When two CMM reports disagree, the question is rarely which report looks better. First establish which coordinate system the drawing authorizes, how datum features were simulated, and whether both teams evaluated identical data and settings. Best fit can be valuable for diagnosis, but it cannot quietly replace drawing-defined mating references for acceptance.
The core distinction: acceptance frame versus diagnostic frame
A CMM does not measure a part in a vacuum. It converts measured points into results relative to an alignment: a defined location and orientation of the physical part in three-dimensional space. Hexagon describes an alignment as the way a measurement system defines part location and orientation; without one, the part has translations and rotations available in space. Alignment Overview also explains that a datum reference frame constrains those degrees of freedom.
For a buyer, the practical difference is simple. A datum alignment asks whether the part meets requirements when located as the drawing says it should be located. A best-fit alignment asks what rigid transformation makes selected measured data agree most closely with nominal data under a chosen mathematical objective.
Those are legitimate but different questions. A report may look materially better after best fitting because the software redistributed deviation across the selected points. That does not automatically show that a hole pattern, sealing surface, or locating interface conforms when assembled from the designated datums.
Drawing-authorized datums or fitting constraints govern acceptance. Best fit can support diagnosis or an expressly permitted evaluation; label its purpose in the report and preserve the approved acceptance basis.
Definitions buyers should require in the report
Datum reference frame
A datum reference frame (DRF) is the coordinate framework established from the drawing’s datum references. It represents how the tolerance is intended to relate the controlled feature to datum features. The primary, secondary, and tertiary datum sequence matters because it determines which motions are restrained and in what order. A DRF is not merely a convenient set of surfaces that happen to be easy to probe.
Alignment transform
An alignment transform is the mathematical operation that maps the measured part coordinates into the nominal coordinate system. It can translate the point cloud, rotate it, or do both. The transform is not inherently correct or incorrect; its appropriateness depends on the drawing requirement and the declared evaluation purpose.
A dispute often starts when one team changes a transform but still labels the output as if it were evaluated under the original datum scheme. Require the report to identify the alignment type and the features or points used to create it.
Constrained and free degrees of freedom
A rigid part has degrees of freedom in translation and rotation. Datum alignment progressively constrains the motions that the datum structure is intended to control. The details depend on the datum features, their order, and any applicable modifiers or allowed mobility.
Do not assume every inspection must constrain every possible motion in the same way. Drawing-authorized profile controls, datum material-boundary modifiers, datum targets, and permitted feature mobility can change the interpretation. A competent GD&T and metrology review is needed where the drawing permits movement or where datum simulation is not obvious.
Best-fit objective
Best fit is not one universal calculation. It is an optimization objective applied to selected data. In PC-DMIS documentation, least-squares fitting minimizes the summed squared deviations between matching measured and nominal points, while min-max approaches optimize a different condition. The software documentation also distinguishes vector-based methods for surfaces where deviation normal to the surface is the relevant direction. The Best Fit Alignment Methods shows why changing the objective can change the residuals shown in a report.
Therefore, “we used best fit” is insufficient information. The buyer needs to know the objective, point set, weights if used, excluded regions, initial alignment, and whether the fit was free or constrained by approved references.
Why the same part can produce different CMM results
A datum-aligned report preserves the relationship of controlled features to functional references. A least-squares fit can reduce average error by spreading mismatch across a group of points. A min-max fit can respond sharply to extreme data, particularly when an outlier is present. PC-DMIS notes that min-max approaches can be sensitive to measurement uncertainty and that a single poor feature can significantly affect results. The Best Fit Alignment Methods
Lower RMS is therefore not a blanket proof of conformance. RMS is an aggregate statistic for the selected data under the selected fit. It does not establish that every feature meets its drawing requirement, that datum relationships are preserved, or that excluded points are acceptable.
Before arguing about the part, compare these inputs:
- Drawing revision, CAD revision, units, and nominal coordinate system.
- The exact datum callouts and any approved interpretation of modifiers, targets, or permitted mobility.
- Actual datum simulation: contact pattern, constructed feature, fixture relationship, and feature extraction rules.
- CMM program revision, alignment transform, fitting objective, selected point set, and any weighting or exclusions.
- Probe configuration, stylus qualification state, scan or point strategy, sampling density, feature construction, and filtering.
- Fixture, restraint condition, part orientation, cleanliness, and temperature treatment.
- Tolerance evaluation method, report template, uncertainty statement where applicable, and pass/fail decision rule.
Four dispute patterns and the question each report answers
A hole pattern looks acceptable only after best fit
Consider a machined plate whose hole pattern locates to a mounting face and locating features. A free best fit using the hole centers can make average pattern error appear smaller by shifting and rotating the cloud. But if the drawing controls hole location relative to the mounting references, the acceptance question remains: where are the holes after the part is located on those references?
The best-fit result may still be useful. It can help distinguish a global setup offset from scattered hole-to-hole variation and can guide process investigation. Label it as diagnostic evidence, retain the datum-aligned acceptance result, and do not substitute one for the other.
A freeform surface has an expressly permitted fit
A sculpted surface, blade-like form, or other complex geometry may be evaluated using a specified fitting method or constrained mobility defined by the drawing or inspection plan. In that situation, a fit may be part of the acceptance logic rather than merely a troubleshooting view.
The buyer should request a written definition of the allowed constraints: which surfaces establish the reference, which directions may move, which points are included, how normal deviation is evaluated, and which model revision is controlling. Avoid assuming that every best fit is free in all directions or that every surface profile requires a rigid datum alignment.
One outlier changes the min-max result
A single measured feature can drive a min-max result in a way that differs substantially from a least-squares result. That is not necessarily evidence of manipulation. The objectives are different. The useful next step is to examine the same underlying data with a documented outlier review: Was the point measured correctly? Does it belong to the specified feature? Is it within the intended sampling area? Is there physical damage, burr, contamination, or a data-processing issue?
Do not delete the point or the failed report simply because it is inconvenient. Preserve both the original report and any reprocessed version, including the reason for any approved exclusion.
The same data is reprocessed without cherry-picking
The cleanest dispute-resolution exercise uses a locked export of the same measured point data. Both parties then apply the same nominal model, units, datum interpretation, and stated alignment alternatives. Each result should list every included and excluded point and identify the person, software version, and program revision responsible for the reprocessing.
This separates data-collection disagreement from alignment disagreement. It also prevents a common failure mode: one report quietly measures or fits different regions and calls the comparison equivalent.
Agree the acceptance basis before inspection
For broader report planning, see drawing-based CMM inspection. The key buyer principle is that alignment choices should be agreed before a disagreement, not negotiated after a result is unfavorable.
Approval workflow for resolving an inspection dispute
- Freeze the evidence. Retain the original reports, raw or exported measurement data where available, program revisions, CAD and drawing files, photographs of the fixture condition, and the identified part. Do not overwrite or delete the failed report.
- Confirm the controlling definition. Identify the drawing and CAD revisions, units, applicable specification, and the exact characteristic in dispute. Determine whether the drawing itself authorizes a datum alignment, constrained fitting approach, or another evaluation method.
- Review datum simulation. Document how each datum feature was contacted, constructed, restrained, and ordered. Where modifiers or allowed motion affect interpretation, obtain qualified GD&T review rather than inferring unrestricted motion.
- Lock a comparison dataset. Reprocess the same points where possible. Do not change sampling, filters, point exclusions, or nominal data without documenting the change and its reason.
- Run declared alternatives. Produce the drawing-authorized acceptance evaluation and, when useful, a separately labeled diagnostic best fit. State least squares, min-max, vector method, or other objective explicitly.
- Investigate differences proportionately. Check fixture effects, probe access, point distribution, feature construction, filtering, surface condition, and environmental handling. A single disagreement is not enough evidence to diagnose a need for recalibration.
- Issue a written disposition. Record the acceptance basis, alternate analytical result, unresolved limitations, corrective action if any, responsible reviewers, and traceable references to both reports. Preserve the audit trail for future lots and revisions.
Buyer guide: RFQ questions, red flags, and checklist
Ask these questions before approving a CMM plan or reviewing a disputed report:
- Which drawing revision, CAD revision, units, and tolerance standard govern?
- Which datums establish acceptance for each critical characteristic?
- How will each datum feature be simulated or constructed on the CMM?
- Is any mobility, datum modifier, target scheme, or profile-fitting approach expressly authorized?
- Which features will be measured, and what point or scan regions are included?
- Will the report distinguish drawing-based acceptance from best-fit diagnostic analysis?
- What report format, data retention, and review path are required for a nonconformance?
Red flags include a report with no alignment description; a best-fit result labeled as datum-based acceptance; unexplained excluded points; different CAD revisions between parties; a color map without its fitting settings; or a request to discard the original failed report.
Use this compact checklist before release:
- The part identifier and revision match the approval package.
- The report states the controlling drawing and CAD revisions and units.
- Datum features, order, and simulation are documented.
- The alignment type and best-fit objective, if any, are visible.
- Point selection, filters, exclusions, and feature construction are traceable.
- Acceptance and diagnostic outputs are clearly separated.
- Both original and reprocessed reports are retained when results differ.
- The final disposition identifies the authorized acceptance basis.
A drawing review can also benefit from a drawing-based datum selection guide, especially before a part reaches first article or supplier transfer.
FAQ
Can a best-fit CMM report be used for final acceptance?
Only when the drawing, contract requirement, or approved inspection plan authorizes that fitting method and its constraints for the characteristic being evaluated. Otherwise, treat it as diagnostic analysis and use the drawing-defined reference frame for acceptance.
Why does least squares often make a part look better?
Least squares minimizes aggregate squared deviation across its selected data. It can distribute error rather than preserve the functional relationship to drawing datums. A lower aggregate residual does not prove every individual requirement conforms.
Is datum alignment always a fully fixed rigid setup?
No. The drawing may authorize constrained motion, datum targets, material-boundary effects, or profile-related conditions that require careful interpretation. The correct approach is the one supported by the controlling definition, not a blanket rule.
What should we do if supplier and customer reports use different alignments?
Freeze both reports and compare the same data, model revision, units, datum simulation, point selection, and software settings. Reissue results with the authorized acceptance alignment and any diagnostic fit clearly distinguished.
Does one CMM disagreement mean the machine needs recalibration?
No. Alignment logic, fixture condition, sampling, feature construction, filtering, probe access, environmental handling, and uncertainty can all contribute. Investigate the complete measurement process before assigning a cause.
Put alignment agreement into the inspection plan
For CNC parts with functional datums, the most economical dispute is the one prevented at drawing review. Specify the acceptance frame, the role of any best-fit analysis, the report fields, and the record-retention expectation before production approval. SUUXIANG can review the drawing, model, critical datums, and required inspection documentation when you discuss your project requirements.