Conducted EMI in Inverters with SiC Transistors

Abstract

Conducted EMI in Inverters with SiC Transistors Electromagnetic Interference (EMI) is the main side effect accompanied with the fast voltage and current switching transients in power electronics applications. Compliance of the Electromagnetic Compatibility (EMC) standard is prescribed for any power electronics product before entering the market. In recent years, the new emerged wide band-gap transistor technology Silicon Carbide (SiC) exhibits great potential to replace Silicon (Si) as the dominant transistor because of its superior qualities (e.g. faster switching, higher blocking voltage and higher operating temperature). However, these advances come at the cost of increased EMI resulting from the SiC transistor’s faster switching speed and higher switching frequencies. In the past, a large variety of EMI suppression approaches have been developed for motor drive systems. However, most of them serve the conventional Si power devices (e.g. IGBT) based motor drive systems. As the result, exploration of corresponding EMI emission mechanism and new suppression approaches is critical. The desired EMC investigations should cover the following features of the SiC power devices based drive systems. • The differences with the Si power devices based motor drive systems and the causes of the differences. • The common EMC analysis and reduction techniques that can be used for both SiC and Si devices based motor drive systems. • The approaches that improve the EMC performance for SiC devices based motor drive systems. In this thesis, with conventional Si IGBTs as the reference, systematic investigations are presented on variable speed drive systems using the latest SiC JFET as the power devices. Main achievements of this thesis are summarized as follows. System equivalent circuit modeling method for EMC performance evaluation This modeling method is used for evaluating the noise suppressing performance on varies noise propagation paths of the drive system. Additionally, by introducing the actual noise source emission instead of using idealized noise source (e.g pure square waveforms) emission, improved noise level prediction is achieved. The method is based on curve-fitting of the impedance-frequency characteristics measured on the leads of individual components and between different portions of the system. Various sub-circuits of the system are modeled as RCL composed equivalent circuits in order to represent details within the conducted frequency range. The model development procedure is presented in Chapter 3. Current transfer ratio measurements are used to verify the model. In addition, the model is also applied in Chapter 5 and 6. Chapter 5 utilizes the model to predict the noise emission levels of the SiC JFETs and Si IGBTs based motor drives respectively. Chapter 6 utilizes the model to predict the filter insertion losses. This method is applicable for both SiC and Si based motor drive systems. Characterization and cancellation of EMI filter parasitics to improve high frequency filtering performance This study explores the parasitics cancellation techniques for EMI filters with three-dimensional spatial layout that utilizes multi-layer PCB technology and employs surface mount technology (SMT) components. The employed SMT components are named x-dimensional (x-dim) components that have the same height (x=14mm) and double sided SMT electrical terminations. In addition to positioning the employed x-dimensional SMT components in the 2D plane, the components are able to be placed in a third dimension by being stacked onto more PCB layers. This extends the conventional parasitic cancellation techniques to three dimensions. Chapter 4 discusses and implements the 3D cancellation techniques in a LC filter for motor drives. The techniques enhance the filter performance especially in the high frequency range, which is critical to handle the increased EMI due to SiC fast switching speed. Comparison and identification of noise emission difference between Si IGBTs and SiC JFETs based motor drives This study compares and identifies the causes of the EMI noise emission differences between Si IGBT and SiC JFET based motor drives. In Chapter 5, two inverter prototypes – with Si IGBTs and SiC JFETs as the power transistors respectively are compared under the same power level and using the same layout. The gate drivers are designed to fully exploit the switching speed for the two types of devices at turn-on transition and to provide the same driving condition at turn-off transition. Their switching waveforms are compared under the inductive switching test condition, using one inverter leg consisting of two switches. The caused EMI level differences are clarified by Fourier analysis transformed from the time-domain measurements. In the system level, their EMI noise levels are compared under unfiltered, C filtered and LC filtered conditions. In order to identify the causes of the noise level differences, the two inverters are operated in the CM testing mode, in which the three top and bottom semiconductors are switched on or off simultaneously at a fixed 50% duty ratio. Hence the maximized CM and DM wave shapes are achieved. Improve EMI filter to accommodate SiC JFETs in motor drives In the second part of Chapter 5, the EMI filter design for SiC JFET motor drive system is improved based on the presence of different noise emissions from the Si IGBT and SiC JFET source. An equivalent circuit model is delivered to predict the noise spectrum emitted from the SiC source and stands as the basis for improving the EMI filter design. The proposed EMI method effectively suppress the increased high frequency noise resulted from the SiC faster switching dv/dts and di/dts. The modeled results agree well with the experiments. Suppression at SiC noise source due to substrate capacitive coupling Two methods to suppress the noise emission due to capacitive coupling are proposed in Chapter 6. One is to use separated substrate, the other one is to use the broadband modeling. Comparing two inverters that use the conventional heat sink and insulated metal substrate (IMS), the emitted noise levels are significantly different due to the different capacitive coupling magnitude. The first part of the Chapter proposes to use separated substrates. The second part presents a broadband modeling procedure to identify the most effective filter design to suppress the capacitive coupling. Both methods effectively suppress the noise emission to comply with the IEC61800-3-C2 standard. How EMI emission is affected by the capacitive coupling is identified.