Needle Control and Metal Detection in Apparel Manufacturing

The tour always includes it. A laminated broken-needle policy on the wall, a binder of logs, a detector at the end of the line, and a small box of taped-down needle pieces. All of it is real and none of it, on its own, answers the question a buyer actually has: for my order, did the system work?

What the law actually says, and what it does not

This gets overstated constantly, in both directions, so it is worth reading the text. In the EU, the General Product Safety Regulation requires that economic operators place only safe products on the market, that is Article 5 in its entirety, one sentence. A safe product is defined as one that, under normal or reasonably foreseeable conditions of use, presents no risk or only minimum risks consistent with a high level of protection. Manufacturers within scope must carry out an internal risk analysis, draw up technical documentation covering the possible risks and the solutions adopted to eliminate or mitigate them, and, the closest provision to this subject, ensure that procedures are in place for products produced in series to remain in conformity with that general safety requirement.

A buyer requirement can be far more specific and more conservative than the statute, and there is nothing wrong with that. It simply does not become law by being written in a vendor manual, and equally, one retailer's procedure is not the industry standard just because it is the one you have seen.

Prevention and detection are not the same control

This is the distinction the whole subject turns on, and the reason a detector at the end of the line is so often mistaken for the answer.

Each of these does a job the others cannot. Buying a machine adds one of them.

A garment passing a detector does not establish that no needle ever broke, that the incident controls were followed, that every sharp object in the building was accounted for, or that the product is safe in any broader sense. It establishes one thing: that under the equipment, settings and setup used, on that pass, nothing produced a response above the threshold.

Which is genuinely useful, and is not a substitute for knowing that a needle broke, that the work in progress around it was held, and that the pieces were accounted for before anything moved on.

A needle detector is not a general metal detector

The terminology in buyer manuals is loose, "metal detector" is used for equipment that is really a needle detector, and the difference matters because it determines what the machine can and cannot find.

Apparel needle detection is magnetic. Equipment manufacturers describe magnetic lines of force formed between a permanent magnet and a detection coil in the sensor head; when a magnetic material such as a broken needle approaches, the field is deformed, the disturbance is converted into a signal, and an alarm triggers above the set sensitivity. The operative property is therefore response to a magnetic field, not being metal. A general metal detector works on a different basis and can respond to a broader range of metals, which makes it less practical for garments that intentionally contain metallic components.

A rated sensitivity is not a guarantee

This is a frequently misunderstood technical point in the subject, and the correction comes from the equipment manufacturers themselves rather than from anyone's opinion.

Detector performance is specified against a ferrous test piece, in practice a small iron ball of a defined size, supplied as a card or a piece. A ball is a useful reference precisely because it presents the same way from any direction. A broken needle does not: it is elongated, and it can enter the field lengthwise, crosswise or point first.

The test piece checks that the machine responds. It does not describe the smallest real fragment the machine will always find.

Manufacturers state plainly that detection response differs depending on the direction in which a fragment enters the sensor head, and, decisively, that it is not always possible to detect a foreign object even where its mass exceeds that of the test piece. So the inference everybody makes, that a machine rated to a given ferrous size will therefore find any needle fragment at least that big, does not hold.

The buyer conclusion is not that detection is useless. It is that a sensitivity figure describes a reference condition, not a promise about real fragments, and that a control system resting entirely on that figure is resting on something narrower than it appears.

Verification and calibration are different things

Buyer manuals routinely say *calibrate the detector before each shift*. What is usually being described is a functional verification: passing a known test piece through the machine, in the positions and at the heights that matter, to confirm it responds under the current setup. That is a useful, ordinary control.

Calibration or formal service is a different activity, belonging to the equipment's own maintenance regime and its manufacturer's terms. Using one word for both hides the fact that a factory doing a daily test-piece check may not have had the machine serviced in years, or the reverse.

When a needle breaks

Needles break. It happens in competent factories on ordinary production, and treating every breakage as evidence of a bad supplier is both unfair and unhelpful, it encourages exactly the behaviour a buyer least wants, which is quiet handling. The question is never whether a needle broke. It is what the system did next.

A conceptual sequence, not a procedure. Note where detection sits: near the end, and after everything it cannot replace.

Expressed as what a control system should be able to demonstrate rather than as steps to follow: that the incident was recognised rather than absorbed; that potentially affected product and work in progress were controlled rather than allowed to continue; that recovered pieces were accounted for; that anything unaccounted for triggered further investigation rather than an assumption; that release rested on an explicit disposition; and that enough was recorded to trace the incident afterwards.

Accounting for the pieces resolves one question, not all of them. Recovering every visible fragment of one needle establishes what happened to that needle. It does not establish that unrelated contamination is absent, that all surrounding work was unaffected, or that the factory's controls work generally. It is a good answer to a specific question.

And when a fragment cannot be accounted for, there is no universal rule about what happens, how much product is treated as potentially affected, what investigation follows, whether detection is used as additional evidence, and what disposition is permitted are all matters for the applicable buyer and factory procedure. What is not defensible is releasing product on the assumption that the missing piece is probably somewhere harmless.

Sharps are broader than sewing needles

The underlying question is not about needles specifically. It is: which sharp objects are deliberately present in this production route, and what stops any of them becoming an uncontrolled foreign object in the product?

None of that means pins or staples are unlawful, no statute prohibits them in a garment factory. Buyer programmes commonly restrict them, and the reason is the same one that makes needle control worth doing: a loose sharp object near product can become an unaccounted-for foreign object. How a factory implements the control is its own business; whether it has one is a fair question to ask.

A written policy and a working system are different claims

This is where much buyer assessment goes wrong. A polished procedure document is easy to produce and easy to show. What a buyer is actually trying to establish is whether the described system operates, and those are answered by different things.

No layer establishes the others. Programmes get into trouble by holding one and believing they hold the system.

The other half of this is coherence. A register showing regular incidents alongside no containment records, or a detector service certificate with no order-level results, or fragments retained for events that never appear in the log, none of those is a forgery, and each of them means the story does not join up. Evidence that contradicts itself is more informative than evidence that is merely thin.

Rejects, and what a second pass proves

A detector reaction is a signal to investigate, not a finding. It might be a fragment. It might equally be intentional hardware, a material that responds to the field, accumulated contamination, or the equipment and its environment.

Running the garment through again is not an investigation. A second pass produces a second data point and may well be part of a defined procedure, but on its own it does not explain why the first reaction happened, and a practice of re-running until something passes converts a safety control into a formality. What should happen instead is whatever the applicable procedure defines, which is a fair thing for a buyer to ask to see.

Repeated reactions on the same style are worth understanding rather than working around. Persistent reactions can indicate genuine contamination, but they can equally indicate that this product and this detection setup are not compatible, which is a real finding, and one that needs resolving rather than tolerating.

Where this sits next to inspection and audit

Ordinary quality inspection is not designed for this hazard and does not substitute for it. Sampled visual inspection is built to estimate the rate of defects that can be seen; a fragment inside a seam allowance is not one of them, and the sample was never intended to answer that question. The quality control guide covers what AQL does establish, and uses a broken needle left in a garment as its example of a critical defect precisely because it belongs to a different category from a crooked seam.

That difference in kind matters. A shade variation and an embedded sharp fragment are not points on one severity scale, one is a quality matter settled against an approved sample, the other is a potential injury, and GPSR for apparel sets out why a quality defect and a safety defect are handled differently.

A factory audit can look at the procedure, the records, the equipment and how sharps are handled on the floor, and it is a reasonable place to assess this. What an audit finding cannot do is establish that a future order was uncontaminated, and neither an audit nor a policy transfers responsibility for what ships.

Where incidents recur, or where the control is found not to operate as described, that is corrective-action territory rather than a matter for the incident log, the distinction between fixing this order and preventing recurrence belongs to supplier performance management.

Children's products, and everyone else

Buyer programmes commonly apply this control most strictly to children's and baby products, and product-safety assessment explicitly takes account of risk to vulnerable consumers. That is a defensible position and it is a buyer and risk-assessment judgement, not a separate legal category with its own detector rule.

The other half is worth saying because it gets forgotten: a sharp fragment can injure an adult. Nothing about this hazard is specific to children's wear, and a control scoped to one product group is a scoping decision that should be made deliberately rather than assumed. Everything else about children's apparel requirements, small parts, attachment, cords and drawstrings, belongs to children's apparel safety and EN 14682.

Specifying it without copying somebody else's manual

The temptation is to lift a retailer's procedure and issue it as your own. It produces a requirement you cannot justify, cannot assess and did not choose, including numbers whose origin nobody in your organisation can explain. The alternative is to ask what the control has to achieve for your product and to specify that.

And a final caution against the two failure modes at either end. Treating a detector as a guarantee is one; the machine answers a narrow question within its own capability. Treating a policy document as a system is the other; a procedure nobody follows produces exactly as much safety as no procedure at all, and rather more paperwork.