What Is Ultrasonic Welding?

A Guide for Medical Parts Manufacturing

I've been running ultrasonic welders for about twelve years now, mostly on medical lines. Catheters, IV sets, filter housings-you name it. People ask me all the time what ultrasonic welding actually is, and I tell them it's basically using sound to melt plastic. High-frequency vibrations, way above what you can hear, generate friction heat right at the joint. No glue, no screws, no mess.

How the Machine Works

Four pieces make up the system. Power supply takes your wall current and bumps it up to 20 or 40 kHz-sometimes 15 or 35 depending on your application. Converter sits on top of the stack; it's got piezoelectric ceramics inside that turn electrical signal into physical movement. Booster in the middle either amplifies or damps down that motion. Horn at the bottom actually touches your part.

Most shops I've worked in run 20 kHz for bigger parts. When you're doing tiny connectors or thin-wall stuff, 40 kHz is better-less force, gentler on delicate geometry. Branson, Dukane, Herrmann-they all make solid equipment. Pick one and stick with it, honestly. The learning curve is steep enough without switching platforms.

Part goes in the fixture, horn comes down. Machine triggers at a preset force or position. Vibrations start. Energy concentrates at the joint interface where you've got your energy director or shear joint. Plastic melts in a fraction of a second. Horn holds pressure while material cools. Done. Pull test it, move on to the next one.

Process Parameters

Amplitude matters more than people realize. It's the peak-to-peak movement at the horn face, measured in microns. Typically 20-100 microns depending on material and joint size. I've seen guys crank amplitude to fix a weak weld when they should've been looking at their energy director geometry. Higher amplitude dumps more energy into the part, sure, but it also risks burning or marking.

Weld time on medical parts runs short. Quarter second, half second. Maybe a full second on a big shear joint. Longer than that and you're probably doing something wrong-either your joint design needs work or your material isn't right for ultrasonic. I've seen people run two-second welds trying to salvage a bad tool design. Doesn't end well.

Hold time keeps pressure on while the plastic sets up. Half your weld time is a decent starting point. Cut it too short on nylon or PP and the joint pulls apart because the crystalline structure hasn't locked in yet.

Force is trickier than it sounds. Not enough and the horn bounces, gives you inconsistent contact. Too much and you're crushing the energy director before it has a chance to melt properly. I usually start around 200-400 N for smaller medical parts and dial it in from there.

Material Selection

Amorphous Plastics

ABS is the easiest stuff to weld. Period. Good energy transmission, wide process window, forgives a lot of sins. If you're prototyping or teaching somebody the process, start with ABS. Polycarbonate welds well too but watch out for residual stress from molding-PC loves to crack at stress risers when you hit it with ultrasonic energy. Had a batch of diagnostic housings crack every time until we traced it back to the gate location.

Polystyrene and acrylic behave nicely. CYROLITE shows up a lot in IV connectors and blood-contact applications. Amorphous materials soften gradually over a temperature range, which is exactly what you want. The melt flows, wets out the joint surface, chains diffuse across the interface. Clean welds.

Semicrystalline Plastics

Now we get into the difficult territory. Polypropylene, nylon, acetal, polyethylene-all semicrystalline. They've got ordered molecular regions that scatter the ultrasonic energy instead of transmitting it. Sharp melting point means the material goes from solid to liquid fast, then resolidifies the instant temperature drops. Very little working time.

PP and PE are everywhere in medical-syringes, containers, closures. They weld, but you need higher amplitude, near-field setup (horn close to the joint), and shear joints for anything requiring a hermetic seal. Energy directors alone won't cut it for leak-tight PP assemblies. I've seen too many people try.

Nylon absorbs water like a sponge. Parts sitting on the shelf for a week will weld differently than parts fresh off the molding machine. Moisture boils at 100°C during the weld cycle and creates porosity in the joint. Dry your nylon or weld it quick.

Quick Reference

On Mixing Materials

People ask about welding dissimilar plastics. Short answer: mostly no. ABS to PC works because they're chemically compatible-that's why PC/ABS blends exist. Melt temps need to be within about 40°F of each other. PP to PE? Forget it. I know they're both polyolefins but they won't bond.

Glass fill changes things. Up to 20% usually helps-stiffer material transmits energy better. Past 30% and there's not enough polymer at the joint interface to make a decent weld. Flame retardants are trouble. Lubricants on the surface are worse. Had a contamination issue once that traced back to silicone mold release on a supplier's parts. Took three days to figure out.

Joint Design

Everything in ultrasonic welding comes down to joint design. Bad joints don't weld, period. You can tweak parameters all day and it won't fix a fundamentally flawed geometry.

Energy Directors

Standard energy director is a triangular ridge molded into one mating surface. Point of the triangle touches the flat opposing face. Small contact area means high energy concentration. Plastic melts fast, flows out to fill the joint. 90° included angle for amorphous stuff. 60° for semicrystalline-sharper point compensates for faster solidification.

Height runs 0.5 to 1.0 mm typically. Smaller parts, smaller ED. The base width matches the height for a 90° director. One thing I see constantly: mold wear rounds off the energy director over time. New tool makes great welds. Six months later, welds are weak. Check your ED geometry when that happens.

Step and Tongue-and-Groove

Step joint adds alignment. One part has a raised step, the other a matching recess. ED sits on the step. Parts self-locate. Flash goes into the clearance instead of outside. Good for cosmetic applications where you don't want visible weld lines.

Tongue-and-groove takes it further with a protruding tongue fitting into a groove. Contains flash on both sides. Downside is tighter tooling tolerances-equal clearance on both sides of the tongue is hard to hold in the mold.

Shear Joints

Shear joints are non-negotiable for hermetic seals on semicrystalline parts. Two walls with slight interference slide past each other during welding. The walls melt as they telescope together. Continuous smearing action eliminates voids and leak paths.

Weld times run 3-4x longer than ED joints. More material melting, more energy required. Interference is typically 0.2-0.4 mm per side. Too loose and you don't get full melt. Too tight and parts won't assemble for welding. Dimensional control matters.

Filter housings, fluid reservoirs, anything holding pressure-shear joint. I've never gotten a reliable hermetic seal on PP with an energy director alone. Some people claim they have. I haven't seen it hold up in production.

Applications in Medical

IV sets are the bread and butter. Drip chambers, connectors, spikes, ports-all welded. No adhesives means no biocompatibility concerns about leaching or residue. A single IV set might have four or five welded joints. High-speed rotary systems can run thousands per hour.

Filtration is huge. Blood filters, dialysis cartridges, respiratory filters. Membrane gets embedded in the housing. Weld seals the perimeter without damaging the filter media. Any leak path and the filter is useless-fluid bypasses instead of going through. Shear joints, 100% inspection, process capability studies. FDA doesn't mess around with this stuff.

Drug delivery keeps growing. Insulin pens, auto-injectors, inhalers. Multiple subassemblies per device. Wearable patches with thin membrane seals-that's where torsional welding systems shine. Regular linear vibration can puncture thin films. The horizontal oscillation pattern of torsional units is gentler.

Diagnostic cartridges run fast. Test strip housings, sample cups, reagent wells. Cycle times under a second. Positioning is critical because these parts are tiny and tolerances are tight. Robotic handling feeds parts to dedicated welding stations.

Quality and Process Control

Modern welders record everything. Actual weld time, energy delivered, peak power, collapse distance. Set up your limits and the machine flags anything outside the window. We run SPC charts on every critical joint. Catch a trend before it becomes a reject.

Pull testing validates the weld. Tensile test fixtures grab the part and measure break force. Run samples at start of shift, periodically throughout production, end of shift. Burst testing for pressure-holding assemblies-fill it up, increase pressure until it fails, record where and at what PSI.

Cross-sections tell you what's happening inside the joint. Mount it in epoxy, cut through the weld, polish the surface, look under magnification. Complete melt and flow looks homogeneous. Incomplete fusion shows a visible interface. Porosity means moisture or contamination. Do this during development to optimize parameters. Do it periodically in production to verify.

Troubleshooting

Weak Welds

First thing I check is the energy director geometry. Grab a part, look at it under magnification. Is the point sharp or has it rounded off? Worn molds round off EDs and nobody notices until pull test numbers drop.

Second is horn contact. Uneven contact means uneven energy delivery. Misalignment puts all the energy on one side of the joint. Look for witness marks on the part-they'll tell you where the horn is actually touching.

Third is material batch variation. New lot of resin, weld parameters that worked yesterday don't work today. Regrind in the mix changes flow characteristics. Got burned on this once when a supplier started adding 15% regrind without telling us.

Timing between molding and welding matters too. Hot parts fresh off the machine behave differently than cold parts that sat in a bin overnight. Standardize it or adjust for it.

Flash Problems

Too much flash usually means too much energy. Back off amplitude first. Then weld time. Then force. One variable at a time or you'll never figure out what actually fixed it.

Joint design can help too. Flash traps-a small recess around the weld perimeter-catch displaced material. Step joints hide flash in the clearance gap. Built-in solutions beat trying to fix it with parameters.

Cracking and Part Damage

PC cracks at stress concentrations. Sharp corners, gate vestiges, molded-in stress from cooling gradients-all potential crack initiation sites. Sometimes the fix is changing the molding parameters, not the welding parameters.

Cold parts are brittle. Warehouse hits 50°F in winter, parts come in cold, they crack instead of melting. Let them warm up or put a heater in the line ahead of the welder.

Horn marks on cosmetic surfaces? Lower trigger force, polish the horn face, or put a thin Mylar film between horn and part. The film takes the abuse instead of the plastic.

Equipment Maintenance

Horn condition is everything. Cracks, pitting, wear grooves-all affect energy transfer. Visual inspect daily, impedance check weekly. A cracked horn sounds different when it runs-you can hear it change. Replace before it fails catastrophically. I've seen cracked horns damage converters and that's an expensive repair.

Converters degrade over time. Piezo ceramics lose efficiency. Output drops gradually enough that you don't notice until welds start failing. Periodic output measurement catches the decline. Replace on a schedule rather than waiting for failure.

Fixtures wear. Parts sit in the same spots, plastic rubs against metal, dimensions drift. Check against drawing specs periodically. Rebuild or replace when tolerances open up. A worn fixture changes part position, which changes weld quality, and you'll be chasing phantom problems if you don't catch it.

Regulatory Notes

ISO 13485 covers quality management for medical devices. Process validation-IQ, OQ, PQ-demonstrates that your welding operation consistently makes acceptable product. Document everything. Change control applies to parameters, tooling, materials. Change one thing without proper documentation and you've got a nonconformance.

Ultrasonic welding works well in cleanrooms. No vapors, minimal particulate, no liquid adhesives dripping or splashing. Equipment integrates easily into controlled environments. Just keep the maintenance schedule and don't contaminate parts with machine grease or cutting oil.

Parting Thoughts

Pick your material based on the application first. Then figure out if it welds. ABS and PC give you the widest process window. PP and nylon need more attention but they're doable. Shear joints for hermetic seals on semicrystalline materials. Energy directors for most everything else.

Get your joint design right before you buy the tooling. Fixing a bad joint design after the mold is cut is expensive and frustrating. Send sample parts to your equipment supplier for welding trials. They do this all day-let them help you find the right parameters.

Most problems trace back to the basics: part dimensions, material consistency, equipment condition. Fancy monitoring systems won't help if your energy director is rounded off or your fixture is worn. Master the fundamentals and the rest follows.