Surface roughness in metal parts and how to specify Ra and Rz
a[data-rs-seo-link]{text-decoration:underline!important;color:#1a56db!important;cursor:pointer!important;}a[data-rs-seo-link]{text-decoration:underline!important;color:#1a56db!important;cursor:pointer!important;}
What surface roughness actually tells you
Surface roughness is the fine-scale pattern of peaks and valleys left on a manufactured surface after casting, cutting, grinding, polishing, coating or other finishing work. For metal parts, it is not simply a cosmetic value. It can influence friction, sealing, wear, fatigue behavior, coating adhesion and inspection repeatability. The practical rule is straightforward: do not specify a roughness number by habit. Choose the parameter, limit, unit, measurement direction and inspection method that match the function of the surface.
In metalworking, surface roughness sits within the broader subject of surface finishes. A finish can refer to appearance, coating, texture, corrosion protection and process route. Roughness is narrower. It is a measurable part of surface texture, usually separated from larger waviness and overall form by filtering during measurement.
That distinction matters. A surface can look smooth but measure poorly. Another surface can meet an Ra value and still fail in service because of directional tool marks, isolated scratches, torn material or unsuitable peak geometry. A single roughness number is useful only when it is tied to the right standard and inspection conditions.
Roughness, waviness and lay are not the same thing
Surface texture is commonly described through three related ideas: roughness, waviness and lay. Roughness covers the short-wavelength irregularities that remain after the measurement system filters out broader shape. Waviness covers wider spacing variations that may come from machine vibration, deflection, heat treatment distortion or fixturing. Lay describes the dominant direction of the surface pattern, such as circular grinding marks, parallel milling marks or cross-hatched honing marks.
A sealing face, sliding guide or bearing seat may need control over more than one of these features. Two surfaces can have the same Ra value, for example, while one has strong directional grooves that encourage leakage along the groove direction. Another surface may have acceptable roughness but excessive waviness, causing poor contact across the assembled joint.
Surface roughness should therefore not be used as a substitute for flatness, profile tolerance, runout or dimensional tolerance. Roughness describes local texture. Geometric tolerances control larger-scale shape and location. Both may be needed on a functional surface, but they solve different problems.
Ra, Rz, Rq and Sa explained without overclaiming
Ra is the most familiar surface roughness parameter, especially on machining drawings. It represents an arithmetic average of the absolute profile height deviations from a mean line over the evaluation length. Ra is popular because it is easy to measure, easy to compare and widely supported by shop-floor instruments.
Ra also hides important detail. A surface with shallow, uniform tool marks and a surface with a few deeper scratches can produce similar Ra values. For this reason, Rz, Rq, material ratio parameters or areal parameters may be more useful when peaks, valleys, contact behavior or fluid retention matter.
| Parameter | What it broadly indicates | Where it is often useful | Main limitation |
|---|---|---|---|
| Ra | Average profile roughness height | General machining, routine inspection, supplier communication | Can mask isolated peaks, valleys and directional texture |
| Rz | Peak-to-valley type height information, depending on the referenced standard | Sealing faces, turned parts, surfaces where deeper valleys matter | Definitions and calculation details can vary by standard generation |
| Rq | Root mean square roughness of the profile | Surfaces where larger deviations should have stronger influence | Less commonly specified on ordinary shop drawings than Ra |
| Sa | Areal arithmetic mean height over a measured area | Optical or 3D surface measurement, textured or non-directional surfaces | Not directly interchangeable with Ra from a single profile trace |
The key point is that Ra and Rz are not reliably convertible by a universal ratio. Some industry charts suggest approximate conversions, but those numbers depend heavily on process type, profile shape, filter settings and measurement method. If a drawing requires Rz, inspect Rz. If it requires Ra, inspect Ra. If both are functionally important, specify both with clear limits.
Units also need attention. Micrometers are common in ISO-based drawings, while microinches remain common in many U.S. machining environments. One micrometer is about 39.37 microinches. As a result, Ra 3.2 µm is roughly Ra 125 µin, but Ra 3.2 µin would be an entirely different and much finer requirement. Ambiguous units are a common source of avoidable disputes.
Standards that shape surface roughness specifications
Surface roughness language is standardized because small differences in definitions and filters can change the measured result. Public standards catalogues and metrology references identify several documents that are especially relevant to metal part drawings and inspection. The following snapshot reflects public catalogue status available in September 2026.
| Standard or reference | Role in roughness work | Practical note |
|---|---|---|
| ISO 21920-1:2021 | Profile surface texture indication | Relevant when calling out profile-based surface texture on ISO-style technical product documentation |
| ISO 21920-2:2021 | Terms, definitions and parameters for profile methods | Important for understanding profile roughness parameters such as Ra and related quantities |
| ISO 21920-3:2021 | Specification operators for profile methods | Helps define how a complete surface texture specification should be interpreted |
| ISO 25178-2:2021 | Terms, definitions and parameters for areal methods | Relevant when surface texture is measured over an area rather than along one profile line |
| ASME B46.1-2019 (R2026) | U.S. surface texture standard covering roughness, waviness and lay | Common reference for North American drawings, inspection planning and metrology communication |
| ISO 1302:2002 | Older indication standard for surface texture in technical product documentation | Listed as withdrawn and superseded by ISO 21920-1:2021, but still appears on legacy drawings |
There is also ongoing standards activity. Public ISO catalogue information in 2026 shows committee drafts under development for newer editions of ISO 21920-2 and ISO 21920-3. A draft should not be treated as a contractual requirement unless a buyer, supplier or regulated project explicitly adopts it. For active production, the safest approach is to name the exact standard and edition on the drawing or inspection plan.
Legacy drawings require special care. If a print calls out an old standard, do not silently reinterpret it under a newer one. Definitions, symbols, filters and defaults may have changed. When a drawing is unclear, the practical answer is not to guess at a conversion. Clarify the governing standard, parameter definition and acceptance rule before parts are made.
How roughness is measured in production
Most shop-floor surface roughness measurements use either a contact stylus instrument or a non-contact optical method. A stylus profilometer drags a fine tip across the surface and records a two-dimensional profile. It is widely used for machined metal parts because it is familiar, portable and well supported by profile roughness standards.
Stylus measurement still has limits. Tip radius, stylus force, traverse direction, skid arrangement, filter settings and part accessibility can affect the result. A stylus may not fully enter very narrow valleys, and it can be unsuitable for delicate coatings or very soft surfaces. Measurement direction is especially important on turned, milled or ground parts because readings taken across the lay can differ from readings taken along the lay.
Optical instruments measure surface texture without dragging a tip across the part. Depending on the instrument type, they can provide areal data and may be useful for fine textures, delicate surfaces, small features or surfaces where a three-dimensional view is more meaningful than a single trace. Optical methods can also face challenges, including steep slopes, reflectivity changes, transparent coatings, contamination and differences in data processing.
For repeatable inspection, the measurement report should not stop at a single number. It should identify the parameter, unit, instrument type, cutoff or filter, evaluation length, measurement direction, location on the part, number of traces or areas measured and the standard used. Without those details, two competent inspectors can measure the same surface and report different results. See also: Buying Guides.
Choosing a roughness value for metal parts
The right surface roughness depends on function. A non-contact cover plate does not need the same finish as a hydraulic sealing face, bearing journal, adhesive bonding surface or cosmetic exposed panel. Over-specifying roughness can add unnecessary machining, grinding, polishing, inspection time and scrap risk. Under-specifying it can create leakage, premature wear, coating failure or inconsistent assembly.
As a planning guide, many metal components use rougher values for non-critical surfaces and finer values for sealing, sliding or precision contact surfaces. The ranges below are not universal capability promises. They are practical starting points that should be confirmed against material, geometry, machine condition, tooling, finishing route and inspection method.
| Surface or process situation | Common planning range for Ra | Design caution |
|---|---|---|
| General non-critical machined surfaces | About 1.6 to 6.3 µm, or about 63 to 250 µin | Usually adequate for clearance surfaces, but not enough information for sealing or sliding performance |
| Finish milling or turning on suitable material | About 0.8 to 3.2 µm, or about 32 to 125 µin | Tool wear, vibration, feed marks and material tearing can move results outside the expected band |
| Grinding and precision finishing | About 0.2 to 0.8 µm, or about 8 to 32 µin | Often used when contact, sealing or bearing behavior matters, but geometry and waviness still need control |
| Lapping, polishing or superfinishing | Below about 0.2 µm may be possible | Cost, edge rounding, contamination and process verification become more important |
| Cast, forged, blasted or coated surfaces | Highly process-dependent | Specify the required condition and inspection method rather than assuming an Ra value from appearance |
Some surfaces need a minimum roughness, not only a maximum. Adhesive bonding, paint, thermal spray, some gasket systems and oil-retaining sliding contacts may perform poorly if the surface is too smooth. In those cases, a one-sided maximum Ra callout may be incomplete. A controlled range, a different parameter or a process-specific texture requirement may be more appropriate.
How to write a clearer surface roughness callout
A good roughness requirement tells manufacturing and inspection exactly what matters. It should be short enough to use on a drawing, but complete enough to prevent conflicting interpretations. The following checklist is useful for buyers, designers and quality teams preparing metal part specifications.
- Name the parameter. Use Ra, Rz, Rq, Sa or another parameter deliberately. Do not assume Ra describes every functional texture problem.
- State the limit and unit. Write values clearly in µm or µin. Avoid unitless notes such as 32 finish unless the drawing standard and company convention make the meaning unambiguous.
- Reference the standard and edition. This is especially important when old ISO or ASME drawings are revised for current production.
- Control the inspection condition where needed. Include cutoff, filter, evaluation length, measurement direction and number of measurements when defaults are not enough.
- Identify the surface condition. State whether the requirement applies before coating, after coating, after heat treatment, after polishing or after final cleaning.
- Address defects separately. Scratches, pits, dents, burrs and tears may not be adequately controlled by Ra alone.
- Use functional zones. Apply tight roughness only where it matters. Avoid blanket fine-finish notes across the whole part unless every surface truly needs them.
A shaft journal, gasket face or sliding rail may need a precise roughness callout tied to a measurement direction. A clearance pocket or hidden bracket face may only need a general machined finish. Separating these requirements helps control cost while protecting performance-critical areas.
Common mistakes that cause roughness disputes
The most common mistake is treating surface roughness as a universal quality grade. Lower Ra is not automatically better. A very smooth surface can reduce coating adhesion, fail to hold lubricant or increase sticking in some contact conditions. A slightly rougher, well-controlled texture may perform better if it matches the application.
A second mistake is mixing parameters across standards. Rz is especially prone to misunderstanding because older and newer standards, regional practices and instrument settings may not always align. If the customer expects one Rz definition and the supplier measures another, both sides may believe they are correct. The solution is to specify the governing standard and confirm instrument setup before production approval.
A third mistake is ignoring location. Surface roughness can vary across a part because of tool entry, tool exit, chatter, interrupted cuts, workholding, coolant access and local geometry. Measuring the easiest flat area may not represent the functional area. When the working zone is small, curved or difficult to reach, the inspection plan should describe where and how to measure it.
Finally, roughness should not be used to hide uncertainty in design intent. If the real concern is sealing, leakage testing, contact pressure, corrosion, visual appearance or fatigue life, the drawing may need additional requirements beyond a roughness value. Ra is a useful number, but it is not a complete engineering description of a surface.
Frequently asked questions
Is surface roughness the same as surface finish?
No. Surface finish is a broader term that may include visual appearance, coating, treatment, texture and process route. Surface roughness is a measurable part of surface texture, usually expressed through parameters such as Ra or Rz.
Is a lower Ra value always better?
No. Lower Ra means a smoother average profile, but the best roughness depends on function. Some sealing, sliding, bonding and coating applications need a controlled texture rather than the lowest possible Ra.
Can Ra be converted to Rz?
Only approximately, and not reliably for acceptance decisions. Ra and Rz describe different aspects of the profile. Their relationship depends on the manufacturing process, profile shape, filter settings and standard used.
Should roughness be measured before or after coating?
It depends on the requirement. If roughness is needed for coating adhesion, the pre-coating substrate may matter. If roughness affects final sealing, sliding or appearance, the finished coated surface may need inspection. The drawing or purchase specification should state the condition clearly.
What should be included in a roughness inspection report?
A useful report should include the parameter, value, unit, standard, instrument type, cutoff or filter, evaluation length, measurement direction, measured location and number of measurements. Those details make the result repeatable and easier to compare between supplier and customer.
