String Sizing for PV Arrays: Why the Temperature Coefficient Is the Number That Actually Constrains the Design

String sizing gets treated as something close to a formality in a surprising number of solar designs: divide the inverter's maximum DC input voltage by the module's rated open-circuit voltage, round down to the nearest whole module, and consider the calculation complete. That approach is necessary but badly incomplete, because it ignores the single variable that actually determines whether a string will remain safe and within the inverter's operating window across its full range of real-world operating temperatures: the module's temperature coefficient of open-circuit voltage.

This article works through why open-circuit voltage measured at Standard Test Conditions is not the number that governs real string sizing, walks through the governing NEC calculation methodology in detail, and covers the specific places where solar designs commonly get this calculation wrong in ways that create genuine code compliance and equipment safety exposure.

Why Voc at Standard Test Conditions Isn't the Number That Matters

A photovoltaic module's rated open-circuit voltage is measured and published at Standard Test Conditions, defined as a cell temperature of 25 degrees Celsius under a specific irradiance and spectral distribution. Real, installed solar arrays operating in the field very rarely sit at exactly 25 degrees Celsius cell temperature. On a cold, clear morning (precisely the condition that tends to occur just before sunrise or during winter operation in many climates) cell temperature can actually drop well below the surrounding ambient air temperature under certain radiative cooling conditions, and critically, a module's open-circuit voltage *rises* as its temperature falls below the 25-degree reference point. This means the coldest expected operating condition at a given site, not the rated Standard Test Condition, is what actually determines the maximum string voltage a design needs to account for.

This voltage correction is calculated using the module's published temperature coefficient of Voc (typically in the range of negative 0.25 to negative 0.35 percent per degree Celsius for standard crystalline silicon modules, a figure published on every module manufacturer's datasheet) applied through the following relationship:

Voc(adjusted) = Voc(STC) × [1 + (β × (T(cell) − 25°C))]

where β represents the temperature coefficient, expressed as a negative decimal fraction per degree Celsius, so that a cell temperature colder than the 25-degree reference point produces a positive correction term that increases the calculated Voc above the STC-rated value.

What the National Electrical Code Actually Requires

NEC Article 690.7 requires this temperature correction be calculated using the lowest expected ambient temperature anticipated for the specific installation site: and critically, the code requires that calculation be based on genuine, location-specific record low temperature data for that site, rather than a generic regional assumption or a rule-of-thumb adder applied uniformly across an entire service territory regardless of local microclimate variation. The code recognizes several acceptable methodologies for determining this design temperature, but the approach that consistently produces the most defensible, genuinely location-specific answer relies on ASHRAE's published extreme minimum design temperature tables, which provide statistically derived low-temperature design values calculated for specific weather stations across a wide range of locations, rather than relying on a single coarse regional figure that may not reflect the particular site's actual climate.

The practical consequence of skipping this location-specific step is straightforward and consequential: a string sized using a generic regional cold-temperature assumption, rather than the site's actual ASHRAE extreme minimum design value, risks calculating a maximum string voltage that understates what the array will actually experience on the coldest mornings the site genuinely sees: a code violation that also carries real equipment safety implications, since exceeding an inverter's rated maximum DC input voltage is not merely a paperwork issue but a genuine risk to the inverter's input circuitry and, in a worst case, a fire safety hazard.

Where Real Designs Commonly Get This Wrong

Using a generic regional temperature adder instead of genuine site-specific ASHRAE extreme minimum design data. NEC 690.7 permits several distinct calculation methodologies, but the ASHRAE design temperature table method is specifically the one that produces a defensible, location-specific answer grounded in actual historical weather data for the site in question. A design that instead uses a flat "coldest month average" temperature, or a single adder applied uniformly across an entire state or region regardless of local elevation and microclimate variation, can meaningfully undersize the required voltage correction and put the resulting string voltage at genuine risk of exceeding the inverter's maximum rated input voltage on an unusually (but not actually unprecedented) cold morning at that specific site.

Ignoring additional string voltage rise contributed by high-albedo site conditions. Sites experiencing unusually cold, clear conditions combined with high ground reflectance from snow cover can push module cell temperatures colder than a straightforward ambient-air-temperature-based calculation would suggest, because the combination of radiative cooling under a clear night sky and reduced convective warming can drive cell temperature below what ambient conditions alone would predict. This is a growing consideration for arrays deployed in snow-country climates where the underlying temperature correction assumption embedded in a standard design workflow may have originally been developed around a milder reference climate.

Treating maximum string voltage as the only constraint requiring calculation. String sizing also carries a genuine minimum-voltage constraint on the opposite, hot end of the operating temperature range: at the site's expected maximum cell temperature (which, unlike the cold-end calculation, is driven by the module heating well above ambient air temperature under full sun exposure) the string's operating voltage under load has to remain within the inverter's maximum power point tracking window, or the inverter physically cannot track the array's maximum power point efficiently during the very peak production hours the system was designed to capture. This failure mode shows up operationally as a measurable production shortfall relative to the modeled expectation, and it is a shortfall that gets misdiagnosed with some regularity as a module underperformance or soiling issue, when the actual root cause is a string sizing error made at the design stage that never gets revisited once the system is operating, because production shortfalls of this kind rarely announce their own root cause obviously.

The Full Voltage Window a Proper String Design Has to Satisfy

Why This Connects Directly to Yield Modeling Accuracy

A string sizing error does not merely create a code compliance exposure (it directly affects the production numbers that a PVsyst or comparable yield model predicts for the system. If the as-built string configuration falls outside the inverter's optimal MPPT tracking window during any meaningful portion of the site's operating temperature range, actual measured production will diverge from modeled production in precisely the manner we described in our earlier post, PVsyst Yield Simulations: What the Numbers Actually Tell You) a genuine production shortfall whose root cause traces back to the electrical string sizing design, not to the solar resource assessment or the yield model's irradiance assumptions, even though a shortfall of this kind often gets investigated first as a resource or equipment performance question rather than a design question.

How We Approach This

Our renewables design team calculates string voltage against site-specific ASHRAE extreme minimum temperature data rather than defaulting to regional assumptions, and we explicitly check both the maximum and minimum voltage constraints against the specific selected inverter's actual MPPT operating range: not merely against its maximum input voltage rating, which addresses only one of the two boundary conditions a complete string design actually needs to satisfy. See our Energy & Renewables Engineering Services page for more on how this work is scoped as part of a full PV design engagement.

Conclusion

Open-circuit voltage measured at Standard Test Conditions is a useful reference figure, but it is not the number that determines whether a real, installed string will operate safely and effectively across the actual temperature range a specific site experiences. The NEC 690.7 requirement to calculate against genuine site-specific ASHRAE extreme minimum temperature data exists precisely because generic regional assumptions have historically produced strings that pass a superficial design check and then risk exceeding inverter input ratings on the coldest mornings a site genuinely experiences: mornings that are entirely foreseeable with proper location-specific data, not unusual outlier events the design could not reasonably have anticipated.

References

• NFPA 70 (National Electrical Code), Article 690: Solar Photovoltaic Systems
• ASHRAE Handbook: Fundamentals, Design Temperature Data
• Related reading: PVsyst Yield Simulations: What the Numbers Actually Tell You
• Related reading: Rooftop Solar Permitting: The Documents That Actually Slow Projects Down
• Pathworks services: Energy & Renewables Engineering Services