Mickael Petit

Le Génie Electrique, en général, et l'électronique de puissance, en particulier, prend une part de plus en plus importante dans les systèmes embarqués. La fiabilité des systèmes électroniques dépend fortement de la gestion de leur température. Les systèmes de refroidissement actuels sont lourds, volumineux, et consommateurs d'énergie, ce qui est en désaccord avec les systèmes embarqués. Il est donc nécessaire de chercher de nouveaux systèmes, plus fiables, plus légers, et moins énergivores. Le sujet de cette thèse porte sur l'utilisation des ferrofluides, suspensions colloïdales magnétiques, à basse température de Curie, dont les propriétés magnétiques varient fortement avec la température entre l'ambiante et une centaine de degrés Celsius afin de les utiliser comme liquide caloporteur dans les systèmes de refroidissement. Les propriétés magnétiques fortement dépendantes de la température d'un tel fluide permettent la mise en mouvement de ce dernier par l'action couplée d'un champ magnétique et d'un gradient de température alors que toutes les pièces solides restent fixes. Le système de refroidissement n'est alors plus assujetti à l'usure de la pompe permettant la circulation du fluide caloporteur. Le système est ainsi globalement plus fiable et moins consommateur d'énergie. L'énergie de mise en mouvement du ferrofluide étant extraite directement des pertes des composants. Le comportement des ferrofluides est trop méconnu à l'heure actuelle pour concevoir et optimiser une pompe statique magnétothermique. Un effort important de modélisation et de caractérisation doit être mené. Ce manuscrit présente une étude pratique vérifiant le principe de création de pression hydrostatique par couplage magnétothermique. Une modélisation de la distribution des forces locales mettant en mouvement le ferrofluide ainsi que la mise en place d'outil permettant la caractérisation des ferrofluides sont également présentées. Les efforts de caractérisation se sont concentrés sur la rhéologie, au regard du champ magnétique, du cisaillement et de la température, ainsi que sur le comportement magnétique du ferrofluide à différentes températures.

Electroactive polymers show promising characteristics, such as lightness, compactness, flexibility, and large displacements, making them a candidate for application in cardiac assist devices. This revives the need for quasi-square wave voltage supply switching between 0 and several kilovolts, that must be efficient, to limit the heat dissipation, and compact in order to be implanted. The high- access resistance, associated with compliant electrodes, represents an additional difficulty. Here, a solid-state Marx modulator is adapted to cope with electroactive polymer characteristics, taking advantage of an efficient energy transfer over a sequential multistep charge/discharge process. To ensure compactness, efficiency, as well as the needs of an implanted device, a wireless magnetic field-based communication and power transfer system has been implemented. This work demonstrates the benefit of this design through simulations and experimental validation on a cardiac assist device. At a voltage of 7 kV, an efficiency of up to 88% has been achieved over a complete charge/discharge cycle.

This article presents local shielding techniques applied to a half-bridge inverter leg with the aim to reduce the common-mode (CM) current noise at converter’s dc input. The research study is conducted for 650-V enhancement mode Gallium Nitride power transistor switches. Main contributors of parasitic capacitances referred to the inverter-leg middle point node are identified. Then, shielding solutions are proposed to reduce the CM current emission by these capacitances. Respecting the precautions concerning the isolation of CM currents of the half-bridge inverter leg, the electromagnetic compatibility measurement setup is developed. Experimental step-by-step addition of local shielding copper layers to different contributors of middle point capacitance shows progressive attenuation of CM noise spectra.

This paper presents the design of a power electronics converter to produce strong current pulses (about 400A) in the core of a HVDC cable in order to use the thermal step method to evaluate the ageing state of the dielectric component. The described converter uses a magnetic coupler with intercells transformers and is current controlled.

The proposed innovative manufacturing process-described in detail-uses metal foam to create a pressed contact between the top side of a printed circuit board-embedded power die and the rest of the circuit. Initial prototypes were constructed using diodes with die dimensions of 4 mm × 6.35 mm. The prototypes were electrically characterized: the chip and contact dc and ac impedance values were measured and compared with those obtained for conventional packaging that uses bond wires. The electrical impedance of the prototypes was found to be similar to that of a state-of-the-art industrial package. Moreover, the proposed process is simple and cost-effective. Although the results presented in this letter are promising, further research is necessary to fully assess the benefits and limitations of the process.

Semiconductor devices incorporated into power electronics systems are proven to be highly sensitive to their thermal environment. Thus, temperature measurement is a key concern when it comes to reliability of power modules. Some Thermo-Sensitive Electrical Parameters (TSEPs) are used to evaluate online temperature of semiconductor components, in operating conditions of power converters. However, the accuracy and the reliability of those TSEPs are not yet fully demonstrated. To overcome this drawback, a power instrumented diode was developed with a Resistance Temperature Detector (RTD) sensor integrated on its top metal surface. This validation tool is characterized in this paper under constant power dissipation. The embedded RTD sensor is conditioned in a resistance bridge circuit. A dynamic characterization of a realistic power module is performed through thermal impedance evaluation. A comparative study is carried out to discuss the static and dynamic performances of the RTD sensor and those of a robust and well-known TSEP (Forward voltage drop under low current Vf_@Ict).

The layout of power modules is one of the key points in power module design, especially for high power densities, where couplings are increased. Stray inductances cause various problems in switching devices and should be minimized at an early design stage to avoid later required countermeasures and redesign after the device prototype has been built. In this paper, using Ansys® Q3D® Extractor, electromagnetic simulation is carried out to extract and evaluate the inductive and resistive parasitics of one phase of a three-phase automotive inverter module with rating of 650 V–400 A. The purpose is to develop and verily the electrical model including parasitic parameters for the studied phase-leg operating as a chopper. The switching behavior of the studied structure is analyzed based on Ansys® Simplorer® simulation, and a dynamic electrical model is carried out. Experimental test measurements are performed. The comparison between the simulation models and the experimental data is reported and conclusions are made.

This paper presents a balanced active and reactive power control, using a Phase Locked Loop for synchronization, and applied to a grid connected Five Level Inverter. The energy source of the system can be a photovoltaic generator or a wind turbine. We size the passive elements of the system and explain the value of the system architecture using a Five Level Inverter when compared to a classical grid connected system. We also compare the balanced active and reactive power control to an unbalanced active and reactive power control. The simulation results obtained by using Matlab Simulink and Simpowersystems are presented and discussed in this paper.

This paper presents the decoupled active and reactive power control (PQ control), using a Phase-Locked Loop (PLL) for synchronization, and applied to the control of Multicellular Parallel Inverter (MPI) for a grid-connected Photovoltaic (PV) system. We explain the value of this multicellular structure of inverter compared to a conventional inverter. We also present and discuss the simulation results obtained using the Matlab software (Simulink and simpowersystems).