Solar inverter offline, or not producing? How to tell

A solar inverter offline in the app and a solar inverter not producing are two different faults, and the app cannot tell you which one you have. If the inverter’s own display or LEDs show a kWh figure climbing at midday, your panels are fine and all you have lost is your monitoring. If the display shows zero, or an error, in good light, you are losing energy right now. And if everything looks normal but the total is below what the same month produced in earlier years, you may have one dead string or one dead panel, the failure none of the three platforms on this page alerts you about by default.

Work through the three checks below in order. It takes about fifteen minutes and it will tell you which of the three you have, before you pay anyone to come and look.

First: is it night, snow, or a power cut?

Rule out the four states below before anything else. Each one puts the symptoms of a dead system on a perfectly healthy one.

• It is dark. An inverter runs on the DC voltage coming off the panels. When that voltage disappears at dusk, the inverter and its wifi stick shut down and the site goes offline in the app. SolarEdge labels this officially as 18x3, Night Mode, “an inverter status indicating that the input voltage is below 6 V DC”. Judge nothing after sunset.
• Snow or heavy cloud. Enphase lists low-light conditions caused by heavy cloud cover or snow first among the causes of its Microinverters Not Reporting error. Snow lying on the array looks exactly like a dead system. Whether to leave it or clear it is covered in snow on your solar panels.
• There has been a power cut. A grid-tied inverter has to disconnect when the grid goes down, and only restarts once its reconnection time has passed. SolarEdge’s code for this is 18x60: “Grid is down. When AC grid voltage returns the inverter will restart after the reconnection time.” What is worth checking once the grid comes back is set out in a power cut and your solar panels. Our guide on a home battery for your solar panels sets out whether one would have kept anything running through an outage like this, and what extra hardware doing so would take.
• The app has updated, or the brand’s cloud is having a day. A stale portal is not a fault on your roof.

So: look again around midday on a clear day. If it still looks dead then, carry on.

The three-way split: no data, no production, or one dead string

Everything that gets called “my solar isn’t working” is one of these three. The tests, the urgency and the person you ring all differ.

The third one sets off no alarm anywhere:

Over Christmas I noticed a panel underperforming. It was maxing out at 36w output! I contacted the installers only to discover that the original installer operated as an independent franchise & MyPlanet who I have the warranties with have washed their hands of it

A reply on the same thread explains why this kind of fault sits undetected for months:

I’d certainly not notice a 10% reduction, however 40%-60% would stand out like a sore thumb

One panel or one string was gone, with no error code anywhere and nothing in the app that named it. What would have caught it is a comparison of what came out against what should have come out.

The check that works on any brand: your generation meter at midday

Before you touch the app again, go and read hardware. This is the test that does not depend on the manufacturer, on your wifi, or on anyone’s cloud.

1. Pick a bright moment, ideally 11:00–15:00 with sun on the panels.
2. If you have a generation meter (look beside the consumer unit and in the loft), write the reading down, wait fifteen minutes and read it again. It should have moved. If the meter has an LED, it flashes only while generation is happening.
3. If you have no generation meter, use the export side instead: your smart meter’s in-home display, your supplier’s app, or the export figure on your account. A household exporting at sunny midday is a household whose panels are producing.
4. If nothing else is to hand, switch on a 2–3 kW load (kettle, immersion) and watch whether your import figure barely moves. It is crude, but a system making 3 kW is obvious.

If the meter moves, your panels and inverter are working and what you have is a monitoring problem. That is a different and far cheaper problem. The meter also catches the opposite case, and this owner found the fault only because he read the meter:

I had got Solar panels installed on our roof in March 2015. In Jan of ’17 I noticed that the solar electricity meter was not reporting any units. … I did the same and found that the inverter was totally blank and seemed dead.

The meter is also the thing your money is paid on. Ofgem states that Feed-in Tariff payments “are based on meter readings that FIT generators submit to their FIT licensee”, and Octopus puts it more bluntly still: “we can’t make FiT payments based on estimated or inverter readings”. A stopped meter is a stopped income, whatever the app says.

SolarEdge: “no communication”, or not producing?

Go and stand in front of the inverter. On the HD-Wave and SetApp generation there are three LEDs, and the official LED table reads:

• Green ON: the system is producing.
• Green blinking: AC is connected but the system is not producing.
• Green flickering (on for 100 ms, off for five seconds): night mode.
• Blue ON: the inverter is communicating with the monitoring platform.
• Red ON: system error.

The combination that matters most here is green on, blue off: producing perfectly, telling nobody. That is a pure communication fault. The reverse, red on while the app looks healthy, is a production fault.

On older units with an LCD, the status screen shows S_OK for as long as the connection to the SolarEdge monitoring portal is up. No S_OK means the inverter is not reaching SolarEdge: comms again, not production. The same screen shows P_OK: XXX/YYY, where XXX is the number of optimisers that have sent data in the last two hours and YYY the number paired. If those two numbers do not match, there may be a problem in one or more of the optimisers. That screen is your partial-fault detector, on the wall, for free.

Error codes appear on the LCD and in the app as hex characters with a literal x separator: 18xB5, 18x3, 2x19 and so on. 18xB5 is DC/AC in-stage overvoltage, with the official action “Contact Support RMA related”. Older firmware showed decimal numbers for the same faults: legacy error 25, “Isolation Fault”, is the modern 2x19. In the monitoring platform the comms wording is “No communication with the site” or “No communication with the Inverter”. That is a platform alert, not something the inverter’s screen ever says. The individual codes are decoded in the SolarEdge error-code guide. More on the brand at SolarEdge monitoring.

Enphase: gateway not reporting, microinverters not reporting, or genuinely not producing?

Enphase splits the same three-way question across two alert names, and the distinction is written into their own documentation.

Gateway Not Reporting: “If your gateway loses its internet connection, an error notification appears in the Enphase App, showing a Gateway Not Reporting alert.” That is comms. The line to remember comes from the microinverter article: “If the gateway is offline, all microinverters may appear as ‘not reporting’ even if production continues.”

Microinverters Not Reporting: some or all panels are not sending data to the gateway, and Enphase says this “does not always mean your system has stopped producing power”. Affected panels render black on the ARRAY screen. The causes they list: low light from heavy cloud or snow, electrical interference on the power lines, a gateway that has been moved or is on an unsuitable socket, a tripped circuit breaker, wiring or equipment issues, and a faulty microinverter in rare cases.

The IQ Gateway (the Envoy, if yours predates the rename) has four separate LEDs, not one, and the one that settles the comms question is the network communications LED:

Amber or red there is a gateway that can see your router but not Enphase: comms, not production. Health states in the app are Normal, Warning and Error; a shortfall shows as Low Solar Production in the app and Production Issue in the installer portal. How many panels are affected narrows it further, and that split is worked through in Enphase microinverter not reporting. More at Enphase monitoring.

Growatt: inverter offline, or datalogger offline?

Growatt splits the job across two devices, and the first thing to settle is which of them you are looking at: the inverter itself, or the wifi datalogger stick (the dongle) plugged into its side. The stick’s indicator reports on the link to your router and to ShineServer; it says nothing about whether the panels are producing. So if that light is doing something you do not recognise, treat it as a monitoring question and go and read the inverter’s own display instead. The stick’s colours are decoded in Growatt datalogger red light.

If there is a number on the screen, do not look it up without your model family: the same fault carries different numbers on different Growatt families, and the same number can mean one thing as an Error and something else as a Warning. The full family-by-family breakdown is in our Growatt error codes 203 and 117 guide, and there is more on the platform at Growatt monitoring.

How long before it counts as a fault

Give it 24 to 48 hours, then stop waiting. Enphase’s own guidance on a low-production system is to monitor it “for the next 24-48 hours” before escalating to the installer.

After that window the rule is simple. Stale app but a display showing kWh climbing at midday: annoying, not urgent, so chase the comms. Display showing zero in good light, or an error code, or a meter that has not moved in two days: that is a fault, and every further day is energy you never get back.

The reason to hold yourself to a deadline is that the alternative is drift. Here is one owner describing that drift, and one describing how narrow the catch is even when it goes well:

Gutted to get back from holiday and find that the inverter stopped with an error on the day we left. 2 weeeks of prime August generation lost :( All it took was a turn off for 10 mins, then back on to resolve it but as we were away we lost out.

my SE 5000H inverter failed last summer (15 months from install) so a bit easier to spot when production fell off a cliff.

A partial failure does not fall off a cliff the way the second one did.

What to say to your installer, and what to attach

Give them the diagnosis you have just made, not the symptom. That is the difference between a diagnosis over the phone and a chargeable visit that turns up a flat wifi stick.

Say which of the three it is, then attach:

• The make, model and serial number of the inverter, off the label on the side.
• The install date and, if you know it, who commissioned it, because the warranty hangs on this.
• The last date and time the system produced anything, from the app or the meter.
• The exact code or LED pattern, copied character for character (18xB5, Error: 203, “green on, blue off”), plus a photo of the screen or the LEDs.
• Two meter readings taken fifteen minutes apart at midday, with the times.
• What you have already done: checked the AC breaker and the RCD in the consumer unit, power-cycled the datalogger, waited 48 hours.

One line worth adding if the system is still in warranty: put the date the fault started in writing. Which? found that 15% of the more than 2,000 members it surveyed in April 2024 had had problems with their solar inverter, the most commonly reported solar problem in that survey. This is an ordinary thing to claim for, and the claim goes more smoothly when the start date is documented rather than guessed at.

Why nobody told you

The alerting in all three platforms was built for installers running fleets, not for you.

On SolarEdge, only the account manager or admin, usually your installer, can set alert rules, and an e-mail reaches you only if they tick a per-site permission called “Expose to alerts notifications”. Its no-communication alert needs at least three hours of silence, and by default it arrives in a daily digest. And SolarEdge has no underperformance alert at all: a system that is up and communicating but producing less than it should raises nothing. If your installer is no longer trading, the account behind that toggle is its own problem: your solar panel company has gone bust.

Enphase does more than the other two, and publishes its thresholds: one microinverter silent for 12 hours of daylight is an in-app Warning with no push, and a notification needs at least 10% of microinverters silent for five consecutive days. On an eleven-panel roof, one dead panel is 9%, which is under the bar, permanently.

The platforms alert on comms and on hard faults, and none of the three alerts a homeowner on gradual underperformance unless the installer has configured a production threshold. Solmiro does: every day it works out what the weather over your roof allowed, and it e-mails you when production falls well short of that or stops. Connecting is free. If you would rather run the check by hand, our guide on how do I know if my solar panels are working sets out the manual routine.

What offline looked like on a real roof

On a 6.09 kWp Enphase roof Solmiro checks, the link to the Enphase cloud dropped for 37 minutes on 26 August 2026 while the panels kept producing, so no production was lost.