Machined surface finish guide for metal parts
What a machined surface means
A machined surface is the surface left on a metal part after material has been removed by turning, milling, drilling, boring, grinding, honing or a similar cutting process. For buyers, designers and machinists, the important question is not whether the surface looks shiny, but whether its texture supports the part’s function. Roughness, waviness and lay can affect sealing, friction, fatigue behavior, coating adhesion, wear and assembly fit.
A practical specification connects the required finish to the work done by that face, bore, shaft, slot or sealing land. This guide explains the main measurements, typical process ranges and drawing practices used for machined surfaces in hardware and metal components.
For related articles in this topic area, see the Surface Finishes category.
Surface texture is more than roughness
In everyday shop language, “surface finish” and “surface roughness” are often used as if they mean the same thing. They are related, but they are not identical. Surface finish, more formally called surface texture in many standards, includes several features at different scales.
- Roughness refers to the fine, closely spaced peaks and valleys left by the cutting edge, abrasive grains or finishing media.
- Waviness describes broader undulations that may come from machine vibration, part deflection, thermal movement or fixturing effects.
- Lay is the dominant direction of the surface pattern, such as the circular marks from turning or the linear marks from grinding.
- Flaws such as scratches, dents, burrs, tears and embedded chips may be visually obvious even when a roughness reading appears acceptable.
Standards such as ASME B46.1 describe surface texture in terms of roughness, waviness and lay. ASME Y14.36 covers how surface texture controls are designated on engineering drawings, while ISO 21920-2:2021 specifies profile-method terms, definitions and parameters. The ISO catalogue for ISO 21920-2 lists earlier ISO 4287 entries as withdrawn, but older drawings and supplier documents may still refer to ISO 4287. For that reason, the drawing standard, parameter and units should be read together instead of assuming that every “Ra” callout is interpreted in the same way.
How Ra, Rz and lay are used
The most common roughness value for a machined surface is Ra. Ra is useful because it condenses many height deviations into one number, which makes it convenient for drawings, inspection reports and supplier communication. Its limitation is that it averages the profile. Two surfaces can have the same Ra while one has sharp isolated peaks and the other has rounded valleys. For sealing, fatigue-sensitive parts or sliding components, that difference can be important.
| Term | What it describes | Practical use |
|---|---|---|
| Ra | Arithmetic average roughness of a measured profile | General-purpose surface roughness callouts for machined parts |
| Rz | Peak-to-valley height measure based on the selected standard and evaluation rules | Useful when high peaks or deep valleys are functionally important |
| Rq | Root mean square roughness | More sensitive to larger deviations than Ra |
| Lay | Dominant direction of tool or abrasive marks | Important for sealing, sliding direction, visual appearance and fluid flow |
| Cutoff and evaluation length | Filtering and length conditions used during measurement | Needed for consistent inspection between supplier and customer |
Ra is often reported in micrometers or microinches. The unit conversion is 1 micrometer equals about 39.37 microinches. A finish of Ra 1.6 micrometers is therefore about Ra 63 microinches. Confusion between micrometers and microinches is one of the easiest ways to create an unrealistic or incorrectly inspected finish requirement.
Rz should not be converted from Ra by a fixed universal ratio. Some rough comparison charts use approximate relationships, but the ratio depends on process, tool marks, material behavior and filtering. A ground surface, a turned surface and an EDM surface can have different peak shapes even if their Ra values are similar.
Typical machined surface ranges by process
The table below gives practical, non-guaranteed ranges for common metalworking processes. Actual results depend on machine condition, tooling, material, fixturing, coolant, tool wear, workholding rigidity and the required inspection method. Treat these values as a planning guide, not as a substitute for a qualified process capability study.
| Process or condition | Typical Ra range | Typical use |
|---|---|---|
| Sawn, rough milled or rough turned faces | About 6.3–12.5 µm | Non-critical faces, stock preparation, hidden surfaces |
| General CNC milling or turning | About 1.6–6.3 µm | Most ordinary machined hardware faces, brackets and housings |
| Fine milling or finish turning | About 0.8–1.6 µm | Visible surfaces, controlled fits, moderately demanding contact faces |
| Grinding | About 0.2–1.6 µm | Precision bearing seats, flatness-sensitive faces and hardened parts |
| Honing, lapping or polishing | Often below 0.4 µm when properly controlled | Sealing, low-friction sliding, precision bores and optical or near-optical requirements |
A tighter finish can change the manufacturing route. Moving from a general milled surface to Ra 1.6 µm may only require a controlled finishing pass on a stable setup. Moving to Ra 0.2 µm may require grinding, honing, lapping or polishing, plus more inspection time. That extra step may be necessary for a bearing, seal or hydraulic component, but it can be wasted cost on a decorative or non-contact face.
What controls the final finish
For a machined surface, final roughness is partly geometric and partly process-related. In turning, the theoretical cusp height is tied to feed per revolution and tool nose radius. A common idealized relationship is that theoretical peak-to-valley height is proportional to feed squared divided by nose radius. A related approximation for Ra is often written as feed squared divided by 32 times the nose radius, when the units are handled consistently. This is a planning model, not an inspection rule.
The real surface also depends on effects that are not captured by a simple formula. Tool wear can smear, tear or plow the material instead of cutting cleanly. A built-up edge can leave random marks. Insufficient rigidity can introduce chatter, increasing waviness and visible patterning. Too light a cut can cause rubbing instead of cutting, while too heavy a finishing pass can deflect the part or tool.
- Feed rate often has a strong influence because feed marks are directly reflected in the surface profile.
- Tool nose radius or cutter geometry affects the spacing and depth of scallops, but a larger radius can increase cutting forces and chatter risk.
- Cutting speed influences heat, built-up edge formation and chip behavior.
- Depth of cut must be large enough for stable cutting but not so large that the finishing operation becomes a roughing pass.
- Material condition matters because gummy aluminum, stainless steel, cast iron and hardened steel respond differently to the same tool path.
- Coolant and chip evacuation help prevent re-cutting chips and dragging debris across the finished face.
Grinding, honing and lapping use different mechanics from turning or milling. Abrasive grain size, wheel dressing, pressure, speed and coolant condition become more important. These processes can deliver smoother surfaces, but they also introduce their own risks, including thermal damage, directional scratches and changes in edge condition.
How to specify a machined surface on drawings
A good surface finish note is specific enough to be inspected and flexible enough to be manufactured economically. The common mistake is to apply one tight finish to every face of the part. A better approach is to identify which surfaces actually need control: sealing faces, bearing seats, sliding ways, gasket lands, press-fit diameters, cosmetic faces or coating-critical areas.
When specifying a finish, include the roughness parameter, limit, unit and drawing convention. For example, a callout such as “Ra 1.6 µm max” is clearer than “smooth finish.” If lay direction matters, specify it. If a surface must be produced by grinding, honing or another process for functional reasons, state that requirement. If the process does not matter, avoid locking the supplier into one method.
ASME Y14.36 is useful because it standardizes surface texture symbols for drawings, specifications and related documents, including controls for roughness, waviness and lay. However, the ASME description also makes clear that the symbol standard does not itself specify how the texture must be produced or measured. Measurement requirements, acceptance rules and process notes still need to be appropriate for the part. See also: Buying Guides.
For general hardware parts, a drawing may use a broad default note for unspecified machined surfaces and tighter individual callouts only where function requires them. This reduces disputes because both the buyer and supplier can see which faces are critical.
Inspection and acceptance should match the callout
Surface roughness can be checked by contact stylus instruments, optical instruments, comparison specimens or specialized laboratory methods. Contact profilometers are common for many machined metal parts because they trace a 2D profile and report parameters such as Ra or Rz. Optical methods can be valuable for delicate, very small or areal measurements, but results may not match a stylus reading unless the method, filtering and parameter definitions are aligned.
NIST publications on surface finish metrology emphasize measurement conditions and uncertainty. In practice, this means the inspector should know the cutoff, evaluation length, stylus condition, calibration status and direction of measurement. For surfaces with a clear lay, the measurement direction is often chosen to capture the relevant profile across the marks, but the applicable drawing or inspection procedure should govern the decision.
Acceptance should also consider visible defects. A face can meet Ra while still containing a scratch, dent or burr that interferes with sealing or assembly. Conversely, a surface can look visually attractive but fail the specified roughness value. The safest inspection plan uses the parameter for measurable texture and separate workmanship or defect criteria where the application requires them.
Common mistakes when choosing a finish
The first common mistake is over-specification. Calling out Ra 0.8 µm on every surface of a simple bracket may force unnecessary finishing passes. The second mistake is under-specification. A shaft seal land or sliding bore may need a controlled roughness range and lay direction, not just a general machined finish. The third mistake is treating Ra as a complete description of performance. For contact mechanics, oil retention, gasket sealing or fatigue, peak shape, valley structure and waviness may matter as much as the average value.
Another mistake is ignoring later finishing steps. Plating, anodizing, painting, blasting, polishing and deburring can all change the surface. If the print requires a machined surface before coating, say so. If the requirement applies after coating, make that clear. A pre-coating Ra value and a final Ra value can lead to different process choices.
Finally, avoid casual conversions between standards and parameters. Older drawings may use symbols, N-grade references or roughness values that need interpretation. When a part is safety-critical, sealing-critical or part of a controlled assembly, confirm the governing standard and inspection method before releasing the drawing or purchase order.
Frequently asked questions
Is a machined surface the same as a smooth surface?
No. A machined surface is simply a surface created by machining. It may be rough, standard, fine or very smooth depending on the process and finishing pass. Smoothness should be specified with measurable criteria such as Ra or Rz, not assumed from the word “machined.”
What Ra value is typical for CNC machining?
Many general CNC milled or turned surfaces fall roughly in the Ra 1.6–6.3 µm range, but this is only a planning range. Fine finishing passes can improve the value, while chatter, tool wear, difficult material or poor chip control can make it worse.
Can milling produce Ra 0.8 µm?
It can be possible on a rigid machine with suitable tooling, light finishing cuts, stable workholding and a machinable material. However, it should not be assumed for every milled face. If the requirement is critical, the supplier may choose grinding, polishing or another finishing operation to make the result repeatable.
Why does the same Ra not always perform the same way?
Ra is an average value. It does not fully describe peak sharpness, valley depth, lay direction or waviness. Two surfaces with the same Ra can behave differently in sealing, sliding, oil retention or fatigue applications.
Should every machined surface have a roughness callout?
No. Critical functional faces should be specified clearly, while non-critical faces can often use a general default note. Selective callouts usually produce better cost control and fewer inspection disputes than applying a tight finish everywhere.
