Scan results different from keyed-in points

After creating a handful of different programs (and comparing results from different alignment strategies), my problem is pretty consistent: profiles from keyed-in points are showing different results from profiles of a scan. The results are similar, but deviations appear exaggerated during the scan (TTP). I had the engineer look at my results, and he basically told me that since it was a test part we would let them go for now. Problem delayed, not solved.

Incidentally, I had been talking with the head engineer about DCI versus translating models into STEP format. He told me that their CAD software has a "rounding" problem. In short, while the print dimensions are neatly rounded, the CAD dimensions may be off a few tenths or so - in the native CAD! A few checks of my translated model compared to his native model compared to the print confirm this: the model does not have the same dimensions as the print.

This is going to start a very interesting discussion regarding whether the model or the print is master. For most of our parts, deviations this small won't matter substantially. On the part I was trying to inspect here (profiles of .0005), this makes or breaks the inspection program.

Thanks to everyone for you help.

After creating a handful of different programs (and comparing results from different alignment strategies), my problem is pretty consistent: profiles from keyed-in points are showing different results from profiles of a scan. The results are similar, but deviations appear exaggerated during the scan (TTP). I had the engineer look at my results, and he basically told me that since it was a test part we would let them go for now. Problem delayed, not solved.

Incidentally, I had been talking with the head engineer about DCI versus translating models into STEP format. He told me that their CAD software has a "rounding" problem. In short, while the print dimensions are neatly rounded, the CAD dimensions may be off a few tenths or so - in the native CAD! A few checks of my translated model compared to his native model compared to the print confirm this: the model does not have the same dimensions as the print.

This is going to start a very interesting discussion regarding whether the model or the print is master. For most of our parts, deviations this small won't matter substantially. On the part I was trying to inspect here (profiles of .0005), this makes or breaks the inspection program.

Thanks to everyone for you help.