PC-DMIS 2022.1. Part has fairly conventional alignment (datum A plane, datum B circle projected onto datum A, datum C line to another circle). Features on interest are holes in different orientations. 40-50% of holes' FCF true positions read "error" on report window, both for size and position. Features' actual values all populate, whether LSQ or max inscribed calculations are used. If I add legacy dimensions (with or without ASME-correct alignment), results populate fine.
My first inclination is that some combination of features is deemed "unacceptable" for use in GeoTol to the math in Y14.5.1-2019 which applies to the Y14.5-2009 standard.
You can align anything you want in an alignment, and you can dimension anything you want in Legacy.
GeoTol gives me issues with some things not being valid, like using a mid-point for a rotational datum spinning around a shaft for the keyway. I guess it wants me to make a mid plane, but keyways are not that big that I can effectively get two opposing planes.
My guess is this is what you are seeing, though why it is selective I couldn't tell you. It is new, could be buggy still, a bit.
You can not use a helix measured cylinder for position in GeoTol (2021.2 anyway) most of the time, once in a while it allows me to do it, so I just stopped using helices. Maybe some of your holes you scanned a helix while others you scanned (or probed) concentric circles?
That is the one thing I know won't fly, I'm sure there are others.
Makes helix scanning fairly useless, since I'm not going to scan a hole with circles to report the position then scan a helix to report the form and I doubt many other people will either.
It is more likely to be the local size option that doesn't allow helical scans (unless you are using ISO). Under ASME, local size defaults to the "Circular Elements" method which requires the hits to be in discrete circular cross-sections, the "opposed points" method may allow a helical scan (depending on the pitch of the helix) because it interpolates between the actual hits to derive the opposed points. Turning local size off would also allow a helically scanned cylinder to be reported for ASME position.
Error message is "Datum A does not contain any features". This really doesn't make any sense, since all of these 100+ holes are exactly the same ABC feature control frame.
The only way this can happen would be if PC-DMIS is recalculating datums for each and every dimension, instead of "DATDEF" simply creating a file to be used throughout the program.
BFRE; it's not the datums, since many of the FCF dimensions work fine. In fact, simply re-running the program, makes different FCFLOCs fail.
The specific error is "Datum A does not contain any features"; I would think that would make any FCF dimension with datum A fail to calculate, not just some.