The CISPR of IEC in European and the ANSI in American use the model of site attenuation to evaluate and quantify the quality of OATS. The ANSI also uses the concept of Normalized Site Attenuation (NSA) to eliminate the effects brought about by different antennas, which may cause the inaccuracy of site attenuation. To stringently require the quality of OATS, the measured values of NSA have to be compared with the theoretical ones. If their differences are within (+/-)4 dB at all frequencies, the test environment of the OATS can be accepted. At present, the theoretical values of NSA are used as a standard for verifying the OATS in the world. However, many non-ideal factors derived from certain assumptions and simplifications of the NSA may cause the measured NSA to deviate from the theoretical values. To understand the manner and extent of the effects, in this thesis we find out the non-ideal factors by studying the NSA by numerically simulating the effects of each factor on NSA measurement by the method of moment (MoM), and the results are analyzed and discussed. We find that there is a difference of 6 dB on the NSA when the impedance of the receive antenna is mismatched. Meanwhile, we also study the relationship between the conductivity of non-perfect electrical conductor and the ideal value of NSA. We conclude that a ground plane made by metal can be regarded as a PEC one. However, the values of NSA will increase if the conductivity of ground plane is below 1000 S/m. Besides, considering the humid Taiwan climate in particular, we investigate the condition of a thin layer of water covering the ground plane after rain and the results show that the effect of the water plane will decrease the values of NSA. As for the measurement at a distance of 3 m, a deviation of 2 dB at low frequencies arises from the effect of near field for the case of horizontal polarization. Finally as for the non-conducting table in OATS, our results show that the larger values of relative permittivity and conductivity of table result in the larger change of NSA value. The difference can be 3 dB for the case with a wooden table. With results of our research, the effects of non-ideal factors on NSA measurement can be comprehended clearly. The comparison of the differences also can be used as a reference for the modification of the NSA measurement in the future.

This thesis first discusses the ground plane effects of a test site on the antenna factors (AFs) of hybrid antenna (biconical log-periodic dipole array). Meanwhile, the effects of mutual coupling between antenna and its image, and the variation of active phase center are also discussed. From these analyses, a hybrid method, based on the modified SSM (Standard Site Method) and the PCPM (Phase Center and Pattern Matching) applied to the hybrid antenna for NSA (Normalized Site Attenuation) measurement is proposed. By this method, the low geometry- dependent AFs of hybrid antenna can be obtained to produce more reasonable NSA values for a test site. Secondly, this thesis proposes a simple, fast, and accurate method to calibrate the free-space AFs of broadband EMC (Electromagnetic Compatibility) antennas. This method adopts a fixed-height configuration and a MUSIC (MUltiple SIgnal Classification) algorithm. This configuration significantly shortens measurement time and removes height-dependent calibration errors. Meanwhile, the MUSIC algorithm can remove unexpected reflections from the ground plane or any other reflecting objects, by which the free-space AFs can be calculated. In addition, this method can also automatically compensate for the phase center shift, which makes measurement easier and more convenient. To verify this method, the calibrated results are compared with other published standard methods: the mean differences can be as low as 0.25 dB for the LPDA (log-periodic dipole array), 0.42 dB for the hybrid antennas, and 0.36 dB for the horn antennas. Finally, this thesis provides a method of using two equivalent negative inductances from two terminals of three coupled inductors to reduce the parasitic inductances of a typical three-capacitor EMI (Electromagnetic Interference) filter. Theoretical analysis and formula deduction for the design of two equivalent negative inductances are demonstrated. The experimental results show that the insertion losses of a three-capacitor EMI filter at 50 MHz can be reduced by 16.8 dB for the DM (differential-mode) and by 19.2 dB for the CM (common-mode). In Appendix A of this thesis, an extended study of the effect of ground plane on antenna¡¦s radiation is described. A simple V-shape edge-groove design for a finite ground plane can effectively reduce the pattern ripples of a monopole. The optimal design of proposed structure can reduce the peak-to-peak pattern ripples from 26 to 4.5 dB.

This thesis is focused on a detailed analysis of the mechanisms of electromagnetic wave propagation in anechoic chambers. The analysis takes into account the influence of the measurement distance, setup of anechoic chamber and other parameters. Furthermore, the work deals with a validation of both types of anechoic chambers and examines the impact of selected parameters on the measurement results. Based on the theoretical analysis and practical measurement, the comparation between semi and fully anechoic chamber, or a three and ten meter measuring distance was made. This thesis also contains a detailed analysis of sources of the uncertainty.