Waterproofing a 3D printed enclosure for a PCB

No, ultrasonic welding is not the only way, and for a 3D printed case it is the wrong way: welding belongs on injection-molded housings on a production line. You make a printed enclosure waterproof in three moves. Fix the print first, with at least 4 perimeter walls and a thin epoxy wipe on the seal faces. Then close the joints: a 2 mm silicone cord in a printed groove at 25% squeeze for the halves, a clear window bonded over the LCD with neutral-cure silicone, and a foam gasket on the battery door so it still opens. Built like that, the case takes splash and jet exposure, IPX5 level. For real immersion — 1 m of water for 30 min — you print the prototype, then order a molded or stainless enclosure.

Is ultrasonic welding the only way to waterproof a plastic enclosure?

No — welding is a production process for injection-molded ABS or polycarbonate halves, and on a one-off printed case a gasketed flange with the cord at 20-30% squeeze seals to the same splash and jet level while staying openable. Ultrasonic welding focuses vibration into a molded energy director and melts the halves into one part; a printed surface gives that process nothing predictable to melt against, and once welded the joint stays shut for good. Ultrasonic welding an enclosure with a sensitive PCB shows where the process does win. So the split is simple. A joint you will reopen — the battery door, the main halves if you service the board — gets a gasket and screws. A joint you close once, like the window, gets sealant. And when the electronics are cheap and the water is serious, potting industrial electronics against water and shock is the zero-maintenance end of the scale, with zero repairs.

Can a 3D printed enclosure be made watertight at all?

Yes, with settings rather than luck: an FDM part weeps along its layer lines, so print at least 4 perimeter walls (about 1.6 to 1.8 mm at a 0.4 mm nozzle) and run 100% infill within about 10 mm of every seal face. The gaps between round extruded strands are the leak path, and more walls close more of them; running the nozzle 5-10 C hotter than normal fuses the layers better. Prove the settings on a coupon before you commit the case: print a 100 x 100 x 25 mm open box at final settings, fill it with water, stand it on dry paper for 24 h, then read the paper. A dry coupon earns the case; a weeping one means one more perimeter and another pass. Print PETG for general use and ASA for outdoor UV — the FAQ has the temperature numbers. Belt-and-braces, wipe thinned epoxy onto the seal faces inside and out, 0.1-0.2 mm per coat, to bridge whatever the layers still pass.

How do you seal the joint between the two halves?

Cut a groove into one half, lay the silicone cord into it, and let the flat mating face of the other half squeeze it 20 to 30% — the standard static face-seal band from o-ring design — with screws every 50 mm [2 in] around the flange. The groove does three jobs: it locates the cord, its depth sets the squeeze, and it stops the screws from crushing the cord flat, which is what kills a printed-flange seal first. Print the groove walls at least twice the cord's thickness, and press brass heat-set inserts into the bosses, because screws in printed threads go slack after a few openings. The tongue and groove gasket design for a sealed enclosure runs the same squeeze math for molded and sheet-metal flanges, and every rule of it transfers.

The arithmetic on a 120 x 80 mm [4.7 x 3.1 in] lid

Tighten those eight to about 0.5 N·m [4.4 lb-in], crosswise in two passes, and the squeeze stays even all round.

How do you seal the LCD seam?

Move the sealing problem off the display: bond a clear acrylic or polycarbonate window, about 3 mm thick, over a printed opening, with the LCD sitting 1 to 2 mm behind it where no water reaches it. The window takes the jet load, and the silicone bead takes the window. Print the opening so the pane lands on a 10-12 mm land all round, lay a bead of neutral-cure silicone the same thickness as the cord — acetoxy releases acetic acid onto solder and pins — and tool a small fillet on the outside edge once it skins over. Keep unsupported spans under about 100 mm [4 in]; wider panes flex under jets and peel the bead, so print a rib, the pattern the long window seal spacing and support article works through. If the display has to take touch through the seal, a bonded bezel display enclosure or a factory HMI housing does that joint better than any printed bead.

How do you seal the battery compartment?

Treat it as a service door, because it has to open: a 1.5 mm closed-cell foam gasket compressed about a third, held down by screws through heat-set inserts, reseals dozens of times where sealant would be torn apart on the first opening. Closed-cell gaskets seal at 30-50% deflection, so print a 1.0 mm stop land inside the door opening: at full closure the foam sits at 33% deflection and the door bottoms on plastic instead of on a crushed gasket. Keep the screws outboard of the gasket line where you can; where a screw must cross the sealing band, put a sealing washer under its head. The logic matches sheet metal enclosure lid fasteners for repeated access, scaled down — and at 0.5 N·m your plastic flange runs out of stiffness long before the screw does.

What IP rating can a printed enclosure actually reach?

Splash and jet level, with immersion out of reach: a printed case built as above sits in IP65 territory — dust-tight gasket, jet-resistant joints — a claim your own water test has to confirm, while IPX7 immersion at 1 m for 30 min belongs to molded or stainless enclosures because print porosity and layer creep work against you. Those digits carry fixed test conditions under IEC 60529: the X5 digit means a 6.3 mm nozzle delivering 12.5 L/min; X6 means a 12.5 mm nozzle at 100 L/min from 2.5-3 m for at least 3 min; X7 means immersion to 1 m for 30 min, and IP66 by itself is not an immersion rating. An IP code is a test result on one sample, so your print carries no rating until it passes the water test — the IP rating chart lays out every digit. The assembly then takes the rating of its weakest part, the same rule behind what NEMA rating a panel keeps after you cut it and mount devices: one cheap cable gland or charge port drags the whole box down to its rating.

Should the enclosure be fully sealed or allowed to breathe?

Fully seal it, and spend the same care on the cable entries as on the lid: thermal cycling turns any small leak into a one-way pump, so a case that almost seals ends up damper inside than one built to breathe from the start. Cool a 1 L box from 30 C to 20 C and the trapped air contracts about 3.3% — some 33 mL — and it pulls that volume in through whichever gap sits lowest. The cure is a gland seat printed and sealed as carefully as the lid groove, which is exactly where condensation in an IP67 project box usually begins. A conformal coat on the PCB — 0.05 to 0.1 mm of acrylic or silicone — buys a second line of defense for the humidity that gets through anyway. If the design genuinely needs a vent or drain, fit one on purpose and accept that the assembly now carries that vent's rating.

When should you stop printing and order an enclosure?

At the first washdown hose, salt air, all-day sun, submersion spec or second unit: a printed case is the prototype route, and for anything with a display the clean answer is a 304 stainless HMI housing with a factory-sealed front. HMI enclosures run displays from 7 in to 23.8 in — the 7 in and 10.1 in sizes cover most handheld-to-panel jobs — with an IP65-type front seal with the display installed, manufacturer rated, not UL listed. Salt air or chlorides move the material to 316L. Without a display, a custom stainless enclosure cut to your layout is delivered in 10 business days from your approval of the drawing, in one piece with every cutout done at the factory, where the rating is still intact. And if this box carries mains wiring past an inspector, print the mock-up by all means, but the installed enclosure needs a listing — what UL 50E listing asks of an enclosure — which a home print does not have.

How do you build it and prove it seals?

Work the nine steps below in order — settings, coating, cord, bead, assembly — and finish with the paper-strip and water checks, which tell you the same day whether the build holds at IPX5 spray conditions.

1. Print the 100 x 100 x 25 mm coupon at final settings, stand it 24 h full of water on dry paper, and add one perimeter plus 5-10 C of nozzle temperature if it weeps.
2. Print the case in PETG, or ASA outdoors: at least 4 perimeters, 100% infill within 10 mm of every seal face, groove walls at least twice the cord's thickness.
3. Wipe the seal faces inside and out with thinned epoxy, 0.1-0.2 mm per coat, and cure overnight.
4. Press brass heat-set inserts into every lid and door boss.
5. Cut the silicone cord 3% shorter than the groove path, butt the ends with a drop of cyanoacrylate, and press it into the groove.
6. Bond the window with a neutral-cure silicone bead on its 10-12 mm land and give it a day of cure before handling.
7. Stick the 1.5 mm closed-cell gasket into the battery door and check the 1.0 mm stop land.
8. Assemble at about 0.5 N·m per M3 screw, crosswise, in two passes.
9. Test with the paper strip first, then with water.

The checks that prove it

• Pull test: close the lid on a strip of printer paper (about 0.1 mm) at the middle of each side; drag on every strip means the cord is compressed all round.
• Spray test: 6.3 mm nozzle, 12.5 L/min, about 3 min from roughly 3 m, every angle, with a sheet of dry paper inside — any damp dot fails the build and marks the joint.
• For an immersion target, dunk to 0.3 m for 30 min first; a fail at 0.3 m saves you the discovery at 1 m.
• Cycle the battery door 10 times and repeat the spray test; a gasket that reseals is the whole point of the design.

The mistakes that undo it

Four slip most often: acetoxy silicone on the window, whose acid attacks solder; screws straight into printed threads, which strip and relax; torque past the groove stop, which flattens the cord into a dead seal; and one cheap cable gland carrying the whole box's rating down with it. The fifth is leaving PLA in the sun, and that one softens on its own schedule.

Enclosures in this article

Questions