Correct metrological calibration is the key to avoiding the main systematic error sources in an interferometer configuration in order to obtain high resolutions and repeatability. The calibration procedure can consist of the acquisition of two fringe patterns with a known phase shift between them, which means the correct adjustment of this phase difference fits the system. Differential phase shifting algorithms, built by combining two known phase shifting algorithms in a non-linear way, obtain this phase difference directly. In this sense, the two-dimensional characteristic polynomial is an innovative tool for qualitatively characterizing the properties of the differential phase shifting algorithms and, particularly, for determining the accuracy of the system. In previous works, it was demonstrated that sensitivities are inherited from precursor phase shifting algorithms. This mechanism of error propagation allows us to design new differential phase shifting algorithms according to standard requirements in each system in order to cancel or at least minimize this known source of errors.