Debris Detection Techniques Compared

A Review of In-Situ Ferrous Debris Sensing

This article assumes the decision to look for metal in the oil has already been made, and asks which detection method should be used. Seven sensor families are covered, along with the fluid properties, installation constraints and electrical environment that decide which of them will work on a given machine.

Two topics are treated as background rather than repeated here. How debris is generated in the first place, including wear mechanisms and the wear-rate curve, is covered in Debris Generation Explanation. The trade between laboratory sampling and continuous in-situ measurement is covered in Offline vs. Inline Oil Monitoring. The short version of the second point matters for everything below: a laboratory analysis is more accurate and more diagnostic than any in-situ sensor. In-situ sensing competes on latency, not on accuracy, and the two belong in the same program.

What a Wear Particle Records

A wear particle is a direct physical record of surface distress. It carries four kinds of information, and these four are the axes every sensor in this article is measured against.

Size:

Indicates wear stage and severity.

Shape:

Indicates mechanism. Rubbing, cutting and fatigue particles look visibly different under magnification.

Material:

Indicates the probable source component.

Rate:

Indicates whether damage is accelerating. This is usually what actually triggers a maintenance decision.

A sensor returns only the features its physics can resolve. A switch does not produce shape. A mass measurement does not produce a count. That is not a quality difference between products. It is a physics difference between methods, and it is why no single sensor type wins across the board.

KasperAero 1030 debris sensor with approximately 10 mg of fine ferrous debris collected on its sensing face — Fine ferrous debris collected on the sensing face of a KasperAero 1030 sensor. The accumulation shown is approximately 10 mg.

There is also no standard unit for debris. Some methods report a particle count, some report size in microns, some in calibration-equivalent microns, and some report accumulated mass in milligrams. Those are not different levels of precision on one scale. They are different quantities, and the inconsistency is what makes sensitivity claims so hard to compare across vendors.

Detection, measurement and diagnosis

Those four features map onto three separate jobs. Keeping them apart is what makes the rest of this article tractable, because the families below are not competing to do the same thing.

Detection:

Did debris appear?

Measurement:

How much, how often, how large, and of what type?

Diagnosis:

Which component is producing it, and by what failure mechanism?

Every family below answers the first question. They differ in how much of the second they deliver. Only laboratory analysis answers the third. Read each family against those three jobs rather than against a single sensitivity ranking.

Two Architectures: Capture and Flow-Through

Every in-situ debris sensor falls into one of two architectures. This choice determines almost everything else about how the sensor behaves, including how it responds to fluid viscosity and where it has to be installed.

Capture

Collects and retains

A magnet collects and holds debris on a sensing face. Particles stay where they landed, in the order they arrived, which gives far more control over the sensing environment. A capture sensor counts what accumulates rather than what passes. Debris can also be physically recovered from the face afterwards for laboratory analysis, so the specimen is preserved. The tradeoff is a wash-away risk and a dependence on debris actually reaching the magnet.

Flow-through

Registers in transit

A particle is registered as it passes, without being removed from the fluid. A downstream filter is still required either way, since neither architecture replaces filtration. Non-ferrous debris reaches only two of the methods in this article: flow-through inductive sensing, which reads eddy currents in conductive particles, and conductive screens, which catch anything that bridges the mesh. Every magnetic method here is blind to it.

Where the sensor sits is a separate question

Architecture describes how a sensor detects. Mounting position describes where it sits in the circuit. The two are independent, and industry practice recognises three positions.

In-line:

The sensor sits in the main flow path and is exposed to full system flow.

On-line:

The sensor sits in a bypass or kidney loop, seeing a defined fraction of total flow at a controlled rate.

Off-line:

A sample is drawn to a bottle or a portable instrument. This is the laboratory route covered in the companion article.

Both architectures can be mounted in-line or on-line. Only a drawn sample is off-line. The distinction changes how a reading is interpreted: a sensor in a bypass loop reports a rate that has to be scaled by the fraction of flow that loop carries, and a capture sensor mounted in the main path is exposed to full flow velocity and the wash-away risk that comes with it. Note also that this is where the word "on-line" properly belongs. It describes a bypass mounting, not a cloud connection, which is why this article uses inline rather than online for a sensor installed in the circuit.

Capacitive sensing: a flow-through method that is not commercial yet

Capacitive debris sensing is a genuine detection method, not simply an oil condition measurement. Debris changes the dielectric constant of the fluid, and a capacitor reads that change. Coaxial and microfluidic designs in the literature resolve individual particles rather than bulk fluid state, with published work reporting detection of iron down to roughly 54 µm and copper to roughly 90 µm, while also separating water droplets and air bubbles from metal [9][10].

That last capability addresses the gap the rest of this article keeps running into. A capacitive channel responds to dielectric contrast rather than to magnetic permeability, so non-ferromagnetic metal is not the blind spot it is for every magnetic method here and a weak point for inductive sensing. Some research devices combine both, using an inductive mode to classify metal and a capacitive mode to reject bubbles and water.

It is not listed as one of the seven families below for a commercial reason rather than a technical one. The work is published, patented and demonstrated on the bench [11], but it has not reached a broadly available product, and it is not treated as a deployable option in the comparison table for that reason. It is also why no debris standard covers capacitive sensing. Standards follow a product base, and there is not one yet.

The Sensor Families

Filter Differential Pressure

As a filter element loads, the pressure drop across it rises. A switch or gauge trips at a set threshold. It is cheap, rugged and universally understood, and on older equipment it is often the only indication of debris present at all. Eaton's helicopter debris patent wires a differential pressure gauge directly to the cockpit alongside the chip detector.

Examples in service: Differential-pressure indicators and switches are a standard option on filter housings from Parker Hannifin, HYDAC, Pall, Eaton (Internormen), Donaldson and Schroeder. Visual pop-up, electrical switch and analogue transmitter variants are all common.

✓ Strengths

✓Already installed on most filtered systems, at no additional cost.
✓Rugged and passive. No electronics in the fluid path.
✓Universally understood by maintenance crews.

✗ Limitations

✗Debris is inferred, not measured. The element loads from ordinary contamination and additive dropout as well as from wear.
✗Warning arrives late. By the time it trips, the filter is near bypass and unfiltered oil is about to circulate.
✗Fine debris can be invisible. A slow leak of fine particles may never move the reading.

Electromesh and Grid Switches

Schematic of an electromesh grid switch in the OK state, showing the filter grid element, debris particles in the flow, two leads and an open indicator circuit — Conductive particles bridge the grid to close a circuit. Shown in the OK state.

Photograph of an electromesh grid switch debris detector — A grid-switch style debris detector

A woven screen, or a pair of interposed electrical grids, sits in the flow. Conductive particles lodge in the mesh, bridge a circuit and raise an alarm. This is one of the few methods here that catches conductive non-ferrous debris such as aluminum, magnesium and phosphor bronze, which every magnetic method in this article is blind to.

Examples in service: Eaton Tedeco electromesh detectors. T.F. Hudgins Spinner II grid switch. Screen-type detectors are also offered by several filtration houses as a filter-bowl option.

✓ Strengths

✓Sees conductive non-ferrous debris. Aluminum, magnesium and phosphor bronze all register.
✓No moving parts in grid-switch designs.
✓Inspectable in place without shutting the machine down.

✗ Limitations

✗Coarse by design. Mesh aperture sets the smallest detectable particle, and fine wear passes straight through.
✗Binary output only. No count, no size distribution, no trend.
✗Geometry-limited. Sensitivity cannot be improved without changing the aperture.

Magnetic Chip Collectors

Photograph of magnetic chip collector plugs — Typical magnetic chip collector plugs

A permanent magnet captures ferrous debris from the fluid stream. The collector is pulled at intervals for visual inspection. It is compact, passive, mechanically robust and inexpensive, and it retains debris for metallurgical analysis, which matters more than it is usually given credit for.

Examples in service: Eaton Tedeco chip collectors, which use a self-closing valve so the probe can be withdrawn without draining the oil. Meggitt chip collectors for gas turbine engines. Debris recovered from a collector does not have to go to a laboratory to be identified: an at-line analyzer such as Gastops ChipCHECK reports particle size, shape and alloy classification by laser spectroscopy in minutes.

✓ Strengths

✓Retains the specimen so captured debris can be sent for laboratory analysis.
✓Inexpensive, compact and passive. Nothing to power or calibrate.
✓Mechanically robust in harsh environments.

✗ Limitations

✗Not a sensor. There is no automatic indication of any kind.
✗No trend data. Nothing is recorded between inspections.
✗Requires a person. It reports nothing until someone physically pulls and inspects it.

Magnetic Chip Detectors

Photograph of magnetic chip detector products — Typical magnetic chip detectors

Add electrodes to a chip collector and it becomes a switch. Two opposed magnetic poles create a narrow gap. Ferrous debris accumulates until it bridges the gap, closes a circuit and triggers a discrete electrical warning. The condition is now reported electrically, with no inspection visit required.

Two structural weaknesses are worth noting. The particle has to physically bridge the gap, which is chance rather than certainty, so the amount of debris needed to trip varies from run to run. The electrodes are also wetted, and therefore exposed to oil chemistry, potting defects and ingress paths. Some designs burn off fine conductive debris with a current pulse to prevent nuisance trips. That works, but it destroys the earliest available evidence in the process.

Examples in service: Eaton Tedeco chip detectors, including the axial radial gap DS200-3B and the Smart Zapper fuzz-burner variants. Meggitt Auto-Fault digital pulsed chip detector systems, used on rotary-wing gearboxes.

✓ Strengths

✓Automatic indication. No inspection visit required to learn something happened.
✓Compact and well proven, with decades of aerospace service history.
✓Drops into existing ports on most gearboxes and engines.

✗ Limitations

✗One bit of output. No count, no size, no trend.
✗Trip point varies run to run, because bridging the gap is a matter of chance.
✗Wetted electrodes are exposed to oil chemistry, potting failure and ingress.
✗Fuzz burners destroy evidence. The fine debris that indicates earliest wear is removed before it is recorded.

Inductive and Eddy-Current Sensing

Diagram of an inductive eddy-current debris sensor showing the coil field around a flow passage — A coil field is perturbed as metal debris passes

Photograph of an inline inductive debris sensor — An inline inductive flow-through sensor

An alternating coil field interacts with passing metal. Ferrous debris changes magnetic permeability, and conductive non-ferrous debris generates eddy currents. The resulting pulse indicates material class and approximate size. This is the richest data of any in-situ method: individual particles, with timing, size, and ferrous versus non-ferrous discrimination.

The catch is scale. Sensitivity falls as bore diameter grows, because signal-to-noise scales with cross-section. A published detection threshold is incomplete unless it names the aperture and flow rate it was measured at. ASTM D7917-14 (reapproved 2018) governs how these devices should be installed, operated and reported, but it is a practice rather than a device qualification, and it applies only to inductive flow-through devices.

Examples in service: Gastops MetalSCAN MS3500 and MS4000, with the 4110 rated for high-temperature service (9.5 to 31.8 mm bore). HYDAC MCS 1000 metallic contamination sensor, and MCS13 at approximately 6.3 mm bore. Parker Kittiwake MWDS and Poseidon Systems Trident DM4500 and DM4600 (10 to 12 mm bore). Gill Sensors WearDetect.

✓ Strengths

✓Richest in-situ data set. Individual particle events with timing and approximate size.
✓Discriminates ferrous from non-ferrous material.
✓Covered by a published standard, ASTM D7917, for installation and reporting.
✓Continuous output suitable for trending.

✗ Limitations

✗Sensitivity falls as bore size rises. The same sensor can be an order of magnitude worse on a larger line.
✗Physically large and costly, because the device must carry the full flow.
✗Specifications are hard to compare unless aperture and flow rate are stated.
✗Does not retain the particle for later laboratory work.

Optical Particle Counting

Photograph of an inline optical particle counter — An inline optical particle counter

A light source illuminates particles passing through a flow cell. Scattered or blocked light lets the sensor count them and estimate their size. Its real strength is that it sees everything, including the non-metallic contamination that every magnetic method misses entirely, and it is backed by a mature calibration standard in ISO 11171.

A particle count is not a wear diagnosis. It cannot establish that the particles are steel from a gear rather than dirt or fiber. Sizes reported under ISO 11171 are given as µm(c), a certified calibration-equivalent diameter traceable to NIST reference material, not a physical dimension that would be measured under a microscope.

This is the most commercially crowded family in the article, and most suppliers report against the same ISO 4406 cleanliness code, which makes cross-shopping easier here than anywhere else. Typical inline units report channels at 4, 6 and 14 µm(c), with wider-channel variants extending to 21, 25, 36, 50 and 70 µm(c). Many combine the particle count with water saturation and temperature in the same housing. Light-blockage designs dominate inline; laser designs are more common in portable and bench instruments where sample conditioning is possible. One instrument class is worth separating out: shape classifiers such as the Spectro Scientific LaserNet Fines image each particle and sort it by shape, which recovers some of the wear-mode information ferrography provides. That is a bench or at-line instrument rather than an inline sensor, but it is the one optical route to morphology.

Examples in service: Inline: MP Filtri ICM 2.0 and ICM 4.0. HYDAC ContaminationSensor CS 1000 series. ARGO-HYTOS OPCom. Parker iCountPD and iCountOS. Eaton PFS 02 (Internormen). Pall PCM. Portable and bench: MP Filtri LPA2 and LPA3. STAUFF LasPaC-3. ARGO-HYTOS OPCount. Shape classification: Spectro Scientific LaserNet Fines Q200, which counts particles and classifies them by shape.

✓ Strengths

✓Sees all particulate, including non-metallic contamination.
✓True count and size distribution, not a single lumped value.
✓Mature calibration standard. ISO 11171 is traceable to NIST reference material.

✗ Limitations

✗Not wear-specific. A count cannot distinguish steel from dirt or fiber.
✗Bubbles read as particles, which inflates counts.
✗Dark or opaque fluid degrades performance.
✗Windows foul over time and need maintenance.

Magnetic Field Distortion

KasperAero NZMS debris sensor, connector end view — The same sensor, connector end view

This family starts the same way as a magnetic chip detector. A permanent magnet captures ferrous debris. The readout is what changes. Instead of waiting for a conductive bridge, a field sensor measures how the accumulated debris distorts the local magnetic field relative to a clean baseline. That converts a binary switch into a continuous trend, and because nothing is wetted, both the bridging problem and the fuzz-burner problem disappear. Fine debris is recorded rather than destroyed.

The tradeoff is that this family measures accumulated mass, not particle count or size. One large chip and a thousand fine particles of equal total mass read nearly the same. Placement matters more here than for flow-through sensors, because a sensor measures nothing if debris never reaches the magnet.

This is not a new idea. Eaton's QDM has used a magnetic flux sensor proportional to captured mass in aerospace service for years. KasperAero's Null Zone Magnetic Sensor is one entrant in this family, and it is held to the same standard as every other product discussed here.

Examples in service: Eaton QDM quantitative debris monitor, in aerospace service. KasperAero NZMS (patent pending).

✓ Strengths

✓Continuous trend rather than a binary trip.
✓Nothing wetted. No exposed electrodes, so no oil chemistry or ingress failure path.
✓Fine debris is recorded, not destroyed. No fuzz burner is required.
✓Retains the specimen on the magnet for later laboratory analysis.
✓Compact and low cost relative to full-flow instruments.

✗ Limitations

✗Ferrous only. No response to bronze, aluminum or babbitt.
✗Accumulated mass only. No particle count and no size distribution.
✗Placement dependent. Debris has to reach the magnet to be seen.
✗Baseline shifts with nearby ferrous structure and requires calibration.

Selecting for the Application

Sensor family is only half the decision. The fluid, the plumbing and the electrical environment decide whether a given family can work at all on a given machine.

Fluid viscosity

Viscosity affects the two architectures in opposite directions, and it is the single most under-discussed selection factor.

A magnetic capture sensor works by pulling a particle across streamlines onto the magnet. That migration is resisted by Stokes drag, which is proportional to viscosity. Double the viscosity and the migration velocity roughly halves for the same magnetic force. Drag scales with particle radius while magnetic force scales with particle volume, so fine particles suffer far more than coarse ones. In a thick oil, a capture sensor still works, but capture efficiency falls and the response becomes biased toward larger debris.

A flow-through sensor does not depend on particle migration at all. The particle only has to pass through the sensing aperture, which it does regardless of viscosity. What rises instead is pressure drop across the bore, which becomes a plumbing and pump sizing problem rather than a detection problem.

Wind turbine gearboxes are the clearest example. Most large turbines specify at least ISO VG 320, meaning a nominal kinematic viscosity of 320 mm²/s at 40 °C [12]. That is roughly ten times a typical VG 32 hydraulic fluid, and at cold start it is far thicker still. This is a genuine reason inductive flow-through sensors dominate wind gearbox monitoring.

One clarification worth making: wind gearbox oil is a very heavy oil, not a grease. Turbine main, pitch and yaw bearings are commonly grease-lubricated, and no in-fluid debris sensor of any family works in grease. The two cases get conflated, and they have different answers.

Temperature

Temperature acts on the sensor and on the fluid at the same time, and the two effects are easy to confuse for one another.

Magnet stability:

Permanent magnets lose remanence as temperature rises, and recover on cooling. In a capture sensor that shifts capture force and, in field-reading designs, shifts the baseline. Sintered NdFeB is the most temperature-sensitive common choice; SmCo holds up better at high temperature at the cost of field strength.

Electronics rating:

The sensing element and its front-end set the real upper limit. Hall effect, TMR, AMR and GMR devices all have published operating ranges, and an amplifier that drifts with temperature drifts the reading with it.

Fluid viscosity:

Viscosity falls sharply with temperature. A capture sensor that performs adequately at operating temperature can capture poorly during cold start, which is also when a machine is most likely to shed debris.

Thermal cycling:

Repeated cycling is what cracks potting compounds and opens ingress paths in wetted-electrode designs. This is a service-life issue rather than an accuracy issue.

Installation and placement

Placement matters more for capture sensors than for flow-through sensors, because a capture sensor measures nothing if debris never reaches the magnet.

Position in circuit:

Upstream of the filter, debris is still in the fluid and available to detect. Downstream, most of it has already been removed. A sensor placed after filtration will under-report, sometimes dramatically.

Bypass and kidney loops:

An offline filtration loop carries only a fraction of total flow. A sensor on that loop sees only that fraction, which has to be accounted for when interpreting a rate.

Low point versus flow path:

Splash and sump systems favor a low-point mount where debris settles. Circulating systems favor a return or scavenge line where debris is actively carried past.

Orientation and drain-down:

A capture face that sits above the static oil level can dry out between runs. A face in a dead zone accumulates sludge rather than wear debris.

Serviceability:

A capture sensor eventually needs cleaning. If the port cannot be reached without draining the system, the maintenance interval is set by access rather than by the sensor.

Electromagnetic environment

Magnetic sensing families read small field changes, and a large motor is a large field source. Proximity is a real constraint.

Motor and drive proximity:

Stator leakage fields and variable frequency drive switching noise both couple into magnetic sensing elements. A VFD is the more troublesome of the two, because its switching harmonics sit in the same band as the sensor's own signal chain.

Nearby ferrous structure:

Steel housings, brackets and fasteners distort the static field a field-reading sensor uses as its reference. This shifts the zero baseline and is why these devices need calibration in their installed position rather than on a bench.

Shielding:

A mu-metal or soft-iron shield redirects external flux around the sensing element. It is effective, but it also perturbs the sensor's own working field, so shield geometry has to be designed with the magnet rather than added afterwards.

Grounding and cable runs:

A single-point ground and a shielded, twisted cable removes most conducted noise. Running sensor cable in the same tray as motor leads reintroduces it.

Signal integrity and impedance

Debris sensor outputs are frequently run over long cables in electrically hostile plant, and the interface choice matters as much as the sensing element.

Source and load matching:

A high-impedance voltage output driving a long cable forms a low-pass filter with the cable capacitance, which rounds off fast particle events. Inductive sensors resolve individual particles in microseconds to milliseconds, so this directly costs detection events.

Coil tuning:

In inductive sensing the coil is part of a resonant circuit. Its impedance has to match the drive and detection electronics, which is why a coil designed for one bore size cannot simply be fitted to another.

Output type:

A 4 to 20 mA current loop is largely immune to voltage drop and induced noise over long runs. A 0 to 10 V output is simpler but degrades with cable length. A digital bus such as CAN or IO-Link avoids both problems and carries diagnostics, at the cost of integration effort.

Bandwidth:

A slow analogue front end will average a genuine particle event into the noise floor. Accumulated-mass sensors are unaffected by this, since the quantity being measured changes slowly by nature.

The Patent Landscape

Debris detection is a heavily patented field, and the density of that protection differs sharply between families. This matters commercially, because it shapes which architectures are practical for a new entrant and which are effectively closed.

Inductive:

The most densely patented family in this space, with sustained filing from RTX, Smiths, Caterpillar and Pruftechnik among others. Coil geometry, drive schemes, particle discrimination algorithms and signal processing are all covered. This is a difficult family to enter without licensing.

Chip detectors:

Long-established and largely off-patent in its basic form, which is why the geometry is effectively standardized across suppliers. Later refinements are still protected. Pratt and Whitney's US10866201B2 measures resistance across the electrode gap rather than treating it as a simple open or closed circuit, and Eaton's fuzz-burner designs are similarly covered.

Optical:

Protection sits mostly in the optics and the counting algorithm rather than in the basic principle, and the field is constrained more by ISO 11171 calibration requirements than by patents.

Capacitive:

Filing is academic and institutional rather than commercial, which is consistent with a method that has not yet productised. US11061010 covers a coaxial capacitive sensor for on-line monitoring of engine lubricating oil abrasive particles.

Magnetic field distortion:

The newest and least crowded of the families. Eaton's QDM established prior art for reading captured mass through magnetic flux. KasperAero's Null Zone Magnetic Sensor is patent pending.

A related point for anyone reading vendor literature: a patent describes what a company protected, not necessarily what it ships. A wider survey of the patent landscape for ferrous debris detection is collected separately in State of the Art (Other Patents).

Sensitivity Across the Families

The chart below places each family and several named products on a sensitivity and size spectrum.

Chart comparing sensitivity against particle size across debris sensing technology families and named products

Optical counters resolve the smallest particles but are not ferrous-specific. Devices that measure accumulated mass rather than individual particles do not sit on a directly comparable micron axis at all.

Debris mass and debris size are different measurements. Flow-through devices resolve single particles. Capture devices respond to fine debris only once it accumulates.

Why Published Specifications Do Not Compare Cleanly

Anyone comparing datasheets across these families will run into the same four problems.

Threshold without geometry:

An inductive sensor's detection threshold means little without the bore diameter and flow rate it was measured at. The same sensor can be an order of magnitude worse on a larger line.

Mass versus dimension:

Some vendors quote the smallest detectable particle in microns, others in milligrams. Converting between the two requires assumptions about particle shape and density that rarely hold for real debris.

Detectable versus actionable:

The smallest particle a sensor can register is not the threshold anyone would actually alarm on. Noise, hysteresis and nuisance-trip margin sit in between, often by an order of magnitude.

Bench versus installed:

Capture efficiency, plumbing, temperature, fluid viscosity and nearby ferrous structure all move real-world performance away from the datasheet figure.

Standards cover only part of this picture. ASTM D7917 governs inductive flow-through devices specifically. ISO 11171 makes optical particle counts traceable to NIST reference material. Neither standard normalizes a capture device against a flow-through one, and there is currently no debris standard at all for capacitive sensing. That gap is real, and it is arguably an opportunity for an industry body to take on.

What Each Family Actually Measures

Method	Count	Size	Ferrous only	Continuous	Composition	Primary output
Filter differential pressure	No	No	No	Yes	No	Threshold trip
Electromesh / grid switch	No	No	No	No	No	Binary trip
Magnetic chip collector	No	No	Yes	No	No	Visual inspection
Magnetic chip detector	No	No	Yes	No	No	Binary trip
Inductive / eddy current	Yes	Approx.	Discriminates	Yes	No	Particle events
Optical counter	Yes	Yes	No	Yes	No	Size distribution
Magnetic field distortion	No	No	Yes	Yes	No	Accumulated mass
Laboratory analysis	Yes	Yes	Both	No	Yes	Full diagnosis

Five Takeaways

First decide what you need to know: that debris exists, how much is being generated, or what is failing. A chip detector can tell you that debris has accumulated. An inductive sensor can count individual particles and trend their production. Neither can identify the failing component or wear mechanism by itself. That requires analysis of the debris.

Choose capture or flow-through before comparing sensor specifications. Capture sensors retain debris and can detect accumulated fine material, but they depend on debris reaching the sensing surface and are affected by viscosity and placement. Flow-through sensors see particles as they pass, but their sensitivity depends strongly on bore size and flow rate.

Put the sensor where the debris will actually reach it. A sensor upstream of a filter can see wear debris that a downstream sensor cannot. A capture sensor in a dead zone may perform poorly regardless of its laboratory sensitivity. In a bypass loop, the measured rate also has to be read against the fraction of total flow being sampled.

Do not compare "10 µm" with "1 mg" as though they are the same specification. Particle size, particle count and accumulated mass are different measurements. Even within one technology, detection thresholds can change substantially with bore size, flow rate, viscosity and installation. A published number is only meaningful when the conditions behind it are known.

The best sensor is the one that changes what you do next. If an alarm only tells you that debris exists, and does not lead to an inspection, a sampling decision, a maintenance action or a shutdown, its detection capability has limited practical value. The maintenance decision should define the measurement, not the other way around.

There is no single best debris sensor. There is only the evidence a given machine needs, and the architecture that can deliver it reliably in that fluid, at that temperature, in that installation.

References and Standards

ASTM D7917-14 (Reapproved 2018). Standard Practice for Inductive Wear Debris Sensors in Gearbox and Drivetrain Applications. ASTM International. astm.org/d7917-14r18.html
ISO 11171:2022. Hydraulic fluid power. Calibration of automatic particle counters for liquids. International Organization for Standardization. iso.org/standard/79769.html
ISO 4406:2021. Hydraulic fluid power. Fluids. Method for coding the level of contamination by solid particles. iso.org/standard/72618.html
US Patent 10866201B2. Chip detector with resistance measurement across the electrode gap. Pratt and Whitney. patents.google.com/patent/US10866201B2
MP Filtri. ICM 4.0 In-Line Contamination Monitor. LED optical technology, particle sizing to ISO 4406. mpfiltri.com
HYDAC. ContaminationSensor CS 1000 Series. hydac.com
ARGO-HYTOS. OPCom Particle Monitor. argo-hytos.com
STAUFF. LasPaC-3 Particle Counters. stauff.com
Zhu, X., et al. Characteristics Study of In-Situ Capacitive Sensor for Monitoring Lubrication Oil Debris. Sensors, MDPI, 2017. doi.org/10.3390/s17122851
A New In Situ Coaxial Capacitive Sensor Network for Debris Monitoring of Lubricating Oil. Sensors, MDPI, 2022. doi.org/10.3390/s22051777
US Patent 11061010. Coaxial capacitive sensor and a method for on-line monitoring and diagnosing engine lubricating oil abrasive particles. patents.google.com/patent/US11061010
ExxonMobil. Mobil SHC Gear 320 WT wind turbine gear lubricant. Source for the ISO VG 320 wind gearbox viscosity figure. mobil.com
KasperAero. Debris Generation Explanation. Wear mechanisms, particle morphology and the wear-rate curve. debris-generation-explanation.html
KasperAero. Offline vs. Inline Oil Monitoring. Laboratory analysis against continuous in-situ sensing. offline-vs-online-oil-monitoring.html
KasperAero. NZMS vs. Traditional Chip Detectors. Prong and pole-style detectors, fuzz burners and nuisance trips. Traditional_Debris_Detectors.html
KasperAero. State of the Art (Other Patents). Patent landscape for ferrous debris detection. StateOfTheArt.html
Kasper, B. A Review of In-Situ Ferrous Debris Monitoring Solutions. Presented at the 2026 MPMA Fall Technical Meeting.

Disclosure

Company and product names are included for reference, based on public product literature. Inclusion is not an endorsement, and the list is not exhaustive. Logos and trademarks are the property of their respective owners. KasperAero manufactures a magnetic field distortion sensor, the NZMS, and has held it to the same standard applied to every other product discussed here.

KasperAero Debris Sensors »
Offline vs. Inline Monitoring » Back to Knowledge Hub »