When it comes time to close two halves of a housing, you have four real options: screws, snap features, adhesive, or welding. Ultrasonic welding is the volume-production answer — it fuses two thermoplastic parts into a single body in well under a second, with no fastener, no adhesive, and no cure time. It is the right choice when you need a sealed joint, a clean exterior with no screw holes, or the shortest possible cycle time on an assembly line. It is the wrong choice when the product has to be opened for service, or when annual volumes are too low to amortize the tooling.

How the Process Works

A power supply generates a high-frequency electrical signal, typically in the 20 to 40 kHz range. A piezoelectric converter turns that into mechanical vibration, a booster scales the amplitude, and a tuned tool called a horn transmits the vibration into the upper part while a press applies static force. The lower part sits in a fixture — the nest — that supports it precisely where the energy arrives. The vibration produces intermolecular friction at the interface between the parts, the material heats to a local melt, the two surfaces fuse, and the joint solidifies under continuing pressure. The whole cycle is usually a fraction of a second, and the part moves straight on down the line.

Three parameters control the outcome: amplitude, force, and weld time or energy. Modern welders run in energy mode or collapse-distance mode rather than fixed time, which means each weld is measured and out-of-family parts are flagged automatically. That is a significant production advantage: the joint becomes a recorded measurement instead of a visual judgment, which is exactly the mindset behind first article inspection and process control generally.

Which Plastics Weld and Which Fight Back

  • Amorphous polymers — ABS, polycarbonate, acrylic, polystyrene, PC/ABS blends — soften over a broad temperature range and weld easily. They also tolerate far-field welding, where the horn contacts the part some distance from the joint.
  • Semi-crystalline polymers — nylon, polypropylene, polyethylene, acetal, PBT — have a sharp melting point and absorb vibration, so they need higher energy and near-field contact, with the horn close to the joint.
  • Glass-filled grades weld well up to a moderate fill percentage; above that the fibers get in the way of polymer-to-polymer contact and joint strength falls off.
  • Dissimilar materials only weld when their melt ranges are close and they are chemically compatible. ABS to polycarbonate can work; ABS to polypropylene will not.
  • Flexible or heavily plasticized materials absorb the vibration instead of transmitting it, and are generally poor candidates.
  • Hygroscopic resins such as nylon weld inconsistently if moisture content varies, so drying and storage become process controls.

Because weldability constrains resin choice, lock this decision alongside the rest of the material selection described in how to choose the right plastic for your product, not after.

Joint Design Is Where It Is Won or Lost

Success is determined on the drawing, not at the machine. There are two dominant joint geometries:

Energy director. A small triangular rib projecting from one mating face concentrates friction into a tiny contact area so it melts first, then flows across the joint. Simple, cheap to mold, and the default for amorphous resins. Best for structural joints where a hermetic seal is not the primary requirement.

Shear joint. One part enters the other with a small interference and melts progressively down the wall as it descends. This gives excellent hermetic sealing and is the preferred geometry for semi-crystalline resins and for anything that must hold pressure or pass an ingress test — closely related to the sealing strategy discussed in building a waterproof prototype.

Either way, several details are non-negotiable: a melt trap or flash pocket to capture displaced material so it does not squeeze out onto a cosmetic surface; alignment features so the halves self-center without relying on the operator; uniform wall thickness in the energy path, because thickness changes redirect where the energy goes; and adequate draft on the joint features so they release from the tool. A useful rule of thumb: if you cannot sketch the joint cross-section on paper and explain where the molten material flows, it is not ready.

Because the joint is molded into both parts, changing it later means changing steel. Design it in the same review as mold design and build, and expect at least one iteration.

Mechanical Limits Worth Knowing Early

Tall parts and parts with thin unsupported ribs can absorb the vibration instead of transmitting it, producing a partial weld — or worse, cracking somewhere far from the joint where a resonance happens to land. Sensitive components inside the housing, such as a display, a crystal oscillator, a MEMS sensor, or a populated board with tall electrolytic capacitors, need to be isolated from the energy path or the weld will damage them. In projects where we see that risk, the practical move is a short series of experimental parts and a parameter sweep before committing to final tooling.

Cost, Volume, and the Alternatives

The economics are typical of tooled processes: a one-time investment in the horn and the nest, then a near-zero marginal cost per unit and a cycle measured in fractions of a second. Horn and fixture tooling generally lands in the low thousands of dollars for a simple part and climbs with size, part count per cycle, and complexity; the welder itself is a capital item usually owned by the contract manufacturer. At high volume, ultrasonic welding is one of the cheapest joining methods available per unit.

Compare it honestly against the neighbors:

  • Heat staking — for attaching a board or a bracket rather than sealing a seam.
  • Laser and spin welding — laser for delicate assemblies and precise energy placement, spin welding for circular joints.
  • Adhesive — handles dissimilar materials and awkward geometry, but adds cure time, fixturing, and a chemical process to control.
  • Screws into bosses or inserts — the choice whenever the product must open, as covered in threads in plastic parts.
  • Snap features — cheapest of all, no secondary equipment, but harder to seal reliably; see the snap-fit design guide.

How to Validate a Welded Joint

Do not accept a weld on appearance. Run destructive pull or burst tests on a sample from the first production run and compare against the parent material strength. Section welds and inspect the melt zone. Leak-test to the declared ingress rating, including after thermal cycling, since a joint that passes fresh can open after expansion and contraction. Then run a small designed experiment across amplitude, force, and energy to find the process window, and set production limits inside it rather than at its edge. Record the weld energy and collapse distance for every unit — that data is what lets you catch a drifting process before it becomes a recall.

If you are deciding how to close a housing, or a welded assembly is failing leak or strength tests, Projects House designs weldable parts and works with the tooling and manufacturing partners that produce them. Send us your design through our contact form and we will review the joint before the steel is cut.