How Food Metal Detectors Work (and What They Cannot Detect)

How does a metal detector for the food industry actually work?

This guide explains the physics in depth — why stainless steel is so much harder to find than iron, what "product effect" really is, how aperture size sets the sensitivity you can achieve, and what to ask a supplier before you buy. If you are deciding between a metal detector and an X-ray system, that decision framework lives in our companion guide in the Resources section, Metal Detection vs X-Ray Inspection; this article covers the metal detection technology itself.

A food metal detector is a balanced coil system. Inside the detector head are three coils wound around the aperture — the tunnel the product travels through. The centre coil is the transmitter: it is driven with a high-frequency alternating current and radiates an electromagnetic field. Either side of it sit two identical receiver coils, wired in opposition. Because they are the same distance from the transmitter, the voltage induced in one exactly cancels the voltage induced in the other. With an empty aperture, the net output is zero.

Any metal entering the aperture breaks that symmetry, through two distinct physical mechanisms. Eddy currents: a conductive object — any metal — has small circulating currents induced in it by the alternating field, and those eddy currents generate their own opposing field, which absorbs energy and disturbs the balance between the two receivers. Magnetic disturbance: a ferromagnetic object — iron, most steels — additionally concentrates and distorts the magnetic field lines passing through it, disturbing the balance a second way.

The electronics watch the receiver output for both the size of the disturbance and its phase — the timing of the disturbance relative to the transmitted signal. Different materials disturb the field at characteristically different phase angles, and that phase information is what lets a modern detector separate a genuine metal contaminant from the signal generated by the product itself. When the disturbance crosses the alarm threshold, the detector fires the reject output and the contaminated pack is diverted off the line.

The whole arrangement is deliberately simple and continuous: no radiation source, no imaging, no moving parts in the detection head itself. That simplicity is why metal detection remains the default inspection technology on most food lines worldwide.

Why are iron, non-ferrous metals and stainless steel detected so differently?

Because the two detection mechanisms — eddy currents and magnetic disturbance — depend on two different material properties: electrical conductivity and magnetic permeability. Each metal family scores differently on the two, so each produces a different strength of signal.

The uncomfortable irony sits in the last row of the table below. The metal a hygienic food factory is built from — non-magnetic stainless steel — is precisely the metal a metal detector finds hardest, because it is both a poor conductor and barely magnetic. That is why every credible specification sheet quotes stainless (SUS) sensitivity separately from iron (Fe), and why the stainless figure is always the larger of the two at any given aperture. When you compare two machines, compare the stainless numbers: that is where the difference between them shows.

One more subtlety explains why sensitivity is conventionally quoted as a sphere diameter: a sphere presents the same cross-section to the field from every direction, so the figure does not depend on how the contaminant happens to be lying. Real contaminants are rarely spheres. A thin wire or a sliver of blade can be substantially harder to detect than a sphere of the same diameter, and its signal changes with its orientation as it passes through the coils. Treat published sphere figures as a like-for-like way to compare machines — not as a guarantee that every fragment larger than the rated sphere will be caught.

What is "product effect", and why are meat and fish the hardest products?

Product effect is the signal generated by the food itself. Wet, salty products conduct electricity — not as well as metal, but well enough that eddy currents form in the product just as they do in a contaminant. Fresh meat, fish, poultry, cheese, and anything packed in brine are the classic cases: high moisture plus dissolved salts make the product itself look, to the detector, like a large diffuse piece of weakly conducting material passing through the aperture. Warm bread straight from the oven behaves similarly.

The detector cannot simply ignore this signal, because it can be far larger than the signal from a small stainless fragment buried inside the product. Three tools manage it.

Phase discrimination. Product effect appears at a characteristic phase angle. The detector learns that angle for each product and suppresses signals arriving at it, while still alarming on signals that arrive at the phase angles typical of metal. The cost: metals whose phase response happens to sit close to the product's — non-ferrous fragments and stainless in wet products are the usual victims — become harder to see.

Operating frequency. As a general principle, lower operating frequencies excite less product effect in wet, conductive products but are less sensitive to stainless steel; higher frequencies see stainless better but amplify product effect. Setting up a wet line is therefore a managed trade-off, not a free lunch — and it is the reason detectors intended for mixed production offer a choice of frequency rather than one fixed setting.

Per-product set-up stored in memory. Rather than asking an operator to tune phase and sensitivity by hand at every changeover, a production detector should learn each product once — run representative product through the aperture, store the settings under that product's name, and recall them automatically at changeover. Hand-tuning at every changeover is the single largest source of set-up error on a real line.

Two practical consequences for a meat and fish operation. First, achievable sensitivity on a wet, salty line is always worse than the laboratory sphere figures on any datasheet — those are measured under ideal conditions, and product effect forces the working threshold up. Second, product state matters: as a general principle, deep-frozen product produces far less product effect than the same product chilled, because the free water that conducts is locked up as ice. The same detector on the same product can behave very differently on either side of a freezer — which is why inspection points are validated with the product in the state it will actually be inspected in, and re-learned whenever a recipe, brine level, or temperature regime changes.

How does aperture size change sensitivity?

The smaller the aperture, the smaller the fragment the detector can resolve — so the correct aperture is the smallest one your product and packaging can physically pass through, not the largest one in the catalogue.

The physics behind that rule is worth understanding, because it is what stops a well-meaning engineer from over-specifying. The coils are wound around the aperture, so the field they generate is strongest close to the windings and weakest at the geometric centre of the opening. A contaminant only produces a signal in proportion to how much of that field it disturbs. Enlarge the aperture and two things happen at once: the product's path through the middle sits further away from every coil, and the fragment disturbs a smaller share of a larger total field. The detector's response to a given fragment therefore falls, and the fragment has to be bigger before its signal clears the alarm threshold. That is the whole relationship — no manufacturer escapes it, which is why every credible sensitivity table is quoted per aperture rather than as a single number for the series.

The limiting dimension is usually the height. In a conveyor detector, the product rides on a belt near the bottom of the aperture and the aperture height sets how far the worst-case path sits from the nearest coil. As a general principle, that is why a taller tunnel costs more sensitivity than a wider one, and why suppliers quote their ratings against aperture height. Specify the lowest aperture that clears your tallest pack with sensible clearance for belt tracking and product wobble — and resist the temptation to buy headroom "for future products", because that headroom is paid for in sensitivity on every pack you run today.

Ratings describe the worst case, deliberately. Sensitivity is conventionally rated at the geometric centre of the aperture, where the field is weakest and detection is hardest. A fragment passing close to the coil walls is easier to see than the rated figure suggests. This is also why test pieces are run through the centre of the aperture during verification: testing anywhere else flatters the machine.

Where does metal detection fail — and where does X-ray take over?

A metal detector has two structural blind spots, and no amount of tuning removes them. The first is non-metallic contaminants: glass, stone, bone, ceramic, dense plastic, rubber and wood generate no eddy currents and no magnetic disturbance, so they pass through undetected — always. The second is foil and metallised film packaging: aluminium foil and metallised film are themselves conductive, so the packaging saturates the detector's signal and masks anything inside. A metal detector cannot inspect product in this packaging, full stop.

There is also a soft failure mode: on high-product-effect lines (wet, salty, dense products), the smallest stainless fragments and fine wire can sit below the achievable working threshold even when the machine is perfectly set up.

X-ray inspection reads density rather than conductivity, which is why it takes over exactly where metal detection fails — foil packs, glass, stone, bone, and dense plastics. It has its own blind spot in return: low-density contaminants such as hair, paper, string and most soft plastics do not image reliably. The full decision framework — which technology suits which packaging and hazard, and a side-by-side selection table — is in our companion guide, Metal Detection vs X-Ray Inspection: How Food Factories Catch Foreign Bodies, in the Resources section. We keep that comparison in one place deliberately; if your question is "which one do I need?", start there.

How do you verify a food metal detector is actually working?

With certified test pieces — small cards or wands carrying a metal sphere of known diameter, in each of the three metal families (ferrous, non-ferrous, stainless steel). Spheres are used because their signal does not depend on orientation, making the test repeatable. The test piece is passed through the detector — through the centre of the aperture, where detection is hardest, and ideally with or inside representative product so that product effect is included — and the machine must both alarm and physically divert the test pack. The result is recorded each time.

How often, and at what sphere sizes, is set by your own HACCP plan and any customer code of practice you operate under — a machine's rated sensitivity is the starting point for that conversation, not the end of it. What the physics in this guide should make clear is why the discipline matters: sensitivity is not a fixed property of the machine but a working outcome of aperture, product, frequency and set-up — and only a routine test with a known sphere proves what your line achieves today. Our companion guide on whether metal detection is a HACCP critical control point sets out the decision logic, the critical limit and the records behind that routine.

One point that is easy to miss: the absence of an alarm proves nothing on its own, because a detector that has failed silently also produces no alarm. The test-piece challenge is not paperwork around the control — on an inspection point, the challenge is the monitoring activity.

What should you ask a metal detection supplier?

The datasheet answers almost none of the questions that decide whether the installation works. These do.

"Will you demonstrate it on my product?" In your packaging, at your line speed, in the state the product will actually be inspected in (chilled or frozen — they behave differently). A demonstration on someone else's dry product tells you nothing about a brine-packed one.

"What sensitivity is achievable on my product — in writing, with the test-piece sizes?" Not the laboratory figure. The honest answer is a demonstrated figure, and it will be larger than the datasheet number on any wet, salty line.

"Which aperture are you proposing, and why that one?" Ask them to justify the height against your tallest pack. If they have specified headroom you did not ask for, you are paying for it in sensitivity every day.

"How does it handle product changeovers?" Per-product settings stored and recalled by name, learned once rather than dialled in by an operator each time.

"What happens when something goes wrong?" Reject bin full, air pressure lost, belt stopped mid-pack, pack jammed in the reject chute. Ask what the machine does — a failure mode that produces no signal is a failure mode your records will never show.

"Who can change the sensitivity settings, and would I know if they had?" This is one of the first things an auditor asks. Get the answer before purchase, not during the audit.

"What records does it produce, and how do they get off the machine?" Records that cannot leave the machine are hard to review, and review is what verification means.

"How is it cleaned?" Washdown rating, belt removal, and whether the construction has places water and product can sit.

"What about test pieces?" Which sizes and materials suit the plan, what certificates come with them, and who supplies them.

"What is around the machine?" Vibration, moving metal (a nearby conveyor frame, a forklift route, a pallet truck), and electrical noise all degrade real-world sensitivity. Installation position is part of the specification, not an afterthought.

"What service and spares support exists in-region?" An inspection point out of service stops the line it protects.

Metal detection from A&D Gulf

A&D builds conveyor metal detection for packaged food lines alongside its in-motion checkweighers and the AD-4991 ProteX X-ray inspection range, so one supplier can cover the whole end-of-line inspection cell. A&D Gulf FZE supplies, integrates and services that food inspection range across the UAE and GCC from Jebel Ali, Dubai.

Because aperture, achievable sensitivity and conveyor configuration are all decided by the product, the packaging and the line — as the sections above explain — we do not publish a sensitivity table. A&D's metal-detection line is available to order, with availability and specifications confirmed per model with your quotation. Send us your product dimensions, pack height, line speed, packaging material and product state through the contact page and we will come back with the right configuration.

For UAE food factories the record-keeping side matters as much as raw sensitivity. Dubai Municipality mandates HACCP-based food safety management, and where your hazard analysis identifies metal as a significant hazard, a metal detection point with logged results is the standard control. Specific documentation and inspection requirements for your licence category should be confirmed with the authority.