On top of UR sign flip of the lower mass of the beamsplitter, I realized that the signs of the coil output gain filters were not following the convention described in  DCC T1200015 (not yet fully updated) and E1100108, which should be +1 for a north magnet (UR and LL), and -1 for a south magnet (UL and LR) (cf medm snapshot attached).

Although, as the test from friday describes it (alog 7677), it works as if they were right -except for UR which has its special case- (sending a positive offset should give a positive response, with the good gains enabled).

Meaning, unless there is any other possibilities, that UL LL and LR magnets of M2 of the beamsplitter are flipped.

Livingston's BS medm screen has been checked, and is correctly following the convention