The failure you can't see from the front of the panel
A panel can look perfect. The dead-front is straight, the wires are dressed, the labeling is clean. And somewhere behind one of those breakers, a lug that was snugged by feel instead of by spec is slowly cooking itself. Nothing trips. Nothing smells. For months, maybe years, the only evidence is a few degrees of extra heat that nobody is standing close enough to feel.
Loose connections are one of the most common root causes in electrical fires, and they almost never announce themselves. Understanding why a slightly-loose terminal is so dangerous — and why "good and tight" is not the same as "correct" — is one of the few pieces of knowledge that changes how you work the moment you understand the mechanism.
A connection is a contact area, not a clamp
When you land a conductor under a lug, you're not really tightening a screw onto a wire. You're creating contact area between two metal surfaces. Current wants to cross from the conductor into the lug, and it can only do that where the two metals actually touch.
Metal surfaces are never perfectly flat. At a microscopic level they meet at a scattering of high points, and current crowds through those tiny contact spots. The proper torque value exists to flatten enough of those high points together that you get a large, stable, low-resistance contact patch. Under-torque it and the conductor rests on just a few points. Over-torque it and you can deform the conductor strands or strip the threads, which is why "as tight as I can get it" is its own failure mode.
The contact resistance of a connection isn't fixed — it's a direct function of how much real contact area you've made. That's the whole reason a number on a sticker matters more than the strength in your wrist.
Where the heat comes from
Every connection has some resistance, and resistance plus current makes heat. The relationship isn't gentle: power dissipated at the connection rises with the square of the current. Double the load and you quadruple the heating at a marginal joint.
A properly torqued connection has such low resistance that this heat is negligible — the lug stays at roughly the temperature of the conductor. A loose connection with a small contact patch has higher resistance, so it runs warmer than everything around it. On a 20-amp circuit pulling near its limit, a connection that should dissipate a fraction of a watt can quietly dissipate several, concentrated on a spot the size of a pencil eraser.
That localized heat is the start of a feedback loop, and the loop is what actually kills.
The runaway no one is watching
Here is the part that makes loose connections genuinely dangerous rather than merely sloppy.
Heat makes metal expand. When a warm connection cycles up under load and cools when the load drops, the metal expands and contracts a little each time. Copper and the brass or aluminum of the lug expand at different rates, so every cycle works the joint a hair looser. A connection that left the shop at 80 percent of spec can walk itself down over months of daily heating and cooling.
Looser means less contact area, which means more resistance, which means more heat, which means more expansion and more loosening. The connection degrades itself, and it speeds up as it goes.
Heat also drives oxidation. As the contact spots get hot, the exposed metal forms an oxide layer, and metal oxides are far more resistive than clean metal. So even as the mechanical joint loosens, the surfaces themselves are turning into a worse conductor. On aluminum this is especially vicious — its oxide forms fast and is a near-insulator, which is exactly why aluminum terminations demand the right torque and often an antioxidant compound.
The endgame is glowing-connection territory: a joint hot enough to discolor insulation, char the surrounding plastic, and eventually ignite it or the dust and debris around it. By then the connection may be running hundreds of degrees hotter than the conductor a few inches away — and a standard breaker never sees it, because the circuit current is still perfectly normal. The breaker protects the wire from too much current. Nothing in a basic panel protects you from a bad joint.
Why "tight" was never good enough
Most people land terminals by feel, and feel is wildly unreliable. Studies and manufacturer testing have shown that hand-tightening produces torque values scattered far above and below spec, with the same person varying from one connection to the next. A screwdriver gives you leverage but no feedback — the connection that feels identically "snug" might be at 50 percent of spec or 200 percent.
Both errors fail. Under-torque gives you the high-resistance, self-loosening joint above. Over-torque can crush fine strands, fracture them, or distort the lug so it no longer seats properly — sometimes nicking the conductor enough to create a new hot spot. The acceptable window is narrower than your hand can sense, which is the entire argument for a calibrated tool.
This is also why the values aren't yours to guess. The torque spec for a given terminal is set by the manufacturer who tested that exact lug with that range of conductors. It lives on a label inside the panel, on the breaker, or in the product instructions — and it's often given in inch-pounds, not foot-pounds, a units mistake that itself causes over-torque when someone grabs the wrong wrench.
What the code now says
The National Electrical Code has tightened the language here precisely because feel kept failing. NEC 110.14(D) requires that where a manufacturer provides a torque value, the connection be tightened to that value using a calibrated torque tool. It's no longer framed as best practice — it's an installation requirement, and inspectors increasingly ask to see that it was followed.
The broader rule, 110.14, has long required terminations to be made so they're suitable for the conductor and won't loosen. The 2020 and later cycles simply made the means explicit: a number, a calibrated tool, applied to spec. The code is doing what good electricians already understood — turning a judgment call into a measurement.
None of this is about distrust of your hands. It's that the physics of contact resistance, thermal cycling, and oxidation operates below the threshold your hands can detect, and the consequences arrive long after you've left the job.
The habit worth building
The practical takeaway is small and almost boring, which is part of why it gets skipped: find the manufacturer's torque value, set a calibrated driver to it, and land every terminal to that number — checking your units before you pull. On critical or high-load terminations, some electricians mark the screw and lug with a paint pen after torquing, so a later thermal-cycle loosening shows as a misaligned mark at the next service visit.
It adds a minute per connection. Against that minute is a failure mode that's invisible at install, accelerates on its own, and surfaces as a charred panel or a thermal-imaging callback a year later. Few habits in the trade have that lopsided a payoff.
This is the kind of detail Voltly is built to keep at your fingertips: when you're standing at a panel and need the torque value, the conductor ampacity, or the box fill behind the connection you're about to make, having the reference offline in your pocket means you tighten to spec instead of to feeling — every terminal, every time. If that's the way you'd rather work, you can find it at https://voltly.lumenlabs.works.