In conventional ToFD, operators are taught that diffracted echoes from the top and bottom of a planar defect have opposite phase and such defects can thus be discriminated from a pair of slag lines. This phase-inversion test is therefore used as a tool for characterizing an indication. When using non-symmetric probe arrangements and less-conventional beam angles, the situation becomes less clear-cut. Early in the development of the ToFD technique in the 1980s and 1990s, theoretical models enabled calculation of the amplitude and phase of diffracted signals from a crack edge for any incident and diffracted angle. Experimental studies at the time gave results that supported the theoretical predictions in idealized situations, but showed that for finite pulses and finite probe dimensions, it was more difficult to confirm the predictions for phase. In an inspection of a weld in a high-integrity component for the Hinkley Point C power station an asymmetric ToFD design was adopted as a solution to an access restriction. Comparison of top/bottom phase of indications was used as part of the procedure for characterization of indications. An analysis is made of the reasons why certain asymmetric configurations could not be used to identify planar defects using top-bottom phase inversion. The resulting re-design of the ToFD inspection configurations for a non-symmetrical inspection surface is presented, in which phase inversion remains a robust tool for defect characterization.