Specifying a roundness tester in 2026 means resolving four sequential decisions, and the order matters: a workshop that locks in a brand before pinning the spindle class or the UPR filter ends up with a precision machine that cannot resolve the lobing signature it bought the instrument to catch.
The class-defining spec is spindle rotational accuracy. Standard models run ±0.025 µm, precision models run ±0.0125 µm, with system accuracy (MZC band, full loop) holding at ≤0.06 µm on standard units and ≤0.03 µm on high-precision units [S1]. A 0.001 µm sensor resolution over a ±250 µm range, paired with a HEIDENHAIN-imported circular grating, is the typical measurement chain on Asian-built units [S1].
Spindle class and accuracy envelope
Spindle accuracy is a hard floor: every downstream reading is the sum of spindle error and probe error, and on a floating air-bearing spindle the bearing error dominates. Workshops chasing bearing-race or crankshaft-journal form should default to the precision class (±0.0125 µm) because the tolerance they are trying to verify is typically 2–5 µm and any spindle error above 5% of that budget corrupts the measurement [S1][S3].
Form tolerance is the most precision-critical callout class in engineering drawings. A bearing race 5 µm out of round produces vibration and premature wear; a crankshaft journal 10 µm barrel-shaped destroys main bearing shells in service; a machine tool spindle 2 µm out of true cannot produce a ground finish [S3]. Hand metrology (calipers, micrometers, bore gauges) measures two-point distances and misses the rotation-dependent form errors that drive these failures, which is the reason a dedicated form tester exists at all.
Reference circle: LSC, MZC, MIC, MCC
The reported roundness number depends on which of the four published reference circles (LSC, MZC, MIC, or MCC) the profile is compared against, so the choice of method affects the result [S3].
LSC (Least Squares Circle) is the workshop default: a best-fit circle minimising the sum of squared radial deviations, balanced, statistically meaningful, and the practical machinist consensus. MZC (Minimum Zone Circle) is the narrowest annular band enclosing all data points; it reports the true peak-to-valley form error and is the ISO 12181-preferred reading for absolute roundness specification. MIC (Maximum Inscribed Circle) is the largest circle that fits inside the profile and gives the effective bearing-fit diameter; MCC (Minimum Circumscribed Circle) is the smallest enclosing circle and gives the effective envelope or shaft diameter [S3].
Selection rule: MZC for drawing-tolerance acceptance, MIC when the downstream question is "will this shaft fit the bore", MCC for sleeve-fit clearance, LSC for batch trending. A buyer who skips this step ends up with a tester that can measure the part but cannot return the value the print asks for.
Filter UPR: the selection most buyers get wrong
Filter cut-off expressed in undulations per revolution (UPR) is the parameter that separates a useful reading from a misleading one, and it must be matched to the manufacturing process generating the surface [S3].
Chuck-induced lobing (a 3-jaw or collet signature) lives at 3 UPR and is typically evaluated at 50 UPR. Honed bearing race finishes carry the cut at 150–500 UPR. Ground surfaces sit higher in the spectrum again. A 50 UPR filter applied to a honed race will smear the cut pattern into an averaged roundness number and hide the very defect the operator is trying to catch; a 500 UPR filter on a chucked part will resolve spindle noise and report an artificially poor result [S3]. ISO 12181-1 governs the Gaussian filter used in 2D form analysis, with ISO 12180-1 and ISO 12180-2 covering cylinder form parameters; the cut-off selection rule lives inside those documents [S3].
Practical selection: chuck turning and shaft grinding default to 50 UPR; honing to 150–500 UPR; superfinishing and lapping above 500 UPR. Buyers should verify the analyser offers 15, 50, 150, 500 and 1500 UPR as selectable bands rather than a single fixed filter.
Workpiece envelope and column reach
The work envelope is the easy spec to read on a datasheet and the easy spec to underspec. A standard bench-top roundness tester accepts Ø3 mm minimum diameter and Ø140 mm maximum diameter, with an optional Ø220 mm column for larger motor shafts, motor commutators, and bearing outer races [S1]. The Z-axis column travel sets the maximum part height; the X-axis radial travel sets the maximum diameter that can be swept without re-fixturing.
Buyers routinely specify the diameter and forget the height. A 1300 mm-tall column instrument (the 750×480×1300 mm cabinet form factor typical of Asian-built units [S1]) accepts a workpiece only as tall as the column travel allows below the probe. For long shaft components, verify the part fits between the table surface and the probe tip before locking the order.
Who a roundness tester is for, and who it is not for
A roundness tester is justified where the print calls out circularity, cylindricity, or runout at tolerances tighter than about 10 µm, and where the part rotates in service. Bearing races, motor commutators, fuel-injector bores, crankshaft journals, hydraulic spools, and machine tool spindles are the canonical applications. Shaft and bearing inspection together drive most capex in this category, with motor commutator roundness, runout, and inter-segment beat a dedicated commutator-measurement software option on instruments aimed at the motor industry [S1].
A roundness tester is not the right tool for first-article inspection of a machined prismatic bracket, for sheet-metal flatness over a 1 m span, or for surface roughness (Ra, Rz) on a milled pocket: those callings live on a CMM or vision system, a flatness table, or a dedicated surface roughness tester, respectively. A buyer looking for a general-purpose shop-floor dimensional tool is overpaying by a factor of 5–10x for an instrument whose spindle accuracy the application will never need.
Selection criteria compared: four realistic configurations
Four configurations cover most 2026 capex asks. (a) Workshop standard, standard spindle class (±0.025 µm), Ø140 mm envelope, LSC + MZC reporting, 50/150/500 UPR filters, commutator option. Suits motor commutator lines, general shaft grinding, workshop batch QA. (b) Bearing-race precision, precision spindle class (±0.0125 µm), Ø140 mm envelope, MZC + MIC + MCC reporting, 15/50/150/500/1500 UPR filters, race-fixture kit. Suits bearing manufacturers and bearing-rebuild shops. (c) Crankshaft/cap-rebuild, precision spindle, Ø220 mm column option, MZC + concentricity + runout, heavy-duty tailstock. Suits engine remanufacture. (d) Lab/research, precision spindle, full MZC/MIC/MCC/LSC set, full filter set, cylindricity + straightness + perpendicularity + parallelism + flatness + slope + spectral analysis in the analysis suite [S1].
Decision shortcut: if the part fits in Ø140 mm and the tolerance is above 5 µm, configuration (a) is the floor. If tolerance is 1–5 µm, jump to (b). If the part exceeds 140 mm diameter or is heavier than 20 kg, go to (c). If the lab needs cylindricity, straightness, and spectral analysis on the same instrument, (d) is the only one that covers the analysis breadth.
Reference data and traceability
Form metrology buyers should require ISO 12181 (filter), ISO 12180-1 / ISO 12180-2 (cylinder form parameters), and a traceable roundness or step artefact in the delivery package for daily verification [S3]. A tester whose accuracy class is quoted as a number on a datasheet but which ships without a traceable artefact cannot be cross-checked against drift; on a 0.03 µm system that is a fatal gap.
Beyond accuracy, the practical evaluation items are: column Z-travel versus the tallest part in production, probe-tip options (standard stylus, deep-reach stylus, needle-tip for grooves), automatic centering and leveling of the workpiece table, pass/fail colour highlighting against stored limits, and segment-by-segment analysis for commutator work [S1]. The commutator software option is the differentiator for motor-industry buyers and worth pricing separately rather than as part of a generic bundle.
Failure modes and constraints
The most common failure mode is environmental. A floating air-bearing spindle on a vibration-isolated table inside a temperature-controlled room (typically 20 ±0.5 °C) is the operating envelope the accuracy class assumes; a tester on a shop floor next to a press is reporting spindle error dressed up as part roundness, and the operator will not know the difference [S3]. The second most common is wrong filter selection, already covered above. The third is column reach: the part is too tall to fit between the table and the probe, and the buyer did not check the Z-travel against the tallest part in production. The fourth is reporting-method mismatch: the print asks for MZC and the analyser is reporting LSC, and the buyer did not verify the method selection on the test certificate before sign-off.
For engineers cross-shopping instruments, the question to bring to a demo is which reference circle the analyser reports by default, which filter UPR values it ships with, and whether the spindle accuracy certificate is a system figure (MZC closed loop) or a spindle-only figure: the two can differ by a factor of two on the same machine [S1].
Bracketing the purchase: shortlist logic
Shortlist by accuracy class first, envelope second, filter set third, software fourth, brand fifth. A precision-class spindle (±0.0125 µm) with 15/50/150/500/1500 UPR filters, Ø140 mm standard envelope, MZC + MIC + MCC + LSC reporting, and a commutator-segment analysis option is the configuration that satisfies roughly 80% of motor and bearing-industry capex asks in 2026. Lift to the Ø220 mm column if the line regularly handles large armature shafts or bearing outer races above 140 mm; drop to a standard-class spindle only if the print tolerance is above 5 µm and budget is binding. Roundness testers sit alongside laser trackers and coordinate measuring machines as form-metrology capital equipment, best served by specialist distributors with applications engineering and demo capability rather than a generic catalog purchase. [S3]
Detailed specification references: deadweight tester, and hardness tester.