The email arrived eight months after delivery and had four photographs attached. An aluminum enclosure we had machined the previous spring, installed outdoors on a loading dock in Rotterdam, had white crust around six of its eight M6 fasteners. One screw had sheared off when the maintenance technician tried to remove it. The customer’s first question was the obvious one: was the part out of tolerance?
It was not. We pulled the inspection report and every dimension sat inside the drawing, most of them comfortably. The housing had been measured on a CMM before it shipped. Nothing about the part was wrong. Something about the joint was.
What the photographs actually showed
The housing was 6061-T6, anodized, with 304 stainless M6 screws holding a lid. Eight fasteners, six with corrosion product, and the corrosion was not spread evenly across the housing. It sat in rings around each screw hole. That pattern is a tell. Corrosion that hugs a fastener is almost never a coating failure. It is a galvanic couple doing exactly what a galvanic couple does.
We asked for the drawing revision and found the cause in the notes. The customer needed electrical continuity between the lid and the housing for grounding, so the threads had been masked before anodizing. Eight bare aluminum threads, eight stainless screws, salt air, and a wet dock. That is a textbook anode-cathode pair, and the anode is the small one.
The number that explains the eight months
Put 6061 and 304 stainless side by side in the galvanic series and the potential difference lands somewhere near half a volt. Half a volt is not much. What makes it destructive is the area ratio. Galvanic attack concentrates on the anode, and when the anode is a few square millimeters of thread flank while the cathode is a large stainless screw with a wide head, the corrosion current gets squeezed into a very small area. Small anode against large cathode is the worst geometry in the table, and it is exactly what a masked thread produces.
The eight months is worth explaining too, because it is why this failure mode gets missed in testing. The rest of the housing was fine. The anodized surface was doing its job. Only the masked threads were dissolving, and a thread losing twenty microns of flank per year does not announce itself. It announces itself when the torque drops and the fastener either strips or shears, which is what happened on the dock in Rotterdam.
What we should have flagged before we quoted
This is the part of the story that matters to us. The drawing was legal, we built it correctly, and the part still failed. The failure was a design decision nobody in the chain questioned, and we had the information to question it. We had the drawing notes. We saw the masking requirement. We treated it as a coating instruction rather than as an electrical requirement with a corrosion consequence attached.
We changed that. Every outdoor job now gets three questions before we release it to the floor. Does this part live outside or in a wet environment? Are there dissimilar metals in the joint? Is any conductive path exposed to that environment? When the answer to all three is yes, the customer hears from us before the chips fly, not eight months later.
Three fixes, and what they cost
We quoted the fix three ways on a 200-piece run, and all three passed the same 1,000-hour salt spray test.
The cheapest was to keep the masked threads and add a nylon isolation washer under each screw head plus a shoulder washer in the hole. That breaks the conductive path without touching the housing at all. On 200 pieces it added about 0.85 USD per part in our shop, mostly in assembly time. That is a budget estimate for planning, not a firm price.
The middle option was to install stainless Helicoil inserts into the eight holes after anodizing. The insert becomes the cathode instead of the screw, and stainless on stainless in a wet environment is a far less aggressive couple than stainless on bare aluminum. That route added roughly 3.10 USD per part, and it is the one we recommend when the joint has to be opened in the field, because inserts tolerate repeated disassembly much better than bare aluminum threads do.
The most expensive option was to abandon the grounding path through the fasteners entirely and run a dedicated braided ground strap with a sealed lug. It was the only option that added no per-part machining cost, but it changed the customer’s assembly process, and assembly changes cost more than parts do.
They chose the inserts. The revised housings have been in service for a year with no recurrence.
What to put on the drawing
If your part goes outdoors and carries fasteners, write the environment on the drawing. Outdoor, coastal, ten-year service is a sentence that changes how a shop thinks about masking, and it costs you nothing to add. Specify the anodize type explicitly instead of leaving it as plain anodize. Type II at 10 to 12 microns is a cosmetic and mild-wear coating, and it is not the same protection as Type III at 25 microns, which is what salt air usually demands.
If you need conductivity, decide where the current flows and give it a dedicated path rather than borrowing the threads. A masked thread is a design decision that looks free and is not. And if you are already in the field with this problem, do not simply swap to 316 screws and call it done. Changing the cathode material moves the potential a little. It does not change the area ratio, and the area ratio is what is killing you.
Send us the joint, not just the part
We machine single parts and we machine production runs, and the failures that cost our customers the most are almost never dimensional. They are the ones where the part was right and the joint was wrong. Our 6061-T6 aluminum page covers how that alloy behaves in outdoor and marine service, the 304 stainless page covers where that grade should and should not be used, and the anodizing page breaks down coating types and their realistic thickness ranges. If your part is headed into a wet or coastal environment, the thread specification guide goes deeper on insert and thread class selection.
Send the STEP file, the 2D drawing with tolerances, the quantity, and one sentence about where the part will live. If we see a dissimilar metal joint that is going to bite you, we would rather raise it on the drawing than in a warranty claim. Send us the drawing and our team confirms within one business day.