Here is part of another post from Rotorheads, my bold.
Maybe I can help clear some of the confusion, which I think comes from the report referring to both nuts involved as ‘the castellated nut’ and part of the servo link as ‘the carrier’.
I will refer to the nut at the spider/ duplex bearing end as ‘the big nut’, and the nut at the servo end as ‘the smaller nut’. The carrier is not to be confused with a bearing carrier. It is actually a hinged block of metal arrangement with a hole in it that the pitch control rod passes through, 'a pin carrier', and forms the attachment of the pitch link rod to the servo feedback link.
The tail rotor rotates anti-clockwise (looking from the righthand side of the aircraft) at about 5000 rpm. It is directly driven (no clutch) by two engines that can provide in the region of 1000 shp each.
The big nut is at the spider/ duplex bearing end of the pitch rod, the smaller nut at the servo link end. Both are cotter pinned, and the smaller nut is also locked with locking wire. They are both conventional RH threads.
The pitch rod doesn’t rotate, as has been explained in previous posts, but the tail rotor and the spider are rotating around it. The duplex bearing in the spider (the pitch rod/ spider interface if you like) enables this to happen.
The tightness of the two nuts at each end is what prevents the pitch rod from rotating freely under normal operation. There is no key, or other method to prevent it turning if there was enough force.
What the report is saying is that the duplex bearing failed, for whatever reason, and stopped doing its job of allowing the spider to rotate freely around a stationary rod. The rod then started rotating with the spider. Because the rod would rotate anti-clock wise, it tightened the big nut up as far as it’s cotter pin would allow, and then the big nut would continue to rotate with the rod. At the smaller nut end, the effect would be the opposite. The locking wire and cotter pin would prevent the nut from undoing, and the smaller nut would turn with the rod. This caused friction against the carrier, eventually welding the smaller nut to it. The nut is now part of the fixed servo link. The rod would be turning against a cotter pin and locking wire, which it seems, from the report, that it sheared, and wound itself out of the nut.
The pitch change rod would now be no longer attached to the servo link to provide feedback of its position, and the servo would continue to move in the direction of its last flying control input.