We are running a rather long probe build and using it some horizontal wrist angles. The problem is that during both calibration and execution of the program, the probe will start "dancing". It might take a hit, but then it will start moving back and forth in a short stroke. Then the light goes out in the head, almost as if the probe because dislodged, but it hasn't because we can see that it is not. See attached probe build. We even swapped out the TP20 and that didn't help. Any ideas? I'm about at my wit's end. 
