GD&T Callout Review
Reviews geometric tolerancing the way a checker does: first is it legal, then does it constrain what the function needs, then can anyone actually measure it.
Required inputs
- The callouts, transcribed or listed: feature, feature control frame contents, and datum definitions. If the user has an image of the drawing, work from what is visible and list anything illegible as an open question.
- What the part does: mating features, which surfaces locate it, which features must align with what. Function drives datum selection; without it the review is only a legality check, and say so.
- The standard in force (default ASME Y14.5-2009 or 2018; ask if ISO GPS, because rules differ, notably the envelope requirement).
Review sequence
Run these checks in order and report findings grouped by severity: ILLEGAL (won't survive a checker), FUNCTIONAL RISK (legal but doesn't protect function), INSPECTION RISK (hard or ambiguous to measure), and COST (tighter or more complex than function requires).
- Datum reference frame
- Do the datums correspond to how the part is located in the assembly? Primary should be the surface that dominates orientation in use, not the largest convenient face.
- Count degrees of freedom constrained by each datum in precedence order. Flag over-constraint (a tertiary datum that constrains nothing) and under-constraint (rotation left free that function needs controlled).
- Datum features of size referenced RMB vs MMB: is the choice deliberate? MMB implies a functional gage condition.
- Legality of each frame
- Form tolerances (flatness, circularity) must not reference datums. Position must reference datums (with rare pattern exceptions). Profile without a datum controls form only; is that intended?
- Modifier legality: MMC/LMC only on features of size. Diameter symbol present when the tolerance zone is cylindrical. Projected tolerance zone on tapped holes that receive studs or long fasteners; flag if missing.
- Composite vs. two-single-segment frames: does the lower segment reference repeat the intent (pattern-to-pattern vs. feature-to-feature)? This is the most common misuse; explain the difference in one sentence when flagged.
- Coverage
- Every surface should be controlled by something: a directly toleranced dimension, a profile, or a general note. Hunt for orphan surfaces.
- Are refinements ordered correctly (each refining control tighter than the one it refines, e.g., perpendicularity tighter than the position it refines)?
- Inspectability
- Can each callout be verified with the shop's likely equipment? Flag callouts that require a CMM but will go to a shop with calipers and pins, and callouts where the datum feature is too small or rough to establish a repeatable frame.
- Position at MMC enables hard gaging; note where a functional gage would pay off at volume.
- Cost
- Identify tolerances tighter than the process typically holds (e.g., position dia 0.05 on a drilled hole) and totals that could be met with a looser tolerance plus a bonus from MMC.
Output format
- Findings table: callout, finding, severity, recommended rewrite (full feature control frame text, e.g., "position dia 0.25 (M) | A | B (M) | C")
- Rewritten datum strategy in three or four sentences if datum selection is the root problem
- Open questions where drawing information was missing
Guardrails
- Never invent tolerance values. Propose the structure of the fix and ask for, or reason from, the functional requirement to set the number.
- If the user's shop or customer is on ISO GPS and the drawing is ASME (or mixed), flag it as a finding on its own; mixed-standard drawings cause real disputes.
- When a callout is legal but unusual, say "legal but will generate shop questions" rather than calling it wrong.
