Why a Breaker That Trips Is Usually Not the Fault

Why this matters

The cheapest thing in the panel is the part everyone blames. A breaker trips, someone swaps it, the customer pays for a part and a trip, and three weeks later it trips again with a different tech standing there. Every one of those visits threw away the same evidence: the device operated because current, or a current difference, or a waveform signature crossed a threshold that somebody engineered on purpose. Treating that as a defect is throwing away the only permanently installed instrument the circuit has.

The one instrument that is always installed

Nothing else on a branch circuit is watching all the time. There is no logger, no trend, no alarm history. There is one device with a published time-current characteristic, sitting in series with everything, that has been integrating current since the last time somebody reset it. When it operates, it has told you something with a shape: a threshold, a duration, and which of its internal elements got there first.

That is worth more than most of what a tech can measure in twenty minutes on site, because it covers hours you were not there.

The gate, stated once

Can the condition the device responds to be shown present at that device's own terminals, at the time it operated?

Unit of analysis is one device on one circuit, on one trip event, not "this panel trips a lot." The condition has to be shown at the device's terminals rather than anywhere on the circuit, because ambient heat at the breaker and current in the conductor are two different inputs to the same thermal element. Answer yes and the device is reporting, and your job is upstream or downstream of it. Answer no, and only after you have measured the load, the ambient and the terminations, the device becomes a candidate - and a candidate is not a finding until a comparison makes it one.

Read which element operated before you read anything else

A common thermal-magnetic breaker holds two elements, and the timing of the trip tells you which one acted.

• Thermal (inverse-time): operates in minutes at modest overcurrent, and the further above rating the faster. It integrates heat, so it answers questions about sustained load.
• Magnetic (instantaneous): operates in a fraction of a cycle at a multiple of rating, commonly in the region of ten times rating for a standard molded-case breaker, though the actual multiple is a property of that product's published curve and not a constant. It answers questions about faults.

Devices with added functions report something else entirely: a ground-fault or arc-fault device can operate at a current far below the handle rating, because magnitude is not what it is measuring. On those, the trip indicator or the reset behaviour is part of the evidence, and the sibling articles on ground faults and on arc-fault devices cover what each one is responding to.

A thermal element that operates at 40 minutes and a magnetic element that operates at 20 milliseconds are two different reports. Recording the trip without recording the timing throws away half of it.

The failure mode that actually dominates is the opposite one

Protective devices do degrade. The direction they degrade in is the part that gets stated backwards.

The mechanisms that age a breaker in the field - contact erosion from interrupting fault current, mechanism lubricant stiffening with age and heat, corrosion in a damp or corrosive enclosure, repeated switching duty when a breaker is used as a light switch - overwhelmingly push toward failing to open, or opening slower, or welding closed. The failure that makes headlines is a device that did not clear a fault it was supposed to clear, not one that opened when it should not have.

That asymmetry is why "the breaker is bad" belongs at the end of the hypothesis list rather than the start. It is also the reason a device that has cleared a serious fault deserves inspection rather than a reset: an interrupting event is exactly the duty that erodes contacts, and what a device can survive interrupting is the subject of a separate article on interrupting ratings.

Case one: the gate answers yes

A 30 A circuit feeding a piece of packaged equipment. It trips the instant the disconnect is closed, every time, and it has already had two breakers.

Live readings here fall under 29 CFR 1910.333(a)(1), which requires de-energizing before working on or near exposed energized parts unless the employer can demonstrate that de-energizing introduces additional or increased hazards or is infeasible for equipment-design or operational reasons; for the isolation work itself the standard is 29 CFR 1910.333(b)(2) in general industry or 29 CFR 1926.417 on construction work, with the live-dead-live proving sequence at NFPA 70E-2021 120.5 as adopted by your employer's electrical safety program. In this case almost nothing needs to be measured live, which is the point.

Apply the gate. The trip is instantaneous, so the magnetic element operated, so the current present was somewhere near ten times 30 A, roughly 300 A, for a fraction of a cycle. Is that condition possible on this circuit for an innocent reason? Starting inrush on this equipment class runs a few times rated current for a few seconds, not ten times for a fraction of a cycle, and it would engage the thermal element rather than the magnetic one if it engaged anything.

So the gate answers yes: a current of that magnitude was genuinely present, and the device is reporting a fault. De-energized, verified dead, an insulation resistance test conductor to conductor and conductor to ground finds one leg low. The conductor is chafed at a strain relief where the flexible connection enters the equipment.

What the two swapped breakers bought: nothing, twice, plus the risk of eventually installing a device that would not clear the fault at all.

Case two: the gate answers no

A 20 A branch circuit feeding a fixed load. It trips after roughly 50 minutes, repeatably, and it has done it four times in two weeks.

Measured steady current: 15.0 A. Against the 20 A rating that is 75 percent. A standard branch breaker is not listed to carry more than 80 percent of its rating continuously, meaning three hours or more, unless the assembly is specifically listed for 100 percent operation, and 80 percent of 20 A is 16.0 A. So 15.0 A is a legitimate continuous load with 1.0 A of headroom under the continuous limit, and the thermal element should not be operating there.

The rest of the gate, before the device is allowed to be a candidate:

• Ambient at the breaker. Molded-case thermal elements are calibrated at a reference ambient, commonly 40 degrees C under UL 489, the listing standard the product was tested to in the edition it was listed under, which binds through the listing and through the adopted NEC's requirement that the device be listed. Above that reference the device carries less before operating. Measured here with the cover on and the load running: 38 degrees C, which is inside the reference, so heat is not the explanation.
• Terminations. A thermal scan of the line and load lugs under load, taken with the enclosure open only for as long as the scan takes and under the same live-work gate above, shows no rise against the neighbouring poles.
• The load itself. Steady, no cycling, no starting event during the 50 minutes.

The gate answers no. Now, and only now, the device is a candidate, and it becomes a finding by comparison rather than by assertion: move this load to an adjacent identical circuit and it holds for a full shift; put a different, comparable load on the suspect breaker and it trips again inside the hour. The device has been shown to be the variable.

And the finding does not stop at the part. The panel record shows this same breaker cleared a fault on a failed compressor two years ago and was reset without inspection. Contact erosion runs both ways depending on the current: it is a fail-to-open mechanism at fault current, and a trip-early mechanism at load current, because the eroded contact's added resistance heats the thermal element from inside. That is the exception the section above did not cover, and it is the mechanism, and it means the replacement is a repair rather than a lucky swap.

The comparison protocol that turns a candidate into a finding

Two-way swap, both directions, or you have proved nothing:

1. Move the load to a known-good circuit of the same rating and hold it for longer than the observed time-to-trip, with a margin. If the observed trip was at 50 minutes, an 8-hour hold is a real test and a 60-minute hold is a coin flip.
2. Put a comparable load on the suspect device so the device is exercised, not just rested. A breaker that is never loaded will always look fine.
3. Record ambient and load current during both halves, because a swap that also changes ambient or load has changed two variables and settles nothing.
4. Keep the removed device until the circuit has run clean through the same duration that used to trip it. If the trip follows the load rather than the breaker, you would rather still have the old part than have a story.

How to verify you got this right

• Write the timing down before you touch anything. Instant, minutes, or an hour. If nobody can tell you, the first job is a reset and a stopwatch, not a part.
• State the ratio, not the amps. Measured current over device rating, and separately over the 80 percent continuous limit where the load runs three hours or more. Fifteen amps means nothing on its own; 75 percent of rating with 1.0 A of continuous headroom is a fact you can reason from.
• Ask what current the operated element requires. If it was the magnetic element, name a multiple of rating that the circuit would have had to see and decide whether the load could produce it innocently. If nothing could, you have a fault whether or not you have found it yet.
• Check that your conclusion survives a reset. If the answer is "the breaker was weak," the circuit should run clean past the old time-to-trip on the new device with the same load. If it trips again, the original device was reporting after all and you have now lost a week.

References

• 29 CFR 1910.333(a)(1) and (b)(2), safety-related work practices for electrical work in general industry; 29 CFR 1926.417 for the construction counterpart
• NFPA 70E-2021, 120.5, as adopted by your employer's electrical safety program
• UL 489, the listing standard for molded-case circuit breakers, in the edition the installed device was listed under
• NEC Articles 210 and 240, including the continuous-load provisions, in the edition your authority having jurisdiction has adopted
• See related: How to Read a Trip as Evidence; What an Interrupting Rating Is and Why It Is Not a Trip Setting; How Fuses and Breakers Fail