When solar production slips, it rarely announces itself. Nothing sparks, nothing beeps — the number on your monitoring app is just a little lower than it should be, and it stays that way. The good news is that the data you already have is usually enough to figure out whether you’re looking at dirt, shade, a failing panel, or an inverter problem. You just need a workflow instead of staring at charts and hoping something jumps out.
Here’s the process I use on my own roof, and what each type of monitoring platform can and can’t tell you along the way.
First, Know What “Normal” Looks Like
You can’t spot underperformance without a baseline. Solar output swings enormously with season and weather — a cloudy week in December might produce a quarter of what a clear week in June does, and neither is a fault.
The habit that pays off: on a handful of genuinely clear days each season, note the daily total and glance at the shape of the production curve. Most platforms keep this history for you (Enphase Enlighten and SolarEdge keep years of it; the Tesla app lets you scroll back through daily and monthly views), but knowing your own typical clear-day numbers by memory makes anomalies obvious at a glance. On my roof — 40 panels split across two Tesla inverters — I have a rough feel for what each inverter’s share of a good day looks like, which means a problem on either one stands out quickly.
Step 1: Rule Out Weather with a Clear-Day Test
Before suspecting hardware, wait for a fully clear day and compare its production curve to a previous clear day from a similar time of year. A healthy system on a cloudless day draws a smooth bell curve: ramp up in the morning, peak around solar noon, taper in the evening.
Two curve shapes that look alarming but are usually fine:
- A flat plateau at midday. If the top of your curve looks sliced off at a constant power level, that’s inverter clipping — your array can momentarily produce more DC power than the inverter’s AC rating. It’s a design tradeoff, not a fault, and it’s common on systems sized with a healthy panel-to-inverter ratio.
- A brief dip at the same clock time every day. That’s almost always shading — a chimney, vent pipe, or neighbor’s tree crossing the array. It’s worth knowing about, but it’s geometry, not failure.
What should concern you on a clear day: a curve that’s the right shape but meaningfully lower than your baseline, a curve with erratic sawtooth drops, or a step down partway through the day.
Step 2: Compare Day-Over-Day and Look for the Shape of the Decline
How production degrades over time tells you a lot about the cause:
- A sudden step change — production drops by a large, consistent fraction overnight and stays there — points to hardware: an inverter offline, a string disconnected, or a breaker tripped. Example: on a two-inverter system, waking up to roughly half your usual clear-day production is a strong hint that one inverter is down, and the first thing to check is whether both units show as reporting in the app.
- A gradual slide over weeks or months — each clear day a bit worse than the last — points to soiling (dust, pollen, bird droppings) or slowly encroaching shade from vegetation. Compare this month’s best clear day against the same month last year if your platform keeps the history.
- Erratic, intermittent drops on otherwise clear days point to a loose connection, an inverter derating in heat, or a component starting to fail. This pattern deserves a call to your installer sooner rather than later.
Step 3: Compare Like Against Like
This is the single most powerful diagnostic move, and it works at whatever granularity your system offers: panels or strings with the same orientation and tilt should track each other closely on a clear day. When one lags its identical neighbors, you’ve localized the problem.
- Panel-level systems (Enphase microinverters, SolarEdge optimizers): open the per-panel view on a clear day and scan the array map. One panel producing noticeably less than the same-facing panels around it — consistently, across several clear days — is your suspect. One low reading on one day could just be a bird’s contribution to your soiling problem; a persistent laggard is a hardware conversation.
- String-level systems: if your inverter or app breaks out production per string or per inverter, compare same-orientation strings against each other. On my roof, the two inverters serve as each other’s control group — if one’s daily curve sags while the other holds its usual shape, I know which half of the system to look at.
- No sub-system data at all: skip ahead to the system-level section below — you can still do real diagnosis, just with coarser tools.
What Your Monitoring Platform Can Actually See
Not all dashboards offer the same visibility, and it changes which diagnostics you can run yourself:
| Platform | Granularity | Best at catching |
|---|---|---|
| Tesla app | System level (per-inverter at best) — no per-panel data | Whole-system drops, an inverter offline, curve-shape anomalies |
| Enphase Enlighten | Per panel (each microinverter reports) | A single failing panel or microinverter, localized shading |
| SolarEdge monitoring | Per panel (each optimizer reports) | A single failing panel or optimizer, string issues |
| Sense / Emporia Vue | Whole-home CTs (production + consumption) | Independent confirmation your system is producing at all; separating solar problems from consumption changes |
One honest correction worth making explicit: the Tesla app does not show individual panel output. If you have Tesla string inverters, panel-level diagnosis requires your installer to pull string data or inspect on-site. A home energy monitor like Sense or an Emporia Vue with a CT on the solar feed won’t add panel detail either, but it gives you a production reading that doesn’t depend on the solar vendor’s cloud — useful for catching a dead system fast and for ruling out “is it my usage or my solar?” questions.
Symptom → Likely Cause
| Symptom (clear day unless noted) | Most likely cause | What to check first |
|---|---|---|
| Production flat at zero | Tripped breaker, inverter offline, or grid outage on a non-battery system | Solar breaker, inverter status lights / error code |
| Sudden step drop that persists (e.g., roughly half of normal) | One inverter or one string down | Per-inverter or per-string reporting in the app |
| Flat-topped curve at midday | Inverter clipping — normal behavior | Nothing; it’s expected with a high DC-to-AC ratio |
| Gradual decline over weeks | Soiling or growing vegetation shade | Visual roof check, compare to same month last year |
| Dip at the same clock time daily | Fixed-object shading (chimney, vent, tree) | Watch the curve across a full day; trim or accept |
| One panel lags same-orientation neighbors | Failing panel/microinverter/optimizer, or localized soiling | Persistence across several clear days; photo the panel |
| Erratic sawtooth output | Loose connection, thermal derating, failing component | Inverter error codes; call your installer |
| Production normal but bill/export worse | Consumption change, not a solar fault | Whole-home monitor (Sense, Emporia Vue) usage view |
Diagnosing with System-Level Data Only
If your app only shows whole-system numbers, you can still get surprisingly far:
- Clear-day curve shape — smooth bell curve at the right height, or not?
- Step-change detection — scroll the daily history and find the exact date production dropped. A date pins down “what changed” (storm, critter, someone in the attic) far better than a vague “it’s been lower lately.”
- Year-over-year clear-day comparison — same month, best day vs. best day, expecting modest degradation but not double-digit losses.
- Physical inspection from the ground — binoculars will reveal heavy soiling, debris, or a slipped panel without anyone climbing on the roof.
Bring the evidence to your installer: the date of the step change, screenshots of the before/after clear-day curves, and anything you spotted visually. That turns a service call from an open-ended investigation into a targeted fix — and if a panel or inverter is failing, your dated production history is exactly the documentation a warranty claim wants.
When to Stop Diagnosing and Call Someone
DIY ends at the data. Anything involving DC wiring, connectors, or opening an inverter is installer territory — string voltages are dangerous, and unauthorized tinkering can void warranties. Your job is to arrive at the call with a narrowed-down story: what dropped, when it dropped, and which part of the array is implicated. Monitoring data gets you those three answers; let a professional handle the wrench.