Specifying surface roughness on precision part drawings: Ra, Rz and their relation to appearance, sealing and contact

A single Ra value on a drawing does not define appearance, sealing and fit at the same time. A surface texture indication is an agreement on four things: the surface, the parameter, the measurement conditions and the acceptance rule. Surfaces with different functions need different questions. Each requirement should carry the conditions that make it measurable.

One precision part, such as a watch case, combines cosmetic faces, seal grooves, press-fit bores and internal walls. The same metal can scatter light evenly on a matte floor and reflect it along the tool path on a machined face. These micro-geometries also control leak paths, friction, wear and press-fit retention. Applying one value to every surface adds operations and inspection time, while critical surfaces may still miss the conditions that matter.

Roughness is separated from form and waviness by wavelength

Height deviations are divided into form error, waviness and roughness. Form covers bending, flatness and roundness. Waviness is a longer undulation from spindle run-out, vibration or coarse tool paths. Roughness is the shortest-wavelength texture left by the tool nose, feed marks, abrasive grains or blasting media.

Instruments process the profile using form removal and filters. The L-filter nesting index Nic in ISO 21920 corresponds to the familiar cut-off λc. Filtering attenuates components through a transmission characteristic rather than separating all wavelengths at a sharp boundary. The S-filter and form-removal method also affect the result. Sealing and cosmetic surfaces may need separate flatness, roundness or waviness requirements.

Ra and Rz summarise different profile information

Ra is the arithmetic mean of absolute roughness-profile deviations from the mean line. It supports process monitoring but does not describe scratch direction or distinguish peaks from valleys. ISO 21920 evaluates Ra over the evaluation length rather than averaging separately calculated values from sampling lengths.

Rz averages peak-to-valley heights of specified sections. Rp, Rv and Rz remain section-based exceptions under ISO 21920. Rt is the highest-to-lowest difference over the entire evaluation length and is not interchangeable with Rz. Separate local-defect requirements may still be necessary.

There is no universal Ra-to-Rz conversion. Periodic marks, sparse peaks and deep valleys can have the same Ra with different Rz values. Ratios obtained from one process must not be transferred to a different material, process, bandwidth or drawing.

Check which Rz the drawing means

ISO Rz since 1997 is the mean maximum height. The 1984 ISO Rz and the former JIS Rz were a ten-point height, now called RzJIS. Older Korean or Japanese drawings and instrument settings should be checked before values are compared.

Shape and material-ratio parameters

Rsk shows whether peaks or valleys dominate. The material ratio Rmr(c) gives the proportion of material at a stated cut depth. Rpk, Rk and Rvk split the material ratio curve into peak, core and valley zones. They suit plateau surfaces such as honed and polished bores.

The interactive model in the Korean section compares six synthetic surfaces and two orthogonal traces. Peaky and plateau examples have matching mean absolute height on the reference X trace, but different section heights and skewness. Parallel tracing of the turned surface captures much less variation. The dimensionless model indices use exaggerated height and simplified processing, not an ISO-compliant instrument simulation. A pair of short traces on a nominally non-directional surface may still differ.

Measurement conditions change the verdict

Under ISO 4288, non-periodic profiles used default cut-offs by Ra range. Above 0.006 to 0.02 µm, λc was 0.08 mm. Above 0.02 to 0.1 µm, it was 0.25 mm. Above 0.1 to 2 µm, it was 0.8 mm. Above 2 to 10 µm, it was 2.5 mm, and above 10 to 80 µm, 8 mm.

Legacy evaluation lengths generally comprise five cut-offs, with prescribed stylus and sampling conditions. Legacy periodic surfaces used RSm for cut-off selection. ISO 21920-3 derives default settings from the drawing tolerance or an explicit class, without that periodic/non-periodic distinction. Sc1–Sc5 use Nic 0.08, 0.25, 0.8, 2.5 and 8 mm, and evaluation lengths 0.4, 1.25, 4, 12.5 and 40 mm. Class-selection tolerance boundaries depend on the parameter; an Ra table must not be used as an Rz table.

Tracing direction

A perpendicular trace generally captures greater variation on a directional surface, but an explicitly specified direction takes priority. Crossed lays do not imply a universal 45-degree trace. Define relevant directions, locations and which values enter the acceptance decision, particularly for concentric and radial patterns.

From the 16 % rule to the max rule

The legacy default for upper limits was the 16% rule, while a max indication invoked the maximum rule. The 16% rule concerns the distribution of values on the specified surface and a prescribed verification procedure. It does not allow 16% defective parts in a production lot or justify an arbitrary sampling rule.

ISO 21920-1 makes the max rule the default. The 16 % rule must now be indicated explicitly. Drawings made under the previous standards are still evaluated under those standards. The same “Ra 0.8” can therefore mean different acceptance criteria.

Guidance for the earlier system excluded surface imperfections from the evaluated profile; newer guidance requires them to be considered in the inspection decision. Location-to-location variation and measurement uncertainty are different issues. A published percentage of typical variation is not a universal tolerance allowance. Near-limit results require attention to calibration, repeatability, location and the agreed rule.

Profile and areal values

Ra and Rz are profile parameters; Sa and Sz are areal parameters. Sa cannot be substituted for an Ra tolerance. Current Sz describes maximum areal height. Optical instruments can also provide an extracted profile for R parameters if processing conditions are appropriate. Instrument principle and parameter type are separate choices.

Optical results depend on lateral resolution, slope, reflectivity and missing-data treatment. Stylus results depend on tip geometry and force, including access to narrow valleys and possible marking of soft surfaces. Neither method always reads higher or is always more accurate. Compare equivalent bandwidths and resolutions.

The drawing symbol carries the value and the conditions

The basic symbol states whether any process is allowed, removal is required or removal is not permitted. Around it sit the first and second requirements, the manufacturing method, the lay symbol and the machining allowance.

The symbol diagram explains the legacy ISO 1302 arrangement. ISO 21920-1 uses a distinguishing symbol for its specification system. Do not mix a legacy symbol with assumed new rules. Existing drawings remain tied to their standards and revision history. State the standard, edition and any deviations from defaults; a KS reference also needs its precise designation and edition.

Sealing surfaces: groove, counterface and scratch direction

Compressed elastomer conforms to surface texture, but a continuous valley may remain a leak path. Moving applications also involve friction, lubricant retention and wear. The appropriate range depends on material, movement, pressure, fluid and temperature.

The second interactive model in the Korean section presses an O-ring section onto a counterface and lets the lay direction, roughness level and seal compression be switched. Grooves running along the seal stay closed, while grooves crossing the contact band that the elastomer cannot fill are highlighted as open paths. It is an explanatory model with exaggerated heights, not a prediction for a specific seal.

Separate counterface, groove and assembly requirements

One O-ring design guide gives Ra ≤ 0.4 µm, Rz ≤ 1.6 µm and Rt ≤ 2.5 µm for radial dynamic mating surfaces, with spiral-free grinding. Its corresponding groove flanks and groove diameter allow Ra ≤ 1.6 µm, Rz ≤ 6.3 µm and Rt ≤ 10 µm. Assembly lead-in chamfers have the example limits Ra ≤ 0.8 µm and Rz ≤ 6.3 µm.

These are different surfaces in the same guidance, not common acceptance values for watch water resistance or every static seal. Gas, vacuum, static and pulsating-pressure applications require their own conditions. There is no universal lower roughness limit for all dynamic seals.

Unit conversion does not convert RMS into Ra

One microinch equals 0.0254 µm, but surface-height RMS normally denotes Rq. Writing 16 µin RMS as 0.4064 µm Ra changes the parameter as well as the unit. For the same profile, Rq is not smaller than Ra, but the ratio depends on the distribution. Confirm the source definition and convert units within the same parameter.

Scratch direction

A continuous scratch across the contact width can connect the pressurised and unpressurised sides. A mark following the groove is not automatically acceptable at any depth. Consider seal compression, width, hardness and pressure, and verify the actual assembly.

Axial scratches on cylindrical counterfaces and radial scratches in face-seal grooves are particularly harmful. A helical feed lead can transport fluid under a rotating seal. Such surfaces need scratch limits and a lay requirement in addition to an Ra limit. Some seal makers also describe dynamic counterfaces with Rmr and Rpk/Rk/Rvk. Because anodising, plating and PVD change roughness and dimensions, state whether the value applies before or after treatment.

Fits and sliding surfaces: peaks touch first and wear first

Contact measurements are affected by asperities and contact conditions. Assembly flattens or shears asperities while the bodies deform. Not all measured interference remains available to generate contact pressure. Specify roughness together with diameter tolerance, wall thickness and material.

Calculations associated with DIN 7190-1:2017 use a smoothing allowance based on Rz. A common diametral form is ΔU ≈ 0.4 × (Rz of shaft + Rz of bore); earlier editions used a different factor. Both mating surfaces and the standard edition must be identified.

For an illustrative 20 µm diametral interference and Rz values of 4 and 6 µm, the allowance is 4 µm and corrected interference is 16 µm. Do not mix diametral and radial formulations. This empirical allowance does not establish retention of a thin case, coated interface or repeatedly assembled joint. Thermal expansion, deformation and friction require separate consideration.

Sliding and rotating surfaces

Initial peak wear in button shafts and slide rails can change clearance. Friction also depends on the mating material, lubrication, load and speed. A smoother surface does not always provide lower friction. Consider lubricant retention and adhesion, then verify the actual assembly through repeated operation.

Critical sliding surfaces may need Rz, Rsk or material-ratio parameters as well as Ra. These values describe limited measurement areas and do not replace running-in and durability tests with the actual mating materials.

Cosmetic surfaces: Ra maintains the process

Bead-blasted matte, hairline and mirror-polished finishes reflect light very differently on the same metal. Matte and hairline surfaces may overlap in Ra. One scatters light in all directions; the other spreads it perpendicular to the lay.

Cosmetic surfaces are therefore accepted with approved samples, viewing distance and lighting, and limits for scratches, dents and pores. Ra confirms that the process is repeated consistently.

Mirror finishes

Mirror appearance depends on image clarity, residual abrasive lines, waviness and local defects. Gloss measurement can complement agreed visual comparison. A low Ra does not guarantee a clear reflection. Mirror texture is measurable, but the method must suit the noise floor, resolution and risk of stylus marking. On MIM components, polishing may reveal pores below the original surface.

Manufacturing process and roughness are decided together

For turning, the theoretical roughness follows from the feed f (mm/rev) and tool nose radius r (mm): R_t,th ≈ f²/(8r) and R_a,th ≈ f²/(32r). This geometric approximation excludes built-up edge, vibration, elastic recovery and tool wear.

With f = 0.10 mm/rev and r = 0.4 mm, R_t,th ≈ 3.1 µm and R_a,th ≈ 0.78 µm. Halving the feed reduces both to one quarter, about 0.78 µm and 0.20 µm. Actual surfaces are often rougher, so trial machining is required.

Halving the feed quarters the geometric terms, but actual improvement need not follow that ratio. Minimum chip thickness, edge condition, built-up edge and vibration also matter. Trial machining determines whether a finer requirement needs changed tooling, better support or an added grinding, lapping or polishing step.

Typical ranges overlap

Typical ranges are roughly 0.2–6.3 µm for turning and boring, 0.8–6.3 µm for milling and 0.03–1.6 µm for grinding. Honing gives about 0.06–0.8 µm and lapping about 0.015–0.4 µm. Feed, tool, lubrication and material can move results outside these bands.

MIM texture depends on powder size and shape, mould finish, debinding, sintering, shrinkage and pores. Fine powder can help texture transfer but does not by itself guarantee Ra. Plan allowances and datums for sealing and fitting faces. Polishing removal must preserve edges; blasting may make appearance uniform without removing pores or deep scratches. Keep as-sintered, polished and finally treated results separate.

Agree on inspection conditions in advance

Confirm both stylus access and travel. The instrument requires start and end allowance beyond the evaluation length, so a land of that width may still be too short. Agree shorter lengths, section count, filters or alternative methods before acceptance. Do not shorten the cut-off to obtain a passing value. Pt may be separately specified but cannot replace an existing Ra or Rz limit. An alternative method needs its own limit and validation.

Clean and support the part, and check the instrument using a suitable reference. Document levelling, form removal and skid conditions where relevant. Optical interpolation or exclusion of pores can change the result and must not be altered arbitrarily.

Reports should identify the parameter and value, standard and edition, locations and directions, instrument and stylus or lens, filters, evaluation length, measurement count, treatment stage and decision rule. Near-limit results require repeatability and uncertainty to be considered. Sealing, retention and durability need separate functional tests.

Items to organise before writing the indication

• Static seals: seal type, compression direction, fluid and pressure; groove and counterface limits with scratch rules.
• Dynamic seals: speed, lubrication and lay; upper and lower limits or material-ratio parameters.
• Press fits: roughness linked to the diameter tolerance.
• Sliding parts: Rz, Rsk or Rpk/Rk/Rvk, verified with cycle tests.
• Cosmetic surfaces: a process-control Ra range with approved samples and defect limits.
• Non-functional surfaces: general requirements only.

When requesting a quotation, mark functional surfaces on the 2D drawing and 3D data. Add the mating parts, seal type, surface treatment sequence, cosmetic acceptance criteria and the surface texture standard the drawing follows. The machining process and inspection method can then be defined precisely.