Most solar owners look at one number. Today's kWh. It goes up, the system works, and that's the end of the inspection.
That's a shame, because the inverter is quietly reporting a dozen other figures, and several of them will tell you about a fault weeks before the daily total drops enough to notice. A string that's slowly corroding at a connector doesn't announce itself. It just shaves a bit off, day after day, until the loss is large enough to be visible against seasonal variation. By then you've paid for it.
So what's worth watching, and what's noise?
DC voltage: the first thing to check when something breaks
Your panels produce direct current. The inverter converts it. Between those two facts sits the most diagnostic number on the whole dashboard.
Each string has an open-circuit voltage that depends on how many panels are wired in series and how cold they are. A string of 14 panels with a 41 V open-circuit rating each sits near 574 V on a cold, bright morning, and sags toward 480 V once the modules warm up. That drop is normal. Silicon loses roughly 0.3% of its voltage per degree above 25 C, which is why your array is at its most electrically aggressive in February, not July.
Here's what matters: the number should be proportional. If you have two strings of equal length and one reads 480 V while the other reads 410 V, something in the second string is wrong. A bypass diode has failed, a connector is high-resistance, or a panel has cracked internally. The absolute figure tells you less than the comparison.
If your inverter shows only a combined value, that comparison is unavailable to you, and I'd argue that alone justifies paying for string-level or panel-level monitoring on any array above about 6 kW. You're buying the ability to answer "which one?" instead of "something, somewhere".
MPPT channels and why two identical strings disagree
Maximum power point tracking is the inverter constantly hunting for the voltage-current combination that yields the most watts. Conditions change, the optimum moves, and the tracker follows.
Most residential inverters have two independent MPPT channels. That's deliberate. If you put an east-facing string and a west-facing string on the same tracker, the inverter has to compromise between two different optima, and both lose. Split across separate channels, each finds its own.
Look at your two channels across a full day. East peaks in the morning, west in the afternoon, and the crossover should be smooth. A tracker that oscillates, jumping between operating points, usually means partial shading is confusing it, or the string voltage is drifting near the inverter's minimum tracking window. Neither is catastrophic. Both cost yield.
Current tells you about dirt, voltage tells you about damage
Voltage and current fail differently, and knowing which one moved narrows the diagnosis enormously.
String voltage is set by how many cells are wired in series. Lose a panel to a failed bypass diode and the voltage drops in a visible step. String current, on the other hand, is set by how much light reaches the cells. It scales with irradiance almost linearly, and it's the number that soiling moves.
So which is it when output falls? If current is down across every string by a similar proportion while voltage holds steady, look at the glass. Dust, pollen, bird mess, a film of agricultural spray. If voltage has stepped down on one string while current is unchanged, look for a dead module or a diode.
There's a third case people miss. Both figures look fine, but the two strings on the same tracker have mismatched currents. Series-wired panels all carry the current of the weakest one, so a single shaded or degraded module drags an entire string down to its level. That's the physics behind module-level electronics, and it's covered properly in our comparison of optimisers and microinverters.
AC frequency and voltage: mostly the grid's problem, occasionally yours
The grid runs at 50 Hz in Europe and 60 Hz in North America, and your inverter must match it within a tight band. It'll report both frequency and AC voltage, and for most owners these figures are dull, which is exactly how they should be.
They stop being dull when your inverter starts disconnecting.
Grid voltage rise is the common cause and it's badly understood. When your system exports, current flows out through the same cable it normally draws through, and that pushes the voltage at your connection point up. If your neighbourhood already sits near the top of the permitted range, your own export can tip it past the limit, and the inverter is legally obliged to disconnect. You see a gap in the middle of a sunny day and assume a fault.
It isn't a fault. It's the network telling you the local cable is undersized. The fix is a conversation with your distribution operator, not a new inverter, and I've seen people replace perfectly good hardware chasing this.
Efficiency and the number nobody reads
Inverter efficiency sits around 96-98% for modern units. It's reported, it barely moves, and it's not where your losses live. So why does every marketing sheet lead with it?
Because it's easy to measure and easy to compare. The losses that actually cost you money are messier: soiling, mismatch, shading, cable resistance, and the slow degradation of the modules themselves.
The figure worth building a habit around is performance ratio: actual output divided by what the array should theoretically have produced given the irradiance it received. It normalises away weather. A cloudy month and a bright month produce very different kWh totals but should produce a similar performance ratio.
A healthy rooftop system runs somewhere around 0.75 to 0.85. If yours drifts from 0.82 to 0.74 over a season while the weather stays ordinary, something has changed physically. Soiling, a failing optimiser, a shading object that grew.
That's the whole point. Absolute output can't distinguish "bad weather" from "broken equipment". Performance ratio can. Our guide to lifting PV yield covers what to do once you've spotted the drift, and the efficiency calculator walkthrough explains the arithmetic behind the ratio itself.
Temperature readings, and why the hot number isn't the panel
Your inverter reports its own internal temperature. People mistake this for module temperature. They're unrelated.
Inverter temperature matters because these units throttle. Push a wall-mounted inverter past roughly 45-50 C ambient and it'll begin derating, quietly reducing output to protect itself. In a south-facing garage in August, that's not hypothetical. If you see your peak clipped on the hottest days but not on merely warm ones, and the clipping tracks inverter temperature rather than irradiance, you've found free yield: shade the unit, or improve airflow around it.
Module temperature is a separate measurement, and only some systems have the sensor. It's what you need for a genuine performance ratio calculation, since panel output depends far more on cell temperature than on air temperature.
Error codes deserve more respect than they get
Every inverter logs faults, and most owners never open the log because the system appears to be running. When did you last look at yours?
Two categories are worth separating. Transient events, such as brief grid excursions or a single tracking fault after a cloud edge, are normal and self-clearing. Recurring events with the same code at the same time of day are a pattern, and patterns have causes.
Isolation faults deserve particular attention. That code means the inverter has detected current leaking to earth, usually through moisture in a connector or damaged cable insulation. It often appears on damp mornings and clears once things dry out, which makes it easy to dismiss as a glitch. It isn't. It's the earliest warning you get of a fault that eventually becomes a permanent shutdown, and occasionally a fire risk. Our notes on how solar panels catch fire go into the failure chain in more detail.
A monitoring routine that takes four minutes a month
You don't need to watch this daily. Watching daily mostly teaches you what clouds look like.
Once a month, with a bright day's data in front of you:
- Compare your two string voltages at the same moment. They should track each other.
- Check the MPPT channels peak at different times if your array faces two directions.
- Note the performance ratio and compare against the same month last year.
- Open the fault log and look for repeats, not one-offs.
- Glance at inverter temperature against the day's peak output.
That's it. Five checks, most of which take seconds once you know where they live in your particular app. Whether that app is any good is a separate argument, and SolarEdge and Enphase take genuinely different approaches to what they show you.
The point of all this isn't to turn you into a technician. It's that solar hardware fails slowly and silently, and the daily kWh figure is the last place a slow failure shows up. Everything else on that dashboard exists to tell you sooner.