While the employment of wide bandgap (WBG) devices in high-frequency and high-voltage applications brings benefits such as reduced system size and improved efficiency, it aggravates the electromagnetic interference (EMI) issue due to fast switching. High-frequency EMI noise suppression relies mainly on the filter design, where the filter's performance is strongly affected by parasitics. Through analyzing the common-mode (CM) equivalent circuit of a half-bridge power module, this letter identifies the key parasitics that dominate the performance of a common-mode filter (CMF) at high frequencies. To minimize the parasitics, the concept of integrating the CMF inside the WBG power module package is developed to improve the noise attenuation. A π-type CMF is integrated with a half-bridge GaN-based power module as a prototype to validate the concept. Experiments are conducted by measuring the CM noise spectrum received by the line impedance stabilization networks (LISNs) from the hard switching of the designed power module under 70 V and 80 kHz. Comparing the measured results of the integrated CMF to the externally-added CMF, up to 50 dBμV more attenuation is achieved by the integrated CMF in the frequency range of 10 MHz to 100 MHz, verifying the theoretical analysis and the established CM equivalent circuit.

Lateral gallium nitride (GaN) high-electron-mobility transistors (HEMTs) present better electrical characteristics compared to silicon or silicon carbide devices such as high switching speed and low gate charge, but also present additional challenges on the module design. This paper discusses a high-density GaN power module with double-sided cooling, low inductance, on-package decoupling capacitors, and integrated gate drivers. The GaN dies as well as the gate drive are sandwiched between the printed circuit board (PCB) and direct bonded copper (DBC) substrate to achieve compact loop and double-sided cooling effect. Design considerations and thermal performance are analyzed. A module assembly procedure is presented utilizing the layer-by-layer attachment process. Finally, a 2.7 cm × 1.8 cm half-bridge GaN power module is fabricated and tested, achieving a low power-loop inductance of 1.03 nH, and the overshoot voltage of the switching waveform is less than 5% under a 400-V/25-A double-pulse test. The thermal resistance is 0.32 K/W, verified by simulation and experimental results. The design and assembly process can be generalized and applied to high power applications to achieve high power density and high performance.

Existing current sensors suffer from insufficient measurement bandwidth or large insertion area to faithfully capture the continuous switching transient current of wide-bandgap devices. The combinational Rogowski coil concept is proposed here, where the self-integrating region of a shielded Rogowski coil is combined with its differentiating region to extend the overall measurement bandwidth. The coil design methodology, including the parasitic element model and practical design boundaries, is discussed. The concept is implemented with printed circuit boards and prototypes demonstrated up to 300 MHz bandwidth. The coil can also be easily integrated in an SiC mosfet module power stage and it is shown that the switching transient current waveform is faithfully captured.

Universal serial bus power delivery (USB-PD) chargers are more attractive to consumers with multiple output ports where more devices can be charged simultaneously. Existing multiple-port charger designs take multi-stage approaches, which increases power loss, charger size, and cost. This paper addresses this issue by presenting a time-division multiplexing (TDM) quasi-resonant (QR) flyback converter that eliminates the buck converter stages with simple and commercial analog implementations without the need of microprocessors. The circuit operation and practical design considerations are discussed in detail. An universal AC input, dual-USB-C GaN prototype validates the concept, while demonstrating USB-PD compliant operation, passed conducted EMI, and superior efficiency which passes the most stringent efficiency standards with margin where the benchmark GaN charger fails. Thanks to the elimination of buck stages, the proposed dual-USB-C charger achieves 1.9 W/cm$^{3}$ uncased power density, close to that of single-USB-C charger.

Universal serial bus power delivery (USB-PD) fast chargers equipped with wide-bandgap devices are driven to higher power density and efficiency. The indispensable high-voltage bulk capacitors used to smooth the rectifier output could take 40% of the total system volume due to the large capacitance value required. This paper discussed a capacitor reduction method using a self-driven thyristor scheme comprised of only three components in total. No extra control circuit is needed. Circuit analysis and design equations are presented, and the design results are implemented in a 60-W GaN-based active-clamp flyback converter. The measurement results on the prototype show a 36.4% reduction of the bulk-capacitor size with similar efficiency compared to the conventional solution.

It is proposed that ferromagnetic resonance (FMR) of thin film ferrites can be utilized to simultaneously improve the radiation efficiency and input impedance matching of electrically small antennas (ESAs). To validate the concept, the role of FMR in radiation is first derived analytically with an ideal thin-film ferrite radiator. It was concluded that the Gilbert damping of the ferrite, determining the quality factor of FMR, directly impacts on the radiation efficiency of the antenna. A practical example is proposed in the form of an electrically small single loop antenna loaded with a thin-film yttrium-iron-garnet (YIG) core. The prototype has been designed, fabricated and evaluated through both full-wave simulations and experiments. The simulation results match well to the experimental results, demonstrating the efficacy and significance of the idea. In addition, broadband equivalent circuit models are derived to model both the electrically small loops with and without the FMR enhancement and to provide additional insights to the design. The circuit models prove to be effective in predicting the input impedance and radiation efficiency of FMR enhanced ESAs at a precision comparable to full-wave simulations.

A physics-based nonlinear circuit model is developed for frequency-selective limiters (FSLs) based on thin-film magnetic materials. The equivalent circuit model is structured and its parameters are determined rigorously from the fundamental physics of electromagnetic waves and spin waves. The spin motions as well as the ferromagnetic resonance (FMR) are modeled by RLC parallel circuits with parameters derived from Polder's tensor and Kittel's equations. The exchange coupling between spins is modeled by an inductor added between adjacent RLC circuits based on quantum spin theory. The nonlinear crossfrequency coupling from signal at ω to spin waves at ω/2 is represented by a nonlinear coupled inductor model that follows the mathematics of a pendulum motion to represent the parametric oscillations of spins. The circuit units are cascaded to describe the spin-wave propagation inside the magnetic material, and transmission line parameters are added finally to describe the electromagnetic wave propagation. An FSL device described in the literature is used as an example to validate the circuit model, and the model successfully predicts the power-dependent insertion loss as well as the threshold power level, time delay, and frequency selectivity of the power limiting effect.

A methodology to develop a subcircuit model for core loss simulation in LTspice is presented. The subcircuit implements the dynamic core loss model used in the transient solver of finite-element analysis (FEA) in order to provide equivalent results in the time domain for ferrite materials. The wipe-out rule is applied in the simulation so that the core loss during the transient behavior can be predicted. A field factor is derived from an effective flux density for a nonuniform flux distribution to simplify the relationship between the current and the field for arbitrary magnetic core shapes. The interface of the subcircuit is capable of being integrated to any power stage simulations. Simulation results of core loss equivalent to the FEA on an exemplary plate-core inductor are obtained from the subcircuit that significantly reduces computational cost.

An inductor with winding sandwiched between two core plates is analyzed to model the nonuniform distribution of magnetic field. The winding is placed near the edge of the core to maximize the energy within the limited footprint such that the amount of energy stored outside the core volume is not negligible. The proportional-reluctance, equal-flux model is developed to build the contours with equal amount of flux by governing the reluctance of the flux path. The shapes of the flux lines are modeled by different functions that are guided by the finite-element simulation. The field calculated from the flux lines enables calculation of inductance, winding loss, and core loss. The inductance is used as a figure of merit to evaluate the modeling accuracy. Prototypes made of flexible circuit for inductors with different layouts are measured to verify the model. The measured inductances agree with the modeled result by less than 13% error.

A “distributed inductor” comprises a core containing multiple winding windows carrying prescribed Ampere-turns. Its magnetic field can be shaped to vary by less than a distribution factor υ. The positions, dimensions, and Ampere-turns of the windings are synthesized to improve the energy density. A design procedure is formulated to accentuate the impact of υ on the tradeoffs between inductance and losses. It is validated by a prototype having half the height of the commercial counterpart for a 30 W converter.

The “constant-flux” concept is leveraged to achieve high magnetic-energy density, leading to inductor geometries with height significantly lower than that of conventional products. Techniques to shape the core and to distribute the winding turns to shape a desirable field profile are described for the two basic classes of magnetic geometries: those with the winding enclosed by the core and those with the core enclosed by the winding. A relatively constant flux distribution is advantageous not only from the density standpoint, but also from the thermal standpoint via the reduction of hot spots, and from the reliability standpoint via the suppression of flux crowding. In this journal paper on a constant-flux inductor (CFI) with enclosed winding, the foci are operating principle, dc analysis, and basic design procedure. Prototype cores and windings were routed from powder-iron disks and copper sheets, respectively. The design of CFI was validated by the assembled inductor prototype.

Nonlinear magnetic properties in power electronics applications, such as permeability and core loss cause significant discrepancies between magnetic designs and prototype measurements, leading to iterative prototyping efforts. To address this issue, a bottom-up approach that is based on the physics of magnetic material is proposed to unveil the fundamental origin of nonlinearity. The physical parameters such as magnetization, gyromagnetic ratio, damping factor, etc., are translated to equivalent circuit parameters. The nonlinear magnetization dynamics of a single magnetic domain are described by an equivalent circuit model derived from the Landau-Lifshitz-Gilbert (LLG) equation. A fictitious magnetic domain is used to demonstrate the developed model’s capability on predicting the nonlinearity that includes magnetization dynamics and hysteresis. Then, the model is further validated by a commercial material, TDK N87 toroidal core. The results such as B-H loop and permeability simulated from the circuit model are in close agreement with the datasheet, validating the feasibility of predicting magnetic material’s nonlinear performance from physics perspective rather than empirical approaches.

Existing Rogowski coil sensors suffer from insufficient measurement bandwidth, which limits their ability to accurately capture the switching transient current of wide-bandgap devices. A combination Rogowski coil is proposed to extend the overall measurement bandwidth to address this limitation. This concept involves combining the self-integrating region of a shielded Rogowski coil with its differentiating region. The selection of suitable geometrical parameters is crucial as it directly affects the measuring performance, including bandwidth and sensitivity. This paper explores the methodology of employing a multi-objective optimization workflow to design Printed Circuit Board (PCB) shielded Rogowski Coils. Eight geometrical parameters are considered as design variables to maximize the objective functions. Utilizing the distributed model of the shielded PCB Rogowski coil, the geometry of the PCB shielded Rogowski coil is optimized using JMP. The factors that have a significant impact on bandwidth and sensitivity are identified through the optimization process. To validate the proposed optimized approach, an optimized current sensor is designed, prototyped, and tested. A comparison is made between the optimized design and state-of-the-art alternatives to assess its performance.

This paper, to the best knowledge of the authors, for the first time reports the packaging of a 100kW three-level active neutral point clamped (3L-ANPC) phase leg power module for electric vehicle (EV) traction inverter applications using 650V/150A e-mode gallium nitride high-electron-mobility transistors (GaN HEMTs). Compared with two-level (2L) half bridge power module, the main challenges of packaging a 3L-ANPC phase leg power module are high number of switches (6 switches vs 2 switches) and multiple commutation loops (1 commutation loop vs 4 commutation loops). In addition, there are four switches involved in the two long commutation loops. Those parasitic loop inductances must be minimized simultaneously. Those challenges are addressed by meticulous packaging of the power module, featuring low power loop inductances, double-sided cooled power module, low junction-to-coolant thermal resistance, symmetrical layout for the two parallel dies of each switch and mechanical robustness, etc. The simulated loop inductances are 1.92nH for the short loops and 5.21nH for the long loops, respectively; resulting in a 471V and 540V of turn-off voltage spikes in a double pulse test simulation at 400V/200A. The simulated junction-to-coolant thermal resistances are 0.37°C/W and 0.53°C/W for different switches, respectively.

Severe electromagnetic interference (EMI) issues in wide-bandgap (WBG) power electronics systems result in larger filter sizes to attenuate noise, which compromises the high power density brought by WBG devices. To mitigate the high-frequency EMI noises of WBG systems, integrating a common-mode filter inside the power module is an effective way of managing parasitics in the noise propagation paths and thereby improving the filter performance. However, the magnetic core of a common-mode choke (CMC) takes up considerable height and weight, which lowers the power density of a module. In this paper, an air-cured and module-compatible manufacturing methodology is developed using permalloy-epoxy magnetic composites for the integrated CMC inside a GaN power module.The magnetic composite is mixed into a paste texture and can conveniently mold the power module to form the CMC, improving the space utilization and power density of the whole module. This also offers a potential magnetic integration solution for other WBG power modules or circuits. One of the composites has a real part of the complex permeability of 13.5 under high frequency with low core-loss density. A 3-D finite-element model is built to predict the inductance of the over-molded CMC cured with the designed magnetic composite. The manufacturing process of the over-molded CMC is explained and the impact of the magnetic material on the power module parasitics is analyzed. The benefits of the over-molded CMC in EMI reduction and power density improvement are verified by test results, and some suggestions for future optimization are provided.

This paper discusses the impact of parasitic inductances on the electromagnetic interference (EMI) performance at radiated frequency and provides a new concept for high-frequency wide bandgap (WBG) power module package design with integrated $\pi$-type common mode filter ($\pi$-CMF). The connection parasitic-inductances of a $\pi$-type CMF model are analyzed, and the parasitic inductances from the CMF to the CM noise source and to the heatsink are minimized to improve the CMF's EMI performance in the radiated frequency range. Therefore, placing the $\pi$-CMF closer to the power module (i.e. in-package CMF) provides a larger noise attenuation compared to placing it outside the module (i.e. external CMF). To verify the theoretical analysis, a half-bridge GaN power module with an in-package $\pi$-CMF is designed, and experiments are conducted by comparing the attenuated noise spectrums of a 70-V/1.75-A hard-switching buck converter built by the designed module with an external CMF and the power-module integrated CMF. According to the experiment results, up to $10\ \text{dB}\mu\mathrm{V}$ more attenuation is achieved by the in-package CMF than the external CMF, validating the analytical conclusion.

In this paper, an embedded GaN half-bridge power module with double-sided cooling, low inductance, low thermal resistance, on-package decoupling capacitors, localized common mode filter, and integrated gate drivers is proposed. The two GaN dies are embedded in a printed circuit board (PCB) with heat dissipation paths to a ceramic substrate from both sides of the devices to achieve double-sided cooling capability. Thermal and electrical performance are fully analyzed and optimized. A low-cost module assembly procedure is presented utilizing standard layer attaching process. Finally, a compact $\mathbf{2.7}\ \mathbf{cm} \times \mathbf{1.8}\ \mathbf{cm}$ half-bridge GaN power module is fabricated to verify both electrical and thermal performance through experiments. The switching performance of the power module is tested under 400 V/25 A double-pulse test that shows the power loop inductance is as low as 1.03 nH and the overshoot voltage of the switching waveform is less than 5% of the dc bus voltage. The thermal resistance is verified to be 0.32 K/W, and the fabricated power module is employed in a buck converter with 500 W output power at 600 kHz switching frequency.

The DC measurement capability and high- frequency bandwidth are vital aspects for current sensors to accurately capture the switching current behavior of wide bandgap devices. However, typical Rogowski coil (RC) current sensors lack DC sensing capability resulting in inaccurate low- frequency measurement results. This paper presents a concept of combinational RC with improved DC measurement capability without intruding the power loop area for double pulse test (DPT) systems. The DC current sensor and the RC are placed at different locations to capture the low-frequency and high-frequency switching current, respectively. An analog adder circuit is used to combine the two sensing results, thereby extending the measurement bandwidth of the RC without increasing the insertion parasitic inductance. The frequency-domain measurement of the combinational RC with DC sensors verifies the extended DC measurement bandwidth. In the paper, the switching current measured from a half-bridge SiC power module under DPT has been used as an example to demonstrate the enhanced measurement capability of this combinational RC avoiding an increase of extra parasitic.

Universal serial bus power delivery (USB-PD) chargers are more attractive to consumers with multiple output ports. Existing multiple port charger designs take multi-stage approach thus increase power loss and charger size. Also, ports are normally designed with different power capability thus the high-power port becomes under-used if the connected device is fully charged while the other lower power ports cannot take over the unused power. This is mainly limited by the power architecture design. This paper addresses this issue by presenting a multiplexing based multi-port quasi-resonant flyback converter. The design achieves true power-sharing with high efficiency, small size, and simple analog implementation. This paper discusses the operating principle and design considerations. A GaN-based 60 W dual port charger verifies the proposed scheme while achieving 76% size reduction and 1% efficiency improvement than state-of-the-art.

An enhanced model for coupled inductors with significant fringing effect is developed in the form of equivalent circuit. The equivalent circuit is derived from a physical model that captures the flux paths through a leakage inductor and two mutual inductors on the primary and secondary side. A mutual winding resistor is added in parallel with the mutual inductor to model the winding loss affected by the fringing flux that penetrates the winding. The equivalent circuit is verified by finite-element simulation (FES) and successfully predicts the open-circuit winding loss and winding loss variation with phase-shift impact.

An unconditionally stable three-dimensional (3D) finite-difference time-domain (FDTD) algorithm has been proposed to solve simultaneously Maxwell's equations and the Landau-Lifshitz-Gilbert (LLG) equation with full nonlinear effects. The proposed algorithm can predict the dynamic interaction between magnetic spins and EM fields. The accuracy of the modeling has been validated by 1. Small signal simulation of a linear ferrite isolator and 2. Large signal simulation of the dispersive permeability of a continuous ferrite film. The simulations agree with the theoretical and experimental predictions, under both linear and nonlinear circumstances. Specifically, the algorithm has fully revealed that sufficiently large RF power can decrease the ferromagnetic resonance (FMR) frequency and suppress the permeability.

A nonlinear circuit model is developed for magnetic material based frequency-selective limiters (FSL). The dominant magnetic-behaviors of FSL devices are translated into equivalent circuits with parameters rigorously determined from fundamental physics. The spin motions as well as the ferromagnetic resonance (FMR) are modeled by RLC parallel circuits with parameters derived from Polder's tensor and Kittel's equations. The exchange coupling between spins is modeled by an inductor added between adjacent RLC circuits based on quantum spin theory. The nonlinear cross-frequency coupling from signal at ω to spin waves at ω/2 is represented by a pendulum model that predicts the parametric oscillations of spins. A FSL device described in literature is used as an example to validate the circuit model. The simulation results match the measurement results published, and the model successfully predicts the threshold power level, nonlinear insertion loss, time delay, and frequency selectivity of the device.

An unconditionally stable three-dimensional (3D) finite-difference time-domain (FDTD) algorithm has been proposed to predict the dynamic interaction between nonlinear magnetic spins and electromagnetic (EM) fields in nonlinear magnetic devices. The proposed modeling solves simultaneously Maxwell's equations and the Landau-Lifshitz-Gilbert (LLG) equation with full nonlinear effects. The accuracy of the modeling has been validated by 1. Small signal simulation of a linear ferrite isolator and 2. Large signal simulation of the dispersive permeability of a continuous ferrite film. The simulations agree with the theoretical and experimental predictions. The fact that the 2-μm-thick film exhibits strong nonlinearity shows the potential of magnetic thin film applied in miniature RF front ends.

The “constant-flux” concept introduced recently is leveraged to distribute magnetic flux to improve energy density, lowering the profile of an inductor. The optimal flux distribution with normalized parameters is identified mathematically, and verified by simulation. It is then applied to reduce the dc resistance of a commercial inductor by a factor of two, keeping the outer dimensions and inductance the same. Thermal-limited current rating is improved by 50%, whereas saturation-limited current rating is improved by 20% thanks to the suppression of flux crowding. The simulation result is verified by measurement results of a prototype under high current bias.

The “constant-flux” concept is leveraged to achieve high magnetic-energy density, leading to inductor geometries with height significantly lower than that of conventional products. Techniques to shape the core and to distribute the winding turns to shape a desirable field profile is described for the two basic classes of magnetic geometries, those with the core outside the winding and those with the core inside the winding. A relatively constant flux distribution is advantageous not only from the density standpoint, but also from the thermal standpoint via the reduction of hot spots, and from the reliability standpoint via the suppression of flux crowding. Design and fabrication results are delineated for a core-outside inductor with same specifications as a commercial counterpart, but with smaller footprint as well as smaller height. Prototype cores and windings were milled from powder-iron disks and copper sheets, respectively. The volume of the prototyped constant-flux inductor is 66% the volume of a commercial inductor with same inductance and dc resistance.