| In this dissertation, a source reconstruction method is proposed to model the emission behavior of an IC by using magnitude-only information during near-field scanning. The phase-retrieval is accomplished during the back-forth iterations between two scanning planes of different heights. The magnitudes of fields on two planes of different heights are recorded and used to constrain the amplitude of the calculated fields during iterations. Three essential transformations among dipoles and fields are continued at each iteration step, until the convergence condition is achieved. The proposed algorithm has been verified by using a set of dipole-array as the original emitting sources, and also validated by simulating the emission behavior of a commercial microcontroller.;In the second paper, an equivalent dipoles based hybrid modeling method is proposed. This hybrid model consists of two groups of equivalent dipoles: one set of dipoles is to represent the radiation from pins and bonding-wires; the other set of dipoles is to mimic the contributions from the DIE and the lead-frame-flag beneath the DIE. The accuracy and flexibility of the hybrid model have been validated by internally exciting a commercial microcontroller under three different operation modes.;The third paper presents a fast and efficient calibration method which uses the same setup and instruments during calibration and measurement, and it allows for easy and economical integration of the calibration hardware and software into the scanning system. Known fields are created by a microstrip trace driven with a comb-generator. By referencing measured data to this known field, the probe factor is obtained over a broad frequency range by capturing one time-domain waveform. |
