Wire harness manufacturing: each stage and how it is checked

A wire harness build can be set out in eight stages, though the split and the order vary with the product and the line: the drawing, cut and strip, terminate, sub-assemble, lay up on a board, bind and protect, label, then test and inspect. Each stage can leave a fault that a later one covers up, and the final electrical test does not see all of them: a terminal short of its lock can pass continuity,¹ and so can a wire with six of its seven strands cut.² So each stage carries its own check, and the table below sets out which check finds which fault.

The same product is called a wire harness, a wiring harness or a cable harness, and in Britain also a wiring loom. Both names appear in UK recall notices, though not equally. Of the 4,566 vehicle recalls launched in the UK from 2016 to 2025, "wiring harness" or "wire harness" appears in 130 and "wiring loom" in 9, six of those from Ford.³

Which check finds which fault

No single source covers every row. NASA-STD-8739.4A comes closest, but it is a workmanship standard for NASA hardware.⁵ Molex and TE write for their own terminals, and Cirris writes as a tester maker. The checks a customer's drawing or standard calls up replace these where they differ.

The build, stage by stage

Drawing and wire list

The drawing and wire list give each wire's size, colour, length and both ends, with the terminal, housing and protection for each part. TE puts the match on the harness maker: terminals, wires and applicators must be compatible, "the use of the correct components during the application has to be ensured by the harness manufacturer", and the wire's conductor area and outer diameter must agree with the terminal drawing.⁴ Size units are one place the match can slip. A drawing in AWG built with metric wire, or the reverse, has no exact pairs: rounding 16 AWG down to 1 mm² loses 23.6% of the copper, as the AWG to mm² conversion works out.

Cut and strip

Wires are cut to length and stripped with hand tools, bench machines or fully automatic wire processors. Molex notes that hand stripping produces more variability than a stripping machine, and bench terminations more than a wire-processing machine.⁶ NASA-STD-8739.4A rejects a conductor nicked, cut or scraped to the point that base metal shows, or reduced in cross-section, and insulation left nicked, cut, crushed or charred.⁵ Molex lists cut or nicked strands among the causes of low pull force.⁶

Strip length matters at the next stage. With the insulation position centred, it sets the conductor brush, the strands that extend past the conductor crimp.⁶

Wire marking, where the drawing asks for it, can go on at this stage too. Komax, a wire-processing machine maker, sells inkjet, hot-stamp and laser markers and a marker-tube module that work in line with its own and Schleuniger's machines.⁷

Terminate

Most harness terminations are crimps. Molex describes the window: the crimp has to press hard enough to break down the oxides on the stripped conductor and the tin plating inside the terminal grip, or resistance can rise, while over-crimping reduces the conductor's area and raises resistance again.⁶ A crimp can look tidy and still be outside that window, so three checks share the work.

• Crimp height is quick and non-destructive, and Molex calls it critical to electrical and mechanical reliability. It is usually control charted, because set-up samples are few and their statistical value is limited.⁶ Tolerances, gauge error and why five pieces cannot give a Cpk are in the crimp height guide.
• Pull force is a quick, destructive test of the crimp's mechanical strength.⁶ NASA-STD-8739.4A requires a pull test at the start and end of each work shift or production run, whichever is shorter, and no production until three samples pass with no failures.⁵ How many to pull, how often and what to do on a fail are in the crimp pull test guide.
• Appearance covers the bell mouth, conductor brush, insulation position, cut-off tab and insulation crimp. Molex makes fanning each bundle and checking these by eye standard practice for the operator, and warns that they interact: a track adjustment for the bell mouth also moves the cut-off tab and the insulation position.⁶

In a NASA review of 780 pull test results, most contact and wire pairs beat the minimum by at least 100%, and the abstract puts averages at 182% of it or more, though the paper's own table of 18 pairs runs from 126% to 300% of NASA's minimum.⁸ That is aerospace data on machined contacts and not a benchmark for a commercial line. The same authors note that at high volume the recorded forces can be analysed for die wear, tool setting changes and improper tool use. They also say pull tests could be run less often if the tool and materials are verified at the start of a run and the settings stay unchanged, provided the project weighs the added risk first.⁸

Splices are terminations too. NASA-STD-8739.4A asks for as few as possible, holds a crimped splice to every requirement for a crimp, sizes its barrel from the combined area of all the wires in it, covers each splice with heat-shrinkable tubing or insulation, and staggers splices so the bundle does not swell in one place.⁵

Sub-assemble

Here groups of wires, such as those ending in one connector, are made up before lay-up, and terminals go into their housings at this stage or later on the board. Loading is where seating faults start. Cirris, a tester maker, describes a contact inserted far enough to pass visual inspection and standard continuity testing that is still not fully seated, so its lock never engages and it can push back later.¹ NASA-STD-8739.4A requires every contact in a connector with retention clips or tines to be push or pull tested for seating.⁵ The terminal seating guide compares the pull forces connector makers publish and what a secondary lock does and does not prove.

Where a later step hides a joint, NASA-STD-8739.4A tests before it: the electrical acceptance tests on potted connectors come immediately before potting and again after final assembly.⁵ Overmoulding hides joints in the same way. Henkel's low-pressure moulding process runs insert, overmould, then test, and Henkel sells one grade for strain relief on cables and wires.⁹ Testing before the moulding as well is an inference from NASA's potting rule, not a Henkel instruction.

Lay up on the board

The wires are routed on a board or a three-dimensional fixture that holds the shape of the finished harness. For complex harnesses NASA-STD-8739.4A requires layout and fixturing, with a full-size form layout fixture to ensure routing, wire lengths, connector configuration, support and access. It also asks for permanent bends and offsets to be built in, so that the installed wire dress is not under continuous stress, and for the layout to limit the bending and pulling a harness takes during installation.⁵ That makes the fixture the check at this stage: each branch length, bend and support position compared with the drawing while the harness is still on the board.

Routing on the board is not the same as routing in the product, and it is the second that reaches the field. In the count of 422 UK wiring recalls, the largest group, 127 recalls, is routing in the vehicle: a harness too close to something, or rubbing on it. Across that count, where the text places a fault, 144 of 193 read as made while building the vehicle.³ The board cannot check clearance in the product. NASA-STD-8739.4A treats it as design: it requires flexing and pressure over sharp or rough edges to be prevented, and protection added where abrasion could occur.⁵

Bind and protect

Tape, straps, sleeving and conduit hold the shape and protect the wires from whatever the harness will rub against. NASA-STD-8739.4A wants straps tight enough not to slide along the bundle but not so tight that they visibly dent or distort the wires, and the tooling that fits them tension-controlled.⁵ It also protects the harness while it is being built: connectors not being worked on are wrapped, and harnesses in fabrication are covered at the end of each shift.⁵

Label

Individual wires may already carry their marking from cut and strip. This stage marks the harness and its connectors: NASA-STD-8739.4A requires each harness and each connector to be permanently identified, the marking to resist abrasion, and every temporary marker to be removed by the end of fabrication.⁵ The same standard treats everything made before a failed pull test as suspect, and requires pull test records traceable to the contact, wire, crimp tool, setting and run or shift.⁵ Read together, those clauses mean the marking on a harness has to lead back to the run its crimps came from.

Test and inspect

For NASA hardware every cable assembly passes continuity, dielectric withstanding voltage (hipot) and insulation resistance tests, in that order, so that a continuity fault is found before full voltage goes on and a latent defect stressed by the hipot shows up in the insulation resistance test.⁵ Commercial drawings vary. The comparison of continuity, 4-wire, IR and hipot tests sets out which faults each one finds and which it misses.

Two gaps are documented. A contact short of its lock can pass continuity.¹ And resistance barely moves with broken strands: in a Cirris experiment on 38 in of 7-strand 26 AWG wire, cutting six of the seven strands raised the reading from 0.1290 Ω to 0.1320 Ω, and warming part of the wire by hand changed it by more.² Faults like these have to be caught at the stages above. NASA-STD-8739.4A adds a 100% visual inspection of assembled connectors for soldering, identification and damage.⁵

The first harness of a new build, the first article, is where the board and the tester are both proved. Cirris's routine for setting up a new test ends with testing the first article: if it passes, the test stands, and if it fails, the failures are compared with the tester's programming to decide whether the fault is in the tester's set-up or in the harness.¹⁰ Lengths and branch positions can be measured against the drawing's tolerances on the same harness.

Whether every unit gets each check, or a sample, is a separate question with its own arithmetic, set out in sampling plans and when to test 100%. First pass yield, counted stage by stage, shows where the rework comes from.

Which standard applies

NASA's pull test paper describes IPC/WHMA-A-620, Requirements and Acceptance for Cable and Wire Harness Assemblies, as the standard levied for commercial and some high-reliability work, with three end-product classes that reflect the level of control applied to the manufacturing process.⁸ Which revision and which class apply comes from the customer's drawing or contract. It is a paid, copyrighted standard, so none of its tables or criteria are given here. Work from a licensed copy.

NASA-STD-8739.4A, which most of the table above draws on, is public and free to read.⁵ It shows what a full set of workmanship checks looks like, but it applies to NASA mission hardware and does not replace the standard the customer names.

If the harness goes into a control panel, the panel has standards of its own. Start with which standard applies to a panel.

Notes

1. Cirris, Connector Contact Retention, 15 June 2023. Supplier guide. ↩ ↩² ↩³

2. Cirris, Can I test for broken strands?, 11 June 2023. A supplier's experiment on one wire, measured 4-wire. ↩ ↩²

3. Driver and Vehicle Standards Agency, Check vehicle recalls, recall data file RecallsFile.csv, downloaded 6 November 2026. Rows grouped by recall number and counted by launch year, searching the concern, defect and remedy text, in research/wire-harness-manufacturing/calc.py. The routing count and the 144 of 193 are from the recall analysis linked above. Contains public sector information licensed under the Open Government Licence v3.0. ↩ ↩²

4. TE Connectivity, Application Specification 114-18022, Rev K, 2011, general guidelines for contacts with open crimp barrels, copy hosted by Farnell. Check TE for a later revision. ↩ ↩²

5. NASA, NASA-STD-8739.4A with Change 4, 2022, clauses 7.2.30 to 7.2.35, 8.1, 8.2, 8.4, 9.6, 10.1, 12.3.5, 13.8, 14.1, 18.2.1, 18.2.6, 18.3.2 and 19.9. Public domain. A US government workmanship standard for NASA mission hardware. ↩ ↩² ↩³ ↩⁴ ↩⁵ ↩⁶ ↩⁷ ↩⁸ ↩⁹ ↩¹⁰ ↩¹¹ ↩¹² ↩¹³ ↩¹⁴ ↩¹⁵ ↩¹⁶ ↩¹⁷ ↩¹⁸ ↩¹⁹ ↩²⁰ ↩²¹ ↩²²

6. Molex, Quality Crimping Handbook, order no. 63800-0029, revision C, 2009, copy hosted by DigiKey. Written for Molex open barrel terminals and Molex tooling, so check the terminal maker's own limits and the current revision. ↩ ↩² ↩³ ↩⁴ ↩⁵ ↩⁶ ↩⁷ ↩⁸ ↩⁹

7. Komax, Wire marking and identification, product overview. Supplier page. ↩

8. A. Constante, C. Fitzgerald and A. Boutte, Analysis of Pull Force Test Results for Crimped Connections, NASA Goddard Space Flight Center, conference paper for IPC APEX EXPO 2023, read in the March 2022 text. Its data are machined contacts for space flight. It quotes IPC requirements, which are not repeated here. ↩ ↩² ↩³

9. Henkel, TECHNOMELT Low Pressure Molding, brochure LT-6746. Supplier document; its sales claims are not repeated here. ↩

10. Cirris, Verify Cable Tester Programs BEFORE Testing, 5 June 2014. Supplier guide. ↩