Ultimately, the purpose of a model for transmission lines is to describe how the transmission properties of a line are frequency dependent. The sources of such dependency can be grouped in two areas: that one related to longitudinal changes on the line geometry and constituting materials (non-uniformities), and that related to electromagnetic phenomena such as the skin effect and the dielectric dispersion (present even when the line is uniform). The contribution of this thesis is related to the above mentioned areas, focused on a specific type of transmission line, the twisted-pair. The models for twisted pairs found in literature assume simplistic considerations which are not always realistic, e.g., they ignore that the dielectric medium is heterogeneous and with losses, ignore the effect of the non-uniformities that are inevitable to all transmission lines, etc. These and other issues are taken into account during the development of a new twisted-pair model. This model is composed for two components, a deterministic one which is a function of the constructive characteristics of the twisted-pair, and a stochastic one which is a function of the inherent defects in twisted-pair cables. Regarding the deterministic component, it employs a realistic and straightforward approach to describe: the proximity effect of the conductors, the presence of conductors’ insulation, the dielectric losses and pair twisting. As a result, the model is more accurate than the models from the literature. Regarding the stochastic component, it was not found in literature similar models for comparison. Nevertheless, it was shown that the proposed stochastic model has good agreement with experimental observation.
|
