This is a twenty-minute check that finds more real faults than any instrument reading, and it needs almost nothing.
Run a substantial load first
Nothing shows at rest. Put the system under a load near what it normally carries, and give it a few minutes for temperature to develop.
Feel along the run, then at every joint
The cable should be uniform. Terminals, lugs, busbar connections, fuse holders and isolators are the places where heat appears, and the fuse holder is the single most common warm component on this shelf.
Measure across anything suspicious
Voltage measured directly across a joint, with the load running, should be negligible. A reading that is not negligible is resistance, and it is located rather than inferred.
Fix it properly rather than tightening it
A joint that has been hot has usually oxidised, and retightening it buys weeks. Remake it: cut back, crimp with the matching die, heat-shrink, and torque to the figure the component maker publishes where one exists.
Then put it in the calendar
Twice a year on a fixed installation, and after any journey that involved vibration. A check that is not scheduled is a check that happens after the failure rather than before it.
What people ask before they order
Why does a voltage drop point at a joint rather than the cable?
Because a cable drops voltage evenly along its length while a joint drops it across a few millimetres. Measuring across the joint alone separates the two immediately.
Is a warm cable always a problem?
A uniformly warm cable under a heavy load may simply be working. A local hot spot is not, and that difference is what makes the comparison along the run worth making rather than a single reading.
Last reviewed 16 September 2026