Surface finish and form measurement

Surface texture (roughness, waviness, lay) and form (straightness, flatness, roundness, cylindricity) definitions, parameters, instruments, filtering, errors, formulas, and solved numericals used in metrology/inspection.

Drafted with Aria, reviewed by the AiCanCode.org team · Updated . How we write and check our notes. Spotted an error? Use Give Feedback at the bottom of the page.

Why it matters

Surface finish and form directly affect friction, wear, sealing, fatigue strength, lubrication, and assembly fit. Measuring them correctly lets you accept good parts, improve processes (e.g., turning vs grinding vs lapping), and diagnose machine-tool errors before they scrap batches.

Key ideas

  • Surface texture: the small-scale geometric features of a surface. It has three components:

    • Roughness: fine irregularities from the manufacturing process (short wavelength).
    • Waviness: more widely spaced undulations from tool/machine deflection, vibration, etc. (longer wavelength than roughness but shorter than form).
    • Lay: the predominant direction of surface pattern (e.g., along feed marks). Always measure roughness perpendicular to lay unless specified.
  • Form (geometric form): the large-scale deviation of a surface from its ideal shape: straightness, flatness, roundness (circularity), cylindricity, etc.

  • Key roughness parameters (profile-based, as commonly used on drawings):

    • Ra (arithmetical mean deviation): average of the absolute profile deviation from the mean line over a sampling length.
    • Rq (root mean square roughness): RMS of profile deviation; more sensitive to peaks/valleys than Ra.
    • Rz (maximum height of the profile): in ISO usage (ISO 4287), within each sampling length Rzi = Rp + Rv, where Rp is the highest peak height and Rv is the deepest valley depth from the mean line; the reported Rz over an evaluation length is the mean of the Rzi values from the consecutive sampling lengths. Note: some legacy definitions use a “ten-point height” average; always check which definition the drawing uses.
    • Rt (total height of the profile): vertical distance between the highest peak and deepest valley over the evaluation length.
    • Other families (when specified): RSm (mean spacing), material ratio (Abbott–Firestone) curve parameters like Rk, Rpk, Rvk (useful for bearing surfaces).
  • Length scales and filtering (why roughness ≠ waviness):

    • Sampling length (λc or cutoff): wavelength boundary between roughness and waviness; selected based on expected roughness.
    • Evaluation length (Le): by ISO 4288, typically Le = 5 × sampling length for stable statistics (unless otherwise specified).
    • Filtering: a standardized filter (commonly Gaussian) separates roughness from waviness. Wrong cutoff leads to wrong Ra.
    • Practical ISO 4288 cutoff selection (contact stylus): standard lc values are 0.08, 0.25, 0.8, 2.5, 8.0 mm. Typical rule vs Ra (µm): 0.006–0.02 → 0.08 mm; 0.02–0.1 → 0.25 mm; 0.1–2 → 0.8 mm; 2–10 → 2.5 mm; 10–50 → 8.0 mm.
  • Contact vs non-contact roughness measurement:

    • Stylus profilometer: a diamond stylus (tip radius typically 2–10 µm) traces the profile.
      • Skidless type references an internal datum (good for waviness and roughness). Skid type references local surface (suppresses waviness; use carefully).
      • Stylus dynamics and tip radius act like an additional filter: features sharper than about 2–3× the tip radius may be underestimated.
    • Optical methods: white-light interferometry, confocal and focus-variation microscopy; good for areal parameters (Sa, Sq) and very smooth surfaces. Sensitive to reflectivity, slope, and contamination.
    • AFM: nanometre-scale characterization (special cases; slow, small area).
  • Form measurement fundamentals and definitions:

    • Straightness (of a line): minimum zone: two parallel lines, with minimum separation, that enclose the line profile. Straightness error = that minimum separation.
    • Flatness (of a plane): minimum zone: two parallel planes, with minimum separation, that enclose all measured points. Flatness error = that separation.
    • Roundness (circularity): minimum zone circle (MZC): two concentric circles with minimum radial separation enclosing the profile; roundness error = radial separation.
    • Cylindricity: minimum zone coaxial cylinders with minimum radial separation enclosing the surface.
    • These “minimum zone” definitions match GD&T (and eliminate bias from least-squares fits when outliers exist).
  • Instruments and setups for form:

    • Surface plate + straightedge + feeler gauges: quick checks of straightness/flatness (resolution limited by gauge leaves and light-gap judgment).
    • Autocollimator or electronic level with a reflective target: measures local slope; integrate slopes to reconstruct vertical deviations and compute straightness/flatness.
    • Optical flat with monochromatic light: fringe patterns indicate flatness and step heights; one fringe step corresponds to a known height increment linked to wavelength.
    • Roundness testers (spindle-based, e.g., Talyround): a high-precision spindle rotates the part while a radial probe records the polar profile; software fits LSC/MZC; can perform harmonic analysis (lobing). Many systems numerically suppress the first harmonic (eccentricity) or re-center the profile; MZC inherently accounts for centering when minimizing the zone.
    • V-block and dial indicator (two-point method): quick assessment of out-of-round; note it can mask certain lobing orders (e.g., 3-lobe on 60° V), so it does not guarantee true roundness.
    • CMM: point or scanning probes fit lines, planes, circles, cylinders; useful for full form analysis but needs correct fitting algorithm selection (LSC vs MZC).
  • Alignment and metrology principles that affect accuracy:

    • Abbe’s principle: the measurement scale must be collinear with the dimension being measured to avoid sine error. Sine (Abbe) error ≈ offset × angle.
    • Cosine error: if the probe’s motion is at angle θ to the true measurement direction, measured length = true length × cosθ; error ≈ (θ²/2) × length for small θ.
    • Environmental effects: vibrations, temperature drift, dirt/oil films, and stylus wear introduce bias and noise.
  • How real devices differ from ideal:

    • Stylus inertia and finite tip radius smooth sharp valleys/peaks.
    • Optical systems can miss steep slopes, saturate on highly reflective or very dark coatings, and be sensitive to refractive index of films.
    • Spindle roundness testers add spindle errors; software often removes first harmonic (eccentricity) to report true roundness.
    • Two-point diameter methods conflate size, eccentricity, and certain lobing; they are not equivalent to roundness per MZC.

Formulas

Ra = (1/L) ∫₀ᴸ |y(x)| dx

  • y(x): profile deviation from mean line, in µm or m; L: sampling length (m). Applies to 2D profile roughness over a sampling length after proper filtering.

Rq = √[(1/L) ∫₀ᴸ y(x)² dx]

  • y(x): as above; L in m. RMS roughness over a sampling length after filtering.

Rz = Rp + Rv

  • Rp: maximum peak height above the mean line within the sampling length (µm); Rv: maximum valley depth below the mean line (µm). ISO usage within a sampling length; over an evaluation length, Rz is the mean of the Rzi values from the consecutive sampling lengths (ISO 4287/4288). Check drawing’s standard before use due to legacy variants.

Rt = y_max − y_min

  • y_max, y_min: highest and lowest ordinates within the evaluation length (µm). Total height of profile across full evaluation length.

Straightness error E_line = max(y_i − y_fit) − min(y_i − y_fit)

  • y_i: reconstructed vertical deviations at measured stations (µm); y_fit: best-fit straight line (method must be stated: least-squares or minimum-zone algorithm). Applies to line straightness from slope or height data.

Flatness error E_plane = max(z_i − z_fit) − min(z_i − z_fit)

  • z_i: measured heights (µm); z_fit: fitted plane (specify least-squares or minimum-zone). Applies when 3+ points measured on a nominal plane.

Roundness error (MZC) Δr = R_max − R_min

  • R_max: maximum radius from center of the minimum zone circles (µm); R_min: minimum radius (µm). Applicable to circular sections.

Δh ≈ L · α (small-angle approximation)

  • Δh: height change between two stations (m); L: horizontal spacing between stations (m); α: measured slope angle in radians. Autocollimator/electronic level based straightness/flatness reconstruction.

1 arcsecond = (π / 648000) rad ≈ 4.848 × 10⁻⁶ rad

  • Unit conversion used in slope-to-height calculations.

h_step = m · (λ / 2)

  • h_step: step height that shifts interference fringes by m bands (m); λ: monochromatic light wavelength (m); m: integer fringe order difference. Applies to optical flat fringe interpretation under monochromatic light.

Abbe (sine) error e ≈ h · θ

  • e: length error (m); h: offset between measurement line and scale (m); θ: small angular misalignment in radians. Applies when scale is not collinear (Abbe’s principle).

Cosine error ≈ L · (1 − cosθ) ≈ L · θ² / 2

  • L: true length (m); θ: small angular misalignment (radians). Applies when probe direction is at angle θ to the measurand.

Worked examples

Example 1 — Direct roughness parameters from discrete data (one sampling length) Given: A stylus profilometer records profile deviations (relative to mean line) at 10 equally spaced points across one sampling length: y in µm = [−0.8, +1.2, −0.5, +0.9, −1.1, +0.6, −0.4, +0.8, −0.7, +1.0]. Compute Ra, Rq, Rz (ISO), and Rt.

Steps:

  1. Ra definition (discrete approximation): Ra ≈ (1/n) Σ |y_i|. Here n = 10.
    • Σ|y_i| = 0.8 + 1.2 + 0.5 + 0.9 + 1.1 + 0.6 + 0.4 + 0.8 + 0.7 + 1.0 = 8.0 µm.
    • Ra = 8.0/10 = 0.8 µm.
  2. Rq definition (discrete): Rq ≈ √[(1/n) Σ y_i²].
    • Squares: 0.64, 1.44, 0.25, 0.81, 1.21, 0.36, 0.16, 0.64, 0.49, 1.00; Σ = 7.00 µm².
    • Mean square = 7.00/10 = 0.70 µm²; Rq = √0.70 = 0.8367 µm.
  3. Rz (ISO): Rz = Rp + Rv over the sampling length.
    • Rp = max positive = 1.2 µm; Rv = max depth = 1.1 µm; Rz = 2.3 µm.
  4. Rt over the evaluation length (here same span as data): Rt = y_max − y_min = 1.2 − (−1.1) = 2.3 µm.

Answer: Ra = 0.8 µm; Rq = 0.837 µm; Rz = 2.3 µm; Rt = 2.3 µm. Final: Ra = 0.8 µm, Rq = 0.837 µm, Rz = 2.3 µm, Rt = 2.3 µm.

Example 2 — Straightness from autocollimator slope readings (GATE-style) Task: A machine guideway’s straightness is checked at 6 stations spaced 100 mm apart (0 to 500 mm). An autocollimator with a flat mirror target gives local slope angles α at each station. Assume each α applies to the following segment (small-angle). Readings (arcsec): at x = 0, 100, 200, 300, 400, 500 mm: α = [0, +5, +5, 0, −5, —] arcsec. The last station has no following segment. Compute straightness error over 0–500 mm using a best-fit straight line through end points and then minimizing the zone by vertical shift only (stated assumption).

Given data and constants:

  • Station spacing L_seg = 100 mm = 0.1 m.
  • 1 arcsec = 4.848 × 10⁻⁶ rad.

Steps:

  1. Convert slopes to height increments between stations using Δh ≈ L_seg · α (radians).
    • α(0→1) = 0 arcsec ⇒ 0 rad ⇒ Δh₀₋₁ = 0.1 m × 0 = 0 m.
    • α(1→2) = +5 arcsec ⇒ 5 × 4.848e−6 = 2.424e−5 rad ⇒ Δh₁₋₂ = 0.1 × 2.424e−5 = 2.424e−6 m = 2.424 µm.
    • α(2→3) = +5 arcsec ⇒ Δh₂₋₃ = 2.424 µm.
    • α(3→4) = 0 arcsec ⇒ Δh₃₋₄ = 0 µm.
    • α(4→5) = −5 arcsec ⇒ −2.424 µm.
  2. Reconstruct cumulative heights h at stations starting from h₀ = 0:
    • h₁ = 0 µm; h₂ = 2.424 µm; h₃ = 4.848 µm; h₄ = 4.848 µm; h₅ = 2.424 µm.
  3. Fit a straight line through end points (assumption) to remove overall tilt: from (x=0, h=0) to (x=500 mm, h=2.424 µm). Line value at x: h_line(x) = (2.424/500) · x µm.
    • Slope = 2.424/500 = 0.004848 µm/mm.
  4. Compute residuals r_i = h_i − h_line(x_i):
    • x₀=0: r₀ = 0 − 0 = 0 µm.
    • x₁=100: r₁ = 0 − 0.4848 = −0.4848 µm.
    • x₂=200: r₂ = 2.424 − 0.9696 = 1.4544 µm.
    • x₃=300: r₃ = 4.848 − 1.4544 = 3.3936 µm.
    • x₄=400: r₄ = 4.848 − 1.9392 = 2.9088 µm.
    • x₅=500: r₅ = 2.424 − 2.424 = 0 µm.
  5. Minimum-zone by vertical shift only (common quick method): choose constant c so that the max positive and min negative residuals are equal in magnitude: c = (r_max + r_min)/2.
    • r_max = 3.3936 µm; r_min = −0.4848 µm; c = (3.3936 − 0.4848)/2 = 1.4544 µm.
    • New extremes: r_max' = 3.3936 − 1.4544 = 1.9392 µm; r_min' = −0.4848 − 1.4544 = ��1.9392 µm.
  6. Straightness error (zone width): E_line = r_max' − r_min' = 1.9392 − (−1.9392) = 3.8784 µm over 500 mm.

Answer: Straightness error = 3.8784 µm over 500 mm. Final: E_line = 3.88 µm (500 mm). Assumptions stated: small-angle; slope applies to following segment; end-point line detrending and vertical equalization used in lieu of full minimum-zone algorithm.

Common mistakes

  • Using Ra alone to judge functional performance where Rz/Rt or material ratio parameters matter (e.g., sealing, plateau honing).
  • Confusing Rz definitions (ISO vs legacy “ten-point”) and comparing apples to oranges.
  • Measuring roughness parallel to lay (gives artificially low Ra for turned or ground surfaces with clear lay).
  • Treating two-point diameter/runout as roundness; not accounting for lobing and eccentricity.
  • Ignoring Abbe and cosine errors in setups, especially with offset probes or tilted parts.
  • In autocollimator straightness, forgetting to convert arcseconds to radians, or using degrees directly; or summing slopes without clear segment association.
  • Not stating the fitting rule (least-squares vs minimum-zone) when reporting straightness/flatness/roundness.

For GATE ME

  • Conceptual: definitions of roughness parameters (Ra, Rq, Rz, Rt), differences between roughness and waviness, sampling length vs evaluation length, lay and measurement direction.
  • Numericals: compute Ra/Rq/Rz from discrete data; convert autocollimator angle readings to straightness; calculate roundness error from max/min radius; interpret optical flat fringes to get step height.
  • Instrumentation: identify which instrument for which task (stylus profilometer vs optical interferometer; autocollimator vs straightedge; roundness tester vs V-block), and pros/cons.
  • Errors: Abbe’s principle, cosine error, stylus tip radius effect, spindle eccentricity (first harmonic) vs lobing (higher harmonics).
  • Filtering/cutoff selection: basic rules and their effect on reported parameters; know that standardized Gaussian filters are typical and the ISO 4288 lc selections listed above are commonly used.

Practise:

  • Quick calculations with arcseconds-to-µm over set baselines.
  • Discrete Ra/Rq/Rz with careful absolute values and squares.
  • Spot which metrology setup would mask a 3-lobed vs 2-lobed profile.
  • Read a simple fringe diagram to compute step height using h = m·λ/2.

Quick check

  1. State the difference between sampling length and evaluation length in surface roughness measurement.
  2. A circular profile has R_max = 20.010 mm and R_min = 19.990 mm (about the same center). What is the roundness error by the minimum-zone definition?
  3. An autocollimator reads 10 arcsec over a 200 mm segment. What is the approximate height change across the segment?

Answers: 1) Sampling length is the length over which roughness parameters are evaluated after filtering (sets the roughness–waviness boundary), while evaluation length is the total length used for statistics (often several sampling lengths; ISO 4288 default Le = 5·lc). 2) Δr = R_max − R_min = 0.020 mm = 20 µm. 3) Δh ≈ L·α = 0.2 m × (10 × 4.848×10⁻⁶) = 9.696×10⁻⁶ m ≈ 9.70 µm.

Finished this topic? Mark it so your progress, study plan and readiness keep up.

Stuck on something here?