How to Detect 14 Geometric Tolerances

• 1
  Key takeaways
• 2
  What the 14 geometric tolerance characteristics actually control
• 3
  Instruments for detecting geometric tolerances on CNC parts
• 4
  Set the datum reference frame before you measure anything
• 5
  Matching measurement method to tolerance value
• 6
  Common errors that invalidate a tolerance check
• 7
  Step by step: how to detect 14 geometric tolerances
• 8
  Detection method by tolerance family

ASME Y14.5 groups the 14 characteristics into five families: form, orientation, location, profile and runout. Form covers straightness, flatness, circularity and cylindricity. Orientation covers angularity, perpendicularity and parallelism. Location covers position, concentricity and symmetry. Profile covers profile of a line and profile of a surface. Runout covers circular runout and total runout. Four plus three plus three plus two plus two equals 14.

The family tells you how to detect the tolerance. Form controls are self-referenced, so you measure the feature alone. The other families sit on a datum reference frame, and the inspection is only as good as that frame. If the datum is wrong, a perfect part can read out of tolerance and a bad part can pass.

Each control also has a shape rule. A flatness callout applies to a plane. A cylindricity callout applies to a cylindrical surface. Applying a gauge or a probe to the wrong feature shape gives a number that means nothing, which is the most common inspection error we see on incoming drawings.

One more thing before instruments. The tolerance value in the feature control frame is a total zone width, not a plus/minus band. A 0.05 mm flatness callout means the entire surface must sit between two parallel planes 0.05 mm apart. Reading it like a bilateral tolerance doubles the allowance you think you have.

• 1
  Form: 4 controlsStraightness, flatness, circularity, cylindricity
• 2
  Orientation: 3 controlsAngularity, perpendicularity, parallelism
• 3
  Location: 3 controlsPosition, concentricity, symmetry
• 4
  Profile and runout: 4 controlsLine, surface, circular runout, total runout

A coordinate measuring machine is the default for how to detect 14 geometric tolerances in one setup. A bridge CMM with a scanning head resolves form to about 1 μm and position to a few micrometres on a 300 mm part. It needs a temperature-controlled room, typically 20 ± 1 °C, and a part that has soaked long enough to reach that temperature.

On the shop floor, a granite surface plate plus a height gauge and a dial indicator covers flatness, straightness, perpendicularity and parallelism fast. A V-block with a dial indicator reads circular runout and total runout without a CMM. A pin gauge or an air gauge checks hole size and, with a functional gauge at MMC, position.

Roundness testers and spindle-mounted rotary tables measure circularity and cylindricity to 0.1 μm when the drawing calls them tight. A laser tracker or a portable arm fits large weldments and frames, though its accuracy budget is looser than a bridge CMM.

For profile of a surface, an optical scanner or a CMM in scanning mode is the only practical route. A few touch points cannot describe a free-form surface. We collect a point cloud and compare it to the CAD model, then report the largest deviation in either direction from the nominal surface.

Instrument choice follows tolerance value, not habit. Below 0.01 mm, use a CMM or a dedicated form tester. Between 0.01 mm and 0.1 mm, manual methods on a surface plate are usually enough. Above 0.1 mm, a height gauge and a good square will do.

• 1
  CMMAll 14 controls, scanning for profile
• 2
  Surface plate setFlatness, straightness, orientation
• 3
  V-block and indicatorCircular and total runout
• 4
  Roundness testerCircularity and cylindricity to 0.1 μm

Most out-of-tolerance reports trace back to datum setup, not to the machine. The primary datum is usually a plane that contacts the datum feature simulator at three points minimum. The secondary datum contacts at two points, and the tertiary at one. That 3-2-1 sequence removes all six degrees of freedom.

Clamp the part so the datum features sit on the simulator without distortion. A thin wall squeezed in a vise will read flat and release out of tolerance. Light contact, then check with a 0.02 mm feeler gauge that the part is seated.

Record the datum targets on the inspection report exactly as they appear on the drawing. If the drawing calls A-B as a compound datum, do not measure against A alone. Reviewers reject reports that quietly substitute one datum for another.

For parts we machine, we inspect to the same datum scheme used in fixturing. That keeps the machining setup and the inspection setup aligned. It also means the first article report reflects how the part will be measured in production, not a cleaner setup that exists only in the metrology lab.

A measurement system needs roughly ten times better resolution than the tolerance it verifies. For a 0.05 mm position tolerance, the CMM needs about 0.005 mm capability, which most bridge machines reach. For a 0.005 mm cylindricity callout, a standard touch CMM is marginal and a form tester is the right tool.

We machine to ±0.005 mm and hold Ra 0.8–1.6 μm on typical surfaces. Inspection at that level needs temperature control, clean probes and a stable setup. A part measured straight off the machine at 28 °C will not match a report taken after 24 hours in a 20 °C room.

Repeatability matters more than a single reading. Measure the same feature three times, re-datum between runs. If the spread is larger than 20 percent of the tolerance, the method is not capable and the number you report is noise.

Keep the measurement uncertainty in the report. A 0.02 mm flatness result with 0.008 mm uncertainty is a different statement than the same number with 0.002 mm uncertainty. Engineers reviewing the report need that context to accept or reject a marginal part.

Probing a rough surface is the first trap. A probe tip of 2 mm radius bridges over small peaks and valleys, so the flatness you read is smoother than the real surface. For fine finishes at Ra 0.2–0.8 μm the effect is small, but on an as-machined surface at Ra 1.6–3.2 μm it can hide real deviation.

The second trap is the material condition modifier. A position callout at MMC allows bonus tolerance as the feature size departs from MMC. Inspecting it with hard gauging and no bonus will reject good parts. Inspecting an RFS callout as if it were MMC will pass bad ones.

Third, do not mix runout and concentricity. Concentricity is about the median axis of a feature; runout is about surface elements during rotation. They can disagree on the same part, and a report citing the wrong one is not evidence.

Fourth, be careful with thin and flexible parts. Clamping loads change the geometry. We fixture thin walls with light pressure and, where needed, support them from behind so the machined shape survives release from the vise.