A batch was rejected at the customer’s incoming inspection and accepted at the factory’s outgoing one. Both inspections were performed, both were recorded, and the two used different limits. The factory had inspected against the figure its line had always used; the buyer had inspected against the specification in the purchase order. The product had not changed. The acceptance criteria had never been agreed.
Charger quality control: where the process breaks and how to see it
Four causes explain most quality disputes on this product, and each produces a recognisable pattern.
Acceptance criteria that were never written down, test limits taken from a nominal figure rather than from characterisation data, records that summarise instead of recording, and sampling that never covers a full shift. Each is a process defect rather than a product defect, and each is confirmed by reading the paperwork rather than by testing the unit again.
Cause 1: acceptance criteria that were never written down
Two inspections, two limits, one argument.
The acceptance criteria are the numbers that decide pass or fail: the output voltage tolerance, the current limit, the no-load behaviour, the temperature rise limit and the visual standard for the finish. Where they live in a purchase order and in a line instruction that were never compared, the disagreement is structural. Confirm it by placing the two documents side by side and asking which limit is binding; where the answer requires a discussion, the criterion is not defined.
Cause 2: a test limit set from a nominal figure
A limit that looks reasonable can be wrong in both directions.
A tolerance drawn from a nominal value rather than from measured characterisation data will pass marginal units and can reject good ones, depending on how the distribution sits. Confirm it by asking for the measurement series the limit was derived from. Where the answer is a datasheet figure, the limit describes the component rather than the product, and the difference shows up as customer returns on a product that passed every inspection.
Cause 3: records that summarise rather than record
A pass rate is not a record.
The record that allows a complaint to be answered is the per-unit or per-lot result against the limit, with the instrument and the date. Where the record is a summary — ninety-eight percent pass — nothing can be traced afterwards, and a defect pattern cannot be seen because the individual values were never kept. Confirm it by asking to see the raw results for the previous shipment rather than the report built from them.
Cause 4: sampling that does not cover the shift
When a sample is taken decides what it catches.
A sample drawn at the start of a shift describes the line when it was warm, calibrated and unhurried. Where a defect is associated with a changeover, the end of a reel or the second shift, a start-of-shift sample will never see it. Confirm it by asking when the samples were taken and on how many shifts; where the answer is one point in time, the sampling plan is a formality rather than a control.
| Pattern | Cause | How to confirm it | What the fix changes |
|---|---|---|---|
| Customer rejects what the factory accepted | Two sets of acceptance criteria | Compare the purchase order with the line instruction | One agreed criteria document |
| Marginal units shipped, returns later | A limit from a nominal figure | Ask for the measurement series behind the limit | Limits set from characterisation data |
| A complaint that cannot be traced | Summarised records | Ask for the raw results for the shipment | Per-unit or per-lot records |
| Defects appearing in clusters | Sampling from one point in the shift | Ask when samples were taken | A plan covering shifts and changeovers |
Setting the acceptance criteria before the first order
Six parameters, agreed in writing, before production starts.
Output voltage and current behaviour across the load range; the multi-port sharing rule where one applies; no-load consumption where the market requires it; the temperature rise limit with the ambient it applies at; the mechanical and finish standard with a retained reference; and the marking and packaging checks. Each is written as a number or a physical reference, because a criterion expressed as “comparable to the sample” changes meaning as the sample wears.
The record pack that makes a reorder repeatable
Four records, requested once and expected with every shipment.
The criteria document, the test results against those limits, the incoming verification records for the critical components and the retained reference samples for the finish and the packaging. With those four, a second order is a repeat of a defined product; without them, every order renegotiates what “acceptable” means, and the negotiation happens after the units are built.
FAQ
Who should write the acceptance criteria?
The buyer states the requirement and the factory confirms that the limits are measurable and achievable, and both sign the same document. A criteria set written by only one side is the reason two inspections can disagree on the same batch. The buyer states the requirement and the factory confirms that the limits are measurable and achievable, and both sign the same document.
Is AQL sampling enough for a charger?
Sampling plans handle visual and mechanical attributes well, and safety-critical and functional parameters are usually tested on every unit. The plan should state which parameters are tested one hundred percent and which are sampled, and over what period the sample is drawn. Sampling plans handle visual and mechanical attributes well, while safety-critical and functional parameters are usually tested on every unit.
How much inspection is too much?
Inspection adds cost and does not improve the product; process control does. Where a parameter is drifting, the fix is in the process or the design, and adding inspection to a drifting process only moves the discovery closer to the customer. Inspection adds cost and does not improve the product, while process control does, so adding inspection to a drifting process only moves the discovery.
What should we do when a batch fails the agreed criteria?
Quarantine it and establish the cause before deciding on rework, sorting or replacement. A sorted batch without a cause will produce the same defect in the next shipment, and the sorting cost repeats. Quarantine the batch and establish the cause before deciding on rework, sorting or replacement, because a sorted batch without a cause produces the same defect again next time.
Can the criteria change during a programme?
They can, and a change is recorded with the reason and the date, because it moves the definition of the product. A limit changed quietly on the line is indistinguishable from a defect when the customer applies the original criterion. Record a change with the reason and the date, because it moves the definition of the product, and a limit changed quietly is indistinguishable from a defect.
Next step
Send the specification, the parameters that matter to your channel and the market requirements to Wecent Group through the contact page. The reply returns the acceptance criteria for the product, the test stages behind them and the records that travel with each shipment. Related reading: our charger factory, OEM and ODM charger manufacturing and charger manufacturing services. External references: IEC standards, IEC Webstore, UL standards and certification, TUV testing services, Intertek testing services and IQNet certification network.
{“@context”:”https://schema.org”,”@type”:”FAQPage”,”mainEntity”:[{“@type”:”Question”,”name”:”Who should write the acceptance criteria?”,”acceptedAnswer”:{“@type”:”Answer”,”text”:”The buyer states the requirement and the factory confirms that the limits are measurable and achievable, and both sign the same document. A criteria set written by only one side is the reason two inspections can disagree on the same batch.”}},{“@type”:”Question”,”name”:”Is AQL sampling enough for a charger?”,”acceptedAnswer”:{“@type”:”Answer”,”text”:”Sampling plans handle visual and mechanical attributes well, and safety-critical and functional parameters are usually tested on every unit. The plan should state which parameters are tested one hundred percent and which are sampled, and over what period the sample is drawn.”}},{“@type”:”Question”,”name”:”How much inspection is too much?”,”acceptedAnswer”:{“@type”:”Answer”,”text”:”Inspection adds cost and does not improve the product; process control does. Where a parameter is drifting, the fix is in the process or the design, and adding inspection to a drifting process only moves the discovery closer to the customer.”}},{“@type”:”Question”,”name”:”What should we do when a batch fails the agreed criteria?”,”acceptedAnswer”:{“@type”:”Answer”,”text”:”Quarantine it and establish the cause before deciding on rework, sorting or replacement. A sorted batch without a cause will produce the same defect in the next shipment, and the sorting cost repeats.”}},{“@type”:”Question”,”name”:”Can the criteria change during a programme?”,”acceptedAnswer”:{“@type”:”Answer”,”text”:”They can, and a change is recorded with the reason and the date, because it moves the definition of the product. A limit changed quietly on the line is indistinguishable from a defect when the customer applies the original criterion.”}}]}