PV Temperature Coefficients and Cold-Weather Voc

A PV module’s rated voltage is measured at Standard Test Conditions: 25 °C cell temperature. Cell temperature almost never sits at 25 °C in the field — it swings from well below freezing on a clear winter morning to 60–70 °C on a hot roof at midday — and voltage moves with it. That movement is not a minor correction. It decides how many modules you can safely wire in series, and it is the reason NEC 690.7 exists.

Verified temperature coefficients

Module	Cell technology	Voc tempco	Pmax tempco
Canadian Solar HiKu6 CS6R-MS-HL	Mono PERC	−0.26 %/°C	−0.34 %/°C
REC Alpha Pure-R	Mono HJT (heterojunction)	−0.24 %/°C	−0.24 %/°C
Renogy RNG-100DB-H	Monocrystalline	−0.31 %/°C	−0.42 %/°C

All three are read directly from the manufacturer’s own datasheet. The Voc coefficients cluster in a −0.24 to −0.31 %/°C range, consistent with the −0.25 to −0.35 %/°C range commonly cited across mono-Si module lines generally. REC’s heterojunction cell shows a notably flatter Pmax coefficient (−0.24 %/°C, matching its Voc coefficient exactly) than the two PERC-type modules, which is a known characteristic of HJT cell technology — its amorphous silicon passivation layer gives it better high-temperature behavior than conventional PERC.

Vmp temperature coefficients are deliberately not published here. None of the three datasheets lists a separate coefficient for voltage at maximum power point — only Voc and Pmax. Publishing an unverified Vmp figure to fill the gap would mean inventing a number no manufacturer stands behind, so this page does not do it. If you need a Vmp estimate, Pmax and Vmp coefficients are close in practice for most crystalline modules, but treat any such approximation as an estimate, not a datasheet figure.

The cold-Voc formula

As cell temperature falls, Voc rises — the opposite direction from Pmax, which falls with rising temperature and only rises modestly, if at all, as temperature drops toward 25 °C. The relationship manufacturers publish is linear around the STC reference point:

Voc(cold) = Voc(STC) × [1 + (T_cell − 25°C) × (tempco_Voc / 100)]

Since tempco_Voc is negative, a T_cell below 25 °C makes the bracketed term greater than 1, raising Voc above its STC rating.

Worked example. Suppose a module has Voc(STC) = 40 V and a manufacturer Voc temperature coefficient of −0.26 %/°C — the HiKu6 figure from the table above. On a clear, cold morning the array is at open circuit (no current flowing, so there is no I²R self-heating), and irradiance is low enough that the cells sit close to ambient temperature: −10 °C.

ΔT = −10°C − 25°C = −35°C
Fraction = 1 + (−35 × −0.0026) = 1 + 0.091 = 1.091
Voc(cold) = 40 V × 1.091 = 43.64 V

For a string of 10 such modules in series:

String Voc(cold) = 43.64 V × 10 = 436.4 V

That figure — not the STC-rated 400 V the string would show on paper (40 V × 10) — is the number that has to stay under the voltage rating of every component in the string: the charge controller’s maximum input voltage, the disconnect, the fuse, the wire insulation rating.

NEC 690.7: why the coldest morning sets the limit, not normal operation

NEC 690.7(A) requires that the maximum PV system voltage be calculated as the sum of the series-connected modules’ rated open-circuit voltage, corrected for the lowest expected ambient temperature at the site — not the temperature the system will experience most of the time. The rule’s own informational note points to the Extreme Annual Mean Minimum Design Dry Bulb Temperature from the ASHRAE Handbook — Fundamentals as a source for that design temperature.

The reason the rule is written around the coldest morning rather than typical conditions is mechanistic, not conservative-by-convention. Voc occurs at zero current, so a string sitting open-circuit before a charge controller connects, or momentarily open-circuited by a tripped breaker or a disconnected combiner, presents its full open-circuit voltage to every downstream component — and that voltage is highest exactly when the array is coldest, typically at dawn before the array has warmed under load. A system that is only checked against typical operating voltage can be safely wired for 300 sunny days a year and still oversize a string past its equipment’s voltage rating on the one cold, clear January dawn that matters.

The code gives two paths to the corrected voltage: use the manufacturer’s own temperature coefficient (the formula above, “shall be used… instead of using Table 690.7” when available), or use Table 690.7, a fixed correction factor by ambient temperature range, when a manufacturer coefficient is not available.

Table 690.7 — voltage correction factors (2014 NEC edition)

Multiply the rated open-circuit voltage by the correction factor for the lowest expected ambient temperature at the site.

Ambient temperature	Correction factor
24 to 20 °C (76 to 68 °F)	1.02
19 to 15 °C (67 to 59 °F)	1.04
14 to 10 °C (58 to 50 °F)	1.06
9 to 5 °C (49 to 41 °F)	1.08
4 to 0 °C (40 to 32 °F)	1.10
−1 to −5 °C (31 to 23 °F)	1.12
−6 to −10 °C (22 to 14 °F)	1.14
−11 to −15 °C (13 to 5 °F)	1.16
−16 to −20 °C (4 to −4 °F)	1.18
−21 to −25 °C (−5 to −13 °F)	1.20
−26 to −30 °C (−14 to −22 °F)	1.21
−31 to −35 °C (−23 to −31 °F)	1.23
−36 to −40 °C (−32 to −40 °F)	1.25

Below −40 °C, or for modules that are not crystalline or multicrystalline silicon (thin-film, for instance), the code requires using the manufacturer’s own instructions rather than this table.

Comparing the two methods on the same example. Using the illustrative 40 V module at −10 °C ambient again, but this time with the Table 690.7 factor for the −6 to −10 °C row (1.14) instead of the manufacturer coefficient:

Voc(cold), Table 690.7 method = 40 V × 1.14 = 45.6 V
Voc(cold), manufacturer-coefficient method = 43.64 V (calculated above)

The code table gives a higher, more conservative figure — 45.6 V versus 43.64 V, a 1.96 V difference per module that becomes nearly 20 V across a 10-module string. That gap is exactly why the code prefers the manufacturer’s own coefficient when it is available: the generic table has to cover every crystalline module on the market, so it errs conservative, while the manufacturer’s number describes the actual part being installed.

The hot-weather consequence: Pmax at real cell temperatures

The same physical relationship works in reverse for power. A module at 65 °C cell temperature — a realistic figure for a dark-framed module on a low-airflow mount in full summer sun, well above the 25 °C STC rating — loses output according to its Pmax coefficient:

Module	Pmax tempco	Power loss at 65 °C cell temp (ΔT = 40 °C)
Canadian Solar HiKu6	−0.34 %/°C	13.6%
REC Alpha Pure-R	−0.24 %/°C	9.6%
Renogy RNG-100DB-H	−0.42 %/°C	16.8%

Power loss % = ΔT × |tempco_Pmax|
             = (65°C − 25°C) × 0.34 %/°C = 13.6%  (HiKu6 example)

A 400 W-rated HiKu6 module at 65 °C cell temperature is producing roughly 346 W from temperature alone — before accounting for anything else in the loss chain covered on our PV system losses page. This is a real, physical, unavoidable derate, not a defect: it is one reason a system’s actual peak summer output typically runs below its cumulative nameplate rating even under clear skies, and one reason for leaving airflow behind and around modules where the mounting method allows it.

Before you rely on this

The Voc formula and the three module coefficients are read directly from manufacturer sources and cross-checked against two independent explanations of the same method. Table 690.7 is transcribed from a 2014-edition NEC mirror; the current 2023 NFPA edition is paywalled and could not be checked, and code numbering and content occasionally shift between editions. Confirm against the edition your local authority having jurisdiction has adopted before using this for a permitted, inspected installation — and always prefer a manufacturer’s own published coefficient over the table when your module’s datasheet provides one.