Sample Acceptance Checklist
Use this to record what you checked on a sample, and to carry the same checks into batch acceptance. A passing sample proves the sample — write the result down so the batch is judged by the same rules.
A capacity test proves the card holds its claim; it does not prove speed or endurance. See the written guide: Bulk Order Acceptance Checklist.
What each group of checks actually catches
- Capacity — a device that reports more storage than it holds. It passes in File Explorer and fails part-way through a full write, which is why write-and-read-back is the only test that answers the question.
- Performance — the gap between the quoted speed and the measured one, and the gap between burst speed and sustained write. A drive can hit its headline number for a few seconds and still fall over on a continuous write.
- Compatibility — working on the test bench but not in the host device. Interface, power draw and file system are the usual causes, and the host device is the only valid test rig.
- Branding and packaging — drift between the approved sample and the batch, plus packaging that does not match the purchase order. Cheap to check, expensive to discover at the customer.
How to run it in practice
- Test several units, not one. A single good unit proves the unit, not the batch.
- Record the tool and version used for every measurement, so a future batch can be judged the same way.
- Keep the signed checklist with the order file. When a dispute starts, this is the document that decides it.
Where this checklist stops
It verifies what you can observe on arrival: capacity, speed, behaviour in the host device and appearance. It cannot measure endurance, retention over years or failure rate under load. Those are controlled by the NAND grade and the test plan agreed before production, not by an incoming inspection.
Frequently asked questions
Do I need to test every unit in a batch?
No. Sample a batch and record the sampling plan, so the result means something repeatable. Testing every unit is only realistic on small, high-value orders, and even then it verifies capacity rather than endurance.
Which single check matters most?
Capacity verification by write-and-read-back. Misrepresented capacity is the most common defect in the flash market and the one that hurts latest — it surfaces when a customer fills the device, long after you accepted the batch.
What if the sample passes but the batch does not?
That is a specification problem, not a luck problem. The approved sample defines the standard; if the batch misses it, the gap is between what was agreed and what was produced, and the checklist is the record you use to resolve it.