High Voltage Synchronous Rectifier Design Considerations
The advent of wide band-gap semiconductors in power electronics has led to the scope of efficient power conversion being pushed further than ever before. This development has allowed for systems to operate at higher and higher voltages than previously achieved. One area of consideration during this high voltage transition is the synchronous rectifier, which is traditionally designed as an afterthought. Prior research in synchronous rectifiers have been limited to low voltage, high current converters. There is practically no research in high voltage synchronous rectification. Therefore, this dissertation focuses on discovering the unknown nuances behind high voltage synchronous rectifier design, and ultimately developing a practical, scalable solution. There are three main issues that must be addressed when designing a high voltage synchronous rectifier: (1) high voltage sensing; (2) light load effects; (3) accuracy.
The first hurdle to designing a high voltage SR system is the high voltage itself. Traditional methods of synchronous rectification (SR) attempt to directly sense voltage or current, which is not possible with high voltage. Therefore, a solution must be designed to limit the voltage seen by the sensing mechanism without sacrificing accuracy. In this dissertation, a novel blocking solution is proposed, analyzed, and tested to over 1-kV. The solution is practical enough to be implemented on practically any commercial drain-source SR controller.
The second hurdle is the light load effect of the SR system on the converter. A large amount of high voltage systems utilize a LLC-based DC transformers (DCX) to provide an efficient means of energy conversion. The LLC-DCX's attractive attributes of soft-switching and high efficiency allure many architects to combine it with an SR system. However, direct implementation of SR on a LLC-DCX will result in a variety of light load oscillation issues, since the rectifier circuitry can excite the resonant tank through a false load transient phenomena. A universal limiting solution is proposed and analyzed, and is validated with a commercial SR controller.
The final hurdle is in optimizing the SR system itself. There is an inherent flaw with drain-source sensing, namely parasitic inductance in the drain-source sense loop. This parasitic inductance causes an error in the sensed voltage, resulting in early SR turn-off and increased losses through the parallel diode. The parasitic will always be present in the circuit, and current solutions are too complex to be implemented. Two solutions are proposed depending on the rectifier architecture: (1) multilevel gate driving for single switch rectifiers; (2) sequential parallel switching for parallel switch rectifiers.
In summary, this dissertation focuses on developing a practical and reliable high voltage SR solution for LLC-DCX converters. Three main issues are addressed: (1) high voltage sensing; (2) light load effects; (3) accuracy. Novel solutions are proposed for all three issues, and validated with commercial controllers.