Factors affecting ultrasonic welding

• 1
  How factors affecting ultrasonic welding generate heat
• 2
  Material modulus and glass content
• 3
  Joint geometry: energy director, shear joint, staking
• 4
  Amplitude, frequency and weld parameters
• 5
  Fixture support, part tolerance and surface condition
• 6
  Factors affecting ultrasonic welding: material and joint checklist

A converter turns a 20 kHz or 15 kHz electrical signal into mechanical vibration. The horn presses that vibration into the top part, the vibration travels down to the joint, and the two surfaces rub against each other at microscopic scale. Friction at the interface, plus bulk viscoelastic heating, melts a thin layer of polymer in tens of milliseconds.

No external heat is applied. The energy is generated exactly where you want it, at the joint line, which is why cycle times of 0.2–1.0 s are normal and why the surrounding part stays cool. That is the whole appeal of the process.

The melt layer is only 0.1–0.5 mm thick. Everything outside that layer stays solid, so the part keeps its shape and any molded-in features survive. A CNC-machined horn or fixture has to hold the parts flat against each other while this happens, to within a few micrometres of parallelism.

One consequence matters for design. Because heating depends on friction between two surfaces, the process is sensitive to anything that changes how those surfaces meet: glass content, mold release, parting-line flash, sink marks. Those are the factors affecting ultrasonic welding that catch people out, not the machine settings alone.

Elastic modulus decides how well a resin transmits vibration. Amorphous resins such as ABS, PC, PMMA and PS have a wide softening range and transmit energy efficiently. Semi-crystalline resins such as PA, POM, PP and PEEK melt sharply and absorb more energy near the joint, so they need higher amplitude and tighter process windows.

Glass-filled grades are the harder case. At 10–20% glass the material still welds, but the fibers do not melt and they sit at the interface as a barrier. At 30% glass or above, weld strength typically drops well below the unfilled grade, so a mechanical interlock or a different joining method is worth considering.

Filler also wears the horn. Glass and mineral fillers erode titanium and aluminium tooling, which changes the resonant frequency over a production run. If you weld 30% glass-filled PA, plan for horn inspection every few thousand cycles rather than treating the horn as a permanent tool.

Two dissimilar resins rarely weld. The classic exception is ABS to PC, which works because their solubility parameters are close. PP to ABS, POM to PA, or anything to a fluoropolymer will not form a melt bond no matter how you tune amplitude.

The energy director is a small triangular ridge, usually 0.25–0.5 mm tall with a 60° or 90° included angle, molded onto one half of the joint. It concentrates the first contact into a tiny area so melting starts there and spreads outward. Get the apex too blunt and the weld starts late and unevenly.

A shear joint is for semi-crystalline resins and for parts that need a hermetic seal. The walls interfere by 0.2–0.4 mm and the vibration melts the interference as the parts telescope together. It needs a rigid support fixture because the joint is loaded sideways, not just in compression.

Staking and spot welding use the horn directly against a molded stud or a flat pad. Stud diameter should be roughly 1.5–2 times the horn contact diameter, and the stud should have a small pilot hole to control where the melt collapses.

Wall thickness at the joint should be at least 1.5 mm, and 2.0–3.0 mm is comfortable. Thin walls flex under the horn, absorb energy, and give you a weld that looks closed but peels apart under load. Add a stiffening rib near the joint before you add amplitude.

Amplitude is peak-to-peak displacement at the horn face, typically 20–60 μm at 20 kHz. More amplitude means more frictional energy per cycle, but it also means more stress on the part and more flash squeezed out of the joint. Far-field welding, where the joint sits 6 mm or more from the horn contact, needs higher amplitude to compensate for attenuation.

Frequency is chosen by part size and by material. 20 kHz is the standard for small and medium parts. 15 kHz gives longer wavelength and better for large, long parts. 35 kHz or 40 kHz suits delicate assemblies such as electronics housings and medical device components where you want low displacement and low stress.

Weld mode matters as much as the numbers. Time mode is simple but blind to part variation. Energy mode stops when a set number of joules has been delivered and is more repeatable for molded parts with consistent geometry. Collapse distance mode stops when the parts have closed by a set distance, which is the most direct control of final joint height.

Trigger force, weld force and hold force are separate settings and each one changes the result. Low trigger force lets the horn touch down softly; low weld force lets the parts slip and burn; too much hold force after the vibration stops can crush a still-soft joint. Hold time of 0.3–1.0 s is a common starting point.

The fixture has one job: stop the lower part from moving while the upper part vibrates. Any compliance in the nest absorbs energy, and the joint at that spot stays cold. Support the part directly under the joint line, not at the far end of the part.

Molded parts carry tolerance. A warp of 0.3 mm across a 60 mm joint means the energy director only touches on one side, so the weld starts there and never closes the gap. Fixture design should push the parts into contact with a light preload rather than relying on the parts being flat.

Contamination kills the bond. Mold release spray, silicone from a dispenser, cutting oil from a secondary machining step, and skin oils all sit at the interface and act as a release layer. Wipe with isopropyl alcohol and let it flash off before welding. Do not assume a freshly molded part is clean.

Moisture matters for polyamides. PA absorbs water, and absorbed moisture flashes to steam at weld temperature, which produces porosity and weak joints. Dry PA parts to below roughly 0.2% moisture and weld them soon after, or the same settings will give different results on different days.

If the parts are CNC-machined rather than molded, the surface texture is different in kind. A machined face at Ra 1.6–3.2 μm has cutting marks and a different real contact area than a molded face, and the energy director has to be machined to the same 0.25–0.5 mm height. Machined prototypes weld, but expect to re-tune amplitude once the molded parts arrive.