The aim of this Ph.D. is to improve the convelter conception. The technical evolutions and the recent restrictions increase the number of constrains and run through software solutions. In a first parl, the nowadays knowledge level will be examined. This study is based on a Flyback and Forward realisations. The design and the realisation of these convelters point out several problems. Consequently, studies about transformer are developed. This conception approach doesn't yet involve EMC considerations at aIl. In the second part, transformer-modelling problems are managed. A pri01"y magnetic representation of a Litz-wire winding layer is outliried. Comparison between theoretical and experimental results for several transformers validates the model. An analyticaI 2D-caIculation method is developed. It improves the leakage inductance representation. The ground plat proxirrlity consideration and a more accurate identification method improve the electrostatic transformer represen tati on. In the third part, EMC considerations ru'e inc1uded in the design concepts. A state of art about now available models is realised. Three methods to evaluate the perturQation levels in frequency domain are consldered. The disturbance levels analyse with regards* to variations of model parameters makes possible to deteet the most sensitive pru"ameters in order to use them to improve the EMC disturbance levels. In the last paIt, the first expeliences with recently developed converter conception software are exposed. This software 1S in expelimental stage and it isn't nowadays possible to extract any interesting results. During the Ph.D. we have outlined and eliminated several eITers, a future version of this software package will integrate significant modifications.
