In this paper, a novel numerical method called interface finite volume method (I-FVM) for calculation of step-varying Electro-quasistatic (EQS) field is proposed. First, the principle of I-FVM is derived. Then, with numerical example of double layers parallel plate structure under step voltage which has an analytical solution, effectiveness and correctness of the I-FVM are verified. It can be found the calculating time of I-FVM is only 30% of normal FVM without decreasing accuracy during the whole calculating process. Furthermore, an engineering example about the electric field of DBC (Direct Bonding Copper) structure in a high voltage IGBT device is given. It can be found that accuracy of the I-FVM is the same as normal FVM, while time cost of I-FVM is only 20.8% of normal FVM. At last, the I-FVM is extended to one dimension based on the two-direction tri-diagonal matrix algorithm (TDMA) method given in this paper which can save processing of LU decomposition compared to one-dimensional traditional TDMA. In conclusion, the novel method called I-FVM proposed in this paper can decrease calculating amount for a step size varying electro-quasistatic field calculation problem. It may be a good method for large-scale EQS field calculation.

Plasma extraction transit time (PETT) oscillation might appear in IGBT devices, which is harmful to the electromagnetic compatibility (EMC) of a renewable energy system. To eliminate this oscillation, its frequency-domain characteristics in wire-bonded IGBT devices have been extensively studied. However, the time-domain analysis of PETT oscillation, especially in Press Pack IGBT (PPI) devices, has not attracted enough attention yet. In this paper, PETT oscillations with multi-chips in PPI devices are systematically investigated by experiments. It is first reported there are multiple resonant oscillations at the tail period when multi-chips turn off. Oscillations overlap in the time domain waveforms, which lead PETT oscillation to be more serious in multi-chips. Then, PETT oscillation is divided into three different feedback states for the first time. For the IGBT chip in the PETT oscillation, its physical based model and behavior model are proposed, which further form the equivalent circuit as the two-port network. Moreover, it is indicated that only parallel resonances can lead to PETT oscillation, which is consistent with experiment results.

This paper focuses on the finite element method in the complex frequency domain (CFD-FEM) for the transient electric field. First, the initial value and boundary value problem of the transient electric field under the electroquasistatic field in the complex frequency domain is given. In addition, the finite element equation and the constrained electric field equation on the boundary are derived. Secondly, the indirect algorithm of the numerical inverse Laplace transform is introduced. Based on it, the calculation procedures of the CFD-FEM are illustrated in detail. Thirdly, the step response, zero-state response under the positive periodic square waveform (PPSW) voltage, and the zero-input response by the CFD-FEM with direct algorithm and indirect algorithm are compared. Finally, the reason for the numerical oscillations of the zero-state response under the PPSW voltage is analyzed, and the method to reduce oscillations is proposed. The results show that the numerical accuracy of the indirect algorithm of the CFD-FEM is more than an order of magnitude higher than that of the direct algorithm when calculating the step response of the transient electric field. The proposed method can significantly reduce the numerical oscillations of the zero-state response under the PPSW voltage. The proposed method is helpful for the calculation of the transient electric field, especially in the case of frequency-dependent parameters.

The high-voltage high-power press-packed IGBT (PPI) devices are the key component of the DC transmission apparatus. A PPI device is composed of several PPI submodules. In general, the PPI submodule works in a state of repetitive turn-on and turn-off, and the corresponding working voltage is the positive periodic square waveform (PPSW) voltage, which is much different from the conventional AC or DC voltage. In addition, insulation capability is one of the most critical challenges in the design and fabrication of PPI devices. To improve the insulation capability of the device, composite insulation structures with multiple dielectrics are usually employed. Under the PPSW voltage, it is essential to analyze the transient electric field to solve the insulation challenge of the PPI devices. However, the electric field of the PPI is often calculated under the electrostatic field or DC field. Moreover, the transient characteristics of the electric field are ignored. Therefore, this paper focuses on the analysis of the transient characteristics of the electric field of the PPI submodule under the PPSW voltage. The influences of the waveform parameters of the PPSW voltage on the transient characteristics are demonstrated in detail. This study is significant for the insulation analysis and design of the PPIs.

In order to evaluate the insulation of two-phase immersion cooling in the HV power electronic package, the insulation degradation of the dielectric interface induced by bubbles is investigated. In this paper, a test strategy with 50 Hz unipolar DC and AC combined voltage for partial discharge (PD) at boiling interface of AlN ceramic is proposed. The insulation threshold of an AlN ceramic surface is acquired in several dielectric environments, such as air, FC-72 liquid (FC-72, a Fluorinert^TM from 3M^TM), FC-72 vapor, and boiling state of FC-72. This reveals the deterioration of boiling on the insulation of the surface immersed in the dielectric refrigerant. To investigate the mechanism of the PD feature at the boiling interface, the PD patterns of the unrestricted bubble and the accumulated bubble are acquired and contrastively analyzed. Combined with the feature of the back discharge and the bubble behavior, the charged vapor-ceramic interface is relatively stable due to the accumulated vapor layer. This stability of the charged vapor-ceramic interface is broken if the bubble is unrestricted. Besides, it is discovered that the vapor-liquid interface inside the bubble may be another charged interface, which can also trigger a back discharge.

This paper is devoted to solving the transient electric field and transient charge density on the dielectric interface under the electroquasistatic (EQS) field conditions with high accuracy. The proposed method is suitable for both 2-D and 3-D applications. Firstly, the governing equations represented by scalar electric potential are discretized by the nodal finite element method (FEM) in space and the finite difference method in time. Secondly, the transient constrained electric field equation on the boundary (TCEFEB) is derived to calculate the normal component of the transient electric field intensities on the Dirichlet boundary and dielectric interface as well as the transient charge density on the dielectric interface. Finally, a 2-D numerical example is employed to demonstrate the validity of the proposed method. Furthermore, the comparisons of the numerical accuracy of the proposed method in this paper with the existing FEMs for electric field intensity and charge density on the dielectric interface are conducted. The results show that the numerical accuracy of the proposed method for calculating the normal component of transient electric field intensities on the Dirichlet boundary and dielectric interface as well as the transient charge density on the dielectric interface is close to that of nodal electric potential and an order of magnitude higher than those of existing FEMs.