Selecting between capacitive, eddy-current, and LVDT displacement sensors comes down to four variables: measurement range as a multiple of probe diameter, target material and size, contamination in the gap, and required bandwidth. The three technologies do not overlap much in their sweet spots, and the wrong choice costs resolution or buys unnecessary hardware.
For a quick rule of thumb drawn from manufacturer guidance: capacitive probes need a sensing element roughly 8x the desired range, eddy-current coils need about 3x the range, and an LVDT sensor can stroke from a few microns to plus or minus 1 m on the same electrical topology [S5]. A capacitive sensor will resolve sub-nanometre motion on a grounded metal target, while an eddy current tester-class displacement probe tolerates oil, coolant, and dust that would swamp a capacitive field [S3][S6].
Operating envelope: range versus probe diameter
Capacitive probes have a sensing range to sensing element diameter ratio of 1:8 for typical calibrations, so a 500 micrometre range needs a 4 mm sensing element; the spot size is only about 30% larger than the element because the guard ring focuses the electric field [S3]. Eddy-current probes, by contrast, run 1:3 range to coil diameter and produce a spot roughly 3x the coil diameter, since the magnetic field is harder to focus and wraps the end of the probe [S3].
LVDT bodies ignore the spot-size rule entirely because the moving core sits inside the bore: a typical industrial LVDT strokes plus or minus 0.5 mm to plus or minus 250 mm without changing probe diameter, and specialty units reach plus or minus 1 m [S5]. The trade-off is that LVDT bodies are physically long relative to their range, and eddy-current losses in the laminated core limit high-frequency excitation, which is why LVDT datasheets rarely quote bandwidth above a few kHz [S1].
Target material: conductors, insulators, and magnetic steel
Eddy-current sensors require an electrically conductive target, and the signal grows with the target's conductivity and the relative velocity between probe and target, so non-ferrous aluminium and copper, plus ferrous steel, all work with the right calibration [S5]. Capacitive sensors measure any material that can act as one plate of a capacitor, including metals, plastics, fluids, and wafers, but the dielectric constant of the target changes the calibration constant and the effective range [S3].
LVDTs use a separate ferromagnetic core moving inside the coil assembly, so the measured object's material is irrelevant as long as the pushrod is mechanically attached; this is the cleanest fit when the measurand is a moving machine part with no electrical access to the target [S5]. Capacitive sensors can sense insulators that eddy-current sensors cannot, while eddy-current sensors are better suited to dirty conductive targets where capacitive sensors are not a viable option.
Environment: dirt, fluids, EMI, and temperature
Eddy-current displacement probes are widely specified for harsh industrial service with dirt, dust, oil mist, and humidity in the gap, because the magnetic field passes through non-conductive contamination and the coil is potted in epoxy inside a stainless housing [S6]. Capacitive probes degrade in dirty environments, with the manufacturer comparison table rating capacitive as 'not an option' for dirty gaps while eddy-current is rated a functional choice and LVDT is rated a best choice with a sealed bore [S3].
LVDTs sit between the two on EMI: the differential secondary cancels common-mode pickup, but stray capacitance between windings and the effect of nearby ground loops have to be designed out, and the laminated core itself is a source of eddy-current loss that the sensor designer has to budget for [S1].
Resolution, bandwidth, and linearity
Capacitive probes are the resolution leaders on small ranges, with sub-nanometre resolution typical for 50 to 500 micrometre spans and bandwidths up to tens of kHz depending on the front-end electronics [S3]. Eddy-current probes deliver nanometre to sub-micrometre resolution on conductive targets with bandwidths of 1 to 50 kHz, which is why they are the default for shaft vibration monitoring on turbo machinery [S6].
LVDTs are specified for linearity, with 0.1% to 0.25% of full scale common and better than 0.1% available on premium units, but their bandwidth is limited to a few hundred Hz to a few kHz because raising the excitation frequency pushes eddy-current losses in the laminated core up faster than the signal grows [S1][S5]. The Drandic FEM study on LVDT eddy-current losses confirms that core displacement, frequency, and lamination material all change the loss budget, so a 5 kHz LVDT is not a drop-in replacement for a 20 kHz eddy-current vibration probe [S1].
Decision matrix: which technology for which job
The three options line up against four decision criteria as follows. Range to probe ratio: capacitive 1:8, eddy-current 1:3, LVDT independent of probe diameter but limited by body length [S3][S5]. Target material: capacitive any material acting as a capacitor plate, eddy-current conductors only, LVDT independent of target [S3][S5]. Dirty or wet gap: eddy-current best with epoxy-potted coil, LVDT best with sealed bore, capacitive worst unless the gap is clean and dry [S3][S6]. Bandwidth: capacitive and eddy-current both reach 10 kHz and beyond, LVDT typically below 5 kHz [S1][S3].
Use this to shortlist. For wafer thickness, film tension, or any non-conductive film measurement on a clean machine, choose capacitive. For turbo-machinery vibration, bearing position, or shaft runout on conductive rotors in oil and coolant, choose eddy-current, including modern differential-compensated probe designs that extend the measurement range beyond the old single-coil limit [S9]. For hydraulic cylinder position, valve stem lift, or any long stroke on a moving mechanical part, choose LVDT, and verify that the required dynamic response fits inside the LVDT's bandwidth budget. For a discussion of how these non-contact choices compare to other industrial sensing problems, the non-contact temperature measurement spec piece walks through a parallel trade-off.
Failure modes and limits the datasheet hides
Capacitive drift with temperature and humidity is real because the gap dielectric changes, so a 100 micrometre capacitive probe in a 0 to 50 degC shop can show several micrometres of full-scale drift without active compensation [S3]. Eddy-current readings are sensitive to target material changes, and a calibration done on aluminium will be wrong on stainless steel; multi-calibration lookup tables in the controller are mandatory when the target grade can change [S5].
LVDTs fail in two characteristic ways: a bent or seized pushrod from side load, and a shorted primary-to-secondary from moisture ingress, both of which destroy the differential cancellation that gives the LVDT its linearity [S5]. The Drandic FEM work also flags that open-type cores let stray flux drive large tangential eddy-current loops in the laminations, so a 50 Hz LVDT design cannot simply be re-excited at 5 kHz without re-doing the loss budget [S1]. When the measurement target is itself a non-conductive fluid film or a semiconductor wafer, the only one of the three that works is capacitive, and a confocal displacement sensor or laser displacement sensor may be needed if the gap exceeds capacitive range.
Standards, sourcing, and selection checklist
No single IEC or ISO standard mandates one displacement technology over another, but several reference documents are commonly cited in procurement specs: ISO 376 for high-accuracy displacement calibration, ASME B89.1.10 for dial and digital indicator calibration, and IEC 61010-1 for the electrical safety of the associated signal conditioner. ATEX and IECEx certification matters when the probe is installed in zone 1 or zone 0 hazardous areas, and an eddy current tester-derived displacement probe is often easier to certify because the coil and cable can be fully potted in a single stainless body, matching the certification patterns described in the hazardous-area oxygen analyzer spec map [S6].
For sourcing, match the probe diameter to the required range using the 1:8 and 1:3 rules before looking at accuracy; a 6 mm capacitive probe cannot give 1 mm of range, and a 6 mm eddy-current probe will only give about 2 mm of range [S3]. Then check that the controller supports the target material, that the gap environment is compatible, and that the required bandwidth fits the technology, before any vendor-specific accuracy claim is considered.
Final selection still hinges on three trackable signals: published linearity versus full scale in the OEM datasheet, the calibration lookup table count for multi-material eddy-current jobs, and the controller's filter and demodulator bandwidth versus the highest frequency component in the measurand. For a step-by-step spec example on a related inductive device, the load cell class mismatch spec guide shows how a single misread number in the datasheet changes the equipment class on the bill of materials.