TP confusion

[MFCguy]

We are running a 5 hole pinion carrier on our CMM with 3.5.
I am confused on whether it is allowable for the X,Y coordinates to change from when the location was reported and when the TP (with MMC) was calculated. Shouldn't the center of the feature being measured to the datums be in the same location?
I am also confused on why some of the angles of the holes are reading exactly the same from part to part? (isn't possible with our process) We have tried to turn the part 72 degrees, to keep one of the pinion holes at the 12o'clock position.
I read the TP posts below in the forum and it is still hazy to me.

Thanks for any help

[Matthew D. Hoedeman]

Pcdmis will 'shift' the part around, away from your alignment, to use the MMC of the 'datum' features instead of adding the datum MMC bonus to the feature you are reporting.

[kbotta]

And I'll add, yuo really should not be using {M} on the datum feature(s) for varification in 3.5. Jan claims it works correctly in 4.1, but I have yet to do my own assesment. In 3.5 as Matt said, pcdmis will shift the feature's, and if there is enough tolerance (M) the parts will show perfect. Is is trying to resemblt what a functional gage wuold do. Attribute "pass" if you will.
Kev

[than_357]

A simple check- look at the dimensional info for the feature, then at the coordinates on the True P dimension. If they don't match, PC-DMIS is putting modifiers into the equation and usually simulating gaging. If it's a small batch inspection it might be easier to report it as RFS and manually figure bonus tolerances if the results are out of tolerance.

Tom H>