A harness drawing that says "build to IPC-A-620" and nothing else is not a specification. It names a standard with three acceptance classes, several revision levels and dozens of criteria that change between them, and it leaves every one of those choices to whoever happens to be building the part. Two suppliers can both claim IPC-A-620 compliance and deliver harnesses that would not pass each other's inspection.
The class you specify is a contract term with a price attached. Class 2 is what most industrial and consumer harnesses are actually built to. Class 3 adds measurable work to every unit: crimp-height monitoring, a tighter pull-test regime, first-article cross-sections, and operator certification that has to be current. Buyers who write "Class 3" onto a drawing without knowing what it triggers usually discover the cost when the first order ships, and buyers who write nothing discover the quality when the first field return arrives.
What IPC/WHMA-A-620 Actually Covers
IPC/WHMA-A-620 is the joint standard for cable and wire harness assemblies, published by IPC with the Wire Harness Manufacturers Association. It covers crimped, soldered, welded and insulation-displacement terminations, plus the assembly-level requirements that follow: wire routing, bundling, marking, shielding termination and strain relief. It is a separate document from IPC-A-610, which covers board-level assembly, and the two are commonly confused on drawings written by people who work mostly on boards.
The standard exists in three classes, and the class is declared by the customer, not chosen by the supplier:
| Class | Definition | Typical product | What it means for acceptance |
|---|---|---|---|
| Class 1 | General electronic products | Internal appliance wiring, low-cost consumer cable assemblies | Function is the only requirement; cosmetic and process evidence are not |
| Class 2 | Dedicated service electronic products | Industrial controls, HVAC, telecom cabinets, most instrument harnesses | Continuous service expected; no cosmetic defects that could reduce life |
| Class 3 | High reliability electronic products | Automotive safety, medical, aerospace, defence | Continuous high performance; equipment downtime is not acceptable; process evidence required |
The practical difference is not that Class 3 has different defect types. It has the same defect types with tighter limits, applied to a larger sample, with more documentation that the process is in control rather than merely the output being acceptable on the day.
Crimp Acceptance: Where the Numbers Live
A crimp is a controlled metal-forming operation, and its quality is set by four dimensions that a tool either produces consistently or does not. The inspector's job is to verify those dimensions, not to form an opinion about how the joint looks.
| Criterion | Class 2 practice | Class 3 practice | How it is verified |
|---|---|---|---|
| Crimp height | Within the terminal maker's tolerance band | Same band, monitored per shift with a crimp-height gauge, records retained | Micrometer or dedicated crimp-height gauge on a sample per setup and per shift |
| Bell-mouth | Visible flare at both ends of the barrel | Required, with flare length inspected against the terminal drawing | Magnification 10x to 20x |
| Brush / strand protrusion | Strands visible in the transition zone, not beyond the barrel edge | Same, with protrusion measured on the first article | Magnification, first article cross-section |
| Insulation clearance | Insulation not inside the wire barrel; a small gap is acceptable | Gap controlled; insulation must not be nicked or deformed | Magnification on every operator setup |
| Cut-off tab | Not protruding beyond the barrel envelope | Flush to the barrel, no burr | Magnification, tactile check |
| Strand damage | No severed strands; nicks limited | No nicks permitted in the crimp zone | Magnification, plus pull test as the functional check |
| Terminal orientation | Square to the wire axis | Same, with bend radius controlled at the transition | Go/no-go gauge on the first article |
The single most useful number on that table is crimp height, because it is the only one that can be measured on a production line without cutting the part open. A crimp that is too high leaves the strands loose inside the barrel and fails in vibration. A crimp that is too low severs strands and fails in thermal cycling. Both look acceptable to a casual visual check, and both are caught by a gauge in ten seconds.
Pull Force: The Test That Settles Arguments
Crimp height predicts; pull force proves. The tensile test pulls the wire out of the terminal at a controlled rate until the joint separates, and the peak load is compared against a minimum for that wire gauge and terminal series. It is the acceptance criterion that survives a dispute, because it is a number both parties can agree on before the order starts.
| Wire gauge | Typical minimum tensile (stranded copper) | When it is run |
|---|---|---|
| 28 AWG | ~13 N (3 lbf) | First article and per setup |
| 26 AWG | ~22 N (5 lbf) | First article and per setup |
| 24 AWG | ~36 N (8 lbf) | First article and per setup |
| 22 AWG | ~58 N (13 lbf) | Per shift sample |
| 20 AWG | ~90 N (20 lbf) | Per shift sample |
| 18 AWG | ~140 N (31 lbf) | Per shift sample |
| 16 AWG | ~200 N (45 lbf) | Per lot |
Those figures are indicative minima for common terminal series and copper conductors; the drawing or the terminal maker's specification governs, and the values shift with alloy, plating and strand count. The important structural point is that the test is destructive. Every crimp pulled is a crimp destroyed, so the sample has to be drawn from the lot rather than from the shipment, and the rejection rule has to be written before the first pull rather than after.
A separate non-destructive test exists for finished assemblies, where a proof load below the crimp's yield strength is applied and the joint must show no movement. It is useful for auditing a finished harness but it cannot replace the destructive test, because a crimp that holds at proof load can still be far from its specified compression.
Strand Damage, Nicks and Birdcaging
Strand damage is the defect that passes inspection and fails in the field. A nick on the outside of a strand is a stress riser, and in a harness that sees vibration or thermal cycling the strand fractures at the nick rather than at the crimp. The failure appears months later, at a location nobody inspected, and is almost always misdiagnosed as a bad crimp.
Birdcaging is the visible version: strands separating and bowing outward between the insulation and the barrel, caused by pushing the wire into the terminal rather than inserting it cleanly, or by a stripping blade that has nicked the conductor. Once strands are separated they cannot be re-compressed into the barrel without damage, so the termination is scrapped rather than reworked. A Class 3 operation treats birdcaging as a process indicator, not an isolated defect, because it usually means the strip tool is worn.
Solder Terminations Inside a Harness
Solder cups and solder-cup connectors remain common in low-volume and military-style harnesses, and they bring their own acceptance criteria. Wicking is the main one: solder must wet the cup and the strands, and must not travel up the wire under the insulation beyond a defined limit. Wicking stiffens the wire at the point where it needs to flex, and a harness that fails a flex test usually fails at a wick line rather than at the termination.
Class 3 restricts solder terminations more tightly than Class 2, both on wicking limits and on the inspection evidence required. Where a design has a choice, a crimped termination is easier to control at volume because the process is a machine setting rather than an operator skill, and machine settings can be monitored. That is the practical reason most automotive and medical harnesses are crimp-only.
What Class 3 Actually Costs
The premium for Class 3 is not a margin adjustment; it is a set of additional operations. The table below expresses them as relative adders against a Class 2 baseline on the same drawing, at a volume where tooling is amortised. They are planning figures, not quotations, and the labour-rate dependence is significant.
| Cost driver | Class 2 | Class 3 | Relative effect on piece price |
|---|---|---|---|
| Crimp-height monitoring | At setup | Per shift, logged, gauge-controlled | +5 to 10% |
| Destructive pull test | First article + per setup | Per shift sample per terminal series | +3 to 8% |
| First-article cross-section | Optional | Required per terminal/wire combination, photographed | One-time tooling, $150-400 per combination |
| Operator certification | Training record | Current IPC/WHMA-A-620 certification per operator | +2 to 5% |
| Inspection sampling | AQL sample | 100% visual plus AQL dimensional | +8 to 20% |
| Traceability | Lot-level | Unit-level, tied to operator and tool | +3 to 6% |
| Documentation package | CoC | CoC + process records + test data | +2 to 4% |
On a simple two-wire harness the total adder is often 20 to 35 percent. On a 60-circuit harness with mixed terminal series the inspection and documentation share grows, because the number of distinct crimp setups grows with it. That is why the class decision belongs on the drawing before the quote is requested rather than after, when it becomes a change order.
Hand Tool Versus Applicator: The Hidden Class 3 Failure
A certified operator with a hand crimp tool can produce a Class 3 crimp. What a hand tool cannot do is produce the same crimp ten thousand times. Hand tools depend on operator force and die alignment, and the variation between operators and between shifts is larger than the Class 3 tolerance band on small terminals.
Production Class 3 work runs on a bench applicator or a fully automatic crimping machine with a controlled ram, a ratchet that prevents partial cycles, and a crimp-height adjustment that is set with a gauge and locked. If a supplier quotes Class 3 pricing on a harness that will be built with hand tools at volume, either the price is wrong or the claim is. The question to ask is not whether the supplier is certified, but which tool will physically make the crimps on this order.
The Four Drawing Notes That Prevent the Argument
Most disputes about harness quality trace back to a drawing that omitted one of four notes. Adding them costs nothing at release and settles the argument later.
- Acceptance class — "Class 2, IPC/WHMA-A-620" or "Class 3". Not "IPC-A-620 compliant", which specifies nothing.
- Standard revision letter — criteria change between revisions, and a supplier building to an older revision is not automatically wrong. Name the revision on the drawing.
- Sampling and test plan — AQL for visual and dimensional, pull-test frequency, and who keeps the records for how long.
- Repair and rework authority — whether a rejected crimp may be cut back and re-terminated, how many times, and whether the rework has to be re-tested or scrapped.
The rework note is the one buyers most often forget. A harness that can be re-terminated twice is cheaper than one that must be scrapped, but only if the rework is a controlled operation with its own pull test. Reworded loosely, it becomes permission to reuse a terminal, which is a defect in every revision of the standard.
What to Ask a Harness Supplier Before You Release the Drawing
Five questions separate a harness cell from a harness claim. Which acceptance class will you build to, and which revision of the standard? Which tool will make the crimps, and how is crimp height set and monitored? What is your destructive pull-test frequency, and will you send the data with the shipment? How do you handle a rev change on the wire list, and who pays for the retest? And which of your operators currently hold certification, with expiry dates?
A supplier that answers those five in writing has a process you can audit. One that answers with a certificate and a reassurance has a document. For the enclosure side of the same box-build programme, where the harness has to pass through a sealed wall and survive the same environment, see our guide to IP rating and enclosure sealing. The sealing and the crimp fail for different reasons but in the same field return, and the acceptance evidence should be reviewed together.
Our own harness cell builds to IPC/WHMA-A-620 with UL 1007/1015/1061 PVC wire, silicone UL 3122/3135 up to 200 degrees Celsius, and PTFE UL 1330/1332 for chemical exposure, with crimped and soldered terminations tested before they are integrated. The method for the pull test itself, and how to read a marginal result, is covered in our guide to crimp quality and pull testing. Where the harness feeds a board that is itself being assembled, the print and placement side is treated in the IPC Class 2 versus Class 3 guide for board-level acceptance.
Get Your Harness Drawing Reviewed Before Release
Send the wire list, the connector schedule and the drawing as it stands. We will flag the missing class and revision notes, identify terminal and wire combinations that will not reach the pull force you need, and tell you where a hand tool versus applicator decision changes the price. That review takes a day and happens before the drawing is released to anyone.
