Surface Roughness Guide: Choosing Ra for Prototypes

Why surface roughness matters

Surface roughness is the fine, microscopic texture left on a part after manufacturing, and it quietly controls how that part behaves. A smoother surface lowers friction and wear in sliding contacts, seals more reliably against a gasket, and reflects light more evenly for a premium look.

Roughness also affects function you cannot see. Sharp peaks act as stress concentrators that shorten fatigue life; rough faces trap contaminants and resist bonding or coating. Choosing the right finish is therefore an engineering decision, not a cosmetic afterthought.

Reading the numbers: Ra and Rz

Ra, or arithmetic average roughness, is the most common value on drawings. It is the average absolute deviation of the surface profile from its mean line, usually given in micrometers (microns, µm) or micro-inches (µin). Lower Ra means smoother; Ra 0.8 µm is mirror-like, while Ra 12.5 µm is clearly matte and grainy.

Rz measures the average height of the five tallest peaks and deepest valleys in the sampling length. It captures the worst spots better than Ra, so it matters for sealing and wear where a single deep groove can leak. Many drawings specify both: Ra for the general feel and Rz for the extreme peaks.

Common roughness grades at a glance

Rough machined surfaces sit around Ra 3.2 to 6.3 µm; a good as-machined CNC finish is typically Ra 1.6 µm. Precision milling or grinding can reach Ra 0.4 to 0.8 µm, while lapping or polishing drops below 0.2 µm for optical or sealing faces.

Printed parts begin where machining ends: FDM shows layer lines near Ra 6 to 12 µm, SLA resin around Ra 1 to 3 µm, and SLS nylon around Ra 4 to 10 µm before finishing. Knowing these baselines lets you predict finish before you commit to a process.

Roughness by manufacturing process

Subtractive CNC gives the most controllable roughness because the cutter path, feed, and tool nose radius are all tunable. A finishing pass with a small stepover and a sharp tool yields a clean, consistent surface straight off the machine.

Additive processes are governed by layer height and orientation rather than tool marks. FDM exposes visible ridges; SLA and SLS are smoother but still show stair-stepping on curves. Casting and sheet metal carry their own textures, typically rougher, that you must plan around if appearance matters.

How roughness is measured

The reference method is a contact profilometer, a diamond stylus dragged across the surface that records the profile and computes Ra and Rz from a defined sampling length. It is precise and traceable, ideal for first articles and supplier acceptance.

For quick shop-floor checks, comparison specimens or roughness bars let an inspector match a surface by eye and touch. Optical and laser scanners are faster for large or fragile parts but need careful filtering. Always agree on the measurement method with your shop, because results vary between techniques.

How material changes the achievable finish

Material decides how fine you can go and how the surface will hold up. Soft aluminum and free-machining brass take a bright, smooth cut but can smear or burr; hardened steel and titanium resist cutting and need slower, finer passes to look good.

Plastics behave differently again: amorphous polymers like ABS and PC polish and vapor-smooth cleanly, while semi-crystalline nylon and PE are harder to gloss. A surface that is easy to finish in metal may be impossible to match in the chosen polymer, so pick material and finish together.

Finishing operations that change Ra

Mechanical finishes shift Ra on purpose. Bead blasting or brushing raises an otherwise smooth machined face to a uniform Ra 1.6 to 6 µm matte that hides tool marks and fingerprints. Tumbling smooths batch parts but rounds edges you may want to keep sharp.

Chemical and thermal routes go finer: anodizing adds a controlled micro-texture on aluminum, vapor smoothing melts printed ridges down toward Ra 1 µm, and electropolishing brings stainless to a bright, low-Ra medical finish. Each adds cost and a small dimensional change you must allowance for.

Where to specify tight Ra, and where not to

Call out a tight Ra only where it changes function or saleability: dynamic seals, O-ring grooves, sliding bearings, optical windows, and bonding or welding faces. On these, a controlled surface prevents leaks, galling, and weak joints.

Leave hidden, non-contact, and structural faces at the default process finish. Over-specifying Ra 0.4 µm across an entire housing multiplies machining time and cost for zero benefit. The discipline is to spend finish budget only where the surface actually works.

Cost and lead-time trade-offs

Every step toward a finer finish costs more. A hand-polished Ra 0.2 µm face can take many times longer than an as-machined Ra 1.6 µm face, and adds inspection. Finer also means higher scrap risk on delicate features.

A matte bead blast is cheap insurance for a uniform, scratch-hiding look on visible parts, and often the best value. Reserve the finest, most expensive finishes for the few faces that truly need them, and accept the process default everywhere else.

How to call out Ra correctly on a drawing

Mark Ra with the standard symbol and a value, applied only to the specific faces that need it, not the whole part. State the sampling length and, if critical, the Rz limit, so suppliers measure the same thing you intend.

Pair the symbol with a finish method note only when necessary, because mandating a process limits your shop's flexibility. A target Ra with a tolerance band is usually enough; let the supplier choose bead blast, brush, or polish to hit it.

A practical selection guide and how FabProto helps

Start from function: if the face seals, slides, bonds, or is seen, pick Ra 0.4 to 1.6 µm and a matching finish; if it is hidden or structural, accept the as-processed default around Ra 1.6 to 6.3 µm. Confirm the achievable number with your shop before freezing the drawing, because some finishes are impossible on certain materials.

FabProto machines, prints, and finishes parts in one facility, so we can tell you the realistic Ra each process delivers on your specific material, and propose the cheapest route to the look and seal you need. Send the model with the faces that matter marked, and we will return a finish plan with achievable Ra, Rz, and cost.

FAQ

What is a good Ra for most prototypes?

For hidden or structural faces, accept the as-processed default, about Ra 1.6 to 6.3 µm from CNC and 6 to 12 µm from printing. Specify finer Ra only on faces that seal, slide, bond, or are seen.

What is the difference between Ra and Rz?

Ra is the average deviation of the whole profile from its mean line; Rz is the average of the five tallest peaks and deepest valleys. Rz captures worst-case peaks that matter for sealing and wear, so critical drawings often list both.

Can 3D-printed parts reach a smooth finish?

Yes, with post-processing. As-printed FDM or SLS shows layer lines around Ra 6 to 12 µm, but vapor smoothing, sanding, or painting can bring resin or nylon parts down toward Ra 1 µm. The base process alone is rarely smooth enough for sealing.

Does a smoother surface cost more?

Almost always. Each finer step adds machining, polishing, or inspection time, and raises scrap risk on delicate features. A matte bead blast is usually the best-value finish for visible parts.

How should I specify roughness on a drawing?

Apply the Ra symbol with a value only to the specific faces that need it, state the sampling length, and add an Rz limit if critical. Avoid mandating a process; a target Ra with tolerance lets the shop choose the cheapest route.

Related Guides

Start a ProjectAll articles