Solar inverter shutting off on sunny afternoons? Grid overvoltage, the trip limits and the wire math

Is it voltage? Read the shape of the dropout

Several different things can make a monitoring graph look wrong at midday, and only one of them is an overvoltage trip. The shape of the curve is the quickest way to tell them apart, before anyone opens a panel. An overvoltage trip produces a cliff: output falls from near its peak to zero for that inverter, stays there for a short period while the inverter waits to reconnect, then ramps back up. It happens most on clear days with low household demand, often on weekends and holidays when neighbours on the same transformer are also exporting and nobody is running air conditioning.

Inverters log the reason for every disconnection. The wording varies by manufacturer, but look for event names that mention grid overvoltage, AC voltage out of range, or grid voltage high. If the inverter or its app keeps a voltage history, export it; the log is worth more than any description of the problem, because it shows the voltage the inverter measured at the moment it stopped.

The three numbers that set your margin

Whether an inverter trips depends on three numbers stacked on top of each other: the voltage the utility delivers to your meter, the voltage your own conductors add between the meter and the inverter while it exports, and the trip setting inside the inverter. The first and third are written down in public documents. The second is arithmetic, and it is the one an owner can most often change.

The utility's number comes from its tariff. Pacific Gas and Electric's Electric Rule 2 (sheet 4, effective December 3, 2023) commits to operating its distribution circuits, under all normal load conditions, so that service voltage on a 120 V nominal service stays between 114 and 126 V, and on a 240 V nominal service between 228 and 252 V. The same rule lists exceptions: weather, infrequent short fluctuations, service interruptions, and causes beyond the utility's control. It also states that, for energy conservation, the utility regulates distribution voltages to the extent practicable to keep service voltages at the low end of the band. Other utilities publish their own service rules; the 114 to 126 V band on a 120 V base is the same Range A band the PG&E table reflects, and your own utility's rule is the document to read.

The inverter's number comes from the interconnection standard. The National Rural Electric Cooperative Association's guide to IEEE 1547-2018 reproduces the default trip values: the first overvoltage threshold, OV1, is 1.10 per unit of nominal voltage, and the second, OV2, is 1.20 per unit, the same thresholds as in the 2003 standard but with longer clearing times. OV2 must clear in 0.16 seconds in every category. OV1 must clear in 2 seconds for Category I and II equipment and 13 seconds for Category III. The guide adds that installations generally use these defaults unless the utility instructs otherwise, and many do: following the IEEE 1547a-2020 amendment, many utilities and ISO New England specify 2 seconds for Category III as well. Your interconnection approval states the setting that applies.

Read the table from the top down and the problem becomes concrete. A meter sitting legally at 252 V leaves 12 V before the default OV1 threshold. Everything between the meter and the inverter's terminals, the service conductors inside your property, the interconnection breaker, and the run of wire to the inverter, raises the voltage the inverter sees while it pushes current out. If that rise plus the utility's voltage reaches 264 V for longer than the clearing time, the inverter must stop.

Why exporting pushes the voltage up

Current only flows from a higher voltage to a lower one. When your house imports power, the utility's side is higher and the voltage at your outlets sags slightly below the meter. When your panels export, the direction reverses: the inverter has to raise its own terminal voltage just above the grid's to push current out through the wires. The higher the current and the more resistance in the path, the bigger that difference, which is Ohm's law, voltage equals current times resistance, applied to every conductor between the inverter and the utility transformer.

That path has segments on both sides of the meter. On the utility side are the service drop or underground service lateral and the distribution transformer that serves your street. On your side are the service entrance conductors, the main panel, and the circuit from the interconnection point to the inverter. Neighbours matter too. When several homes on one transformer export at the same moment, they all push current back through the same transformer, and the voltage at every one of those homes rises together. That is why trips cluster on clear, mild weekends: high production, low demand, and many systems exporting at once.

The utility controls the starting voltage, and some feeders start high. The PG&E rule's commitment is to keep voltage within the band under normal load conditions; a feeder set high so that homes at the far end do not sag at peak demand can leave homes near the start of the circuit close to the upper limit on quiet afternoons. That is a utility question, and the evidence log later on this page is how to raise it.

The wire-rise arithmetic, from the NIST copper tables

The part of the rise inside your property can be calculated. On a 240 V split-phase circuit, the inverter's current flows out on one hot conductor and back on the other, so the resistance that matters is the conductor's resistance per foot times twice the one-way length. Multiply by the current, and the result is the voltage rise in that run.

The resistance values come from the National Institute of Standards and Technology's Copper Wire Tables, Handbook 100 (1966), Table 6, which lists ohms per 1,000 feet of standard annealed copper at temperatures from 0 to 200 degrees Celsius. Conductors carrying an inverter's full output on a summer afternoon run warm, so the calculation below uses the 75 °C column: 0.7639 ohms per 1,000 feet for 8 AWG, 0.4806 for 6 AWG, and 0.3022 for 4 AWG. The inverter is a 7.6 kW AC unit at full output on a 240 V service, which is 31.7 A.

Two features of the table matter for a homeowner. Rise scales directly with distance, so an inverter mounted in a detached garage or at the far side of a ground-mounted array 150 feet from the panel spends three times the margin of one mounted beside the panel. And moving up one or two wire sizes cuts the rise sharply: at 150 feet, going from 8 AWG to 4 AWG removes about 4.4 V, more than a third of the 12 V margin. The calculation covers only the run you can see on the plan set. Loose or corroded terminations add resistance the table does not show, which is one reason an electrician measures rather than calculates when a problem appears after years of normal running.

A worked case: 250 volts at the meter

This illustration uses the figures above, with every input stated. It describes no particular home.

A home's meter reads 250 V on a clear Sunday afternoon, inside the 252 V ceiling in the utility's rule. The 7.6 kW inverter sits 100 feet away, fed by 8 AWG copper. At full output, the inverter's terminals sit about 4.8 V above the meter, roughly 254.8 V. That is well short of the 264 V OV1 threshold, so on these numbers the inverter keeps running. Under California's default smart inverter settings, it would already be absorbing some reactive power, because 254.8 V is above the 247.2 V point where the default volt-var curve begins.

Now change one input. Several neighbours export at the same time and the meter voltage climbs to 259 V, above the utility's band. The inverter sees about 263.8 V, a fraction of a volt below the trip line. A few more watts of export, a slightly warmer conductor, or one loose termination is enough to cross 264 V, and the inverter must stop within 2 seconds. Once it stops, its own contribution disappears, the voltage falls, the inverter waits and reconnects, and the cycle repeats. That on-off rhythm is the signature in the monitoring graph.

The lesson from the arithmetic: in this case the owner's wiring contributed under 5 V, and the decisive change was on the utility side. In another home, a 150-foot run of smaller conductor would make the owner's side the main contributor. Only measurements at both the meter and the inverter separate the two, which is why the evidence log comes before any fix.

Smart inverter settings: volt-var, volt-watt and why output may sag instead of trip

Modern inverters can do more than trip. IEEE 1547-2018 requires inverters to have voltage regulation capabilities, including volt-var and volt-watt functions, but leaves the decision to use them, and the settings, to the utility. The NRECA guide notes that, by default, the standard puts an inverter in fixed power factor mode at unity power factor, meaning no reactive power support unless the utility asks for it.

California went further. PG&E's Rule 21 advice letter 6635-E-C (submitted May 4, 2023, effective August 29, 2023) sets the default activation states for UL 1741 SB inverters: dynamic volt-var and volt-watt are both activated, and fixed power factor is deactivated. The default volt-var curve holds unity power factor between 97 and 103 percent of nominal voltage, ramps to 30 percent reactive power absorption at 107 percent, and injects reactive power at low voltage. Absorbing reactive power pulls the local voltage down without cutting the energy you produce, which can keep an inverter below its trip line on a marginal circuit.

Volt-watt works differently: as voltage rises, the inverter reduces its real power output. That prevents trips, but the energy is still lost; it just disappears more gradually, as a sag below the expected curve rather than a cliff to zero. An owner in a state with these settings active can therefore have a voltage problem without ever seeing an overvoltage event in the log. If the curve sags only on the brightest, quietest afternoons, ask whether volt-watt is active and pull the voltage history for those hours.

Build the evidence log before calling anyone

A utility will check your service against its rule; an electrician can check your side. The habit that makes this easy is the monthly monitoring check described in the solar maintenance routine: an owner who already reads the event log notices the pattern weeks earlier. Both work faster with data, and a few weeks of records usually settles which side the problem sits on.

1. Export the inverter's event log and voltage history for every day with a dropout. Note the time of each event and the voltage the inverter recorded just before it.
2. Record the conditions: weather, the time of day, and whether household demand was low. A cluster on clear, mild weekends points to export-driven rise.
3. Ask a licensed electrician to measure voltage at the meter and at the inverter terminals at the same time, while the system is exporting near full output. The difference is your side's rise; the meter reading is the utility's side. Do not open the meter or the main panel yourself.
4. Compare the meter reading with your utility's published service band. On PG&E's rule, anything above 252 V on a 240 V service is outside the normal-condition range.
5. Check whether the dropouts started after a change: a neighbour's new system, utility work on the street, a battery or an expansion added to your own system, or an electrical repair.
6. Report to the utility with the log attached. PG&E's voltage complaint process sends a field technician to check the service against Rule 2, may leave a recording volt meter for a few days if nothing is found at first, and has its voltage reliability team respond within 48 hours of receiving that meter data. The utility asks customers to say whether the complaint involves a solar system.

Who fixes which part

Once the measurements are in, the fix follows the side that owns the problem. PG&E's own page states both outcomes: if the high voltage is caused by customer-side generation, the utility notifies the customer of its findings and helps resolve the issue; if the service voltage is within the rule, it tells the customer the problem may be on the customer's side.

One remedy is off the table for an owner: changing the inverter's trip settings on your own. The settings are part of what the utility approved when it let the system interconnect, and the NRECA guide describes default values staying in place unless the utility instructs otherwise. Any change goes through the utility and the installer, in writing. If the system is under a lease or PPA, the owner of the equipment handles it, and the contract's performance terms may apply; the ownership comparison explains who holds which obligation.

Raising the trip limit as the first fix

Forum threads often end with someone advising that the inverter's overvoltage setting be raised a few volts. In the US, that setting is part of the interconnection the utility approved, and changing it without the utility's agreement can breach that agreement. It also hides the evidence: if the utility's voltage is outside its own published band, raising your trip point lets the problem continue on the whole street. Measure first, report to the utility, and change settings only through the utility and installer, in writing.

What repeated trips cost

Overvoltage trips cost energy, not hardware, and the amount depends on how often they happen and when. A labelled illustration shows the scale. Suppose a system producing about 6 kW at midday loses a combined 90 minutes on each of 40 sunny days in a year. That is 6 kW times 1.5 hours times 40 days, or 360 kWh. Against a planning example of 11,000 kWh a year, it is about 3.3 percent of annual output. At a labelled 30 cents per kWh of avoided purchases it would be about $108 a year; if most of the lost energy would have been exported at a much lower credit, it is worth less, which is why your export rule decides the dollar value.

That cost is small enough that an expensive fix on the owner's side needs to be weighed against it, and large enough to justify the free steps: the evidence log, the utility complaint, and a check of terminations. The trips also distort any comparison with the production estimate. If you are measuring the system against its estimate or a production guarantee, keep the event log; the production estimate guide explains how a guarantee clause treats shortfalls, and whether grid-caused downtime counts is a question for the contract's exact wording.

Where the simple explanation breaks

• 208 V services. Some apartments, condominiums and small commercial buildings are served at 208 V from a three-phase system, not 240 V split phase. PG&E's rule gives 197 to 218 V for that service. An inverter must be configured for the service it is connected to, and a mismatch is an installer question, not a utility one.
• Battery systems. A battery charging from the panels at midday reduces export and can reduce voltage rise; a battery discharging to the grid adds to it. A trip pattern that changed after a battery was installed may reflect its operating mode. The guide to adding a battery to existing solar covers how coupling choices change what the inverter sees.
• Expansions. Adding panels or a second inverter raises export current on the same conductors, so a circuit that never tripped can start. The guide to expanding an existing system covers the interconnection checks an expansion triggers.
• Flickering lights or appliances behaving oddly. Lights that brighten and dim with load changes, or appliances failing, can indicate a loose neutral or another wiring fault, which is a safety problem rather than a solar setting. Call the utility and an electrician; do not wait for a monitoring pattern.
• Low voltage instead of high. Undervoltage events at the end of long rural feeders are a different problem with different trip settings; the same logging steps apply.