A CNC inspection result close to a tolerance limit needs an agreed decision rule. The displayed dimension alone does not explain the measurement uncertainty or how a buyer and supplier should handle a borderline result. Before ordering, define the feature, the measurement method, the uncertainty information required and the acceptance rule. That preparation makes a later disagreement easier to resolve without changing the drawing after parts have been made.
This FUMING guide addresses purchasing and quality teams sourcing CNC machining services. It focuses on near-limit dimensional decisions, rather than the complete first article inspection process. The examples below are hypothetical; they are not statements of FUMING production capability, measured performance or customer acceptance policy.
What measurement uncertainty tells a CNC buyer
A measurement produces an estimate of a defined quantity. Uncertainty describes the dispersion associated with that estimate, using the information available. It is not simply the last digit on a display, and it is not permission to enlarge a drawing tolerance. The JCGM vocabulary entry for measurement uncertainty provides the formal definition and distinguishes uncertainty evaluation from a casual guess about measurement error.
Three numbers therefore have different jobs. The drawing limits describe the acceptable product characteristic. The reported result describes the measurement estimate. The uncertainty describes knowledge of that estimate. Keep all three visible when discussing a critical diameter, thickness or location result. A purchase order that requests only a green pass cell may conceal the information needed for a defensible release decision.
Define the feature before comparing results
Start with the drawing revision and characteristic identifier. A local two-point shaft diameter, an average of several sections and a fitted cylinder can describe different aspects of the same physical surface. Measurements taken at different locations may legitimately differ. Asking which instrument is more accurate will not resolve a disagreement if the teams measured different quantities.
Specify the part condition as well. Record whether inspection occurs before or after finishing, whether a thin component is restrained, and which datum system applies. Refer to the CNC tolerance and datum review guide when the drawing itself leaves these choices unclear. Resolve definition questions before asking a laboratory to quantify uncertainty.
Ask for uncertainty that belongs to the measurement task
An instrument calibration certificate is useful evidence, but the uncertainty of a production measurement may include more than the calibration contribution. The practical question is whether the evaluation represents the actual method, part and conditions. NIST’s measurement uncertainty overview explains why uncertainty is associated with a measurement result and a measurement model.
For a critical feature, ask the inspection team to identify the applicable uncertainty estimate and explain its scope. A short controlled method reference is more useful than an unexplained number copied from a machine brochure. The following review prompts help a buyer obtain that explanation; they are not a substitute for a metrologist’s uncertainty budget.
| Review item | Question for the inspection team | Record to retain |
|---|---|---|
| Instrument and reference | Which calibration and reference artifact apply? | Instrument identity and relevant calibration reference |
| Method | Which locations, alignment and evaluation settings define the result? | Controlled inspection instruction |
| Environment | How are temperature effects and stabilization addressed? | Applicable conditions and compensation assumptions |
| Part interaction | Could contact, restraint, surface condition or geometry affect the reading? | Setup and feature-specific considerations |
| Reporting | Is the value standard or expanded uncertainty, and what coverage information accompanies it? | Units, uncertainty and stated coverage basis |
Do not impose a universal instrument-to-tolerance ratio without checking the contractual requirement and measurement task. A ratio can be a screening tool, but it does not explain every risk in a particular measurement. Equally, more displayed decimal places do not establish that an inspection method is suitable.
Agree on the decision rule before production
A decision rule states how the measurement result and its uncertainty enter a conformity decision. Some agreements use simple acceptance at the specification limits. Others move an acceptance boundary inward through a guard band. Different rules allocate the risk of accepting a nonconforming part or rejecting a conforming part differently. The choice belongs in the agreed inspection requirements, not in an improvised discussion after delivery.
ILAC G8 guidance addresses decision rules and statements of conformity for laboratory work. The JCGM guides collection includes JCGM 106 on uncertainty in conformity assessment. Use the applicable contract and standard to select the rule with competent quality personnel; this article does not prescribe one rule for every CNC order.
A hypothetical near-limit example
Suppose a drawing permits a shaft diameter from 9.980 to 10.020 mm. The reported result is 10.018 mm with expanded uncertainty U = 0.004 mm under the laboratory’s stated coverage conditions. The estimate lies inside the drawing limits, while the interval formed by the result plus and minus U extends beyond the upper limit.
Under a simple acceptance rule based on the reported value, that result may be accepted. Under an illustrative inward guard band equal to U, the upper acceptance boundary would be 10.016 mm, so 10.018 mm would not satisfy that acceptance condition. This does not prove that the shaft’s actual diameter is outside specification. It shows why the agreed decision rule matters. Treatment of a result that is not accepted must also follow the documented rule.
Neither party should choose whichever rule produces the preferred commercial outcome. If the contract is silent, pause the disposition and agree how the uncertainty will be handled. Preserve the original report and document any subsequent decision.
Use a controlled response to a borderline result
The first response should be an evidence review, not repeated measurement until a passing value appears. Retain the initial reading and identify the affected part or lot. Confirm units, feature definition, rounding, revision and inspection condition. Then decide whether additional measurement is justified and what it is intended to establish.
- Contain: identify the affected items and their release status without assuming every item in a lot shares the result.
- Compare: review the supplier and buyer methods against the same characteristic definition.
- Investigate: distinguish a setup issue, a part variation issue and a decision-rule disagreement.
- Re-measure deliberately: use an agreed method and preserve both the original and follow-up records.
- Authorize disposition: record who may accept, reject, rework or approve a concession.
Where bore measurements disagree, the bore gauge correlation guide provides related process context. Correlation work and uncertainty evaluation answer connected but different questions. A close match between two readings alone does not establish a complete uncertainty evaluation.
Put the agreement into the purchase and inspection records
Create a compact characteristic register for features whose acceptance could affect assembly or product function. Each entry should name the drawing requirement, inspection stage, method reference, required uncertainty information, decision rule and escalation owner. Review this register alongside the first article inspection checklist so the first delivery establishes a usable reporting format.
Ask the supplier to state exceptions before order acceptance. For example, a supplier may be able to measure a feature but need outside laboratory support for the requested reporting scope. Resolve the scope, cost and timing explicitly. Do not mistake a general calibration statement for an agreement to provide every requested conformity statement.
NIST’s traceability guidance also explains that traceability alone does not guarantee fitness for purpose. For a buyer, the practical implication is to review whether the evidence supports this feature and this decision, rather than treating a reference to a national standard as a complete acceptance argument.
Frequently asked questions
Does uncertainty change the drawing tolerance?
No. The specification limits remain the product requirement unless the authorized design owner changes them. An acceptance boundary used by a decision rule is a separate concept. Keep the original limits, the chosen rule and any approved concession separately documented.
Should every dimension receive the same reporting requirement?
Not automatically. Prioritize characteristics according to function and contractual obligations. Apply consistent requirements where required, but avoid creating a large report that obscures the few decisions needing special attention. The quality team should approve the scope before production.
What should accompany an inquiry to FUMING?
Provide the controlled drawing and model, material and finish, order quantity, critical characteristic list, inspection stage and any mandated decision rule. Include the expected report format and unresolved questions. Use the FUMING quotation form to request a project review so the measurement and reporting scope can be discussed with the manufacturing requirements.
