Most electromagnetic interference standards specify that measurements of radiated emissions must be performed in the far field (FF), e.g. at an open-area test site or in a semi-anechoic chamber. Since near field (NF) measurements are cheaper, quicker and more flexible compared to FF tests, establishing a correlation between NF and FF data is of great research interest. One strategy to achieve this goal is to find a set of basic radiators comprising electric and magnetic dipoles that generate the same NF as the original source at selected observation points. This set of dipoles, based on the uniqueness theorem, can then be used to predict the FF radiation patterns. The uniqueness theorem requires that electric or magnetic fields are matched on a closed surface with respect to the magnitude and phase. The focus of this thesis is the investigation of FF prediction based on NF magnitude-only data. In this thesis, a robust NF-FF conversion model based on Genetic Algorithms (GAs) is built up to predict the radiation of printed circuit boards (PCBs). This is done by introducing a dipole moment magnitude range pre-selection before the initialisation step of GAs, customising the processes of selection, crossover and mutation for anti-sticking and checking the correlation between NF and FF fitness values. Since the performance of GAs is tightly related to the number of dipoles in the GA model, FF characteristics of generic radiation sources (such as a long wire and a large loop) are analysed using both analytical calculation and source modelling by GAs. For structures with simple FF patterns, if more dipoles than necessary are used, the computational cost of GAs is unnecessarily high. On the other side, for structures with complicated FF patterns, the GA modelling may not be able to well approximate the FF radiation, due to the limitation for GAs to tackle too many unknowns. Therefore the scope of the model applicability is discussed, and a dipole number N, depending on the electrical size of the source, is recommended for GA modelling. By applying GAs to get the equivalent dipole set of a radiating PCB from the magnetic NF magnitudes, NF sampling approaches are investigated in detail, including where to locate NF sampling planes, what plane coverage angle to choose, how many points to observe, what type of data to collect, what dynamic range to allow for the data, and how many planes to choose. Two case studies are presented for predicting the FF radiation of PCBs from magnetic NF magnitude-only observations, and validate the NF sampling approaches in this thesis.
| Identifer | oai:union.ndltd.org:ADTP/246428 |
| Date | January 2009 |
| Creators | Fan, Hongmei |
| Publisher | Western Australian Telecommunications Research Institute, University of Western Australia. School of Electrical, Electronic and Computer Engineering |
| Source Sets | Australiasian Digital Theses Program |
| Language | English |
| Detected Language | English |
| Rights | Copyright Hongmei Fan, http://www.itpo.uwa.edu.au/UWA-Computer-And-Software-Use-Regulations.html |
Page generated in 0.0018 seconds