The Heat Pump That Tripped the Main When the Car Charged

Why this matters

A main breaker that trips is the whole house going dark, and on a converted house in January that means no heat. The first call on this one came in as "the heat pump keeps killing the power," and the first two techs treated it as a heat pump problem. It was not. The heat pump was running exactly as designed, and so was the car charger, and so was the main breaker. What failed was a load calculation that assumed the car and the backup heat would never run at the same time, and a phone app that quietly made sure they did.

First, when a main trips

Tell the customer to leave the main off if there is any burning smell, heat at the panel door, buzzing, or visible scorching, and to call back; that panel is an electrician's, not a reset. With none of those signs, one reset is reasonable, but a main that trips again under the same conditions stays off until someone finds out why, because a breaker tripping on overload is a breaker doing its job. Nobody but a qualified electrician removes the panel cover: the conductors feeding the main are live with it off, and there is no disconnect above them except the utility's.

The house and the complaint

A 1,900 square foot house with a 125-amp, 120/240-volt service, converted two winters earlier: a ducted cold-climate heat pump with 10 kilowatts of resistance backup, and a 240-volt heat pump water heater. Last spring the shop added a Level 2 charger on a 60-amp circuit, set to 48 amps. The main had tripped four times in two weeks, every time between 6:30 and 7:00 in the morning, every time on a day below about 20 F. Afternoons and mild mornings were fine.

The pattern was the first real clue, and it was nearly thrown away. "Cold mornings only" sounds like a heat pump symptom, so both early visits went straight to the heat pump.

What the office now asks on a main-trip call

The shop wrote down the questions that would have sent the first visit to the right place, and the office now asks them before anyone is dispatched:

• What time did it trip, each time? A cluster at one time of day points at coincident loads, not at a failing part.
• What was the outdoor temperature? Cold-only trips on a heat pump house point at backup heat being part of the stack.
• What was running, including things that run themselves: a car on a schedule, a water heater on a timer, a dryer someone started?
• What has been added or changed since the last trip-free winter, including a new car, a new charging plan from the utility, or a new household member?
• Did any branch breaker trip, or only the main?

On this house the answers were 6:30 to 7:00, below 20 F, the car plus whatever the kitchen was doing, a new charger since last winter, and only the main. That is a load story before anyone opens a truck door.

The candidates, and what killed each

A heat pump fault. The second tech checked the obvious: did the heat pump's own breaker or the backup heat breaker trip? No; only the main. A heat pump with a locked compressor, a shorted strip or strips staging all at once would usually trip its own circuit breaker first, because that breaker is far smaller than the main. The tech clamped the backup heat in operation: about 41 amps, which is 10,000 watts divided by 240 volts, 41.7 amps, as the nameplate says. The compressor drew what its nameplate allowed. Defrost brought the strips on for a few minutes, as designed. Nothing here was drawing more than it should. Eliminated.

A failing main breaker. Breakers can weaken and trip below their rating. That needed an electrician with the cover off, and the shop sent one with a thermal camera and a current logger. Under a morning load, the main and its lugs showed no hot spot against the conductors on either side, and the torque on the lugs was at the panel's labeled value. A weak breaker would have shown up as a trip at a current below 125 amps. What the logger showed, below, was a trip well above it. Eliminated, and the reason why matters: the breaker was not the fault, it was the witness.

The utility. Low voltage makes some motor loads draw more current. The logger's voltage channel read 238 to 242 volts leg to leg all morning, steady. Eliminated.

A ground fault somewhere. A residential main breaker of this size usually carries no ground-fault function at all, so a ground fault would trip a branch GFCI or a breaker, not the main. The charger's own ground-fault protection never opened. Eliminated.

That left load, which the logger had been recording the whole time.

What the logger recorded

The electrician installed a recording clamp meter on each service conductor, working on the energized panel under the shop's NFPA 70E 130.5 risk assessment and arc-rated PPE, and left it for a week. The fourth morning was 18 F.

LOGGER  service conductors, 125 A main, 1 min avg
time    L1 A   L2 A   notes
04:50    73     73    HPWH element + car start
05:40    99     99    + compressor, blower
06:05   141    141    recovery, strips on
06:20   164    155    kettle, toaster, microwave
06:47     0      0    MAIN TRIPPED

Build L1 at 6:20 from the nameplates. The loads that run on both legs at 240 volts: compressor 22 amps, backup strips 41.7, air handler blower 4, charger 48, water heater element 18.8. That is 134.5 amps on each leg. The 120-volt loads add the rest: about 6 amps of background on each leg all morning, plus a kettle, toaster and microwave at 6:20 that happen to sit on L1 circuits (23.5 amps) and a hair dryer and bath fan on L2 (14.5 amps). That is 29.5 amps on L1 and 20.5 on L2. L1: 134.5 plus 29.5 is 164.0 amps. L2: 134.5 plus 20.5 is 155.0 amps. Both match the logger.

164 amps on a 125-amp main is about 1.3 times its rating, and a thermal-magnetic breaker at that overload opens on a delay that depends on its own trip curve (within the limits UL 489, the molded-case breaker standard, sets) and how warm it is. Here the main ran over its rating from 6:05, at about 1.13 times (141 amps), reached about 1.3 times at 6:20, and opened at 6:47, 42 minutes after it first went over. That is the breaker working.

The calculation that let it happen

The charger quote's load calculation was in the file. It used a year of the utility's 15-minute demand data, which included the heat pump's winter, and it came out over 125 amps once the charger was added at its 48-amp setting. The estimator then wrote the line that decided the job: "EV charges 11 pm to 5 am off-peak; backup heat runs at recovery after 6 am; treat as noncoincident." With the charger dropped from the total on that basis, the calculation passed.

Two things were wrong with that line.

The code does not allow it. The NEC's noncoincident-load allowance (220.60, in the edition your jurisdiction has adopted) covers loads that will not run together, the classic case being heating and cooling. A charger whose schedule the owner can change from a phone is not that kind of load. Without a listed energy management system controlling it, the calculation has to count the charger whenever it might run, which is any time.

The schedule changed anyway. Over the summer the utility moved its off-peak window to midnight to 7 am. The owner set the car to "ready by 7:30," and the car's software did what that setting does: it delayed charging to finish as late as the window allowed, which started it around 4:50 and ran it straight through the thermostat's 6 am recovery. Nobody had changed anything the estimator controlled. The assumption broke on its own.

The fix

The shop offered a service upgrade or load management. The upgrade was quoted at many weeks because the utility needed a transformer study. The owner took load management, which is what the arithmetic supported.

A listed energy management system monitoring both service conductors, set to shed controlled loads when either leg passes a setpoint. The 2023 NEC at 220.70 lets the calculation count the controlled service at the system's setpoint, as adopted by your jurisdiction; earlier editions handle the charger case through 625.42(A). The shop set the setpoint at 110 amps, below the 125-amp rating, as a margin of its own choosing.

The shed order, checked against the logged morning. At 6:20, L1 without the charger is 164 minus 48, which is 116 amps: still over 110. So shedding the charger alone would not have held the setpoint. The second shed is the water heater element: 116 minus 18.8 is 97.2 amps, under 110. The order is charger first, water heater second, and never the heat.

Schedules moved to match. The water heater was set to heat-pump-only mode except on request, which keeps its 18.8-amp element off on ordinary mornings. The car was set to charge midnight to 4 am instead of "ready by," and the water heater's off-peak recovery was moved to start at 1 am. The thermostat's recovery was set to start at 5 am with a slower ramp, so the compressor carries more of the climb before the strips stage in. Those changes mean the system rarely has to shed at all; the system is the backstop when someone changes a schedule again.

Confirmed on a colder morning. With the logger left in for two more weeks, the coldest morning was 12 F. L1 peaked at 108 amps at 6:25. The charger had finished at 3:40, so there was nothing to shed, and the water heater was idle. The main has not tripped since. See related: Alternatives to a Service Upgrade, which owns how the code counts each of these tools.

Where this case would have gone another way

If the logged fixed load with the charger and water heater both shed had still been over the rating, load management would have meant switching off heat on the coldest mornings, and the upgrade would have been the answer. That is the test that separates the two, and it has to be run on a logged cold morning, not on a summer peak. A house with a second car on order would fail it within a year. See related: Measuring Existing Load Before You Sell a Service Upgrade.

References

• NFPA 70, National Electrical Code, 220.60, 220.70 and 625.42 (2023 edition), as adopted by your jurisdiction, and Article 750 on energy management systems
• NFPA 70E (current edition), Standard for Electrical Safety in the Workplace, 130.5 risk assessment
• UL 489, Molded-Case Circuit Breakers, for how trip time varies with overload
• See related: Alternatives to a Service Upgrade
• See related: Time-of-Use Rates and Equipment Settings