What a Crimp Actually Does to the Conductor

Why this matters

A crimp that holds when you pull on it can still be electrically defective, and the shop that treats the pull test as the acceptance criterion will never find out until the season turns. The pull test measures friction and mechanical interlock. What makes a crimp last is something else entirely: a controlled amount of plastic deformation that tears the oxide film off the strands, presses fresh metal against fresh metal hard enough to cold-weld it, and squeezes the air out so no new oxide can form. Those two properties reach acceptable values at different amounts of deformation, and the window where both are good is narrower than the window where the pull test passes.

Before any of the measurements below

The drop readings in this article are taken with the circuit energized and loaded, which is energized diagnostic work under your employer's electrical safety program, inside the shock and arc-flash boundaries that program sets on NFPA 70E in the edition the employer adopted. Any cutting, re-crimping or terminal replacement happens de-energized, locked, tagged and proven dead under 29 CFR 1910.333(b)(2) for general industry or 29 CFR 1926.417 for construction, proven with live-dead-live per NFPA 70E-2021, 120.5.

The call

A shop had crimped twelve ring terminals onto control and small power conductors during a rooftop unit changeout in spring. By late summer, three of the twelve were showing brown discoloration at the barrel and one had opened a control circuit intermittently on hot afternoons. The terminals were the right size for the conductor, the right ring size for the stud, all from the same bag. Nobody had over-tightened anything, because nothing on a ring terminal's barrel gets tightened.

The tech who made them had used a plier-style crimper from the truck, the kind with a die profile stamped into the jaws and no ratchet. The tech who made the other four, on a different day, had used a ratcheting tool with the die the terminal manufacturer specifies.

What the pull test proved, and what it could not

The shop's acceptance step was a pull test: hold the terminal, pull the conductor, and reject anything that comes apart. All twelve had passed. So had the four survivors, and so had the three that were now cooking.

That result is not a failure of the pull test, it is the pull test doing exactly what it does. Pull-out force in a crimp comes from the barrel gripping the strand bundle: friction plus the mechanical interlock of deformed metal. A barrel that is only lightly compressed still grips. It grips well enough to exceed a minimum tensile figure long before the interface inside it has gone gas-tight, which is why crimp height, not pull force, is the production control everywhere crimps are made in volume.

Push the deformation further and the two properties separate again at the other end. Pull force rises to a peak and then falls, sharply, because the die starts severing strands. Contact resistance falls to a low plateau and then rises for the same reason. The usable window sits below the pull-force peak and inside the resistance plateau, and it is a few thousandths of an inch wide on the crimp height dimension for small terminals. That number belongs to the terminal manufacturer's specification, transcribed from their sheet for that terminal and that die, and there is no general table for it.

What the crimp is physically doing

Strip the process down and it is a cold weld made under a hydrostatic squeeze.

Copper strands arrive with an oxide skin on every one of them. Copper oxide is a poor conductor, and it re-forms in air within minutes of being exposed. As the die closes, the barrel wall and the strands both yield. Metal flows. The oxide layers, being brittle and thin, crack and are dragged apart, and the fresh copper exposed underneath is pressed into fresh copper on the neighbouring strand with nowhere for air to get back in. Where that happens the two pieces of metal are no longer two pieces of metal.

Three consequences follow, and they are the whole of why crimps work.

• The contact area is not the barrel's inside surface. It is the sum of the cold-welded patches, which is far smaller and far more conductive than a merely mechanical contact would be.
• The interface is sealed, not protected. A gas-tight crimp resists corrosion because there is no gas at the interface, not because anything was coated. That is why an underdone crimp cannot be rescued by weatherproofing the outside of it.
• The strand bundle is consolidated. Voids between strands are closed. Voids that remain are where moisture condenses and where the joint's resistance later grows.

The two ways the deformation goes wrong

Not enough. The barrel grips, the terminal stays on, and the oxide is only partly broken. Current crosses at fewer and smaller welded patches, plus a scatter of oxide-separated contact points that behave like tiny resistors. Voids remain between the strands. This joint reads slightly high on day one and then gets worse every thermal cycle, because the remaining unwelded contact points oxidize and the barrel, which was never work-hardened into its final shape, relaxes a little each time it heats.

Too much. The die severs strands where they cross the barrel edge. You lose conductor cross-section right at the point of highest current density, and the joint becomes a fuse with a long time constant. The tells are visible: a bright flash of extruded metal at the die parting line, a barrel that has split, crushed insulation forward of the barrel, and a few loose strand ends brushing out at the wire entry. Pull it and it may still hold, because the surviving strands are welded hard, which is exactly why the pull test does not catch this either.

The measurement that separated them

The shop measured the drop across each crimp barrel while the circuit was loaded, recording the current alongside every reading, because a millivolt figure without its current is not a measurement.

• The four made with the matched die and ratcheting tool: 1.2 millivolts at 20 amps. That is 0.06 milliohms of joint and 0.024 watts dissipated.
• The eight made with the plier crimper: 9 millivolts at the same 20 amps. That is 0.45 milliohms and 0.18 watts.

Seven and a half times the resistance, and on day one that difference is invisible: a fifth of a watt in a ring terminal on a stud, with a metal bus behind it to conduct heat away, warms nothing you would notice by hand.

The season that decided it

They re-measured in late summer under the same 20 amps.

The four matched-die crimps read 1.2 to 1.5 millivolts, which is a joint that has not moved. The eight plier crimps had spread: the best still read 11 millivolts, the worst three read 34 millivolts. At 34 millivolts and 20 amps the joint is 1.7 milliohms and dissipating 0.68 watts, which is nearly four times what it dissipated in spring, in the same terminal on the same stud carrying the same current.

That growth is the signature the pull test cannot reach. A joint that was fully cold-welded has nothing left to oxidize at the interface and no voids to relax into, so it holds its value. A joint that was underdone has both, and it moves. The direction is worth stating both ways: at low current these joints stay quiet for years because there is not enough heat to drive the oxidation, and in a circuit that only pulls real load two months a year the same eight terminals could have gone five seasons before anybody smelled anything.

What the fix had to include

All eight plier crimps were cut off and replaced, not re-crimped. A crimp cannot be re-made on the same conductor end, because the strand ends have already been work-hardened and partly severed and will not flow the same way twice. Cut back to sound conductor, re-strip, new terminal.

Two things they were tempted to do and correctly did not. They did not flow solder into the barrels to "make sure", because solder wicks up between the strands, creates a stiff section that ends abruptly, and puts the fatigue point on a vibrating rooftop right where the conductor leaves the solder. It also defeats the crimp's own design, which relies on metal-to-metal flow rather than a filler. And they did not seal the outside of the suspect crimps with adhesive-lined tubing, because sealing the outside of a crimp with voids inside it traps whatever moisture is already there.

Where tubing was used on the new terminals, it was shrunk with the heat gun on its lowest effective setting with the enclosure open to outside air, because overheated shrink tubing decomposes, and fluoropolymer tubing in particular releases decomposition products that are an inhalation hazard rather than a contact one. A glove does nothing about that route.

Keeping it from happening again

The shop changed three things, none of which is the pull test.

They matched tools to terminals: one ratcheting tool per terminal family, the die marked on the tool, and a rule that a terminal whose manufacturer's die is not on the truck does not get installed that day. Where a crimp height gauge exists for the terminal, they measure one crimp per bag and record it, which is the only check that reads the actual control variable rather than a proxy.

They sectioned a sample. Cutting one practice crimp lengthwise and looking at the cut face with a loupe shows the voids and the strand deformation directly, and once a tech has seen an under-crimp and a good crimp side by side, the visual difference on the outside of the barrel becomes obvious to them.

And they kept the pull test, demoted. It no longer certifies a crimp. It detects a tool that has drifted badly enough to stop gripping at all, which is a real failure worth catching in ten seconds, and nothing more than that.

References

• Terminal and connector manufacturer specifications, which own the crimp height, the die number and the crimp inspection criteria for that specific terminal.
• UL 486A-486B (wire connectors) and UL 486C (splicing wire connectors), which reach you through the connector's listing and the code that requires listed connectors, not on their own authority.
• 29 CFR 1910.333(b)(2), general industry, and 29 CFR 1926.417, construction, for de-energizing and verifying before cutting or re-terminating; NFPA 70E-2021, 120.5, for live-dead-live, binding through the employer's program in the edition adopted.
• See related: How to Make a Terminated Connection That Lasts; Why a Loose Electrical Connection Heats and Then Fails; What a Soldered Joint Actually Is.