Here’s the puzzle every solar skeptic eventually lands on: a 20-panel array with one panel shaded loses far more than 5% of its output — but nobody who quotes you a loss figure explains why. The reason is wiring, and understanding it tells you both what shade actually costs and what can fix it.

Why panels in a string are hostages to each other

Residential panels wire in series into “strings” feeding the inverter. In series, every module must pass the same current — and the weakest-lit cell sets it. A panel shading to half output doesn’t politely contribute half; it throttles the string’s current toward its own. Ten percent shade on one module can cost 30% or more of that string’s power — the “nonlinear” part that makes shade both a physics problem and a quoting problem.

Bypass diodes are the built-in pressure relief: each panel carries (usually) three, letting current skip a shaded cell-group instead of flowing through it. When they work as designed, a fully shaded section gets bypassed cleanly — the panel drops out rather than dragging everyone down, and losses stay closer to proportional. The catch is real-world mess: partial shade that leaves a panel producing some voltage can keep the diodes from tripping cleanly, which is when the string-level tax is worst.

The three fixes, in ascending price

1. Design around the shadow. Hour-by-hour shade modeling (installers run this from a photo tool; you can sanity-check with a season of observation) tells you whether the chimney matters at 9 a.m. or 4 p.m. — a shadow that only lands at 4:30 costs far fewer sun-hours than the morning one.
2. String-level mitigations. Put shaded panels on a different string or a different MPPT input so the clear string runs at its own best point. Cheap where the layout allows it.
3. Module-level electronics. DC optimizers (SolarEdge-style) or microinverters put per-panel tracking on each module — the shaded one degrades alone. Typical add-on cost: roughly $40–100/panel equivalent; on a patchy roof it pays for itself in recovered output, on a clear roof it’s mostly a nicer monitoring app.

Honest shade percentages — the calculator’s range

Our itemized loss field defaults to 3% for shading/mismatch/downtime — the “my roof is basically clear” figure. For real shade:

Roof reality	String inverter	+ Microinverters/optimizers
Clear all day	0–3%	0–3%
Edge shade ≤1 h/day, or light morning haze	5–10%	4–8%
Recurring shadow line midday hours	10–20%	8–14%
Heavy tree cover	don’t build	still don’t build

The last row is sincere: a roof shaded for a third of its sun-hours is usually a ground-mount or a “no” — no inverter topology recovers photons that never arrived.

Shade vs soiling vs clouds — don’t confuse the budgets

Three different theft mechanisms people lump together:

• Shade is geometry — objects between you and the sun. Partially recoverable by electronics; it lives in the shade line of the loss stack.
• Soiling is film on the glass — dust, pollen, salt, ash. ~1–4% where rain does the cleaning; a cleaning service exists for a reason in dusty climates.
• Clouds are already inside your PSH — the regional number is an average of clear and cloudy days, so adding a “cloudy days” loss on top double-counts them. That’s the most common double-loss error in DIY estimates.

Measuring before you pay

Free and cheap options, in order:

1. Stand on the roof line (or your yard eye-level) on a clear day near solar noon in different months — shadows move more than intuition suggests; December reveals objects July hides.
2. Solar-pathfinder-style site tools — every serious installer runs a dome-camera or LIDAR shade survey that yields a % shading figure per year. Ask for the number and the map, not just “it’ll be fine.”
3. PVWatts’ shade input — the same NREL tool that estimates output lets you plug a shade fraction; compare its shaded and unshaded runs for your address.

Then put the percentage into the calculator’s itemized losses and watch the payback move. An honest 10% shade on a marginal deal turns a 10-year payback into 11.3 — still a working project. An honest 25% turns it into “call the tree service first,” which is also a useful answer.

Frequently asked questions

One of my 20 panels is shaded an hour a day — how much do I lose?

On a basic string inverter, that panel's shaded cells can drag its substring down far beyond the 5% a naive share suggests — losing 15–30% of that hour's string output is common. With bypass diodes working properly and short shade duration, the hit stays closer to proportional. With microinverters or DC optimizers, it's nearly proportional: ~5% for that hour plus the diffuse light the panel still makes.

Are microinverters worth it for a shady roof?

Often, yes — they put the maximum-power-point tracking on each panel, so one shaded module stops taxing its neighbors. Typical premium on a 20-panel job is $800–2,000 vs a string inverter; on genuinely patchy roofs it earns that back in a few years of recovered output. On an unshaded roof it's mostly a monitoring convenience, not an output upgrade.

Is shade loss the same every season?

No — it usually worsens in winter. A tree or chimney's shadow lengthens dramatically when the sun rides low: an edge that's clear at June noon can sit in shade at December noon. Honest shade estimates weight the sun-hours lost, not the clock hours.

What can I do about shade besides cutting trees?

Re-layout to dodge the shadow line (designers model hourly shadows on photos), split shaded and clear panels onto separate strings/MPPT inputs, use optimizers on the affected modules only, or accept it as a bigger shade-loss percentage in the calculator and see if the payback still works. Sometimes the honest answer is a different roof plane.

How do I enter shade in the calculator?

Turn on itemize losses — the shading/mismatch/downtime field defaults to 3%. For a roof with a real shadow line, 8–15% is the honest range for a string system; 5–10% for the same roof with module-level electronics. Measure first where you can: a $30 irradiance meter reading at the worst hour beats any guess.