An Iterative Interpolated DFT to Remove Spectral Leakage in Time-Domain Near-Field Scanning

Time-domain near-field scanning is gaining more and more interest within EMC engineering to analyze electromagnetic near-fields of, e.g., quasi-stationar devices. When using a digital oscilloscope to scan the near-fields of an electronic device, the oscilloscope measures time-domain signals that comprise in most cases a large number of frequency components. For many of these components, noncoherent sampling occurs, resulting in spectral leakage when calculating the frequency spectrum of the time-domain signals with the discrete Fourier transform. This paper proposes an improved method to detect the presence of such noncoherently sampled signals as well as an iterative algorithm to obtain accurate approximations of all the frequency components with their accompanying amplitudes and phase angles. The algorithm excels over existing algorithms in obtaining these values especially in situations where several sinusoidal components are close to each other in the spectrum. This is achieved, thanks to an iterative process of removing the influence of the multiple sinusoidal components on each other. This paper contains the mathematical description of the algorithm and a numerical example evaluating the accuracy of the algorithm. The algorithm has a higher accuracy than the existing approaches, e.g., multipoint Interpolated Discrete Fourier Transform (IpDFTs), with only a slight increase of the computational cost.