Your CNC parts passed dimensional inspection, but surface roughness failed. This usually happens after machining is already complete, leaving you with finished parts that can’t be assembled, coated, or approved—yet no clear explanation from the supplier.
Wrong surface roughness almost always results from missing or incorrect process control, not unclear drawings. Typical causes include the wrong machining method, skipped finishing steps, or a supplier using equipment that cannot reliably achieve the specified Ra.
The sections below break down exactly where surface roughness failures occur, how to determine whether parts can be salvaged without breaking tolerances, what proof a supplier must provide to justify the failure, and when this issue signals a deeper process control problem that requires switching suppliers.
Table of Contents
Why did CNC parts miss surface roughness spec despite clear drawing?
CNC parts miss surface roughness specs because the supplier did not treat Ra as a controlled process requirement, only as a by-product of machining.
Many suppliers assume that if dimensions are correct, surface finish will “fall into range.” In reality, Ra depends on how the final surface is generated—tool condition, cutting strategy, pass sequencing, and machine stability. If none of these are intentionally defined for Ra-critical surfaces, roughness becomes unpredictable even with a perfectly clear drawing.
This is why suppliers often discover Ra failures only after parts are finished or delivered. Without in-process verification, surface roughness is never confirmed until inspection—and by then, responsibility is pushed back to the specification instead of the process that produced it.
When we review parts with Ra requirements, the first question is not what Ra is specified, but how that Ra will be achieved and verified. If a supplier can’t answer that clearly, the failure was already built into the process before machining started.
Sourcing takeaway:
If your drawing clearly specified Ra and parts still failed, the practical issue to investigate is not documentation clarity, but whether surface roughness was ever planned, controlled, and checked during machining. That distinction determines whether rework is realistic or the batch was destined to fail.
Was wrong machining process used after parts were already cut?
Yes. Surface roughness failures usually occur because the final machining pass was not a finish-grade process capable of meeting the specified Ra.
This happens when roughing strategies are left as final passes, worn tools are reused, or cycle time is prioritized over surface control. These choices often preserve dimensions while permanently imprinting tool marks that no amount of polishing can fully undo.
Once parts are cut this way, suppliers frequently propose polishing as a fix. What’s often left unsaid is that polishing removes material unevenly and can compromise flatness, edge definition, or functional geometry—especially on sealing faces or precision fits. At that stage, Ra correction becomes a risk decision, not a simple rework.
The safer approach is preventing the issue entirely: locking the surface-creating operation in advance and checking roughness before parts move to finishing, coating, or assembly. If Ra is wrong at that point, the problem is caught early—before tolerances and geometry are put at risk.
Sourcing takeaway:
If Ra failure is discovered only after parts are complete and polishing is suggested as the solution, the real question is whether the original machining process was ever capable of meeting Ra. That answer should guide whether rework is approved or a controlled remake is the safer path.
Did supplier skip a required step that caused Ra failure?
Yes. Ra failures often happen because the supplier skipped the final surface-creating step that was required to meet your Ra—then shipped parts based on dimensional acceptance alone.
The most common “skipped step” isn’t some exotic process. It’s usually a missing finish pass, a missing controlled deburr, or a missing post-machining surface treatment that the supplier assumed was optional. If the supplier’s route sheet doesn’t explicitly call out how the Ra surface is produced, operators will default to whatever is fastest and “good enough” visually.
When this happens, you’ll notice a pattern in the supplier’s explanations: they talk about how hard the Ra is, how it “varies,” or how it “depends on material,” but they don’t show exactly what they did on the final pass. That’s the red flag. A capable supplier can point to the final operation that created the surface and explain why it should hit Ra—then back it up with measurement data.
On our side, we prevent this by forcing a simple discipline: every Ra-critical surface must have a defined final operation and a verification method before parts move downstream. If a surface finish matters functionally (seal, bearing, cosmetic face), we treat it like any other critical feature—planned and checked.
Sourcing takeaway:
If your supplier can’t clearly name the final operation that produced the Ra surface, assume they skipped or improvised the required step. At that point, the fastest way forward is to review the drawing + the failed surface photos/inspection notes and decide whether controlled rework is safe—or whether it’s smarter to remake with a defined finish process.
Your Parts Failed. Don’t Guess
Wrong Ra often gets worse after rework.
Confirm feasibility first.
Was Ra failure caused by supplier limits or unachievable spec?
Most of the time it’s supplier limits—not an unachievable spec. Truly unachievable Ra requirements are usually obvious before machining, and a competent supplier flags them with a specific technical reason.
Here’s the key difference: when the spec is genuinely unrealistic, the supplier can explain the constraint in concrete terms—surface access, geometry that prevents finishing, material behavior, or an interface that would require a different manufacturing method. When it’s a supplier limitation, you get vague language like “Ra varies,” “polish can’t guarantee,” or “this is normal,” because they don’t have a repeatable process or measurement discipline.
In practice, “unachievable” is rarely about the Ra number itself and more about the combination: Ra + access + tolerance sensitivity. For example, if the only way to improve Ra is polishing but the geometry can’t tolerate material removal, then the spec may be incompatible with the part design unless you change how the surface is generated in machining. That’s not a paperwork problem—it’s a capability and process-planning problem.
What closes the gap is a clear feasibility decision: identify the Ra-critical surfaces, identify the surface-creating operation that can hit Ra, and confirm whether rework would violate geometry. Once you have that, the decision becomes simple: salvage, controlled remake, or redesign.
Sourcing takeaway:
Don’t accept “unachievable” without a concrete explanation tied to your geometry and finishing access. If the supplier can’t explain the specific limitation and the exact process barrier, treat it as a capability gap and start evaluating alternatives immediately.
Why do suppliers claim “surface roughness varies” after parts are delivered?
Because “surface roughness varies” is often a cover for missing process control and missing verification. It’s the language suppliers use when they can’t prove what happened—or can’t repeat the result reliably.
Surface finish does vary if the shop allows it to vary: tool wear not monitored, feeds/speeds adjusted informally, finishing passes not standardized, or no roughness measurement until final inspection (or never). After delivery, that variability gets reframed as an unavoidable manufacturing reality—because admitting “we didn’t control it” is much harder.
As a buyer, the practical problem isn’t the phrase itself—it’s what it implies about the supplier’s system. If they treat Ra as “variable,” you should assume future batches will drift as well, especially when operators, machines, or tooling change. This is why Ra failures often repeat on the second order even after a “we’ll be more careful next time” promise.
The fastest way to cut through the debate is to ask for evidence that connects the result to a controlled process: what created the final surface, what changed versus prior batches, and what measurement confirms Ra before shipment. If they can’t provide that trail, you’re not negotiating a one-time mistake—you’re negotiating with a system that doesn’t protect you.
Sourcing takeaway:
When a supplier says “Ra varies” only after delivery, treat it as a process transparency warning. If you’re under deadline pressure, it’s usually safer to line up a backup quote based on a controlled finish plan than to wait for the same uncontrolled process to repeat.
Can parts with wrong Ra be salvaged — or must they be remade?
Parts with wrong Ra can only be salvaged if improving surface finish will not remove material from functional geometry or tolerance-critical features. If material removal changes how the part fits, seals, or aligns, salvage becomes higher risk than a remake.
The mistake many teams make is treating Ra correction as cosmetic. In reality, any process that improves roughness—polishing, lapping, secondary finishing—removes material unevenly. That’s acceptable on non-functional faces, but dangerous on bearing seats, sealing lands, datum surfaces, or edges that define assembly position.
This is why suppliers often say “we can polish it” without addressing consequences. Polishing may improve the Ra number while quietly breaking flatness, roundness, or edge definition. The part may measure smoother but fail later during assembly, leakage testing, or field use.
The correct decision hinges on one question: was the original surface generated by a process capable of meeting Ra without material removal? If yes, controlled rework may be viable. If no, polishing is just masking a process failure—and the safest option is a controlled remake using the correct surface-creating operation.
Sourcing takeaway:
Before approving rework, you should identify which surfaces are function-critical and whether Ra improvement requires removing material from them. If that risk can’t be clearly ruled out, salvage is usually the more expensive mistake—even if the remake feels painful upfront.
What evidence should supplier provide to explain the Ra failure?
A supplier must provide objective inspection data and a documented process explanation—not verbal assurances—to credibly explain a surface roughness failure. If they can’t show evidence, you should assume the failure is not understood and may repeat.
The first requirement is measurement proof. A capable supplier should be able to show Ra values measured on the actual failed surfaces, using a calibrated roughness instrument, with sampling locations and measurement direction documented. Statements like “it looks smooth” or “our machines normally hit that Ra” are not evidence—they’re opinions.
The second requirement is process traceability. The supplier should clearly identify:
- Which operation created the final surface
- Tool type and condition used on that operation
- Whether a dedicated finish pass was applied
- Any changes made compared to prior successful batches
If Ra failed, something in that chain changed—or was never controlled.
Finally, there must be causality, not guesswork. A legitimate explanation links process decisions to measured outcomes: what created the surface, why that method was selected, and why it failed to achieve Ra this time. Without that link, promises like “we’ll be more careful next time” have no technical value.
Suppliers who truly understand the failure will often be uncomfortable—but precise. Suppliers who don’t will stay vague, defensive, or overly reassuring.
Sourcing takeaway:
If your supplier cannot produce surface roughness measurements and a clear description of how the final surface was generated, stop negotiating on opinions. At that point, the fastest path forward is to reassess supplier capability based on evidence, not explanations.
How do you determine whether re-machining will break tolerances or geometry?
You determine rework risk by identifying whether the Ra-critical surface also controls fit, alignment, sealing, or datum relationships. If it does, re-machining almost always introduces hidden tolerance and geometry risk.
A reliable assessment doesn’t start with “can we make it smoother?” It starts with mapping surfaces:
- Which faces control position, roundness, or sealing?
- Which tolerances are bilateral vs functional?
- Which features stack into assembly performance?
If improving Ra requires touching any surface that controls these relationships, you’re no longer “fixing finish”—you’re changing geometry. That’s where parts pass visual inspection but fail functionally later, often outside the supplier’s responsibility window.
Suppliers who understand this will evaluate rework by process impact, not optimism. They’ll explain exactly where material will be removed, how much, and which tolerances could drift as a result. Suppliers who don’t will reassure you verbally—and leave you holding the risk.
The safest teams make this decision once, clearly: if rework alters controlled geometry, stop. Remake under a defined finish process is slower emotionally, but faster in total project risk.
Sourcing takeaway:
If your supplier can’t clearly explain how re-machining improves Ra without touching functional geometry, you don’t have a rework plan—you have a gamble. At that point, moving to a controlled remake is usually the most defensible engineering decision.
When does wrong surface roughness signal deeper process control problems?
Wrong surface roughness signals deeper process control problems when the supplier cannot explain, measure, or reproduce the finish consistently across batches. At that point, Ra failure is no longer isolated—it’s symptomatic.
Surface roughness sits downstream of multiple upstream controls: tooling condition, finish-pass discipline, inspection timing. When Ra fails without a documented cause, it usually means these controls are informal or operator-dependent. This is why Ra issues often resurface on the second or third order—even after a “fix.”
A reliable indicator is how quickly the supplier can diagnose the failure. Shops with real process control can usually identify the root cause within one working day, because finish strategy and inspection points are already defined. When diagnosis drags on or remains vague, the problem is rarely limited to Ra alone.
Sourcing takeaway:
If Ra failure cannot be diagnosed quickly and concretely, treat it as a sign of systemic control gaps—not a one-off miss.
How do you ensure a new supplier won’t repeat the Ra failure?
You ensure a new supplier won’t repeat the Ra failure by confirming that surface roughness is planned, verified, and controlled—not assumed.
Before committing, a capable supplier should be able to answer three questions immediately, not after production starts:
- Which operation creates the final Ra-critical surface
- Where roughness is measured before shipment
- What changes if Ra approaches the limit
Suppliers who truly control Ra answer these without escalation or internal debate. If explanations rely on averages, experience, or reassurance instead of process description, the risk profile hasn’t changed—only the supplier has.
Sourcing takeaway:
When comparing suppliers, clarity and response speed matter more than confidence. The one who can explain Ra control cleanly and quickly is usually the safer choice.
Stop Re-Machining Blind
Rework can break tolerances.
Check recovery before committing.
When should you stop negotiating and switch to a new supplier?
You should stop negotiating when explanations replace evidence and future promises replace accountability.
Negotiation is productive only while new information appears—inspection data, process changes, or clear corrective actions. Once discussions cycle through reassurance, discounts, or verbal guarantees, the supplier has reached the limit of their control system.
A practical signal is whether each conversation produces new technical facts. If it doesn’t, you’re no longer evaluating a solution—you’re managing risk exposure.
Sourcing takeaway:
Switching suppliers becomes the rational choice when negotiations stop generating new data. At that point, staying is an active risk decision.
Should you get backup quotes while current supplier attempts rework?
Yes. Getting backup quotes during rework is basic risk control—not disloyalty.
Rework consumes time and schedule buffer. If it fails, teams without alternatives lose weeks simply restarting sourcing. Backup quotes let you compare rework optimism versus a defined finish plan, before deadlines force the decision.
Suppliers with strong process control rarely object to this. Those who discourage backups often know their explanation won’t stand up to comparison.
Sourcing takeaway:
Parallel sourcing doesn’t signal distrust—it preserves options. Waiting until rework fails is the most expensive time to seek alternatives.
Conclusion
Surface roughness failures are rarely cosmetic—they expose whether a supplier truly controls their process. When explanations lack evidence, risk compounds fast. Upload your drawing and inspection report for a surface-finish feasibility review and confirm whether your Ra issue is salvageable or requires a controlled remake—within 24 hours.
Frequently Asked Questions
Often, yes—especially if the supplier frames Ra as “variable” instead of controllable. Without a defined finish process and verification step, the same issue tends to repeat when operators, tools, or machines change.
No. Polishing removes material unevenly and can compromise flatness, fits, or sealing surfaces. It’s only safe when the Ra-critical surface is non-functional and geometry isn’t tolerance-sensitive. Otherwise, polishing trades a finish problem for a performance risk.
You should expect measured Ra values taken from the actual functional surfaces, using calibrated roughness equipment, with locations documented. Visual checks or “experience-based” assurances aren’t sufficient to explain or prevent repeat failures.
No. Lining up backup quotes during rework is basic risk management. It lets you compare rework optimism against a defined finish plan and prevents schedule loss if rework fails. Strong suppliers understand this and don’t discourage parallel sourcing.
Suppliers with real process control can usually identify the cause within one working day because finish strategy and inspection points are already defined. Long delays or vague answers often indicate Ra was never controlled in the first place.
In most cases, it’s a supplier process problem. Clear Ra callouts still fail when surface finish isn’t planned, controlled, and verified during machining. If the supplier can’t show how Ra was achieved and measured, the issue isn’t documentation—it’s execution.
