Fast spherical near-field antenna measurement methods

Spherical near-field antenna measurement is an established method for characterizing the electrical properties of antennas. Compared to far-field measurements, however, a mathematical transformation of the near-field data is necessary to determine the desired far-field characteristics. In order to obtain accurate and complete transformation results, it is necessary to acquire near-field data on a closed surface (e.g. a sphere) around the test antenna. This significantly increases the measurement time compared to far-field measurements, which is particularly relevant in view of the increase of mobile communication devices and the associated measurement burden. In this thesis, different methods for accelerating spherical near-field antenna measurements are investigated and evaluated. First, the theory of the spherical wave expansion is explained. It is shown that the transformation time is not a relevant factor for most antennas. Usually the measuring chamber is the limiting resource and, therefore, it is beneficial to increase the complexity of the transformation to reduce measurement chamber utilization. After introducing the theory and the transformation algorithms, several possibilities for accelerating the near-field measurements are investigated. A simple approach is the reduction of measurement points by truncation of the measurement surface. This method aims to determine only certain, selected areas of the far field correctly. Due to the lack of information however, approximation errors always occur. The main contribution of this thesis to the state-of-the-art is the investigation of non or minimally redundant sampling grids and the development of the associated transformation methods. It is shown that the number of measurement points can always be minimized by a suitable selection of the transformation origin. Furthermore, it is comprehensively analyzed which scanning grids enable time-efficient and accurate measurements. For this purpose, two typical measurement systems (roll-over-azimuth and robotic arm system) are considered. It is shown that the respective measuring system has a significant influence on the measurement duration depending on the different sampling grids. Using the proposed methods, a measurement time reduction between 5% and 50% has been achieved for the investigated examples. The determined uncertainties show that this does not reduce the measurement accuracy significantly. A further reduction of the measurement points leads to an underdetermined system of equations whose solution is generally not unique. However, using Compressed Sensing methods, it is still possible to correctly determine the spherical mode spectrum under certain circumstances. These methods are not covered by this thesis. In summary, different acceleration methods can be used and combined. Truncation can considerably reduce the measurement time in certain cases, but has the disadvantage that the antenna properties can only be approximated and not completely determined. In comparison, non-redundant sampling grids reduce the measurement time without significantly reducing the measurement accuracy. In general, the spherical mode spectrum can be determined with the developed methods from measured data on any closed measurement surface. The methods presented can therefore also be used to develop novel measuring systems that are adapted to the corresponding measuring task, since a spherical measuring geometry is generally not required