November 1, 2026 · Technical Guide
A crimped contact can pass a visual inspection, pass a crimp-height gauge, pass an insulation-resistance check and pass a pull test — and still be wrong. It can be crimped onto fifteen strands of a nineteen-strand conductor. It can have a strand folded back inside the barrel. It can have the wire inserted only halfway. It can have the two wings of the barrel touching at the tips but never actually closing around the bundle, so the joint holds by friction on the flash rather than by a cold weld on the wire.
None of those defects is visible from outside. None of them changes the crimp height enough to trip a go/no-go gauge on a well-set die. Most of them will still survive a pull test, because a partially filled barrel can still grip. But all of them change the cross-section, and the cross-section is the only inspection method that looks at the thing you actually care about: how much metal is touching how much wire, where, and under how much compression.
That is the trade: to see it, you have to destroy the joint. So a cross-section is not a product test at all — it is a process test. It answers "is this die, this tool, this operator and this wire lot set up correctly?", not "is this particular joint good?". Confusing the two is the most common way shops either over-test (sectioning sample after sample and learning nothing new) or under-test (running a whole harness build on a setup that was never verified). This article covers how to take the section, how to read it, what the numbers mean, and how it fits with the non-destructive checks that guard production.
It assumes the termination method is already decided — if you are still choosing between crimp, solder and a PCB tail, read the termination guide first. It also assumes the contact itself is the right part; the stamped vs machined construction guide covers why an open-barrel stamped contact and a closed-barrel machined contact need different tooling and different section criteria.
Every metal surface in air carries an oxide film — on copper it is a few nanometres, on tin it is thicker and grows with time and temperature, on aluminium it is immediate and stubborn. That film is a semiconductor at best and an insulator at worst. A crimp works by deforming both surfaces past the point where the film can survive: the barrel walls are squeezed inward until the metal underneath the oxide is extruded into contact with similarly fresh wire surface, and the two cold-weld at the asperities while the surrounding metal locks them mechanically in place.
Two consequences follow directly, and they explain almost every acceptance criterion in this article:
Cross-sectioning is the last and most expensive rung. Put it in the right place or you will use it to answer questions the cheaper tests already answer.
| Test | Destructive? | What it proves | What it cannot see |
|---|---|---|---|
| Visual (10×, both ends) | No | Barrel closed, no cut-off tab spikes, insulation not in the wire barrel, brush present, no obvious strand damage | Anything inside the barrel: voids, missing strands, insertion depth beyond the bellmouth |
| Crimp height (micrometer or dedicated gauge) | No | That the die closed to the dimension the tool maker's chart specifies for this wire and this contact | Whether the wire in that barrel was the wire the chart was set up for |
| Crimp width | No | That the wings formed to the expected width — a width outside expectation means metal volume changed (wrong contact, wrong strip thickness, wrong die) | Whether the wings met, overlapped, or crossed |
| Pull / tensile test | Yes (the joint) | Mechanical strength of the joint, and by implication that some cold welding occurred | Electrical quality of the joint, void content, strand count, insertion depth — a friction grip can pass |
| Low-level contact resistance (LLCR) | No | That current can cross the joint with a small voltage drop, measured at a low test current so the film is not broken down by the meter | Margin. A joint can read acceptable today and degrade rapidly under thermal cycling if the true contact area is small |
| Cross-section / microsection | Yes (the joint) | Barrel geometry, wing closure and symmetry, void content and distribution, strand count and placement, insertion depth, bellmouth, cracks, wall thinning, and the actual compression achieved | Nothing about the joint's strength — you destroyed it — and nothing about the next joint off the tool |
The pattern is worth stating plainly: the non-destructive tests measure the setup, the destructive tests measure the joint, and the section is the only one of the five that measures the inside of the joint. That is why a section is used to qualify a setup and then replaced by height, width and pull for day-to-day production.
Most bad cross-section conclusions are preparation errors, not reading errors. A section that is taken in the wrong plane, or polished until the wings smear, will show a defect that does not exist — or hide one that does.
Two planes matter and they answer different questions:
A single transverse cut is the standard production check. The longitudinal cut is what you reach for when a transverse section looks acceptable but the joint fails a pull test — because pull strength is governed by engagement length and bellmouth as much as by area.
A 10× to 20× toolmaker's microscope or an optical comparator with a measuring reticle is enough for the dimensions; the photograph for the record is better taken at higher power so the wing seam and void distribution are legible.
Measured across the outside of the barrel, at the middle of the crimp, in the transverse plane, perpendicular to the base. This is the number the die sets and the number the tool maker's chart publishes for a given wire gauge, insulation thickness and contact part number. Because the die nest depth fixes it, crimp height is the variable a shop controls; nearly everything else follows from it and from the wire.
Measured across the barrel at the same station, parallel to the base. Crimp width is not independently adjustable — on a given die it is a consequence of how much metal is present and how far the punch travels. That makes it a diagnostic rather than a setting: width outside the expected band means the metal volume changed, which means the wrong contact, the wrong strip thickness, a die that is worn, or a barrel that was already deformed before it entered the tool.
On an open-barrel (F-crimp) contact the two wings are rolled inward and should meet or overlap at the bottom of the barrel, forming a closed seam roughly on the centre line. Read three things: do they meet, are they symmetric about the centre, and is the seam at the bottom rather than rotated to one side. Asymmetric or crossed wings ("rabbit ears" that pass each other rather than closing) mean the contact was not seated squarely in the die, the wire was off-centre, or the tool is worn — and a crossed seam is both a mechanical weak point and a path for moisture into the barrel.
On a closed-barrel (machined) contact there are no wings; the equivalent readings are wall thickness uniformity around the circumference and the absence of the classic four-indent or circumferential-crimp marks cutting into the wire bundle.
Void area is the empty space inside the barrel that is neither wire nor barrel metal. Read it as distribution before you read it as a total: scattered small voids around the periphery are normal in a stranded crimp and are not a defect, because the strands cannot pack perfectly. A chain of connected voids across the barrel, or a void that sits between the wire bundle and the seam, is a signature of under-crimp or of wire that never filled the barrel. Isolated voids adjacent to the wing tips are a signature of wings that closed on each other instead of on the wire.
This is the number people argue about, so define it before you quote it. The version used in this article is fill factor: the cross-sectional area of conductor metal inside the barrel, divided by the internal cross-sectional area of the crimped barrel, expressed as a percentage. Both areas are read directly off the transverse section (or computed from the wire's strand count, for the numerator).
Worked example, with the arithmetic shown so you can substitute your own wire:
The acceptable band is a product-specification number, not a universal constant: it depends on whether the contact is open- or closed-barrel, on the wire stranding, and on the tool design. Published workmanship criteria commonly land in a band in the region of 60–80 % fill, but treat that as orientation only — read the band from the contact's product specification or the tool maker's crimp chart for your exact part number and wire gauge, and do not carry a number from one product family to another. The method is the transferable part; the number is not.
This is the practical core of the article. Each row is a thing you can see on a section, the mechanism behind it, and the correction — in order of how often each one shows up.
| Section signature | Mechanism | What it costs in service | Correction |
|---|---|---|---|
| Void chain across the barrel; bundle loose; wings meet but do not compress the wire | Under-crimp — die did not bottom, wrong die for the wire, worn tool, or positioner set to the wrong depth | Oxide film never fully broken; high and unstable resistance; fretting opens the joint under vibration | Reset crimp height to the tool maker's chart for that wire/contact; verify with a new section before running |
| Barrel wall visibly thinned; flash extruded at the seam; wing tips driven into the bundle; individual strands severed | Over-crimp — height set below the chart value, or a die/contact mismatch closing further than designed | Reduced conductor area (a hot spot), cracked wings, stress-corrosion initiation, shortened flex life | Raise crimp height; replace the die if it is worn or incorrect; scrap the run made at that setting |
| Crack at the outer bend radius of a wing, penetrating the wall | Over-crimp, or base-metal with insufficient elongation for the form | Fatigue crack growth under vibration; a path for electrolyte into the joint | Correct height first; if it recurs at the correct height, raise it with the contact supplier — the material or the forming die is the issue |
| Wings asymmetric, seam rotated off the bottom centre line, or wings crossed past each other | Contact not seated squarely in the die; wire off-centre; worn or damaged die nest | Mechanical weak point; moisture path at the seam; unpredictable pull values | Check contact seating and positioner; inspect the die nest; re-qualify with a section |
| Strand count short, or a strand folded back inside the barrel | Strip/cut process damaged or lost a strand; wire was dressed badly before insertion | Reduced current capacity and a locally hotter joint — and nothing else in the inspection plan will find it | Fix the strip-and-cut setup; add a strand-count check at the wire prep station, not at the crimp station |
| Wire not inserted to full depth — barrel only partly filled along its length | Insertion depth stop set short; operator did not push home; strand flare hit the barrel and stopped the wire | Pull strength below spec and a joint that relies on the front half of the barrel only | Set and lock the insertion depth stop; check bellmouth and flare before insertion; verify on a longitudinal section |
| Insulation inside the wire barrel | Over-strip or under-insertion | Zero conductor-to-barrel contact in the affected region; a joint that may pass a weak pull test and fail electrically | Correct strip length; move the insulation crimp wings to the insulation |
| Bellmouth absent, or barrel edge sheared into the strands | Bellmouth formed incorrectly at the tool, or the contact was pre-deformed | Severed or nicked strands at the entry — a fatigue origin, and reduced conductor area | Check the tool's bellmouth/ramp geometry; reject contacts with a squared-off entry |
| Cut-off tab spike visible inside the barrel | Poor cut-off in the strip; worn cut-off punch | Locally reduced fill and a sharp inclusion against the strands | Die maintenance; add cut-off tab limits to the incoming contact inspection |
Note what has no entry in that table: "the wire was the wrong gauge". It belongs there, but it does not produce a signature — it produces a section that looks acceptable at the measured height because the die compressed an undersized bundle until the wings met. This is the single most important limitation of the method, and the next section is about it.
Crimp height is a dimensional criterion: it proves the tool closed to the right depth. It does not prove that the correct amount of metal was inside the barrel when it closed. Four situations produce a section that contradicts a good height reading:
The practical rule: a cross-section qualifies a specific combination — this contact part number, this wire part number and lot, this die, this tool, this crimp height setting. Change any one of them and the section is stale. That is why the record has to name all of them (see the evidence package below), and why a section on file from last year proves nothing about today's build.
You cannot section every joint, and you should not try. The point of a destructive test is to place it where it buys the most information:
Between sections, production is guarded by the non-destructive trio — crimp height at a defined frequency, pull test on a sample, and LLCR on the finished assembly — which is exactly the set the electrical testing guide covers in detail. Treat those three as the trend you watch and the section as the explanation you reach for when the trend moves.
One more placement point: because a section is destructive and tooling-specific, it belongs in the harness shop's control plan, not the connector supplier's incoming inspection. If you are buying finished cable assemblies, the section evidence is something you ask the supplier to produce — which is a receiving-inspection document question, and the receiving inspection guide covers how to make that request enforceable in the purchase order.
The criteria shift with the contact design, and the shift matters because a criterion copied across families is usually wrong:
| Contact family | Barrel type | The reading that matters most | Watch for |
|---|---|---|---|
| MIL circular signal contacts (#22D, #20, #16 — M39029 family) | Closed barrel, machined | Wall thickness uniformity and fill factor; indent depth if a four-indent crimp | Indent crushing the bundle rather than compressing it; barrel ovality |
| Power contacts (#12, #8, #4) | Closed barrel, heavy wall | Fill factor and absolute barrel area, because current density is the service risk | Stranded-wire fill dropping as gauge rises; a section that looks fine but has measurable voids at the seam |
| Open-barrel stamped terminals and splices | Open barrel, F-crimp with separate insulation wings | Wing closure and symmetry, bellmouth, cut-off tab, insulation wings gripping insulation | Wings crossed instead of closed; insulation inside the wire barrel; tab spike |
| Environmental splices (heat-shrink sealed) | Crimp barrel inside a sealed sleeve | Fill factor read before sealing — the sleeve hides the barrel afterwards | Sectioning a sealed splice and reading the melted adhesive as a void |
| Thermocouple and small-gauge contacts | Very small closed barrel | Both conductors fully inside and both bottomed; no strand loss | Two wires in one barrel read as a single oversized bundle; unequal insertion depth |
The family you are building determines which product drawing governs, and the drawing determines the numbers. For circular mil-spec work that is the contact's slash sheet plus the connector's product specification; for the assemblies those contacts go into, the J599 (MIL-DTL-38999), ZH83723 and MIL-DTL-26500 families all use removable rear-release contacts that are sectioned the same way, and the criteria come from the contact, not the shell. The two contact pages sectioned most often in this range are the size 20 socket and the size 16 socket — both closed-barrel, both with a barrel wall thick enough that an over-crimp shows as thinning rather than as a crack. If the joint is being made in a finished harness rather than at a bench, the wire harness assemblies page covers how those builds are documented.
A section photograph on its own answers nothing six months later, because nobody can tell what it was made from. A usable record contains, at minimum:
That last line is what turns a folder of photographs into a process record. A section without a disposition is a picture; with one, it is the evidence that the setup was qualified on a date, by a tool, with a wire lot — which is exactly what an audit or a first-article review will ask for.
No — and doing so would be a poor use of a destructive test. Section at setup, at first article, at the frequency your control plan sets, and after any change of contact, wire, die, tool or operator. Day-to-day production is guarded by crimp height, pull test and low-level contact resistance, all of which are non-destructive or nearly so.
No. They measure different properties. A section proves the geometry and the amount of metal contact; a pull test proves mechanical strength, which depends heavily on insertion depth and bellmouth — features a transverse section barely shows. The two disagree for a reason, and the disagreement is diagnostic: good section plus poor pull usually means insertion depth or bellmouth, not compression.
Crimp height is an external dimension of the finished barrel and is set by the die. Compression ratio (or fill factor) is a cross-sectional area ratio read from a destroyed joint. Height is the variable you control in production; the ratio is what you measure to prove the height was correct for the wire that was actually in the barrel.
Check three things in order: whether the failed joint itself was sectioned (it should not have been — use a sibling), whether the wire lot matched the section's wire lot, and whether the plating or the contact surface — not the crimp — was the failure origin. A crimp section says nothing about the mating interface, and in a connector the mating end is often where the resistance actually lives.
No. It guarantees that the barrel was filled and compressed as designed at the moment of crimping. It does not guarantee the joint stays that way — thermal cycling, vibration and contamination all change a joint after it leaves the bench, which is why the electrical and environmental qualification evidence sits alongside the section, not instead of it.
Airoadcon supplies J599 (MIL-DTL-38999), ZH83723 and MIL-DTL-26500 connectors with M39029 removable contacts, plus the matching crimp tooling, positioners, extraction tools and finished harness assemblies built to a documented control plan.
Send us the wire specification and the contact part number and we will confirm the crimp height setting, the die and positioner for that combination, and the pull-test and resistance evidence your receiving inspection should expect. If you are chasing a joint that reads good and fails in service, send the section photograph and the wire lot — that pair usually identifies the cause in one pass.
Email: info@airoadcon.com | WeChat: 18991927716