Proximity Probe

What is a Proximity Probe

A proximity probe is a non-contacting eddy-current transducer that measures the distance between its tip and a conductive target, almost always a rotating steel shaft. It belongs to the displacement family of vibration sensors, distinct from the velocity sensors and accelerometers that measure the motion of stationary parts. Because the probe never touches the shaft, it has no wear, no resonance of its own, and a flat response down to zero frequency, which lets it measure both a steady position and a vibration in the same signal. This dual capability is the reason it dominates protection of machines that run in fluid-film (journal or sleeve) bearings.

The distinction from a seismic sensor is fundamental, not cosmetic. An accelerometer mounted on a bearing housing measures how the housing moves; on a heavy machine with a stiff casing and a relatively light, flexible rotor, the shaft can whirl violently inside the bearing clearance while the casing barely trembles. The seismic sensor would report a healthy machine while the journal is hammering the babbitt. A proximity probe looks past the casing and the oil film straight at the shaft, so it sees the motion that actually threatens the bearing. ISO 20816-1, which consolidates the former ISO 7919-1 (shaft vibration) and ISO 10816-1 (housing vibration), recognises that a complete assessment of a fluid-film machine often needs both measurement types together.

The output of the system is two superimposed signals carried on one wire. A negative DC voltage is proportional to the average gap between probe and shaft, giving a direct position reading used for axial thrust float and rotor centring. Riding on top of it is an AC voltage proportional to the dynamic shaft motion, which feeds the vibration channel. A single probe therefore protects against two unrelated failure modes at once: a position channel that catches a failing thrust bearing before metal-to-metal contact, and a vibration channel that catches imbalance, misalignment, rub, and oil whirl.

The technology has a clear lineage in turbomachinery monitoring. Donald Bently commercialised the eddy-current proximity transducer in the late 1950s, and the Bently Nevada name became synonymous with shaft monitoring on steam and gas turbines through the following decades. When the American Petroleum Institute issued Standard 670 for machinery protection systems, it codified the proximity probe convention that the rest of the industry now follows: a nominal sensitivity of 200 mV/mil, a defined linear range, and strict runout limits. Today the same architecture protects the largest rotating assets in refineries, petrochemical plants, power stations, and LNG trains, where a single unplanned trip can cost millions of dollars per day.

Four engineering realities govern proximity-probe practice and recur throughout this guide: the system is a matched three-part chain that cannot be mixed casually; the target material and surface dictate accuracy as much as the electronics; runout, both electrical and mechanical, is the dominant source of false readings; and the gap must be set within a defined window so the shaft has room to move without clipping the signal. Mastering these four removes most of the field problems that plague new installations.

Measurement Applications and Probe Sizes

Proximity probes are deployed in a handful of well-defined configurations, each tied to a probe size and a mounting geometry. The probe tip diameter sets the linear sensing range and the minimum usable target, so the application drives the size choice rather than the other way around. The four standard tip diameters and their roles are summarised below.

Radial shaft vibration is the most common duty. Two probes are mounted on the bearing 90 degrees apart, conventionally labelled X and Y. Each measures shaft motion along its own axis; the monitor combines the pair into a shaft orbit, a direct picture of the path the journal centreline traces every revolution. The orbit is the single most diagnostic plot in rotating machinery: a clean ellipse is healthy, a figure-eight points to a 2X fault such as misalignment, and an inward-spiralling loop signals oil whirl. The 8 mm probe is the default here because its target footprint tolerates minor surface defects better than a 5 mm tip, while still fitting most bearing housings.

Axial (thrust) position uses the DC term of the signal to measure how far the rotor has floated within the thrust bearing clearance. A probe aimed at a thrust collar or the shaft end watches the gap grow or shrink by fractions of a millimetre; an unexpected drift is the earliest warning of thrust-bearing wear before catastrophic axial contact. Because a thrust failure is sudden and destructive, this measurement is usually duplicated with two probes in a voting arrangement. The longer range of the 11 mm probe suits axial duty because thrust float can exceed the 2 mm window of an 8 mm tip.

Phase reference, or Keyphasor, is a single probe aimed at a keyway, notch, or projection on the shaft. It fires one pulse per revolution, giving a highly reliable once-per-turn timing mark used for rotor speed, for the phase angle of vibration, and for field balancing. Related niche duties include eccentricity (slow-roll bow on a turbine before startup), rod drop on reciprocating compressors, and differential expansion between a turbine rotor and casing during thermal transients. The 11 mm and 25 mm probes cover these longer-travel measurements, with the 25 mm reaching about 12.7 mm (500 mil) for the largest steam-turbine supervisory instrumentation channels.

System Architecture and the Eddy-Current Principle

A proximity probe system is always three matched parts: the probe, the extension cable, and the proximitor driver. The reason for the matching lies in the physics. The driver houses a radio-frequency oscillator that excites a flat coil potted into the probe tip. The coil radiates an RF magnetic field into the conductive shaft, inducing circulating eddy currents in the metal surface. Those eddy currents dissipate energy and reflect a counter-field that loads the coil, lowering its effective amplitude. As the shaft moves closer, more energy is absorbed and the coil amplitude drops further; as it moves away, the coil recovers. The driver demodulates this amplitude change into the negative DC-plus-AC output voltage. Because the oscillator is tuned to a specific coil inductance and a specific total cable capacitance, swapping in a probe or cable of different length detunes the system and shifts the calibration.

The probe and the extension cable are selected so that their combined electrical length equals the value the driver is calibrated for. A 5 metre system, the most common, might pair a 0.5 metre probe lead with a 4.5 metre cable; a 9 metre system extends the cable proportionally. This is why datasheets specify a total system length rather than independent component lengths, and why field engineers must never substitute a 5 metre cable into a system tuned for 9 metres. Bently Nevada designates the parts as the 330101 probe, the 330130 extension cable, and the proximitor; Metrix supplies an equivalent set under the MX2033 and MX2034 Digital Proximity System designation.

The eddy-current principle gives the technology its defining strengths. It is genuinely non-contacting, so there is no wear and an essentially unlimited life on the probe itself. It responds from DC up through several kilohertz, which is what allows a single channel to carry both a static position and a dynamic vibration. It is immune to the oil film, dirt, and pressure inside the bearing, because the RF field passes through non-conductive media unaffected. And it is largely insensitive to temperature drift within the rated envelope once the driver compensation is applied.

The same physics imposes the technology's constraints. The eddy-current response depends on the target's electrical conductivity and magnetic permeability, so the system must be calibrated to a known material, and any change of target alloy or plating shifts the reading. The field penetrates only a thin skin of the surface, so localised electromagnetic anomalies in that skin appear as false motion. The probe must see a flat, adequately large target; a curved or undersized surface truncates the field. And the linear range is finite, typically 2 mm for an 8 mm probe, so the installed gap must be set near mid-range to leave headroom in both directions. These constraints are not flaws but the boundary conditions that Chapter 4 turns into hard installation rules.

Digital proximitors such as the Metrix MX2033 add field configurability on top of this classic chain. Where a traditional analog driver is fixed at one material and one cable length, a digital driver can store calibration curves for several target materials and several system lengths, selectable in the field, and can drive long monitor wiring, up to about 1,500 m between driver and monitor in the MX2033 case. The sensing physics is identical; the flexibility lives entirely in the demodulation and output stage.

Standards, Runout, and Target Requirements

API Standard 670, Machinery Protection Systems, is the governing document for proximity-probe installations on critical rotating equipment. Its 5th edition (November 2014) defines the proximity system interface so completely that probes, drivers, and monitors from compliant vendors interchange. The headline requirement is the sensitivity: the system output must be 200 mV/mil and must stay within plus or minus 5 percent of that value across the linear range, which for a standard radial system runs from 10 to 90 mil (0.25 to 2.3 mm) of gap. API 670 also dictates X and Y probe orientation, dual voting for axial position, and the runout limits below.

Runout is the central installation problem and the reason a perfectly good sensor can report a false alarm. It has two independent sources. Mechanical runout is real geometric imperfection of the probe track: out-of-roundness, scratches, dents, or eccentricity that physically changes the gap once per revolution. Electrical runout is more insidious because nothing is moving: local variations in the shaft's conductivity, permeability, or residual magnetism cause the eddy-current response to change around the circumference, so the probe reads apparent displacement where the surface is actually concentric. Welds, chrome plating boundaries, hard spots from grinding, and magnetised patches are the usual culprits.

API 670 caps the combined electrical plus mechanical runout at 25 percent of the allowable peak-to-peak vibration, or 0.25 mil (6.4 micrometres) peak-to-peak, whichever is greater. Meeting that limit is a shaft-preparation task, not an electronics task. The remedies are practical: degauss (demagnetise) the shaft before commissioning, burnish the probe track to a fine, uniform finish free of plating seams and stamped marks, and record a slow-roll runout vector at low speed that the monitor subtracts vectorially from the running signal. The slow-roll compensation is standard practice on turbine startups and is the difference between a believable orbit and a noisy one.

The target requirements follow directly from the eddy-current physics. Proximitors are calibrated from the factory to AISI 4140 steel, the default shaft alloy, because the eddy-current response is set by the target's resistivity and permeability. Any other material, such as 300-series stainless, Inconel, titanium, or chrome plating, has a different response and requires a material-specific recalibration or a published correction factor; ignoring this can shift the apparent sensitivity by tens of percent. The observed surface must also be flat, concentric, and large enough that the eddy-current field is not clipped by an edge, generally a target diameter of at least twice the probe tip diameter. The comparison below maps the major standards an engineer encounters.

Key Specification Parameters

Reading a proximity-transducer datasheet means checking that the matched set meets the application across a small number of parameters. Using the Bently Nevada 3300 XL 8 mm system and the Metrix MX2033 as representative, the parameters that actually drive selection are sensitivity, linear range, frequency response, supply voltage, output range and recommended gap, and operating temperature. Each is explained below.

Sensitivity is fixed by convention at 200 mV/mil (7.87 V/mm) for 5 mm and 8 mm probes and 100 mV/mil (3.94 V/mm) for 11 mm probes. This is not a free design choice but a compatibility constraint: choosing a non-standard value forces the monitor scaling and every alarm setpoint to be reconfigured and breaks plant-wide spare interchange. The API 670 tolerance of plus or minus 5 percent across range is what makes one vendor's 8 mm system a valid spare for another's.

Linear range is the span of gap over which the output stays linear within the sensitivity tolerance. For an 8 mm system it is 2 mm (80 mil), spanning a gap of roughly 0.25 to 2.3 mm (10 to 90 mil). Outside this window the curve flattens and the reading is no longer trustworthy, which is why the installed gap must be centred. The 11 mm probe doubles the range to 4 mm (160 mil) at the cost of halved sensitivity, and the 25 mm reaches about 12.7 mm (500 mil) for long-travel position.

Frequency response defines the dynamic bandwidth of the vibration channel. A typical system runs from 0 Hz (DC, for position) to about 10 kHz, often specified as 0 to 10 kHz at plus 0, minus 3 dB. DC response is what enables the position measurement; the 10 kHz upper limit comfortably covers the running speed and the first several harmonics of any turbomachine. Supply voltage is conventionally a negative rail, nominally minus 24 Vdc, with drivers typically accepting a range around minus 17.5 to minus 26 Vdc; the negative convention is why the output and gap voltages are themselves negative.

Output range and recommended gap together define the installation target. An 8 mm system swings roughly from about minus 1 Vdc near the shaft to about minus 17 Vdc at the far limit, and the recommended installed gap is around minus 9 Vdc, which sits the shaft near the centre of the 2 mm window at about 1.27 mm (50 mil). Centring leaves headroom for the shaft to move toward and away from the probe before the signal clips. Operating temperature must be checked separately for the probe and the driver: the probe tip endures the bearing-area heat, with a typical 8 mm probe rated to about minus 35 to plus 120 degrees C, while the driver electronics live in a cooler junction box. The table below consolidates the headline figures for the common probe sizes.

Selection Decision Factors

To turn the preceding five chapters into a specific order, follow the sequence below. Most proximity-probe field failures trace not to a faulty sensor but to a decision skipped early, a wrong target material, a mismatched cable length, or a gap set off-centre. These steps double as an RFQ template for a machinery protection package.

1. Measurement type: Decide radial vibration (X-Y pair, 90 degrees apart), axial thrust position (dual voting), Keyphasor phase reference, or a long-travel duty such as differential expansion or rod drop. The type fixes the probe count and mounting geometry before anything else.
2. Probe tip diameter: 8 mm for standard radial vibration, 5 mm where the housing is tight, 11 mm for axial and other 160 mil-range duties, 25 mm for steam-turbine supervisory long travel. The diameter sets both the linear range and the minimum target size.
3. Target material and surface: Confirm the shaft alloy. If it is not AISI 4140, obtain the manufacturer's material correction or order a recalibrated driver. Verify the probe track is flat, concentric, free of keyways and plating seams in the field of view, and at least twice the tip diameter.
4. System length: Specify the total electrical length (commonly 5, 7, or 9 m) so probe and extension cable sum to the value the driver expects. Never field-substitute a cable of different length into a tuned system.
5. Standards and certification: API 670 compliance for the 200 mV/mil interface, runout limits, and voting; hazardous-area approval (ATEX, IECEx, or local) for the driver if it sits in a classified zone; SIL rating where the channel is part of a safety function.
6. Runout budget: Plan shaft degaussing, track burnishing, and slow-roll runout compensation into commissioning so combined runout stays under 25 percent of allowable vibration or 0.25 mil (6.4 micrometres), whichever is greater.
7. Environment: Verify the probe temperature rating against the bearing-area heat and the driver rating against the junction-box ambient. Confirm cable armor and ingress protection suit the installation, and that the gap can be physically accessed to set near mid-range.
8. Interchange and spares: Standardise on one vendor family per channel so probe, cable, and driver remain a matched set. Stock spares by total system length, and record each installed gap voltage so a replacement can be set identically.

A final, frequently overlooked dimension is serviceability and interchange: whether the chosen family offers field-configurable digital drivers, documented material correction factors, long-monitor-wiring support, and a spares pipeline that will still exist a decade into the plant's life. Bently Nevada (Baker Hughes) 3300 XL and Metrix MX2033/MX2034 both provide API 670-compliant systems with established service networks, while Meggitt Vibro-Meter and CTC PRO Line supply interchangeable probe sets for retrofit. Because the probe, cable, and driver are a tuned chain, mixing brands within one channel is discouraged unless the vendor explicitly certifies interchange compensation.