The short version for the service desk.
- The problem: the monitoring portal tells you the inverter is offline and which code it threw. It cannot tell you whether the DC isolator is off, the RCD tripped at dawn or the display is dark, and those are the details that decide whether a van goes out.
- The fix: the owner or site contact records six things on their own phone, guided step by step, with no call to schedule. The desk reviews the case when it lands, separates the alerts that need a visit from the ones that do not, and loads the van with the right part.
- Who this is for: service managers and support leads at solar installers, PV maintenance providers and battery storage installers, on residential and commercial rooftops. Utility-scale plants are a different post.
Monday morning, forty alerts in the portal. Twelve say "inverter offline". Three say "ISO fault". One customer has already phoned twice, and what she can tell you is that the light is red and the app "says nothing is coming in".
Your planner has a choice. Book a van on the strength of a red light, or spend twenty minutes on the phone trying to talk someone through the garage in the dark. Most desks do the first, because the second rarely works.
The van arrives on Wednesday. The DC isolator is off. The customer's electrician turned it off last week while fitting a car charger.
That visit was never a diagnosis problem. The code was right there in the portal all along. What the desk lacked was a picture of the physical state around the code, and nobody on site knew which picture to take.
If you run service at a solar installer or an O&M provider, this is the post for that gap. It covers the six things the person on site should film, how the desk uses them to split visits from non-visits, what changes when the fault is in the battery unit, and how the same case documents the fix afterwards. Venta Capture, a product of VentaVid, is the guided capture tool we describe, but the process works in any tool that lets you define the steps. If you would rather test it than read about it, start a free account and build the inverter flow before the next Monday queue.
In this post:
The alert is not the diagnosis
Most of what gets sold as solar inverter fault remote diagnosis is a portal login. A monitoring platform is very good at one thing: telling you that an inverter has stopped exporting and which fault family it logged. Grid overvoltage, isolation resistance, communication lost, offline.
What it cannot see is everything around the box. Whether the AC isolator is on. Whether the consumer unit RCD dropped at 6 am when the roof was wet. Whether the display is blank because the unit is dead or because the DC switch is off. Whether the "communication lost" is a failed data logger on a router someone unplugged.
The same alert in the portal can be produced by six different physical situations, and only one of them needs an electrician. The other five need a conversation, a switch or a reset by the right person.
The inverter is where the losses concentrate. Across 67 GW of utility and commercial projects analysed in 2024, inverters caused 39% of equipment-related power loss, ahead of strings at 22% and combiners at 19%, according to Raptor Maps data reported by pv magazine USA. That dataset is bigger than a rooftop portfolio, but the pattern holds on residential and commercial roofs: when production stops, the inverter is the first suspect, and the first phone call.
The nuisance-trip case shows how far a code can be from a cause. PV inverters leak a little current to earth, typically 5 to 10 mA and more on a wet morning, so a shared RCD trips at dawn and the system looks dead until someone resets it, as electricians on the Electricians Forums explain. The portal shows "offline at 06:12". The customer shows a tripped breaker. Those are two different jobs.
The asset-management side has the same problem at a bigger scale. SolRiver Capital describes an inverter that tripped on a grid overvoltage event and needed multiple visits, because nobody had mapped that code to a test plan, so each contractor arrived to find out what to test. A code-to-action map fixes half of that. The other half is knowing what the site looked like before the van left.
I have sat next to solar service desks working an alert queue, and the pattern is consistent. The desk almost never lacks the fault code. It lacks the answer to "is the DC isolator on?", and that question quietly decides more visits than any code table does.
The six things the site contact records
Every homeowner guide in the search results says the same thing: photograph the display before you call. GreenLancer's reset guide tells owners to gather the brand, the exact code or light pattern, when it happened, whether weather or an outage was involved, and any smell or scorching, then take a photo of the code.
Good advice. It sits in the wrong place. It relies on the owner reading a blog before phoning you.
The desk should own the request: which views, in which order, with which question attached, sent as a secure, personal capture link the moment the alert fires. That is what guided capture means in practice, and it is the difference between "send us some photos" and a submission you can act on. The mechanics of writing a flow are covered in the guided photo capture post and on the guided video capture feature page, so this section stays on what a solar flow contains.
Two design details matter more than the list.
- Look, do not touch. The flow instructs the site contact to film switches, not operate them. The safety faults (arc, insulation, earth) must not be reset without investigation, and the STS fault-code guide is blunt about that. Your flow carries the same instruction in the customer's own language.
- A yes/no question steers the steps. "Is there a battery?" adds step 5. "Can you see any scorching or smell burning?" ends the flow and flags the case red for the desk. That is the organisation's expertise, built into the flow so a homeowner never has to guess what matters.
The capture is asynchronous. Send the link at 8:10 when the alert arrives; the owner films it at lunchtime in the garage; the desk reviews at 14:00 with the code, the isolators and the meter all in one case. Nobody scheduled a video call, and nobody spent twenty minutes narrating a consumer unit over the phone. The pre-visit assessment post covers why that timing matters for any field service desk; for solar it matters twice, because the owner is usually at work when the alert fires.
Splitting alerts into visit and no visit
With the six views in hand, the desk sorts the queue in a way a code table alone never allows.
The published reset-versus-technician splits are a decent starting point. The 2026 inverter error code guide from Energy Solutions Intelligence puts grid over- and undervoltage, frequency codes and one-off communication errors in the "one clean reset, then stop" group, and puts ground or isolation faults, arc faults, persistent codes, overheating, hardware failures and any visible damage in the technician group.
Now add the evidence, and the queue turns into four piles.
- Talk it through. DC or AC isolator off, RCD tripped, display in standby. The desk calls the customer with the picture in front of it and walks them through the one switch that matters, or dispatches their own electrician if the RCD keeps dropping. No van from you.
- Reset once, then watch. Grid code, isolators on, LEDs showing a fault, no damage. The customer resets once under instruction; the desk watches the portal for 24 hours. The case stays open with a note.
- Visit, prepared. Blank display with every isolator on and the meter at zero. That is an inverter, and the case already holds the model, serial and firmware from the rating plate. Book the van with the replacement on it.
- Visit, urgent. Arc fault, isolation fault, scorching, smell, water in the enclosure. The flow flagged it red at step 6, and the case explains to a technician what they are walking into.
The point of the sort is that the first pile is large, and until now it was invisible to the desk. Venta Capture is used for exactly this triage in customer support and field service, where 23% of customer problems are solved without dispatching a technician. Your share will depend on how many of your alerts are switches and breakers rather than silicon, and the only way to know is to run the flow against a month of alerts.
The remote diagnostics post explains the general shape of remote triage without sensors, and the truck roll post covers how to measure what you avoid. Both apply to solar unchanged. What is specific to solar is that you usually do have the sensor data, and it still cannot see a switch.
One caveat the tool will not remove: the qualified electrician decides. A guided capture gives the desk a clearer view of a code and an enclosure. The judgement about whether an ISO fault is a wet connector or a damaged cable stays with a person who is allowed to make it.
Preparing the right replacement
The second pile of value sits in the visits you still make. A solar van that arrives without the right part is a second visit by definition, and on rooftop work the second visit often means a second scaffold or a second access appointment with a facilities manager.
The evidence narrows the part list before departure.
- DC isolator. Water in the isolator, discoloured housing or a switch that will not stay in position on camera. The van carries the isolator and the glands, and the technician knows to check the string on the same side.
- String or optimiser. Isolators on, inverter running, the app shows one string or a cluster of panels at zero. The case points at a string, not the inverter, and the van carries connectors, fuses and, where fitted, a spare optimiser.
- Inverter swap. Blank display with everything upstream live, or a hardware code that returned after one reset. The serial and model plate from step 6 lets the desk open the manufacturer RMA before the visit, so the replacement is on the van rather than on order.
Model, serial and firmware in the case before the van leaves is what turns a diagnostic visit into a repair visit. That matters commercially because the labour is mostly yours to carry. EnergySage notes that most manufacturers do not reimburse the labour on a warranty replacement, and Solar Insure adds that truck rolls, shipping and labour are typically limited or excluded for the term of the parts warranty. Every wasted trip on a warranty job comes off your margin, not the manufacturer's.
Track the result as first-time fix rate per alert type. The inverter-swap pile should sit near 100% once the evidence flow is in place, because the only unknown left is on the roof.
Battery storage units
Battery storage changes the fault picture in one important way: there are now two boxes and a link between them, and a "communication error" in the portal can mean the battery, the hybrid inverter or the cable.
The site contact's flow gets one extra step and two extra questions.
- The battery's own state. LEDs on the battery unit, its own breaker or isolator, the state of charge on the inverter display. A battery at 0% with its breaker off is a different job from a battery at 60% that the inverter cannot see.
- The link. Where the communication cable enters both units, filmed once at each end. Loose plugs and chewed cables show on camera; a failed board does not, which is itself useful: it pushes the case towards a visit with the right module.
- What changed. A yes/no on whether anything was installed, moved or powered down recently. The answer explains more battery "faults" than the code does.
Warranty adds a wrinkle worth documenting. Communication equipment tends to carry about five years of cover against 10 to 12 years on a string inverter and 25 on microinverters, per EnergySage. The desk wants to know which box is the suspect before it decides who pays.
For battery faults the physical state decides the visit more often than the code does, because three components share one alarm. The same six-plus-one flow, sent by SMS or WhatsApp the moment the alert fires, gives the desk that state without a call. The field service page shows how the same link-based capture is used across equipment types.
Documenting the fix, and the maths
The visit happens. The isolator, the string or the inverter is replaced. Then the most under-documented moment in solar service arrives: the technician closes the lid, the portal turns green, and nobody records what was actually done.
Run the same flow from the other side. Staff can start a capture themselves without sending a link, so the technician films the old serial, the new serial, the final display with the code cleared, every isolator in the on position and the meter reading, and it lands as a case linked to the original alert. The proof of work post covers the mechanics; for solar the payoff is specific.
- The RMA. Manufacturers ask for logs and photos with a warranty claim, and the Energy Solutions guide advises keeping them from the first fault onwards. A case with the original code, the enclosure, both serials and a server-verified receipt time is that evidence, in one place, rather than a technician's camera roll.
- The O&M contract. Commercial customers want to see what was found and fixed. Copy the view link into the ticket, or copy the case for the CRM, and the facilities manager has it without a login.
- The next alert. Six months later the same site throws the same code. Two linked cases show what the enclosure looked like before and after the last visit. Cases remember what people forget.
The evidence standard behind all of this, timestamps, receipt time and a sealed submission, is laid out in the inspection photo evidence post. The timestamp proves when the submission was received, not when the fault occurred, and the post explains why that distinction matters in a dispute.
What the sort is worth, in your numbers
Treat the figures below as an example to redo with your own costs.
Say the desk handles 40 inverter or battery alerts a month that currently produce a visit. A single avoided truck roll saves an installer roughly $225 on average, according to NREL data cited by Tigo. If the evidence flow moves one alert in five into the "talk it through" pile, that is eight visits, or about $1,800 a month, before counting the second visits it prevents on the remaining 32.
For a regional or commercial portfolio the tiers are steeper. Power Factors' break-even method puts a local roll at about $125, a regional one at $400 and an overnight one at $1,500. Avoiding one regional visit a week is a different business case from avoiding a local one, and you will know which you have after the first month of cases.
Neither of those numbers is a Venta Capture result. They are industry figures to plug your own alert count into. If you want to see the flow against your real alert types before you build anything, book a demo and bring last month's queue.
For utility-scale sites and wind, see remote fault diagnosis for wind and solar; for charge points, remote EV charger fault diagnosis.
Frequently asked questions
Can a solar inverter fault be diagnosed remotely without portal access?
Partly. The code itself is on the display, so a 20-second video of the screen gives the desk the same fault family the portal would, and the isolators, LEDs and meter give it the physical state the portal never has. The diagnosis still belongs to a qualified electrician; the capture only decides what they need to see and whether they need to travel.
Should the customer reset the inverter before filming?
No. Film first, then reset once only if the desk says so and the fault is in the reset-safe group. Repeated resets wipe the pattern support needs to see and are never appropriate for arc, isolation or earth faults.
What if the site contact is a facilities manager rather than the owner?
The flow is the same; only the recipient changes. A commercial rooftop usually has a plant room or a roof-level inverter station, and the facilities team can reach it far faster than your van can. Send the link to whoever holds the keys.
Does this replace the monitoring platform?
No. The portal fires the alert and keeps the fault history; the capture supplies what the portal cannot see. Most desks send the capture link as the first response to the alert and treat the two together as one triage picture.
Is the footage enough for a manufacturer warranty claim?
It is a strong start. A case carries the original code on screen, the rating plate, both serials after the swap and a server-verified receipt time, which is the kind of record manufacturers ask for. Whether a specific claim is accepted remains the manufacturer's decision.
What does Venta Capture cost?
Prices are not published. There is a free plan with no credit card required, which is enough to build the inverter flow, send it to your own phone and then to real customers.
Start with your noisiest alert
Every solar desk has one alert type that books the most vans and fixes the fewest faults. For most it is "inverter offline"; for battery installers it is the communication error.
Pick the alert type that books the most vans and make it the first flow. Six steps, one yes/no question about the battery, one about scorching. Send it as the first response to the next 40 alerts of that type, and count how many cases closed with a phone call and how many vans arrived with the right part.
What it takes to start.
- Free plan, no credit card, and a first standard flow is live in about ten minutes.
- Nothing for the customer to install. The link opens in the phone browser, in their own language.
- The electrician still decides. The capture changes what the desk can see, not who signs off the fault.
Stuck on the flow? Book a free setup call and we build your first inverter flow together.
The VentaVid team builds Venta Capture for service, claims and inspection teams.
Start for free and build the inverter flow
If you would rather see it against last month's alert queue first, book a demo and bring your worst alert type. How the review and retake step works is on the video review and retake page.