---
title: "Solar panel rated power vs real output | ComparePV"
description: "Why a solar panel rarely delivers its rated power, and how much heat, weaker light and power tolerance take away, from catalogue figures and PVGIS."
source: "https://comparepv.com/articles/stc-power-versus-real-output"
image: "https://comparepv.com/static/images/og/articles/stc-power-versus-real-output.jpg"
updated: 2026-10-09
---

# STC power versus what you actually get

Why a solar panel rarely delivers its rated power, and how much heat, weaker light and power tolerance take away, from catalogue figures and PVGIS.

By [Bart Szablowski](https://comparepv.com/authors/bart-szablowski) · Updated 9 October 2026 · 10 min read


## The analysis

On this page

1. How the rated power is measured
2. Heat and light on a sunny hour
3. A whole year in three cities
4. What this means when you buy
5. Example: the most-offered module
6. What these figures don't tell you

Figures and denominatorsRelated readingReferences

**In short:** A panel's rated power is measured at standard test conditions: 1000 W/m² of light on cells held at 25 °C, a combination a roof rarely offers. In full sun the cells run above that temperature and every degree costs power, so the median module in our catalogue gives 96.7 % of its rating on a cold, clear hour and 75.4 % in the NOCT hour. Weaker light and the power tolerance on the label move output further.

The label isn't wrong. It's the one figure every manufacturer measures the same way, which is why you compare modules on it. Just don't expect to read it off your inverter at noon.

### How the rated power is measured

Standard test conditions fix three things: 1000 W/m² of light, a cell temperature of 25 °C, and the AM1.5G spectrum. That spectrum (air mass 1.5) is the mix of wavelengths in clear-sky sunlight after it has crossed one and a half times the thickness of the atmosphere.

Nobody waits for the sky to deliver that. A solar simulator makes the light: a lamp filtered to the AM1.5G spectrum, set to the right strength, with a meter recording current and voltage.

![Laboratory solar simulator, a large dark enclosure on a metal rack with measurement instruments underneath](https://comparepv.com/static/images/articles/solar-simulator-lab-am15.webp)

*A laboratory solar simulator that reproduces the AM1.5G spectrum so a device can be measured under standardised light, on a rack with its source-meter below. Image: [ECTran71](https://commons.wikimedia.org/wiki/File:Solar_simulator_at_U-Dub.jpg), [CC BY-SA 4.0](https://creativecommons.org/licenses/by-sa/4.0).*

A factory does the same at line speed. Each cell, and later each finished module, gets a short flash while a tracer records its I-V curve, current against voltage. The flash is too brief to warm the cells above 25 °C. The highest point of current times voltage on that curve is the rated power.

![Diagram of a flash lamp lighting a solar cell on a conveyor belt, with STC conditions and an I-V tracer](https://comparepv.com/static/images/articles/flash-lamp-stc-test-diagram.webp)

*Sketch of a flash-lamp test line: the cell is lit at 1000 W/m2, 25 C and AM1.5G, an I-V tracer records the curve, and cells are sorted by current. Image: [Pablo García-Linares](https://commons.wikimedia.org/wiki/File:Solar_cell_characterization_system_based_on_a_flash_lamp.jpg), [CC BY-SA 4.0](https://creativecommons.org/licenses/by-sa/4.0).*

#### Power tolerance: the rating is a bin

After the flash, modules are sorted into power classes. The power tolerance is the width of the class: how far a module sold as 450 W may sit from 450 W. The Clean Energy Council's listing terms in Australia require that the "binning tolerance shall be no more than -5W from nominal", so "-3W is accepted, but -10W is not".

Datasheets print it in two formats. Here is what each allows on a 450 W module:

| Tolerance | Power range | Can start below the label |
| --- | --- | --- |
| 0/+5 W | 450 to 455 W | No |
| ±5 W | 445 to 455 W | Yes, by up to 5 W |

The second format allows the most those terms accept. The same terms cap the uncertainty of the power measurement itself at "+/-5% for crystalline PV modules", and warn that "a common misconception of this clause is to consider this uncertainty as a sorting criteria for module power classes".

So two things sit under the label: the bin the factory chose, and the error of the tester that chose it. The tolerance line covers only the first.

### Heat and light on a sunny hour

On a roof, the irradiance is rarely 1000 W/m² and the cells are rarely 25 °C. Both pull output away from the rating.

#### How the example hours are built

Of the 14,005 modules in the catalogue, 12,850 state both a Pmax temperature coefficient, the share of power lost per degree above 25 °C, and a NOCT. For each we calculated the output at six example hours, which I chose to pull heat and light apart.

The model has two steps, both from PVEducation; the [heat yield comparison](https://comparepv.com/tools/heat-yield-comparison) runs the same one. First, the cell temperature. NOCT is the temperature cells reach under 800 W/m² of light, 20 °C air, a 1 m/s wind and an open back side. PVEducation explains that cells warm "above ambient levels, with increasing solar irradiance", so the rise above the air scales with the light.

Second, the output: scale the rating by the light as a share of 1000 W/m², then take off the coefficient for every degree above 25 °C. The exact formula and counting rules are under "Figures and denominators" below the article.

#### A worked example: a 450 W module

Take a module rated 450 W with a coefficient of -0.30 %/°C and a NOCT of 45 °C. It isn't a catalogue module, just round numbers.

On a cold clear morning, 5 °C air and 1000 W/m²:

- cell = 5 + (45 - 20) / 80 × 100 = 5 + 31.25 = 36.25 °C
- loss = (36.25 - 25) × 0.30 % = 3.4 %
- output = 450 × 0.966 = 434.7 W

Even on a cold morning, full light puts this module's cells eleven degrees over the rating condition.

In NOCT conditions, 20 °C air and 800 W/m²:

- cell = 20 + (45 - 20) / 80 × 80 = 45 °C
- light alone = 450 × 0.8 = 360 W
- loss = (45 - 25) × 0.30 % = 6 %
- output = 360 × 0.94 = 338.4 W, or 75 % of the rating

Light took 20 % of the rating, and heat took 6 % of what was left, another 4.8 % of the rating.

The same split shows in the shape of a cell's I-V curve. Less light lowers the current, a hotter cell lowers the voltage, and at NOCT both happen at once. The curves below are for a generic cell.

![Chart of two cell I-V curves: STC at 1000 W/m2 and 25 C, NOCT at 800 W/m2 and 47 C](https://comparepv.com/static/images/articles/iv-curve-stc-vs-noct.webp)

*Cell I-V curves at Standard Test Conditions and at the milder NOCT conditions, showing both current and voltage drop away from the lab rating. Image: [Squirmymcphee](https://commons.wikimedia.org/wiki/File:I-V_Curve_NOCT.PNG), [CC BY-SA 3.0](https://creativecommons.org/licenses/by-sa/3.0).*

To try other hours, set the light and the cell temperature yourself; the figure starts from a module made of catalogue medians.

**The I-V curve under real light and heat**

Move the light and the cell temperature and watch the current, the power and the maximum power point move with them.

A module's I-V curve runs from the short-circuit current Isc at zero volts to the open-circuit voltage Voc at zero amps. Power is voltage times current, so it is zero at both ends and peaks at the maximum power point, Vmp times Imp, which the inverter holds. Less light lowers the current almost in proportion: at 500 W/m² a module gives about half its STC current. Hotter cells lower the voltage, by the Voc coefficient for every degree above 25 °C, and the current barely moves. Power takes both changes, which is why a hot, bright roof and a cool, dim morning both land below the label.

Interactive on the web page.

#### Six hours across the catalogue

The table runs the same two steps for every module in the pool and gives the median. Heat loss is the shortfall caused by cells warmer than 25 °C, as a share of what the same light would give at 25 °C. Read down the output column and hold each row against 100.

| Hour | Light (W/m²) | Cell temperature (°C) | Output (% of rating) | Heat loss (%) |
| --- | --- | --- | --- | --- |
| Cold, 5 °C air | 1000 | 36 | 96.7 | 3.3 |
| Mild, 20 °C air | 1000 | 51 | 92.4 | 7.6 |
| NOCT, 20 °C air | 800 | 45 | 75.4 | 5.8 |
| Hot, 30 °C air | 1000 | 61 | 89.4 | 10.6 |
| Heatwave, 40 °C air | 1000 | 71 | 86.1 | 13.9 |
| Overcast, 20 °C air | 200 | 26 | 19.9 | 0.4 |

The median hides a spread. In the cold hour the middle eight in ten modules deliver between 96.1 % and 97.8 %; in the heatwave hour, between 83.8 % and 88.0 %.

That gives you two distances to compare. The width of one range is how much modules differ from one another. The step from the cold range to the heatwave range is how much one module differs from itself between two sunny hours. If the ranges don't touch, the weather moves output more than the choice of module does.

The overcast row has the same air as the mild row and a fifth of the light, so the cells warm a fifth as much over the air and the heat loss shrinks with it. The light itself is a fifth of the test condition, and that row assumes output falls in proportion. How well a module keeps its efficiency in weak light is a separate datasheet figure our catalogue doesn't record.

#### The manufacturer's own NOCT figure

10,193 modules in the catalogue also state a power figure measured at NOCT conditions, which gives the model something to be checked against. The median is 0.754 of the rated power, with the middle eight in ten between 0.744 and 0.764:

*Based on 10,193 modules with a stated value; the line marks the median, 0.754.*

The numbers behind the chart

| Stated power at NOCT as a share of the rating | Modules |
| --- | --- |
| \<0.73 | 100 |
| 0.73-0.74 | 319 |
| 0.74-0.75 | 2,162 |
| 0.75-0.76 | 5,332 |
| 0.76-0.77 | 2,079 |
| 0.77-0.78 | 132 |
| 0.78+ | 69 |

That figure is light and heat together, as the manufacturer measured it. Divided by what the coefficient predicts at the same point, it lands 0.1 % from the prediction for the median module, with the middle eight in ten between -1.0 % and 1.3 %. The closer those figures sit to zero, the better the simple model matches the datasheets at 800 W/m². Below that light level, nothing in the catalogue checks it.

### A whole year in three cities

Single hours don't add up to a year, so we asked PVGIS, the European Commission's yield calculator, for 1 kWp on an open rack at the best fixed angle in Warsaw, Munich and Seville. It reports each loss separately. Its "temperature and irradiance loss" covers the same two effects as the table above, from its own module model, over a year of real weather.

In these three runs, the first column grows from Warsaw to Seville:

| City | Temperature and irradiance (%) | Sun angle on the glass (%) | Spectrum (%) |
| --- | --- | --- | --- |
| Warsaw | -5.72 | -2.93 | +1.77 |
| Munich | -6.23 | -2.87 | +1.59 |
| Seville | -9.81 | -2.6 | +0.53 |

On top of these sits a flat 14 % for cables, the inverter and ageing. That's PVGIS's default, which its manual describes as "all the losses in the system, which cause the power actually delivered to the electricity grid to be lower than the power produced by the PV modules". We chose it as an input; nobody measured it.

With everything included, Seville loses a larger share than Warsaw or Munich and still produces more per kWp (specific yield), because it gets more sun. The performance ratio is the energy delivered divided by what the label alone would give for the sunlight that reached the modules:

| City | Yield (kWh/kWp) | Sunlight on modules (kWh/m²) | Performance ratio | Total loss (%) |
| --- | --- | --- | --- | --- |
| Warsaw | 1,052.59 | 1,314.12 | 0.80 | 19.9 |
| Munich | 1,140.73 | 1,433.66 | 0.80 | 20.43 |
| Seville | 1,671.76 | 2,201.32 | 0.76 | 24.06 |

Spread over the 8,760 hours of a year, Warsaw's 1,052.59 kWh/kWp averages 12 % of the rating, nights included.

### What this means when you buy

Plan a year with yields, not with the label. PVGIS and similar tools give output per kWp for your site; our [article on how many panels a household needs](https://comparepv.com/articles/how-many-solar-panels-does-a-household-need) works from those.

Compare modules on the rating anyway; it's measured the same way for all of them. In the model above, light scales every module the same way, and what separates them is heat loss, set by the coefficient and the NOCT. Our [article on panels for a hot climate](https://comparepv.com/articles/panels-for-a-hot-climate) weighs the two.

Read the tolerance line on the datasheet (linked from the panel page where we have it; our [datasheet guide](https://comparepv.com/articles/how-to-read-a-solar-panel-datasheet) shows where it sits). Positive-only means the factory didn't sort the module below its label; symmetric allows a few watts less. Between two otherwise equal modules I would take the positive-only one.

Leave air behind the modules if you can. Every figure here assumes an open back. The PVGIS manual notes that without airflow, modules "can get considerably hotter (up to 15°C at 1000W/m2 of sunlight)". At -0.30 %/°C, 15 °C more costs another 4.5 % in such an hour.

### Example: the most-offered module

The card shows JA Solar JAM54D40-470/LR, the module with the most shop offers on its ComparePV page when this was written (ties broken alphabetically). Under it are its three cheapest offers, ranked on price before VAT:

Scorecard for this module: [open its page on ComparePV](https://comparepv.com/panel/ja-solar-jam54d40-470-lr).

Cheapest current offers for [JA Solar JAM54D40-470/LR](https://comparepv.com/panel/ja-solar-jam54d40-470-lr), prices read 9 Oct 2026.

| Shop | Market | Price | Availability |
| --- | --- | --- | --- |
| [PV Zenit](https://comparepv.com/go/65264) | Poland | 335 zł incl. VAT | In stock |
| [Eco System Projekt](https://comparepv.com/go/181) | Poland | 339 zł incl. VAT | In stock |
| [Solar-EM](https://comparepv.com/go/1383) | Poland | 341 zł incl. VAT | In stock |

*Each price is the figure the shop printed, on its own side of VAT.*

### What these figures don't tell you

Every catalogue figure here is a manufacturer's claim, measured on sample modules under controlled conditions. We haven't tested any module ourselves, and the yearly figures come from PVGIS's model of real weather; no metered system went into them.

The example hours are steady states: one wind speed, no passing clouds, no thermal mass, no soiling, no inverter. They aren't a yield forecast.

The coefficient is treated as a straight line around 25 °C. The datasheets' own NOCT power can check that at 800 W/m², and nothing checks it in weak light.

The tolerance part rests on one listing body's terms and two example formats, because the catalogue doesn't record tolerance. Prices change daily; the offers shown are as of the last pull.


## 01. Figures and denominators

How the figures above were computed. Every figure is recomputed from the catalogue and the live shop offers when the page is served, and each one counts only the modules or offers that publish the value.

The catalogue figures cover every listable module, 14,005 of them, as each manufacturer's datasheet states it. A module counts toward the operating-point figures when its datasheet states a Pmax temperature coefficient between -1 and 0 %/°C and a NOCT between 35 and 55 °C; values outside those ranges are treated as data entry errors. The stated power at NOCT counts when it is between 0.65 and 0.85 of the rated power.

Each operating point is an example hour we chose, not a measured one: an air temperature and a sunlight level on the module, an open rack, the NOCT wind of 1 m/s. Cell temperature follows the NOCT formula from PVEducation, air temperature plus (NOCT - 20) / 80 times the sunlight in mW/cm². Output is the rated power scaled by sunlight over 1000 W/m² and by one plus the coefficient times the degrees above 25 °C. The heat loss is that second factor alone. This is the model the heat yield comparison tool uses.

The yearly figures are from the European Commission's PVGIS 5.3 calculator for 1 kWp of crystalline silicon modules, free-standing, at the optimal fixed tilt and orientation, with its default 14 % system loss. The links are in the sources.


| Figure | Value | Computed over |
| --- | --- | --- |
| Modules in the catalogue | 14,005 | 14,005 modules |
| Modules stating both a Pmax coefficient and a NOCT | 12,850 | 14,005 modules |
| Operating point, cold clear hour (5 °C air, 1000 W/m²): output as a share of the rating, 10th percentile | 96.1 % | 12,850 modules |
| Operating point, cold clear hour (5 °C air, 1000 W/m²): output as a share of the rating, 90th percentile | 97.8 % | 12,850 modules |
| Operating point, heatwave hour (40 °C air, 1000 W/m²): output as a share of the rating, 10th percentile | 83.8 % | 12,850 modules |
| Operating point, heatwave hour (40 °C air, 1000 W/m²): output as a share of the rating, 90th percentile | 88.0 % | 12,850 modules |
| Modules stating power at NOCT | 10,193 | 14,005 modules |
| Pmax at NOCT as a share of Pmax at STC, median | 0.754 | 10,193 modules |
| Pmax at NOCT as a share of Pmax at STC, 10th percentile | 0.744 | 10,193 modules |
| Pmax at NOCT as a share of Pmax at STC, 90th percentile | 0.764 | 10,193 modules |
| Stated power at NOCT against the coefficient's prediction, median | 0.1 % | 10,004 modules |
| Stated power at NOCT against the coefficient's prediction, 10th percentile | -1.0 % | 10,004 modules |
| Stated power at NOCT against the coefficient's prediction, 90th percentile | 1.3 % | 10,004 modules |


## 02. Related tools

[Heat yield comparison](https://comparepv.com/tools/heat-yield-comparison) [Solar payback calculator](https://comparepv.com/tools/solar-payback-calculator)


## 03. Related reading


#### How to choose panels for a hot climate

[Open](https://comparepv.com/articles/panels-for-a-hot-climate)

For a hot climate, compare the temperature coefficient before NOCT. A worked example, catalogue spreads and what a better coefficient costs in shops.


#### How many solar panels does a household need

[Open](https://comparepv.com/articles/how-many-solar-panels-does-a-household-need)

Yearly kWh, your site's yield in kWh/kWp and one module's power give the module count. Worked for Warsaw, Munich and Seville, with the catalogue spread.


## 04. Reference list

1. [PVEducation, Nominal Operating Cell Temperature](https://www.pveducation.org/pvcdrom/modules-and-arrays/nominal-operating-cell-temperature)
2. [European Commission JRC, PVGIS user manual](https://joint-research-centre.ec.europa.eu/photovoltaic-geographical-information-system-pvgis/getting-started-pvgis/pvgis-user-manual_en)
3. [PVGIS 5.3 PVcalc, Warsaw, 1 kWp, 14 % system loss, optimal angles, free-standing](https://re.jrc.ec.europa.eu/api/v5_3/PVcalc?lat=52.23&lon=21.01&peakpower=1&loss=14&optimalangles=1&mountingplace=free&outputformat=json)
4. [PVGIS 5.3 PVcalc, Munich, 1 kWp, 14 % system loss, optimal angles, free-standing](https://re.jrc.ec.europa.eu/api/v5_3/PVcalc?lat=48.14&lon=11.58&peakpower=1&loss=14&optimalangles=1&mountingplace=free&outputformat=json)
5. [PVGIS 5.3 PVcalc, Seville, 1 kWp, 14 % system loss, optimal angles, free-standing](https://re.jrc.ec.europa.eu/api/v5_3/PVcalc?lat=37.39&lon=-5.98&peakpower=1&loss=14&optimalangles=1&mountingplace=free&outputformat=json)
6. [Clean Energy Council, PV module listing terms and conditions, clauses 28 and 29](https://cleanenergycouncil.org.au/industry-programs/products-program/products-terms-and-conditions)


---

Source: https://comparepv.com/articles/stc-power-versus-real-output