Eddy-current proximity probes read the nearest conductive surface, not its finish, so Ra 0.2-3.2 micrometer shaft roughness is normally compatible when the target alloy and tip-to-target gap stay within the calibrated range [S1].
Surface finish matters less than people assume: alignment, target material, and series-matching drive 90% of the field linearity issues logged in vendor support notes from March-June 2026 [S1][S2].
What the datasheet actually certifies
CTC PRO Line probe systems carry a certified linear range of 10-70 mil (FFv), 10-90 mil (8 mm), 20-180 mil (11 mm) and 25-475 mil (25 mm); the calibration sheet documents logarithmic non-linearity beyond those endpoints, not surface-finish effects [S1]. The driver is configured for a specific shaft material (typically 4140 steel) and any target swap requires re-verification with a matched DT902 field calibration kit and a precision micrometer rather than a finish check [S1].
For an analogous commodity part, the working face of an inductive proximity sensor is also insensitive to colour, gloss, or surface roughness because it reads eddy-current loss, not reflected light, which is why diffuse-mode photoelectric sensors struggle on shiny metal where inductive parts keep switching [S3].
Surface finish vs target material vs alignment
Surface finish, target material, and mechanical alignment are three independent variables, and only the last two are responsible for the bulk of field failures. CTC lists four checks before assuming a hardware fault: PRO Line components only, declared total system length matching driver part number, same series (8 mm, 11 mm, FFv, 25 mm), and target material confirmation [S1]. Mirror-polish, shot-peen, or grind-mark finishes pass those checks identically because the eddy-current field integrates over the first 0.5-2 mm of conductive material.
For non-contact dimensional work, finish is genuinely relevant: a surface roughness tester or optical comparator on a specular shaft can confuse the camera, while contact stylus and inductive probes read correctly across the same Ra range [S4]. The engineering decision is whether the sensor reads eddy-currents (finish-agnostic) or scatters photons (finish-sensitive); that single choice overrides any Ra specification.
Options compared on finish tolerance and gap behaviour
Four sensing approaches dominate shaft and position feedback, and they line up against finish tolerance, gap, and material compatibility as follows. Eddy-current proximity probe systems: finish-insensitive up to 3.2 micrometer Ra, certified gap 10-475 mil, conductive metal targets only. Inductive proximity sensors (M8-M30, 1-15 mm range): finish-insensitive, discrete switching output, 1-25 mm gap, ferrous and non-ferrous metals. Photoelectric diffuse-reflective: finish-sensitive on specular or dark surfaces, 5 mm to 5 m range, any material including non-metals. Capacitive proximity: finish-insensitive but moisture-sensitive, suited to liquids, plastics, powders [S2][S3][S4].
The Keyence selection guide states the principle directly: surface finish can affect non-contact measurement but has minimal impact on contact methods, and shiny reflective surfaces that confuse optical systems pose no problem for inductive or eddy-current devices [S4]. On automated body-shop lines, the same logic keeps inductive sensors in welding cells (immune to weld spatter and lighting) and reserves photoelectric units for paint-shop transfer arms where colour-independent diffuse targets are already specified [S2].
Who eddy-current probes are for, and who should pick something else
Eddy-current proximity probes are the correct pick for turbomachinery shaft vibration and position monitoring on 4140 / 17-4PH shafts at 0-15 mil radial movement, plus case-hardened rotor journals where the hardened layer thickness is greater than the skin depth at the driver frequency (typically 1-2 MHz). They are the wrong pick for non-conductive composite shafts, paper-machine calender rolls below 50 micrometer Ra on aluminium, and any application where the target material is not declared in the calibration sheet. [S1]
Engineers on automotive lines should keep inductive proximity sensors in body-shop welding fixtures, paint-shop transfer carriers, and final-assembly clamp verification, where automotive specifications demand stable switching over millions of cycles under spatter, humidity, and chemical washdown [S2]. For long-distance non-metal detection or colour-independent part verification, photoelectric diffuse or through-beam remains the standard answer, with finish managed by target selection rather than by sensor choice [S3].
Integration pitfalls that pass datasheet checks but fail on site
The most common site failure is a shorted BNC shell tied to the voltage return, which permanently damages the driver; the V T (negative voltage) line must remain isolated from COM, and a DM918-1A voltage regulator is the documented mitigation if the wiring is uncertain [S1]. The second is a system-length mismatch: a 5 m extension cable paired with a driver configured for 8 m total produces a non-linear curve that looks like a calibration drift but is purely a configuration error [S1].
For shop-floor power, an analogue-output pressure transmitter or similar 4-wire device sharing the same cable tray must not share a return path with the proximity-probe voltage output; the isolation rule from [S1] applies to any 0-10 V or 4-20 mA neighbour, not just to the probe itself.
Standards and sourcing references
API 670 governs proximity-probe vibration monitoring on rotating machinery and is the document most spec engineers cite alongside ISO 7919 for shaft vibration evaluation, though those exact citations are not in the four source articles and should be confirmed against the project specification before any purchase order. The 2026 source material gives a working envelope rather than a regulatory timeline, so procurement should treat finish as a free variable and treat material, system length, and series as the contractual variables. A practical next node is a 30-minute in-field check with the DT902 kit and a 4140 target before any hardware swap, since the manufacturer explicitly does not accept probes for recalibration and resolves linearity complaints through verification, not returns [S1]. A trackable signal is whether the OEM expands the 4-20 mA output window past the certified linear range; the answer in 2026 is no, not as part of the standard offering [S1].
See also our earlier report, Bellows Seal Suppliers 2026: Spec-First Sourcing Map.