AQL sampling is a statistical tool, and it answers one question: given a batch of a certain size, how many defective units may be found before the batch is rejected. Choosing the level is therefore a statement about acceptable risk, not a formality.

This article maps defect classes to sampling plans for chargers and explains which checks should never be sampled.

What AQL level should a charger programme use?

Tighter for cosmetics, zero for safety.

Critical defects such as safety failures require 100% testing rather than sampling, while major and minor defects are usually sampled at tighter and looser AQL levels respectively.

The convention is to define three defect classes and assign a sampling level to each. Critical defects, which include anything affecting safety or compliance, are not sampled – they are tested on every unit. Major defects, which affect function or make the product unsaleable, attract a tighter level. Minor defects, which affect appearance within acceptable limits, attract a looser level.

For a charger, the electrical and safety checks – functional test, hi-pot pressure test and load aging – sit in the critical class, which is why they are specified as 100% tests rather than as sampling plans.

Mapping defect classes to a charger

Classification is what makes a sampling plan usable. Without an agreed list, two inspectors will classify the same scratch differently and the statistics become meaningless.

Sampling sizes at different batch quantities

Sampling plans scale in steps rather than linearly, so a small batch and a large one may use the same sample size while the acceptance number stays the same. Buyers sometimes mistake this for leniency at volume, when the effect is the opposite: a fixed sample size detects a rising defect rate more sharply as batch size grows.

For pilot quantities, the sample size can approach the batch size, which makes the distinction less relevant; the plan matters most from the first full production run onward.

Which checks are never sampled

Safety and electrical performance checks are never sampled on a charger, because a single failure in those categories is not an acceptable statistical outcome. That includes the functional test across the output range, the insulation and hi-pot test, and the load aging that verifies behaviour under sustained operation.

Where a specification requires a shorter ageing duration on the production line than in the qualification test, that difference should be documented rather than discovered.

Inspection reporting and rejection handling

The report should record the sample size, the defect classification, the defects found and the disposition. A rejection then triggers a defined response: re-inspection of the remaining units, sorting, rework, or a batch hold, depending on the class of defect involved.

Keeping the reports with the batch records turns them into evidence for later claims, which is why the inspection plan and the claim policy should be written at the same time.

Cost of tightening AQL

Tightening a sampling plan increases inspection cost and, in some cases, the rejection rate. The trade is worth making where a defect is expensive in the field: a connector failure in a cable bundle, for example, generates a return that costs more than the inspection that would have caught it.

Conversely, over-tightening a cosmetic plan produces rejections that have no customer impact. The practical approach is to set the cosmetic plan with limit samples, so that acceptable variation is defined physically rather than described in words.

Writing the plan into the programme document

A usable inspection plan names the defect classes with examples, the sampling level for each class, the 100% tests, the reporting format and the disposition rules. With those five items, an inspector can act without seeking clarification.

WECENT’s quality control process covers incoming material verification, ESD-controlled PCBA handling, 100% functional testing, hi-pot pressure testing, load aging and per-batch inspection reports, and programme documentation is available to buyers under NDA through the OEM/ODM programme; the platform range is listed in the product catalogue. Transit performance for packaging can be validated against a published protocol (ISTA), and market documentation follows the destination’s rules (Regulation (EU) 2019/1782 and 47 CFR Part 15).

Programmes in this area are decided by documentation as much as by hardware. The practical discipline is to keep one file per model containing the specification, the test and inspection records and the market documents, and to review it whenever the bill of materials or the production site changes. WECENT produces per-batch inspection records at its quality control gates and provides model-level specifications and certification planning through its OEM/ODM programme, with the platform range listed in the product catalogue.

For inspection planning, the same file answers most questions that arise during qualification, audit or listing review: which standard applies, which model was tested, and which batch the shipped units came from. Reference documents for the relevant requirements are published by the standards and regulatory bodies, including the official text and the related requirement, with the equivalent market rule and https://www.ecfr.gov/current/title-47/chapter-I/subchapter-A/part-15 applying where the product is sold into those markets.

Two further habits keep the file usable over time. The first is to record the review date and the reviewer against each document, so that an outdated report is visible rather than assumed current. The second is to keep the commercial documents – quotation, agreed specification and packaging artwork – in the same folder as the technical evidence, because a question about a shipped unit usually requires both.

A further point concerns the production reality behind the paperwork. Claims that survive scrutiny are the ones a factory can reproduce batch after batch, which means the specification, the test method and the acceptance criteria have to travel together with the order. Where they do not, the same product can be built to two different understandings of the same document.

Finally, buyers should decide in advance what evidence they will need if a question arises later, and request it at the sampling stage rather than after the shipment. Requesting records retrospectively is possible, but it is slower and it removes the opportunity to correct a specification before tooling. WECENT supports programmes with model-level specifications and per-batch inspection records produced at its quality control gates, available under NDA through its OEM/ODM programme, with the platform range listed in its product catalogue catalogue.