← Journal Design · · 7 min

IP67 enclosure design: seals, vents and leak testing

IP67 enclosure design comes down to gland geometry, stiff lids, venting and early leak tests. Here is what we check before a sealed device goes to a lab.

Milo Perović
Founder & lead engineer

  • DFM
  • Enclosures
  • Environmental
  • Manufacturing
IP67 enclosure design: seals, vents and leak testing

Most sealed products leak for boring reasons: a gland that is slightly too deep, a lid that bows between screws, a USB port nobody thought about until the prototype came out of the water tank wet. Good IP67 enclosure design is less about exotic materials and more about getting a handful of geometric and process decisions right before the first mould is cut.

This post is for founders and engineers building connected devices that live outdoors, in vehicles or anywhere near a hose. It covers what the rating actually asks for, how we design the seal, why venting matters more than people expect, and how we test a small batch before paying for a lab.

What IP67 actually asks of an enclosure

The IP code comes from IEC 60529. The first digit covers solids and the second covers water. A 6 for solids means dust-tight: no ingress of dust after a chamber test, usually run with the enclosure under slight vacuum. A 7 for water means temporary immersion, with the lowest point of the product at 1 m depth and the top at least 0.15 m below the surface, for 30 minutes.

Two things follow from that definition that catch teams out:

  • IP67 is not a higher version of IP65 or IP66. Immersion and water jets are different tests. A product can pass 30 minutes in a tank and still let a pressure washer force water past a lip seal. If your product will be hosed down, it needs to be designed and tested for jets as well, and labelled with both codes.
  • The rating applies to the product as sold. Every opening counts: connectors, buttons, the battery door, the light pipe and the screw holes. A sealed housing with an unsealed charging port is not an IP67 product.

Choosing a seal strategy for IP67 enclosure design

There are four common ways to close a two-part housing, and each pulls the rest of the design in a different direction.

ApproachStrengthsWatch out for
O-ring or cord in a grooveCheap, predictable, serviceableNeeds accurate gland geometry and a stiff flange
Form-in-place gasketFollows complex outlines, no loose parts in assemblyDispensing equipment, cure time, bead consistency
Overmoulded or two-shot TPE sealSeal cannot be lost or misassembledHigher tooling cost, material bonding compatibility
Ultrasonic or laser weldingPermanent, no seal to ageNo servicing, batteries sealed in for life, joint design must suit the process

For prototypes and small batches we usually start with a cord or O-ring in a groove, because it is the easiest to iterate. Moving to an overmoulded seal later is then a tooling change rather than a redesign.

Gland geometry

For a static face seal, the groove depth sets how much the elastomer is squeezed. Seal manufacturers’ design guides typically recommend somewhere around 15–30% compression for static seals, with the groove volume large enough that the seal fills well under 100% of it at maximum tolerance and temperature. Too little squeeze and the seal leaks as it relaxes; too much and the lid will not close evenly, or the seal takes a permanent set.

The trap is tolerance stacking. Moulded parts shrink, warp and vary between cavities, and a groove that looks perfect in CAD can swing several tenths of a millimetre on real parts. We run the stack-up at both extremes and check that compression stays within range at each, rather than designing to nominal.

Clamping and lid stiffness

A seal is only as good as the force holding it. Plastic lids bow between screws, and the gap is largest exactly halfway between fasteners. A thicker flange, ribs along the seal path and closer screw spacing all help. How close depends on the lid material, wall thickness and seal stiffness, so we check flange deflection in simulation rather than relying on a rule of thumb, then confirm it on printed or machined parts.

Material choices for seals and housings

The elastomer has to match the environment, not just the datasheet hardness:

  • Silicone (VMQ): very wide temperature range and good compression set, but weak in tear and abrasion. Good for static seals that are rarely disturbed.
  • EPDM: excellent resistance to water, ozone and weathering. Poor with oils and fuels, so a bad fit for anything near engines.
  • NBR (nitrile): good with oils and fuels, but degrades in ozone and sunlight unless protected.

The housing matters too. PC and PC/ABS are common for outdoor products, but glossy surfaces show sink marks at thick sections and UV stabilisation is not optional for sun exposure. We covered how wall thickness and draft drive these decisions in designing for injection moulding before you have a mould budget.

Openings: connectors, buttons and sound

Every hole is a separate sealing problem, and first-prototype leaks come from here at least as often as from the main lid seal.

  1. Connectors. Use parts that are rated for sealing when mated and, if needed, when unmated. Panel-mount sealed USB-C receptacles exist, but they rely on their own gasket against the housing, so the panel surface must be flat and the mounting stiff. Cable glands need the right cable diameter range; a gland clamping a cable at the edge of its range is a slow leak.
  2. Buttons. A moulded silicone keypad over tactile switches, a sealed membrane, or a capacitive sensor reading through a continuous wall. The last one removes the opening entirely, at the cost of firmware tuning and some sensitivity to water on the surface.
  3. Sound and air. Buzzers, microphones and pressure sensors need acoustic or venting membranes, which brings us to the part most teams skip.

Why sealed enclosures need to breathe

A perfectly sealed box is a pressure vessel. When a warm device in the sun is hit by cold rain, the air inside contracts and the enclosure pulls a partial vacuum. That pressure difference works on the seals every day, and over many cycles it can draw moisture in past seals that would pass a static immersion test.

The usual fix is a microporous membrane vent, typically expanded PTFE, which lets air and some water vapour through while blocking liquid water and dust. Vents come as snap-in, screw-in or adhesive parts with their own IP ratings. They need to be placed where water will not pool over them and protected from direct jets if the product is cleaned that way.

Venting does not solve condensation on its own. If the device is assembled in humid air and then cooled, that moisture condenses inside. Assembling in a controlled environment, adding a small desiccant where space allows, and conformal coating on the PCB are the usual layers of defence.

Testing a sealed design before the lab

A formal IP test is worth booking only once you are confident the design will pass. On a small batch we usually work through these steps first:

  • Tolerance check on real parts. Measure gland depth and flange flatness on several parts, not one.
  • Pressure decay test. Pressurise or evacuate the enclosure slightly and watch for pressure loss. It finds leaks quickly and without getting electronics wet.
  • Immersion with indicator paper. Line the inside with water-sensitive paper, immerse at the rated depth and time, then look for where water came in.
  • Temperature cycling, then immersion again. This is where compression set and lid creep show up.

We fold these into the wider environmental plan described in environmental testing on a small-batch budget, so sealing, thermal and vibration results are read together.

What we would do on a new sealed product

  • Write down the real exposure first: immersion, jets, sun, oils, cleaning chemicals. Pick the IP codes from that, not from a competitor’s box.
  • Count every opening and give each one a named sealing method before layout is frozen.
  • Design the gland from the seal manufacturer’s guide and run the tolerance stack at both extremes.
  • Stiffen the flange and check deflection between screws before committing to a fastener count.
  • Plan a membrane vent from the start; adding one after tooling is painful.
  • Run pressure decay, immersion and cycling on prototypes before booking a lab slot.

None of this is glamorous, but it is the difference between a product that passes once in a lab and one that is still dry after two winters on a pole.

Working on something similar?Free scoping call →