Determination of Conducted EM Emissions on DC–AC Power Converters Based on Linear Equivalent Thevenin Block Circuit Models

Abstract

This article deals with circuit modeling of power systems and more particularly of DC/AC converters used for controlling electric motors. The objective of the model is to be able to assess conducted emission currents at both inputs and outputs in order to apply appropriate mitigation design rules such as filtering. First, we propose a linear Thevenin block-model of the converter considering it as a 5-port linear black-box allowing possible interaction between the DC and AC ports. To this extent, the block-model is made of an impedance matrix and a Thevenin voltage generators vector. We also propose theoretical relationships allowing the processing of data in order to be able to generate this model from measurements. Then, we test the linear simplifications on a real DC/AC converter and compare the results to measurements. For this, S-parameter measurements in offmode and currents at the 5 ports in on-mode are used for the characterization of the model on a first test setup. The model is then applied and checked by comparing circuit simulated and measured current responses on a modified test setup for which the lengths of the connection cables are modified. Finally, the capability of this block-model to extrapolate by calculation the response of other installation configurations is showed considering the design of a filter and its insertion at input/output ports of the converter with much longer cables.

Black-box Modeling of Converters in Renewable Energy Systems for EMC Assessment: Overview and Discussion of Available Models

“…There were four main approaches for characterizing the passive part of the model. The most commonly used method was to directly measure the S-parameter matrix of the converter with a vector network analyzer (VNA) and to convert the S-parameter representation into impedance/admittance notation [22,29] . It was possible to measure the S-parameters through an LISN and then de-embed the effects of the LISN and interconnecting cables [42] .…”

“…In both cases, it was necessary to simultaneously measure the voltage/current waveforms for each port to preserve phase information. An attempt to measure the currents directly in the frequency domain was described in a study by Guibert et al [29] . This method required the use of a spectrum analyzer and preliminary assumptions (usually not satisfied in the entire frequency interval) regarding the phase shift between different sources.…”

“…The choice between these two models depends on the two modeling strategies and objectives of the analysis. In other words, resorting to unterminated models is necessary when the goal is to predict conducted emissions (e.g., to design suitable EMI filters) at both sides of the power converter and/or when there is a need to connect several power converters in series [28][29] . On the contrary, if the analysis focuses on the emissions at one side only, and/or on converters connected in parallel, resorting to terminated models is preferable because the procedure for model derivation requires fewer measurements [18] .…”

“…For instance, Spadacini et al used the black-box model topology in Figs. 5d and 5e to predict the conducted emissions in the presence of twisted-pair cables or high-voltage direct current (HVDC) cables modelled by 2-N port networks [22,29] . The topologies in Figs.…”

“…It is possible to model it from either the direct current (DC) or alternating current (AC) side using a two-port or three-port terminated model, respectively. Alternatively, it is possible to model it using a five-port unterminated model with both the three ports on the AC side and the two ports on the DC side accessible [28][29] . The minimum number of sources and impedances required depends on the number of ports.…”