What flatness measurement tells you
Sealing faces, mounting surfaces, wafers, optical components, heat-sink interfaces, PCBs and moulded housings all depend on being flat. Too much form error and a seal leaks, a joint carries load on its edges, a wafer cannot be chucked, or a component stresses its solder joints. Flatness and warpage measurement quantifies that error and shows its shape, so its cause can be found.
Full-field rather than point-by-point
Traditional flatness checks sample a surface at a handful of points with a probe or indicator, then infer the plane. Any warpage between the points is missed. A profilometer measures the whole surface as a dense 3D map, so the flatness deviation is calculated from every point and the pattern of distortion (bow, twist, saddle, edge roll-off) is visible directly.
Flatness parameters
ISO 12781-1 defines flatness as deviation from a reference plane:
- FLTt: total peak-to-valley deviation, the value usually compared with a tolerance;
- FLTp: the highest point above the reference plane;
- FLTv: the deepest point below it;
- FLTq: the root-mean-square deviation, a more stable measure of overall form.
The reference can be a least-squares plane or a minimum-zone plane. Beyond flatness, the software fits planes and polynomial surfaces to describe warpage, and filters the residual into waviness and roughness so each scale can be assessed separately.
Why stitching matters for flatness
Instruments with a small field of view, such as microscopes and interferometers, measure large surfaces by stitching tiles together. Each seam carries a small tilt or offset, and over many tiles these errors add up to apparent form error: the very thing a flatness measurement is meant to detect. Chromatic confocal profilometers scan continuously, with travel up to 400 × 400 mm on the ST500, so the whole surface is measured without seams and the flatness result reflects the part rather than the stitching.
Flatness in production
Flatness is often a go/no-go criterion. Automated recipes measure a defined area, calculate the deviation and apply pass/fail limits, and NANOVEA’s machined parts inspection shows the same approach applied to dimensions against a CAD model. For high volumes, in-line sensors check parts on the line.

















