How Should Buyers Correlate CMM, Fixture and Functional Results for Precision Castings?
Buyers should correlate CMM, fixture and functional results by measuring the same identified parts in defined states, using documented datum alignments and restraint conditions, then comparing each result to the requirement it represents. When results disagree, contain the affected parts, verify setup and method variation, run a controlled cross-method study and apply a preapproved disposition rule. A majority vote among methods is not a technical decision.
CMM coordinates, a production fixture or gauge, and assembly, leak or motion results can all be valid while answering different questions. Correlation begins by stating whether each method represents the drawing datum reference frame, the manufacturing setup, a boundary condition or the delivered function. The buyer's released drawing and acceptance plan remain the authority unless engineering approves a controlled change.
Lock Alignment, Restraint and Part State Before Comparison
A CMM program may derive datums from measured datum features, use a mathematical alignment or apply a best fit. A production fixture may locate on selected contacts and clamp a flexible casting. An assembly may pull the part into another condition through fasteners, seals or mating components. Results cannot be compared until these differences are explicit.
Record drawing revision, feature IDs, datum precedence, datum feature simulation, support points, clamp condition, measurement temperature where relevant, machining/finish state and time after release from restraint. A broad cast face that deflects under clamping may pass a fixture check and show a different free-state CMM profile. Neither result explains function without the specified acceptance state.
The post-machining review should also identify burr removal, cleaning, plugs, inserts and finishes that can change seating or leakage. Use the same serialized parts across methods so part-to-part variation is not mistaken for method disagreement.
Disagreement Pattern | Correlation Question | Controlled Check | Disposition Direction |
|---|---|---|---|
CMM fails; production gauge passes | Do datum simulation, restraint or boundary definitions differ? | Measure the same parts free and in the gauge condition; map contacts to drawing datums | Hold acceptance until the governing condition and gauge correlation are approved |
CMM passes; assembly binds | Does the program miss a combined relationship, local form or mating-part condition? | Index the assembly interference and compare it with feature-level coordinate data | Investigate drawing coverage, mating condition and method sensitivity; do not waive function |
Fixture result changes after unclamping | Is clamp-induced deformation masking free-state geometry? | Use before/during/after measurements with controlled support and clamp sequence | Define the required state or revise support/process before release |
Dimensions pass; leak result fails | Is the final boundary affected by surface damage, connected discontinuity or closure state? | Preserve the failure, verify test configuration and map the leak location to machined features | Follow functional-failure containment and root-cause evidence rather than coordinate voting |
Repeat CMM results disagree | Are fixturing, alignment, access, operator or program effects significant? | Repeat the same-part sequence under a controlled plan; apply GR&R when appropriate | Correct the measurement system before using results for product disposition |
Golden sample drifts in one method | Did the method, fixture, sample condition or environment change? | Check sample preservation and rerun reference artifacts or independent verification | Suspend routine correlation until the source is identified |
Separate Measurement-System Variation From Part Behavior
Repeat the simplest informative sequence first: same part, same operator, same setup; then deliberate changes in operator, fixture loading or program alignment. Where measurement-system risk is material and the project requires it, a gauge repeatability and reproducibility study can quantify selected equipment, appraiser and fixture contributions. GR&R design, metrics and acceptance remain project-specific; the acronym alone does not validate a method.
Next test part behavior. Compare free and restrained states, rotate the loading order, and record whether the same zone moves. Correlate feature coordinates with actual gauge contacts or assembly interference rather than comparing only pass/fail labels. For a leak discrepancy, preserve test settings and final boundary configuration before sectioning or changing the part.
The testing equipment overview can support a capability discussion, but it does not establish that a program, gauge or functional fixture is suitable for this part. Correlation evidence must come from the approved setup and represented features.
Resolve a Three-Way Conflict on a Servo Pump Housing
If a servo pump housing passes a production locator fixture but CMM reports the shaft bore tilted to the mounting plane, compare how each system establishes the plane. The fixture may clamp three feet flat while the CMM measures the part free and derives a plane from the full machined face. Matched measurements can show whether localized foot height or released warpage creates the difference.
If the assembled shaft still turns freely, that result informs consequence but does not automatically cancel the drawing requirement. Engineering reviews mating clearance, load and sealing relationships before approving any acceptance change. If the same housing fails a port leak test, the leak is handled as a separate functional nonconformance unless evidence links it to the geometric disagreement.
The controlled outcome may be a revised CMM simulation, a fixture-support correction, a drawing-state clarification or a product/process correction. Each requires approval and re-correlation on represented parts. The result is not “CMM wins” or “assembly wins”; it is one defined method for each acceptance question.
Maintain Golden Samples and Routine Correlation With Disposition Rules
During low-volume validation, select reference parts that span meaningful conforming conditions rather than one unusually favorable sample. Record their coordinate data, fixture/gauge response and applicable functional results. Preserve identity, condition and storage; a damaged or altered golden sample cannot remain an authority.
Procurement and quality should require a correlation plan that names routine check frequency or triggers, responsible parties, allowed method changes, containment steps and disposition authority. Fixture repair, CMM program revision, datum change, gauge wear, test-fixture change or unexplained golden-sample drift reopens correlation. Product acceptance resumes only after the disagreement is understood and the governing evidence path is restored.
