Ultrasonic Wire Splice Welding: Wire Stack-Up, Process Window and RFQ Evidence

An ultrasonic splice is not defined by total conductor area alone. Two joints can both total 10 mm^2 yet behave differently. One may use two equal copper conductors. The other may combine fine wires, different strand classes, tin plating, an aluminum branch, uneven strip lengths or a different stack order. A recipe that makes one combination pass cannot be assumed to qualify the other.

For a comparable RFQ, freeze the splice family, wire stack-up, prepared-wire condition, weld-nugget geometry and tooling set. Also freeze the machine and software identity, validated process window, monitoring limits, destructive-test plan and traceability record. Ask how the nominal recipe was developed, how boundary samples were challenged and how tool wear and material changes are controlled. Require a clear reaction when a production cycle alarms.

Ultrasonic metal welding is a pressure-and-friction joining process. Schunk Sonosystems explains that its converter, booster and sonotrode create and transfer high-frequency mechanical oscillation. Its official technology page says the metals remain in the solid state and identifies aluminum, copper and their alloys as common application materials. TELSONIC similarly describes pressure plus high-frequency relative motion and lists wire splices among the applications. These manufacturer explanations establish the process principle, not suitability for a particular splice. Schunk ultrasonic metal-welding principle and TELSONIC metal-welding technologies.

This guide addresses bare-conductor ultrasonic splices inside cable and wire harnesses. It complements the sealed-splice RFQ guide. That guide covers the environmental seal around a completed branch. The conductor strand-class guide covers conductor construction and downstream interfaces. This article does not prescribe a universal weld recipe or acceptance value. It does not claim an unverified SINAWATTS machine, laboratory, process approval, material, test capability, certification, volume, price, MOQ or lead time.

Direct answer: what should an ultrasonic wire-splice RFQ require?

Require one controlled splice specification for every approved wire combination. At minimum, each return should identify:

• harness part number, splice ID, drawing revision and electrical function;
• wire manufacturer and part number, material, plating, nominal area, strand construction and insulation;
• number of conductors entering each side, branch direction, order in the weld pocket and permitted orientation;
• cut and strip dimensions, insulation setback, exposed-strand overlap and allowable strand disturbance;
• total nominal metallic area while preserving the identity of every constituent wire;
• machine manufacturer, model, generator, frequency, actuator and controller/software revision;
• sonotrode, anvil, gather tool and insertion-aid part number, serial or revision and approved working surface;
• weld width, final height or deformation, length and location relative to insulation;
• nominal amplitude, force or pressure, energy, time and trigger or end-condition settings;
• upper and lower monitoring limits and the response to each alarm;
• development trials and boundary samples used to establish the process window;
• visual, dimensional, pull, electrical and metallographic evidence required at qualification;
• environmental and durability validation tied to the actual equipment duty;
• start-up, changeover, periodic verification and tool-life controls;
• production record fields and retention period;
• repair policy and disposition of a rejected weld; and
• a signed exception list covering every item that is proposed rather than proven.

Do not accept "ultrasonic welded per standard" as the complete definition. The order must state the applicable IPC or customer document revision and product class where relevant. It must also state drawing precedence, sampling frequency, test method and numerical acceptance criteria.

Why total cross-sectional area is not a recipe

Total area helps select an initial machine and tooling range, but it hides variables that affect coupling and deformation. A splice of 2 x 5 mm^2 is not automatically equivalent to 10 x 1 mm^2. The smaller wires have more outside surface, more strand interfaces and a different stack geometry. A fine-stranded flexible conductor does not necessarily behave like a coarse-stranded conductor of the same nominal area. Tin-coated copper, bare copper and aluminum present different surfaces. Wire hardness, temper, strand lay, coating thickness, oxide condition and supplier tolerances can also change the response.

Create a constituent-wire table instead of only a "total mm^2" field:

On small screens, swipe or scroll sideways to read every column. Keyboard users can focus the table and use the arrow keys.

If alternates are allowed, state whether each alternate needs a separate recipe, equivalence study or full requalification. "Same gauge" is not equivalence evidence.

Freeze the stack-up and loading orientation

The weld pocket must receive the intended conductors in the intended three-dimensional arrangement. Flattening wires side by side can create a wide joint with limited bonded interfaces. A misplaced small wire can escape the active zone. Excessive overlap can form flash, while too little can leave loose strands. Insulation too close to the active tool can soften or become trapped. An excessive setback can expose an unsupported conductor length.

Branson's official Ultrasplice 40 instructions show conditions for an excessive gather gap, an incorrect side splice and a good splice. The manual says a pull test should challenge the smallest-gauge wire. When all wires are equal, it points to the wire nearest the anvil. It also lists broken or loose strands and excessive insulation damage as rejection conditions under that equipment guidance. Use these points as examples of what an RFQ must define, not as universal limits for another machine. Branson Ultrasplice 40 instructions.

A loading control should include:

1. a cavity or fixture that supports the approved stack without forcing strands out;
2. a diagram from the operator viewing direction;
3. distinct instructions for inline, end, Y and multi-branch joints;
4. insulation-stop or insertion references;
5. confirmation that conductor ends reach the controlled overlap zone;
6. prevention or detection of missing, wrong-gauge and reversed-side wires;
7. criteria for strand splay, nicked strands and contamination before welding; and
8. a completed-splice image showing acceptable geometry.

Schunk's official Cable Fit-In page says overlap length and distance from the weld area to insulation are influenced by manual insertion. It describes a configurable insertion aid for those dimensions. This supports treating placement as a controlled input, but it does not require that accessory. Schunk Cable Fit-In insertion guidance.

Define weld geometry before developing settings

A recipe needs a target joint shape. Record weld-pocket width, weld length, pre-weld stack height, final height or deformation and insulation setback on every conductor. Define where dimensions are measured and whether loose fringe strands count.

Geometry confirms that the intended metal volume entered the tool. It can reveal a missing or misloaded wire and gives a repeatable specimen for testing. It is not a substitute for bond evidence. A final height inside limits can still conceal contamination, an incorrect material or weak interfaces.

Keep design dimensions separate from machine readouts. A controller height value may be a position relative to a reference. Its display resolution does not equal measurement uncertainty. State the zeroing method, gauge or artifact, verification interval and reaction when the reference shifts.

Where packaging is tight, include the splice plus seal or protective covering in the envelope. Check whether flash, sharp edges or strand ends can damage heat-shrink tubing or adjacent insulation. Use the protective-sleeving guide when the joint sits inside an abrasion-control system.

Establish a real process window

A nominal setting is one point. A process window demonstrates acceptable output across controlled input and equipment variation. Develop it with the exact wires, tooling, joint arrangement and equipment intended for production.

Begin with a trial plan covering variables that the chosen equipment controls or measures. Typical candidates include:

• oscillation amplitude;
• normal force or pressure;
• weld energy;
• weld time;
• trigger force or trigger position;
• weld width and gather position;
• pre-weld and final height;
• hold time where the equipment uses it;
• wire stack and insertion depth;
• tool surface condition; and
• material lot, conductor condition and temperature.

Do not vary parameters independently without considering interaction. Higher amplitude and longer time may both increase deformation. Higher force can alter coupling. A development study should deliberately include low and high boundary combinations, not only visually acceptable nominal welds.

For every trial, retain machine curves or monitored values, joint dimensions, photographs, failure mode and destructive or electrical results. Select a nominal region with margin from failure boundaries. Document why the chosen center and alarm limits are appropriate. A broad equipment setting range is not an approved process window.

The official Schunk GS-40-plus page lists process monitoring, documentation and position measurement for joint geometry on that named splicer. Branson's Ultrasplice instructions describe monitoring weld time and peak power on its named system. These examples show that machines expose useful signals, but signal names and algorithms differ. Ask for the exact controller manual and exported record definition. Schunk GS-40-plus and Branson Ultrasplice 40 instructions.

Monitoring limits are screens, not proof of every joint

Energy, time, peak power, displacement, pre-height and final height can reveal missing wires, bad loading, tool damage or abnormal coupling. They do not directly prove every internal interface or long-term performance. Different defects can sometimes produce similar aggregate signals. A sound joint can also shift as a tool wears while remaining acceptable.

For every monitored characteristic, require:

• definition and unit;
• source sensor and resolution;
• sampling or calculation method;
• nominal value and upper and lower limits;
• whether the value controls the weld or only reports it;
• alarm, lockout and bypass behavior;
• authority needed to change limits;
• record precision and timestamp;
• linkage to machine, recipe, tool and product; and
• containment action following an out-of-limit cycle.

Challenge monitoring with known disturbances. Omit the smallest wire, change stack order, vary insertion and use approved boundary material lots. Introduce a controlled worn-tool condition and move settings toward both process limits. Do this with development samples and approved safe procedures, not saleable production.

Record both setpoints and actual cycle results. A saved recipe name without its parameter revision is weak evidence. Protect recipes with access control, back them up and audit changes.

Validate mechanical strength by failure mode as well as force

A pull result needs a defined specimen, direction, rate, gauge length, grip method and conditioned state. Pulling the whole branch at once can hide a weak small conductor. Test each critical conductor or use the customer-approved configuration.

The Branson manual focuses on the smallest wire and, for equal-gauge splices, the anvil-side wire. This supports a buyer question: which constituent is most difficult to join and how is it challenged? The numerical minimum must come from the customer drawing, applicable standard or validated design.

Record:

• peak force and force-displacement curve where required;
• which wire and side were loaded;
• break location;
• strand pullout, interface separation, conductor break or mixed failure;
• visible strand and insulation damage;
• specimen conditioning and age; and
• splice recipe, material lots and tool identity.

A high peak force with an unacceptable brittle break at the nugget may not meet the requirement. A conductor break outside the weld can show that the joint exceeded that specimen's conductor strength. It does not by itself establish electrical or environmental durability.

Use peel or torsion tests only when the product or customer method defines them. Do not convert one test into another using an unsupported factor.

Use cross-sections and microscopy for development and audits

A polished cross-section can reveal consolidation, unbonded regions, strand distortion, asymmetry, contamination and cracks. It can help correlate geometry and monitoring signals during development. Define section location and preparation so the sample is not smeared or torn in a misleading way.

Do not impose a generic void percentage or compression ratio unless the governing specification defines it. Ultrasonic splices are multi-strand solid-state joints. Image thresholds and what counts as a void need a controlled method. State magnification, lighting, etch if any, measurement software and acceptance rule.

Use cross-sections for:

• initial recipe development;
• low and high process-window comparison;
• new wire or plating evaluation;
• tool-surface or geometry changes;
• periodic audit where risk requires; and
• failure analysis after electrical, environmental or mechanical testing.

Keep the micrograph linked to the exact cycle record. A "representative section" without splice ID, recipe and material lots cannot validate the offered combination.

Verify electrical performance at the correct boundary

Measure resistance across defined conductor points outside the disturbed zone. Use a four-wire Kelvin method when lead and contact resistance matter. Document current magnitude and direction, stabilization time, temperature and probe positions. The Kelvin resistance guide explains measurement-boundary and uncertainty controls that also apply here.

Do not report only "continuity pass." Compare splice contribution with the conductor baseline and design limit. If the harness requirement includes voltage drop under operating current, validate the complete circuit at its stated temperature. Do not infer that result from a room-temperature continuity threshold.

For qualification, evaluate resistance before and after the applicable environmental sequence. A joint can pass pull strength yet develop resistance through oxidation, fretting, corrosion or interface damage. The sequence must come from the customer or end-product specification and intended environment.

Production resistance measurement may be impractical on every small splice. If so, connect process monitoring, periodic destructive and electrical verification, and end-of-line harness tests in the control plan. The harness end-of-line fixture guide helps separate product resistance from fixture error.

Treat copper, aluminum and mixed-material joints separately

Copper-to-copper results do not qualify aluminum or copper-to-aluminum combinations. Aluminum develops an oxide surface and has different mechanical, thermal and electrochemical behavior. Platings, lubricants or drawing compounds may also change weld response. TELSONIC and Schunk describe copper and aluminum joining as applications. Neither statement qualifies an arbitrary wire pair.

For any aluminum or mixed joint, require:

• exact conductor alloy, temper, plating or coating and strand construction;
• approved storage and surface-condition limits;
• stripping method that does not damage strands;
• dedicated recipe and tooling evidence;
• joint geometry and insulation setback;
• electrical resistance and temperature-rise validation;
• mechanical tests before and after conditioning;
• a corrosion and environmental sequence appropriate to installation;
• sealing and moisture-control design where used;
• galvanic-interface assessment; and
• change control for material source or coating.

A solid-state process does not eliminate corrosion risk. The finished splice still operates in an environment and can contain dissimilar metals. The seal is a separate engineered feature. The galvanic-interface guide provides a structured approach to material pairs and environmental evidence.

Control wire preparation and cleanliness

Wire preparation begins upstream of the welder. Define cut quality, strip length, maximum strand nicking or loss, twist disturbance, exposure time, handling and contamination controls. Identify whether the approved wire contains process lubricants and whether cleaning is permitted.

Never add a solvent, abrasion or chemical cleaning step without material, EHS and process approval. Residue can alter welding and later sealing. An aggressive method can remove plating or reduce strand area.

Verify stripping equipment, blades and settings at start-up and after changeover. Use vision or dimensional checks where justified. If conductor ends are precompacted, state whether compaction is part of the qualified sequence and how it is controlled.

Trace a suspect weld backward to the cut and strip station and wire lot. A welder alarm cannot detect every upstream defect.

Manage sonotrode, anvil and gather-tool life

Tool surfaces transmit vibration, constrain the nugget and establish geometry. Wear, contamination, chipped teeth, polishing, looseness or incorrect assembly can move the process. A cycle count copied from another joint is not enough.

Create a tool-control plan that includes:

• tool part number, material or coating where relevant, traceable ID and revision;
• approved working faces and rotation or indexing rules;
• installation torque and stack assembly from the equipment maker;
• frequency, tuning or health check after installation;
• cleaning method and prohibited abrasives;
• start-up reference welds;
• inspection features and magnification;
• cycle counter and joint-family usage;
• monitored trends that trigger service;
• rework, resurfacing or retirement route; and
• containment since the last known-good check after damage is found.

A tool-life limit can combine a validated maximum cycle count with earlier condition-based removal. Document actual life by wire family and settings. Do not extend life only because joints still pass a visual check.

Run first-article and capability studies on the offered combination

A first article should use released production wires, machines, tools, operator instructions and downstream sealing. Include samples from each approved stack orientation and material alternate. If several machines will run the joint, demonstrate machine-to-machine equivalence or qualify each route.

A strong initial package includes:

1. released splice drawing and constituent-wire matrix;
2. parameter-development report with boundary trials;
3. measurement-system checks for dimensions, pull and resistance;
4. dimensional, visual, mechanical, electrical and cross-section results;
5. applicable environmental and durability report;
6. monitoring challenge study;
7. short-run stability evidence for critical outputs;
8. tool-life baseline;
9. production control plan and reaction plan; and
10. traceability linking every sample and result.

Capability statistics are useful only when the process is stable and the measurement system is suitable. The specification must also have a meaningful one-sided or two-sided interpretation. Do not accept an index calculated from mixed splice families, reworked data or too little history without context.

Use a bounded bid comparison

Consider a hypothetical four-wire copper splice. The drawing defines wire part numbers, stack order, maximum resistance contribution, per-wire pull requirements and insulation setbacks. It also defines an environmental sequence. This example contains no recommended numerical limits.

Bid A quotes a machine capacity covering total area and promises automatic quality control. It does not list wire constructions, tool IDs, monitored signals, destructive tests or alarm containment. Its machine range is not product evidence.

Bid B supplies a nominal recipe and average pull results from five samples. The wires have the correct gauges but an unidentified supplier and different strand construction. There is no boundary study or post-environment resistance result. The evidence is not transferable to the offered BOM.

Bid C returns the exact wire matrix and stack drawing. It validates low, nominal and high process conditions and correlates monitoring with known disturbances. It records each conductor's pull failure mode, measures resistance at a defined boundary and sections window samples. It controls tooling, shows post-environment evidence and lists revalidation triggers.

Bid C provides the strongest decision evidence. The buyer must still check governing requirements, sample adequacy and supplier capability. The example illustrates method, not approval of a company.

RFQ evidence matrix

On small screens, swipe or scroll sideways to read every column. Keyboard users can focus the table and use the arrow keys.

Production controls and reaction plan

At changeover, verify recipe, wire IDs, tool set, stack aid, dimensional reference and monitoring status. Run approved reference pieces and required destructive checks before releasing production. Prevent production if recipe and harness revision do not match.

For every splice, retain enough information to identify:

• product serial or batch;
• splice ID;
• timestamp and operator or station;
• wire lots;
• machine and software revision;
• recipe number and revision;
• tool set and cycle count;
• actual monitored values;
• pass or fail alarms and overrides; and
• downstream seal or protection batch where relevant.

When an alarm occurs, stop the affected route and preserve the part and original record. Verify wire and loading, check tools and equipment, find the last known-good check and define containment. Do not weld the same splice repeatedly until it passes. A second ultrasonic cycle changes an already formed joint and needs an approved repair method or rejection.

Trend alarms and destructive results by joint family, machine, shift, tool life and wire lot. A rising final-height drift or repeat alarm rate can justify maintenance before an escape.

Standards, customer specifications and evidence boundaries

The official IPC revision table checked on 2026-09-30 lists IPC/WHMA-A-620 Revision F from October 2025. It also lists IPC-D-620 Revision A. IPC describes IPC/WHMA-A-620 as requirements and acceptability for electronic wire harnesses and cables. The purchased standard, contract, addendum, customer drawing and product class determine the criteria. IPC document-revision table.

Technician training, company certification or a generic IPC statement does not qualify the offered splice. Do not assume the standard supplies the buyer's sampling frequency, recipe or every product-specific pull and environmental limit. State document revisions and precedence in the RFQ.

The Schunk, TELSONIC and Emerson or Branson sources are original equipment-manufacturer materials. Their process descriptions, monitored variables and machine features apply to the named technologies or equipment. They do not validate SINAWATTS, another supplier, another machine or the buyer's splice. The supplier must return evidence for the exact production combination.

Change control and revalidation

Reopen the approved evidence when any of the following changes:

• wire maker, part number, alloy, plating, strand construction or insulation;
• number of wires, side entry, stack order, branch direction, strip or overlap;
• splice geometry or sealing design;
• machine, generator, actuator, transducer stack or controller software;
• recipe setpoint, monitoring algorithm or alarm limit;
• sonotrode, anvil, gather tool, insertion aid or working-face geometry;
• preparation or cleaning method;
• production site, line or utilities outside the qualified range;
• test method, acceptance criterion or governing specification; or
• field failure, unusual alarm trend, tool damage or failed verification.

Classify changes by risk and define evidence before release. "No effect expected" is a hypothesis, not closure. Compare the changed input against the approved window. High-risk material, geometry and tooling changes may require full mechanical, electrical and environmental requalification.

Source boundaries checked on 2026-09-30

The official Schunk technology, GS-40-plus and Cable Fit-In pages were checked on 2026-09-30. The TELSONIC metal-welding page, Emerson or Branson Ultrasplice 40 instructions and official IPC revision table were also checked. Manufacturer ranges belong only to named equipment. IPC revision information identifies the current document family but does not replace the purchased standard. No universal setting, pull force, resistance, sampling interval or tool life was inferred.

Send the splice diagram, exact wire part numbers, environment and acceptance requirements for a structured RFQ. Include volume, traceability, customer standard revision and qualification sequence. Bidders can then return an evidence package instead of only a machine name and total area. The actual supplier must confirm capability, price, quantity and lead time.

Buyer FAQ

Is total conductor area enough to select a recipe?

No. Total area does not preserve wire count, individual size, strand construction, material, plating, stack order or surface condition. Freeze every constituent and the stack-up.

Is ultrasonic wire welding a melting process?

The cited Schunk and TELSONIC sources describe a solid-state pressure and friction process below the joined metals' melting point. The exact thermal history still depends on the application and settings.

Which parameters should production monitor?

Use signals supported and validated for the exact machine. Examples include energy, time, peak power, pre-height, final height or deformation. Correlate them with product tests and known disturbances.

Does an in-limit machine cycle prove a good splice?

No. It is a process screen. Material identity, internal interfaces and long-term durability still require qualification, periodic checks and a reaction plan.

Which wire should be pull-tested?

Use the customer-approved method. Challenge the smallest or otherwise weakest constituent and each critical location. Record failure mode, not just peak force.

Are copper and aluminum recipes interchangeable?

No. Treat copper, aluminum and mixed joints as separate combinations. Require exact alloy, coating, surface, tool, recipe, electrical, mechanical and environmental evidence.

Should every splice be cross-sectioned?

Usually cross-sections are destructive and used for development, qualification, audit and failure analysis. Sampling frequency must follow product risk and contract requirements.

Can a rejected splice be welded again?

Only if a validated and contract-approved repair method explicitly allows it. Reapplying ultrasonics changes the existing nugget and must not be an informal retest.

What is the most important tooling evidence?

Require exact tool IDs, working-face condition, installation and health checks, cycle history, cleaning, inspection and replacement rules. Also require containment after damage.

What should trigger requalification?

Changes to wire identity, stack-up, preparation, geometry, machine, recipe, tooling, site, test method or specification need formal review. A significant quality signal also needs review.