A shipment arrives, a batch fails in test, and somebody opens a claim. The supplier replies that the assemblies were built to IPC/WHMA-A-620, which is the standard the purchase order named, and that they passed the tests the standard requires.
That reply is usually true, and it is not a defense. The standard the supplier is quoting contains a clause that settles most claims, and it is a clause that very few buyers have ever read.
The clause almost nobody has read
In the general section of IPC/WHMA-A-620F, under verification of dimensions, the standard says this:
Where not specifically invoked by this standard, actual measurements, e.g., of specific solder fillet dimensions, determination of damage and wrap percentages, are not required except for referee purposes.
Read that twice, because the consequence is uncomfortable. The standard deliberately does not require a supplier to measure and record every dimension on every assembly. Dimensional evidence is not owed as routine paperwork. It becomes required for referee purposes — that is, when there is a dispute to referee.
So the measurement you will need at the start of a claim is the measurement that was not required at the moment you could still easily have recorded it. This is not a loophole; it is a design decision. The standard’s stated intent is to rely on process control rather than on inspection, which means the evidence a claim needs is evidence about the *process*, not about the individual part.
That reframing is the whole article: a claim is not won by the failure but by what the process was told to do.
The four decisions that have to be recorded before production starts, because none of them can be reconstructed afterwards
Four things that had to be written down before production
The standard is explicit that four decisions belong to the buyer, and in each case it states a default that applies if the buyer stays quiet.
Decision | What the standard says | Consequence of silence |
|---|
The acceptance class | "Use of this standard requires agreement on the Class to which the product belongs. The User has the ultimate responsibility for identifying the Class to which the assembly is evaluated. If the User does not establish and document the acceptance Class, the Manufacturer may do so." | The manufacturer picks, and the claim has to argue against a class the supplier was entitled to choose |
|---|
The revision in effect | "The use of an amendment or newer revision is not automatically required. The revision in effect shall be as specified by the User." | Two parties can each name a different revision and both be defensible |
|---|
The inspection frequency | The standard "can be used as a stand-alone document for purchasing products, however it does not specify frequency of in-process inspection or frequency of end product inspection." | Nobody has specified how much inspection was owed, so nobody can show it was owed and missed |
|---|
The defect and its disposition | The inspection conditions section defines what is acceptable, what is a defect, and what disposition means in that context | "Defect" becomes an argument about vocabulary rather than a finding |
|---|
The class row is the one that decides most disputes, because class is what the acceptance criteria are keyed to. Section 1.5 states that where the words "shall" or "shall not" indicate a requirement for at least one class, the requirements for each class appear in brackets next to the requirement. One clause therefore holds up to three different thresholds, and which one applies depends entirely on a class assignment that the standard hands to the buyer.
If the class was never documented, the argument is not about workmanship. It is about which rulebook applies, and the standard already says the manufacturer was allowed to choose.
"It passed our test" is not a defense
Section 1.1 contains a sentence that changes what a test report proves:
No limit is placed on the number of process indicators or the number of allowable repair/rework of defects.
The same paragraph then says such information "should be developed with a statistical process control plan."
So the standard sets no ceiling on how many times a defect may be repaired or how many process indicators a process may show. An assembly that failed its first test, was reworked, and passed on the second attempt is compliant. It will not carry a mark saying so, and after shipment it is not distinguishable from an assembly that passed first time.
That is the gap a claim usually falls into. A buyer reads a passing certificate as proof the assembly was never in trouble. The standard’s model is process control, in which rework is a permitted tool. The standard does structure the two operations differently — rework and repair appear as separate headings, with post-rework and post-repair cleaning as a third — which tells you the distinction matters enough to name, but the standard does not make either one visible after the fact.
There is a related detail worth knowing if your product has plated contacts. Gold removal appears as a defined operation in the soldering section of the standard. That is an operation the standard requires in some circumstances, it changes the surface of the part, and it is exactly the kind of process step that a claim file should be able to show was or was not performed.
The identity evidence is on the assembly
Everything above concerns what the process was told. The next question in a claim is which units the problem applies to, and the only evidence that travels with the product is the marking.
The standard treats marking as a chapter rather than an afterthought. Section 12 covers content, legibility, permanency, location and orientation, and direct marking, then specifies marker sleeves in wrap-around and tubular forms, adhesive flag markers, and tie wrap markers.
That list is a checklist for the buyer as much as for the supplier. If a failed assembly’s marking is illegible, that is not merely a cosmetic finding; it breaks the link between the unit in your hand and the production records that would tell you when and how it was made. If the marking’s permanency was never specified, then nothing in the shipment can be shown to have failed a permanency requirement.
One point from the standard’s own front matter is worth carrying into any claim. The development committee "recognizes that different parts of the industry have different definitions for some terms used herein." Appendix A carries the terms and definitions. If your claim and your supplier’s reply use the same word with different meanings, the definitions appendix is where that gets settled, and it is better to settle it at purchase order stage than at claim stage.
What a claim file should contain
Working backward from the standard, a file that can actually support a claim looks like this:
Item | Why the standard makes it necessary |
|---|
The documented acceptance class | Without it, the supplier was entitled to choose the class |
|---|
The revision of the standard named in the contract | The revision in effect is the one the user specified, and it is not automatically the newest |
|---|
The statistical process control plan or equivalent, defining inspection frequency | The standard states no frequency, so the frequency is whatever the parties agreed |
|---|
The written definitions of defect and disposition you are applying | The standard provides the framework and expects the parties to apply it |
|---|
Test records for the tests the standard publishes minimum requirements for | Continuity, shorts, dielectric withstanding voltage, insulation resistance, VSWR, insertion loss, reflection coefficient and pull force each have a published table |
|---|
The assembly marking specification, including permanency | Marking is the only identity evidence that ships with the product |
|---|
Any required process operations that alter the part, such as gold removal | They are requirements in some circumstances and they leave a physical trace |
|---|
A first article record if the class is high | The standard’s intent is process control, and process control needs a baseline |
|---|
None of these is exotic. Every one is something a competent purchase order can demand, and most cost nothing except the discipline of writing them down.
What the standard cannot do for you
Three limits are worth stating plainly, because a claim that assumes otherwise will fail.
First, the standard does not provide criteria for cross-section or X-ray evaluation. X-ray guidance sits in a separate appendix and is guidance rather than acceptance criteria. If your failure mode requires a microsection to prove, the standard is not going to supply the pass and fail line for it.
Second, the standard does not set inspection frequency. It says so itself. A claim that a supplier under-inspected cannot be built on this standard alone.
Third, the revision question is contractual, not technical. The publisher’s own position statement is that revising a standard "is not automatic unless required by the contract," while recommending the latest revision. Those two halves pull in opposite directions, and the only thing that resolves them is what the order said.
How a passing test certificate and a never-reworked assembly are two different claims, and the standard does not let you tell them apart after shipment
If the claim is already live
Most people read this after the argument starts, so triage matters more than theory.
Fix the vocabulary first. Agree which acceptance class and which revision you and the supplier are both measuring against, and put it in writing. Until that is settled there is no criterion, and every subsequent exchange is two people reading different documents with the same title.
Protect the failed unit and its marking. The marking is the only identity evidence that shipped with the product, so a unit whose marking has become illegible has already lost its link to the production records. Photograph it before anyone handles it further.
Ask for the process data rather than the inspection data. The standard’s stated intent is process control, and it places no ceiling on the number of allowable repairs or rework. The question that has an answer is what the process indicators showed across the lot, not what one final test said.
Ask directly whether any assembly in the lot was reworked. If no record exists either way, that is itself a finding: the same section that permits unlimited rework says the information should be developed with a statistical process control plan.
Raise the measurement question in writing rather than assuming it will be produced. The standard does not require those measurements as routine paperwork; it requires them for referee purposes.
The documentation timeline: the decisions that must precede production, the marking that travels with the product, and the measurement obligation that switches on when a dispute begins
FAQ
Can a supplier defend a claim simply by saying the assemblies were built to IPC/WHMA-A-620?
Not by itself. The standard requires agreement on the acceptance class, and if the buyer does not establish and document it, the manufacturer may choose. It also leaves inspection frequency and the revision in effect to the buyer. Naming the standard is the start of the argument, not the end of it.
The assemblies passed the electrical tests. Does that settle it?
It settles that the tests were passed. The standard places no limit on the number of allowable repairs or rework of defects, so a passing result is consistent with an assembly that failed earlier and was corrected. Nothing in the shipped product distinguishes the two cases.
What does "referee purposes" mean in practice?
It means the standard does not require routine dimensional records, but measurements become required when there is a dispute to resolve. The practical reading is that you record what you would need if the argument happened, because at that point the opportunity is usually gone.
Where do I find the test requirements the standard publishes?
Section 19 carries minimum requirements and test parameters for continuity, shorts, dielectric withstanding voltage, insulation resistance, VSWR, insertion loss, reflection coefficient and pull force, with further tables in an appendix. Naming the specific table you are testing against removes an argument about what "tested" meant.
Conclusion
A cable assembly claim is a documentation contest that happens after the documentation window has closed. The standard is candid about this: measurements are not owed except for referee purposes, inspection frequency is the buyer’s to specify, the class is the buyer’s to identify and document, and a repair is not a disqualification. Every one of those sentences is a reason to write the purchase order more carefully and to keep the records that the order was supposed to generate.
The defect is the event. The file is the case.
Building the file before you need it
ZONG Cable builds cable assemblies and cable to the class, revision and inspection plan each customer specifies, and we would rather agree those in writing at the quotation stage than argue them later. If your order does not currently name an acceptance class, a standard revision, an inspection frequency and a definition of defect and disposition, we will tell you which of those we cannot answer for you. The routines behind each production batch are described on our test and inspection page, the range sits under cable, and OEM support covers how a specification is agreed for private-label programs.
Sources
- IPC/WHMA-A-620F, Requirements and Acceptance for Cable and Wire Harness Assemblies, published front matter, including section 1.1 Scope, 1.2 Purpose, 1.3 Classification, 1.4.1 Verification of Dimensions, 1.5 Definition of Requirements, section 1.5.1 Inspection Conditions with 1.5.1.1 Acceptable, 1.5.1.2 Defect and 1.5.1.2.1 Disposition, the table of contents covering rework and repair, gold removal, marking and labeling, section 19 test tables and the appendices, and the revision history and revision position statement. Copyright 2025 IPC International, Inc. https://www.electronics.org/TOC/IPC-WHMA-A-620F-TOC.pdf
- IPC-9191, referenced within IPC/WHMA-A-620F section 1.1 as the source for statistical process control planning. Cited as named in the standard; the document itself was not retrieved.
- IPC/WHMA-A-620D product description, ANSI Webstore, describing the standard as the industry-consensus standard for requirements and acceptance of cable and wire harness assemblies. https://webstore.ansi.org/standards/ipc/ipcwhma620d2020
- ZONG Cable range. https://zh-zong.com/cable/
- ZONG Cable test and inspection. https://zh-zong.com/test-inspection/
- ZONG Cable OEM support. https://zh-zong.com/oem-support/
