Skip to content
UAV MFG

Guides / Design

Machining a drone enclosure that actually keeps water out

August 16, 2026

An IP rating on a drone enclosure is not a property of the gasket. It is a property of the groove, the face, the fasteners and the vent, most of which are machined features.

Almost every sealed housing we see that leaks, leaks for one of four reasons.

1. The groove was drawn from the cord, not from the squeeze

An o-ring seals because it is compressed. The groove has to be shallower than the cord diameter so that closing the lid squeezes the seal by the right amount, and wider than the cord so the displaced rubber has somewhere to go.

For a static face seal, a typical target is 15 to 30 percent squeeze, with the groove width around 1.1 to 1.3 times the cord diameter. The exact numbers come from the seal manufacturer table for that cord and that application. What you must not do is model the cord in CAD, cut a semicircular groove that fits it perfectly and expect a seal. That groove gives zero squeeze and zero sealing.

Put the groove dimensions on the drawing as dimensions with tolerances, and put the o-ring part number on the drawing as well.

2. The sealing face was machined before the pockets

A housing loses most of its mass when the inside is cut out. The part then relaxes, and a face that was flat when it was machined is no longer flat when it ships.

The fix is process, not tolerance. Rough the pockets, let the part rest, then finish the sealing face last with light cuts. If a shop quotes a sealed housing without mentioning this, ask how they plan to hold the flatness, because the answer tells you whether they have made one before.

Call flatness out on the sealing face. Something around 0.05 mm across the face is realistic for a machined enclosure, and it should be inspected on a surface plate rather than assumed.

3. The corners are tighter than the seal can follow

An o-ring in a rectangular groove has to bend around every corner. Tight corners stretch the cord on the outside and bunch it on the inside, and that is where a leak path opens.

  • Keep internal corner radii on the groove generous, several times the cord diameter where you can.
  • Keep the groove path in one plane. A groove that steps between levels needs a moulded gasket, not an o-ring.
  • Avoid a groove that runs right up to a fastener boss. There has to be material for the groove wall.

4. The box cannot breathe

A sealed enclosure heats up in the sun and cools in flight. Air inside expands and contracts. With nowhere to go, the pressure difference works against the seal, and each cycle pulls a little moisture in.

A protective vent, a small membrane element that passes air but not liquid water, removes the problem for the price of one machined threaded hole. Any enclosure that is genuinely sealed and sees temperature swings should have one.

Fasteners and lid stiffness

The seal is squeezed by the lid, and the lid is held by screws. Between screws, a thin lid bows and the squeeze drops.

  • Space the screws closer than you think, typically four to eight cord diameters apart on a small enclosure.
  • Make the lid stiff near the seal, with a flange or a rib, rather than uniformly thick.
  • Give the screws real thread engagement in the base. Threads in thin aluminium are covered in threaded inserts in thin UAV plates.
  • Use a captive or retained screw where a technician opens the box in a field with long grass.

Connectors, cables and the other holes

The groove is usually not where the water gets in. The connector is.

  • Use connectors with their own rated seal, and machine the panel cutout to the connector drawing, including the flat land the seal needs.
  • A cable gland needs a flat, square faced boss, not a curved surface.
  • Every hole through the wall is a leak path. Count them at design review.

Material and finish

6061-T6 is the standard. It machines cleanly, it anodises evenly and it conducts heat out of the electronics inside. Hard anodise for aircraft that spray or fly near sea, type II for everything else.

Two finish details that matter here:

  • Do not anodise inside the o-ring groove if the drawing dimensions are pre coating and the squeeze is marginal. Decide, and say so.
  • Mask any internal face that has to conduct heat to a component or carry a ground. Anodise is an insulator.

A groove call out that quotes without questions

  • O-ring part number, cord diameter and material.
  • Groove width and depth with tolerances, and the squeeze percentage they came from.
  • Groove corner radii.
  • Surface finish on the groove and the mating face.
  • Flatness on the sealing face.
  • Whether the groove is masked during anodising.

Send the model with those six lines through the RFQ page and the sealed housing quote comes back within 24 business hours, with the groove checked against the seal table before anyone cuts metal. There is a worked version of this part on the worked examples page.

Frequently asked questions

How deep should an o-ring groove be?

Deep enough to give the seal the compression its maker specifies, typically 15 to 30 percent squeeze on a static face seal. The groove depth comes from the cord diameter and that squeeze figure, from the seal maker table, not from the CAD model of the cord.

Why do sealed enclosures leak after a temperature change?

Because a sealed box is a pressure vessel. Air inside expands and contracts, and if the box cannot breathe it pushes past the seal on the way out and pulls moisture in on the way back. A protective vent solves it.

What surface finish does a sealing face need?

Smoother than general machining but not polished. Around Ra 0.8 to 1.6 on the groove walls and the mating face works for a static seal. Machining marks running across the seal are the ones that leak.

Need this part quoted?

Send the model and the quantity. Price and lead time come back within 24 business hours.

Send a drawing

Keep reading