With the development of power electronic technology, the thermal characteristics of high-power insulated gate bipolar transistor (IGBT) module are vital information for thermal management of power electronic equipment, reliability analysis, and thermal design of power electronic system. However, the present commonly used lumped thermal network model based on thermal resistance and heat capacity still has some limitations in accurately predicting the junction temperature of IGBT modules, significantly once considering the degradation of base solder in high-power IGBT modules. In this paper, the thermal behavior of high-power IGBT chips under different base solder degradation is studied by the finite element method (FEM). The results show that the degradation of the base solder has effects on the thermal impedance between the junction and the case, which should be fastidiously thought about within the thermal network model modeling. As a result, a dynamic three-dimensional thermal network model that takes into account the thermal interaction between IGBT chips, as well as the degradation of the base solder, is proposed in this paper. The proposed thermal network model can estimate the junction temperature of a high-power IGBT module correctly and fast based on the real degradation of the base solder. Finally, finite element simulation and experimental findings are used to validate the proposed dynamic three-dimensional thermal network model. The results show that the dynamic three-dimensional thermal network model has comparable accuracy to finite element simulation and test results. At the same time, the dynamic three-dimensional thermal network model responds faster than the finite element simulation.

This paper proposes a new auxiliary power supply (APS), which is used to feed the controllers, drivers, and sensors of the sub-modules for medium-/high-voltage and high-power solid state transformer (SST). In medium-/high-voltage and high-power applications, the APS is usually connected to the dc bus of the sub-module to obtain electric energy. The voltage of the sub-module's dc bus is often as high as several thousand volts, so that the APS must have the ability to handle such high input voltage. An input-series capacitor structure, which makes the voltage stress of the power semiconductor devices reduced, is adopted in the proposed APS. Therefore, the high input voltage can be handled by the low voltage power semiconductor devices. Furthermore, the input voltage sharing (IVS) can be achieved without any special control strategy, which makes the control system simplified and the stability improved. Meanwhile, zero-current switching (ZCS) is guaranteed for all the switches, so that the power density and efficiency can be improved. Finally, the simulation model and prototype are built, by which the feasibility and effectiveness of the proposed APS are verified.

Power electronic traction transformer (PETT) is a module-cascaded converter with high-frequency-link technology to realize high power conversion. A typical structure of PETT contains a single-phase cascaded H-bridge rectifier in the first stage and several dual active full-bridge DC-DC converters in the second stage. For PETT adopting such complex topological structure, the control strategy also becomes complicated and should be carefully designed. In this paper, a power-linked predictive control strategy for PETT is proposed based on the predictive controls of H-bridge rectifier and dual active full-bridge DC-DC converter. A power link is brought into the control strategy to connect H-bridge rectifier with DC-DC converter, through which the input power of PETT can match the output power. Moreover, the voltage balance control and power balance control are also combined with predictive control by power-linked predictive control strategy. Using such strategy, the fluctuation of H-bridge rectifier DC voltage is suppressed, the circumfluence among modules is reduced, and the dynamic and steady-state performances are both improved compared with traditional strategy. Simulation and experimentation platforms with both resistance load and motor load are built to show the effectiveness and correctness of power-linked predictive control strategy.

Insulated gate bipolar transistor (IGBT) is the most important power electronic component in the traction converters of trains. During the operation of the IGBT, the solder layer is continuously subjected to alternating stress, which eventually leads to the fatigue failure of the solder layer. Therefore, an accurate assessment of the lifetime of the IGBT module solder layer is of great importance for the safe and reliable operation of the trains. Firstly, the micromechanism of the plastic deformation of the base solder layer of the IGBT module is introduced, as well as the lifetime model based on the plastic strain energy density. Then the total plastic strain energy density of the base solder layer is determined by conducting the accelerated aging tests on the IGBT module and combining the finite element simulation results. Meanwhile, the relationship between different aging degrees of the base solder layer of the IGBT module and the plastic strain energy density is studied by finite element simulation. Then, the lifetime of the base solder layer at different aging stages is calculated based on the above results so that the total lifetime of the IGBT module is obtained. Finally, the predicted lifetime of the IGBT module by the proposed method is compared with the lifetime obtained by the accelerated aging test and compared with the traditional lifetime prediction method based on the Coffin-Manson-Arrhenius model. The comparison results verify the accuracy of the proposed lifetime prediction method in this paper.

LLC resonant converters possess excellent soft-switching characteristics and are extensively applied in the dc–dc converter field. To achieve closed-loop control of the LLC resonant converter, it is crucial to establish an accurate small-signal model. Traditional modeling methods suffer from the problems of decreasing accuracy away from the resonant frequency or high model complexity in practical applications. This article proposes a small-signal modeling method for an LLC resonant converter based on time-domain correction, which provides practical analytical small-signal models below, at, and above the resonant frequency. The proposed method strikes a balance between accuracy and complexity. Equivalent models of components and the average equivalent circuit of the LLC converter are established using simplified time-domain analysis to describe the average characteristics of the LLC converter. Small-signal perturbations are applied to the average equivalent circuit to derive an analytical second-order average small-signal model that can guide closed-loop regulator design and stability analysis of the LLC converter. Finally, the correctness of the proposed small-signal modeling method is verified through PLECS ac sweep simulations and experiments.

The traditional control strategy for the L-LLC resonant converter is the synchronous unequal width pulse frequency modulation (SUW PFM). However, this strategy causes a large circulating power in the forward buck and reverse boost modes, especially when the voltage gain deviates from 1. To solve the limitations of the SUW PFM, a novel constant width PFM (CW PFM) is proposed in this paper, which can reduce the circulating power and improve the converter efficiency in the forward buck and reverse boost modes. Furthermore, a hybrid PFM control strategy is proposed by merging the appropriate working conditions of the CW PFM and the SUW PFM, which can flexibly switch the two modulations depending on the gain and load conditions. Therefore, the L-LLC resonant converter has good soft-switching characteristics in entire load and gain ranges. Finally, a 7.5kW L-LLC resonant converter prototype is built to verify the validity and superiority of the proposed control strategy.