Etude et developpement du guide d'ondes integre au substrat pour la conception de systemes en ondes millimetriques

The launch of a new communication system is greatly affected by its production costs. The quest for reducing such costs remains a research area in constant evolution. The common school of thought regarding this type of research is to minimize the value of every component contained in the system. Despite that, we propose a novel approach: a complete integration platform allowing the design of low-loss passive components and the assembly of surface mounted active circuits on a single dielectric layer or substrate. To create this platform we developed and used the Substrate Integrated Waveguide (SIW), which is synthesized directly on a planar substrate with two rows of metallic cylinders. This thesis presents an in-depth study of the SIW and its characteristics. It also describes the design techniques of passive circuits and the integration of active circuits. The obtained results are applied to the design of a millimetre-wave Radio-over-Fibre (RoF) interface.;Planar transitions are key components in the integration platform. We proposed three different topologies: the first uses a microstrip line; the second, a coplanar waveguide with a voltage probe and the last, a coplanar waveguide with a current probe.;Millimetre-wave filters are one of the most expensive passive components because they need to be tuned manually. We studied two different filter topologies: the inline inductive post filter and the dual-mode filter. Although the inline topology is more suitable for the SIW, it does not allow the insertion of transmission zeros. The dual-mode topology requires more work but can be used to synthesize the generalized Tchebyshev transfer function. (Abstract shortened by UMI.);The first chapter of this thesis deals with the calculation of complex propagation constant in the SIW. Three different techniques are studied: the first is based on the solution of an eigenvalue problem, the second, on the solution of a transcendental equation and the third, on the surface impedance concept. We found that the eigenvalue method is very accurate but tedious and that the transcendental method does not converge. Confronted with these problems, we proposed a new technique combining the Method of Moments (MoM) with the surface impedance concept. We solved this problem with the transverse resonance technique.