The breaker in a Phoenix garage never trips at noon. It trips at four in the afternoon, after the garage has been baking for six hours and the air conditioner has been running since breakfast. The homeowner resets it. It holds through the evening. It holds the next morning. Then at four o'clock it does it again, and the family starts scheduling their lives around a piece of plastic they no longer trust. When the electrician finally arrives at nine in the morning, everything checks out. The meter is happy. The breaker is happy. The house, as far as anyone can prove, is fine.

Here is what nobody tells that family: the breaker isn't defective, and it isn't lying. It is measuring exactly what it was built to measure. The uncomfortable part is that what it measures is not current. It is heat.

A breaker doesn't count amps. It takes a temperature.

A standard thermal-magnetic breaker has two separate ways to trip. The magnetic element is a coil that responds to the violent surge of a short circuit, and it acts in a fraction of a cycle. The thermal element handles the slow stuff, the overload that creeps up over seconds or minutes. It is a bimetal strip: two metals bonded together that expand at different rates when warmed. Current flows through the strip, the strip heats from its own resistance, the two metals pull against each other, the strip curls, and when it curls far enough it knocks the latch loose.

Notice what actually moves the latch. Not amperes. Deflection. And deflection depends on the temperature of the strip, which is the sum of two things added together: the heat the current generates inside it, and the heat of the air around it. The bimetal has no way to tell those two apart. A strip warmed by a hot enclosure needs less current to reach its trip point than the same strip sitting in a cool basement.

The engineers who designed these devices knew this, so they picked a reference. UL 489, the standard that molded-case circuit breakers are listed to, calibrates them at an ambient of 40°C, which is 104°F. A 20-amp breaker is a 20-amp breaker at 40°C. Warmer than that, it trips at less. Cooler, it holds more. The number on the handle is not a promise. It is a measurement taken at one temperature, and the panel in that garage is not at that temperature.

What a hot panel does to the number on the handle

Breaker manufacturers publish ambient rerating curves for exactly this reason, and they are worth pulling for whatever brand is in the panel. The shape is consistent across makers: as the enclosure climbs past 40°C, the current the breaker will carry without tripping falls, and at 50°C a typical residential breaker is holding somewhere in the neighborhood of 90 percent of its nameplate. A 20-amp breaker in a 50°C panel is, for practical purposes, an 18-amp breaker. The air inside a panel on a west-facing garage wall in August gets there without much effort.

The heat has three sources, and they stack. The first is the room: a garage or attic can run 15 to 20 degrees above outdoor air on a sunny day. The second is the sun itself, if the panel cover catches direct light through a window or sits on an exterior wall. The third is the breakers. Every breaker under load is a small heater, and a two-pole air-conditioner breaker sitting in the middle of a full panel is being warmed on both sides by its neighbors. The temperature that matters is not the thermostat in the hallway. It is the air touching the bimetal.

Why the load gets bigger on the exact same afternoon

This is where the physics turns cruel. The circuit that trips is almost always the air conditioner, and the compressor does not draw a constant current. The condenser coil has to push heat into outdoor air, and the hotter that air is, the higher the refrigerant pressure the compressor has to work against. Higher head pressure means more mechanical work, and more work means more current. The rated load amps on the unit's nameplate are a hot-weather number, and on the worst days of the year the compressor finally gets close to it.

Meanwhile the copper feeding it has warmed up too. Copper's resistance rises roughly 0.4 percent for every degree Celsius, so the conductor running through a 60°C attic is dissipating more heat than it did in April. It is a small effect, but it points the same direction as everything else.

So you have two curves. The breaker's effective capacity falls as the afternoon heats up. The load rises as the afternoon heats up. At nine in the morning they are comfortably apart. At four o'clock they cross. Nothing is broken. Two correct numbers simply met.

There is a reason this particular failure wears people down in a way that a dead circuit never does. A thing that is broken gives you a stable story. A thing that works most of the time and fails at unpredictable hours keeps reversing its own verdict, and each reset that holds quietly teaches you the breaker is fine, right up until it isn't. People in this situation often start to feel faintly unreasonable, as if the house is arguing with them. It helps to know the house is simply doing arithmetic.

The one thing that imitates ambient heat and isn't

Before blaming the weather, rule out the imposter. A conductor that was never torqued properly at the breaker lug has a high-resistance joint, and that joint heats. The heat conducts straight into the breaker body, right where the bimetal lives. The result looks identical from the outside: a breaker that trips below its rating, worse in summer, fine in the morning.

The tell is comparative. Point an infrared thermometer at the suspect breaker and at its neighbors under similar load. A panel that is uniformly warm has an environment problem. A single breaker that is noticeably hotter than everything around it has a termination problem, and that is not a nuisance. NEC 110.14 exists because loose connections are how panels catch fire. The two problems call for opposite fixes, a torque screwdriver versus a bigger circuit or a cooler panel, and the thermometer is what tells you which one you are holding.

The mistake that makes the house less safe

The tempting fix is to pull the 20 and install a 30. The breaker stops tripping and everyone sleeps. But the 12 AWG wire in the wall did not get any bigger, and under NEC 240.4 the overcurrent device is sized to protect the conductor, not to accommodate the load. A 12 AWG conductor on a 30-amp breaker can now be overloaded with no device on earth positioned to notice. You have not solved the problem. You have removed the only thing that was reporting it.

If the load genuinely needs more room, the conductor and the breaker grow together, and the continuous-load rules in 210.19 and 210.20 still apply. More often, the honest answer is smaller and less satisfying: measure the real current, measure the real temperature, apply the manufacturer's rerating curve, and discover whether the circuit was ever sized for the hottest hour of the hottest day.

Your next moves

  • Keep a trip log for three days. Write down the time of each trip, the outdoor temperature, and what was running. A breaker that only trips between 2 and 6 p.m. on days above 95°F is telling you its story in plain language.
  • Clamp the circuit at peak heat, not in the morning. Read running amps on the compressor circuit at the hottest part of the afternoon and compare them to the breaker rating and the unit's rated load amps. A continuous load sitting above 80 percent of the breaker is already in the margin the rerating curve eats.
  • Shoot the panel with an infrared thermometer. Compare the suspect breaker to its neighbors and note the air temperature inside the enclosure. One hot breaker means a termination to check and torque to spec. A uniformly hot panel means ambient.
  • Find the rerating curve for the exact breaker in the panel. Every major manufacturer publishes one. Apply the measured enclosure temperature and write down what the breaker is actually rated for on that wall in August.
  • Fix the heat before you touch the breaker size. Shade a sun-struck panel, ventilate a sealed garage, and verify torque on every termination on the circuit. Only if the corrected numbers still don't fit should the circuit be upsized, and then the conductor grows with the breaker.

Carrying the math into the garage

Most of this comes down to numbers that are easy to look up in an office and hard to look up on a ladder in a 115-degree garage with no signal. The ampacity tables, the temperature correction factors, the conductor sizing for a bigger circuit, the voltage drop on the long run to the condenser, and the code sections that decide whether a fix is legal all live in Voltly, fully offline, built for the moment when the panel is open and the homeowner is standing behind you asking why it only happens at four o'clock. If you want the reference and the calculators in one place that works without a connection, it is at https://voltly.lumenlabs.works.