A 3D scanner records the whole surface of a legacy part as a point cloud, and a handheld laser scanner states about 0.025 mm accuracy on its datasheet; a tactile coordinate measuring machine (CMM) measures single features to about 2 µm over 100 mm, and a digital caliper to ±0.02 mm. Scan the part for free-form shape, measure every feature that carries a fit by contact, and take neither the tolerances nor the material from the result.
What does each method measure on a legacy part?
Hand tools measure one dimension at a time by contact: a caliper for lengths, widths and depths, a micrometer for diameters and thicknesses. Their strength is that each reading is a single, traceable number a person can check. Their limit is that every dimension has to be chosen in advance.
A tactile CMM touches the part at discrete points and fits geometry to them: planes, cylinders, cones, spheres. The software then reports sizes, positions and the relation between datums. This suits bores, seats, flats and hole patterns, where the drawing will need exactly such elements.
An optical 3D scanner projects laser lines or a fringe pattern and records millions of points without contact. The result is a mesh of the whole visible surface, which can be compared with a CAD model as a colour map. It sees only what the sensor sees: blind bores, internal threads and undercuts are not captured. ISO 10360-13 defines the acceptance test for optical 3D systems and applies when gloss and colour of the scanned object stay in a cooperative range. A datasheet figure therefore does not automatically hold for a polished or black surface, so ask the supplier how the part is prepared.
How accurate are calipers, CMMs and 3D scanners?
The figures below come from manufacturer datasheets for one instrument per method. They show orders of magnitude, not what a particular supplier achieves.
Table 1. Stated accuracy by method, manufacturer datasheets
| Method and example | Stated accuracy | Range or condition |
|---|---|---|
| Digital caliper, Mitutoyo 500-196-30 | ±0.02 mm (E MPE); ±0.04 mm on the sliding surface (S MPE) | 0–150 mm, step 0.01 mm |
| External micrometer, Mitutoyo 293-240-30 | ±1 µm (E MPE) | 0–25 mm, step 0.001 mm |
| Bridge CMM, Mitutoyo CRYSTA-Apex S 544 | 1.7 + 3L/1000 µm, L in mm: 2.0 µm at 100 mm, 2.9 µm at 400 mm | SP25M probe, 20 ± 2 °C, measuring range 500 × 400 × 400 mm |
| Handheld laser scanner, Creaform HandySCAN 3D EVO | 0.025 mm; volumetric 0.020 mm + 0.040 mm/m, so 0.040 mm over 500 mm | Resolution 0.025 mm; "based on ISO 10360" |
Source: Mitutoyo product pages 500-196-30 and 293-240-30 and Bulletin No. 2202(3); Creaform HandySCAN 3D EVO series page; as of 2026-10-30.
The CMM figure holds under stated conditions. For the same machine, Mitutoyo guarantees 1.7 + 3L/1000 µm in a room at 20 ± 2 °C with a temperature change of at most 2 °C per hour, and 1.7 + 4L/1000 µm in a room at 16 to 26 °C with a change of at most 5 °C per 24 hours. Workshop conditions sit closer to the second case or beyond it.
Why can two accuracy figures not be compared directly?
Because they come from different tests. ISO 10360 is a series with one part per sensor type. ISO 10360-2:2009 covers Cartesian CMMs with contacting probing, ISO 10360-8:2013 covers Cartesian CMMs with optical distance sensors, and ISO 10360-13:2021 covers optical 3D coordinate measuring systems. As of 2026-10-30, ISO has confirmed parts 2 and 8 (in 2026 and 2024) and lists part 13 as under systematic review since 2026-07-15. The Creaform page says its accuracy is based on ISO 10360 without naming the part.
Older optical scanner datasheets quote VDI/VDE 2634. Part 2 (2012-08) and Part 3 (2008-12) of that guideline are withdrawn, and DIN recommends DIN EN ISO 10360-13:2023-11 instead. A figure that cites VDI/VDE 2634 refers to a withdrawn test.
Three other differences matter:
- Resolution is not accuracy. Creaform lists resolution and accuracy separately, both 0.025 mm for the EVO. ZEISS lists a point distance of 0.03 to 0.12 mm for the ATOS Q 12M and gives no accuracy figure on that product page. Point spacing says how dense the points are, not how close they lie to the true surface.
- Temperature. Drawing dimensions are specified at the standard reference temperature of 20 °C (ISO 1:2022). A part measured at another temperature needs compensation, which the Mitutoyo CMM applies from sensors on the scales and on the workpiece.
- Length dependence. The CMM and scanner errors grow with measured length. A figure for 100 mm does not carry over to 1 m.
When you compare quotes, ask for the test the figure comes from, the part of the standard and the room conditions.
Which method fits which feature?
Choose the method for each feature of the part.
Table 2. Method by feature on a legacy part
| Feature | Method | Reason |
|---|---|---|
| Bearing seat or bore with a fit, such as Ø25 H7 | Micrometer, bore gauge or tactile CMM | A 0.025 mm scanner accuracy exceeds the IT7 band of 0.021 mm |
| Datum faces, hole positions, perpendicularity | Tactile CMM | Fitted geometry gives size and relation between features |
| Cast or forged free-form surface, ribs, blade profiles | Optical scan | Millions of points, comparison with CAD as a colour map |
| Overall length, width, thickness with a wide tolerance | Caliper or micrometer | Single readings, error far below the tolerance |
| Blind bore, internal thread, undercut | Contact measurement or gauges | An optical sensor does not see into them |
| Wear distribution over the whole surface | Optical scan of worn and unworn part | The whole surface is compared at once |
Source: working practice; accuracy figures from Table 1; IT7 band for 18–30 mm from ISO 286-1:2010.
How does measuring accuracy relate to the tolerance?
The instrument error has to be small against the tolerance band. Mitutoyo's CRYSTA-Apex S bulletin gives a rule of thumb: the measuring uncertainty should be no more than one fifth of the dimension tolerance, ideally one tenth.
Take a Ø25 mm bore with an H7 fit. For nominal sizes over 18 up to 30 mm, grade IT7 is 21 µm wide (ISO 286-1:2010), and one fifth of that is 4.2 µm. The CMM limit at 25 mm is 1.7 + 3 × 25 / 1000 = 1.8 µm, which fits. A caliper at ±0.02 mm spreads over 40 µm, and the scanner at 0.025 mm is of the same size, so neither can judge this bore. Now take a 20 mm length that relies on ISO 2768-m. The class gives ±0.2 mm for sizes over 6 up to 30 mm, a band of 0.4 mm and a fifth of 0.08 mm. A caliper and a scanner both meet that.
An instrument error limit is one term of the full uncertainty, which also contains probing, fixturing, temperature and the form of the part, so the margin left is smaller than these numbers suggest. ISO 14253-1:2017 sets the rules for deciding conformity when a result lies within the uncertainty of a limit. See ISO 286 fits and IT grades for the band widths and ISO 2768 general tolerances for the general classes.
What can neither method tell you?
Both record geometry in the present state of the part. Neither says what the original dimension was, which deviation is wear and which is design, what the material is, how it was heat treated, or what tolerance the function requires. A scan of a worn bearing seat reproduces the wear exactly. The nominal size has to be taken from unworn surfaces, and the fit from the mating part.
Material grade and treatment come from a chemical analysis and a hardness test, listed as a separate item on any reverse engineering job. Reverse engineering spare parts describes what that analysis can and cannot establish and how tolerances are derived from function.
How do you combine the methods on one part?
Use each method where its error is smaller than the tolerance and its coverage is larger than the alternative:
- Scan the whole part to capture free-form surfaces and the position of features.
- Measure the functional features by contact: fits, datums and threads with micrometer, gauges or CMM, on unworn areas.
- Build the CAD model from fitted geometry and set nominal values, using the mesh as a reference and not as the model itself.
- Check the first part made against the model with the same methods, and report it with its measurement uncertainty. Inspection reports with CMM and Cpk covers what the report should contain.
Frequently asked questions
- Is a 3D scan accurate enough to replace CMM measurement?
- Not for features with a tight tolerance. The handheld laser scanner used as the example here states 0.025 mm, while a bridge CMM from the same comparison states 1.7 + 3L/1000 µm, which is 2.0 µm at 100 mm. For a fit such as H7 on a 25 mm bore, whose tolerance band is 0.021 mm, the scan is too coarse and contact measurement is needed.
- How accurate is a handheld 3D scanner?
- Creaform states 0.025 mm accuracy and 0.020 mm + 0.040 mm/m volumetric accuracy for its HandySCAN 3D EVO series, so about 0.040 mm over 500 mm. These are manufacturer datasheet values, not results from your part, surface and workshop.
- Which standard tests the accuracy of an optical 3D scanner?
- ISO 10360-13:2021 specifies acceptance and reverification tests for optical 3D coordinate measuring systems. The guidelines VDI/VDE 2634 Part 2 and Part 3, which older datasheets cite, are withdrawn, and DIN recommends DIN EN ISO 10360-13:2023-11 instead.
- When is a digital caliper enough?
- When the tolerance band is wide compared with the caliper error. A 0–150 mm digital caliper has a stated maximum permissible error of ±0.02 mm, which is well inside one fifth of the 0.4 mm band that ISO 2768-m allows on a size between 6 and 30 mm.
- Can a scan or CMM report give the material or the tolerances of a legacy part?
- No. Both record geometry only. Material grade, heat treatment and tolerances have to be established separately and written on the drawing.
Sources
- ISO 10360-2:2009, Acceptance and reverification tests for CMMs, linear dimensions (ISO catalogue)
- ISO 10360-8:2013, CMMs with optical distance sensors (ISO catalogue)
- ISO 10360-13:2021, Optical 3D coordinate measuring systems (ISO catalogue)
- ISO 13385-1:2019, Design and metrological characteristics of callipers (ISO catalogue)
- ISO 3611:2023, Design and metrological characteristics of micrometers for external measurements (ISO catalogue)
- ISO 14253-1:2017, Decision rules for verifying conformity or nonconformity with specifications (ISO catalogue)
- ISO 1:2022, Standard reference temperature for geometrical and dimensional properties (catalogue entry, genorma)
- NIST, The 2016 revision of ISO 1: the standard reference temperature remains 20 °C
- ISO 286-1:2010, ISO code system for tolerances on linear sizes (ISO catalogue)
- ISO 286-1:2010 standard tolerance grades, IT7 for 18–30 mm (table reproduced on Wikipedia)
- ISO 2768-1:1989, General tolerances, linear and angular dimensions (ISO catalogue)
- VDI/VDE 2634 Part 2:2012-08, withdrawn, recommended standard DIN EN ISO 10360-13:2023-11 (DIN Media)
- DIN EN ISO 10360-13:2023-11, relation to withdrawn VDI/VDE 2634 Parts 2 and 3 (DIN Media)
- Mitutoyo, Digital ABS AOS Caliper 500-196-30, E MPE and S MPE (Mitutoyo UK shop)
- Mitutoyo, Digital Micrometer IP65 293-240-30, maximum permissible error (Mitutoyo Europe shop)
- Mitutoyo, CRYSTA-Apex S series, Bulletin No. 2202(3), S 500 series accuracy table
- Creaform, HandySCAN 3D EVO series, technical specifications
- ZEISS, ATOS Q, technical data (point distance)
Questions on a drawing? sales@ertas.engineering