A (and probably B, C) seem to probe differently than XYZ. If you give A similar steps/revolution as the Z axis, and give the probe function similar search_distance as Z, the g2 board goes into stat:2 (alarmed) and complains that the move is too small.