Military cable harness testing checklist

A harness that passes a buzz-through is not a harness that has been tested. This is the sequence we run on naval and defence assemblies, and what each step is looking for.

Before any electrical test

Verify against the drawing, not against the last one. Wiring schedule, connector part numbers, pin assignments, backshell types and overall lengths, checked against the released revision. Most harness faults found at acceptance are build-to-wrong-revision, not workmanship.

Inspect the terminations. Crimp height and pull-off where crimped, solder wetting and absence of wicking where soldered, no nicked strands, no exposed conductor beyond the specified length. A crimp that looks acceptable and was made with the wrong die will pass continuity and fail in service under vibration.

Check screening and bonding. Screen continuity through to the backshell, 360° termination where specified, no pigtails where the drawing prohibits them. On naval EMC-critical assemblies this is frequently the difference between a compliant and a non-compliant install, and it is invisible to a continuity test between pins.

Confirm marking. Wire and harness identification legible, correct and durable in the intended environment — and matching the schedule, not merely present.

The electrical sequence

Continuity, end to end, every core. Point-to-point against the schedule, not a spot check. Record the measured resistance rather than a pass/fail, because a rising contact resistance across a batch is a crimp-tooling problem you want to catch on unit three rather than unit thirty.

Isolation between every conductor pair, and to screen and shell. This is where reversed or transposed cores are caught. A harness can be perfectly continuous and completely wrong.

Insulation resistance at the specified test voltage. Applied between conductors and between conductors and screen or shell, held for the specified dwell rather than flashed. Record the value and the ambient conditions; insulation resistance is temperature and humidity dependent and a number without conditions is not a result.

Dielectric withstand where the specification calls for it. Run after insulation resistance, never before — a withstand test on an assembly with an undetected insulation defect turns a repairable fault into a destroyed one.

Screen transfer and bonding resistance where EMC performance is specified, measured rather than assumed from visual inspection.

Fibre-optic assemblies

End-face inspection before anything else. Under magnification, every connector, every time. Contamination is the dominant cause of fibre loss and mating a dirty ferrule can damage the one it mates with.

Insertion loss against the budget for the link, measured at the operating wavelength — not at whatever wavelength the tester defaults to.

Return loss where reflection matters to the system.

OTDR trace for anything long enough for the location of a fault to matter, retained as a baseline. A trace taken at acceptance is what makes a fault two years later diagnosable rather than a guess.

Environmental and mechanical, where specified

Naval and airborne assemblies frequently carry environmental requirements that a bench test does not touch: vibration, thermal cycling, salt fog, or ingress protection at a stated rating. These are specification-driven — do what the contract says, and do not substitute a bench check for a qualification test.

Where a harness is qualified rather than merely tested, keep the distinction explicit in the paperwork. Qualification demonstrates the design; acceptance testing demonstrates the individual unit. Conflating them in a report is a finding waiting to happen at audit.

Mechanical checks that belong on every unit: strain relief engaged and correctly sized, bend radius respected at every fixing, clamping at the specified intervals, and connector keying confirmed so a unit cannot be mated wrongly under time pressure in a dark compartment.

The documentation is part of the deliverable

For defence and PSU work the test evidence is as much the product as the harness. Each unit should ship with the procedure it was tested to, the measured results with instrument identification and calibration status, the ambient conditions, the operator, and the date.

Record measured values, not ticks. A file of pass marks tells you nothing about drift; a file of numbers shows a batch trending toward a limit before any unit fails.

Where an assembly is a reverse-engineered replacement for an imported item, the pack should also carry the investigation record: what the original measured, what was substituted, and the equivalence rationale. That is what makes the substitution defensible rather than merely functional.

Sources.

• Our own build and acceptance work on naval electrical and fibre-optic cable assemblies, including flask and turbine cable assemblies for submarine and torpedo systems, and cable casing, flange and coupling work for Indian Navy ships.
• Test order follows standard electrical acceptance practice: continuity and isolation, then insulation resistance, then dielectric withstand.

Published 16 August 2026. Last revised 25 September 2026. Corrections to sales@vebixautomation.com.

Need harnesses built and documented?

Send the drawing set, or a sample if there are no drawings, and we will reply on reverse engineering, the build and the acceptance test pack. Ask for a quote.