Surface Finish Ra Chart — Classification & Conversion

How to Use This Chart

Theoretical turning Ra ≈ f² / (8R) with feed f and nose radius R (inches → µin). Calculator: surface finish Ra calculator. Guides: surface finish guide, finish callouts.

Ra vs Rz — They Are Not Interchangeable

Ra (roughness average) is the arithmetic mean of the absolute profile deviations from the mean line over a sampling length. It is the number most U.S. prints still mean when they say “32 finish” or show the checkmark with 32. Rz (mean roughness depth, often ISO) is the average of the peak-to-valley heights of successive sampling lengths. Rz is almost always larger than Ra for the same surface — often 4–7× on typical machined profiles — because peaks and valleys dominate Rz while Ra averages them away.

If a European print calls Rz 16 µm and you machine to Ra 16 µin, you are not even in the same unit family, let alone the same roughness. Convert the callout first (see the µin/µm table), then decide whether the designer specified Ra or Rz. Treating Rz as “basically Ra” is how sealing faces leak and how grind shops over-process a part. RMS (Rq) is another average (root-mean-square); older U.S. notes sometimes used RMS, which runs a bit higher than Ra for the same trace. When the print is ambiguous, ask; do not pick the conversion that makes the job cheapest.

Which Processes Hit Which Ra Bands by Default

Process capability is why this classification scale exists. Flame cut and sawed edges live above 500 µin unless you secondary-machine them. Rough turning and heavy milling commonly land 125–500 µin depending on feed and nose radius. General turning and conventional milling with a reasonable finish pass sit in the 63–125 µin neighborhood. Precision turning, fine milling, and a decent grind start to own 32–63 µin. Production grinding, honing, and lapping are how you reliably get under 16 µin.

That mapping is capability, not a promise. A sharp insert, small feed, and large nose radius can theoretically turn a 16 µin Ra (the calculator’s f²/8R estimate). In the vise, the same setup with a worn insert and 0.0005″ of spindle growth will print 40. Conversely, you can grind a surface that still measures 63 µin if the wheel is glazed. Use this chart to sanity-check the process choice on the routing: if the print says 16 µin Ra and the router only has a mill, the issue is the routing, not the operator’s feed.

Castings, weldments, and additive parts have their own as-built roughness that this turning-oriented scale does not describe well. Machine the functional faces; do not chase Ra on a flame-cut web that never sees a seal or a bearing.

Functional Finish vs Cosmetic Finish

A sealing face (gasket, O-ring gland, hydraulic spool) needs roughness in a band that holds the seal, not “as smooth as possible.” Too rough and the gasket weeps; too smooth (especially with no lay specified) and some static seals cannot bite. Bearing fits and press fits care about Ra because peaks shear off on assembly and change actual interference. Hydraulic cylinder rods and valve plates are functional. A painted cover, a bracket that nobody touches, and the OD of a turned shaft that only needs to look even are cosmetic — 125 µin is often plenty, and grinding them is wasted money.

Lay (the direction of the tool marks) is part of functional finish and is almost invisible on this Ra-only chart. A circular lay from facing is not the same as a ground axial lay for a rod seal. If the print shows a lay symbol, Ra compliance without matching lay can still fail the application. That is when you leave this chart and read the finish callout guide.

How Ra Is Actually Measured

Shop reality is a stylus profilometer: a diamond tip dragged across the surface for a defined cutoff length (often 0.030″ / 0.8 mm for typical machined Ra). The instrument filters waviness from roughness and reports Ra, and usually Rz, Rq, and more. Skid-type portable units are what most job shops own. They are good enough for 32–125 µin production checks; they are less trustworthy on very short lands, inside small bores, and on surfaces under ~8 µin where the skid and tip radius start to miss the real profile.

Optical (non-contact) systems and laboratory skidless traces are what you use when the print is 4–8 µin, when the customer is doing a capability study, or when the stylus would scratch a plated face. A fingernail comparison coupon is not a measurement; it is a go/no-go that fails on lay and on anything below ~63 µin. If you are arguing with a customer about a 32 vs 63 callout, bring a trace, not a thumbnail.

Cutoff length matters as much as the number. Measuring a milled face with the wrong cutoff can move Ra a full class. Match the instrument setup to the standard cited on the print (ASME B46.1 / ISO 4287 family), and measure perpendicular to the lay unless the note says otherwise.

Ra Classification Scale

1 µm ≈ 39.37 µin; calculator uses ×0.0254 for µin→µm.

Common Callout Conversions

Related Calculators & Guides