter is predicted to achieve greater than 86% efficiency, and be 60% smaller in size compared to the planar integrated magnetics Ćuk converter.
Further increase in the switching frequency of automotive LED drivers demands exploring new design techniques and the use of high performance semiconductor devices. This thesis presents high efficiency dc-dc converters operating at very high frequencies using custom monolithic GaN-based half-bridge power stages with integrated gate drivers. A new gate driver circuitry is introduced, which enables efficient converter operation at very high switching frequencies, while maintaining very low quiescent power consumption. While using only n-type transistors in the GaN process, the proposed gate driver emulates complementary operation commonly employed in CMOS processes. A family of monolithic GaN chips is designed to operate over switching frequencies in the range of 20-400 MHz,
Sepahvand, Alihossein, "High Frequency DC-DC Power Conversion for Automotive LED Driver Applications" (2018). Electrical, Computer & Energy Engineering Graduate Theses & Dissertations. 181.