Metal Detector False Rejects: Causes and Calibration

Metal Detector False Rejects: Causes and Calibration

False rejects are one of the most frustrating problems on a packaging line using the MTD-001 metal detector. A good pouch, tray, or frozen pack gets kicked off the belt even though there is no metal inside, and every reject represents lost product, lost throughput, and operator distrust in the inspection step. In a high-volume frozen food or ready-meal plant running at 60 to 120 packs per minute, false reject rates above 1% can translate into thousands of dollars of waste per shift and, worse, can mask real contaminant events because operators start ignoring or bypassing the reject station.

This troubleshooting guide explains the root causes of metal detector false rejects on the MTD-001, walks through the wet product effect, vibration, frequency selection, and phase adjustment, and provides a complete calibration workflow that you can run during changeover or after a sensor replacement. The goal is to bring your false reject rate back below 0.3% while keeping your ferrous, non-ferrous, and stainless steel sensitivity at the levels required by your customer spec and by retailer codes of practice.

• False rejects are almost always caused by one of four things: wet or conductive product effect, mechanical vibration, incorrect operating frequency, or out-of-calibration phase and balance.
• The “wet product effect” is electrical, not magnetic — salty, acidic, or high-moisture foods conduct the detector field and look like a metal signature at the wrong frequency.
• Vibration from the rejector, belt drive, or nearby machines couples into the search head and produces a noise floor that crosses the reject threshold.
• Frequency selection is a trade-off: high frequency gives better sensitivity on small non-ferrous contaminants but is more easily fooled by wet product; low frequency is more stable on wet product but less sensitive to thin wire and stainless shards.
• Phase adjustment (the product-compensation angle) must be re-taught every time the product, the belt speed, the tray shape, or the temperature class changes.
• A calibrated MTD-001 should hold a false reject rate under 0.3% over an 8-hour run and pass a daily ferrous / non-ferrous / stainless test piece challenge at the agreed diameter.
• Most false-reject tickets are solved in under 30 minutes if you follow the order: clean, dry, isolate, re-teach phase, verify with test pieces.

Understand Why the MTD-001 Is Rejecting Good Product

Before changing any setting, confirm that the detector is actually rejecting for the wrong reason. Open the MTD-001 event log and look at the last 50 rejects. Each event is tagged with a suspected contaminant type (FE, NF, SS), a signal strength in microvolts, and a phase angle. If the signal strength is hovering just above the threshold and the phase angle is scattered across the full dial, the detector is responding to background noise or product effect, not a real piece of metal. If the signal is sharp, narrow, and well above threshold, you may in fact have a real contaminant or a real mechanical interference such as a metal clip on a film tail.

False rejects fall into four mechanical or electrical categories. Product effect is the most common on frozen, ready-meal, and high-moisture products. Vibration is the second most common and is often misdiagnosed as product effect because it produces a similar drifting baseline. Frequency mismatch is usually introduced when a new SKU is run on a setting left over from a previous product. Phase drift happens slowly over months as the search head coils age and the temperature inside the electrical cabinet cycles. Knowing which of the four you are dealing with determines which calibration step you start with.

The Wet Product Effect on Frozen and High-Moisture Foods

The wet product effect is the single largest source of false rejects on food lines. It is not caused by water being magnetic. It is caused by dissolved salts, acids, and minerals in the product making the food electrically conductive. When the MTD-001 search head transmits its electromagnetic field, a conductive product absorbs energy and shifts the phase of the received signal. The detector sees the shift and, if it is large enough, interprets it as a non-ferrous or stainless contaminant.

Frozen foods are a special case. A block of pure ice is relatively transparent to the detector, but a frozen ready-meal containing sauce, brine, marinated protein, or cheese is highly conductive even at minus 18 degrees Celsius. As the product thaws during a long production run, the conductivity rises, the product effect signature grows, and a detector that was perfectly calibrated at the start of shift starts rejecting at the end. This drift is normal and is the reason that phase compensation has to be re-taught whenever the product temperature class changes by more than about 5 degrees.

Other common product-effect triggers are tomato-based sauces, soy sauce and other brines, cured and salted meats, processed cheese, fruit fillings with citric acid, and any product packed in a thin metallic pouch that the detector can see through but that adds its own signal. The table below summarizes typical product categories and the operating frequency that usually works best as a starting point on the MTD-001.

Vibration and Mechanical Noise Coupling

Vibration is the second most common cause of false rejects and it is frequently misread as product effect because the symptoms look the same: a wandering baseline, intermittent rejects, and signal strength just above threshold. The difference is that vibration-induced rejects tend to happen at the same physical position on the belt, get worse when the line speeds up, and disappear the moment the belt is stopped while the detector stays on. If you see any of those three patterns, suspect vibration before you suspect the product.

The MTD-001 search head is a balanced coil assembly. The transmit coil sits between two receive coils wired in opposition so that, in a perfectly quiet mechanical environment, the two received signals cancel each other out and the detector sees zero. Any vibration that moves the coils relative to each other breaks that balance and produces a signal that is indistinguishable from a small contaminant. The most common sources are the rejector solenoid or pneumatic cylinder firing too aggressively, the belt drive motor sitting on the same frame as the search head, worn belt tensioners causing the belt to slap the deck, and nearby forklift traffic or large mixer motors bolted to the same floor slab.

Run the belt empty at full speed for two minutes and watch the live signal bar. If it moves more than roughly 20 percent of the reject threshold with no product present, you have a vibration problem, not a product problem.

The fixes are mechanical first. Tighten the belt to spec, replace worn tensioners, isolate the search head frame from the rejector frame with rubber dampers, and make sure the detector is bolted to its own pad, not sharing a frame with a 15-kW mixer. After mechanical isolation, re-run the empty-belt test and only then move on to electrical and product-effect calibration.

Frequency Selection on the MTD-001

The MTD-001 is a tunable-frequency detector. Choosing the right operating frequency is the single biggest lever you have for trading sensitivity against false rejects. A high frequency (250 to 400 kHz) gives strong signal amplitude on small ferrous and non-ferrous particles, which makes 1.0 mm and even 0.8 mm test pieces detectable on dry products. The downside is that high frequencies are strongly absorbed by conductive foods, so the same 300 kHz setting that gives great sensitivity on dry crackers will produce continuous false rejects on a marinated chicken line.

A low frequency (40 to 100 kHz) penetrates conductive products much better and stabilizes the baseline, but at the cost of reduced sensitivity to thin wire, stainless shards, and small non-ferrous particles. Most BRC and IFS customer specs therefore require a documented compromise: the lowest frequency that still passes the agreed test piece set on that specific product. If you switch SKUs, you must re-run the test piece challenge at a frequency that works for the new product and store the result as a named product recipe on the MTD-001.

Modern installations increasingly use multi-frequency or simultaneous-frequency detection, where the MTD-001 runs two frequencies at once and the algorithm subtracts the product effect signature captured at one frequency from the contaminant signature captured at the other. If your unit supports this, enable it on high-product-effect lines such as frozen meals and tomato sauces, and confirm with the test piece set that the dual-frequency mode still meets the spec sensitivity. Dual-frequency mode is not free — it slightly raises the baseline noise — but on wet products it usually reduces false rejects by an order of magnitude.

Phase Adjustment and Product Compensation

Phase adjustment is the calibration step that compensates for the electrical signature of the product itself. Every conductive product produces a signal at a specific phase angle relative to the transmit field. The MTD-001 allows you to rotate the detection angle so that the product’s own signature is cancelled out, leaving only signals that look like real metal. Done correctly, phase adjustment can take a high-product-effect line from continuous false rejects down to under 0.2% false reject rate without sacrificing sensitivity.

The procedure is identical on the MTD-001 regardless of product. Run the belt empty and zero the baseline. Then place ten consecutive good product samples on the belt, one at a time, and let the detector measure the product signature at the current frequency. The detector will calculate an average phase angle and a residual signal strength. Rotate the phase to null the product signal, then narrow the detection window around the residual. Finally, run the test piece set — typically a 1.5 mm ferrous, 2.0 mm non-ferrous, and 2.5 mm stainless sphere on most frozen lines — and confirm each is detected 10 times out of 10 passes through the center of the aperture.

Phase needs to be re-taught whenever any of the following change: the product itself, the tray or pouch shape, the product temperature class, the belt speed, the orientation of the pack, or after any service work that moved or replaced the search head. A common mistake is to assume that “the same product” from a different supplier does not require re-teaching. Brine levels, fat content, and seasoning blends all vary batch to batch and can shift the phase angle enough to push a previously stable detector into false reject territory.

What Is a Complete Calibration Workflow?

This is the order in which to troubleshoot and calibrate the MTD-001 when false rejects spike. The sequence matters because each step assumes the previous one is clean. Trying to re-teach phase on a detector that is still vibrating will produce a phase setting that works only at that exact moment and drifts within an hour.

Start with a clean machine. Strip the belt, clean the search head aperture, and inspect the belt surface for embedded metal fragments that may have been causing real but tiny rejects. Run the belt empty at full speed for two minutes and confirm the baseline is below 20 percent of threshold. If it is not, fix the vibration first.

Next, isolate the product effect. Place a sample of the product on the belt and watch the live signal. If the signal spikes as the product enters the search head and returns to baseline after it leaves, you have product effect. Adjust the frequency down one step at a time until the product signature is at or below the level where it would trigger a reject, then re-run the test piece set to confirm sensitivity is still on spec. If you cannot find a single frequency that gives both acceptable product-effect stability and the required sensitivity, switch to dual-frequency mode.

Only after frequency and product effect are under control should you teach the phase. Use ten fresh product samples from the current production batch, not the calibration reference sample that has been sitting in the lab for two years. Save the result as a named recipe tied to the SKU. Then run a 100-piece production trial and confirm the false reject rate is below 0.3% while all three test pieces still detect on every pass.

Daily Verification and Long-Term Stability

Calibration is not a one-time event. The MTD-001 should be challenged with test pieces at the start of every shift, at every product changeover, after any belt stoppage longer than 30 minutes, and at the end of every shift. The standard practice is to use a wand carrying ferrous, non-ferrous, and stainless spheres placed on the leading edge, center, and trailing edge of a representative pack, with each sphere passing through the geometric center and both sides of the aperture. All nine tests must trigger a reject for the detector to be considered in calibration.

Over the long term, search head coils age, the epoxy potting expands and contracts with temperature cycles, and the electronics drift. Expect to need a full service recalibration every 12 months on a single-shift line and every 6 months on a three-shift line. Track the false reject rate weekly. If you see it creeping upward week over week even after phase is re-taught, the search head is likely aging and a service visit is due. Tracking the rate is also your defense in a customer audit: a graph showing 0.1 to 0.3% false rejects with daily test piece passes is strong evidence that your inspection step is under control.

A well-calibrated MTD-001 is invisible to the operator. If your line operators are talking about the metal detector, it is either rejecting too much or failing its test pieces. Both are calibration problems.

Frequently Asked Questions

Why does my MTD-001 reject the same SKU that ran fine last week? Almost always a phase or product effect drift caused by a recipe change at the supplier, a different product temperature class, or a slow vibration that has developed from a worn belt tensioner. Re-run the empty-belt baseline test, then re-teach phase using fresh product from the current batch.

How low should my false reject rate be? On a properly calibrated MTD-001 running a stable product, the false reject rate should sit between 0.1% and 0.3%. Anything above 0.5% indicates a calibration, vibration, or frequency mismatch issue that should be fixed before it normalizes operator distrust of the reject station.

Can I just raise the threshold to stop false rejects? You can, but every increase in threshold directly reduces sensitivity to real contaminants. Raising the threshold without fixing the underlying product-effect or vibration problem is how metal escapes reach the customer. Fix the root cause first, raise the threshold only as a last resort, and document any change with a fresh test piece challenge.

Does freezing the product reduce the wet product effect? Partially. A fully frozen block at minus 18 degrees Celsius is less conductive than the same product at chill temperature, but marinated and sauce-based products remain highly conductive even when frozen solid because the dissolved salts are still mobile. The frozen ready-meal category still requires low-frequency operation and phase compensation on the MTD-001.

How often should I recalibrate phase? At minimum, at every product changeover and at the start of every shift. On lines where the product temperature class changes during the run, such as a thawing ready-meal line, re-teach phase at the start and again halfway through the run. Use the named recipe function on the MTD-001 so the second teach is a one-button recall rather than a full procedure.

What is the smallest contaminant the MTD-001 can reliably detect? It depends entirely on the product and the aperture size. On a dry product in a small aperture, the MTD-001 will reliably find 1.0 mm ferrous, 1.2 mm non-ferrous, and 1.5 mm stainless. On a wet, salty, frozen product in a larger aperture, realistic targets are 2.0 mm ferrous, 2.5 mm non-ferrous, and 3.0 mm stainless. Always agree the test piece set with your customer spec and document it.

If false rejects are eating your throughput or your MTD-001 is no longer passing its daily test piece challenge, contact our engineering team for a calibration visit, a search head health check, or help selecting the right operating frequency for your high-moisture products. Email [email protected] with your line details and the SKU that is causing trouble, and we will send a calibration plan within one business day.

Learn more: cooking oil processing applications

Get a quote: [email protected]