Key Takeaways
- Monitoring caught a fault on my own rooftop system days before I would have spotted it on my electricity bill, saving weeks of quietly lost generation.
- The tell was not a dramatic zero-output alarm but a slow drift: one string reading lower than its twin under the same sun.
- You do not need fancy analytics to catch most faults; a daily glance at per-string or per-inverter numbers does the heavy lifting.
- Set an underperformance alert so the system pings you instead of relying on you to remember to check.
- Document your normal baseline early, so an abnormal day is obvious rather than a guess.
Catching a fault via monitoring is the difference between a shrug at your annual bill and a quick fix that pays for itself. I am Arjun Mehta, and over years of commissioning and living with rooftop solar I have learned that panels rarely fail loudly. They fade. This is the story of one fault on my own 5 kW system that monitoring caught early, what the data looked like, and the exact steps I took to confirm and fix it. By the end you will know which numbers to watch, how to tell a real fault from a cloudy afternoon, and how to set up alerts so your system tells you before your wallet does.
The day the numbers stopped agreeing
My home system runs two strings into a single string inverter, wired so each string sees roughly the same roof, the same tilt, and the same sun. For two years the two strings tracked each other almost perfectly. On a clear morning in the pre-monsoon heat, I opened the app out of habit and something felt off.
String A was pushing its usual healthy figure. String B was sitting noticeably below it, on a cloudless day, with both facing the same sky. That gap between two strings that should be twins is the single most useful fault signal a home system gives you, and it is why I always recommend per-string visibility over a single lumped total.
Why monitoring saved my system
Here is the honest bit: without monitoring I would have missed this for weeks. A single underperforming string on a two-string system does not trip anything. The inverter keeps running, the house keeps drawing power, and the loss hides inside a bill you only read once a month. The fault was quiet by design.
What monitoring gave me was a comparison. One string against the other, today against last week, this hour against the same hour on a similar clear day. That context is what turns a plausible number into an obviously wrong one. If you want the full picture of how these systems work end to end, see our complete guide to solar monitoring.
Tip: On day one of a new install, screenshot a clear-sky afternoon showing both strings at full tilt. That is your baseline. Six months later, when something looks low, you have a real reference instead of a fuzzy memory.
Ruling out the innocent explanations first
Before I climbed onto the roof, I ran through the boring causes, because most low readings are not faults at all. A string can read low for perfectly ordinary reasons, and chasing a phantom fault wastes an afternoon and a ladder trip.
The checklist below is the exact order I work through. It moves from cheapest to check to most involved, so you spend effort only when the easy answers are exhausted.
| Possible cause | How it looks in monitoring | Real fault? |
|---|---|---|
| Passing clouds or haze | Both strings dip together, then recover | No |
| Morning or evening shading | One string low only at fixed times of day | Usually no |
| Dust or bird droppings | Gradual, whole-array decline over weeks | Maintenance, not a fault |
| Loose DC connector or fuse | One string steadily lower in full sun | Yes |
| Failing panel or bad diode | Persistent gap, worse under load and heat | Yes |
String B was low in full sun, all day, not just at shading hours, and both strings had been cleaned the same week. That combination pointed past the innocent rows and toward something physical on the DC side.
A trick I lean on: match the suspect day against a known-good day with similar weather from a few weeks earlier. If String A repeats its old figure but String B falls short of its own past self, weather is off the hook. The reference is the past, not just the neighbour. That is why keeping a little history matters more than any single fancy chart.
Reading the pattern like a performance ratio
The concept that made this obvious is performance ratio: how much energy you actually harvested versus how much that sunlight should have produced. When one string quietly slips below the other, its effective ratio drops while its twin stays put. That divergence is the fingerprint of a localized problem rather than a bad-weather day.
If the whole array had dropped together, I would have suspected weather, soiling, or a hot inverter derating itself. Because only one string sagged, the fault had to live in that string alone. For a deeper walk-through of this metric, read Performance Ratio in Solar, Explained.
The step-by-step I followed to confirm and fix it
Once the data pointed at String B, I stopped guessing and worked methodically. Here is the sequence, and I would follow the same order on any home system.
- Compared both strings side by side across three clear days to confirm the gap was consistent, not a one-off.
- Checked the inverter event log for any DC fault, arc, or isolation warnings around the same dates.
- Isolated the system safely at the DC switch before touching anything, then waited for the panels to sit idle.
- Walked the String B array visually, looking for a discoloured cell, a cracked backsheet, or a scorched connector.
- Found a DC connector on String B that had worked slightly loose, its contact dulled and warm-looking at the housing.
- Had it re-terminated properly, powered back up, and watched the two strings snap back into agreement within the hour.
Safety warning: Solar DC strings stay live whenever there is daylight, and a degraded connector can carry enough energy to burn or arc. Do not open DC connectors or probe live conductors yourself. Isolate at the DC switch and call a qualified installer for any work on the DC side, mains, or roof.
What this fault would have cost if I had ignored it
A single degraded string does not cost you half your output overnight. It bleeds you slowly. Left alone through the sunniest months, a persistently low string can quietly forfeit a meaningful slice of that string's generation, and in the worst case a hot, loose connector becomes a fire risk rather than just a yield problem.
Because monitoring flagged the divergence within days, the actual loss was small and the safety risk was closed out fast. That is the real return on watching your data: not a bigger number on a good day, but avoided losses and avoided hazards on the bad ones.
There is a net-metering angle too. In many Indian states you are credited for what you export, so a string quietly underproducing does not just raise your own consumption, it shrinks the surplus you bank against future bills. A fault you cannot see is a credit you never earn. Over a full billing cycle that quiet gap compounds, which is exactly why I treat monitoring as part of the system, not an optional add-on.
How to make your system tell you next time
I got lucky that I happened to open the app that morning. You should not rely on luck. The fix is to move from checking to being alerted, so an abnormal day reaches you even when you are not looking.
Most portals let you set a threshold: notify me if today falls below a percentage of expected, or if one string diverges from another. Set it a little loose at first so cloudy days do not spam you, then tighten it once you know your baseline. The full method is in our guide on how to build solar underperformance alerts, and if you want to go further, Advanced Solar Performance Analytics Explained covers the deeper trend work.
For the standards and safety background behind DC fault detection, the international framework for PV system monitoring is worth a skim; the IEC maintains the relevant PV performance and monitoring standards, and India's MNRE publishes rooftop programme guidance many installers follow.
Frequently Asked Questions
How did monitoring catch a fault my inverter did not alarm on?
The inverter kept running normally because one low string on a two-string system does not trip a fault. Monitoring caught it by comparing the two strings against each other, which revealed a persistent gap on clear days that no single alarm threshold would have flagged on its own.
What is the earliest sign of a solar fault in monitoring data?
The earliest sign is usually divergence rather than a dramatic drop: one string, inverter, or panel reading lower than its identical neighbours under the same sun. A slow, one-sided decline over several clear days is far more telling than a single bad afternoon.
Do I need expensive analytics to catch faults at home?
No. A daily glance at per-string or per-inverter numbers catches most home faults. Advanced analytics help at scale or for subtle trends, but for a single rooftop system, a clear baseline plus a simple underperformance alert does the vast majority of the work.
Should I fix a low string myself?
No. Solar DC strings stay live in daylight and can arc or burn at a bad connector. Diagnose from the monitoring data, isolate at the DC switch, then call a qualified installer for any physical work on the DC side, mains, or roof. Data work is safe; hardware work is not.
Conclusion
The lesson from my roof is simple: faults hide, and monitoring drags them into the light. A five-minute habit and one alert threshold turned a fault that could have bled generation for a month into a same-week fix. Set your baseline, watch your strings against each other, and let the system nudge you. If you are setting this up for the first time, start with our complete guide to solar monitoring, and if you want to know more about who is behind this advice, our about page lays it out.
