14 November 2018 in IEEE Transactions on Vehicular Technology
In this paper, a bidirectional CLLC converter operating at MHz-level switching frequency is proposed to achieve high power density and high efficiency. An optimized stacked printed circuit board (PCB) winding based planar transformer with adjustable leakage inductance is modeled and designed for the proposed converter. It is enabled using Gallium nitride (GaN) devices to achieve fast switching with low switching losses. A detailed power loss model is established for the efficient thermal design of the converter. A proof-of-concept hardware of 3.3 kW, 400-450 VDC input, 250-420 VDC output CLLC converter with 1 MHz operating frequency is built and tested over a wide range of loading conditions. The power density of the system is 9.22 W/cm $^3$ (151.1 W/in $^3$ ) and the peak efficiency achieves over 97%.
19 October 2018 in IEEE Transactions on Industrial Electronics
Synchronous rectification (SR) is one of the well-known methods to reduce the conduction losses by replacing the power diodes. The control of SR requires the phase information of the device current. In this paper, a novel extended harmonics approximation modeling approach is introduced for CLLC resonant converters to estimate the phase of its secondary side current accurately. Conventionally, the first harmonic approximation (FHA) is used to model any resonant converter; however, FHA works more accurately near the resonant frequency operation. But in case of a set of unmatched LC tank parameters in the primary and secondary side of a CLLC converter, there is no uniquely defined resonant frequency, which reduces the accuracy of the FHA model. Unlike FHA-based approach, our proposed modeling considers the effects of other odd order harmonics present in the square wave voltage waveform towards determining the zero-crossing instant or phase information of the resonant currents. The proposed concept is verified through experimental results obtained at 3.3-kW load condition, and the converter efficiency is improved by 1.8% with the proposed phase tracking technique, compared to FHA modeling approach.