A comprehensive study is performed to investigate the performance of a non-uniform circular array interferometer in a real time 3-dimensional direction finder. The angular range of view is supposed to be 65 degrees vertically and 120 degrees horizontally, which is suitable for airborne applications. Interferometer is designed to work in the S, C and X bands. Regarding optimization process, the interferometer employs an eight element non-uniform circular array along with a phase reference antenna at the center of the array. Several quantities and parameters are studied, e.g., frequency behavior, origins of phase measurement errors, Signal to Noise Ratio (SNR) effect on phase measurement, and the effect of the phase measurement error on direction finding performance. The proposed interferometer is able to tolerate at least 35 degrees of phase measurement error. Radius of the array is determined to be 22 cm in order to have good frequency response in the desired frequency band. Both Generalized Regression Neural Network (GRNN) and Maximum Likelihood (ML) estimation are applied for mapping the phase relationships between antennas to the Direction of Arrival (DoA). The results of two methods are well matched, and therefore validation is performed.