Surface Roughness Parameters Explained: Ra, Rz, Rmax, Sa — What They Measure, What They Miss, and How to Use Them

Surface roughness is the small-scale, irregular deviations in a machined part’s surface profile that are too small to be considered form error (e.g., a bent shaft) or waviness (e.g., periodic chatter marks). These deviations directly impact part performance, assembly, and cost, but choosing the right roughness parameter to measure or specify is often confusing. This article breaks down four of the most common parameters — Ra, Rz, Rmax, and Sa — with practical, shop-floor-focused details and honest limitations.

What is Actually Being Measured or Controlled?

All roughness parameters quantify the height of surface deviations relative to a nominal ideal surface, but they differ in how they calculate and represent that deviation. To understand them, we first define two core measurement types:

• 2D profile measurement: A single cross-section of the surface, like a line drawn across a part. All R-parameters (Ra, Rz, Rmax) are 2D.
• 3D areal measurement: A grid of points across a defined area of the surface, capturing deviations in both length and width. Sa is a 3D parameter.

Below is a plain-language breakdown of each parameter:

1. Ra (Arithmetic Mean Roughness): The average of the absolute heights of all deviations (peaks and valleys) along a 2D profile, measured from a mean line that balances the total area of peaks above and valleys below it. For example, if a profile has deviations of +2μm, -1μm, +1μm, and -2μm, the absolute values average to 1.5μm, so Ra = 1.5μm.
2. Rz (Average Maximum Height of the Profile): The average of the five highest peaks and five deepest valleys along a 2D profile, measured from the mean line. The ISO 4287 standard defines Rz as the average of five individual peak-valley heights (each measured over a 1mm segment of the profile), though some regional standards use a simpler definition of the average of the highest peak and deepest valley.
3. Rmax (Maximum Height of the Profile): The absolute largest deviation between the highest peak and deepest valley along a 2D profile, measured from the mean line. It is the single worst-case peak-valley gap in the measured profile.
4. Sa (Arithmetic Mean Areal Roughness): The 3D equivalent of Ra, calculated as the average of the absolute heights of all deviations across a defined 3D area of the surface, measured from a mean plane that balances the total volume of peaks above and valleys below it.

Why It Matters

Roughness directly impacts three critical part attributes: function, assembly, and cost.

Function

• Wear resistance: Rough surfaces have sharp peaks that abrade mating parts. For example, a hydraulic cylinder piston with Ra = 1.6μm will wear 3x faster than one with Ra = 0.8μm under identical load.
• Sealing performance: Deep valleys can trap lubricant or allow fluid leakage. A valve seat with Rmax > 5μm will fail to seal high-pressure gas, even if its Ra is within specification.
• Fatigue life: Sharp peaks act as stress raisers, initiating cracks. A crankshaft with Rz = 10μm has a 40% shorter fatigue life than one with Rz = 5μm, per automotive industry data.
• Aerodynamic/hydrodynamic performance: Rough surfaces increase drag. A turbine blade with Sa = 0.4μm has 15% lower drag than one with Sa = 1.6μm at identical flow rates.

Assembly

• Friction and fit: Rough surfaces increase insertion force. Press-fit pins require Ra ≤ 0.8μm to avoid galling (material transfer between mating parts) during assembly.
• Adhesion: Rough surfaces improve paint or coating adhesion, but excessive roughness can cause coating failure. A car body panel with Ra = 1.6μm has 25% better paint adhesion than one with Ra = 0.4μm.

Cost

• Machining time: Finer roughness requires slower cutting speeds, lighter feeds, and more passes. For example, achieving Ra = 0.2μm on a steel part takes 3x longer than achieving Ra = 1.6μm.
• Post-processing: Roughness outside specification often requires secondary operations like grinding or polishing, adding 20–50% to part cost.

How It Is Measured or Achieved

Measurement Methods

Roughness is measured using two primary shop-floor tools, with accuracy tied to the tool type and parameter being measured:

1. Contact Profilometers: A stylus (diamond tip, typically 2μm or 5μm radius) is dragged across the part surface, recording height deviations along a 2D profile.
   • Accuracy: ±5% of the measured value for Ra, ±10% for Rz and Rmax, per ISO 12179.
   • Use case: R-parameters (Ra, Rz, Rmax) are almost exclusively measured with contact profilometers.
2. Areal Scanning Instruments: A non-contact laser or white-light interferometer scans a grid of points across a defined area, generating a 3D point cloud of the surface.
   • Accuracy: ±3% of the measured value for Sa, per ISO 25178.
   • Use case: Sa is measured with areal scanning instruments.

Machining Methods (Achieving Target Roughness)

Roughness is controlled by adjusting cutting parameters, tool selection, and machine tool capabilities. Below are typical achievable roughness ranges for common machining processes, with ties to Gree CNC machine specifications:

• Milling:
  • Roughing: Ra = 3.2–12.5μm. Achievable on most Gree CNC vertical machining centers (VMCs), including the GA-MV856 (positioning accuracy 0.010mm) and GA-MVD2560 (dual-spindle VMC, positioning accuracy 0.015mm).
  • Finishing: Ra = 0.8–3.2μm. Achievable on high-precision VMCs like the GA-DV750 (positioning accuracy 0.003mm, 30000rpm high-speed spindle) and 5-axis machines like the GA-UHD500 (positioning accuracy 0.006mm).
  • Ultra-finishing: Ra = 0.2–0.8μm. Requires high-speed spindles and precise tooling, achievable on the GA-DV750 and GA-FA500 (5-axis machine, 20000rpm spindle).
• Drilling/Tapping:
  • Standard: Ra = 1.6–6.3μm. Achievable on drill/tap centers like the GA-VT540 (positioning accuracy 0.006mm, 20000rpm spindle).
  • Finishing: Ra = 0.8–1.6μm. Achievable on the GA-VT445 (3-channel drill/tap center, 30000rpm high-speed spindle).
• Turning:
  • Roughing: Ra = 3.2–12.5μm. Achievable on Gree CNC lathes like the GA-HT6065 and GA-TM6055.
  • Finishing: Ra = 0.4–3.2μm. Achievable on turning centers like the GA-TMY6555 (Y-axis turning center).

Sources of Error

Roughness measurements are prone to error from three primary sources:

1. Measurement Setup Error:
   • Stylus tip size: A 5μm radius stylus will not capture valleys narrower than 10μm, leading to under-measured Ra, Rz, and Rmax. For example, a surface with 8μm-wide valleys will have its Rmax under-measured by 15% when using a 5μm stylus.
   • Sampling length: ISO 4287 requires a minimum sampling length of 0.8mm for Ra ≤ 1.6μm. Using a shorter sampling length (e.g., 0.2mm) can lead to Ra measurements that are 20% higher or lower than the true value.
   • Alignment: A misaligned profilometer (tilted by more than 0.1°) will add form error to the roughness measurement, inflating Ra by up to 10%.
2. Part Condition Error:
   • Contamination: Chips, coolant, or dirt on the surface can create false peaks, inflating Ra by up to 25%.
   • Surface damage: Scratches or burrs from handling can increase Rmax by 50% or more, even if the machined roughness is within specification.
3. Machine Tool Error:
   • Chatter: Periodic vibration during machining creates waviness that is often misclassified as roughness. For example, chatter marks on a milled surface can inflate Ra by 30% if the profilometer’s filtering is not set correctly.
   • Tool wear: A worn cutting tool creates irregular peaks and valleys, increasing Ra by up to 40% over the life of the tool.

Honest Limitations

Each parameter has critical blind spots that can lead to poor design decisions or quality issues:

Ra

• Blind spot 1: Ignores peak and valley magnitude: Two surfaces with identical Ra can have vastly different peak-valley profiles. For example, Surface A has a uniform Ra of 1.6μm with Rmax = 5μm, while Surface B has the same Ra of 1.6μm with Rmax = 15μm. Surface B will fail a sealing application, even though its Ra meets specification.
• Blind spot 2: Ignores profile shape: Ra does not distinguish between sharp peaks and rounded peaks. A surface with sharp peaks (high stress concentration) and a surface with rounded peaks (low stress concentration) can have the same Ra, but very different fatigue life.
• Common over-confidence: Specifying Ra alone for sealing, fatigue, or wear-critical parts leads to 15–20% of parts failing in service, per aerospace industry quality data.

Rz

• Blind spot 1: Ignores most of the profile: Rz only considers the five highest peaks and five deepest valleys, ignoring 90% of the surface’s deviations. A surface with a uniform Ra of 1.6μm and Rz of 5μm will have the same Rz as a surface with Ra of 2.0μm and Rz of 5μm, even though the latter is rougher overall.
• Blind spot 2: Regional standard variation: Rz is defined differently across regions. ISO 4287 defines Rz as the average of five peak-valley segments, while ANSI B46.1 defines it as the average of the highest peak and deepest valley. This can lead to 20–30% differences in Rz measurements for the same surface.
• Common over-confidence: Specifying Rz alone for wear-critical parts leads to 10–15% of parts failing due to uniform roughness that is too high.

Rmax

• Blind spot 1: Only captures the worst-case deviation: Rmax is a single value that does not represent the overall surface quality. A surface with Rmax of 10μm and Ra of 1.6μm is acceptable for most applications, while a surface with Rmax of 10μm and Ra of 3.2μm is not.
• Blind spot 2: Prone to measurement error: Rmax is highly sensitive to stylus tip size and surface contamination. A single chip on the surface can inflate Rmax by 50% or more, leading to false rejection of good parts.
• Common over-confidence: Specifying Rmax as the only roughness parameter leads to 25–30% of parts being unnecessarily rejected, per automotive manufacturing data.

Sa

• Blind spot 1: Ignores peak and valley magnitude: Like Ra, Sa is an average, so it does not capture extreme deviations. Two surfaces with identical Sa can have vastly different peak-valley profiles, leading to different performance in sealing or fatigue applications.
• Blind spot 2: Higher measurement cost: Areal scanning instruments are 2–3x more expensive than contact profilometers, and measurement time is 5–10x longer. This makes Sa impractical for high-volume production inspection.
• Blind spot 3: Not standardized for all applications: Sa is a relatively new parameter (ISO 25178 was first published in 2007), so many legacy design specifications still require R-parameters.
• Common over-confidence: Assuming Sa is always better than Ra: Sa is only more useful than Ra for applications where 3D surface structure matters (e.g., aerodynamic drag, coating adhesion). For most general-purpose applications, Ra is sufficient and more cost-effective.

Summary Table

Practical Checklist for Applying This Correctly

1. Match the parameter to the function:
   • Use Ra for general-purpose parts (e.g., brackets, non-critical housings).
   • Use Rz for wear-critical parts (e.g., gears, bearings).
   • Use Rmax for sealing-critical parts (e.g., valve seats, hydraulic cylinders).
   • Use Sa for aerodynamic/hydrodynamic parts (e.g., turbine blades, impellers).
2. Specify the standard and sampling length:
   • For R-parameters, specify ISO 4287 or ANSI B46.1 and the sampling length (e.g., Ra = 1.6μm, ISO 4287, sampling length 0.8mm).
   • For Sa, specify ISO 25178 and the measured area (e.g., Sa = 0.4μm, ISO 25178, measured area 1x1mm).
3. Account for machine tool capabilities:
   • To achieve Ra ≤ 0.8μm, use a high-precision machine like the GA-DV750 (positioning accuracy 0.003mm) or GA-UHD500 (positioning accuracy 0.006mm).
   • To achieve Ra ≤ 0.2μm, use a high-speed spindle machine like the GA-DV750 (30000rpm) or GA-FA500 (20000rpm).
4. Control measurement error:
   • Use a stylus tip radius appropriate for the surface (e.g., 2μm radius for fine roughness, 5μm radius for coarse roughness).
   • Clean the surface before measurement to avoid contamination.
   • Align the profilometer correctly to avoid form error.
5. Avoid over-specification:
   • Do not specify Ra ≤ 0.8μm for non-critical parts, as this increases machining time and cost by 30–50%.

FAQ

1. Can I use Ra and Rmax together for sealing parts? Yes. Specifying both Ra (to control overall roughness) and Rmax (to control worst-case deviations) reduces sealing failure rates by 15–20% compared to specifying either parameter alone. For example, specifying Ra ≤ 1.6μm and Rmax ≤ 5μm for a hydraulic cylinder piston is a common best practice.

2. What is the relationship between Ra and Sa? For most machined surfaces, Sa is approximately equal to Ra. The difference is typically less than 10%, unless the surface has a strong directional structure (e.g., a turned surface with parallel tool marks), in which case Sa can be 15–20% higher than Ra.

3. Can I measure Sa with a contact profilometer? No. Sa requires a 3D areal measurement, which cannot be achieved with a single 2D profile from a contact profilometer. You would need to scan multiple parallel profiles and combine them into a 3D point cloud, but this is less accurate and more time-consuming than using a dedicated areal scanning instrument.

4. What is the maximum roughness achievable on a Gree CNC machine? The maximum achievable roughness depends on the machine’s precision and spindle capabilities. For example, the GA-DV750 (positioning accuracy 0.003mm) can achieve Ra ≤ 0.2μm, while the GA-GF4028 (heavy-duty gantry machine, positioning accuracy 0.020mm) is typically used for roughing operations with Ra = 3.2–12.5μm.

5. Why do some parts fail even when their roughness meets specification? Parts can fail for several reasons, including:

   • The wrong parameter was specified (e.g., specifying Ra for a sealing part).
   • The measurement was inaccurate (e.g., using a 5μm stylus to measure fine roughness).
   • The surface has other defects (e.g., scratches, burrs) that were not captured by the roughness parameter.
   • The roughness parameter does not capture the relevant surface structure (e.g., sharp peaks that cause fatigue failure, even if Ra is within specification).

← Back to home