EMI from Switched Converters – Simulation Methods and Reduction Techniques
Doktorsavhandling, 2011

In this thesis, the conducted EMI from switched power converters has been analyzed using various existing models, own-derived models as well as measurements. The ingoing passive components in a switching converter have been modeled with respect to their high frequency behavior and the static and dynamic behavior of the active semiconductors has been analyzed. To understand the origin of EMI, the sources of EMI within a switching converter are investigated and measures how to reduce conducted emissions are analyzed. Once the background theory is known, all models of the ingoing components are verified as standalone units; i.e. the high frequency behavior of resistors, inductors, capacitors, printed circuit boards and relevant semiconductors are presented. The verified components are then put together to complete converters from which the emission levels are analyzed. As a measure to reduce EMI directly at the source, a new method of reducing EMI referred to as active gate control is investigated. The principle aims at reducing the high frequency content in the otherwise sharp voltage and current transitions by controlling the gate voltage during a switching event. It has been found that many component models are insufficient for the purpose of EMI simulations; e.g. is the diode model in the widespread electric simulator SPICE unsuitable due to snappy reverse recovery behavior. Other simulation languages such as Simplorer and SABER® have solved this by incorporating more advanced diode models with adequate reverse recovery behavior that can be adapted to different diodes by more or less advanced parameter extraction procedures. Regarding MOSFET models, the black-box models provided by manufacturers have shown to exhibit good static and dynamic performance which in general makes them suitable for analysis of switching applications. Most analyzed IGBT models showed incorrect switching times and the most suitable model found in this thesis is the Simplorer model that requires more than 100 hand-tuned elements to show sufficient switching performance. When a complete converter is to be simulated, it is concluded that it is essential to include parasitic elements (e.g. stray inductances and capacitances in the PCB and in the components) in the simulations to obtain correct switching behavior. By such a consideration by e.g. Ansoft Q3d that determine the parasitic elements within a PCB, the often unwanted oscillations in the circuit can be determined to a large extent. However, the conducted emission levels are not just determined bystray elements and it has been has been shown both experimentally and in simulations that the levels are strongly determined by the mutual couplings in the input filter. The need for a diode model with correct reverse recovery behavior was shown to be important since the reverse recovery event affects the conducted EMI in the frequency region of 5MHz-30MHz. The connection between reverse recovery current and emission levels has been verified in simulations and measurements for two different types of converter topologies. The derived strategy of active gate control was verified both theoretically and by measurements that establish the operation principle. The method was shown to give a reduction in the conducted EMI of a MOSFET switching circuit and the most suitable unit to control was found to be the drain current. Finally, a reducedorder MOSFET model was shown to be sufficiently accurate to model static as well as switching applications due to its thorough characterization of the gate-drain capacitance as a function of applied voltages.

state space methods


semiconductor device modeling

diode reverse recovery

diode modeling

conducted emissions

Semiconductor device measurements

DC/DC converters

Opponent: Hans Bängtsson


Andreas Henriksson

Chalmers, Energi och miljö, Elkraftteknik




Annan elektroteknik och elektronik



Doktorsavhandlingar vid Chalmers tekniska högskola. Ny serie


Opponent: Hans Bängtsson

Mer information