Ultrasonic Welding Design Guide for Stronger Joints

A product engineer once watched a flawless-looking medical housing pass every drop test, then pop open at its seam on day 30 of a shelf-life trial — the parts were fine, the joint had simply been designed without anywhere for ultrasonic energy to concentrate. That is the quiet truth of ultrasonic welding: the machine rarely fails, the drawing does.

Why Ultrasonic Welding Success Is Decided on the Drawing Board

A product engineer once watched a flawless-looking medical housing pass every drop test, then pop open at its seam on day 30 of a shelf-life trial. The parts were dimensionally perfect. The weld settings were dialed in. The joint had simply been designed without anywhere for ultrasonic energy to concentrate. That is the quiet truth of this process: the machine rarely fails, the drawing does.

If you are new to the topic, the short definition helps: ultrasonic welding joins workpieces by pressing them together while subjecting them to low-amplitude, high-frequency vibration, typically in the 20–70 kHz range, so the parts fuse without bolts, nails, solder or adhesives (TWI’s process overview). No consumables, no cure time, no solvents. Which is exactly why the geometry of the joint has to do so much of the work.

What a Good Ultrasonic Weld Design Actually Requires

The principal question this ultrasonic welding design guide answers is simple: what does a good joint need before the machine ever switches on? The answer is four things working together — a place for energy to enter, a place for it to stop, a controlled first-contact point that melts before anything else, and solid support directly under the seam. Get those right on the drawing, and parameter tuning becomes straightforward. Get them wrong, and no amount of force, time or amplitude will rescue the part.

The Machine Stack in Plain Language

Ultrasonic welding equipment converts electrical energy into mechanical vibration and applies it under pressure. The typical stack includes:

• Generator (power supply) — converts line power to high-frequency electrical output and regulates the cycle.
• Press or actuator — supplies the weld force and brings the tooling down onto the part.
• Converter / transducer — piezoelectric elements that turn electrical signals into mechanical vibration.
• Booster — amplifies or reduces amplitude and acts as the mounting interface for the horn.
• Horn (sonotrode) — the tuned tool that contacts the part and delivers vibration into it.
• Anvil / nest — backs the part up from below so the joint sees squeeze instead of deflection.
• Controller — sets and monitors time, energy, collapse and power limits.

The converter, booster and horn are machined as a matched set to resonate at one frequency — commonly 15, 20, 30, 35 or 40 kHz — and a production weld cycle is usually over in under one to two seconds (Wikipedia, ultrasonic welding). That speed is a large part of why the process shows up on high-volume lines.

Why Geometry Beats Settings

Because the cycle is so fast and the energy so precisely directed, the outcome is governed mostly by what the vibration meets when it enters the plastic: wall thickness, distance from horn to joint, whether an energy director exists, whether the anvil supports the seam. Weld parameters can fine-tune a sound design, but they cannot invent a joint feature that was never drawn. This is the same lesson welders learn elsewhere in the trade, whether they are working through a aluminum MIG welding guide or prepping base metal for an arc process: preparation and geometry decide the result long before the arc, or the horn, fires.

A Short History and Where the Process Is Used

Robert Soloff and Seymour Linsley patented ultrasonic welding of rigid thermoplastics in 1965, and in 1969 the first car made entirely of plastic was assembled using the process. From that starting point the technology spread into medical devices and disposables, automotive assemblies, electronics and connectors, packaging and blister sealing, and battery tab joining for cells and EV modules. Each of those industries uses the same physics; only the joint design changes.

What This Guide Covers Next

The sections ahead work through the physics you must design around, joint geometry such as energy directors, shear joints and tongue-and-groove features, near-field versus far-field distance rules, horn and anvil tooling, material weldability, process parameters, metal welding for foils and tabs, a troubleshooting catalogue of real mistakes, and a closing validation checklist. It is the sequence a design engineer actually follows — from first sketch to qualified production joint.

How Ultrasonic Welding Works: The Physics You Must Design Around

Before you sketch a single feature, you need a working model of what the machine does to your parts. From the operator’s chair it looks like sleight of hand: two halves are pressed together, a horn descends, there is a brief squeal, and the parts come out joined. No glue, no flame, no screws. TWI defines the process as joining materials under pressure using high-frequency vibrations, and that one sentence hides most of what a designer needs to know. The vibrations do real mechanical work on your geometry, and geometry decides whether that work ends in a strong joint or a scorch mark.

The Three Stages of Every Ultrasonic Weld

Every cycle, whether it runs in 0.2 seconds or two, moves through the same three phases.

1. Surface adaptation and initial contact. The press clamps the parts under static force before vibration starts. Nothing is perfectly flat at the microscopic level, so first contact happens only at the high points, called asperities. As force builds, those points deform and crush, real contact area grows, and the joint begins to seat. A warped part or a proud energy director changes everything downstream.
2. Vibration and energy input. The horn transmits axial oscillation, typically 20 to 70 kHz with amplitudes measured in tens of micrometres, into the part stack. Friction at the mating surfaces plus internal deformation converts that motion into heat, and a thin layer at the interface softens and flows.
3. Consolidation and hold. Vibration stops, but clamp force is held. The molten layer intermixes, penetrates surface texture, and cools under pressure into a monolithic joint. Releasing early invites weak welds and voids; holding too long just adds cycle time.

Where the Heat Actually Comes From

Two mechanisms generate heat, and they hand off control mid-weld. Interfacial friction dominates the first few milliseconds: surfaces rub at high frequency, asperity tips heat quickly, and surface texture, clamp force and flatness all shape the result.

Once the polymer nears its glass transition or softening range, viscoelastic dissipation takes over. Thermoplastics are viscoelastic, part spring and part thick syrup. Bend them rapidly and the chains slide against one another, turning vibration energy into heat inside the material. Research shows this dissipation scales with frequency, the material’s loss modulus and the square of the strain amplitude (analysis of viscoelastic heat generation). The consequence matters for designers: because heating accelerates as the material softens, melting stays concentrated at the joint instead of spreading through the part, which is what makes the process fast and cosmetically clean.

Plastic Welding vs Metal Welding: Two Different Kinds of Motion

This is the fork many designers miss. In plastic ultrasonic welding, oscillation runs roughly perpendicular to the faying surfaces. The parts are squeezed and shaken axially, the interface melts, and the joint forms by material flow and re-solidification. In metal ultrasonic welding, vibration is transmitted parallel to the interface as shear. There is no melting at all. Pressure plus shear ruptures brittle surface oxides and contaminant films, exposing clean metal that forms atomic-scale metallic contact with its neighbour. This solid-state mechanism, confirmed in the literature on ultrasonic metal joining (PMC7142629), keeps base metals below their melting point, which is why heat-sensitive assemblies survive.

That difference dictates your prep work. Fusion processes such as laser welding melt the base metal and can bridge dirty surfaces; ultrasonic metal welding cannot. Oxide, oil and hydration block clean contact, so aluminium and copper must be degreased and de-oxidised before welding, as covered in how to prep aluminium for welding. It also explains why only soft, thin metals qualify: the process relies on outer layers deforming enough to break through contamination, which thick or hardened sections resist.

The Energy Equation: E = F × f × A × t

Total ultrasonic energy delivered to a joint follows E = F × f × A × t, where F is weld force in newtons, f is frequency in hertz, A is amplitude in micrometres, and t is vibration time in seconds (ultrasonic welding overview). Read it as a design tool, not a formula to memorise. The four variables are partly interchangeable levers, since a weak weld can often be fixed by raising amplitude, extending time, or adding force, but they are not equal partners. Amplitude and time usually dominate, while brute force mostly risks crushing or marking the part.

Frequency sets the character of the process. A 20 kHz stack gives larger stroke and suits bigger parts and deeper reach; a 40 or 70 kHz stack gives finer, lower-amplitude motion for miniature electronic components. Your job is to give those levers a controlled place to act. Without a stress concentration, a stiff load path, or margin for collapse, adding energy does not build a weld; it smears the part. Every joule needs somewhere to go.

Joint Design for Plastic Parts: Energy Directors, Shear Joints and Tongue-and-Groove

In plastic ultrasonic welding, joint geometry beats brute force every time. No amount of extra weld pressure or cycle time will rescue a badly designed interface. The goal is straightforward: give vibrational energy a controlled place to concentrate, generate heat exactly where the bond should form, and do it all without distorting the surrounding part. That means building intentional features into the mating surfaces, and the most common of them is the energy director.

Energy Directors: Creating a Melt Zone

An energy director is a small, resin-rich protrusion molded into one mating surface that concentrates ultrasonic stress at its apex. Without it, flat mating faces fail to weld consistently, because vibration spreads across the whole contact area and never reaches melt temperature anywhere. By raising only a fraction of the surface, all initial pressure and shear land on that tiny feature, so it softens and collapses first, feeding a thin, controlled film of molten polymer across the seam.

Shape matters as much as presence. The most used forms are triangular, semicircular, rectangular/flat, and multiple EDs for longer seams. Research has shown that triangular energy directors gave better weld quality in virgin and 10% glass-filled polypropylene, while semicircular energy directors performed better at higher fibre loadings. In high-performance resins, tests have also shown that a 0.45 mm flat PEEK energy director produced nearly 50% higher lap shear strength than the same joint welded without an energy director at a 0.9 s weld time. Orientation is equally important: energy directors should run perpendicular to the direction of ultrasonic vibration so stress stacks up at each apex rather than sliding along the ridge. For the underlying mechanics, the TWI technical FAQ on ultrasonic welding explains how this stress concentration initiates the weld.

Energy-Director Joints vs. Shear Joints

Not every plastic responds to vibration the same way. Energy-director (butt and lap) joints are the standard choice for amorphous plastics such as ABS or polycarbonate, whose broad softening range lets melted resin flow into the joint with little resistance.

Semi-crystalline resins behave differently. Polypropylene, nylon and acetal have a sharp melting point and a damped, spring-like response to vibration, so the melt forms quickly and resists flow. For these materials, shear joints are recommended, particularly for structural or hermetic sealing. Instead of a raised director, a shear joint uses an interference fit between a male and female feature, and the ultrasonic vibration runs parallel to the overlapping interface. Friction builds along the fitted walls until they melt and fuse under pressure, producing a tighter, leak-resistant seal that is far more forgiving of damping. This distinction is echoed in the broader explanation of plastic welding methods, where shear-style motion is key to consistent fusion for these resins. For demanding, leak-proof assemblies such as those found within aerospace welding applications, this seal integrity is often non-negotiable.

Secondary Features: Alignment, Flash Control, and Melt Distribution

Once the fusion style is chosen, secondary features protect the weld while it forms. Tongue-and-groove and step joints lengthen the sealing path, so any leak has further to travel. Crash/dovetail features and molded ribs hold the parts in register throughout vibration, and molded flash traps capture excess melt squeezed out during collapse so it cannot mar the visible surface or jam moving assemblies.

All of these work together with the energy director’s geometry. A properly sized director dictates collapse behaviour, melt distribution, void risk, and weld-line appearance. One that is too large generates excessive flash and a brittle, over-welded joint; one that is too small risks an incomplete, starved weld with no real bond.

Practical Do/Don’t for Joint Design

Do	Don’t
Match the director to your resin: triangular for virgin or lightly filled PP, semicircular for higher fibre loadings.	Rely on flat, unfeatured mating faces to weld consistently.
Use a shear joint for semi-crystalline resins and for any structural or hermetic seal.	Force an energy-director butt joint onto a heavily damped semi-crystalline part and expect the same process window as an amorphous resin.
Add flash traps, ribs, or tongue-and-groove features for registration and melt containment.	Leave a long, unsupported seam with no means of registration, which amplifies tolerance stack-up.
Size the director for controlled collapse, so melt flows rather than being squeezed out instantly.	Over-size the energy director, which increases flash and the risk of a brittle joint.

The joint is essentially the moldable blueprint for the whole process. Size the directors for the resin, match the fusion style to the material’s melt behaviour, and add features that hold everything in register, and the process window widens dramatically. Drop any one of those pieces, and no amount of parameter tuning will make up for it.

Near-Field vs Far-Field Welding and the Wall Thickness Rules

If you remember one number from this entire ultrasonic welding design guide, make it 6.35 mm, or a quarter inch. That dimension separates near-field from far-field welding, and it decides whether a joint fuses cleanly or refuses to take a weld. It is also the fastest way to catch a bad design on paper rather than on the production floor.

Ultrasonic welding sends mechanical vibration through solid material until it reaches the joint interface, where friction and localized softening create the bond (TWI’s process overview). Every millimetre traveled before the seam costs energy: scattered at wall transitions, absorbed by soft or thick sections, diverted into ribs and bosses that were never meant to join.

What Near-Field and Far-Field Actually Mean

The classification is purely geometric. It describes the distance from the horn’s contact face, the surface the sonotrode touches, to the joint interface. It is not overall part height or top-to-bottom distance.

Mode	Distance from horn face to joint	Best-suited materials	Design implication
Near-field	0.25 in (6.35 mm) or less	Soft, flexible and porous thermoplastics; semi-crystalline resins such as nylon, polypropylene, polyethylene and acetal	Energy reaches the seam with little loss, so the joint tolerates damping and sharper melt behaviour
Far-field	Greater than 0.25 in (6.35 mm)	Rigid, amorphous resins: ABS, polycarbonate, polystyrene, SAN, acrylic	Vibration must travel through the part wall, so the path needs to stay stiff, thin and uninterrupted

The material logic matters more than the table. Soft and porous polymers convert vibration into heat internally and damp it quickly, so the horn has to sit close to the action; near-field welding is the recommendation for these resins. Rigid amorphous polymers transmit vibration efficiently over distance and tolerate a wide amplitude range, which makes them the forgiving choice for far-field joints.

Wall Thickness: Keep It Uniform and Keep It Under 3 mm

Wall thickness is where far-field designs live or die. Aim for uniform walls along the entire energy path. Abrupt steps, thick transition bands and heavy bosses near the seam all act as energy sinks, absorbing amplitude before it reaches the interface. A boss that looks harmless in CAD can double the local section and starve the joint of input.

There is also a practical ceiling. With standard ultrasonic equipment, weldable thickness is currently limited to roughly 3 mm, because vibration cannot effectively penetrate thicker sections (Ultrasonic welding reference). This is a hardware limit, not a settings problem. No amount of extra weld force will push useful amplitude through a 6 mm slab.

The mental model is simple: stiff, thin, rigid sections transmit vibration; thick, soft or heavily damped sections absorb it (plastic welding overview). Design the path accordingly.

Part Depth, Rib Placement and Horn Reach

Distance on the drawing is only half the equation. The other half is whether a horn can physically get there. Deep drawers, tall side walls, undercuts and snap-fit features can block the sonotrode long before the joint interface becomes a problem. Ribs placed close to the seam are a second trap: they stiffen the wall locally, but they also siphon amplitude sideways and can leave the weld line under-driven while the rib tips chatter.

Practical ways to fix a long or blocked path:

• Move the joint, not the material. Raise the seam into a shallow step so the interface sits within 6.35 mm of the horn face.
• Thin the path. Hold a constant, modest wall between horn and joint; draft the wall rather than stepping it.
• Relocate ribs and bosses. Keep them out of the direct energy corridor or gusset them away from the seam.
• Recontour the horn. A stepped or contoured sonotrode can reach past tall walls, but a booster cannot fix an obstructed path.
• Switch joint style. A shear joint collapses material progressively and tolerates deeper sections better than a simple energy-director butt.
• Split the operation. Two accessible near-field welds often beat one heroic far-field weld.

If none of those work, ultrasonic welding may be the wrong tool for that geometry. Thick-walled castings and metal assemblies follow entirely different rules; for a sense of how much surface condition governs success in aluminium work, see Aluminum Cleaner for Welding: How to Prep Aluminum Right.

Run the check early: measure the horn-to-joint distance, match it against the resin family, then audit wall thickness along that path. Before tooling, a fix costs a CAD revision. Afterward, it costs a mold change.

Designing the Horn, Anvil and Fixture Around Your Part

You can draw a textbook energy director, specify the resin correctly, and still end up with a weak or ugly weld if the tooling does not cooperate. In practice, the horn, anvil and fixture are half of the joint design. The part drawing decides where energy should go; the tooling decides whether it actually gets there. Treat them as one design task, not two.

Horn (Sonotrode) Principles

The horn, or sonotrode, is the part of the stack that touches your product. It has four jobs, and each one is a design constraint:

• Reach the weld line. The horn face must physically arrive at the joint. Tall side walls, deep bosses, undercuts and nearby obstructions all limit reach before anything else matters.
• Match the part surface. If your seam is curved or stepped, the horn face must be contoured to match it, otherwise contact is partial and energy enters unevenly.
• Avoid crushing what it touches. A horn that lands on ribs, thin walls or cosmetic features will mark or deform them long before the joint melts.
• Resonate with the stack. Horns are machined to a specific resonant length for a specific frequency. A horn cut for 20 kHz will not run correctly at 15 kHz; changing frequency means new tooling, not a setting change.

The ultrasonic generator behind it all converts mains power, commonly cited around 5,000 W at 50–60 Hz into a 20–40 kHz output, with 20 kHz the most widely used welding frequency. That electrical output becomes mechanical swing through the transducer, is scaled by the booster, and is shaped by the horn.

Horn material choice follows the same logic. Titanium gives excellent fatigue life and handles high-duty cycles, aluminium is lighter and cheaper with good amplitude transfer but wears faster, and steel survives abrasion and clean-room wear better at the cost of higher internal losses. Critically, amplitude is not a free dial on the machine: it is set by the booster ratio and the horn’s gain geometry. Designers who need more amplitude at the part are really asking for different tooling, which is why amplitude requirements should be settled at the drawing stage.

The Anvil and Nest: Backing the Joint

The anvil (often a custom nest) is the anvil-side partner of the horn, and it is the most overlooked item on the drawing. Its job is simple to state: back up the joint line directly so the joint sees squeeze instead of deflection. If the anvil supports the part an inch away from the seam, the wall flexes under weld force, vibration leaks into the structure, and the energy director never gets the concentrated stress it needs.

A good nest replicates the part’s shape for support, stays rigid and flat, and remains parallel to the horn face throughout the stroke. It should also be wear-resistant, since a nest that wears into a groove will shift collapse readings over time and quietly erode process capability.

Fixtures, Clamping and Noise Control

Fixtures hold alignment before, during and after the weld. Parts must not slide while vibrating: slippage shows up as polished witness marks, asymmetric flash and scatter in collapse distance. Large parts and assemblies hanging near the machine deserve special attention. Sections like housings with long attached panels can vibrate at sub-harmonics of the driving frequency, creating loud audible ringing. Clamping those parts at one or more locations dampens the vibration and stabilises the weld.

Do not overlook acoustics either. Airborne output at 15–20 kHz is squarely in the range that damages hearing over time, so enclosures, curtains and proper hearing protection belong in the fixture design conversation from day one, alongside guarding that keeps hands clear of the descending press.

One last habit worth building: keep the aluminium horn face as disciplined about cleanliness as any workpiece prep routine, since residue or galling changes how energy couples into the part, as covered in Aluminum Cleaner for Welding: How to Prep Aluminum Right. When horn, anvil and fixture are designed with the joint rather than after it, most “parameter problems” never appear at all.

Material Selection and Design Rules That Decide Weldability

Before a single energy director goes on the drawing, answer an earlier question: can this pair of materials even be welded? Ultrasonic welding is forgiving about geometry and unforgiving about chemistry. A well-designed joint in an incompatible pair will still fail, and no parameter tweak on the floor will rescue it.

Amorphous vs Semi-Crystalline: The Two Behaviours That Matter

Polymers behave very differently under ultrasonic vibration, and the split falls almost perfectly along the amorphous/semi-crystalline line.

• Amorphous resins (ABS, polycarbonate, polystyrene, acrylic, PVC) soften over a broad temperature range rather than melting at one point. Their random molecular structure lets vibration travel through the part with relatively little loss, so they tolerate far-field joints and a wide amplitude window. They are the easiest materials to design for.
• Semi-crystalline resins (nylon, acetal, PBT, PET, polypropylene, polyethylene) have ordered molecular regions that behave like tiny springs, absorbing and damping vibration before it reaches the joint. They also melt sharply at a distinct temperature, which leaves less margin for error. In practice this means higher amplitude, near-field placement, or a shear joint rather than a simple energy-director butt.

That is why designers move the joint close to the horn whenever the material is semi-crystalline or flexible, and why the process window stays narrow: a small amplitude increase jumps from too cold to burned through very quickly.

Dissimilar Materials, Moisture and Contamination

Joining two different thermoplastics is possible but only within limits. As a working rule, the two resins should be chemically compatible and their melting temperatures should differ by no more than about 22 °C. Beyond that gap, one side liquefies while the other is still solid, and the interface never builds a continuous bond. When in doubt, check a compatibility chart or run a trial weld on scrap parts before committing to tooling.

Moisture is the quiet saboteur. Water trapped at the joint interface flashes to steam at 100 °C, well before most engineering resins melt, producing a porous, foamy weld that looks fine from outside but fails leak and tensile tests. Hygroscopic materials such as nylon and PET must be dried before welding, and parts should run soon after leaving the dry box. Surface chemistry matters too: mold-release agents, plasticizers, silicone coatings and rubber impact modifiers all form a weak boundary layer that blocks energy transfer. Specify a release-free mold texture on mating faces, and keep painted, printed or oily surfaces off the mating line.

Fillers follow a curve rather than a rule. Rigid fillers such as glass fibre or mineral actually improve ultrasonic transmission up to roughly 20 percent loading, because stiffness carries vibration efficiently. Past that point there is not enough free resin left to flow and form the bond, and weld strength drops. Check the total loading on a filled grade; near the upper end, use a resin-rich joint zone or a shear joint that forces flow locally.

Metals: Hardness, Oxide, Conductivity and Cleanliness

For metal joints, four variables decide the outcome before tooling is cut. Hardness limits what standard equipment can deform: soft, ductile metals such as aluminium, copper, nickel and thin foils weld well, while hard or thick sections simply will not move. Oxide layers must be broken up mechanically so clean metal reaches atomic contact, which is why knurled horn faces and generous weld force matter. Thermal conductivity drains heat away from the interface, so high-conductivity metals need more energy delivered faster; the same physics that makes aluminium tricky in arc processes, covered in Aluminum MIG Welding: The Complete Practical Guide, shows up here in another form. Cleanliness is a design requirement, not a shop-floor afterthought: oxide, oil and drawing compounds contaminate the bond line, so the drawing should call for a solvent or abrasive wipe as described in Aluminum Cleaner for Welding: How to Prep Aluminum Right.

The process definition and its limits are summarised by TWI’s ultrasonic welding FAQ, while polymer behaviour appears in this open-access study on ultrasonic welding of polymers and the general overview of ultrasonic welding.

Process Parameters and How Design Creates a Usable Process Window

Once the joint geometry is right, tuning feels easy: a handful of variables, a stack light, and an afternoon of trials. The catch is that a poorly designed part compresses those variables into a sliver of settings that works, usually only in the lab. A well-designed part gives you room to move. That room is the process window, and it is created on the drawing long before anyone touches the controller.

The Five Parameters Engineers Actually Tune

Every ultrasonic weld cycle comes down to five levers, and each one is really a proxy for a geometric decision you already made:

Parameter	What it controls	What the drawing decides
Amplitude	Peak distance the horn travels, typically 10 to 100 micrometres	Set by booster and horn geometry, but an energy director determines how much of it the joint actually feels
Weld force (pressure)	How hard the parts are squeezed during vibration	The anvil must sit directly under the seam, or force goes into deflecting the part instead of crushing the joint
Weld time or energy	How long vibration runs, or how many joules are delivered	Stiff, uniform sections transmit energy; thick or damped sections swallow it and stretch the cycle
Hold (solidification) time	Pressure held after vibration stops	Needs a flat, rigid backup so the joint sets without being disturbed
Control mode	The rule that ends the weld	Collapse-distance control only works if the geometry creates a measurable crush event

Amplitude and Time Do the Heavy Lifting

Pressure gets blamed for weak welds more often than it deserves. In glass-filled polypropylene trials, weld pressure had a comparatively minor effect on weld quality when measured against weld time, amplitude and energy-director type. That corrects a very human instinct: when a weld is weak, do not reach for more force. Extra force typically pushes the part out of parallel, marks the surface, or bottoms the joint before the resin has flowed. The basic energy relationship (energy = force × frequency × amplitude × time) makes the point: the four inputs are interchangeable contributors to one result, so the geometry has to give them a controlled place to act.

Amplitude decides how aggressively surface asperities are sheared; time decides how long that energy keeps feeding the joint. Too much of either produces flash, non-uniform melting, polymer degradation, and fibre breakage in composites. More energy is not automatically better: in a glass-fibre nylon study, raising weld energy from 200 J to 1000 J increased bond efficiency, yet pushing beyond that reduced strength as pores formed from horn indentation. Every design has a peak; the job is to find the broad flat region around it, not the summit.

Control Modes and the Energy Director’s Crush Signal

Four control modes are standard on modern machines, each with a different tolerance for part variation:

• Time control runs vibration for a fixed duration. Simple, but it does not react to how much the joint actually collapsed.
• Energy control delivers a fixed number of joules, absorbing some part-to-part friction variation.
• Collapse distance control ends the weld when the parts have travelled a set distance together.
• Peak power control stops on a power spike, a reliable sign the joint has bottomed out.

This is where joint design becomes process control. A triangular energy director gives a crisp, repeatable displacement signature: a gentle ramp as the tip contacts, a steeper yield as it collapses, then a plateau when the flat faces meet. That knee is a measurable signal the controller can key on. Design a perfectly flat mating face instead and there is no knee at all, which quietly removes collapse control from your toolbox.

Over-Welding, Under-Welding and the Collapse Margin

Over-welding shows up as flash on cosmetic faces, horn witness marks, discoloured or brittle resin at the seam. Under-welding shows up as a starved weld line and an incomplete seal that leak testing catches weeks later. Both are usually geometric answers rather than parameter answers.

The fix for over-welding is a collapse margin: deliberately reserved space so that molten energy-director resin has somewhere to go without being squeezed onto a visible surface. Ribs, flash traps and a small offset between the parting line and the joint provide that room. For deep sections, a shear joint is the better answer, because it collapses progressively along the overlap instead of meeting everything at once, giving consistent crush and a stable power curve through the whole cycle.

A quick symptom-to-design map:

Symptom	Check first	Design fix
Weak or inconsistent weld, little collapse	Amplitude, contact sequence	Make the energy director the first and highest contact along the entire seam
Flash on the cosmetic face	ED volume, collapse distance	Reduce ED size, add a flash trap, widen the collapse margin
Parts bottom out, witness marks	Thick section under the horn	Add relief, even out wall thickness behind the joint
Brittle, discoloured weld	Amplitude and energy too high	Drop amplitude, shorten weld time, re-check filler and moisture content
Erratic collapse, power spikes	Anvil support	Back up the joint line directly so squeeze is not deflection
Incomplete seal at corners	Distance to horn	Bring the joint into near field or switch to a shear joint

Contamination behaves the same way at the parameter level: mold release, plasticizer or surface oil raises the energy a joint needs until the window disappears. The discipline is identical on the metal side, which is why surface-prep resources like Aluminum Cleaner for Welding: How to Prep Aluminum Right read like a checklist you could borrow for plastics.

The practical lesson of this stage: parameters can optimize a joint, but they cannot rescue one. Set the geometry so the melt happens where you want it, give the displaced material somewhere to go, and the controller will find a wide, forgiving window on its own.

Designing for Ultrasonic Metal Welding: Foils, Wires, Battery Tabs and Dissimilar Metals

Plastic ultrasonic design is about managing melt flow. Metal ultrasonic design is about managing contact. That single distinction changes almost every decision on the drawing, because ultrasonic metal welding is a solid-state process: the temperature at the interface stays well below the melting point of either material, and the joint forms when high-pressure shearing disperses the surface oxides so clean metal touches clean metal and bonds at atomic level. Nothing puddles. Nothing solidifies from a liquid. If your design assumes a weld pool, you are designing for the wrong process.

Why Only Thin, Soft Materials Qualify

Because bonding comes from displacing surface layers rather than from bulk heating, the materials must be thin enough and soft enough to deform under the horn. In practice this means small welds of thin, malleable metals such as aluminium, copper and nickel: foils, wires, ribbons, precision strip and light sheet. Thick structural sections, hardened alloys and stiff high-strength stacks are out of reach for standard 20–40 kHz equipment — the horn simply cannot scrub the interface hard enough before the machine runs out of force or stroke.

• Good candidates: battery tabs, cell foils, wire splices, sensors, lead frames, thin busbar laminations, shielding cans.
• Poor candidates: heavy plate, thick-walled tube, precipitation-hardened alloys, heavily work-hardened strip.

The engineering upside is impressive: linear vibration systems have successfully joined stacks of 60 layers of copper foil for battery and energy-storage work, a feat no fusion process handles cleanly. The constraint is equally clear — you are limited to cross-sections the tooling can compress uniformly.

Geometry the Horn Actually Sees

Unlike a plastic energy director, you are not designing a stress riser. You are designing a flat, backed, accessible lap.

1. A stack that presses flat. All layers must contact each other before the horn contacts the top layer. Any bowed foil, burr or curl creates a gap that never closes, and the weld stalls at the first unbonded interface.
2. A knurled or patterned anvil. A textured anvil surface traps material, prevents slip and forces the shear to happen inside the joint rather than at the horn face. Smooth-on-smooth stacks tend to skate instead of bond.
3. Lap area sized to the load path. Undersized overlap gives a weak joint; oversized overlap demands more energy than the stack can absorb without crushing edges. Design the overlap from the mechanical load, then check it against horn coverage.
4. Two-sided access. Both horn and anvil need a straight approach. On a wire harness or busbar, if the anvil cannot sit directly beneath the weld zone, the part flexes instead of shearing and the weld never initiates.

Cleanliness Is a Design Requirement

Oxide, oil, drawing compound and even fingerprint moisture all sit between the two metals and must be broken up or expelled. That makes surface preparation part of the drawing, not a shop-floor afterthought: specify cleaned, protected surfaces, a maximum storage time before welding, and handling with gloves. For aluminium specifically, the prep chemistry matters, so follow a proven process such as aluminum cleaner for welding: how to prep aluminum right rather than improvising with solvents.

Dissimilar Metals and Foil Stacks

Aluminium-to-copper joints are the backbone of modern battery and interconnect assembly, and ultrasonic welding is one of the few processes that joins them without brittle intermetallic growth from a melt pool. Design rules for these joints are specific: keep the softer metal against the horn where possible, size the lap generously because the interface is the weak link, and avoid geometry that concentrates peel force at the weld edge. With multi-layer stacks, arrange foils symmetrically, keep total thickness within the machine’s rated range, and remember that inner layers are the last to see displacement — if they slip, the outer layers look perfect and the stack still fails an electrical test.

Get those four things right — flat stack, trapped shear, sized lap, two-sided access — and the process parameters become a tuning exercise rather than a rescue mission.

Common Design Mistakes, Troubleshooting and How to Fix Them

Field troubleshooting has a habit of ending at the machine: turn up the amplitude, add pressure, run it longer. More often than not, the real answer sits in the part. Weak or missing welds trace back to a handful of design causes more than any others — no feature to direct energy, a joint placed too far from the horn for that resin, a contaminated or wet interface, and no anvil backing directly under the joint line. Ultrasonic welding joins materials through high-frequency, low-amplitude vibration applied under pressure, and as TWI’s process overview explains, the energy has to be delivered to a specific interface to work. The table below maps the ten mistakes we see most, with the fix for each.

Ten design mistakes, symptom by symptom

Mistake	Symptom	Root cause	Design fix
Flat mating faces, no energy director	No visible weld line, thin or absent bond	Energy spreads across the whole surface	Mold a triangular or semicircular energy director on one face
Semi-crystalline joint in far field	Weak, inconsistent weld on long seams	Semi-crystalline resins damp vibration	Bring the joint within 6.35 mm of the horn or use a shear joint
Non-uniform wall thickness	One area welds, another flashes	Thick sections absorb energy, thin ones transmit it	Even out walls; step thickness away from the seam
Horn cannot reach the seam (tall walls, undercuts)	Horn strikes walls, cosmetic damage, no weld	Tooling never contacts the joint area	Lower the weld-zone wall, add horn clearance, change the parting line
No anvil support under the joint line	Parts flex, collapse varies, underside marks	Joint sees deflection instead of squeeze	Back up the joint line with the nest directly beneath it
Dissimilar resins with mismatched melt ranges	One side melts, the other stays dry	Melt temperatures do not overlap	Match resins or specify a compatible, closely melting pair
Ignoring moisture, mold release, plasticizer	Brittle weld, bubbling at the seam	Contamination blocks energy transfer	Dry the resin, drop release agents, avoid plasticized or heavily impact-modified grades
Weld line too close to the edge	Flash off the edge or short shot at the seam	Too little material to contain the flow	Inset the energy director and add a flash trap
Expecting more force to fix a weak weld	Crushed features, marking, still weak	Pressure matters less than weld time, amplitude and ED type	Fix the geometry, then tune amplitude and time
No crush margin	Parts bottom out, collapse stalls early	Molten material has nowhere to go	Leave crush space and a flow relief or flash trap

Symptom-based troubleshooting: parameter fix or design fix?

When a weld goes wrong on the floor, split the response into two columns. Parameters are the quick test; design answers are the permanent ones.

Symptom	Parameter answer	Design answer
Weak or no weld	Raise amplitude slightly, extend weld time	Add an energy director, shorten horn distance, add anvil backing
Flash	Reduce time or force	Add a flash trap, shorten the ED, thicken the wall near the seam
Cosmetic marking	Lower pressure, soften nest contact	Support the marked area, redesign nest footprint
Brittleness	Reduce amplitude to stop degradation	Dry the resin, remove contamination, check material pairing
Voids at the joint	Extend time, switch to energy mode	Lower ED height, add venting and flow space
Inconsistent collapse	Use collapse-distance or energy control	Even the walls, tighten ED height tolerance, fix stack-up

When a weld still looks wrong, cut it. A cross-section shows wet-out, voids, ED crush and bond-line consistency that no power curve reveals — and destructive evaluation is standard practice for thermoplastic welds. Bend, tensile and impact test methods are defined in DVS 2203, while AWS G1.6 sets plastic welding inspector qualification levels, renewed every three years. On the metal side, contamination is the usual culprit, and aluminum cleaner for welding shows how proper surface prep before joining changes results. If you are still chasing a stubborn weak weld after geometry, moisture and tooling are confirmed, the section on aluminum MIG welding techniques is a useful reminder of the same principle in another process: preparation beats parameter.

Ultrasonic Welding vs Other Joining Methods, Validation and a Design Checklist

No joining process wins every job. A useful ultrasonic welding design guide has to be just as honest about when ultrasonic welding is the wrong tool as it is about how to design for it.

Where Ultrasonic Welding Wins, and Where It Does Not

Ultrasonic welding is one of the fastest joining methods available. Cycle times often fall well below one second, and computer disk welds have been reported at under 300 ms (Wikipedia, Ultrasonic welding). There is no cure time, no surface activation, no consumables and no bolt holes, which is why it dominates high-volume plastics assembly and battery tab joining.

Method	Cycle time	Joint preparation	Tooling cost	Thickness / material fit
Ultrasonic	Under 1 s typical	Molded energy director or lap	Moderate (horn + nest)	Thin walls, foils, tabs
Adhesive bonding	Seconds to cure minutes	Cleaning, priming, fixtures	Low tooling, high labour	Thick, dissimilar, large area
Mechanical fastening	1–5 s per fastener	Holes, bosses, inserts	Low	Thick and serviceable joints
Hot plate / spin	Roughly 1–20 s	Flat faces or rotational symmetry	Low to moderate	Large parts, rods and tubes
RF / high frequency	Seconds	Electrode tooling	High	Polar plastics, coatings, films
Laser / induction	Fast, point by point	Clean surface, line-of-sight	High capital	Metals, fine detail

Two contrasts matter most when you are choosing a process. Spin and friction welding run at roughly 100–300 Hz with 1–2 mm of amplitude, while ultrasonic welding runs at 15–40 kHz with very low amplitude, so the movement direction and the joint geometry each process wants are completely different (Wikipedia, Plastic welding). And while fasteners still rule thick structural assemblies, mechanical fastening can account for 19–42 percent of aircraft assembly cost, with Boeing reporting labour savings greater than 61 percent from fusion bonding a composite wing structure versus a bolted one — context worth knowing if you work in aerospace welding.

Ultrasonic is the wrong choice when sections are very thick, when the resin is a thermoset, when the horn physically cannot reach the joint, or when a soft, heavily damped material sits far from the horn face. For fine metallic detail and line-of-sight work, laser processes remain the better answer, as covered in Alpha Laser Welding: Systems, Uses, and Safety Guide.

Validation: Prove the Joint Before the Line Runs

Validation works on three levels:

• Destructive tests: tensile and shear pull tests, peel tests on lap joints, impact/drop tests, and burst or leak testing for sealed enclosures.
• Non-destructive checks: visual flash inspection, leak testing, and ultrasonic testing of the joint. Be aware that conventional ultrasonic NDT struggles with attenuative, semi-crystalline plastics, so it works better as a trend check than a absolute pass/fail on polymers (TWI, What is ultrasonic welding).
• In-process monitoring: collapse distance, power/time curves, peak-power limits and, increasingly, machine-learning systems that classify each weld from its signature in real time.

12-Point Design-for-Ultrasonic-Welding Checklist

1. Is the resin thermoplastic, and amorphous or semi-crystalline?
2. Is the joint within 6.35 mm (0.25 in) of the horn face if it is soft or semi-crystalline?
3. Is there an energy director, shear joint or intentional stress-raiser at the interface?
4. Does that feature make first and highest contact everywhere along the seam?
5. Are walls uniform, with no abrupt steps or thick bosses near the weld line?
6. Can the horn reach the seam with clearance for tall walls and no undercut traps?
7. Is the anvil directly backing the joint line so the seam sees squeeze, not deflection?
8. Are alignment features (tongue-and-groove, step, crash ribs) holding register?
9. Is there a flash trap or crush margin so molten material has somewhere to go?
10. Are contaminants controlled: moisture, mold release, plasticizers, plasticized coatings?
11. Does the process window have room in amplitude, force, time and hold before cosmetic damage appears?
12. Is there a validation plan specifying destructive, non-destructive and in-process checks?

This ultrasonic welding design guide began with a housing that looked perfect and failed anyway, and the lesson is unchanged: the machine rarely fails, the drawing does. Get energy into the joint, stop it there, melt the right material first and support the seam, and the process window takes care of itself.

Frequently Asked Questions

What is the most important rule in ultrasonic welding design?

Concentrate energy at the joint and let it stop there. Use a molded energy director or shear joint, keep the joint close to the horn for soft or semi-crystalline resins, and back the seam with a rigid anvil so the vibration does work instead of flexing the wall.

What is the difference between near-field and far-field ultrasonic welding?

Near-field means the joint interface sits 0.25 inches (6.35 mm) or less from the horn contact surface; far-field means it sits farther away. Soft, flexible and semi-crystalline materials generally need near-field geometry, while rigid amorphous plastics such as ABS and polycarbonate tolerate far-field.

Do I always need an energy director?

Almost always on flat joints, because two perfectly flat faces give ultrasonic energy nowhere to concentrate. Shear joints, tongue-and-groove features or textured interfaces can replace a traditional energy director when you need a hermetic or semi-crystalline weld.

How long does a typical ultrasonic weld take?

Weld cycles are commonly well under a second to about two seconds including the hold phase, which is why the process suits high-volume automated lines. Actual time depends on resin, weld area, amplitude and whether you control by time, energy or collapse distance.

Does ultrasonic welding melt the metal being joined?

No. Ultrasonic metal welding is a solid-state process: pressure and shear break up surface oxides and create metallic bonding while the temperature stays well below the melting point of the base metals. That is why it works well for dissimilar pairs like aluminium and copper.

Why is my ultrasonic weld weak even at maximum power?

Power cannot compensate for geometry. Common causes are missing energy-directing features, a joint too far from the horn, moisture or mold release at the interface, unsupported walls that flex, or a joint stack that bottoms out before the melt forms.

Which plastic materials are hardest to weld ultrasonically?

Semi-crystalline resins such as nylon, polypropylene, PET and PEEK are the most demanding because their sharp melting point and spring-like damping absorb vibration before it reaches the interface. They typically need near-field placement, higher amplitude or a shear joint design.

How do I verify that an ultrasonic weld design is good?

Combine process monitoring with testing: watch collapse or energy curves for consistency, then cut sections and run destructive checks such as tensile, peel, impact or burst tests, plus visual or leak testing for sealed parts. Requalify whenever resin, tooling, amplitude or moisture conditions change.

When should I choose another joining method instead?

Consider adhesives, hot-plate, spin, laser or mechanical fastening when sections are too thick, the material is a thermoset or heavily damped, the horn cannot reach the joint, or you need large structural seams. Ultrasonic welding is strongest for small, thin, repeatable, automated joints.