Checkweigher Reject: Meaning, Evidence and Controls
Follow a weight exception from measurement to removal, containment and human disposition while keeping production decisions separate from sampled-lot acceptance.

On this page · 11 sections
- 1What a checkweigher reject means
- 2Gross weight, representative tare and estimated net contents
- 3Regulatory scope and the separate question of lot acceptance
- 4Set operating limits without copying inspection tolerances
- 5Reject-station failure modes and physical controls
- 6Where weighing, rejection and supervisory review fit
- 7Build records that distinguish decisions from outcomes
- 8Challenge the complete detection-to-removal chain
- 9Classify the risk before interpreting reject trends
- 10Contain failed confirmation and recover the affected product
- 11Protect configuration and review readiness before restart
How does Checkweigher 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 a checkweigher reject means
A checkweigher reject is the controlled diversion of a package from the accepted product stream in response to a weight exception identified by an in-line checkweigher.
A configured exception can concern an underweight result, an overweight result or an invalid measurement handled through the reject route. These categories have different meanings and should remain distinguishable in records. A classification or actuator command does not establish successful physical removal.
| Term | Decision or event | Evidence | What it does not establish |
|---|---|---|---|
| Weight measurement | Instrument produces a result | Value, units, validity status | Acceptability or removal |
| Weight exception | Result meets configured exception criteria | Classification and applicable settings | Successful ejection |
| Reject command | Controller requests diversion | Command event | Package left accepted stream |
| Confirmed reject | Physical removal confirmed | Package-correlated reject-path sensing or observation | Final product disposition |
| Containment hold | Potentially affected product restricted | Hold scope, location and status | Every held unit is defective |
| Inspection-lot rejection | Lot fails an inspection assessment | Sampling results and decision criteria | Every package was weighed or ejected |
Package-weight screening is not inherently a product identity test, an assay, a package-integrity inspection or a sampled-lot acceptance decision. A fill weight checker concerns weight assessment; the reject function concerns routing the affected package.
02Gross weight, representative tare and estimated net contents
A checkweigher weighing a sealed package measures gross weight: product plus packaging and anything else carried through the weighing zone. Packaging tare is the weight attributable to the container, closure, label and other packaging components. Representative tare is a value selected to represent those components across a defined population, rather than measured separately for every package.
Estimated net weight = measured gross weight − representative tare.
That calculation estimates product weight; it does not establish each package’s actual tare. If the package’s actual tare is heavier than the representative tare, too little packaging weight is subtracted, making estimated net contents look better than actual contents. Actual packaging tare lighter than the representative tare produces the opposite bias. This distinction underpins tare and net weight calculations.
Changes in bottles, caps, liners, labels or packaging supply can shift tare. Material retained during tare preparation can also change the value obtained. Define the packaging population and preparation method behind the value, and reassess representativeness when either changes; tare-verified weighing addresses this operational distinction.
For US packaged-goods assessment using NIST Handbook 133 guidance, Section 2.3.6.1 calculates nominal gross weight as average tare weight plus labeled weight. For the same US packaged-goods assessment, Section 2.3.6.2 calculates package error by subtracting nominal gross weight from each package’s gross weight.
In an operational uncertainty assessment, keep these contributions distinct:
- Measurement repeatability: variation when the same package is weighed repeatedly under comparable conditions.
- Dynamic weighing effects: influences from speed, spacing, vibration and product movement during transit.
- Tare representativeness: differences between the selected tare and the packaging actually processed.
A decision margin is a site-defined response to uncertainty, not a universal legal allowance. Document its basis rather than treating all variation as scale noise. Sealed-package gross weight alone cannot establish every declared quantity: a mass calculation does not independently establish volume, count, drained contents or active-ingredient content.
03Regulatory scope and the separate question of lot acceptance
Start with legal applicability, then consider technical usefulness. A method can help structure a weight assessment without being the governing acceptance basis for that market. Keep guidance, international recommendations and binding pharmaceutical regulation distinct.
| Jurisdiction and product | Instrument and status | Relevant topic | Boundary for production use |
|---|---|---|---|
| US packaged goods | NIST Handbook 133; legal-metrology guidance | For US packaged goods assessed using NIST guidance, Handbook 133 Section 2.1 describes methods for determining net contents of packages labeled by weight. | Establish national or local applicability separately; do not copy inspection criteria into production reject settings. |
| International prepackage assessment | OIML R87; international recommendation, not universal law | For international prepackage assessment using OIML recommendations, OIML R87 Section 2.2 defines average error as signed individual errors summed and divided by the number of prepackages in the sample. | Identify the applicable market method before using the recommendation as an acceptance basis. |
| US finished pharmaceuticals within Part 211 scope | 21 CFR 211.110; binding regulation | For US finished pharmaceuticals within Part 211 scope, assess in-process-material controls separately from finished-package routing and rejection. | Do not extend this pharmaceutical provision to other product classes or infer a mandated checkweigher. |
As a practical assessment sequence, distinguish:
- Inspection population: the defined group whose acceptance is being assessed.
- Selected sample: the packages examined to inform that assessment, not the entire population.
- Population-average question: what the applicable method concludes about average contents from the sample results.
- Individual short-package question: how that method treats packages with negative quantity errors, separately from the average.
Identify the method before interpreting acceptance. A sampled-lot decision concerns a defined population; a tracked-unit reject decision concerns routing a particular package under configured production limits. Handbook sampling criteria are not automatic production reject settings. Keep those decisions separate in the batch record, including which method or configuration supported each conclusion.
For supplements, food, devices and chemicals, establish the market, product classification, quantity declaration and applicable inspection method before choosing a legal acceptance basis. The Part 211 pharmaceutical framework is not a substitute for that selection. As a control-design boundary, infer neither a universal numerical tolerance nor assured reject performance from these instruments.
04Set operating limits without copying inspection tolerances
As an engineering practice, separate settings by purpose rather than treating every weight value as a reject threshold:
- Labeled quantity: the declared net contents, not the gross package weight.
- Filling target: the process aim, with its gross or estimated-net basis explicitly identified.
- Production acceptance limits: the boundaries used for unit decisions, identified as gross or estimated net.
- Diagnostic alarm limits: thresholds prompting investigation of drift or equipment behavior; distinguish these from unit rejection.
- Inspection criteria: the separately applicable assessment rules, not another name for the production recipe.
What the handbook calculation establishes
For US packaged-goods testing under NIST Handbook 133 guidance, the Section 2.2 scale example concerns packages labeled 113 g and states a maximum allowable variation (MAV) of 7.2 g. In that US packaged-goods example, the calculation is 7.2 g ÷ 6 = 1.2 g. For the same US packaged-goods example, NIST identifies a 1 g scale division as the maximum appropriate division for weighing those packages.
The engineering distinction is resolution versus a production decision boundary. This calculation selects inspection-scale resolution; it does not authorise a line’s lower reject setting. Neither subtracting the MAV from the label nor adopting the scale division as an allowable shortfall establishes a defensible dynamic limit. Production settings need a separate rationale addressing tare variability, measurement uncertainty and actual line performance. Assess measurement capability against tolerance, rather than relying on display resolution alone.
Control the recipe and operating conditions
A practical setup record identifies product and package, representative tare basis, weighing mode, belt conditions and package spacing. Preserve each configured value’s unit, weight basis and configuration version. Approve recipes and manage revisions through change control. Define handling for unstable or invalid measurements instead of silently treating them as acceptable weights. When rejects increase, check recipe selection against the running product before diagnosing a genuine filling deviation: a correctly functioning comparator can still apply the wrong limits.
05Reject-station failure modes and physical controls
Common packaging-line practice treats rejection as a tracked physical transfer. From weigh point to reject point, package identification or positional tracking links the decision to the intended unit. Transport delay, package pitch (spacing), actuator response and the confirmation window need to remain aligned as speed or package behaviour changes.
Define what each sensor observes: a pressure switch indicates available pneumatic pressure; actuator-position feedback indicates mechanism position, not package removal. A reject-route photoeye observes beam interruption, not necessarily the intended package entering collection. Shared sensors or counters can lose attribution when packages overlap or travel close together.
| Failure mode | Observable signal | Control response | Evidence to retain |
|---|---|---|---|
| Missing actuator energy | Low pressure or drive fault | Inhibit feed; contain uncertain units | Energy status; command timing |
| Package jam or bounce | Blocked route or inconsistent passage sequence | Stop transport; inspect collection path | Sensor sequence; recovered packages |
| Package overlap | Abnormal occupancy or spacing | Separate flow; contain ambiguous units | Spacing signals; affected interval |
| Incorrect tracking | Arrival outside expected window | Stop; re-establish package association | Tracking state; speed; event times |
| Obscured confirmation sensor | Persistent beam state or failed challenge | Suspend reliance on confirmation; inspect | Sensor state; challenge result |
| Full or unsecured reject collection | Level indication or access-switch state | Secure collection; prevent further accumulation | Collection status; access events |
| Bypass left active | Bypass status remains enabled | Restrict operation; investigate affected production | User; reason; activation interval |
Suggested workflow: measurement → decision → tracked actuation → package-route confirmation. Missing or ambiguous confirmation → exception alarm → containment of potentially affected units → investigation → authorised human disposition.
The table gives candidate responses, not a universal interlock specification. Line stop, upstream inhibition, guarded collection and alarm escalation are risk-selected design options; coordinate escalation with alarm flood prevention. A mechanism, sensor arrangement or fail-safe label does not establish successful rejection under every operating condition. Installation-specific challenges establish the operating envelope. A machine event is not quality release: final product disposition remains with authorised people through the QA disposition step.
- Approve setupLink product, recipe, tare basis and challenge status.
- Measure packageCapture gross weight, validity, units and tracking identity.
- Apply criteriaRecord the decision against the active configuration version.
- Command RejectAssociate the actuation request with the tracked package.
- Confirm removalRecord removal evidence against the expected confirmation window.
- Contain UncertaintyHold potentially affected units when tracking or confirmation fails.
- Review dispositionReconcile events and product before authorised human restart.
06Where weighing, rejection and supervisory review fit
A common practical arrangement separates measurement, physical control and supervisory records. The weighing instrument produces a weight result and validity information; line control tracks packages and coordinates rejection; supervisory systems associate events with production context and review records. These are practice-based roles, not a prescribed architecture: actual boundaries depend on the installed design.
- Assign an authoritative owner for each state: active recipe, weight decision, reject timing, confirmation state and production-run identity. For example, the instrument may own the decision, the line controller timing and confirmation, and the supervisory system run identity.
- Designate a single configuration master for the active recipe, even when other systems distribute or display it. Document which component detects a failed exchange, initiates the response and records the unresolved state.
Distinguish a physical package identifier, such as a readable serial number, from a temporary tracking identifier assigned at entry. Without serialization, sequence or conveyor position provides attribution only while tracking remains reliable. Manual removal, insertion or lost position can break that association; a sequence number alone does not establish permanent unit traceability. Use production context tagging to retain run and product associations without overstating package identity.
Define message contracts covering event identity, timestamps and clock basis, units, configuration references and acknowledgement meaning. Receipt acknowledges communication, not physical product movement. In MES and SCADA integration, keep enterprise reporting and remote review outside time-critical rejection control. Specify the locally maintained operating state during supervisory communication loss: which configuration remains active, whether local tracking and rejection remain available, and when the line holds rather than continues. Detailed reconnection handling belongs in integration recovery.
07Build records that distinguish decisions from outcomes
A useful checkweigher evidence model links related records rather than flattening each event into a single rejected status. As an operational traceability practice, use a shared event identity and production context to connect:
- Run setup: product and run identity, recipe version, operating limits and tare reference.
- Weight result: raw measured gross weight, measurement validity, unit and resolution; separately identified derived net estimate and decision.
- Reject attempt: command time, confirmation state and associated alarms, linked to the relevant weight result.
- Containment: affected-product references, hold location and the boundary used to identify potentially affected units.
- Human disposition: reviewer identity, decision, rationale and references to supporting checks.
Keep original records distinguishable from annotations, corrections and derived summaries. A later review outcome should not overwrite the original measurement or machine decision. Corrections are more useful when their author, reason and relationship to the original remain visible in the process event log.
For US packaged-goods inspection records, NIST Handbook 133 guidance in Section 2.3.6.2 illustrates recording package error in scale increments. In its example, weighing increments are 0.001 lb and the package error is +0.008 lb. The recorded positive error is 8 increments: 0.008 lb ÷ 0.001 lb = 8.
Importing 8 without its scale factor and unit destroys the intended meaning, even if the event identity survives. Preserve sign, unit, resolution and conversion provenance so reviewers can reconstruct the quantity rather than guess its interpretation. Review missing events and conflicting states, not only completed rejects.
Keep this operational traceability design separate from any Part 11 or Annex 11 applicability assessment. Do not label an event log compliant merely because it contains timestamps or user names.
Run setup
Approved settings for the production run.
- run_id
- sku_id
- recipe_version_id
- tare_version_id
Weight Result
Measured value and its interpreted result.
- event_id
- run_id
- track_id
- measured_at
- gross_weight
- weight_unit
- estimated_net_weight
- decision_code
Reject Attempt
Command and confirmation for that event.
- reject_attempt_id
- event_id
- commanded_at
- confirmation_state
- confirmed_at
Containment Hold
Affected scope and its reconstruction basis.
- hold_id
- run_id
- first_affected_track_id
- last_affected_track_id
- scope_basis
Human Disposition
Authorised decision for the held product.
- disposition_id
- hold_id
- decision
- approved_by
- approved_at
08Challenge the complete detection-to-removal chain
Risk-based engineering practice separates measurement verification from operational challenges. Passing one test does not substitute for the others. A practical operational qualification protocol distinguishes these checks:
- Static measurement: check indication against independently established reference values under stationary conditions where the instrument supports this test. This does not demonstrate performance on moving packages.
- Dynamic weighing: run identifiable packages with reference values through the weighing zone; assess repeatability and differences from reference under intended operating conditions.
- Decision logic: test values below, at and above configured acceptance boundaries, including displayed rounding and invalid measurements. Define the expected decision for each case beforehand.
- Physical rejection: observe whether the intended package reaches the reject collection point, whether confirmation corresponds to that package, and whether adjacent acceptable packages are disturbed or removed.
- Record retrieval: retrieve each challenge event and compare package identity, measurement, decision, confirmation and fault status with independent observations—not merely with the reject command.
Choose conditions that represent intended package formats, operating speeds, spacing, product movement and relevant environmental influences. Record reference units and uncertainty where appropriate. Define acceptance criteria before execution rather than treating whichever response occurs as acceptable.
Include safely simulated station failures identified in the installation assessment, such as loss of reject-actuator supply or missing confirmation where applicable. Check the specified fault response, restoration after an authorised bypass, and recovery from an interrupted challenge. Identify, segregate and reconcile challenge materials so they cannot inadvertently enter saleable production.
Retest scope and routine verification frequency are site decisions informed by intended use, changes, failure history and risk-based validation. Retain the configuration, test conditions, observations, failures, corrective work and authorised acceptance. After corrective work, repeat affected checks and assess whether the change could invalidate previously accepted results.
09Classify the risk before interpreting reject trends
Start with the consequence of the weight exception for the specific product: commercial quantity, a missing component, a dose-related concern, downstream handling or another documented concern. The same checkweigher can serve different purposes on different lines. Record which concern the weighing step addresses and what evidence connects weight to that concern.
A checkweigher is not universally a critical control point, preventive control or safety instrumented function. Classification depends on the product hazard analysis and the role actually assigned to the control, not the equipment name. Treat this as a product-specific risk assessment rather than a machine designation.
Weight is a useful proxy only where its relationship to the attribute of interest is established. A satisfactory package weight alone does not demonstrate active-ingredient content, component identity or absence of contamination. For a proposed link to a critical quality attribute, document both the relationship and conditions under which it could fail.
Keep process-monitoring trends separate from acceptance decisions. Examine the measurement distribution before rejection, including its spread and low-weight tail, and review invalid results alongside reject counts. Removing low results can improve the surviving stream without correcting the filler. A falling reject count can also reflect changed settings rather than improved filling.
For comparable trends, define the numerator and denominator, production period, recipe boundaries and excluded challenge events. Use production context tagging to distinguish these populations. Interpret changes alongside setting changes, tare changes, throughput, downtime and confirmation faults. Route unexplained shifts through out-of-trend handling; investigate the underlying process rather than treating a cleaner accepted-stream chart as proof of improvement.
10Contain failed confirmation and recover the affected product
Consider this illustrative recovery case: the line uses package tracking, the reject route has confirmation sensing, and downstream product remains accessible. A weight exception creates a reject command, but the expected confirmation does not arrive. The recovery approach below is operational practice, with boundaries determined by the installation and available evidence.
Apply the defined protective response and contain potentially affected units, not just the package associated with the weight decision. Work forward from the last trustworthy tracking and confirmation state, accounting for transport positions, controller queues and downstream access. In this case, that could include packages on the discharge conveyor and in the receiving tote. Identify held containers through container status tracking.
If package attribution is unreliable, expand the hold to a defensible physical or production boundary. Missing confirmation can mean unsuccessful removal or unavailable evidence of removal; neither supports assuming safe segregation. Preserve the reject-container contents for investigation rather than emptying them during fault clearance.
Reconcile measurement events, reject attempts, confirmed removals, reject-container contents and held product. These counts need not correspond directly: a package may have repeated measurements, while confirmation signals may not establish package identity. Record uncertainty and unexplained differences instead of forcing totals to match.
Keep controlled reinspection, rework, scrap and continued hold as separate disposition options. A later acceptable reweigh does not itself resolve the original fault or authorise return to production. Maintain the rejected inventory status until an authorised person decides the product’s disposition.
Document the containment basis, investigation, authorised product decision, correction, successful recovery challenge and restart approval. Keep product disposition distinct from equipment restart: restoring the reject function does not decide the fate of held packages.
11Protect configuration and review readiness before restart
The following are industrial integration and controlled-change practices, not universal regulatory requirements. On reconnection, preserve original event identity and distinguish retransmission from a new event. Detect duplicates, gaps and out-of-order delivery in the process event log. Flag unresolved gaps for review; never fabricate missing confirmations or silently replace original records. A restored connection alone does not establish record completeness.
Restrict access to recipes, tare values, operating limits, bypasses and remote maintenance by assigned responsibility. Make changes attributable to a person, with the previous and revised settings, timing and reason retained in an audit trail. Review effects on both physical operation and recorded interpretation: a mapping change can alter the meaning of a reject status without moving any hardware.
Use change control to assess packaging materials, product behaviour, conveyor mechanics, controller logic, firmware, sensors and interface mappings. Select verification work according to the affected functions and failure paths; do not assume every change needs identical testing.
- Confirm that product and active configuration match.
- Confirm that the measurement basis is understood.
- Review challenge results against established acceptance criteria.
- Check reject collection and access controls are ready.
- Keep unresolved product holds controlled.
- Demonstrate predictable interface recovery, including interrupted delivery.
- Obtain restart approval from authorised people.
Keep acceptance of the equipment state, reconciliation of affected product and approval to restart distinguishable in the review record. Equipment readiness does not resolve a product hold; product reconciliation does not establish restart readiness.
Frequently asked questions
Q.How should a tare study account for a change in bottle, closure or packaging supplier?+
Treat the change as a potential shift in both packaging mass and its variability. A practical tare study covers complete packaging assemblies from the new supply, including relevant labels, liners and closures, across representative lots. Compare the distribution, not just the mean, with the existing tare basis. Gross weight minus representative tare estimates net contents; individual packaging differences remain an uncertainty. Through change control, assess whether mixed packaging stocks, revised settings or additional verification need attention before routine use.
Q.Can a passing inspection sample justify returning previously rejected packages to the line?+
Not by itself. A passing sample describes the population and acceptance criteria it was designed to assess; it does not resolve each previously rejected package’s exception. Keep rejected units identifiable in a rejected inventory bucket while reviewing their original rejection reasons. Any proposed recovery should use an authorized assessment suitable for those reasons, with traceable results and protection against mixing unresolved units into accepted output. The quality disposition decision belongs to authorized personnel, not to the sampling result alone.
Q.What should an operator do when a rejected package passes a later bench-scale reweigh?+
Keep the package segregated and retain both results rather than replacing the original failure with the later pass. Check package identity, whether either result represents gross or estimated net weight, the tare basis, and any leakage or material change between measurements. Then investigate dynamic influences such as spacing, vibration and conveyor transfer, alongside bench-scale suitability. Use the site’s exception-handling process to document the discrepancy. Repeated weighing until a passing value appears is not a sound basis for disposition.
Q.How should the investigation proceed when a package is found in the reject bin but its confirmation event is missing?+
Separate physical recovery from proof that the intended rejection sequence worked. A package in the bin may have arrived through successful ejection, manual placement or another reject route. Preserve its identity and examine the process event log, tracking sequence, sensor condition and available operator observations. Contain potentially affected units across the uncertain interval, including downstream output where appropriate. Record the issue as a deviation; finding a package in the bin does not establish that every intended reject was removed.
Q.When is an overweight package a quality concern rather than simply excess giveaway?+
Overweight becomes a quality concern when the excess could indicate the wrong contents, an extra component, an incorrect concentration, inadequate headspace or interference with package function. First distinguish heavier packaging from additional product. For filled containers, investigate whether excess fill affects closure application or seal integrity; for kits, investigate unintended duplicate components. Assess consequences against the product’s intended use and specifications. Weight alone neither establishes potency nor assigns a universal safety-control classification; the product-specific failure mechanism determines the significance.
Q.Can package weight reliably detect a missing tablet, sachet or device component?+
Only when the missing item produces a mass difference distinguishable from normal product variation, packaging variation and measurement uncertainty under operating conditions. Evaluate complete and deliberately incomplete packages across representative materials, speeds and orientations. If their measured distributions overlap, some incomplete packages can resemble complete ones; heavier packaging or another extra item can also mask the loss. A measurement-system study helps characterize discrimination, but weight does not establish component identity. Consider complementary counting or presence checks when mass separation is inadequate.
Q.Does a lower reject rate demonstrate that the filling process has improved?+
No. Fewer recorded rejects can result from wider limits, a shifted tare value, disabled detection, missing records or a changed product mix rather than better filling. Compare runs using consistent settings, populations and counting rules. Examine the weight distribution before rejection, its position relative to the target, and independent fill checks. An EWMA control chart can help reveal drift hidden by a headline rate. Also review giveaway and verify reject-system performance before attributing the reduction to process improvement.
Q.How should a run be reviewed when physical rejection continued but the supervisory event connection was interrupted?+
Treat this as an event-reconstruction problem without assuming either that rejection failed or that the run remained fully controlled. Preserve available local records, counters, alarm history and settings. Reconcile their sequence with the batch execution history, accounting for clock differences, buffered messages, duplicates and missing events. Distinguish rejection commands from physical confirmation. Review the supervisory interface recovery before relying on restored totals. Where local evidence cannot establish control during the interruption, contain the uncertain output for authorized assessment.
Primary sources
Want to see how Checkweigher 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.
