Metal Detector Selection: X-ray Inspection and Thai Plant RFP

“Our existing metal detector has too many false rejects. Complaints about bones and hard foreign bodies are not going down. In audits, our Japanese customers ask to see the verification records for our inspection equipment. Should we replace the metal detector, switch to X-ray, or install both?” We often hear this from quality assurance and production engineering staff at Japanese-owned food factories in Thailand. Here is the short answer: metal detector selection is not decided by comparing sensitivity figures such as “what size of test sphere it can see.” It is decided by 3 things: (1) the combination of which foreign bodies you want to stop and the properties of your product; (2) whether detected product can be reliably rejected and isolated; and (3) whether verification runs every day as part of operations. X-ray foreign body detection equipment is not all-powerful either. Metal detectors are better at thin metal fragments, and foreign bodies with a density close to that of the product are difficult for both methods. That is exactly why you work backward from “what you want to stop” to choose the equipment.

All amounts, pack counts, headcounts, false reject rates, complaint counts, and payback years in this article are original estimates and assumptions (placeholder values) created for this article, based on the model factory described later. They are neither industry averages nor survey results. Please read them as a “calculation template” and replace them with your own measured figures.

Why Thai Food Factories Are Now Reviewing Foreign Body Detection Equipment

Thai food exports are shifting from a phase of earning through volume to a phase of earning through added value.

According to comments by a Thai processed food industry representative reported by the business newspaper Thansettakij (September 17, 2026), Thailand’s food exports in January-June 2026 were THB 693,376.26 million, down 7.8% year on year. The main factors cited were the strong baht, war, and rising freight costs. When it is hard to compete on price or exchange rates, quality assurance carries more weight in “continuing to be chosen” by customers. A single foreign body complaint can shake the continuation of a business relationship itself.

Thailand’s total exports, meanwhile, are moving differently. According to The Nation Thailand (August 9, 2026), Thailand’s total exports in the first half of 2026 were USD 196.74 billion, up 17.6% year on year. However, this figure was pushed up by electronics and similar products and runs opposite to food exports. You cannot say “Thai exports are strong, so food is strong too.”

On the investment side, the Thailand Board of Investment (BOI) announced in a press release on August 4, 2026, that investment applications in January-June 2026 totaled THB 1.47 trillion (up 37% year on year). The same release states that the agriculture and food processing sector “continues to move up the value chain.” For high value-added processed foods and pet food, the level of quality assurance demanded by customer audits also rises.

Labor costs cannot be ignored either. Thailand’s minimum wage has been THB 337-400 per day since July 1, 2025 (at the time of research). In a process where workers lined up beside the conveyor pick out bones and hard foreign bodies by eye, costs grow heavier the more people you add, and missed items still cannot be brought to zero.

Overseas, cases of metal fragments that were not stopped in the process continue. In the United States, Ukrop’s Homestyle Foods recalled a total of about 23,000 pounds of Baked Spaghetti and Chicken Cobbler due to possible metal fragment contamination. This was announced by the USDA Food Safety and Inspection Service (USDA FSIS) on August 2, 2026, and covers products made from July 1 to 29, 2026. It came to light when consumers reported finding metal fragments, and no injuries have been reported. Because the fragments were not found in the process but in the hands of consumers, it is a case that prompts a review of inspection equipment and verification practices.

Equipment is evolving too. At interpack 2026 in May 2026, Mettler-Toledo presented the M50 R-Series AdvancedLine metal detector, the X56 DXD+ dual-energy photon-counting X-ray system, the Eagle MAXIMIZER RMI X-ray system for poultry, the C35 AdvancedLine checkweigher, and the ProdX data management software. The trend of proposing metal detectors, X-ray, checkweighers, and inspection data management as a combination is becoming clear.

How Metal Detectors and X-ray Foreign Body Detection Equipment Differ in Principle

Before comparing equipment, let us establish “what each of the 2 methods is looking at.” If you get this wrong, no amount of catalog comparison will prevent a selection mistake.

Metal detectors: looking for an “imbalance” in balanced coils

The metal detectors common in food factories use the balanced coil method. According to the background art section of a patent specification from Mettler-Toledo Safeline, 3 coils are wound parallel to one another on a non-metallic frame, and a high-frequency current is passed through the central transmitter coil to create a magnetic field. On either side of it, identical receiver coils are placed at the same distance. The 2 receiver coils are wound in opposite directions and connected in series, so when no metal is present the induced voltages cancel each other out and the output is zero.

When a metal particle passes through, the magnetic field is disturbed first near one receiver coil and then near the other, and the voltage in each receiver coil changes. This change is typically very small, in the nanovolt range. The imbalance is amplified and processed to determine “metal present.”

What matters is that the nature (phase) of the signal differs by type of metal. According to the same specification, the signal from ferrous metals is mainly a reactive component, while the signal from stainless steel is mainly a resistive component. Conductive products containing moisture or salt also produce a strong resistive component. In other words, the signal from stainless steel and the signal from wet product are similar in nature. For this reason, an operating frequency is chosen at which the foreign body signal and the product signal separate by phase (the specification gives 300-850 kHz as a typical example, but this is an illustration in a single patent, not an industry standard).

X-ray foreign body detection equipment: looking at the “shadow” of density and thickness

X-ray foreign body detection equipment creates an image from differences in the intensity of X-rays that pass through the product. According to a technical document from Anritsu, high-density foreign bodies that X-rays do not pass through easily appear as dark shadows, so they can be detected regardless of magnetism. Not only metal but also non-metallic foreign bodies such as rubber and glass can be detected if they appear darker than the product.

On the other hand, the same document also shows the limits. Even a high-density foreign body lets X-rays pass more easily the thinner it is, which weakens the cue available for detection. It also states that for very thin metals such as rust or aluminum foil, metal detectors are superior to X-ray. It then concludes that the 2 methods complement each other well, so using them together in series makes for the safest inspection process. This is a manufacturer’s technical document, but the point that you cannot say “once you install X-ray, a metal detector is unnecessary” is a fact that forms the foundation of this whole article.

Furthermore, X-ray is said, in principle, to have difficulty finding foreign bodies whose density differs little from the product. A product introduction article on an X-ray system also states that the equipment detects foreign bodies that absorb more X-rays than the surrounding product; put the other way around, foreign bodies with about the same density as the product are hard to detect.

Dual-energy X-ray is a method to make up for this weakness. It uses 2 kinds of X-rays, high energy and low energy, to distinguish materials of different density, and it is explained that software subtracts the product from the image so that only the foreign body stands out. It is said to suit cases aiming to detect glass, plastic, rubber, calcified bone, and the like, but how much can be seen depends on the combination of product and foreign body.

Decide by the Foreign Bodies You Want to Stop and the Product: Comparison with X-ray Foreign Body Detection Equipment

This is the core of selection. What you decide first is not the equipment but “which foreign bodies you want to stop” and “what properties the product has.”

Stainless steel makes metal detector selection difficult

Ease of metal detection is generally said to follow the order “ferrous > non-ferrous > stainless steel.” Ferrous metal is magnetic and conducts electricity well, so it is the easiest to detect; non-ferrous metals such as brass, copper, aluminum, and lead are not magnetic but conduct electricity well. Stainless steel, especially non-magnetic grades such as 304 and 316, also has low conductivity and is said to be hard to detect.

Much of the equipment in food factories is made of stainless steel. Considering that many of the metals that can get into product, such as chipped blades, mesh fragments, bolts, and washers, are stainless steel, comparing metal detectors only by a figure like “ferrous sphere X mm” is meaningless. In the RFP, the starting point is to write the target size for each of ferrous, non-ferrous, and stainless steel, with your own product passing through.

Product effect: wet, salty, and partially thawed product creates false rejects

Many metal detector false rejects are caused by “product effect.” According to a Thermo Fisher Scientific blog post (January 16, 2019), in products high in salt or moisture, combined with air bubbles and density differences, it becomes hard to distinguish the true signal of non-ferrous metal from false signals coming from the product. Cheese is highly conductive and is said to be one of the most difficult products to inspect.

Temperature state matters too. According to the same blog, frozen bakery products show less product effect than in their fresh state, while partially thawed products are said to cause false detections easily. This is because the unfrozen core produces a metal-like signal. This is also why, in factories that pack and freeze cooked meat products, false rejects tend to increase when product temperature at the time of packing varies.

One countermeasure is a method that uses multiple frequencies simultaneously or by switching between them. The same blog explains that metal hidden at one frequency can sometimes be seen at another. In the company’s example, a multiscan type operates in the 50-1000 kHz range and selects and uses up to 5 frequencies, allowing sensitivity to be optimized for each type of metal (manufacturer’s explanation). Loma’s variable frequency type is also introduced as a method that automatically analyzes product effect, scans the band, and selects a frequency suited to the application. However, how much the difference in method reduces false rejects cannot be known without trying it on your own product.

Aluminum-packaged products and metal detectors

For products in aluminum foil packaging or aluminum trays, a standard metal detector reacts to the packaging itself. What is used in this case is a dedicated machine that detects only ferrous metal (ferrous-in-foil). Fortress Technology’s Stealth In-Foil is a ferrous-only detector built to handle signal interference from aluminum packaging and aluminum trays, and its listed applications include packaging for bakery, ready meals, confectionery, and protein products such as meat.

However, the company clearly states that this model cannot detect aluminum or non-magnetic stainless steel. If the risk of non-magnetic stainless steel in an aluminum-packaged product cannot be ignored, a metal detector alone cannot protect you. This is one of the reasons to choose X-ray, or to place a standard metal detector before packaging.

Read sensitivity in terms of “sphere diameter” and “orientation”

Metal detector sensitivity is generally said to be expressed as the diameter of a test sphere made of a specific metal. Because a sphere has no orientation, passing it through the same position in the aperture always produces the same signal, which makes it an easy basis for comparison. On the other hand, it is explained that for long, thin foreign bodies such as wire or swarf, ease of detection changes with the type of metal and the orientation in which they pass through.

Aperture size and the surrounding environment also affect sensitivity. Generally, the larger the aperture, the lower the sensitivity is said to be, and an area around the detector head where no metal is placed (a metal-free zone) must be secured. The required distance depends on the model and aperture dimensions, so follow the manufacturer’s installation instructions. If a detector is squeezed into an existing line afterward, the conveyor frame or reject device may be too close and the sensitivity cannot be achieved.

The option of installing “both”

To sum up so far, whether to choose a metal detector or X-ray is decided by the combination of foreign bodies you want to stop and the properties of the product.

• You only want to stop metal, and product effect can be controlled → metal detector (multi-frequency if needed)
• You also want to stop bone, glass, and stone, or you want to stop non-magnetic stainless steel in aluminum-packaged product → X-ray foreign body detection equipment
• Products with a high risk from aluminum foil packaging or thin metal fragments → there is the option of placing a metal detector and X-ray in series

Note that whether to manage metal foreign bodies as a HACCP critical control point (CCP) is not uniformly decided by international standards. The 2020 revision of the General Principles of Food Hygiene (Codex) is said to have added a tool for determining CCPs, but it does not say that metal detection must be a CCP; it is decided by each company’s hazard analysis.

Detection Alone Is Not Enough: Rejection and Fail-Safes

No matter how sensitive the equipment you choose, it is meaningless unless detected product reliably leaves the line. If product is detected but not physically rejected, that inspection is a fail.

Match the reject method to the product

Under BRCGS, metal detectors and X-ray systems are said to require mechanisms such as an automatic reject device that removes contaminated product from the product flow into a locked unit that only authorized personnel can open, and, for large packs and other cases where automatic rejection is not possible, a belt stop with an alarm. Reject methods include kickers (pushers), air blast, belt stop alarm, and overhead sweep, and you choose one that suits the product’s weight, shape, spacing, and line speed.

Reject timing is easily overlooked. A contributed article by Mike Bradley of Mettler-Toledo (FoodTechBiz, August 10, 2022) says that retailers want to confirm that reject timer settings are correct. On high-speed lines with tight product spacing, an accident can occur in which the product next to the contaminated one is removed while the contaminated product itself passes through.

Fail-safe: “fail toward the safe side when something breaks”

A Thermo Fisher Scientific blog post (July 10, 2019) describes the standard of the UK retailer M&S as among the particularly strict ones and introduces it as requiring failure to the safe side under all conditions. Examples of fail-safes given include detector fault detection, reject confirmation alarms, full reject bin notification, low air pressure alarms, and equipment failure detection.

The fail-safes to write into the RFP are at least the following 4:

1. Reject confirmation: after the reject device operates, a sensor confirms that the product actually entered the reject bin. If this cannot be confirmed, the line stops
2. Full bin detection: prevents the reject bin from overflowing and rejected product returning to the line
3. Air pressure monitoring: stops the line if air pressure for the air blast or pusher drops
4. Locked reject bin: ensures that rejected product can only be taken out by authorized personnel

This flow of “judging by inspection, rejecting via the PLC, and confirming the rejection” is a design shared with other 100% inspections such as visual inspection. The design approach from judgment to rejection is explained in detail in “AI visual inspection on high-speed lines (inspection and PLC rejection).”

Verification in Operation: Test Pieces, Frequency, and Records to Protect the HACCP Metal Detection CCP

A metal detector or X-ray system performs best on the day it is installed. After that, sensitivity and the reliability of rejection may decline unnoticed due to product changeovers, setting changes, wear of parts, and impacts during cleaning. Daily verification is what finds this.

Pass test pieces together with real product

An article by CMMS vendor Oxmaint (July 3, 2026) explains that BRCGS Issue 9 requires periodic checks of metal detection equipment, with the frequency decided by the site’s HACCP team. The same article cites passing certified ferrous, non-ferrous, and stainless steel test pieces together with real product, not on an empty belt, at the leading edge, middle, and trailing edge of the product, and states that a test piece that is detected but not physically rejected counts as a fail.

If you run only a test piece on an empty belt, there is no product effect, so sensitivity looks better than it really is. Verification is therefore premised on real production conditions, with product effect present.

Types of tests

The contributed article by Bradley mentioned above lists the following 4 tests for meeting retailer requirements:

• Standard test
• Consecutive test
• Memory test
• Large metal test (for side reject devices with a photo gate)

According to the same article, test packs use ferrous, non-ferrous, and stainless steel test pieces, placed so as to pass through the geometric center of the aperture. The consecutive test runs packs in succession at normal pack spacing and normal production speed. On high-speed lines, it serves to confirm that only the contaminated product is correctly removed without taking the products before and after it along with it.

Decide frequency by risk, not by “someone’s recommendation”

Various “recommended values” circulate for check frequency. The Oxmaint article mentioned above gives examples such as at start-up, every 30 minutes to 2 hours during operation, after a stop of more than 15 minutes, after a product changeover, and after maintenance that touches the detector, but this is that article’s recommendation, not a requirement of the standard. Some major UK retailers are said to set check frequency in their own codes of practice.

Frequency is decided by your own HACCP team, based on risk, considering the product, line, and past failures. Then check whether customers or retailers in export markets have their own requirements, and follow the stricter one.

Records and traceability

For each verification, record when it was done, by whom, with which test piece and at which position, and whether detection and rejection were carried out correctly. If verification fails, the product that passed between the last pass and the current check must be identified, held, and re-inspected. Whether that range can be narrowed down depends on whether you have records that link lots and times.

Operation is possible with paper check sheets, but if verification results and reject history are output as data from the equipment and linked to lot information, both showing records in audits and narrowing the scope of impact when a problem occurs become faster. The approach to lot management in food factories is explained in “Food factory traceability,” and a mechanism for automatically collecting inspection data from equipment in “Automatic inspection data collection.”

Metal Detector and Checkweigher Combination Units, and X-ray Inspection of Missing Items and Counts

When choosing foreign body inspection equipment, deciding at the same time how to combine it with weight inspection (checkweighers) makes it easier to keep line length and costs down.

Side by side or combined

There are also combination units that put a metal detector and checkweigher into a single machine. Fortress Technology’s Raptor BBK Combination combines metal detection and case-level weighing in one machine, covering bags and cases from 25 kg up to 50 kg for heavy products. The company explains that, compared with placing 2 machines separately, it has a smaller footprint, is easier to integrate and maintain, and allows both to be operated from a single operator screen (HMI). The reject method is said to be selectable from kicker, air blast, belt stop alarm, overhead sweep, and others.

The advantages of a combination unit are footprint and unified operation. On the other hand, a failure of one function stops the whole line, and it is hard to replace just one of the two. Decide by your own circumstances: for example, a combination unit for an existing factory without spare line length, or side by side if you may replace only one of the two in the future. For equipment layout across the whole packing line, please also see “Packaging line automation.”

X-ray can also check missing items, counts, and shape

X-ray foreign body detection equipment can be used not only for foreign bodies but also for quality inspection. According to the product brochure for Anritsu’s XR75 (January 2018), the machine can detect high-density foreign bodies such as metal, wire, glass, stone, and calcified bone, and at the same time perform inspections for missing items, count, packaging, voids, and so on.

The same brochure includes an easy-to-understand example. In a product containing 12 cupcakes of 20±2 g each, if only 11 cupcakes of 22 g are packed, the total is 242 g and the weight falls within specification. This “short count,” which a checkweigher would pass, can be found in the image by X-ray. For assortment products and products with a stated count, the effect of introducing X-ray can be evaluated not only for foreign bodies but also for missing items.

Cost and ROI: Comparing a Metal Detector Upgrade and X-ray at a Model Factory

All figures from here on are placeholder values set independently by this article. They are neither industry averages nor survey results.

Common assumptions (Model Factory E)

Configuration A: Upgrade to a multi-frequency metal detector + reject with fail-safes

Initial investment is THB 1,100,000 for the multi-frequency metal detector, THB 300,000 for the reject mechanism, fail-safes (reject confirmation, full bin, air pressure monitoring, locked reject bin), and conveyor, THB 200,000 for installation and verification with real product, and THB 200,000 for test pieces and training, for a total of THB 1,800,000. Annual operating cost for verification, test piece renewal, and maintenance is set at THB 120,000 per year.

The benefits (assumed values) are as follows:

• False rejects: 0.5% → 0.1% (15,000 packs → 3,000 packs); the reduction of 12,000 packs × THB 25 = THB 300,000
• Foreign body complaints: metal 1 → 0, THB 120,000 (bone fragments and soft plastic cannot be stopped by a metal detector, so they do not change)
• Staff: visual sorting remains as a measure against bone fragments, so 0

Total benefits are THB 420,000 per year, annual net benefit after operating cost is 420,000 − 120,000 = THB 300,000 per year, and the simple payback period is 1,800,000 ÷ 300,000 = 6.0 years.

Configuration B: Dual-energy X-ray foreign body detection equipment (also serving as the metal detector) + reject with fail-safes

Initial investment is THB 3,800,000 for the X-ray foreign body detection equipment, THB 400,000 for the reject mechanism, fail-safes, and conveyor, THB 300,000 for OAP notification preparation, shielding checks, and radiation safety training, and THB 300,000 for installation and verification with real product (test pieces and product learning), for a total of THB 4,800,000. Annual operating cost for maintenance, a reserve for X-ray tube replacement, and dose checks is set at THB 350,000 per year.

The benefits (assumed values) are calculated as replacing the benefits of Configuration A (they are not added to Configuration A’s benefits). The reduction of 1 metal complaint is the same 1 complaint as in Configuration A.

• False rejects: X-ray is not affected by product effect, but false rejects remain due to density variation, so 0.5% → 0.2% (15,000 packs → 6,000 packs); the reduction of 9,000 packs × THB 25 = THB 225,000
• Foreign body complaints: 6 → 2 (metal 1 → 0, bone fragments 4 → 1 (small bone fragments remain), soft plastic 1 → 1 (remains because the density difference is small even for X-ray)); the reduction of 4 complaints × THB 120,000 = THB 480,000
• Staff: visual sorting 6 → 2 people (1 person per shift is kept); 4 people reduced × THB 240,000 = THB 960,000

Total benefits are THB 1,665,000 per year, annual net benefit is 1,665,000 − 350,000 = THB 1,315,000 per year, and the simple payback period is 4,800,000 ÷ 1,315,000 = about 3.7 years.

Key point 1: B, with about 2.7 times the initial investment, pays back faster. The reason is “foreign bodies a metal detector cannot stop”

Looking at the 5-year cumulative figure, Configuration A is 300,000 × 5 − 1,800,000 = −THB 300,000, so it does not fully pay back in 5 years. Configuration B is 1,315,000 × 5 − 4,800,000 = THB 1,775,000.

Taking only Configuration B’s increment, the additional initial investment is 4,800,000 − 1,800,000 = THB 3,000,000, the additional annual net benefit is 1,315,000 − 300,000 = THB 1,015,000 per year, and the payback on the increment is about 3.0 years.

Looking only at false reject reduction, Configuration A has the larger benefit (THB 300,000 versus THB 225,000). This is because of the assumption that false rejects from product effect can be greatly reduced by revisiting the metal detector’s frequency. Even so, Configuration B pays back faster because the largest item in Configuration B’s benefits of THB 1,665,000 is staff, at THB 960,000. And whether visual sorting can be reduced is decided by whether the equipment can stop bone fragments. A metal detector does not see bone fragments, so Configuration A cannot reduce visual sorting by even 1 person.

That does not make Configuration A a “bad investment.” The Configuration A upgrade is not an investment payback project but a project to reliably protect the metal CCP. When an old metal detector has many false rejects, the floor is tempted to lower the sensitivity, and the handling of rejected product also tends to get careless. The value of Configuration A lies not in its payback years but in the reduction of false rejects and the reliability of fail-safes and verification records. For a factory where the only foreign body to stop is metal and there are no bone fragment complaints, Configuration A is a rational choice.

Key point 2: How does the conclusion change if visual sorting cannot be reduced?

The most uncertain element in this estimate is Configuration B’s staff reduction. Let us recalculate, considering cases where customers require visual sorting to continue or where X-ray bone fragment detection is not sufficient on real product.

With a staff reduction of 0, the annual net benefit falls to 1,315,000 − 960,000 = THB 355,000 per year, payback is 4,800,000 ÷ 355,000 = about 13.5 years, and the 5-year cumulative figure is −THB 3,025,000. With a reduction of 2 people (instead of 4), the net benefit is 1,315,000 − 480,000 = THB 835,000 per year, and payback is about 5.7 years.

In other words, the basis for choosing Configuration B is the verification result itself: “down to how many mm can bone fragments be stopped in real product.” Rather than relying on the catalog detection size, always carry out a manufacturer’s test (sample test) using your own product and actual bone fragments and test pieces before ordering, and based on the result, agree with the customer’s quality assurance staff on how many visual sorters can be reduced. In the 90-day plan below, this is placed in days 31-60.

The approach of “whether to automate 100% inspection,” common to capital investment decisions in general, is also organized in “Criteria for automating 100% inspection.”

12 Items to Write in a Metal Detector RFP

These are the items you should write, at a minimum, in the RFP (request for proposal) when asking manufacturers or distributors for quotations.

1. Foreign bodies to stop and target sizes (by material): ferrous, non-ferrous, stainless steel, and for X-ray also bone, glass, stone, etc. Specify with which product passing through, and at which position in the aperture, the size is guaranteed
2. Product specifications: temperature (frozen, partially thawed, or ambient), moisture, salt, packaging material (film, aluminum foil, tray), dimensions and weight
3. Line speed and product spacing: speed at maximum and product spacing at its tightest
4. Aperture dimensions: maximum product dimensions and constraints on belt width and installation space
5. Reject method and isolation: method (kicker, air blast, belt stop alarm, etc.) and a locked reject bin
6. Full set of fail-safes: reject confirmation, full bin detection, air pressure monitoring, detector fault detection
7. Test pieces: material, size, whether calibration certificates are provided, and the renewal cycle
8. Verification frequency and recording method: data output of verification results and reject history, and integration with lot traceability
9. Cleanability and ingress protection rating: construction that withstands high-pressure and chemical cleaning, and consideration of condensation
10. Support with X-ray notification documents (for X-ray): the scope of model information and documents provided for notification to OAP
11. Consumable parts and local stock: stock of X-ray tubes, belts, sensors, etc., and lead times within Thailand
12. FAT/SAT criteria: test items, the products and test pieces used, acceptance criteria, and the scope of witnessing

Items 1 and 8 are especially important. If item 1 is left at “ferrous sphere X mm,” the proposals from different companies cannot be compared on the same basis. Item 8 links directly to narrowing the scope of impact when records are requested in an audit or when a verification fails. The overall process of ordering and accepting inspection equipment is explained in detail in “Ordering and acceptance (FAT/SAT) of inspection equipment in Thailand.”

What to Check at FAT/SAT

In the factory acceptance test before shipment (FAT) and the on-site test after installation (SAT), confirm with your own product, not catalog values, that the system “stops, rejects, and records.”

• Test pieces in real product: attach ferrous, non-ferrous, and stainless steel test pieces to real product and pass them at the leading edge, middle, and trailing edge of the product
• Consecutive test and memory test: run product in succession at normal spacing and normal speed, and see whether only the contaminated product is removed without taking the products before and after it along
• If detected but not rejected, it is a fail: confirm not only the detection display but that the contaminated product entered the reject bin
• Simulated fail-safe tests: deliberately create abnormalities such as blocking the reject confirmation sensor, filling the reject bin, and lowering air pressure, and confirm that the line stops
• False reject rate across product temperatures: run product across the range of temperatures that actually occur, such as around freezing and immediately after packing, and count how many packs are falsely rejected without a test piece
• For X-ray, detection rate and false reject rate with actual bone fragment samples: confirm the detection rate using not only test pieces but bone fragments actually found in your own product, and at the same time record the false reject rate for good product
• Missing item detection (for X-ray): run samples with short counts or chips and confirm the judgment

As the Thermo Fisher Scientific blog points out, partially thawed product causes false detections. FAT tends to be run only with product in good condition, but at SAT it is important to test including the temperature variation that actually occurs on site.

Issues Specific to Thailand and ASEAN

Notification for X-ray foreign body detection equipment

In Thailand, possession and use of radiation-generating equipment such as X-ray systems is regulated under the Nuclear Energy for Peace Act. According to an English translation by the legal group of the Office of Atoms for Peace (OAP), a 2020 ministerial regulation (B.E.2563) provided that equipment such as devices whose maximum radiation energy is below 1 MeV, used in a fully shielded state, and not used on people, does not require a license application and only requires notification of possession and use to the Secretary-General. In OAP explanatory material (lecture slides), “X-ray equipment for product quality inspection” is given as an example of this notification category, and it is explained that the notification form is to be submitted within 30 days of the date of possession and that there is no notification fee.

On the other hand, according to the same material, equipment below 1 MeV that is used without shielding, or that is used on people, is subject to licensing. Modifications such as removing the shielding curtains, or open-type use, may take the equipment out of the notification category.

Food X-ray foreign body detection equipment is considered in many cases to fall into the notification category, but please confirm for each model with OAP (Office of Atoms for Peace) or the vendor. Note that OAP’s English translation has no legal force; the authoritative text is the Thai original. As for safety requirements for equipment in the notification category, the ministerial regulation was described as a “draft” at the time of the material, so please also confirm the latest requirements with OAP. Incidentally, equipment of 5 kiloelectron volts or less is said to be outside the scope of the Act, but food X-ray systems operate at several tens of kV, so they do not fall under this. For reference, the manufacturer’s stated values for the Anritsu XR75 are a tube voltage of 30-80 kV and X-ray leakage of 1.0 μSv/h or less at maximum, but these are values for one model and do not apply to all models.

Food GMP (Ministry of Public Health Notification No. 420)

Thailand’s food GMP consolidated multiple earlier notifications into one with Ministry of Public Health Notification No. 420 (2020). According to the provisional translation by JETRO Bangkok, the objectives of the basic requirements include reducing and eliminating physical, chemical, and microbiological hazards. However, no explicit provision mandating the installation of metal detectors can be found in the text of the notification. Please confirm individually with the competent authority or certification body how records of foreign body control should be kept.

Heat, humidity, and high-pressure cleaning

Thai food factories routinely carry out high-pressure and chemical cleaning in a hot and humid environment. Include in the RFP and test conditions the ingress protection rating of the detector head and control panel, measures against condensation, and how product effect changes with temperature changes. On frozen product lines, it is also important not to place equipment where condensation readily occurs due to movement in and out of freezers.

List the requirements of each export market

Requirements for verification frequency and test pieces differ by export market. For the United States, the FDA Compliance Policy Guide (CPG Sec. 555.425) serves as a guide for foreign body size. Hard or sharp foreign objects less than 7 mm in maximum dimension are said to rarely cause trauma or serious injury, except in special risk groups such as infants, surgery patients, and the elderly. On the other hand, hard or sharp foreign objects of 7-25 mm in food eaten without cooking or with minimal cooking are said to be a basis for actions such as seizure. Note that this does not mean “less than 7 mm is safe”; products for special risk groups and the like are evaluated separately.

Attaching a list of each customer’s requirements, such as those of Japanese customers, the United States, and UK retailer codes of practice, to the RFP prevents mismatches in manufacturers’ proposals. Factories handling halal-certified products should also consider integration with halal control records (see “Halal traceability“).

Maintenance support

Check calibration certificates for test pieces, local stock of consumables such as X-ray tubes and belts, and maintenance support in Thai from the manufacturer or distributor. The decision of whether to continue production without inspection or stop the line when the inspection equipment stops should also be written into the HACCP procedures in advance.

Equipment procurement in BOI projects

For projects receiving BOI investment promotion, please confirm the treatment of equipment procurement and changes individually with the competent authority (BOI) or experts. The same applies to tax treatment.

Examples of options on the market

Manufacturers handling metal detectors, X-ray foreign body detection equipment, and checkweighers (including combination units) include, for example, Mettler-Toledo (Safeline metal detectors, Safeline/Eagle X-ray, checkweighers), Anritsu (X-ray inspection equipment, quality control systems), Fortress Technology (metal detectors, ferrous-only machines for aluminum packaging, combination units), Thermo Fisher Scientific (multiscan metal detectors), Loma Systems (variable frequency metal detectors, X-ray), Sesotec (X-ray), and Eriez (X-ray). They are listed here only as examples and are not a recommendation or an evaluation of relative merit.

90-Day Plan

Days 0-30: Measure foreign body complaints, false rejects, and visual sorting workload, and organize the foreign bodies to stop

Classify and tally the past 1-2 years of foreign body complaints into metal (by ferrous, non-ferrous, and stainless steel), bone fragments, glass and stone, plastic, and so on. At the same time, measure the number of false rejects from the current metal detector and the number of people and time spent on visual sorting. From this tally, create a list of “foreign bodies to stop” and draft target sizes. Replace the assumptions of the model estimate with your own figures and form an outlook on whether the direction of Configuration A or Configuration B fits.

Days 31-60: Manufacturer sample tests with real product, and agreement with the customer’s quality assurance staff

Send your own real product, along with actual bone fragments found and test pieces, to multiple manufacturers and request sample tests. For a metal detector, check false rejects across product temperatures; for X-ray, check the bone fragment detection rate and the false reject rate for good product. Based on the results, agree with the customer’s quality assurance staff on how many visual sorters can be reduced and on the frequency and recording method for verification.

Days 61-90: RFP, FAT/SAT criteria, and order decision (for X-ray, preparation of notification to OAP)

Based on the test results, finalize the 12 RFP items and the FAT/SAT test items and acceptance criteria, and obtain quotations from multiple companies under the same conditions. Compare them against your own version of the estimate and decide whether to order and in which configuration. If you choose X-ray, confirm the category for each model with OAP or the vendor and begin preparing the notification.

The 5 deliverables you want to have in hand at the end of the 90 days are: (1) a tally of classified foreign body complaints, false rejects, and visual sorting workload, and your own version of the investment estimate reflecting it; (2) a list of foreign bodies to stop and target sizes; (3) the results of sample tests with real product; (4) the verification and recording method agreed with the customer; and (5) the RFP and FAT/SAT criteria. If the tally in (1) shows that there are almost no bone fragment complaints, it is also rational to focus investment on upgrading the metal detector and strengthening fail-safes. If you are looking ahead to integration with production planning and lot information, “Production management systems for the food industry” may also be helpful.

Frequently Asked Questions

Which should I choose, a metal detector or X-ray foreign body detection equipment?

It depends on the foreign bodies you want to stop and the properties of the product. If you only want to stop metal and product effect can be controlled, a metal detector is a candidate. If you also want to stop bone, glass, and stone, or want to stop non-magnetic stainless steel in aluminum-packaged product, X-ray becomes a candidate. However, manufacturer documents explain that metal detectors are better at thin metals such as rust flakes and aluminum foil pieces, and for high-risk products there is also the option of placing both in series.

Why is stainless steel hard to detect?

Non-magnetic stainless steel such as 304 and 316 is not magnetic and has low conductivity, so it is generally said to be harder to detect than ferrous or non-ferrous metals. In addition, the signal from stainless steel is similar in nature to the signal from products containing moisture or salt, so it is easily buried in product effect. In the RFP, we recommend writing a target size for stainless steel as well, with your own product passing through.

Can metal detectors be used for aluminum-packaged products?

A dedicated machine that detects only ferrous metal (ferrous-in-foil) can be used. However, the manufacturer clearly states that this method cannot detect aluminum or non-magnetic stainless steel. If there is a risk of non-magnetic stainless steel, consider placing a standard metal detector before packaging, or X-ray foreign body detection equipment.

How often should metal detector checks (test piece verification) be done?

Frequency is decided by your own HACCP team based on risk. Then check whether customers or retailers in export markets set their own frequency, and follow the stricter one. Checks are premised on passing test pieces together with real product, not on an empty belt, and confirming not only detection but also rejection.

Is a license required to use X-ray foreign body detection equipment in Thailand?

Equipment below 1 MeV that is fully shielded and not used on people was placed in the notification category, not the license category, by a 2020 ministerial regulation. OAP explanatory material gives X-ray equipment for product quality inspection as an example and explains notification within 30 days of possession and no fee. Food X-ray foreign body detection equipment is considered in many cases to fall into the notification category, but please confirm for each model with OAP (Office of Atoms for Peace) or the vendor. The English translation is for reference only and has no legal force; the authoritative text is the Thai original. Modifications or open-type use may make the equipment subject to licensing.

What do metal detectors and X-ray systems cost, and what is the payback period?

In this article’s model estimate (all placeholder values), Configuration A, upgrading to a multi-frequency metal detector, had an initial investment of THB 1,800,000 and a simple payback of 6.0 years, and Configuration B, dual-energy X-ray, had THB 4,800,000 and about 3.7 years. However, Configuration B’s payback assumes that visual sorting can be reduced by 4 people; if the reduction is 0, it extends to about 13.5 years. You need to recalculate with your own complaint counts, false reject counts, and number of sorters, and confirm the assumptions through sample tests with real product. For the treatment of capital investment in BOI projects and tax matters, please confirm individually with the competent authority or experts.

Summary

• Metal detector selection is not decided by comparing sensitivity figures. It is decided by 3 things: the foreign bodies you want to stop and the properties of the product, reliable rejection after detection, and verification that runs every day.
• Metal detectors are weak on stainless steel and on wet, salty, and partially thawed product. X-ray can see bone, glass, and stone, but is weak on thin metal and foreign bodies with a density close to the product. Ferrous-only machines for aluminum packaging cannot detect non-magnetic stainless steel.
• If product is detected but not physically rejected, it is a fail. Write fail-safes for reject confirmation, full bin, air pressure monitoring, and a locked reject bin into the RFP.
• Verification frequency is decided by the HACCP team based on risk; confirm both detection and rejection with real product, and record results linked to lots.
• In the model estimate (placeholder values), the X-ray configuration, with about 2.7 times the initial investment, paid back faster. What made the difference was whether bone fragments could be stopped and visual sorting reduced. If no reduction is possible, payback extends to about 13.5 years, so sample tests with real product before ordering are the crux of the decision.

TOMAS TECH supports Japanese-owned food factories in Thailand from the preparation stage, such as tallying foreign body complaints and false rejects, organizing the foreign bodies to stop, and arranging manufacturer sample tests, through to RFP drafting and attendance at FAT/SAT. Even if you are at the stage of “first wanting to understand how many complaints and false rejects we currently have,” please feel free to contact us through our contact form.

References

• Google Patents, “US8841903B2 Metal detection apparatus” (Mettler-Toledo Safeline Ltd, September 23, 2014): https://patents.google.com/patent/US8841903
• Mettler-Toledo, “Safeline Metal Detection”: https://www.mt.com/gb/en/home/products/Product-Inspection_1/safeline-metal-detection.html
• Mettler-Toledo, “Metal Detector Sensitivity”: https://www.mt.com/us/en/home/library/know-how/product-inspection/metal-detector-sensitivity.html
• Thermo Fisher Scientific, “How is metal detection affected by wet, salty and high-mineral foods” (January 16, 2019): https://thermofisher.com/blog/food/how-is-metal-detection-affected-by-wet-salty-and-high-mineral-foods
• Food Engineering (Loma variable frequency type): https://foodengineeringmag.com/articles/86227-variable-frequency-detector
• Oklahoma State University, “Metal Detectors for Food Processing”: https://extension.okstate.edu/fact-sheets/metal-detectors-for-food-processing.html
• Fortress Technology, “Stealth In-Foil Metal Detector”: https://fortresstechnology.com/solutions/stealth-in-foil-metal-detector/
• Fortress Technology, “Raptor BBK Combination System”: https://fortresstechnology.com/solutions/raptor-bbk-combination-system/
• Anritsu, “Technical Note Vol.2”: https://dl.cdn-anritsu.com/anritsu-infivis/en-gb/technical-note/technicalnote-vol2-en.pdf
• Food Safety Magazine, “RAYCON for Food Contaminant Detection”: https://www.foodsafetymagazine.com/products/raycon-for-food-contaminant-detection/
• Packaging Digest, “How to choose between single or dual energy X-ray systems”: https://www.packagingdigest.com/packaging-technologies/how-to-choose-between-single-or-dual-energy-x-ray-systems
• Anritsu XR75 product brochure (January 2018): https://cdn.sigmaequipment.com/files/attachment/anritsu-xr75-x-ray-brochure-xr.pdf
• U.S. FDA, “CPG Sec. 555.425 Foods, Adulteration Involving Hard or Sharp Foreign Objects”: https://www.fda.gov/regulatory-information/search-fda-guidance-documents/cpg-sec-555425-foods-adulteration-involving-hard-or-sharp-foreign-objects
• IFSQN forum (BRC reject requirements for foreign body detectors): https://www.ifsqn.com/forum/index.php/topic/22328-questions-regarding-brc-issue-6-41033
• Oxmaint, “Metal Detector CCP Sensitivity Verification” (July 3, 2026): https://oxmaint.com/industries/food-manufacturing/metal-detector-ccp-sensitivity-verification-cmms-guide
• Thermo Fisher Scientific blog (retailer standards and fail-safes, July 10, 2019): https://www.thermofisher.com/blog/food/?p=10254
• FoodTechBiz, “How to meet retailers’ metal detection testing requirements” (August 10, 2022): https://www.foodtechbiz.com/opinion/how-to-meet-retailers-metal-detection-testing-requirements
• Certifico, “Codex Alimentarius CXC 1-1969”: https://certifico.com/haccp/documenti-haccp/codex-alimentarius-cxc-1-1969
• OAP English translation of ministerial regulation (radiation-generating equipment subject to notification, 2020): https://www.oap.go.th/wp-content/uploads/2022/12/Ministerial-Regulation-RadGensToBeNotified.pdf
• OAP lecture material on licensing and notification of radiation-generating equipment (in Thai): https://www.oap.go.th/wp-content/uploads/2023/01/03_13-15-12.pdf
• OAP English translation of ministerial regulation (radiation-generating equipment not under control, 2020): https://www.oap.go.th/wp-content/uploads/2024/03/MinRegRadGenNotUnderControl.pdf
• JETRO Bangkok, provisional translation of “Ministry of Public Health Notification No. 420”: https://www.jetro.go.jp/ext_images/thailand/food/MoPH420.pdf
• Thansettakij (first-half 2026 food exports, September 17, 2026): https://www.thansettakij.com/economy/trade-agriculture/669128
• The Nation Thailand (first-half 2026 exports, August 9, 2026): https://www.nationthailand.com/business/trade/40069585
• BOI press release No.124/2569 (August 4, 2026): https://www.boi.go.th/upload/content/No.124_2569.pdf
• DLA Piper, “Thailand increases minimum wage across all provinces” (July 21, 2025): https://www.dlapiper.com/insights/publications/2025/07/thailand-increases-minimum-wage-across-all-provinces
• Tilleke & Gibbins, “Thailand Increases Minimum Wage Across All Provinces” (January 2, 2025): https://www.tilleke.com/insights/thailand-increases-minimum-wage-across-all-provinces
• interpack exhibitor news (Mettler-Toledo, May 7, 2026): https://www.interpack.com/en/Media_News/News/Exhibitor_News/Mettler-Toledo_Showcases_Smart_Product_Inspection_Solutions_for_Food_Manufacturers_at_Interpack_2026
• FOX 29 (Ukrop’s recall, August 2, 2026): https://www.fox29.com/news/ukrops-baked-spaghetti-chicken-cobbler-recall