Introduction to Electromagnetic Compatibility

Due to the characteristics of low cost and high efficiency, the transformerless photovoltaic (PV) grid-connected inverters have been popularized in the application of solar electric generation system in residential market. Unfortunately, the leakage current through the stray capacitors between the PV array and the ground is harmful. This paper focuses on the leakage current suppressing method, in which all common-mode paths are considered. First, the common-mode analytical model at switching frequency is developed, and the rules of eliminating switching frequency common-mode source are summarized based on this model. The existing full-bridge- and half-bridge-type converters have been analyzed by using the developed model and rules, and then, a new full-bridge-type converter structure and a compensation strategy for half-bridge-type inverter have been presented finally.

Leakage Current Analytical Model and Application in Single-Phase Transformerless Photovoltaic Grid-Connected Inverter

A theory of overall quantization noise for nonsubtractive dither was originally developed in unpublished work by J.N. Wright and by T.J. Stockham and subsequently expanded by L.K. Brinton, S.P. Lipshitz, J. Vanderkooy, and R.A. Wannamaker. It is suggested that since these latter results are not as well known as the original results, misunderstanding persists in the literature. New proofs of the properties of quantizer dither, both subtractive and nonsubtractive, are provided. The new proofs are based on elementary Fourier series and Rice's characteristic function method and do not require the use of generalized functions (impulse trains of Dirac delta functions) and sampling theorem arguments. The goal is to provide a unified derivation and presentation of the two forms of dithered quantizer noise based on elementary Fourier techniques. >

http://ieeexplore.ieee.org/iel4/557/4025/00695075.pdf

Dithered Quantizers

Silicon carbide (SiC) MOSFETs and gallium nitride (GaN) high-electron mobility transistors are perceived as future replacements for Si IGBTs and MOSFETs in medium- and low-voltage drives due to their low conduction and switching losses. However, it is widely believed that the already significant conducted common-mode (CM) electromagnetic interference (EMI) emission of motor drives will be further exacerbated by the high-speed switching operation of these new devices. Hence, this paper investigates and quantifies the increase in the conducted CM EMI emission of a pulse width modulation inverter-based motor drive when SiC and GaN devices are adopted. Through an analytical approach, the results reveal that the influence of dv/dt on the conducted CM emission is generally limited. On the other hand, the influence of switching frequency is more significant. Lab tests are also conducted to verify the analysis.

Comparative Analysis on Conducted CM EMI Emission of Motor Drives: WBG Versus Si Devices

A generalized terminal modeling technique was proposed earlier to predict conducted electromagnetic interference from a dc-dc boost converter. The predictions of these conducted emissions showed that there was a good agreement up to 50 MHz. This paper extends the generalized terminal modeling approach to converters with the buck-type input. Both dc and ac applications are discussed. The technique is developed for the electromagnetic interference modeling of switched power converters in aerospace applications where the requirements on electromagnetic pollution are very strict. The model is shown to successfully predict conducted emissions for a buck converter and a three-phase voltage source inverter up to 100 MHz with an error of 6 dB or less at most frequencies.

Analysis of EMI Terminal Modeling of Switched Power Converters

An improved and simplified electromagnetic interference (EMI) modeling method based on multiple slope approximation of device-switching transitions for EMI analysis of power converters is presented. The traditional noise source modeling method, which uses single slope for rise and fall transition, is studied, and the criteria for reasonable modeling in the frequency range is analyzed. The turn-on and turn-off dynamics are investigated by dividing the nonlinear transitions into several stages based on an insulated gate bipolar transistor (IGBT) behavior circuit model. Real device-switching voltage and current waveforms are approximated by piece-wise linear lines and modeled by multiple dv/dt and di/dt slopes. The predicted EMI spectra suggest that high-frequency EMI noise is modeled with an acceptable accuracy. The proposed method was verified experimentally for a dc-dc buck converter

Multiple Slope Switching Waveform Approximation to Improve Conducted EMI Spectral Analysis of Power Converters

In this paper, a general lumped circuit modeling method is proposed to describe the conducted electromagnetic interference (EMI) coupling mechanism for the switching power converters. The EMI characteristics of the converters can be analytically deduced from a circuit theoretical viewpoint. The shunt and series impedance insertion method is introduced to identify the differential-mode (DM) and common-mode (CM) noise impedances and voltage sources. The procedure of parameters estimation for the noise models comprises several simple measurements and is convenient to be implemented. Experimental illustrations are also included to verify the validity of the proposed method. Comparison between the measured and predicted results shows that the EMI modeling method can provide adequate prediction of the EMI feature for power-switching converters

Noise Source Lumped Circuit Modeling and Identification for Power Converters

Automatic modulation classification (AMC) lies at the core of cognitive radio and spectrum sensing. In this Letter, the authors propose a novel convolutional neural network (CNN)-based AMC method with multi-feature fusion. First, the modulation signals are transformed into two image representations of cyclic spectra (CS) and constellation diagram (CD), respectively. Then, a two-branch CNN model is developed, a gradient decent strategy is adopted and a multi-feature fusion technique is exploited to integrate the features learned from CS and CD. The proposed method is computationally efficient, benefited from its simple neural network. Experimental results show that the proposed method can achieve identical or better results with much reduced learned parameters and training time, compared with the state-of-the-art deep learning-based methods.

Convolutional neural network and multi-feature fusion for automatic modulation classification

In this paper, a digital-domain reconstruction based radio-frequency (RF) interference cancellation scheme is proposed to cancel the nonlinear wideband self-interference (SI) signal in co-site radio systems. First, an auxiliary cancellation chain is employed to reconstruct and suppress the SI signal received at the receiver chain. Then, in order to generate the SI replica more accurately, we derive two nonlinear models to predicate the coupling characteristics for the nonlinear wideband SI signal. The proposed models combine the transmit chain and wireless SI channel characteristics, and employ both the Hammerstein model and Wiener model for the transmit chain, respectively. Next, the coefficients of each linear and nonlinear component in the SI chain are integrated into a coefficient vector, which is linear with respect to the output samples, with the proposed models. Based on the linearity, a dynamic least-squares (LS) type algorithm is developed to efficiently estimate and update the coefficients of the proposed models. Finally, experimental results obtained from a FPGA-based testbed have validated that the proposed scheme has a superior interference cancellation performance for nonlinear wideband SI.

A Digital-Domain Controlled Nonlinear RF Interference Cancellation Scheme for Co-Site Wideband Radios

Digital Self-Interference Cancellation (SIC) is the key technique for enabling In-Band Full-Duplexing (IBFD). Commonly, a nonlinear model is required to approximate the circuity nonlinearity so as to achieve SIC. Artificial neural network (ANN) is a potential candidate due to its nature of nonlinear approximation. Although real-valued neural network (RVNN) is more widely used in the field of artificial intelligence, the counterpart, complex-valued neural network (CVNN) seems to be more appropriate for digital SIC since the communication baseband signals are essentially complex. With this hypothesis, we conduct comprehensive comparisons of RVNN and CVNN for digital SIC in this paper. It is interesting to find that CVNN does perform better but only when non-holomorphic activation functions are used, and the number of hidden layer neurons can be reduced by maximally a half. When the number of neurons is larger than a certain value, RVNN and CVNN both achieve their optimal performance and provide approximately the same cancellation result.

Jin Meng

Papers

Multiple Slope Switching Waveform Approximation to Improve Conducted EMI Spectral Analysis of Power Converters

Noise Source Lumped Circuit Modeling and Identification for Power Converters

Convolutional neural network and multi-feature fusion for automatic modulation classification

A Digital-Domain Controlled Nonlinear RF Interference Cancellation Scheme for Co-Site Wideband Radios

Performance Comparison of Real and Complex Valued Neural Networks for Digital Self-Interference Cancellation