The enclosure that leaked at the cable and not the lid

Sealing everything except the part that actually mattered.

The enclosure leaked at the cable entry rather than the lid because every effort at controlling seal quality had gone into the O-ring around the lid, the part that looked, by every visible measure, like the obvious place a housing would fail, while the cable entry, a different sealing method entirely, was trusted on the strength of being a bought, off-the-shelf component rather than checked with anything like the same scrutiny. The lid seal, by the time the leak was finally traced, had already been checked so thoroughly that it was the very last place anyone thought to look.

What is really going on

An enclosure's lid is the joint everyone naturally focuses attention on, since it is large, visible, and clearly load-bearing, the exact kind of feature a careful design review circles back to repeatedly, while a cable entry is small, often supplied as a finished, pre-rated component, and easy to treat as a solved problem the moment it is bolted or screwed into place. That difference in visible importance has nothing to do with the actual risk each joint carries, and a cable entry relies on its own separate sealing mechanism, typically a compression gland squeezing directly onto the cable's outer jacket, a mechanism with its own failure modes entirely distinct from an O-ring's, and one that depends heavily on the cable's actual diameter, jacket material, and how correctly the gland itself was tightened, none of which is guaranteed simply by the component being commercially rated for the job. A gland rated for a cable within a certain diameter range still needs that cable's jacket to be round, smooth, and consistent along the exact stretch being clamped, and a jacket with even a small manufacturing rib, seam, or print marking sitting right where the gland squeezes can open a leak path a purely visual inspection of the finished assembly would never catch. None of that is a defect in the gland as a product, a correctly rated gland clamped onto a genuinely round, correctly sized cable seals reliably, the actual risk sits entirely in the gap between the cable the rating assumed and the cable that was actually threaded through it on the day the enclosure was built.

The weakest-link comparison

A chain is exactly as strong as its weakest link regardless of how impressively over-built every other link happens to be, and a team polishing and reinforcing nine links out of ten while leaving the tenth exactly as delivered has spent real effort without moving the chain's actual breaking point at all, since the chain will always part at whatever link is weakest, not at whichever one received the most attention. An enclosure's overall watertightness works the same way, a leak occurs at whichever sealed joint is weakest, and that joint is under no obligation to be the one that looked most important or received the most design scrutiny, which is precisely why the cable entry, quietly under-examined the whole time, turned out to be the actual weak link in a housing whose lid seal had been checked and rechecked. Every hour spent refining the lid seal beyond what it already needed was, in a real sense, wasted relative to the housing's actual watertightness, since none of that extra polish moved the point at which the housing was actually going to fail. A chain inspector who spends the whole afternoon admiring nine well-forged links has learned nothing at all about whether the chain will hold, since the only link whose condition actually matters is the one nobody has looked at yet, and the same blind spot is what let a cable gland sit unexamined for as long as the lid seal kept absorbing everyone's attention instead.

What this changes in practice

Because a housing's overall watertightness is set by its weakest sealed joint rather than by the average quality of all its joints combined, a proper review has to walk every single sealed interface with equal seriousness, however small, however seemingly minor, or however confidently a supplier's rating suggests the component can simply be trusted, rather than concentrating scrutiny on whichever joint happens to look most structurally significant. This means treating a bought, pre-rated component's rating as a starting assumption to be verified under the specific conditions it will actually be used in, rather than as a finished guarantee, since a cable gland rated for a certain depth and a certain cable diameter can still leak if the actual cable used differs even slightly from what the rating assumed. A rating printed on a datasheet describes what the component achieved under the conditions the manufacturer tested it against, and it is only ever as trustworthy as how closely the actual application matches those original test conditions, a match that has to be checked rather than assumed.

Where this stops being true

Not every joint on a housing genuinely carries equal risk, and a sensible review still has to prioritise, since spending unlimited scrutiny on every fastener and every seam is neither practical nor a good use of limited time, the actual lesson is not that every joint deserves identical attention regardless of context, it is that the attention paid to each joint should be set by an honest assessment of its actual risk of failure, rather than by how visually prominent or structurally impressive that joint happens to look. A cable entry looks minor and is, in fact, frequently a genuine risk, which is exactly the mismatch between appearance and actual risk this whole article is about, and correcting that mismatch, rather than eliminating the need to prioritise at all, is what an honest review process actually has to achieve. The final article in this set turns from this specific failure back to a more general question the whole set has been circling, why the pressure trying to break a seal open can, understood correctly, be turned into the very thing helping it stay closed.

My key error with this

Almost all of my attention during that design went to the main housing seal, because a large lid with a captive o-ring is the part that looks like it might let water in, and I checked it, calculated it and pressure tested it accordingly. The cable penetrations I treated as catalogue items, since a gland is a bought part with a published rating, and I selected them by cable diameter and moved on. What I had not accounted for is that some of the cables we were running were unusually large and unusually stiff, so the gland was being asked to seal around a bundle that resisted being compressed evenly, and the mismatch left a path that no amount of tightening ever closed properly. Water came in through a fitting I had never really designed, next to a lid I had designed twice over. What replaced the belief is that a penetration is a seal in exactly the same sense the lid is, that a bought part with a rating still has to be checked against the specific thing being passed through it, and that the fitting nobody thought about is reliably the one that leaks.