The cert for a finished lot, drafted from your records and held against the PO.

When a lot finishes, the proof that it was done right is scattered: gauge readings or hardness checks, a tank log or furnace record, perhaps a salt spray panel or a coating test, and the customer's PO saying which of those it wants to see. Quality staff assemble that into a certificate naming the spec, the results and the lot, and their signature goes on it.

In the test, AI reads copies of those records for lots you already shipped and drafts the certificate in your layout, sets every reading beside the range it was held against, notes the method used, and lists what the order required but the records don't show. It never marks a lot as passing, and nothing goes to a customer or onto a live certificate.

A salt spray result needs a definition of failure. ASTM B117 tells you how to run the cabinet, and its scope leaves the interpretation of results to someone else, while warning that reproducibility depends heavily on the specimens and the evaluation criteria chosen. So a certificate reading "salt spray per B117" says how the cabinet was operated and nothing about what counted as a failure. The hours and the failure point must come from the order, the customer's spec or your own test record; the draft cites where it found each, and lists it as a gap when it can't.

Adhesion has two methods and a strict scale. ASTM D3359 rates adhesion by tape test, with Method A aimed at coating systems over 125 micrometers (5 mils) in total and Method B at thinner ones, and it tells users to establish the film thickness first if it isn't known, with a practice such as D7091. Ranks run 0 to 5 with no values in between, and the test doesn't separate higher levels of adhesion from each other. It also isn't technically equivalent to ISO 2409, so a record citing one doesn't quietly answer a PO asking for the other. The draft compares the method in your record with the film thickness and flags a 4.5 if one shows up.

A thickness number needs its method beside it. Zinc over steel is commonly read magnetically under B499, anodize on aluminum by eddy current under B244, and a cross section under B487 means cutting the part. The X-ray method in B568 measures the mass of coating per unit area and yields a thickness only when the coating's density is known. Each reading on the draft carries its method and the units it was taken in, and one with no method behind it goes on the list for quality staff.

Hardness scales don't convert cleanly. ASTM E140's tables relate Brinell, Vickers, Rockwell and other scales, but the standard calls conversion only an approximate process, to be used when the material can't be tested under the conditions specified. A print in Rockwell C and a certificate reporting Brinell gets called out, even where a table might put the number in range. A furnace chart, a quench record and a hardness check each prove something different too, so the draft won't let one stand in for another.

Public buyers spell out their certificates. Iowa DOT's Materials IM 568, for steel products powder coated for its projects, has an authorized representative of an approved shop certify the coating and that the powder came from approved sources, with a copy going to the project engineer and the District Materials Engineer at shipment. The same memo puts powder film at 3.0 to 4.0 mils unless otherwise specified and checks cure by solvent rub. Those terms govern Iowa's work only, and they show how one buyer's written rule becomes fields on a certificate. Which spec governs your lot stays with your quality staff; nothing on this page is legal or engineering advice.

With your reviewer's yes, certificates get drafted on newly finished lots for about 30 days.

Quality staff review, correct and sign each one before it goes anywhere. The drafts never stand in for their judgment, and scoring stays on through the month.