Designing Enclosures for IoT Devices

The gasket arithmetic below is hypothetical teaching maths. It is not a seal rating, not an ingress result and not a BrahmWorks test. An ingress mark on a drawing is a wish until a named build is tested to a named method.

A 2.4 GHz radio, or any other, has a keep-out. That keep-out is a volume: no metal, no conductive coating, no battery, no shield can, no cable, and no screw boss you forgot was metal. Metallic paint and some carbon-filled or metal-filled resins are conductors for this purpose. A rib across the keep-out changes the plastic in front of the antenna. The board can move the antenna; the enclosure can ruin it without moving the board at all.

Put the antenna, the keep-out and the cell on one section before you freeze the outline. A kit that connected lid-off is not evidence. If the design wants metal, give the radio a plastic window or a strain-relieved external antenna on purpose. A module maker's keep-out, ignored, is your redesign. Light pipes that puncture a seal, or leak into the next indicator, are the same class of geometry problem.

A cell the user replaces needs a door, a cycle count, polarity that is hard to reverse, and a contact that still meets after wear. A pack the user never sees still needs a factory connection, and the space the cell maker requires, including growth if they specify it. This article gives no growth figure. A glued door is neither a service method nor a factory method. Give wires a channel. Tape across a keep-out is a material you did not specify.

If the product is honestly indoors and dry, do not design a waterproof enclosure you will not test. A casual gasket that is not compressed to a stop becomes a leak and a false sense of a rating. If the product does see water, dust or a wash-down, the seal is a designed joint: gasket section, groove, compression, a hard stop so screws cannot crush it further, and a parting line that does not cross the seal.

Closed boxes still breathe. A cold wall and warm air inside will condense on the board. A membrane vent can pass air and stop liquid if you seat it and do not paint it over. A drain that is also a leak is not a shortcut. Polymer conducts heat poorly. Metal helps the regulator and hurts the antenna unless you planned both. Put the thermal limit and the ingress claim on one page so they collide in review.

A tactile button needs a travel and a return. A membrane needs a flat land. Either fails when the stack was drawn before the board height was known, so the key is always half pressed or never reaches. A cable exit needs strain relief that takes the yank, not the solder joint. If the cable is the antenna, the exit is also keep-out.

Mounting screws need a load path into a rib. Specify the wall and the anchor. Adhesive fails by surface, temperature and time. "Any wall" is not a surface.

Programming and the functional test have to see the unit you ship. A header under a glued lid means you tested a different product. Test before the seal, then check the seal, and leave a path that service can use a year later if you offer service. If the product is disposable, say so, and do not pretend the door is a service joint.

Invented device: a wall sensor, primary cell, 2.4 GHz module, plastic housing, indoor plasterboard. The first brief says dry office. A later slide says "weatherproof" because a customer asked.

The teaching disposition is to refuse the slide or to redesign. The office design has a cell door, no gasket stop, a parting line wherever the print was split, and a light pipe with a gap. None of that becomes a seal by editing the brochure. If the real environment is the office, delete the weatherproof word and spend the design on the keep-out, the cell door and the mounting holes. If the real environment is a wet plant room, stop and design the seal, the vent and the antenna window as one change, then test that build. No radio result and no ingress result are claimed here.

Assume a round cord of 1.6 mm diameter, and a teaching target of 25 percent compression. Compressed height = 1.6 × (1 − 0.25) = 1.2 mm. With a hard stop, the groove floor is 1.2 mm below the stop, and the screws cannot pull the joint tighter than that.

Now assume the lid has no stop and the screws can close the joint another 0.3 mm. Height becomes 1.2 − 0.3 = 0.9 mm. Compression = (1.6 − 0.9) / 1.6 = 0.4375, about 44 percent.

The sum does not say 44 percent leaks or that 25 percent seals. Groove width, fill, cord tolerance, relaxation and the test method are all outside it. It says the screw, not the designer, is setting the compression once the stop is missing. A seal designed as a torque is a seal designed as a mood. Add the stop, then ask a seal supplier what compression that section wants. Do not copy 25 percent onto a drawing.

The enclosure is ready to leave the designer only when someone else can build this seal, this keep-out and this test sequence. That handoff is How to Take a Hardware Prototype to Production.

• Antenna keep-out is a volume on the assembly, including screws, cell, cables and coatings.
• The lid-on, cell-in configuration is the one that must connect. Lid-off does not count.
• Metal, paint and filled resin are marked as conductive or not.
• Cell space follows the cell maker's requirement, including growth if they specify it.
• Door cycles and polarity are a requirement, or the cell is not user-replaceable.
• Wires have a channel that is not the keep-out and not a screw path.
• Any seal has a groove, a compression, a hard stop and a parting line off the seal face.
• Ingress words on the drawing match a test you will run, or they are deleted.
• Vent, drain and condensation are explicit, including "none, because indoor".
• Heat and ingress are on one page.
• Button travel is in the same stack as the board height.
• Programming and end-of-line test survive the assembly sequence.
• Mounting loads go into ribs. The wall type is named.

No. Plastic is usually kinder than metal, which is not the same as invisible. Wall thickness, ribs, coatings, the cell and nearby metal still detune or shadow the antenna. The only evidence is the closed unit, in the material you will ship, at the supply voltage you will ship. A module datasheet range, measured on a kit, is not that evidence.

No. Seal the environment you actually have. An untested gasket is worse than an honest vented office box, because it traps condensation and invites a claim you cannot support. Seal when the requirement is water or dust, and then test the joint. Do not inherit a rating from a similar-looking product.

Distance in a manual is not a keep-out in the tool. If the cell is in the volume, move the cell or move the antenna. A note to the user will not fix a shadow inside the housing. The same applies to a "keep metal away" instruction aimed at a screw the user never sees.

Not before the keep-out, the seal story and the board outline have survived one honest build. Tooling a shape that still has the cell in the antenna volume locks the failure in. A printed shell is enough to see the clash. It is not enough to sign the moulding, which has its own draft and wall rules, handled as a separate review.

Bring the section that shows antenna, cell, seal and test access on one view. BrahmWorks can tell you whether that view is a product or a lid drawn around a kit.

Review the next stage with BrahmWorks