Why a waterproof connector leaks when it is fitted too tightly

Panel mount torque, and why the correct figure is small and specific.

A waterproof connector leaks when it is fitted too tightly because the rubber seal doing the actual sealing has a working range of compression, and forcing it well past that range does not seal it more firmly, it deforms the seal out of the shape and the position it needs to be in to seal at all.

Why tighter is not automatically safer

The instinct to keep turning a fitting until it feels properly tight is a reasonable one for most ordinary hardware, since a loose bolt is a genuinely common and genuinely serious failure, and most people's experience with fastening things teaches that more torque means more security right up until the point something actually breaks. A rubber seal behaves completely differently from a plain metal-on-metal joint, though, because its sealing action depends on being compressed by a specific, fairly narrow amount, not simply squeezed as hard as possible. Compress it too little and it never fills the gap between the two surfaces it is meant to bridge, leaving a path water can still find. Compress it too much, past the point the material and the groove holding it were designed for, and the seal can be forced to extrude sideways out of its own groove, thin unevenly across its face, or simply take a permanent compression set that leaves it unable to spring back and re-seal itself once the load that created it changes even slightly, which an outdoor connector's temperature swings and vibration guarantee it eventually will.

The screw-top-bottle comparison

A screw-top water bottle with a rubber gasket seals perfectly well tightened only to a firm finger-tight pressure, and most people, worried about it leaking in a bag, give the cap one or two extra determined turns beyond that for good measure. Do that too many times, or too hard, and the gasket can be felt to change: it stops sitting flat and starts to feel slightly pinched or rolled at one edge, and the bottle that sealed reliably at a gentle tightness can begin weeping at exactly the seam the extra turns were meant to make safer. Nothing about the cap loosened; the seal itself was physically deformed by pressure well beyond what it needed, and the deformation, once it has happened, does not reverse itself just because the cap is backed off again afterward. A panel-mounted waterproof connector's gland or nut is doing precisely the same job as that bottle cap, and the same over-eager extra turn that seems like careful practice on a water bottle is exactly what damages the seal on a connector installed the same way.

Why the seal fails from the inside of its own groove

An O-ring or a gasket seals by being squeezed into intimate contact with both of the surfaces it sits between, and that squeeze is normally engineered to a compression of only a modest fraction of the seal's own relaxed thickness, enough to guarantee full contact without asking the material to absorb more strain than it can recover from. A groove is cut to hold the seal at exactly that intended compression once the two halves of the joint are brought together to their designed final position, and the torque specified for the fastener is really just the number that reliably brings the joint to that position, no more and no less. Over-tightening past that point does not increase the squeeze in any useful way, because once the two mating faces are already touching metal to metal, or plastic to plastic, around the seal, further turning stops compressing the seal at all and instead starts loading the fastener itself, the housing, or the seal's trapped edges well beyond their design intent, which is exactly where extrusion and permanent set begin.

The one number worth remembering

The correct tightening torque for a small waterproofing gland or panel-mount connector is very often only a couple of newton-metres, an amount that can feel almost suspiciously light to a hand used to tightening ordinary hardware, and easy to exceed without noticing by simply continuing to turn the fitting until it merely feels finished rather than stopping at the specific figure the manufacturer actually specified.

What this changes in practice

Specifying and actually achieving the correct torque are two different problems, and the second one is where this most often goes wrong in the field rather than in the original design. A datasheet can state the exact figure a connector's gland nut should be tightened to, and that figure still has to survive being read, remembered, and applied correctly by whoever is doing the physical installation, often months or years later and often without the datasheet anywhere nearby. Some connectors solve this by building a torque-limiting feature directly into the fitting itself, a nut that audibly clicks or slips once the correct compression is reached regardless of how much further force is applied to it, which removes the need for the installer to trust their own sense of feel or to have a calibrated torque wrench to hand at all. Where that feature is not available, the only real substitute is discipline: a torque wrench actually used rather than estimated, and the specified figure treated as a hard limit rather than a rough guide to be exceeded whenever a fitting looks like it could plausibly take a little more.

What this does not explain

None of this means loose is ever the safer default, since an under-tightened connector fails for the opposite reason, never reaching the compression its seal needs in the first place, and the two failure modes can look almost identical from the outside, a connector that leaks despite appearing to be properly fitted. The only reliable way to tell them apart, and the only real defence against either one, is to stop guessing at feel entirely and torque the fitting to the specific figure its manufacturer actually published, since feel is precisely the sense this particular joint was never designed to be judged by. A field technician diagnosing a leaking enclosure is often better served by checking the gland's actual tightness against its published figure before assuming the seal itself has failed, since a connector re-fitted slightly too enthusiastically during an earlier repair is a genuinely common cause of exactly the fault it was tightened in an attempt to prevent.