CNC Machining Surface Roughness: A Practical Guide to Reading and Specifying It

Surface roughness is one of the most important characteristics of a CNC-machined part. It directly affects how the part performs, seals, wears, and fits, and it has a real impact on machining cost. Matching the finish to what each surface actually does is what keeps a part both functional and affordable.

The challenge is that finish is easy to over- or under-specify. A tighter callout than a surface needs adds cycle time and secondary operations for no functional gain. Too loose, and a sealing or mating face can leak or fail at incoming inspection. The skill is knowing which surfaces need tight numbers and which don’t.

This guide explains what CNC surface roughness actually mean in practical terms, how cutting parameters drive those values, the achievable ranges by material, and how to specify finish on drawings without overpaying.

What is CNC Machining Surface Roughness?

Surface roughness refers to the fine-scale irregularities left by the cutting tool’s motion—peaks and valleys produced by cutter geometry and the path it traced through the material. These irregularities are invisible to the naked eye. A surface that looks and feels smooth can still fail a roughness spec when a profilometer traces across it.

Surface finish is the broader term, covering roughness, waviness, and lay direction. When a drawing calls out Ra 3.2, it is specifying roughness specifically—the measurable component that drives machining time and cost.

Machining Surface Roughness Chart: Ra Values by Application

The table below maps common Ra values to their ISO N-grade equivalents, typical applications, the process required to achieve them, and relative cost. Cost multipliers are approximate ranges, not fixed numbers. They vary by part geometry, material, feature count, and supplier. Use them for budgeting conversations, not for quoting.

The cost jump from one Ra step to the next is not linear. Stepping from Ra 3.2 to Ra 1.6 is modest—a parameter change within the same CNC process: slower feeds, lighter cuts, and sharper tooling. Stepping from Ra 0.8 to Ra 0.4 is where the economics shift, because secondary operations like grinding or honing enter the workflow. Every time a part leaves one machine and goes to another, cost and lead time compound.

A hydraulic manifold drawing that specifies Ra 0.4 on every internal surface, including non-sealing passages that only carry fluid flow, requires precision grinding after machining across the whole part. Relaxing the non-contact passages to Ra 1.6, achievable with fine CNC turning, while keeping Ra 0.4 only on the sealing surfaces and valve seats, can reduce total machining time by roughly 30 to 50 percent on a typical part of this type. Hydraulic performance stays identical.

What Does Ra 3.2 Surface Finish Mean?

Ra 3.2 is the typical default at most CNC shops when no surface finish callout is specified. Parts machined to Ra 3.2 show clean surfaces with fine, uniform cutting marks visible under close inspection but are smooth to the touch.

Ra 3.2 suits structural brackets, housings, enclosures, mounting plates, non-contact surfaces, and surfaces hidden inside assemblies. It covers the widest range of general-purpose machined parts and is the most economical Ra grade—no secondary operations required beyond standard finishing passes.

Ra 3.2 falls short on dynamic sealing surfaces, precision bearing fits, coated surfaces that depend on finer Ra for adhesion, and any surface that handles friction or fluid retention under load. When any of those conditions apply, Ra 1.6 or tighter is the appropriate next step.

What Does Ra 1.6 Surface Finish Mean?

Ra 1.6 is the first step into precision finishing on CNC parts. Tool marks are minimal, barely visible without magnification. The surface feels smooth to the touch.

Ra 1.6 suits precision mating surfaces, slow-moving fits, stressed parts, pre-coating surfaces, and static gaskets at the rougher end of their allowable range. When Ra 3.2 is not quite good enough for the function, but Ra 0.8 would be over-engineering it, Ra 1.6 is the practical middle ground.

Ra 1.6 is reliably achievable on standard milling and turning equipment with optimized parameters: slower feeds, finer cuts, and sharper tooling. No grinding is required in most cases. Cost typically increases by 20 to 50 percent on the affected surfaces compared to Ra 3.2, depending on feature type and material. That increase comes from longer cycle time per feature, not from additional process steps—which means no added setup time or secondary-operation lead time.

Achievable Surface Roughness for Common CNC Materials

The table below shows typical achievable Ra ranges for common materials, separated by milling and turning. All ranges are approximate and assume standard CNC equipment in good condition. Results depend on geometry, fixturing, feature size, and tool condition.

The pattern is consistent: softer, free-machining materials reach finer Ra values at standard parameters. Harder, tougher, or work-hardening materials need more effort, more tool changes, and sometimes secondary operations to hit the same targets. If a drawing calls for Ra 1.6 on Titanium Grade 5, expect the quote to reflect that. On Aluminum 6061, Ra 1.6 is a routine finishing pass.

Surface Roughness Parameters: Ra, Rz, and When Each Matters

Three or four parameters carry most of the surface roughness language used on drawings. Ra is by far the most common, but understanding what Ra hides and which other parameters fill in the gap is useful when specifying functional surfaces.

Ra: roughness average

Ra is the most common parameter on engineering drawings worldwide, and for most CNC-machined parts, it is the only roughness value needed. It is the arithmetic mean of surface height deviations across a sampling length, as defined in ASME B46.1 and ISO 21920.

What Ra hides matters as much as what it captures. Two surfaces with identical Ra values can have very different peak-to-valley profiles. A surface with one deep scratch surrounded by smooth terrain reads the same Ra as a surface with uniform, shallow tool marks. Ra averages everything out, which is both its strength and its blind spot.

Most CNC-machined parts fall between Ra 0.4 and Ra 6.3 micrometers. Ra 1.6 is the first step into precision finishing. Below Ra 0.8, secondary operations like grinding or lapping usually enter the picture.

Rz: average maximum height

Rz is the average of the five largest peak-to-valley heights across five sampling lengths, per ISO 21920. Where Ra smooths everything into a single average, Rz pays more attention to the extremes.

Rz matters on sealing surfaces, bearing seats, coating substrates, and any application where outlier peaks or valleys affect function. A gasket face with acceptable Ra but poor Rz might have a single deep valley that creates a leak path. For sealing and bearing applications, specifying both Ra and Rz gives a more complete picture of what the surface will actually do. On machined surfaces, Rz values commonly run four to seven times the Ra value, with significant variation depending on the cutting process.

Other parameters: Rq, Rt, Rsk

Rq shows up in optical and semiconductor specifications. Rt captures the single largest peak-to-valley distance across the entire evaluation length, relevant for safety-critical aerospace surfaces where any single outlier matters. Rsk tells you whether the surface has more peaks or more valleys, which affects oil retention in bearing and lubrication applications. For most buyers, Ra with Rz added on functional surfaces is enough.

How Machining Parameters Affect Surface Roughness

Surface roughness on a machined part comes from four inputs: feed rate, tool geometry, cutting speed, and material behavior.

Feed rate and its direct relationship to Ra

Feed rate is the single most controllable input to surface roughness, especially in turning. The theoretical relationship is straightforward: Ra is approximately f² divided by 32r, where f is feed per revolution and r is tool nose radius. Doubling the feed roughly quadruples the theoretical Ra.

Measured Ra runs consistently higher than theoretical, commonly 1.5 to 3 times the calculated value. Vibration, tool wear, and built-up edge all push the real number above the formula. The formula is still useful as a directional tool, not as an absolute predictor.

In milling, feed per tooth and stepover both affect the achievable Ra. Reducing stepover in face milling or ball-end finishing produces finer finishes, at the cost of more passes and more machine time.

The practical takeaway: reducing feed from roughing to finishing values is how most shops step from Ra 3.2 to Ra 1.6 without changing the process. The machine stays the same. The tooling stays the same. The programmer adjusts the feed rate on the finishing pass. Reaching Ra 0.8 typically requires more than a feed-rate change alone.

Tool geometry, condition, and material behavior

Tool nose radius in turning, or cutter geometry in milling, directly affects achievable Ra. A larger nose radius produces a smoother surface at the same feed rate, up to where chatter or chip evacuation problems emerge.

Tool condition matters more than most engineers expect. Flank wear above approximately 0.2 to 0.3 mm produces measurably rougher and less consistent surfaces. For production runs where finish consistency matters across hundreds or thousands of parts, tool life management becomes a finish control strategy. A shop that changes inserts on a schedule rather than waiting for visible wear holds tighter Ra across the run.

Material behavior adds another layer. Free-machining materials finish cleanly at standard parameters; work-hardening materials need more controlled conditions to hit the same Ra targets:

• Aluminum 6061, Brass C360: clean finish at standard feeds; sharp tools improve consistency
• Stainless Steel 316L: built-up edge and heat generation push Ra higher; slower speeds and sharp tooling required
• Titanium Grade 5: galling and chip re-welding risks; sharp tools and slower speeds required; secondary operations often needed for Ra below 1.6

When a supplier quotes a premium for Ra 1.6 on titanium versus aluminum, this is the reason.

Machine rigidity, fixturing, and coolant

Below roughly Ra 1.6, machine condition and fixturing stability matter as much as cutting parameters. Vibration, long tool overhangs, and resonant frequencies produce chatter marks that parameter changes alone cannot fix.

Coolant delivery affects finish by reducing thermal distortion and preventing chip re-cutting. When chips are not evacuated cleanly, they drag across the freshly cut surface and leave marks. For precision finishing, flood coolant or high-pressure through-tool coolant makes a measurable difference.

How Surface Roughness is Measured

Three measurement approaches cover most parts: contact profilometers for the standard range, optical methods for very fine finishes, and drawing-level conventions that determine what gets inspected and how.

The standard shop-floor tool is the contact profilometer. A diamond-tipped stylus traces the surface, recording vertical deviations, and the instrument calculates Ra and Rz from the profile. These instruments cover Ra 0.2 to 12.5 micrometers, which spans nearly all machined surfaces.

Measurement direction matters. A stylus traced parallel to tool marks gives a lower Ra than one traced perpendicular. On a turned part, tracing along the axis reads differently from tracing around the circumference. For functional surfaces where orientation matters, specifying the measurement direction on the drawing prevents disputes at inspection.

Non-contact and optical methods

Laser profilometers, white-light interferometers, and confocal microscopes handle surfaces below Ra 0.2, delicate materials, or applications where stylus contact would damage the surface. These are typically metrology lab tools rather than shop-floor standards. If parts need non-contact measurement, factor the inspection into lead time.

Drawing callouts and defaults

Surface roughness is specified via ISO 1302 or ASME Y14.36 symbols. Many suppliers default to around Ra 3.2 when no callout is present. If a drawing does not specify a finish, do not assume the supplier will aim for anything tighter than Ra 3.2.

How to Specify Surface Roughness Without Overpaying

The drawing is the place to control the surface finish cost. A few practical habits at the specification stage save money without giving up function.

Specify the roughest finish that meets the functional requirement

Start from the function:

• Ra 3.2 to 6.3: non-contact and non-sealing surfaces
• Ra 1.6: mating surfaces with light static loads
• Ra 0.8 or tighter: dynamic sealing surfaces and bearing surfaces
• Ra 0.4 or below: optical and ultra-precision surfaces; secondary operations are always required

The instinct to specify a tighter finish just in case adds cost that compounds across every surface and every part in the order.

Apply different Ra values to different surfaces on the same part

A drawing rarely needs the same Ra everywhere. Specifying Ra 0.8 on two sealing faces and Ra 3.2 on the housing body, mounting holes, and rear surfaces is normal practice. Use a general note for the default: “Unless otherwise specified, all surfaces are Ra 3.2 max.” Then call out the few surfaces that need tighter Ra individually. The savings come from not applying precision finishing on surfaces that do not need it.

Confirm achievability at the DFM review

Some geometry-material combinations make tight Ra targets expensive or impractical. Deep pockets, thin walls, internal radii, and hard-to-finish materials all constrain achievable Ra. A feature that is easy to finish on the outside of a part might be difficult to finish at the same Ra on the inside. A DFM review with your machining supplier should flag any Ra callouts likely to drive cost without adding function.

Specifying Surface Roughness with Your Manufacturing Partner

Getting surface roughness right saves cost and prevents rejects at inspection. The most practical place to catch it is early. If your drawings carry finish callouts, a DFM review is where specs that add cost without adding function get flagged before they reach the floor.

This is part of how Yijin Solution approaches custom CNC machining services. We review surface finish callouts at the quoting stage and point out anything likely to drive cost out of proportion to its functional benefit. Upload your CAD file with your target finish, and our engineers will return a DFM review and quote within 24 hours

FAQs on CNC Machining Surface Roughness

How do I read a surface finish symbol on a drawing?

Surface finish callouts use a check-mark symbol that holds the Ra value, lay direction, and sampling length in defined positions. The Ra value sits above the long leg of the check mark. A circle on the vertex means no material removal is allowed, which usually applies to cast or molded surfaces. Two short legs across the symbol mean material removal is required. ASME Y14.36 and ISO 1302 cover the full symbol library, including lay direction marks and waviness callouts.

What Ra should I specify for a static O-ring or gasket face?

For static O-ring face seals, Ra 0.8 to 1.6 is the typical specification, with Rz held under 6.3 micrometers to prevent leak paths through deep valleys. Soft gaskets compress under load and can tolerate Ra 3.2 in many applications. Dynamic O-ring grooves typically need Ra 0.4 or better on the dynamic surface and Ra 1.6 on the static groove walls. ISO 3601-2 covers standard O-ring face groove geometry.

Does anodizing or powder coating change Ra?

Anodizing builds an oxide layer of roughly 5 to 25 micrometers and follows the underlying machined finish closely. The post-anodize Ra is usually within 10 to 20 percent of the pre-anodize value, so the underlying finish still matters. Powder coating builds a thicker polymer film of 40 to 100 micrometers and smooths out fine machining marks, often improving Ra slightly. For surfaces with a post-coating Ra requirement, specify both the as-machined Ra and the post-finish Ra on the drawing to avoid ambiguity at inspection.

Do I need to specify Rz, or is Ra enough?

For most CNC parts, Ra is enough. Add Rz when the surface seals against another surface, runs in a bearing application, or accepts a thin coating where a single deep valley would cause failure. A common practical pairing is Ra 1.6 with Rz 8 on sealing surfaces. Specifying Rz on every surface adds inspection time without adding function.

Is Ra the same as RMS roughness?

No, Ra is the arithmetic average of surface height deviations. RMS, also called Rq, is the root-mean-square average of the same deviations. RMS values run roughly 1.1 to 1.2 times the Ra value for typical machined surfaces. Older US drawings sometimes specify RMS instead of Ra. Dividing the RMS value by approximately 1.11 gives a working conversion to Ra for practical purposes.

Back to Top: CNC Machining Surface Roughness: A Practical Guide to Reading and Specifying It