iterative versus best fit

huangyhg

We have some parts that are mostly curves (airfoil shaped).
They lie in the aircraft coordinate system at very odd angles.
When these parts are set up on CMM table, aircraft X plus may
be rotated and elevated at extreme angles to the machine axies (say 45 and 45 degrees).
As in the past this presented problems when using iterative alignment,
we generally have to go with a small section of the part and expand
from there.
We created the first two iterative alignments but when we came to
the third PC D-MIS would not align.
The part is still constrained in the fixture, so I don't think distortion is
an issue.
The CMM operator decided to use a best fit through the same six points he was trying to use for the third iterative alignment.
The best fit alignment worked.
When we re-inspect the six points they are all within .0004
Question: with a tolerance of .002, why would the iterative alignment not work?
I'm assuming it's because the iterative and best fit alignment use different algorithms?
V3.7 MR-2
Comments?