A computational and experimental investigation of λ/2 and λ/4 sampling step in phaseless planar near-field measurements at 60GHz

While complex planar near-field measurements involve a formal sampling step requirement (less than 0.5λ, typically 0.45λ) phaseless planar near-field measurements with the two-scans technique do not. In phaseless measurements only the magnitude is measured; while the plane wave spectrum of the square of the magnitude has twice the bandwidth of the complex signal, the magnitude itself may have an even much larger bandwidth.

However, the spectrum of the magnitude is not of significance for phaseless measurements with the two-scans technique where the measured magnitude is combined with an initial guess for the phase to form a complex signal. But this combined complex field may also have a plane wave spectrum that is broader than that of the measured complex signal and thus extend beyond the visible region of the spectral domain.

This will happen if the initial guess for the phase does not fit properly to the measured magnitude; e.g. if the phase jumps by 180deg. at a point where the measured magnitude in non-zero. For this reason, the phaseless measurement may require a sampling step notably smaller than λ/2 to avoid aliasing effects in the visible region of the spectral domain.

This work, based on experimental and simulated planar near-field measurements of a 60GHz horn, investigates the effect of the initial guess for the phase on the plane wave spectrum of the combined complex signal. Furthermore, it investigates the effect of the initial guess for the phase and different sampling steps on the phase retrieval, using the Iterative Fourier Technique, and thus on the resulting far-field radiation pattern.