KEY TAKEAWAYS

✓
Ra is an average roughness value — useful for general surface quality but blind to peaks and valleys that matter in sealing and fatigue applications.

✓
Rz measures the average of the five highest peak-to-valley distances and is more sensitive to surface defects than Ra alone.

✓
Rmax (or Rz1max) captures the single worst peak-to-valley measurement in a trace — critical for applications where one bad spike can cause failure.

✓
Always specify both the parameter (Ra, Rz, Rmax) and the cutoff length on your drawing — omitting the cutoff length makes the spec ambiguous and unenforceable.

✓
Tighter surface finish specs increase cost and lead time — only call out what the function actually requires, and confirm manufacturability before releasing your drawing.

Every machined, ground, or formed surface has texture — microscopic peaks and valleys left behind by the cutting tool, abrasive, or forming process. That texture isn’t just cosmetic. It governs how two surfaces mate under load, how well a seal conforms to a bore, how a coating bonds, and how stress concentrations develop under cyclic loading. Engineers who treat surface finish as a default callout — slapping ‘Ra 125’ on every surface because it’s the default in their CAD template — routinely over-specify low-stress geometry and under-specify critical interfaces. Both errors cost money. Over-specification drives unnecessary grinding or lapping operations. Under-specification produces leaking hydraulic cylinders, fretting failures in press-fit bores, and coatings that delaminate in service. Surface roughness parameters are the language that translates functional requirements into measurable, inspectable values. Understanding what Ra, Rz, and Rmax actually measure — and where each one is the right tool — is foundational to writing drawings that get built correctly the first time.

Contact profilometers are the standard instrument. A diamond stylus drags across the surface over a defined evaluation length, recording vertical displacement as a continuous trace. That trace is then mathematically processed to extract roughness parameters. The cutoff wavelength (λc) is a high-pass filter applied to the raw trace — it separates roughness (short-wavelength texture) from waviness (longer-wavelength form error). A typical cutoff length for general machining is 0.8 mm; finer surfaces use 0.25 mm; very rough surfaces may use 2.5 mm. The evaluation length is typically five times the cutoff length, giving five consecutive sampling lengths from which statistical parameters are derived. Non-contact optical profilometers (white-light interferometry, confocal microscopy) are increasingly common for soft or delicate surfaces where stylus contact could alter the measurement. Both methods produce the same Ra, Rz, and Rmax outputs — the physics of what those numbers represent doesn’t change with the instrument. What matters is that the measurement conditions match the standard called out on the drawing, typically ISO 4287 or ASME B46.1.

⚠️ Always specify the cutoff wavelength alongside your roughness value. A drawing that says ‘Ra 0.8 µm’ without a cutoff length is technically incomplete — different cutoff settings on the same surface can produce meaningfully different Ra readings.

Ra is the arithmetic mean of the absolute deviations of the roughness profile from the mean line, measured over the evaluation length. In plain terms: take all the peaks and valleys in the trace, flip the valleys above the mean line, and average all of those absolute heights. Ra is the most widely used roughness parameter in manufacturing drawings worldwide, primarily because it is simple to calculate, easy to communicate, and stable — a few isolated peaks or valleys don’t dramatically skew the average. Common Ra values and their typical process associations:

• Ra 3.2–6.3 µm (125–250 µin): General CNC milled or turned surfaces, no finish pass required
• Ra 1.6 µm (63 µin): Standard finish pass on CNC lathe or mill — suitable for most non-critical mating surfaces
• Ra 0.8 µm (32 µin): Fine turning or grinding — O-ring grooves, bearing seats
• Ra 0.4 µm (16 µin): Precision grinding — hydraulic cylinder bores, precision shafts
• Ra 0.1 µm (4 µin) and below: Lapping, honing, superfinishing — optical surfaces, high-pressure seals

Ra’s weakness is its insensitivity to the shape of the texture. Two surfaces can share an identical Ra value while one has sharp, spiky peaks and the other has broad, rounded valleys — a distinction that matters enormously for fatigue, wear, and sealing performance.

Rz is defined under ISO 4287 as the arithmetic mean of the five individual maximum peak-to-valley heights (Rz1 through Rz5), each measured within one sampling length. Because it averages five of the worst local events rather than all deviations, Rz is more sensitive to significant surface defects than Ra while still being statistically robust enough to be repeatable. A surface with occasional deep scratches or sharp spikes will show a much higher Rz relative to its Ra than a uniformly textured surface — that ratio is itself diagnostic. The Rz/Ra ratio for a random, isotropic surface typically falls between 4:1 and 7:1. Ratios significantly above 7 suggest the presence of isolated defects; ratios below 4 indicate a very periodic, regular texture (like a honed crosshatch). European standards and DIN drawings have historically favored Rz over Ra for this reason — it provides better discrimination for critical surfaces. In aerospace and hydraulic applications, Rz is often the governing specification on sealing surfaces and dynamic interfaces. If your drawing only calls Ra and your supplier is in a region that defaults to Rz, confirm that both parties are reading the same parameter.

💡 Rule of thumb: for sealing surfaces and fatigue-critical features, always specify Rz in addition to Ra. An Ra callout alone won’t catch an isolated deep scratch that falls within the statistical average.

Rmax (also written as Rz1max in ISO 4287 terminology, or Ry in older JIS standards) is the single largest peak-to-valley height measured within any one sampling length across the entire evaluation length. It is the worst-case number — the single most extreme surface event captured in the measurement. Rmax is the right parameter when one bad spike can cause a functional failure. Examples include: optical surfaces where a single scratch causes diffraction artifacts, hard-coated surfaces where a peak taller than the coating thickness will protrude through and cause contact wear, and fuel system components where a surface defect can initiate a leak path at pressure. Because Rmax is by definition a single extreme event, it is inherently less repeatable than Ra or Rz — two measurements on the same surface can yield different Rmax values depending on exactly where the stylus traverses. For this reason, Rmax is typically used as a reject threshold rather than a primary specification. It answers the question: ‘Is there any single defect on this surface that exceeds the limit?’ rather than characterizing the overall surface quality.

Selecting the right parameter is a functional decision, not an arbitrary one. Use Ra when you need a general characterization of overall surface texture — non-critical mating surfaces, cosmetic requirements, or when specifying a process capability range. It is the most widely understood parameter and the easiest to measure and communicate. Use Rz when the functional requirement is driven by the amplitude of significant surface events rather than the average texture. Dynamic seals, bearing surfaces, fatigue-critical features, and surfaces that will receive thin coatings (anodize, plating under 25 µm) all benefit from an Rz callout. The relationship between Ra and Rz also lets you audit the consistency of a surface’s texture. Use Rmax as a hard upper bound on isolated defects — not as a standalone specification, but as a supplemental limit on drawings where a single surface anomaly is intolerable. In practice, a well-engineered drawing for a critical surface might call all three: for example, ‘Ra 0.4 µm max, Rz 2.0 µm max, Rmax 4.0 µm max.’ That combination defines the average texture, the typical worst-case event, and the absolute rejection threshold in a single, complete specification. When Nimble’s team conducts DFM reviews on incoming drawings, conflicting or incomplete surface finish callouts are one of the most common sources of supplier ambiguity and inspection disputes — getting these three parameters right up front eliminates that risk.

Surface finish specifications have a direct, nonlinear relationship with cost. Moving from Ra 3.2 µm to Ra 1.6 µm typically adds a finish pass — modest cost increase. Moving from Ra 1.6 µm to Ra 0.4 µm usually requires a process change from milling or turning to grinding — significant cost and lead time increase. Moving below Ra 0.2 µm into lapping or superfinishing territory can multiply part cost by 3x–5x on small features. The cost drivers are process capability, cycle time, and inspection. Grinding requires different fixturing and machines than milling. Lapping is often a manual, labor-intensive operation. And tighter finish specs demand more sophisticated measurement — profilometer traces rather than visual comparator checks, with documented CMM or surface metrology data in the inspection report. Specifying Ra 0.4 µm on a bolt clearance hole because it happens to be the drawing default wastes money without adding function. Conversely, leaving a hydraulic valve body bore at Ra 1.6 µm because nobody challenged the original drawing will generate warranty returns. The correct approach is to work backward from function: what does this surface need to do, and what is the roughest surface that will still perform reliably? That number goes on the drawing — nothing tighter. Nimble’s free DFM review process routinely flags over-specified surfaces where a relaxed callout would cut cost without compromising performance.

⚠️ Cost escalation by finish tier (approximate, relative to an as-machined Ra 3.2 µm baseline): Ra 1.6 µm = 1.1–1.3x | Ra 0.8 µm = 1.3–1.8x | Ra 0.4 µm = 1.8–2.5x | Ra 0.1 µm and below = 3x–6x. These multipliers compound when multiple surfaces on a part require fine finishing.

Surface finish is called out on engineering drawings using symbols defined in ASME Y14.36M (US) or ISO 1302 (international). The symbol — a check-mark-like glyph placed on the surface line or a leader — carries the parameter value, the parameter type (Ra, Rz, etc.), the cutoff wavelength, and any lay direction requirements. A bare triangle symbol with no value historically meant ‘machined, no specific requirement’ — this convention is obsolete under current standards but still appears on legacy drawings. Always verify which standard governs before interpreting an unfamiliar callout. Inspection of surface roughness is typically performed with a contact profilometer at receiving inspection or in-process. For aerospace and defense components manufactured through AS9100-certified facilities, surface roughness measurements are part of the documented first article inspection (FAI) record. CMM reports confirm dimensional tolerances, but surface finish data comes from profilometer traces — both sets of data belong in the inspection package. When sourcing through a managed partner network like Nimble, CMM inspection reports are included as standard, and surface finish data can be specified as a required deliverable in the quality plan. Knowing that your supplier has calibrated profilometry equipment and documented measurement procedures — not just a visual comparator and a wall chart of surface samples — matters when you are building flight hardware or hydraulic assemblies.

READY TO SOURCE?

Table of Contents

Related Services

Get a Quote in 24h

Upload your drawing and get a detailed quote with free DFM review.