High voltage (HV) generator with wide load range is widely adopted for X-ray beam excitation and density adjustment. SiC based LCC HV inverter at high switching frequency will benefit fast transient for more efficient utilization of X-ray beam and smaller size. A precise prediction of SiC switching characteristics provides important prerequisites for efficiency, thermal, power density and reliability co-optimized design. Understanding SiC switching loss at different switching conditions will help resonant network design to achieve required gain and efficiency over wide operation ranges. SiC switching characteristics is strongly depended with its parasitic capacitors. Miller capacitor, existed across gate-source loop and drain-source loop, presents a strong non-linear characteristics. This paper proposes an accuracy enhanced miller capacitor modeling approach. An internal cascaded configuration of a fixed capacitor and a voltage dependent capacitor with segmentations based on different doping concentrations is employed. The maximum mismatch between the proposed approach and datasheet’s C-V data is within 28%. Prediction accuracy improves two times compared with device SPICE model. An enhanced DPT circuit is used for comprehensive switching behavior verification. The proposed model has highest accuracy in switching loss prediction with up to 15X less mismatch compared to state-of-arts. The hard turn on and off loss are 157.7µJ and 48.1µJ at 500V/40A. The ZVS turn on loss is 1.67µJ at 40A. Soft turn off loss is reduced to 4.43µJ with 3.2nF snubber capacitor. SiC transistor with ZVS-ON and soft turn off has the lowest overall switching loss of 6.1µJ. Such switching condition corresponds to the region of switching frequency higher than resonant frequency in LCC inverter. Therefore, both high efficiency and high power density will be achieved under this optimal region.

Utilizing coil inductor of electromagnetic vibration energy harvester (EVEH) to form single stage ac–dc boost converter can make the self-powered milli-power system more compact. However, large output current harmonics of EVEH significantly degraded the output power in such system. This article proposes a new approach of adopting hysteresis current control to realize maximum power point tracking (MPPT) and reduce current harmonics in inductor-less converter by using auxiliary coils. The output characteristics and correlations between multiple coils are firstly studied to verify the feasibility of using auxiliary coil to represent the output of main coil. The optimal winding strategy for main coil and auxiliary coils is revealed for no degradation on maximum output power of main coil. Based upon these, a completely self-powered multicoil EVEH system is designed with auxiliary coils as reference for MPPT controller and as auxiliary power supply in cold-start. Experiments show that the designed prototype can obtain 9.54 mW output power at acceleration of 3 m/s ² , with the overall efficiency of 77.6%. To charge a 10 mF supercapacitor from zero to 5 V, an average output power of 7.27 mW can be obtained with an average charging efficiency of 62.4%. In addition, the start-up method adopting auxiliary coil allows the system to cold-start when the output voltage is lower as 0.5 V. The proposed approach has been demonstrated to achieve high overall efficiency in milli-power self-powered EVEH system and also has significant advantages in cost and volume compared with using external devices to obtain reference.

Most of electromagnetic vibration energy harvester (EVEH) systems are only focused on resistance matching at resonant frequency. However, the EVEH will have complex output impedance at nonresonant frequencies. Pure resistance matching cannot fully extract the maximum power at nonresonance. This article is the first to propose an automatic and fully self-powered complex impedance matching EVEH system. The characteristics of the output impedance of the EVEH are analyzed, and it is found that the impedance will be distributed on a circle. An impedance-circle-based perturb-and-observe (P&O) method with dual time-scale is proposed. It perturbs the central angle with a short time-scale for fast response to vibration frequency change, and perturbs both angle and diameter with a long time scale to adapt to the slow aging and wear-out of the EVEH. Such dual time-scale P&O improves the speed and accuracy of complex impedance determination. Besides, a fully self-powered milliwatts impedance control scheme with an inductor-less H-bridge ac–dc converter is designed. Experiments show that the designed system can realize complex impedance matching control stably and quickly. The maximum improvement of the output power of the converter reaches 25% compared to resistance matching scheme at nonresonant frequencies. Under the excitation acceleration amplitude of 3 m/s ² at the resonant frequency, the system generates the maximum power of 10.11 mW with the overall system efficiency reaching 82.2%. It is the first time the automatic complex impedance matching is realized in millipower fully self-powered EVEH.

Benefited from the fast switching speed, Gallium nitride (GaN) high electron mobility transistors (HEMTs) have been widely used in high switching frequency converters. However, due to the significant correlation between the turn- off time and operating conditions (more than 20 times difference of the turn- off time between the high load current and small load current for GaN HEMTs), using a fixed dead time will introduce extra dead-time (DT) losses in inverters where the output voltage and current are constantly varying. Therefore, this article proposes a high-efficiency adaptive method to dynamically adjust the GaN-inverter DT with operating conditions. An improved transient model including the parasitic inductance and output voltage with bidirectional solution flow has been proposed for GaN HEMTs to increase the accuracy of DT adjustment. Based on this model, the dynamic DT adjustment can be realized without extra sensors for GaN-inverters. Using the dynamic DT adjustment, the experimental results show a peak efficiency of 98.25% in a 1200 W inverter with triangular current mode modulation. Compared with using the fixed DT, the dynamic DT method can reduce the power loss by 26.7% under full load and 49.14% under light load.

For gallium nitride (GaN) based triangular current mode (TCM) applications, the dead-time has a significant effect on the switching loss. However, previous GaN high electron mobility transistor (HEMT) models focus on the turn- on/off process without fully considering the effect of the dead-time. Hence, this article analyzes the switching transients under the superfluous and insufficient dead-time and evaluates the dead-time loss with an improved GaN HEMT model. The proposed model improves an existing GaN HEMT model by adding the voltage rising/falling time of the gate driver, the dynamic threshold voltage of Schottky-type GaN HEMTs, and an equivalent gate-drain capacitance obtained from the datasheet. Verified by a GaN-based double-pulse test, the proposed model can more accurately calculate the gate-source voltage and the self-commutated reverse conduction voltage. Verified by a GaN-based TCM Buck converter, the proposed model can predict the dead-time loss well and has higher simulation accuracy for the turn- on process induced by the inappropriate dead-time.

Realizing zero turn- off loss (ZTL) of SiC mosfet s in zero voltage switching (ZVS) power converters will further break the limitation of the switching frequency and push for higher power density. However, quantitative models analyzing ZTL phenomenon were not revealed in state-of-arts and the assumptions of switching transients under ZTL were not correct. Current ZTL implementations either have limited application range or sacrificed transition speed. To overcome these limitations, this article proposes an improved determination criterion considering no Miller plateau for an accurate prediction of ZTL and its boundary based on the Miller's theorem. A more accurate multicurve parameter extraction approach is presented combining the C-V curve with the gate charge and transfer characteristic curves for parameters extraction under dynamic switching transients, further improving the ZTL determination accuracy. For ZTL implementation, a novel approach of coadjustment of gate and channel current based on the gate charge compensation is first proposed, with advantages of wide adaptive range and high d v /d t . Experiments show the quantitative determination of the ZTL boundary has prediction error of less than 9.5%. It is also verified that the proposed ZTL implementation can reduce the turn- off loss up to 81% with only 20% increasement in the drain-source voltage rising time at ZTL boundary transition. The proposed implementation is also capable to adjust d v /d t based on demand with adjustment error less than 7.2%.