Every terminal test starts the same way: strip the conductor, insert it, tighten to the value in the table, measure. Half the failures in the field happen after that point, in the second and third thermal cycle, when the conductor has yielded and the screw has not moved.
The retighten test exists to reproduce that. The conductor is heated, cooled, and the screw is checked again with the same torque wrench. What you measure is not force, it is how much the joint has relaxed.
What the table values mean
Torque values are given by screw size, not by current rating. A 3.5 mm screw and a 4 mm screw can carry the same current and still need different torques, because the torque limit is set by the screw and the plastic boss behind it.
| Screw | Typical torque | Why this order of magnitude |
| M3 | About 0.5 N·m | Small head, plastic boss takes the reaction |
| M3.5 | About 0.8 N·m | The common wall-plug terminal |
| M4 | About 1.2 N·m | Larger clamp, still a moulded boss |
| M5 | About 2.0 N·m | Terminal blocks rather than accessories |
| Screwless (push-in) | No torque; 60 N pull | Clamping is by spring, tested by pull |
| After 5 cycles | At least 60 % of the first value retained | The number that fails quietly |
Two details decide whether a reading is meaningful. The wrench must be applied in the tightening direction with a steady hand, and the reading must be taken as the screw starts to move, not after it has turned. A torque wrench used as a lever will read high every time.
The five cycles, and what each one is for
- Cycle 1, cold. Establishes the baseline. Record it even when it is inside tolerance.
- Cycles 2 and 3, after heating. The conductor and the brass clamp expand at different rates; the screw preload drops. Most of the loss happens here.
- Cycle 4, after cooling. The joint is now at its loosest in normal service. This reading predicts field behaviour best.
- Cycle 5, verification. If the joint still holds the minimum value, the design is forgiving. If it has fallen below, the terminal relies on the installer rather than on the design.
A terminal that drops to a third of its first reading is not "still tight". It is a joint that runs hotter, and the temperature rise test will show it as a 10 K to 15 K penalty on that pole.
Damage the test can do, and how to avoid it
Every retighten consumes a little of the thread and a little of the plastic boss. Repeating the five cycles on the same specimen for a second time produces numbers that look like design failures but are test artefacts. Use a fresh sample for the temperature rise test, and log which sample went through how many cycles.
The conductor itself matters too: a stranded conductor settles under the clamp after the first tightening, a solid conductor does not. Mixing the two in one comparison produces a 20 % difference that has nothing to do with the terminal. Our terminal and cord anchorage test equipment includes the torque wrenches and pull rigs for both cases; the temperature rise method that follows this test is described in the article list.
Standards referenced: IEC 60884-1 clause 26 and Table 5, IEC 60998-1 for terminal blocks, IEC 61210 for flat quick-connect terminations.
How to know a terminal will pass before you test it
Two design numbers predict the outcome better than the test itself: the thread engagement, which should be at least 3 full threads, and the boss wall thickness around the insert, which should be at least 1.5 mm for an M3.5 screw. A terminal with 2 threads and a 0.8 mm boss will lose 70 percent of its preload in five cycles even when the first reading is perfect.
Check both on the drawing before the sample is moulded; it is cheaper than a failed type test.
