Calipers vs. micrometers: tolerance, accuracy and uncertainty

Start with the feature and tolerance

A caliper is versatile for checking several kinds of dimensions quickly. An appropriate micrometer provides a more specialized contact arrangement for its intended feature. Neither tool name guarantees that a reading is good enough to accept a part. Start with the drawing: what feature is being measured, what tolerance applies and what decision depends on the result?

A rough stock check and a close-tolerance shaft inspection are different jobs. Access, surface condition, alignment and the measurement procedure can be as important as the instrument specification. Select the method before selecting the number of display digits.

Describe what “diameter” means in the inspection plan. A single two-point reading is a local size between contacts. It does not establish roundness, cylindricity or every diameter along a shaft. Likewise, an outside micrometer cannot verify the position of a hole relative to a datum. A drawing may require several different measurement methods even though all results use millimeters.

Separate process feedback from final acceptance. A quick caliper reading can show whether roughing stock remains; a controlled measurement can decide whether the finished bearing seat meets its requirement. This division avoids forcing one general-purpose tool to answer every question. Browse the metalworking and machining department with the inspection method in mind, not just the machine purchase.

Three numbers that should not be confused

A concrete example is Mitutoyo’s MDH-25MB brochure: it specifies selectable 0.1 µm resolution and ±0.5 µm instrumental error at 20°C, with the stated instrumental error excluding a further ±1 display-count quantization term. Those values belong to that instrument. They demonstrate the distinction; they are not typical specifications for every micrometer.

Contact and alignment can change the reading

Mitutoyo’s caliper guidance identifies jaw tilt, measuring force, temperature differences and small-hole geometry as potential error sources. Ordinary calipers lack a constant-force mechanism. Clean contacts and a zero check are useful, but zeroing does not calibrate the instrument across its full range.

Use a consistent, trained measurement technique. A burr, chip or tilted jaw can make a neat digital reading misleading. For a micrometer, use the force-control method specified for that model; squeezing harder is not a way to obtain a more certain result. Confirm that the contact geometry is suitable for the feature rather than forcing a standard outside micrometer into an unsuitable job.

Why jaw depth creates an alignment problem

In an idealized offset measurement, angular motion θ creates an error approximately e = hθ for small angles measured in radians, where h is the distance from the scale axis to the contact line. At an illustrative 40 mm offset and 0.0002 rad tilt, e ≈ 0.008 mm. This is a geometry example, not an error assigned to every caliper. It explains why identical display electronics can give different results when contact position or jaw loading changes.

NIST’s dimensional metrology lecture describes Abbe offset, contact geometry and the need to define the measured quantity. A micrometer aligns its measuring screw with the contacts more directly, but still has alignment, force, surface and temperature influences. Instrument architecture reduces particular errors; it does not abolish metrology.

A warm part is a different measurement condition

A machined part and the measuring instrument may not be at the same temperature. The usual dimensional reference is 20°C. NIST research by Dennis Swyt explains why uncertainty in both temperature and the coefficient of expansion matters when measuring away from that reference. Applying a correction does not make either uncertainty disappear.

For a tight tolerance, record part and instrument conditions and allow the approved stabilization process. A universal ten-minute waiting period is not defensible: geometry, material, airflow and starting temperature differ. Avoid subtracting 0.00575 mm from an actual reading unless the material, temperatures, reference and compensation model have been established.

Choose a measurement method

Avoid an automatic “ten times better” slogan as the complete decision. Quality requirements may specify the acceptance method, uncertainty or gauge capability. Follow those requirements and involve metrology support for consequential measurements.

What goes into an uncertainty budget?

NIST Technical Note 1297 describes combining standard uncertainties with covariance where appropriate. This is different from adding every manufacturer’s maximum error as though it were a standard deviation. A useful budget starts with a model of the measurement and includes the influences relevant to that method.

Before using a real budget, establish how each value was obtained and whether contributors overlap. A repeated reference check may already include several environmental effects; adding the same effects again can double-count them. Conversely, short-term repeatability can omit long-term drift and operator differences. Reporting expanded uncertainty also requires a justified coverage factor and stated interpretation. A bare root-sum-of-squares calculation does not establish a probability of conformity.

A number inside the limits can still require review

Suppose a drawing permits 24.980–25.020 mm. Assume the approved measurement process reports 25.015 mm with expanded uncertainty U = 0.008 mm. Its illustrative interval is 25.007–25.023 mm, which crosses the upper limit. This is not proof that the part is bad. It means the reported measurement and its uncertainty do not support an unqualified decision under a rule that requires the whole interval to fall within the limits.

That rule is an example for explaining risk, not a prescribed replacement for a customer contract or quality procedure. Specify the decision rule before measuring. Also record what U represents and how it was evaluated; an unexplained plus/minus number should not be treated as a confidence statement.

When the caliper and micrometer disagree

Do not average two conflicting instruments and call the midpoint more accurate. First determine whether they are measuring the same feature under comparable conditions. Repeated readings can reveal short-term scatter, but repeating the same biased setup does not cancel systematic error. Preserve the original results so the investigation can explain the discrepancy.

Calibration and checks between measurements

Calibration establishes information about an instrument’s behavior; it does not prevent subsequent damage or misuse. Check condition, cleanliness, zero and required reference checks under the organization’s process. Record the instrument identity when traceability matters. A dropped instrument or an unexplained change needs evaluation before further acceptance work.

For workshop planning, pair this article with workholding and lathe tooling. Better measurement can reveal a setup problem; it cannot correct the machining process by itself.

NIST’s calibration interval guidance does not prescribe one interval for all instruments. It points to required accuracy, stability, environmental conditions and external requirements. Keep as-found results: a correction made during calibration matters for earlier measurements as well as future use.

Questions from the inspection bench