Quality inspection for cable assemblies: what to require and how to read results

Not every characteristic needs the same intensity of check. Critical polarity on multi-way connectors, protective-earth continuity on power assemblies, and high-energy DC terminations usually justify 100% verification. Cosmetic braid pattern or minor jacket scuffing may sit at sample visual review — or photo-standard comparison — if they do not affect function or safety. Classify each characteristic as critical, major or minor, then assign a method: visual, gauge, electrical, or destructive sample.

Align the plan with drawing notes and any clauses in your quality manual or customer quality agreement. Requirements that appear only on the purchase order create conflict when production and receiving disagree. If a clause matters, put it on the released package or a numbered quality requirement under revision control. Orphan checks also waste capacity: hipot on every low-voltage signal loom “because we always do” can create false failures and obscure the tests that actually protect the product.

Visual criteria cover jacket damage, incorrect wire colour, missing or illegible labels, poor tape or lace coverage, connector damage, contamination, incorrect keying orientation and incomplete backshell hardware. Borderline conditions — “acceptable scuff,” “slight braid flare,” “label edge lift” — consume more meeting time than hard failures. Publish photo standards for those edge cases so operators and incoming inspectors share the same reference.

Dimensional checks focus on critical lengths, strip dimensions, breakout positions and connector clocking. Soft goods need fixtures, form boards or controlled layouts to be fair; measuring a formed harness without the same support used in production will reject good product or accept bad product depending on who pulls harder. Use the measurement method defined on the drawing. Prefer envelope tolerances on free spans and reserve tight tolerances for clamp-to-connector or connector-to-connector interfaces.

Crimp quality may be verified by pull force, microsection, crimp height monitoring, or a combination defined per terminal family. Define which method applies at first article, after applicator or die change, and during serial production. Tool identification and calibration status should be available for review when your programme requires them — not as a generic claim, but tied to the work orders that built your lots.

Soldered terminations, ultrasonic welds and bolted lugs need their own criteria: wetting and fillet appearance, weld process windows, torque and marking. Mixing “termination looks fine” language across technologies invites subjective escapes. Point the inspection plan to the acceptance criteria you actually invoked — for example IPC/WHMA-A-620 class callouts where used — rather than vague “good crimp” notes.

For sealed connectors, visual inspection should include seal presence, correct cable OD range and absence of nicks on glands or boots. An electrically perfect assembly that fails ingress in the field is still a quality escape; sealing steps deserve checklist lines, not tribal knowledge.

Continuity and short testing catch miswires, swapped cavities and unintended bridges. Insulation resistance or dielectric withstand (hipot) validates insulation integrity when specified. Shield continuity deserves its own step on shielded assemblies; treating the shield as an afterthought is a common field EMC complaint that started as a skipped test.

Parameter selection must match the product. Excessive hipot voltage or dwell can create failures that would not appear in service, especially on fine wire or connectors with limited creepage. Under-specified tests miss insulation damage from strip blades or crushed jackets. State voltage, limits, dwell and whether tests run dry or conditioned, and keep those parameters under the same revision control as the drawing.

For complex harnesses, automated test fixtures reduce escape rates compared with hand probing. Ask whether electrical tests run 100% or sample, how failures are quarantined, whether rework returns to full retest, and how fixture maintenance is controlled. A fixture that drifts silently is worse than no fixture at all if everyone trusts its green light.

Agree AQL or sampling logic where 100% is not required, and agree what happens when a sample fails — lot reject, tightened inspection, or screen of remaining units. Ambiguity here turns every defect into a commercial negotiation. Record formats should be usable by your receiving team: part number, revision, lot or work order, quantities, test results, inspector identity or stamp, and disposition of any nonconformances.

Certificate of conformance language should reference the drawing revision actually built. Generic “meets all requirements” statements without revision identity are weak evidence when ECOs are frequent. If your customer audits you, they will ask how you know which revision shipped; your supplier interface should be able to answer from retrievable records.

Align packaging and labelling inspection with the same package. Wrong ship labels, missing kit bags or mixed revisions in one carton are receiving failures even when the assemblies inside are electrically correct. Kit completeness checks belong in the plan when you buy electromechanical kits, not only discrete cables.

First article inspection should freeze the process: materials against AVL, tools and applicators, fixtures, test scripts, labelling artwork and work instructions. Serial lots then compare to that baseline. When engineering change orders land, decide in advance whether partial or full delta FAI is required — a label-only change is not the same as a connector family change.

Photograph critical orientations, breakouts and label placements at FAI and keep those photos with the approved package. They resolve disputes faster than paragraph arguments. Re-verify after significant ECOs, tool transfers or long production gaps where skill fade or material substitution risk rises.

Casablanca Cable Company agrees inspection and documentation scope per programme before production is allocated to manufacturing partners, so the evidence package matches what your receiving team will check. Partner allocation follows that agreed plan; it does not invent a parallel inspection story after the fact.

Inspection data only improves the product if someone acts on it. Trend pull-force margins, hipot failures and visual defect codes. Clusters point to strip setup, wrong die, training gaps or drawing ambiguity. Single random defects still need disposition, but chronic codes need process correction.

Share calibrated defect photos with the manufacturing partner and with your own design team when the root cause is documentation. Many “supplier quality” issues are actually measurement-method mismatches or missing notes. A joint review of the first three lots after FAI often prevents months of friction.

Finally, keep the inspection plan living but controlled. Do not let temporary firefighting checks become permanent without deciding whether they earn their keep. Empty paperwork slows delivery without protecting the field; focused checks tied to risk do both.