KasperAero: A Review of In-Situ Ferrous Debris Monitoring

How Do Components Fail?

Oil debris sensing is not oil quality sensing, and it is related to but different from vibration sensing. It starts with the five ways a lubricated component sheds material:

Physical interface failures. Abrasive, fatigue and adhesive wear where surfaces in pumps, gears, valves or actuators contact under load.
Corrosion and chemical attack. Water or incompatible fluids attack metal surfaces and leave them prone to flaking.
Fatigue and fracture. Repeated stress cycles grow micro-cracks that eventually release fragments.
Impact or foreign object damage. Solid contaminants entering the system chip away at internal components.
Manufacturing or maintenance residue. Debris left from assembly or repair becomes a long-term contamination source.

Over a component's life, debris generation follows the familiar bathtub curve: a bed-in period at commissioning, a long stable stretch with low generation, and a rise as the component begins to fail. The full picture is covered in Debris Generation Explanation.

Bathtub curve of wear rate against running time, showing running-in, stable and increased wear phases

What Information Does Debris Carry?

Three samples of ferrous wear debris increasing in size from fine powder, to flakes, to a broken gear tooth fragment

Wear debris is a direct physical record of surface distress. That is why debris is a more direct indication of failure than vibration: the debris is the failure, not a symptom of it. Each particle carries four pieces of information.

Vibration analysis also struggles where it is needed most. On a high-vibration platform the defect signature is buried in everything else shaking the structure, the data has to be collected and reviewed continuously, and interpreting it is a difficult, highly skilled job. See Vibration Monitoring vs. Debris Monitoring.

Size

Indicates the wear stage and severity. Fine powder is normal running wear. Flakes and chunks point to a component that is coming apart.

Shape

Indicates the mechanism. Rubbing, cutting and fatigue each leave a characteristic particle shape: powder, slivers, granules, flakes or chunks.

Material

Indicates the probable source component. The alloy tells you whether the bearing, the gear or the housing is producing the debris.

Rate

Indicates whether the damage is accelerating. A steady trickle and a sudden rise can carry the same total mass and mean very different things.

Debris feature to wear feature mapping after Han et al. (2023).

In-Situ vs. Laboratory Analysis

There are two branches of debris analysis. In-situ means on the machine or vehicle, measured while it runs. Laboratory analysis means a sample is pulled, shipped and examined. The laboratory is more accurate and far more diagnostic. In-situ sensing competes on latency, not on accuracy.

In-Situ: Real Time

A trend, not a snapshot. Debris is tracked continuously
The rate of accumulation is visible
Alarms at the asset in real time, with no shipping and no turnaround

What you give up: far fewer measured features, with no composition and no morphology.

Laboratory: Periodic

Composition: which alloy, and so which component is failing
Morphology: the wear mechanism read from particle shape
Full fluid chemistry: viscosity, TAN/TBN, water, oxidation and varnish

What you give up: discrete points instead of a curve. A critical asset can degrade to failure inside a one to three month sampling interval.

Downtime is counted in hours. Sampling intervals are counted in months. For the full comparison, including turnaround times and cost, see Offline vs. Inline Oil Monitoring.

Two Types of In-Situ Sensor

Every in-situ debris sensor either captures debris or counts it as it flows past. Each family is evaluated on sensitivity, particle size limits, ferrous selectivity and integration burden. The keyword for capture devices is capture efficiency.

The architecture and the readout are separate choices. Inductive and eddy-current sensing is usually thought of as a flow-through technology, but it is also built in capture form: a magnet pulls ferrous debris onto the sensing face and an inductive coil registers each chip as it arrives. Eaton's QDM and Gill's WearDetect both work this way, counting captured chips from a sensor that installs like a chip detector.

Capture

A magnet collects and retains debris on the sensing face
Particles stay where they landed, in the order they arrived
Debris can be recovered from the face for laboratory analysis, so composition is answered by the lab from your own specimen
Read out by an electrode gap, an inductive / eddy-current coil, or a magnetic field sensor
Ferrous only, and subject to wash-away if the hold is weak

Flow-Through

Much more control over the sensing environment, which makes for better particle-level measurement
The sensor counts each particle as it passes but does not remove it
A downstream filter is always still required, and the filter and its debris are normally discarded at change-out
Traditionally required for non-ferrous detection, and harder to integrate

The Sensing Families, Legacy to Emerging

The talk walked through seven families. Each is covered in depth, with principle diagrams and full pros and cons, in Debris Detection Techniques Compared. The short version, with inductive sensing split into its flow-through and capture forms:

Family	Type	Strength	Limitation
Filter Differential Pressure	Capture, legacy	Already fitted, cheap, rugged	Rises from ordinary loading, binary, no size, count or rate
Electromesh and Grid Switch	Capture, legacy	Ferrous and non-ferrous, no moving parts, cleanable	Mesh aperture sets the limit, binary, no trend
Magnetic Chip Collector	Capture, legacy	Passive, inexpensive, high temperature, self-closing valve	Not a sensor: manual inspection only
Magnetic Chip Detector	Capture, legacy	True electrical trip, high temperature, self-closing valve	Binary, debris must bridge the gap, wetted electrodes and epoxy
Inductive / Eddy Current (Flow-Through)	Flow-through, current	Real-time counts, ferrous and non-ferrous (ASTM D7917)	Expensive, harder to integrate, sensitivity falls with bore size
Inductive / Eddy Current (Capture)	Capture, current	Counts each captured chip, retains debris, remote electronics suit hot locations (Eaton QDM, Gill WearDetect)	Counts arrivals above a mass threshold, misses fines and thin flakes, ferrous only
Optical Particle Counter	Flow-through, current	Counts and sizes metallic and non-metallic particles (ISO 11171)	Bubbles, dark fluid and dirty windows, no composition
Magnetic Field Distortion (NZMS)	Capture, emerging	Continuous accumulated mass, fine particles, low cost, no wetted electrodes	No particle size, placement sets capture, baseline calibration

Who Makes Debris Sensors

A reference list of the companies working in this field, grouped by approach. It is not exhaustive.

Capture and Chip Detection (ferrous only)

Eaton Tedeco (QDM, Zapper, Electromesh), Gill Sensors & Controls (WearDetect), Poseidon Systems (MIDAS, offline), Schaeffler / FAG (Wear Debris Check), T.F. Hudgins (Spinner II grid switch), KasperAero (1020 / 1030 NZMS), Joinwe (JWMA magnetic plug). Eaton QDM and Gill WearDetect are capture-style inductive / eddy-current sensors.

Flow-Through Inductive (ferrous and non-ferrous)

Gastops (MetalSCAN), Parker Kittiwake (MWDS), Poseidon Systems (Trident DM4500), HYDAC (MCS 1000), CM Technologies (Metallic Particle Sensor), Prüftechnik, INZOC (ISL / IFM series).

Optical and Spectroscopic (contamination and elements)

MP Filtri (ICM 2.0), HYDAC (CS 1000), STAUFF (LasPaC), Gastops (FluidSIGHT inline fluorescence, ChipCHECK at-line elemental), AMETEK Spectro (LaserNet Fines, Spectroil), Parker contamination monitors.

Oil Condition (adjacent, not debris sensors)

Tan Delta Systems (OQSx), ARGO-HYTOS (LubCos), STAUFF Oil Quality Sensor, b2 electronic (OilQSens), Voelker Sensors (Oil Insyte), Lubrigard (dielectric loss factor), Poseidon (EIS oil quality).

Large aerospace OEMs, including GE Aerospace, RTX / Pratt & Whitney, Collins Aerospace, Honeywell and Safran Aircraft Engines, also build in-house or integrated solutions. The field is consolidating: Smiths Aerospace went to GE in 2007, Kittiwake to Parker in 2012, and Poseidon acquired Manor Technology (2010) and MACOM (2013).

Sensitivity vs. Size

Chart placing each debris sensing family by detectable particle size, from greater than 300 microns to less than 10 microns, against sensor size

Compare these ranges with care. Debris mass and debris size are different measurements, and there is no standard debris unit. Flow-through devices resolve single particles. Capture devices respond to fine debris only once it accumulates. Inductive sensitivity also falls as bore diameter grows, which is why large lines are often monitored through a sampling loop. Eaton QDM is placed at its published count threshold of 50 µg, equivalent to a 230 µm M50 steel sphere. These are the ranges KasperAero targeted when designing the NZMS.

Counting Events vs. Measuring Accumulation

Two capture sensors can sit in the same port, hold the same debris, and still report different things. The difference is whether the readout responds to a chip arriving or to what is sitting on the face.

Event Counting (Capture Inductive)

A sense coil around the magnet produces a pulse as each chip snaps onto the pole
Each pulse above a mass threshold is counted. Eaton publishes 50 µg for QDM, about a 230 µm sphere
Pulse height depends on arrival speed and shape. A 250 µm flake can weigh about 10 µg and go uncounted
Fine debris that drifts in slowly, or a chip that lands while unpowered, is never counted

Accumulated Mass (Field Distortion, NZMS)

A field sensor reads how the debris on the face distorts the magnet's static field
The reading reflects everything collected, including fines and flakes, however slowly they arrived
The state survives a power interruption. The next reading includes anything collected in the meantime
It does not resolve individual particle sizes. One large chip and many fine particles of equal mass read nearly the same

Neither is strictly better. Event counting targets the large chips of an advanced failure. Accumulated mass trends the whole progression, starting with the fine debris of early wear.

Magnetic field lines around a ring magnet in a null zone magnetic sensor, showing the north and south poles and the symmetric field

Is Magnet Strength the Only Advantage Over a Chip Detector?

Engineers evaluating debris sensors often ask some version of this: how is an NZMS sensor better than a traditional chip detector, and is a stronger magnet the only advantage? It is a fair question, because at a glance both are a magnet in a port that collects ferrous debris.

The answer is no. Magnet strength is one advantage of several, and it matters mostly because of what it enables. The NZMS keeps everything a chip detector does well and changes how the debris is read. Instead of waiting for particles to bridge an electrode gap, a field sensor behind a sealed face measures how the accumulated debris distorts the magnet's field.

Rate Data, Not a Trip

A chip detector reports one bit: tripped or not. The NZMS reports a continuous measure of accumulated debris, so the rate of accumulation is visible. Rate is what tells you whether damage is accelerating, and it is the information a maintenance decision actually needs.

No Fuzz Burner

There is no electrode gap to bridge, so fine conductive fuzz cannot cause a nuisance trip and nothing needs to be burned off. The fine debris that marks the earliest stage of wear is recorded instead of destroyed, and the burner circuit and its power supply are not needed.

More Pull, More Debris, More Robust Signal

NZMS sensors generate significantly more magnetic flux at the collection face than traditional chip detectors. That pulls in more of the passing debris and holds it against wash-away. Capturing more debris is what makes the measurement robust: the signal is built on a larger, more representative sample. See NZMS vs Traditional Chip Detectors.

Same Ports, Same Valves

NZMS sensors remain compatible with self-closing valves, so the sensor can be pulled for inspection without draining the oil. They fit traditional chip detector form factors for installation in gearboxes and scavenge lines, including AS5202 ports. The captured debris can still be removed and sent to a lab.

No Wetted Electrodes

The sensing element sits behind a sealed face. There are no electrodes or potting exposed to oil chemistry, removing the ingress, cracked epoxy and electrode degradation failure modes that affect chip detectors over a long service life.

Simple Electronics, No Expert Required

The principle is simple enough to explain to a child: a magnet collects the debris, and a field sensor reports how much is there. Because the physics is that simple, the output needs almost no signal conditioning. There is no pulse detection, no threshold tuning and no alert algorithm to develop, and reading the signal takes no specialist. More debris means a larger output.

Light Weight, Lower Cost, Lower Integration Burden

In aircraft, avionics are often discussed by the pound, and the cost of electronics tracks closely with their weight anywhere on the platform. Simple physics means little electronics, and little electronics means little weight. A KasperAero 1030AM601 sensor weighs 0.175 lb (about 79 g) with the electronics included, and delivers a low-power analog output directly.

By comparison, Eaton's QDM signal conditioner alone measures about 4 × 4 × 2 inches and weighs 0.95 lb, before the sensor head and harness, and the counting and alert algorithms are left to the engine or aircraft OEM. An inline flow-through system such as Gastops MetalSCAN carries the full oil flow through its sensor body and needs its own signal processing electronics. An NZMS sensor costs a small fraction of either and drops into an existing chip detector port. Lower weight, lower cost and a lower integration burden are what make wide adoption practical, across a fleet rather than on a single critical asset.

Where a Chip Detector Still Wins: Temperature

The honest limitation is temperature. A chip detector is a magnet and two contacts, and it can go almost anywhere the oil goes. The NZMS carries active electronics in the sensor body, and the 1030 series is currently rated to 125°C (257°F).

Getting past roughly 300°F (150°C) is possible, but it is not easy. The sensing element is not the barrier: Hall-effect sensors rated to 170°C are available. Everything around it is. Circuit board substrates and solder joints, and the epoxies that seal and pot the electronics, all have to survive continuous heat and thermal cycling. The magnet has to change too. Neodymium magnets lose strength as they get hot, so a high-temperature sensor needs samarium cobalt magnets, which hold their field at temperature but cost more and start out somewhat weaker. A higher-temperature NZMS is a realistic path, but it will take development work.

That covers the large majority of gearbox, rotorcraft transmission, UAV powertrain and small turbine applications. It does not reach the hottest zones of large turbine engines. For those locations, a chip detector, a chip collector, or a capture inductive sensor with remotely mounted electronics such as Eaton's QDM remains the right tool today.

Five Things to Take Away

Debris is direct evidence. Ferrous debris is physical evidence of surface distress, a more direct detection of failure than vibration monitoring.
Latency, not accuracy. Laboratory analysis remains more accurate and more diagnostic. In-situ sensing competes on how quickly you know.
Every method has limits. Every application is different, and there is no one size fits all sensor.
Debris is not a unit of measure. With no standard debris unit, sensor accuracy and sensitivity are hard to compare across vendors.
Start from the maintenance decision. Before choosing a sensor, decide what you would do if and when it detects debris.

Continue Learning About Condition Monitoring

For a deeper treatment, AGMA and MPMA offer a Gearbox Condition Monitoring education course covering gearbox health and failure mechanisms, condition monitoring techniques, diagnosis and decision-making, and practical gearbox experience. It is taught by Raymond J. Drago, P.E., Chief Engineer of Drive Systems Technology, Inc., who spent 37 years at Boeing Helicopters and is the author of Fundamentals of Gear Design. Contact [email protected] for details.

AGMA, Ray Drago and Drive Systems Technology, Inc. have no affiliation with KasperAero, have received no payment or consideration from KasperAero, and do not endorse KasperAero or any of its products.