Powder Coating Subcontracting in China: Pre-Treatment, Film Thickness and the Salt Spray Number That Means Nothing

The first time I quoted a powder coating line in Ningbo, I did the thing most buyers do: sent a drawing, asked for a price per square metre, and compared three suppliers. The quotes came back at 6.8, 7.4 and 9.2 yuan. I picked the middle one because I suspected the cheap one was cutting corners and the expensive one was padding. Six weeks later, a third of the brackets inside the cartons had chipped edges and two hinges looked like they had been painted with a brush.

What I had actually bought, without knowing it, was not a coating. It was a process window. The supplier who quoted 7.4 hit it. The one who quoted 6.8 skipped one rinse stage and made the line move faster. The one who quoted 9.2 used a better resin and could have justified it. Price per square metre told me nothing about any of that.

Since then I quote by process, not by number. Here is what that looks like in practice.

Pre-treatment decides the next three years

Powder coating has no solvent to carry away contamination. Whatever sits on the steel when the powder goes on is trapped under the film forever. That is why the pre-treatment line matters more than the powder brand, and it is also the part a factory is most tempted to shorten when they are short of capacity.

A proper multi-stage line for steel goes through degreasing, rinse, derusting (or a mechanical blast before that), rinse again, a conversion coating, a final rinse, and then a drying oven that has to be genuinely dry, not just warm. Conversion coating is usually zinc phosphating, or a nano-ceramic zirconium-based film on newer lines. Both work. What does not work is a two-stage "wash and go" line, which is common on small job shops and almost never shows up in a quote.

Ask how many stages the line has and ask for a photo of the tunnel. If the answer is three, walk. You can also read it later on the parts: poor phosphating shows up as rust creeping out from a scratch or a cut edge within a few hundred hours of salt spray, even when the coating itself still looks fine.

The film thickness number matters less than the distribution

Everybody specifies 60–80 µm now, because that range is what the standards hint at. What nobody specifies is where the measurement is taken. Coating on a flat face pools thick. Coating on an edge, a thread, a weld seam or inside a corner thins out, and those are exactly the places that rust first.

I ask for a film thickness map: five readings per part, at least two of them on an edge or a corner, on the first article and again on a part pulled from the middle of the run. Cheaper suppliers will hand you five readings from the flat centre of the same part, and it will read 72 µm every time. If the same part also has an edge reading of 38 µm, you now know something useful.

Weld seams and laser-cut edges are the classic failure points. If you need corrosion resistance on cut edges, either specify a zinc-rich primer under the powder or move to a different substrate. Powder alone will not seal a sheared edge, no matter how thick you make it.

Cure window, and why the oven is the real production limit

Most architectural powders cure at 180 °C or 200 °C for 10 to 15 minutes at metal temperature. Metal temperature, not oven air temperature. If the part is thick, or the basket is dense, the oven air might be at 200 °C while the middle of a heavy casting sits at 150 °C for the first six minutes. Under-cured powder looks perfect and behaves badly: it chalks, it loses gloss in a season outdoors, it scratches like chalk.

The check is a solvent rub on a cured part from the middle of the load, not from the first one in. MEK or acetone on a rag, twenty double rubs. Fully cured powder barely softens. Under-cured powder comes off on the rag. It takes two minutes and has saved me two complaints.

Over-curing is the other direction and it is just as bad — yellowing on whites and light greys, brittleness on the film, and a colour drift you cannot fix later. If a factory runs a mixed load with thin and thick parts in the same basket, you will get both problems in one batch.

Salt spray: read the test report, not the number on the quote

"500 hours salt spray" appears on almost every Chinese powder coating quote and it is close to meaningless on its own. A salt spray result depends on which powder, which substrate, which pre-treatment, the film thickness, and whether the part was scribed. A scribed panel tested to ASTM B117 or ISO 9227 will fail earlier than an unscribed one, always.

What you want in the report is: the standard used, whether the panel was scribed, the average and maximum creep from the scribe in millimetres, the blister rating, and the substrate and pre-treatment that were actually used. Compare that against what the factory is actually running. A report from a powder manufacturer's lab on their best pre-treatment line is not evidence about a job shop's three-stage wash.

My working thresholds for outdoor steel parts in a mild European climate: unscribed 500 hours with less than 2 mm creep, scribed 480 hours with creep under 3 mm and no blistering at the scribe. For a coastal installation, double the hours. For anything you cannot easily replace, put a zinc-rich primer underneath rather than asking powder to do a job it was not designed for.

Colour consistency across runs is a batch problem, not a colour problem

Powder is bought in batches. Two batches of the same RAL number from the same manufacturer will be close but not identical, and on a satin or textured finish the difference is visible side by side. If your product ships in sets — a matching door handle and hinge, a bench frame and legs — colour drift between batches is a returns problem.

The practical fix is boring and it works: settle the colour with a physical approved sample, keep your own retained panel from the first approved run, and ask the factory to declare the powder batch number on the inspection record. When they change batches, you get told. If you run a seasonal product, buy the whole season's powder up front and have the factory store it — that is often cheaper than re-shooting a batch.

Judge colour under D65 light with a gloss meter or a spectrophotometer if you have volume, and under a plain window with your eyes if you do not. Metallic and textured finishes cannot be measured reliably by eye and half of the "your colour is wrong" arguments I have seen were metallic finishes compared under different lights.

What I actually put on the purchase order

These days the PO line for a powder coated steel part reads roughly like this, and it costs me nothing but was the single biggest quality improvement I made in this category:

• Pre-treatment: multi-stage with conversion coating, minimum stage count stated, no substitution without written approval
• Film thickness: 60–80 µm nominal, minimum 50 µm at any measured point including edges, five-point measurement per part, first article and mid-run audit
• Cure: metal temperature 180 °C or 200 °C per powder TDS, solvent rub test per batch
• Adhesion: ISO 2409 cross-cut, Gt0–Gt1 on the production parts, not on a lab panel
• Corrosion: ISO 9227 NSS, hours and scribed/unscribed stated, report submitted with first article
• Colour: approved physical sample retained by both sides, gloss range stated, powder batch number on the inspection record
• Impact and handling: parts packed with edge protection, no bulk dumping into cartons, chipped parts counted as defect not as "acceptable"

That list will lose you a supplier or two when you first send it out. The ones who sit down and answer each line with a number or a document are the ones worth keeping. The ones who reply "yes, all standard, no problem" and then send a three-stage line's output are the ones who cost you a container of rework in March.

One last thing that took me longer to learn than it should have: powder coating is a service, and services are bought on process documentation. Nobody in this industry will volunteer that their line has four stages instead of seven. Ask. It is a normal question, and the good factories are not offended by it — they are usually a little relieved that someone asked.