14.2 Datum Shift and Worst-Case Boundary Calculations

14.2 Datum Shift and Worst-Case Boundary Calculations

Quick Answer: When a datum feature of size is referenced at MMB (or LMB), the datum simulator can shift relative to the part. The available datum shift = |actual datum size − MMB|, capped by the datum feature's own size tolerance. Datum shift translates the entire datum reference frame (DRF) — it does not add to each controlled feature's position tolerance the way bonus does.

The datum shift formula

For a datum feature of size referenced at MMB:

If the datum is referenced RMB (RFS), datum shift is zero — the simulator conforms to the actual datum feature with no looseness. If referenced LMB, replace MMB with LMB throughout; shift then comes from features departing from LMC.

How shift differs from bonus

• Bonus is per-feature. It grows as that one controlled feature departs from MMC and is added to that feature's position tolerance only.
• Datum shift is per-DRF. It translates the entire DRF (and therefore every feature controlled to that DRF) by the same amount in the same direction. You cannot give different features different datum shifts.
• Both are bounded by a size tolerance, but they apply at different levels of the tolerance stack.

This distinction is the single most-tested datum concept at Senior level. A typical distractor adds datum shift directly to a feature's bonus and reports a wildly large position tolerance; the correct answer applies bonus to the feature and shift to the DRF separately.

Worked example — hole at MMC with datum B at MMB

A plate has datum A (planar primary), datum B = hole Ø8 ± 0.1 referenced at MMB (secondary), datum C (planar tertiary). A controlled hole is Ø5 ± 0.2 with position Ø0.1 at MMC to A|B(M)|C. Actual measured values: controlled hole = Ø5.1, datum B hole = Ø8.0.

Step 1 — controlled feature (the Ø5 hole)

• MMC = 4.8, LMC = 5.2, size tol = 0.4.
• VC = MMC − pos tol = 4.8 − 0.1 = Ø4.7 (mating pin boundary).
• Bonus at actual Ø5.1 = |5.1 − 4.8| = 0.3; total position tolerance = 0.1 + 0.3 = Ø0.4.
• The hole's axis may lie within a Ø0.4 zone centered on its true position (defined by the DRF).

Step 2 — datum B (Ø8 hole at MMB)

• MMB of datum B = MMC of datum B = Ø7.9 (smallest datum hole; no datum geo tol given).
• Actual datum B = Ø8.0, so datum shift = |8.0 − 7.9| = Ø0.1 of DRF translation.
• Maximum datum shift (if datum B produced at LMC Ø8.1) = |8.1 − 7.9| = Ø0.2 = full datum size tol.

Step 3 — total location freedom (worst case)

• The hole's axis can deviate from its true position by Ø0.4 (its own zone with bonus).
• The true position itself can shift with the DRF by Ø0.1 (current datum shift) up to Ø0.2 (max).
• For assembly, the mating pin must clear the controlled hole's VC = Ø4.7 AND the locating datum's shift. The worst-case location of the controlled hole's axis relative to the part's other features = radial 0.2 (own zone) + radial 0.1 (current datum shift) = 0.3 radial, or Ø0.6 diametral equivalent.

Step 4 — why you don't simply add them as "position tolerance"

If two holes are both controlled to A|B(M)|C, the datum shift moves them together. Hole 1 going its way and hole 2 going the opposite way is not possible from datum shift alone — both true positions shift by the same vector. The exam often offers an option that doubles the shift per feature; that is wrong because the DRF moves once, not per feature.

Worst-case boundary reasoning for assembly

For a fixed-fastener assembly ( mating pin in two holes ), the worst case is:

1. Both holes at MMC (smallest) → tightest position requirement (no bonus).
2. Both holes at their VC for assembly clearance.
3. Datum shift on the locating datums in each part moves the two patterns in opposite directions in the worst case.

The assembly boundary is the sum of each part's VC clearance plus each part's datum shift. If the parts share locating datums built at MMB, the assembly can still go together because datum shift accommodates the variation. This is exactly why functional gaging (14.3) works.

Common traps

1. Treating datum shift as per-feature bonus. It is a DRF movement; one shift applies to all features on that DRF.
2. Forgetting the cap. Datum shift is bounded by the datum feature's size tolerance, not by the controlled feature's tolerance.
3. Adding shift to bonus and reporting a single "mega tolerance." They combine as a stack with different units of application, not as a simple sum on the feature.
4. Using MMC instead of MMB. MMB is the boundary the datum simulator is built at, which usually equals the datum's MMC but can be modified (e.g., MMB modified by a datum's own geo tol, or MMB = LMC for an LMB reference).
5. Ignoring RMB default. Per Y14.5-2009, a datum feature of size with no modifier defaults to RMB — zero shift. Don't add shift when no MMB/LMB symbol is shown.