This paper exploits high-resolution time-frequency analysis of servo error signals in the investigation of position-dependent dynamics in a motion system with iterative learning control. Based on the matching pursuit algorithm signals are decomposed with respect to a multiple complex dictionary of atoms. A high-resolution signal energy distribution is derived in the time-frequency plane, which does not include cross-terms, like Wigner distribution. The proposed method, applied for analysis of servo errors of a real system, provides insights in the appearance and propagation of certain low energetic system features, which can be related to position-dependent dynamics.