A profilometer measures the texture and roughness of a surface by tracing or scanning across it and recording tiny height variations. Whether you’re using a contact stylus instrument or an optical system, the basic workflow involves calibrating the device, setting the right measurement parameters, running the scan, and interpreting the resulting roughness values. Here’s how each step works in practice.
Contact vs. Optical: Choose Your Method
Contact (stylus) profilometers drag a fine-tipped needle across the surface and record a 2D height profile along that line. They’re fully standardized, relatively inexpensive as handheld devices, and widely used in manufacturing quality control. The trade-off is that the stylus physically touches the part, which can scratch very soft materials, and you only get data along a single line per pass.
Optical profilometers use light (usually white-light interferometry or confocal techniques) to scan the surface without touching it. They capture full 3D area maps rather than single-line profiles, giving you a more complete picture of the surface. They’re better for delicate or coated materials and produce statistically more representative measurements. Optical systems cost more and require more setup, but they’re increasingly common in labs and advanced production environments.
If you’re checking a machined metal part on the shop floor, a handheld stylus profilometer is usually the fastest option. If you need a detailed 3D map of a coating, thin film, or additive-manufactured surface, an optical instrument is the better choice.
Calibrate Before You Measure
Every profilometer needs regular calibration to produce trustworthy numbers. The standard approach uses reference specimens: physical artifacts with precisely known dimensions that let you verify your instrument is reading correctly. You’ll typically need three types of reference artifacts:
- Step height standard: A flat specimen with a precisely machined step, used to calibrate the vertical (Z) axis. This is the most critical calibration artifact, classified as a Type A reference specimen under ASME B46.1.
- Periodic pattern: A specimen with evenly spaced grooves (sinusoidal or triangular), used to verify the horizontal (X) axis spacing.
- Roughness artifact: An irregular surface with a certified roughness value, used to verify that the instrument’s software and overall system are calculating parameters correctly.
For contact instruments, you also need to check the stylus tip itself for wear. A worn or chipped tip will round off sharp features and produce artificially smooth readings. Run calibration checks at the start of each measurement session, and verify more frequently during long production runs. If your readings on the reference specimen drift outside the stated uncertainty, replace the stylus tip or recalibrate.
Set Up the Measurement Parameters
Before scanning your part, you need to select the right cutoff length and evaluation length. These settings determine what counts as “roughness” versus the broader waviness or form of the surface, and choosing wrong values will give you misleading results.
The cutoff length (also called the sampling length) acts as a filter. Short cutoff lengths capture only fine-scale roughness, while longer ones include broader features. ISO 4288 provides standard tables that match the cutoff to the expected roughness of your part. For non-periodic surfaces like ground metal, the standard links the cutoff to the expected Ra range. For periodic surfaces (like turned parts with visible tool marks), the cutoff is based on the spacing of those marks. If your drawing or specification doesn’t state a cutoff, use the tables as your default.
The evaluation length is the total distance the stylus travels (or the optical system scans) during the measurement. The standard evaluation length equals five consecutive sampling lengths. So if your cutoff is 0.8 mm, your evaluation length is 4.0 mm. Using fewer than five sampling lengths reduces the statistical reliability of your result.
On a contact profilometer, you’ll also set the traverse speed. Slower speeds improve accuracy but take longer. Follow your instrument’s manual for recommended speeds at each cutoff setting.
Using a Stylus Profilometer
Place your part on a stable, vibration-free surface. Even light vibrations from nearby machinery can introduce noise into the measurement. If your lab doesn’t have an isolation table, at minimum avoid measuring while heavy equipment is running nearby.
Clean the surface first. Dust, oil, or particle buildup will create false peaks in your profile. Use a lint-free wipe with an appropriate solvent for the material. Position the stylus at your starting point, making sure the tip is in contact with the surface but not pressing against an edge or feature that could damage it. Align the measurement direction perpendicular to the dominant surface texture (across the machining marks on a turned part, for example) unless your specification says otherwise.
Start the measurement. The stylus will traverse the surface at a constant speed while a sensor records the vertical displacement of the tip. Most instruments display the raw profile in real time. Once the scan completes, the software applies your selected cutoff filter and calculates the roughness parameters.
Take at least three measurements at different locations on the part, then average the results. Surface roughness can vary across a single component, and a single trace may not represent the whole surface.
Using an Optical Profilometer
The setup for an optical instrument involves more steps but follows a logical sequence. Start by placing your sample on the stage and lowering the light intensity to zero to protect the detector. Select your initial objective lens (lower magnification first to find your area of interest).
Use the XY stage controls to position over an edge or identifiable feature on your sample, then bring the surface into focus by adjusting the Z axis. Start with faster Z movement to get close, then switch to slower speeds as you approach focus. For interferometric systems, you’ll see interference fringes appear on the live image. Adjust the Z height until you reach the point of highest fringe contrast.
Once roughly focused, run the instrument’s autofocus and auto tip/tilt functions. These automatically optimize the fringe pattern so it spreads evenly across the field of view, which is essential for accurate height measurement. If fringes are still bunched after one pass, run auto tip/tilt again. Then adjust the light intensity so the image is well-exposed without saturating the detector.
Switch to your final magnification and refocus. Higher magnification gives finer lateral resolution but covers a smaller area. If you need to measure a larger region than a single field of view, use the stitching function: select a rectangular area, define the corners by moving the stage, set an overlap percentage (typically 10 to 20 percent), and let the instrument scan and stitch the images together automatically. For multiple measurement sites on the same sample, the automation function lets you define a grid or list of XY coordinates and run them in sequence.
Understanding Your Results
After the scan, your software will calculate a set of roughness parameters from the profile data. The three you’ll encounter most often are Ra, Rq, and Rz.
Ra (roughness average) is the most commonly specified parameter in engineering drawings worldwide. It’s the arithmetic average of all the height deviations from the mean line across the evaluation length. Think of it as the “average bumpiness” of the surface. Ra is easy to understand and compare, but it can hide important details. Two surfaces with the same Ra can look very different: one might have many small peaks and valleys, while the other has a few deep scratches.
Rq (root mean square roughness) is calculated similarly to Ra but squares the height values before averaging, then takes the square root. This makes Rq more sensitive to occasional large peaks or valleys. For a perfectly random surface, Rq is about 1.11 times larger than Ra. It’s preferred in optics and semiconductor applications where extreme points matter more than the average.
Rz (average maximum height) captures the most extreme features. It averages the largest peak-to-valley heights within each sampling length across the evaluation length. If your concern is whether a coating will fill all the valleys or whether a seal will leak past the tallest peaks, Rz tells you more than Ra does.
If you’re using an optical profilometer that captures 3D data, you’ll also see areal parameters prefixed with “S” instead of “R” (Sa, Sq, Sz). These are the 3D equivalents and provide a more complete characterization of the surface.
Common Sources of Error
Vibration is the most frequent cause of noisy or unreliable measurements. Even footsteps or air conditioning drafts can affect sensitive instruments. Use a vibration isolation table for optical systems, and avoid placing contact profilometers on the same bench as running equipment.
Temperature changes cause thermal expansion in both the instrument and the sample. If the room temperature shifts during a long measurement session, your results can drift. Allow parts to reach room temperature before measuring, especially if they’ve just come off a machine or out of storage.
High-frequency noise from electronic interference or environmental sources adds artificial texture to the measured profile. Most software includes filtering tools to remove it, but heavy filtering can also erase real surface features. Apply only the filters specified by your measurement standard.
Surface contamination, as mentioned earlier, creates false data points. Optical instruments are particularly susceptible to transparent films or residues that scatter light unpredictably. Spikes and non-measured points (common in optical data from highly reflective or very dark surfaces) need to be identified and handled in software before calculating parameters, since even a few outlier points can distort values like skewness and kurtosis significantly. Repeated measurements of feature-based parameters tend to show the largest variation, so take extra care and additional scans when reporting values like peak density or feature volume.