When a supplier and a receiving inspector report different surface roughness values on the same CNC part, compare the measurement recipe before changing the machining process. The parameter, filter, evaluation length, stylus, trace direction and exact measurement location all belong in that comparison. A second reading is useful only when both parties know what was actually measured.
This guide explains how to build a repeatable roughness correlation record for a purchasing or quality dispute. It complements a drawing specification; it does not select a universal roughness limit. For a new CNC machining project, agreeing on the inspection method before first-article approval is usually easier than negotiating it after shipment.
Separate the part requirement from the measurement recipe
The drawing states the intended surface requirement. The measurement recipe explains how an instrument evaluates that requirement on a particular feature. Keep those two records connected by drawing revision and feature identifier, but do not assume that a parameter name by itself fully defines a test.
The ASME B46.1 contents distinguish roughness, waviness and lay, with sections covering several measurement and comparison methods. Use the standard and edition specified by the contract. A reference to an older edition should not silently become a different specification because one instrument has newer default software settings.
For example, an Ra requirement on a sealing face is not automatically a requirement for the same measurement on every outside surface. Mark the controlled region on a drawing image. Identify whether the result applies before coating, after finishing, or after another operation. Link that region to the existing surface roughness callout rather than extending its scope informally.
Build a side-by-side settings record
Ask both laboratories to export their settings with the results. A photograph of the displayed number rarely includes enough information to reproduce it. A useful comparison sheet has a column for each instrument, a difference column, and an owner for resolving each difference. Record automatic settings explicitly instead of writing “default.”
| Record | What to capture | Why it helps |
|---|---|---|
| Part and feature | Serial or sample ID, revision, marked trace location | Separates sampling differences from equipment differences |
| Parameter and units | Full parameter name, definition and reporting units | Prevents comparisons between different quantities |
| Filtering | Filter type and all enabled wavelength settings | Makes the separation of profile components reproducible |
| Length | Evaluation length, traverse length and excluded regions | Shows whether comparable surface area was evaluated |
| Detector | Instrument, stylus specification and skid configuration | Identifies differences in how the profile was acquired |
| Setup | Trace direction, fixture, leveling and surface condition | Documents the practical conditions of the reading |
| Evidence | Profile file, individual results, date and operator | Allows review without repeating an undocumented test |
Check filters before comparing the final number
A roughness instrument processes a measured profile. The cutoff and filter determine which spatial components contribute to the reported result. Mahr’s surface texture parameter reference describes why the selected cutoff and scanning distance matter. Consequently, two reports with the same parameter label can still describe different evaluations.
Do not solve this by selecting whichever cutoff produces a passing result. First identify the governing specification and the instrument’s documented interpretation. If the drawing is incomplete, have the design authority approve a method clarification before using it for acceptance. Preserve the original results so the clarification does not erase the history of the disagreement.
A practical comparison begins with one controlled surface and one agreed recipe. If instruments cannot use equivalent settings, document that limitation and obtain a suitable correlation method from the metrology team. A mathematically convenient conversion between unlike measurements is not evidence that they are interchangeable.
Control direction, location and access
A trace along machining lay does not cross the same topography as a trace across it. On a turned shaft, a convenient access direction may differ from the specified inspection direction. On a narrow shoulder, the available surface may be shorter than the required traverse. These are method questions, not reasons to conceal an abbreviated measurement.
Use the actual instrument manual to establish detector access, supported measurements and setup restrictions. A fixture that works for a broad planar coupon may not provide a valid arrangement for a small curved feature.
Choose repeatable trace positions using features that both teams can identify without marking the functional surface destructively. Record a photograph or sketch of each track. Note whether the sample was cleaned, whether protective oil was removed and whether damage was present. Distinguish a local scratch from the general texture question instead of allowing the scratch to disappear from the inspection record.
Run a correlation study that preserves individual results
Select retained samples that can travel between the two inspection locations without being refinished. Use stable identification and protective packaging. Include samples relevant to the disputed process rather than relying only on a smooth reference specimen. A reference check can support instrument confidence, while the actual component still tests fixturing, access and location control.
Have each team measure the agreed locations with the agreed recipe and save the individual readings. Repeat placement where the inspection plan requires it. Compare within-setup variation before discussing the difference between laboratories. A large spread from one setup calls for investigating that setup before calculating a correction factor between sites.
Calibration status also belongs in the evidence package. NIST’s traceability guidance treats traceability as a property supported by a documented measurement chain and associated uncertainty. A sticker alone does not establish that the selected method is suitable for this feature. Consider the uncertainty and decision rule alongside the result, as discussed in FUMING’s measurement uncertainty acceptance guide.
Use a simple disposition table: settings mismatch, location mismatch, unstable measurement, confirmed part variation, or unresolved. Assign the next action to the appropriate owner. This makes it harder for an unresolved metrology issue to become an unsupported demand for re-machining.
Keep roughness acceptance separate from cosmetic approval
Two surfaces with similar numerical roughness can look different under directional light. Conversely, a visually consistent surface does not prove that its measured texture meets a functional requirement. Keep the dimensional and surface-texture report separate from the approved cosmetic sample, lighting conditions and viewing criteria.
The Mahr surface parameter reference lists different parameters for different aspects of a profile. That variety is a reason to specify the quantity needed for the function, not to substitute a more favorable parameter after a rejection. If the current specification fails to describe a real functional problem, revise the engineering requirement through controlled approval.
For a first-article package, include the original recipe, results and any agreed changes in the same controlled folder as the CAD and drawing revision review when that document applies. The essential requirement is a traceable connection between the tested part and the accepted definition.
What to send with a FUMING inquiry
Send the drawing revision, material and finish requirements, marked measurement regions, expected inspection method and any sample correlation reports. State whether the concern is sealing, sliding contact, appearance or another function. Include both disputed results rather than only the rejected value. Preserve the sample identity alongside the material and lot traceability record when material condition is relevant. FUMING can review the manufacturing inquiry using the supplied requirements; a quotation should identify any unresolved method or access questions before production begins.
Frequently asked questions
Can I compare Ra values from two different instruments?
Yes, when their methods and settings are suitable and sufficiently comparable for the stated requirement. Confirm parameter definition, units, filters, length, detector configuration, direction and location. Different instrument models alone neither prove nor disprove agreement.
Should we average supplier and customer results?
Not simply to resolve a disagreement. Averaging results from different recipes can hide the cause. Establish a common method, understand the variation and follow the approved acceptance rule before deciding disposition.
What if the feature is too short for the specified trace?
Record the access limitation and request an approved method or specification clarification. Do not report a shorter trace as though the full prescribed method was completed.
Does passing a reference specimen prove the part passes?
No. It supports a defined instrument check. The part still needs its own suitable setup, trace locations and acceptance evaluation. Preserve both records because they answer different questions.
