How to Make a Terminated Connection That Lasts

Why this matters

Most callbacks on a termination are not workmanship failures in the way the phrase suggests. The screw was tight. The strip length was fine. What went wrong is that somebody managed one property of the joint, usually torque, and left three others to chance, and the joint failed on one of the three. A torque value controls clamping force at the moment you set it. It does not control what metal is under the clamp, whether the terminal was ever meant to hold that conductor, whether the conductor is carrying the weight of the cable, or what the joint does the first time the equipment runs a full cycle and cools down again.

This is a specification you complete as you work, one field at a time. Each field is a decision, and each one is locked by the field above it: once you have chosen the terminal, you cannot fix a preparation problem with torque, and once you have torqued it, you cannot fix a terminal-selection problem with heat shrink.

Before you touch a terminal

De-energize, lock, tag and prove dead before any contact work on a utilization circuit, under 29 CFR 1910.333(b)(2) for general industry or 29 CFR 1926.417 for construction, and prove it with the live-dead-live sequence at NFPA 70E-2021, 120.5 (that standard reaches you through your employer's electrical safety program in the edition the employer adopted, or through a contract). 29 CFR 1910.147 is the wrong citation for utilization-equipment electrical work; it excludes that exposure at (a)(1)(ii)(C).

A conductor that has been landed on a bus and is now free in your hand can still be backfed. Prove the conductor you are about to strip, not just the panel it came out of.

Field 1: conductor material, size and stranding

Write down what you are actually terminating: material (copper or aluminium), size, and solid or stranded, and if stranded, whether it is the fine-strand flexible type. This is the field everything else answers to.

Skipping it is how the classic mismatch happens: a fine-stranded flexible conductor landed in a mechanical lug listed only for Class B stranding. The lug's set screw is designed to bite a small number of large strands; on fine strand it extrudes them sideways out from under the screw, and the joint loses clamping force over the first few heat cycles without anybody doing anything wrong afterward. The fix is a terminal identified for the stranding, or a properly crimped ferrule, and neither of those is available to you once the screw is set.

Field 2: terminal identified for that conductor

The terminal has to be identified for the material, the size range and the number of conductors you are putting in it. All three, not just the size.

Two conductors under one screw is the most common violation and it is not a judgment call: a terminal is listed for one conductor unless it is marked for more. Under two conductors the screw seats on the larger one, and the smaller one is held by whatever is left. It reads as tight and carries almost nothing until the larger one relaxes, at which point the current redistributes and the small conductor becomes the hot element.

Under NFPA 70 (National Electrical Code) Article 110, in the edition your jurisdiction adopted, terminations must use connectors identified for the conductor material, and the installed assembly's ampacity is limited by the lowest temperature rating in the path. That last part is a whole-path rule: a 90 C conductor landed on a 75 C-rated lug in a 60 C-rated device is a 60 C connection, and sizing it as though it were 90 C is how a correctly made joint runs hot forever. OSHA gives the same requirement teeth at 29 CFR 1910.303(b) for general industry and 29 CFR 1926.403 for construction, both of which require listed equipment to be installed and used according to its listing and labeling.

Field 3: preparation

Record the strip length and what you did to the metal.

Strip to the terminal's own strip gauge, not by eye. Too short and insulation goes under the clamp, which is a soft, creeping material directly in the load path. Too long and bare conductor stands proud of the terminal, which is a clearance problem and, in a vibrating enclosure, a fatigue point.

Use a stripper that indexes on the insulation rather than a knife wherever the conductor size allows, because a circumferential nick in a strand is a stress riser that will crack under thermal cycling long after you have gone. Where a knife is the only option on a large cable, cut along the jacket away from your body with the blade travelling away from your supporting hand, and pencil the jacket rather than ringing it.

If the conductor or the terminal calls for an oxide-inhibiting compound, apply it per the terminal's instructions. Treat the compound as a chemical: it is a skin and eye irritant, so wear chemical-resistant gloves and eye protection and read the safety data sheet before opening the tube, and do not use it as a substitute for wire-brushing an oxidized surface where the instructions call for both.

Field 4: how contact pressure is made, and where the number came from

Two ways to make contact pressure, and each has exactly one legitimate source for its number.

Threaded terminals. Torque to the value marked on the terminal, the device, or in the instructions shipped with it. That value belongs to the manufacturer, not to a general table, and NEC Article 110 in the adopted edition requires the listed torque where one is provided. Use a calibrated torque tool. A screwdriver-feel estimate lands wide in both directions, and both directions fail: light leaves the contact spots too small, heavy strips the threads or extrudes the conductor out from under the screw so the joint relaxes as soon as it is warm.

Crimped terminals. The number is the die, not a force. The crimp tool, the die and the terminal are a matched system specified by the terminal's manufacturer, and a tool that closes fully on a terminal it was not made for produces a joint that looks right and is not. Keep your hand clear of the die closing line, because a full-cycle ratchet tool will not release mid-stroke.

There is no third source. A torque figure from a bolt chart, a die selected by eye, or a number remembered from another job is a guess wearing the clothes of a specification.

Field 5: where the mechanical load is taken

The contact interface should carry current and nothing else. Anything else in the load path - the weight of the cable, the pull of a door that swings, the tug of a strain-relieved cord - belongs on a clamp, a cable tie to a fixed anchor, or a connector's own strain relief, positioned so the conductor between the anchor and the terminal is slack.

Leave enough slack for the enclosure to be opened without loading the terminal, but not so much that a loop of conductor can vibrate against a sharp edge. This is the field most often left blank, and its failure mode is slow: the terminal holds fine, then a fine crack opens in the conductor a few strands at a time just behind the lug, the remaining strands carry the full current, and the joint that was never loose starts behaving exactly like a loose one.

Field 6: environment

Note the temperature swing, the moisture exposure and the vibration source. These change the other fields rather than adding a step. Vibration argues for a crimped connection over a threaded one where the terminal type is a free choice. A wide temperature swing argues for keeping dissimilar metals out of the clamp stack, because differential expansion inside a threaded joint is the main way preload gets lost over seasons rather than years. The dissimilar-metal joint has its own article; do not try to solve it here with more torque.

Field 7: verification, recorded as a number

• Crimped: a pull test on a sample from the same tool and die, and the crimp is judged by the die's own inspection criteria, not by whether it survived your hands.
• Threaded: the torque tool reading, plus a witness mark across the screw head and terminal body so the next visit can see movement rather than guess at it.
• Under load, once energized: a millivolt drop across the joint with the load current recorded beside it. A drop without its current is not a measurement. That reading is energized diagnostic work under the arc-flash and shock requirements of your employer's program.

The card, filled in

A 6 AWG stranded copper conductor landing on a mechanical lug inside a rooftop-unit disconnect, fed from a 60-amp circuit that actually pulls close to 45 amps in cooling season.

• Field 1: 6 AWG, copper, Class B stranded, 7 strands. Not flexible cord.
• Field 2: lug is marked for one 14 to 4 AWG copper conductor, 75 C. The conductor insulation is rated 90 C, the disconnect is marked 75 C. Lowest rating in the path is 75 C, so 75 C is the column used for ampacity. Nothing about the 90 C insulation is available to me.
• Field 3: stripped to the lug's gauge mark, indexed stripper, no knife. No nicks visible on any of the 7 strands under a light. No compound: the lug is copper-only and the instructions do not call for it.
• Field 4: torque value marked on the lug body, set with a calibrated torque screwdriver at the marked figure. I did not use a bolt chart, because the lug is not a bolt.
• Field 5: cable clamped at the enclosure entry, about a hand's width of slack between the clamp and the lug, routed so the door closes without touching it.
• Field 6: rooftop. The daily surface swing on a summer roof is large, and the unit's own fan puts a continuous small vibration into the disconnect. Crimped would have been the better terminal type had the disconnect come with one; since it did not, the witness mark and a re-check at the next service become part of the deliverable rather than optional.
• Field 7: witness mark applied. The follow-up reading is energized diagnostic work: it proceeds only where de-energizing is infeasible or would introduce a greater hazard under 29 CFR 1910.333(a), with the shock and arc-flash boundaries established and PPE selected for the incident energy at that enclosure under your employer's electrical safety program, leads placed once and your body outside the boundary. Taken that way at the following visit, in cooling season with the unit running, the drop across this lug read 5 millivolts at a measured 44 amps. That is 0.22 watts, and the witness mark had not moved.

Now read the last field against Field 6. The 5 millivolts is 0.11 milliohms of joint, which is a good joint. But the value only means something because the current is stated beside it. Had the same 5 millivolts been read in spring with the unit idling at 4 amps, the joint would have been 1.25 milliohms and dissipating 0.02 watts, and at the summer 44 amps that same joint would drop 55 millivolts and dissipate 2.4 watts. Same recorded millivolts, an order of magnitude apart in what they mean. That is why Field 7 has two blanks and not one.

What would change this procedure

The three things that genuinely rewrite it: aluminium anywhere in the joint, which changes Fields 2, 3 and 4 together and is covered in its own article; a terminal that will be inaccessible after the equipment is closed, which moves the verification effort forward because there is no second chance; and a listed assembly whose instructions contradict any of the above, which wins, because that is what the listing rules in Article 110 and 29 CFR 1910.303(b) actually require.

References

• NFPA 70 (National Electrical Code), Article 110, in the edition your authority having jurisdiction adopted, for connectors identified for the conductor material, temperature limitations at terminations, and installation to the listed torque.
• 29 CFR 1910.303(b), general industry, and 29 CFR 1926.403, construction, requiring listed equipment to be installed and used in accordance with its listing and labeling.
• 29 CFR 1910.333(b)(2), general industry, and 29 CFR 1926.417, construction, for de-energizing and verifying before contact work; NFPA 70E-2021, 120.5, for live-dead-live, binding through the employer's program in the edition adopted.
• Terminal and connector manufacturer instructions, which own the torque value, the die selection and the crimp inspection criteria.
• See related: What a Crimp Actually Does to the Conductor; Why Aluminium and Copper Terminations Differ; The Connection Checks Worth Doing Before You Close It Up.