Metal Detector Reject: Challenge Testing, Reject Verification and Records
Follow the detector signal through physical removal, challenge testing, failure containment and documented human decisions.

On this page · 11 sections
- 1What it is: detection is not confirmed rejection
- 2U.S. food requirements, FDA guidance and scope boundaries
- 3Specify the complete detector-to-reject chain
- 4Challenge testing: capability, routine checks and acceptance
- 5An operating sequence from setup to controlled restart
- 6Failed challenges: contain, assess and decide disposition
- 7Evidence records and the required review
- 8Connect line events to batches and recover exceptions
- 9Failure modes and controls to challenge deliberately
- 10Use reject trends without confusing them with safety evidence
- 11Implementation, change and restart checklist
How does Metal Detector Reject apply to your shop floor?
Pick your industry and scale — Ask V5 rewrites the definition in your context, gives a worked example, and shows what V5 does on day one.
01What it is: detection is not confirmed rejection
Metal Detector Reject is the system function that separates product flagged by a metal detector from the accepted product stream. This is an operational definition, not a statutory definition. Its boundary includes identifying the intended product and physically separating it—not merely generating an alarm or sending an instruction to a reject device.
| Term | What it establishes | What it does not establish |
|---|---|---|
| Detector alarm | An indication that the detector flagged a response. | That any product left the accepted stream. |
| Reject command | An instruction to operate the reject device. | That the device moved or selected the correct product. |
| Confirmed rejection | Observed removal of the intended product from the accepted stream. | Recovery of the challenge piece unless separately checked. |
| Challenge test | A controlled test of defined detection and rejection functions. | Performance outside the tested conditions or functions. |
For practical acceptance criteria, separate four observations: detection, correct product selection, physical separation and recovery of the challenge piece. Record each outcome against the intended result. A detection-only result does not demonstrate the other three outcomes. For example, an alarm followed by removal of the next pack rather than the challenged pack demonstrates neither correct selection nor successful rejection of the challenged pack.
Metal detector reject verification means, in this operational usage, the site's documented check that the intended rejection occurred under specified conditions. The broader regulatory meaning of verification is addressed later.
Place the mechanism within the site's foreign-material control approach and, where used, its HACCP hazard-control plan. The physical reject mechanism is one operating function, not the plan itself.
02U.S. food requirements, FDA guidance and scope boundaries
Start with the product and jurisdiction, not the detector model. The regulatory discussion here concerns U.S. human-food operations where the relevant provisions of 21 CFR 117 apply. Keep binding requirements, FDA explanations and site operating specifications distinct when writing the detector procedure.
| Jurisdiction and product context | Authority or reference | Application to this article | Scope check |
|---|---|---|---|
| U.S. human-food operations where the relevant Part 117 provisions apply | 21 CFR Part 117 — binding regulation | Preventive-control management framework for the discussion below. | Check the current provisions for the particular operation and control. |
| FDA hard-or-sharp-object guidance for food | CPG Sec. 555.425 — FDA guidance | Injury context, not an equipment specification. | Preserve the qualifications concerning object dimensions and special-risk consumers. |
| U.S. supplements and other regulated product categories | Current product-specific requirements | Do not transfer this human-food regulatory map to another product category. | Identify the applicable requirements for the product and activity. |
| Non-U.S. manufacturing | Current jurisdiction- and product-specific requirements | Use a separate local regulatory assessment. | Check manufacturing and destination-market requirements without assuming equivalence. |
Binding framework and explanatory references
21 CFR 117.140(a) lists three preventive-control management components: monitoring under § 117.145; corrective actions and corrections under § 117.150; and verification under § 117.155. Their detailed requirements belong in the challenge-testing, failure-handling and records sections below.
FDA’s Preventive Controls for Human Food final-rule page is an explanatory agency summary of the rule’s food-safety-plan and risk-based preventive-control framework. Read it separately from the binding Part 117 provisions. CPG Sec. 555.425 is FDA guidance; recommended operating practices below are a third, distinct category.
Injury context is not detector acceptance
In CPG Sec. 555.425, Background, FDA reports the Health Hazard Evaluation Board’s finding that foreign objects below 7 mm in maximum dimension rarely cause trauma or serious injury, except in special-risk groups such as infants, surgery patients and the elderly. Retain those exceptions whenever using the finding.
As risk-based specification practice, do not convert that injury statement into detector sensitivity, challenge-piece size, an acceptance limit or permission to return rejected food to production. Establish operating specifications through site-specific justification and validation rather than adopting an injury-context dimension as a machine setting.
Likewise, the site’s hazard-control classification and the equipment’s physical function are separate decisions. In the site’s HACCP assessment, justify any critical control point designation; do not label every metal detector a CCP merely because it detects metal and actuates a reject device.
03Specify the complete detector-to-reject chain
Use the following as site-specific design and acceptance questions, not legal equipment specifications. Carry them into the factory acceptance test, with observable pass criteria for the complete arrangement rather than the detector alone.
- Detector indication: What signal identifies a detection, and how can the assessor distinguish it from a fault or status indication?
- Product identification or tracking: Which unit is targeted? When products are close together, how is that unit tracked from detection to the rejection point?
- Reject command and actuation: How are the command and the mechanism’s actual movement observed separately? What demonstrates that reject timing matches product arrival?
- Removal confirmation: What observation demonstrates that the intended unit left the accepted-product stream, rather than merely that the actuator moved?
- Collection and unsuccessful removal: Where does rejected product land? How does that arrangement prevent an unresolved reject from mixing with accepted product, and what response follows unsuccessful removal?
Assess product, packaging, aperture, operating speed, product spacing, challenge placement, site recipe selection and reject timing together. Define site-determined, validated acceptance conditions; do not assume a universal sensitivity or a numerical performance effect from any one variable.
For automatic diversion, propose confirmed capture plus a defined stop or containment response if capture fails. For stop-and-segregate arrangements, propose stopping followed by identification and physical segregation of the affected unit or group. Judge success by demonstrated containment: a stopped conveyor alone does not prove removal.
04Challenge testing: capability, routine checks and acceptance
Three different questions
For a metal detector used as a preventive control subject to the applicable 21 CFR 117 requirements, distinguish capability evidence, routine monitoring and verification. They answer different questions: can the control work under defined conditions, is it being performed consistently, and is monitoring being conducted as required?
FDA’s verification explanation includes scientific validation of process preventive controls as an example of establishing capability to control an identified hazard. This is not a statement that every detector has an identical validation obligation. For site planning, connect capability evidence to defined operating conditions, as in performance qualification, rather than treating one successful challenge as proof for every product and setup.
21 CFR 117.145(a) requires written monitoring procedures, including their frequency. Section 117.145(b) requires monitoring frequently enough to provide assurance that preventive controls are consistently performed; it does not prescribe an hourly, shift-based or batch-based detector challenge schedule. Section 117.145(c)(1) requires monitoring to be documented in records.
Section 117.155(a)(2) separately addresses verification that monitoring is being conducted as required by § 117.140. Operationally, an operator performing a scheduled challenge and a reviewer checking whether required monitoring occurred are different activities. Name each activity’s purpose in the site procedure rather than relying on the word “test.”
Proposed site challenge protocol
The following is a proposed operational protocol, not a regulatory checklist. Test-piece sizes, sensitivity settings and challenge schedules are site-determined and justified through applicable validation, not universal values.
- Identify each test piece and justify its size and material for the challenge objective.
- Specify the product, detector recipe and operating conditions covered by the test.
- Define placements and repetitions before testing so the intended combinations are evaluated.
- Record detection and required physical removal separately for every trial.
- Account for and recover every challenge piece; distinguish normal rejection from manual recovery.
- Predetermine acceptance criteria, including whether every trial must pass both detection and removal.
Worked case: detection passes, rejection fails
Assume a hypothetical approved protocol specifies 3 identified test pieces, 3 placements per piece and 2 repetitions per placement: 3 × 3 × 2 = 18 trials. Its acceptance criterion requires every trial to achieve both detection and physical rejection.
All 18 trials produce detection, but only 17 achieve the required physical rejection. Every test piece is recovered, including manual containment of the unsuccessful reject. The overall challenge fails: recovery accounts for the challenge pieces but does not turn the failed removal into a successful reject. Reporting only “18 detections” would conceal the failure against the stated acceptance criterion.
05An operating sequence from setup to controlled restart
Build the site procedure around explicit decisions, not just test results. Assess setup, product change, scheduled operation, intervention and run completion as possible check triggers; these are procedure-design choices, not mandatory legal events or frequencies. Assign responsibilities locally; the roles below are examples, not universal job-title requirements.
- Select configuration. The operator selects the approved product configuration and confirms that the detector and reject arrangement match the intended run.
- Establish initial evidence. The operator performs and records the procedure’s initial checks; the line lead confirms readiness before production begins.
- Run with defined checks. The operator completes checks at site-defined triggers. The line lead manages line status and carries unresolved matters into the shift handover.
- Investigate exceptions. For a failed or missed check, the line lead stops the affected operation and opens a deviation. An assigned food-safety assessor determines the affected scope and containment.
- Restore and reassess. Maintenance restores equipment; the operator repeats the designated checks. The assessor separately evaluates affected product.
- Document the outcome. The assigned reviewer examines the evidence; authorised people record restart and product-disposition decisions separately, including any continuing restrictions.
| Suggested state | Meaning |
|---|---|
| Operating | Production permitted under the defined checks. |
| Stopped | Affected operation halted; assessment may remain pending. |
| Awaiting assessment | An equipment or product question remains unresolved; retain applicable restrictions. |
| Ready-for-authorisation | Technical recheck accepted; restart awaits approval. Any unresolved product decision remains open. |
Track equipment and product status separately. This sequence illustrates procedure structure; it does not replace the monitoring, verification or disposition requirements discussed elsewhere.
06Failed challenges: contain, assess and decide disposition
Separate correction from corrective action
FDA's Corrections explanation describes timely steps to identify and correct a minor, isolated problem during food production. For a failed reject challenge, do not use a quick equipment repair alone as the basis for classifying the event as minor.
FDA's Corrective actions explanation lists these elements:
- Identify and correct a problem implementing a preventive control.
- Reduce the likelihood the problem will recur.
- Evaluate affected food for safety.
- Prevent that food from entering commerce if you cannot ensure that the affected food is not adulterated.
The same FDA summary states that corrective actions must be documented with records. Use root cause analysis to support the recurrence assessment rather than treating replacement of a failed component as the whole investigation.
Worked case 2: establish a preliminary hold
Assume the last successful challenge was at 09:00, the failed challenge and line stop occur at 09:24, production was constant at 80 packs/minute, and no additional in-flight product falls outside that interval. The preliminary interval contains 24 minutes × 80 packs/minute = 1,920 packs.
This is illustrative containment reasoning, not a universal containment rule. Establish the actual boundary using site evidence: account for uncertainty about when rejection failed, transfers to other locations, and any other affected product. Identify held lots, quantities and locations, and record why the boundary includes or excludes particular product. Keep the hold traceable when product moves.
Restore equipment; decide product disposition separately
As site practice, keep equipment restoration evidence separate from the affected-product assessment. A later passing challenge demonstrates the result of that check; it does not by itself release held product. Neither does an event count. Record the assessment rationale and the authorised person's disposition decision within the site's foreign material control process.
For applicable food preventive controls, 21 CFR 117.155(a)(3) requires verification that appropriate corrective-action decisions are being made as required by § 117.140. That decision-focused verification is distinct from demonstrating that the reject mechanism works again.
- Challenge fails or rejection unconfirmed
- Stop or Contain the Affected Operation
- Establish preliminary hold back to last confirmed pass
- Restore and re-challenge equipment; record evidence separately
- Assess held product: examination, further testing or other disposition basis
- Authorised Disposition Decision with Documented Rationale
- PCQI or oversight review within the applicable timeframe
07Evidence records and the required review
Required review under Part 117
Where these preventive-control requirements apply, 21 CFR 117.165(a)(4) requires review by, or under the oversight of, a preventive controls qualified individual (PCQI). The provision specifies four review purposes:
- Records are complete.
- Recorded activities occurred in accordance with the food safety plan.
- Preventive controls are effective.
- Appropriate corrective-action decisions were made.
For monitoring and corrective-action records, 21 CFR 117.165(a)(4)(i) requires review within 7 working days after creation, or within a reasonable timeframe, provided that the PCQI prepares—or oversees preparation of—a written justification for a timeframe exceeding 7 working days.
Separately, 21 CFR 117.165(a)(4)(ii) requires review of calibration records, testing records (for example, product testing and environmental monitoring), supplier and supply-chain verification records, and other verification records within a reasonable time after creation. Keep these record categories and their review timing distinct.
Suggested evidence schema—not a regulatory field list
The following data-model diagram proposes entities and fields for reconstructing a metal detector challenge. These are recommended record-design choices, separate from the four regulatory review purposes above. Use production context tagging to connect the observation to the actual production run.
As an evidence practice, preserve the original observation when adding a later correction or assessment. Link the addition to that observation and record who made it, when, and why; do not overwrite an initial failure with a later successful result. Keep the physical-rejection result independently readable rather than burying it in a general pass/fail field. A decision reference should lead to the authorised person's documented assessment, while review findings should identify the records examined and any unresolved questions.
Production Run
Identifies the operating context.
- run_id
- batch_id
- line_id
- product_id
- recipe_version
Challenge Test
Separates detection from rejection.
- challenge_id
- run_id
- test_piece_id
- challenge_position
- detected
- reject_confirmed
- performed_at
- operator_id
Reject Event
Links indication, command and confirmation.
- event_id
- run_id
- detected_at
- reject_command_at
- confirmed_at
- collection_id
Hold and disposition
Defines affected product and its decision.
- hold_id
- run_id
- affected_from
- affected_to
- quantity
- quantity_unit
- disposition
- authorised_by
Record review
Links evidence to the review outcome.
- review_id
- record_id
- record_type
- reviewer_id
- reviewed_at
- outcome
08Connect line events to batches and recover exceptions
As integration design practice, assign stable identifiers to production runs, product or recipe versions, challenge tests, reject events, holds and reviews. Use production context tagging to link these records without relying on whichever batch is active when a message arrives. Record occurrence time separately from receipt time, with the originating equipment identity.
In the process event log, model detector indication, reject command and physical rejection confirmation as separate event types linked by a correlation identifier. A command alone should leave removal unconfirmed, not successful. Preserve an explicit unknown outcome when confirmation cannot be matched reliably.
- Communication loss: flag the gap, retain available local events and reconcile missing sequences after reconnection.
- Duplicate messages: use stable event identifiers to prevent double counting while retaining evidence of repeated receipt.
- Delayed confirmation: attach it to the original event; retain the earlier unconfirmed state and record the reconciliation.
- Equipment restart: assign a new session identifier so reused counters cannot collide with earlier events; carry unresolved exceptions forward.
- Clock disagreement: flag the discrepancy and use sequence and correlation data; do not silently rewrite original timestamps.
- Batch change: retain the event’s original context and show every potentially affected batch association until an authorised reviewer resolves the ambiguity.
Retain original events alongside reconciliation decisions, reviewer identity and rationale. Keep unresolved product associations visible during review. Restoring communications should not clear a product hold: hold disposition remains a separate, recorded decision by an authorised person.
09Failure modes and controls to challenge deliberately
Use this table as a suggested challenge and review aid, not a mandatory failure catalogue or prevalence ranking. These hypothetical scenarios can inform a site-specific failure-mode assessment. Keep mechanical performance, procedural execution and missing evidence separate when recording findings.
| Failure scenario | Evidence to inspect | Suggested control or challenge |
|---|---|---|
| Detection without removal | Mechanical: observed product path after detection. | Track a controlled challenge through the actual reject route. |
| Removal of the wrong unit | Timing: product identity against detection and actuation events. | Compare the intended unit with the unit actually collected. |
| Missing confirmation | Evidence: confirmation signal, observation and event record. | Inspect confirmation evidence separately from the reject command. |
| Unsuitable recipe or setup | Procedural: active settings against approved setup. | Reconcile product, recipe revision and physical configuration before testing. |
| Unsecured collection route | Physical control: access, collection point and possible return paths. | Check collection control from diversion through retrieval. |
| Unidentified or unrecovered challenge piece | Traceability: issued piece identity and recovery entry. | Reconcile test-piece identity and recovery after each planned challenge. |
| Missed or incomplete scheduled check | Procedural/evidence: planned check against completed results. | Identify the missing check or unfinished step explicitly. |
Link findings to the relevant batch record, distinguishing an observed equipment failure from an undocumented result. A missing confirmation alone should not be recorded as proof of either successful removal or mechanical failure. Refer unresolved product decisions to the earlier disposition section.
10Use reject trends without confusing them with safety evidence
Worked case 3: Assume 60,000 production packs were inspected and 12 production packs were confirmed rejected. Deliberate challenge packs are excluded from both counts. The reject rate is 12 ÷ 60,000 × 100 = 0.02%, equivalent to 200 rejected packs per million inspected packs.
For operational trending, keep production rejects, challenge failures, unconfirmed reject events and missed checks in separate categories. Label each counted unit explicitly: packs, events or checks. An event count is not a pack count; do not divide a mixed total of alarms and failed checks by inspected packs and label it a pack reject rate. Report challenge failures against challenges performed, separately from production volume.
Review rates alongside product, recipe, operating conditions and maintenance context, using production context tagging to support like-for-like comparisons. Treat this result as a reject-rate measure—not a measured contamination rate or proof that accepted product is safe. Investigate changes in context rather than assigning a universal acceptable reject percentage.
Handle missed or late checks as their own category rather than folding them into pass rates. A monitoring procedure that specifies an interval also needs a defined response when the interval is exceeded: which production interval is potentially affected, who is notified, and what containment applies while the position is established. Record the gap itself, the reason where known, and the assessment outcome. Trending missed checks over time often reveals staffing, access or scheduling problems that reject-rate figures alone would never show, and it gives the site an honest measure of whether the monitoring plan is actually being executed as written.
Finally, resist reading safety conclusions directly from a low reject rate. A low rate is consistent with effective upstream control, but it is also consistent with a detector that has stopped detecting. Pair reject trends with challenge results and maintenance history so that a quiet period is interpreted correctly: confirmed capability with few findings is reassuring; unconfirmed capability with few findings is a gap in evidence, not a success signal.
11Implementation, change and restart checklist
Use this suggested checklist for change and restart decisions. It is an implementation aid, separate from the statutory record-review purposes covered above; software and maintenance preferences do not expand that regulatory list.
- Changed assumptions: after any intervention, list which operating assumptions (product, recipe, speed, spacing, aperture, reject timing) actually changed.
- Reassessment scope: reassess only the functions those assumptions affect; record why untouched functions remain covered by existing evidence.
- Restart evidence: define what demonstrates return to the approved operating state before production resumes, and who authorises it.
- Record linkage: attach the change, reassessment and restart evidence to the affected runs through change control, preserving earlier results.
For proposed changes or maintenance, use change control to ask: Which operating assumptions changed? Which functions need reassessment? What evidence supports return to the approved operating state? Base reassessment on the intervention and its consequences, rather than a blanket revalidation interval.
Revisit HACCP planning and CCP classification if the control’s role needs reconsideration. The FAQs below apply these distinctions to operating scenarios.
Frequently asked questions
Q.What should happen when a production pack enters rejected inventory but no metal fragment is found?+
As an operating practice, keep the pack in rejected inventory while the event is assessed; an unsuccessful search is not a release decision. Record the pack identity, detector settings, reject event and examination method. Use the investigation to distinguish an unexplained detector response from a confirmed contaminant finding. Define any further examination or controlled retesting in the site procedure, and have an authorised person document the disposition rather than returning the pack simply because a repeat pass succeeds.
Q.Can a stopped conveyor count as a successful challenge if the test pack never reached the reject collection point?+
Assess the result against the function being challenged, as established in the equipment section above. A stopped conveyor alone does not demonstrate delivery into the reject collection point for an automatic-diversion function; for a stop-and-hold function, assess whether the pack remained controlled and the specified recovery sequence worked. Record detection, conveyor response, pack location and recovery separately, and link acceptance criteria to the intended function established during operational qualification, rather than recording a generic pass.
Q.How should a passing challenge be assessed when it used a different recipe from the production run?+
Treat the result as evidence for the configuration actually tested, not automatically for the production configuration. As a site practice, compare the selected recipe, relevant detector and reject settings, product presentation and test conditions with the production run. Accept transferability only where the documented qualification rationale supports it; otherwise challenge the correct configuration and assess the intervening production. Record both recipe identities through recipe management, preserving the original result rather than changing its recipe label retrospectively.
Q.Does the CPG's reference to objects below 7 mm justify choosing a 7 mm challenge piece?+
No. FDA’s CPG 555.425, Regulatory Action Guidance, criterion d identifies objects less than 7 mm long where a special-risk group is among the intended consumers as a criterion for recommending legal action. That is enforcement guidance, and the regulatory section above explains why it is not a detector acceptance basis. For foreign material control, determine and validate challenge-piece selection for the site’s product, equipment and hazard analysis.
Q.How can rejected product be traced when two batches share the same collection bin?+
Use a documented bin allocation that preserves each pack’s batch identity, or separate the batches physically before collection. As a traceability practice, record bin identity, contributing batches, collection intervals, quantities and subsequent movements in bin-level lot genealogy. If individual packs cannot be assigned reliably, treat the unresolved contents as a combined assessment population rather than guessing their origin. Have an authorised person document the disposition scope and reconcile the contents when the bin is emptied.
Q.How should a customer's hourly testing specification fit into the site's monitoring procedure?+
Treat the hourly interval as a customer specification, not automatically as a regulatory frequency. As a procedure-design practice, reconcile it with the site’s validated monitoring approach and describe when the hourly clock starts, who performs the challenge, what constitutes completion and how missed tests are handled. Address startup, interruptions and changeovers explicitly rather than assuming the hourly instruction resolves them. Process revisions through change control, with customer commitments and the technical basis recorded separately.
Q.How should a challenge be assessed when the detector clock and the [batch record](/glossary/batch-record) disagree?+
Preserve both original timestamps and investigate the discrepancy before assigning the challenge to a production interval. As a records practice, compare the process event log, pack or batch identifiers, sequence of events and documented clock settings. Record any established offset and its basis without overwriting the original entries. If timing remains uncertain, identify the potentially affected production interval and obtain an authorised assessment; do not choose whichever timestamp makes the challenge appear on time.
Q.How should a metal detector be described in a [HACCP plan](/glossary/haccp) when the site has not designated it a [CCP](/glossary/ccp)?+
Describe its actual role and the site’s classification, applying the classification caution in the regulatory section above. Identify the metal hazard addressed, process location, intended detection-and-reject function, operating checks, failure response and responsible personnel, and reference the hazard-analysis rationale supporting the classification. If the site also maintains a HARPC framework, explain how the detector is represented there so readers can follow the relationship between plans without assuming that their control categories are interchangeable.
Primary sources
Want to see how Metal Detector Reject could fit into your own records and workflows? Explore the related V5 pages or talk to our team about what applies to your operation.
