Where the shift comes from
When a datum feature of size is referenced at a material boundary, its simulator is fixed at that boundary rather than closing on the produced feature. If the produced datum feature has departed from the boundary, the part is loose on the simulator by that departure, and that looseness is the datum shift.
The number is a diametral clearance, bounded by the size tolerance of the datum feature, and it is zero when the datum feature is referenced with no modifier or produced outside its own limits. A planar datum feature can never contribute shift at all, because a plane has no size to depart from.
Why it is not simply additive
Shift is available in some direction, not in every direction at once. A part that is loose by 0.08 mm on diameter can move 0.08 mm one way, and doing so uses up the freedom that would have helped a feature displaced the other way.
There is also only one part and one frame, so a pattern of six holes shares a single displacement. Letting each hole borrow the shift in whichever direction suits it accepts patterns that no gauge and no assembly would accept, because the part cannot be in six places at once.
How DatumLab reports it
The strict acceptance limit is the stated tolerance plus the bonus, and that is the number the verdict uses. The shift is reported separately, with a clearly labelled best case bound that adds it, and a result sitting between the two comes back as a review rather than an accept.
That is the honest presentation. A tool that folds shift into one acceptance number cannot tell the user whether a part passed on solid ground or on an assumption about direction that nobody checked.
Settling a shift assisted case properly
That is the procedure a functional gauge performs mechanically, and on a coordinate measuring machine it is a best fit within the shift available, often done as a paper gauge analysis.
- Judge every feature against the strict limit first: the stated tolerance plus its own bonus, with no shift.
- If any feature fails that, look for a single displacement of the datum reference frame, inside the shift available, that brings the whole pattern inside its zones at once.
- Apply that one displacement to every feature in the pattern. There is one part and one frame, so a feature cannot borrow the shift in a direction that suits it alone.
- If no single displacement works, reject the pattern, however good the individual features look.
- If one does, record which displacement was used. The next inspector has to be able to reach the same verdict.
A hole at MMC with a datum feature at MMB
The controlled hole is 10.00 to 10.30 mm with position diameter 0.20 mm at MMC, produced at 10.10 mm and 0.17 mm off in X. Datum feature B is a hole 8.00 to 8.10 mm at MMB, produced at 8.08 mm.
| Quantity | Value |
|---|---|
| Bonus on the controlled feature | 0.100 mm |
| Strict acceptance limit | 0.300 mm |
| Datum shift available | 0.080 mm |
| Best case limit if the shift lies the right way | 0.380 mm |
| True position measured | 0.340 mm |
| Verdict | review-datum-shift |
The measured value sits above the strict limit and below the best case, so this is a review rather than an accept. Nothing about the feature has been decided until the whole pattern has been checked in one displaced frame.