Surface Roughness: Do You Really Understand It?

• 1
  What surface roughness actually measures
• 2
  Ra, Rz and Rq: which number belongs on the drawing
• 3
  How cutting parameters set surface roughness
• 4
  Why the same insert gives different results on different metals
• 5
  Measuring surface roughness without fooling yourself
• 6
  Typical Ra ranges and what they suit

Surface roughness describes the small height deviations left on a machined face. A stylus or optical sensor traces a short line across the surface, and the instrument reports the average deviation from a mean line. That average is Ra, written in micrometers or microinches. The line is only a few millimeters long, so the number describes the sampled path, not the whole part.

This is why two parts can feel identical to a finger yet carry different Ra values. The finger reads texture over a broad area at low sensitivity. The instrument reads a filtered profile at micron resolution. When a drawing says Ra 1.6 μm, it is asking for a measured number under a defined cutoff, not a visual impression.

Ra also says nothing about direction. A face milled with a fine feed shows parallel tool marks; a bead-blasted face shows a random pattern. Both can land near the same Ra. If the function depends on sliding direction, sealing, or fluid flow, the lay pattern matters as much as the average height.

• 1
  RaArithmetic mean of profile height over the sampling length. The default callout on most drawings.
• 2
  RzMean peak-to-valley height of five sampling segments. Sensitive to deep scratches and single high peaks.
• 3
  LayDirection of the dominant tool marks. Critical for seals, bearings and flow surfaces.

Ra is an average, so a single deep scratch barely moves it. Rz takes the mean of peak-to-valley heights, so one scratch moves it a lot. For a static locating face, Ra is usually enough. For a sealing face, a bearing race, or a fatigue-critical fillet, Rz or Rmax catches the defects that actually cause leaks and crack initiation.

Cutoff length is the other half of the callout. The instrument applies a high-pass filter to strip out waviness and keep roughness. A 0.8 mm cutoff suits fine finishes; a 2.5 mm cutoff suits rougher ones. Measure the same surface with two cutoffs and the Ra values differ. Quote the cutoff or the number is not reproducible.

Rq is the root-mean-square height. It weights peaks more heavily than Ra and shows up in optics, tribology and some aerospace specs. Rsk tells you whether the profile leans toward peaks or valleys. Rku flags spiky versus rounded profiles. These are second-tier numbers; add them only when the function justifies the extra inspection time.

• 1
  Ra for general facesLocating pads, covers, brackets, non-sealing surfaces.
• 2
  Rz for sealing and fatigueO-ring grooves, hydraulic bores, fillets under cyclic load.
• 3
  Rq when peaks matterOptical surfaces, wear couples, some coating adhesion specs.

On a turned or milled face, the dominant pattern comes from the tool path. The tool leaves a scallop each time it passes. Feed per tooth and tool nose radius set the height of that scallop through a simple geometric relation: a larger nose radius and a smaller feed produce a shorter scallop. This is why finishing passes use small feeds and radiused inserts rather than sharp corners.

A sharp corner leaves a tall scallop even at a light feed. A large nose radius flattens it, but a radius that is too large for the feature can rub instead of cut, which raises cutting forces and can chatter. The practical window is a compromise between the geometric finish and the stability of the cut.

Vibration and tool wear add a second layer. Chatter prints a periodic waviness that the roughness filter may partly remove, but the underlying surface is still damaged. A worn edge rubs and burnishes instead of shearing, producing a smeared finish that looks bright and reads low on Ra while hiding torn metal underneath. Inspect under magnification, not just by number.

• 1
  Smaller feedLowers scallop height roughly with the square of the feed.
• 2
  Larger nose radiusFlattens scallops but risks rubbing on tight features.
• 3
  Rigid setupRemoves chatter that no parameter change can hide.

Aluminum 6061 and 7075 cut clean and take a fine finish easily. Built-up edge is the main enemy. A sharp, polished tool with plenty of rake and a light feed keeps the surface bright. Stainless 304 and 316 work-harden at the cut, so a dull edge or a dwell in the cut raises both roughness and tool load. Sharp edges and constant feed matter more than chasing a low Ra number.

Titanium Ti-6Al-4V conducts heat poorly, so the edge runs hot and wears fast. Surface finish drifts as the tool wears, which means the first part and the last part in a run can differ. Inconel behaves the same way but harder. On these alloys, plan for tool change intervals based on measured finish, not on a fixed part count.

Plastics and copper alloys have their own rules. POM and PEEK can be cut to a very fine finish but scratch easily during handling. Brass C36000 machines freely and takes a near-mirror finish with a sharp tool. Beryllium copper needs care because of dust control. In every case, the finish on the drawing must be achievable in the chosen material, not just in the chosen machine.

• 1
  AluminumWatch built-up edge; use polished, high-rake tools.
• 2
  Stainless and titaniumWatch work hardening and tool wear; change edges on measured drift.
• 3
  PlasticsFinish is easy, damage during handling is not.

A portable skidded gauge is the shop floor standard. The skid rides the surface and the stylus follows it. On a curved or short face the skid can tip, and the reading drifts. For bores, grooves and small features, use a skidless pickup or a flexible fixture. Clean the surface first; a fingerprint or a chip changes the result.

Direction matters. Measure across the lay for the worst-case height, and along the lay when you care about flow. State which one you used. On a milled face, a reading taken across the tool marks can be several times higher than one taken along them.

For production control, measure at defined locations on a defined sample size. A single reading on one part proves nothing about a run. At GreatLight, we run 100% inspection before shipment and can supply roughness reports on request, with raw material checks and in-process monitoring feeding into the final record.

• 1
  Fix the partLoose fixturing adds vibration to the reading.
• 2
  State the directionAcross lay versus along lay changes the number.
• 3
  Sample the runOne reading on one part is not process control.