Count how many times the connector is unmated over the life of the product, then double it. Under 20, tin is fine. Between 20 and 100, specify 10 µin gold. Above 100, or in a test fixture, specify 30 µin gold and treat the mating half as a consumable.
That is the whole decision, and it is made with one number that most drawings leave blank. Instead the drawing says "gold plated" or nothing, and the part that arrives is whatever the shop had in the plating line that week.
What actually wears out
The contact spring and the plating wear together, but they fail differently. The spring relaxes slowly, which raises contact resistance gradually. The plating wears through locally, which exposes the base metal and lets it oxidise, which raises resistance quickly.
The visible damage is a polished patch on the pin where the socket rides. On a tin contact that patch turns dull and then dark. On gold it stays bright until the gold is gone, then fails suddenly.
| Plating | Cycles before resistance change | Failure mode |
|---|---|---|
| Tin, 100 µin | 20–50 | Fretting corrosion, dark patch, rising mΩ |
| Gold flash | 10–20 | Pore corrosion through the thin layer |
| Gold 10 µin | 50–100 | Gold worn through at the contact line |
| Gold 30 µin | 100–500 | Spring relaxation before plating wear |
Why the number on the datasheet is optimistic
Cycle ratings are measured in clean air, at room temperature, with a straight insertion and no side load. Real assemblies insert at an angle, in a cabinet at 50 °C, with dust on the connector. Each of those cuts the number.
We quote the clean-air number because that is what the industry quotes, and we tell customers to plan on half of it if the connector is in a machine shop or an outdoor cabinet. That is not pessimism, it is the difference between a warranty claim and a service visit.
The service-interval test
Take the number of times a technician unplugs the connector per year, multiply by the warranty years, then add the factory test cycles. If the total is over 15, gold flash is already marginal.
- Machine controller, unplugged for board replacement twice in ten years: tin is fine.
- Test fixture, mated every shift: 30 µin gold, replace the harness yearly.
- Field-serviceable module, plugged in on site: 10 µin minimum, and specify a mating half with matching gold.
Mismatched plating is worse than thin plating
Gold on one side and tin on the other is a known bad combination. The tin side transfers material to the gold side, and the transfer layer oxidises. The connection degrades faster than tin-to-tin.
Specify the pair together. If the board header is 10 µin gold, the cable socket should be gold too, at a similar thickness. This is the most common plating mistake we see on customer drawings, and it is free to fix.
Does a mating cycle include the factory test?
Usually not. Datasheet cycles count field mating. Add one or two for the test fixture at the end of the line, and one more if the unit is retested after repair.
Can I re-tin a worn contact?
No. Replating a contact at the bench destroys the spring temper. Replace the connector or replace the cable assembly.
Is 30 µin gold worth it on a $0.30 header?
Only if the header is mated more than 100 times. Below that you are buying insurance against a failure that will not happen. See how the three thicknesses compare.
Tell us the service interval and the environment and we will quote the plating that matches it, on either half of the pair. Cable-side and board-side parts are both in the box header and IDC section.