String Sizing Verification Technique

Why this matters

A string that was sized correctly on paper does not stay correct just because the permit set says so. A crew short a module on the truck adds one from a different pallet, a wiring error strings two partial runs into one long series loop, or a design gets copied from a similar job without rechecking this site's coldest expected morning. Any of those turns a compliant string into one that pushes open-circuit voltage past what the inverter, and the code, allow it to see. The consequence is not cosmetic: exceeding the inverter's rated input voltage risks damaging it and can void its listing outright, and it is a fault that will not show up on a mild afternoon, only on the cold, clear, high-voltage morning the design was supposed to be checked against in the first place. This technique verifies what got built, not what got drawn.

Confirm what is actually in the string before you calculate anything

Count modules in the string as wired, not as designed, and confirm every one is the same model shown on the plan set. A substituted module with a different Voc or a different temperature coefficient invalidates any calculation built on the original datasheet, and the substitution is often invisible from the ground, a different sticker on the back of a module that otherwise looks identical from the front. Trace the series path physically, positive lead to negative lead, module to module, rather than assuming the string count matches the design because the module count on the roof matches the pallet count delivered.

Pull the real inputs from this module's datasheet, not a remembered number

Get open-circuit voltage and the temperature coefficient of open-circuit voltage from the datasheet for the model actually installed. Crystalline silicon modules commonly publish a Voc coefficient in the range of roughly a quarter to a third of a percent per degree Celsius, negative, meaning voltage rises as the cell gets colder, but the exact figure is model-specific and a different cell technology can sit outside that range entirely. Using a coefficient from a different module, even a similar-looking one from the same manufacturer's other product line, produces a corrected voltage that looks exactly as confident as a right one while being wrong.

While tracing the series path, confirm polarity at every connector rather than assuming a matched MC4 pair guarantees correct wiring; a connector can physically mate positive to negative between two strings from different sub-arrays if the runs were pulled and landed by different people on different days. A string with one module wired backward does not read as an oversized voltage, it reads as a voltage short by roughly twice that module's contribution, the opposite direction from the failure this technique exists to catch, but it is caught by the same physical trace and is worth confirming here rather than discovering it only once the sizing check comes back mysteriously low.

Get the site's actual design-low temperature, not the coldest day anyone remembers

Pull the lowest expected ambient temperature from the design data source the jurisdiction uses for this calculation, typically an ASHRAE extreme minimum design dry-bulb figure for the site, rather than a number a tech recalls from a hard freeze two winters ago. This is the temperature the maximum-voltage check is built around, and it is a published design value, not a personal memory of the coldest morning worked.

Check the maximum-voltage ceiling at the cold extreme

Correct the module's nameplate Voc to the site's design-low temperature using its own coefficient, then multiply by the module count to get the string's corrected maximum voltage. Compare that figure against two numbers and take whichever is lower: the inverter's rated maximum input voltage, and the code ceiling for PV DC circuits on the building type, commonly 600 volts for a one- or two-family dwelling in many adopted editions, higher for other occupancies and installation types under the code your AHJ has adopted. A string that clears the inverter's number but not the code's, or the reverse, still fails; both have to hold, and a string that fails either one is not re-checked, it is redesigned.

This check uses the coldest expected temperature because Voc rises as cells get colder, so the worst case for this ceiling is the coldest morning the system will actually see, not a mild afternoon and not the hottest day of the year.

Check the MPPT floor at the opposite extreme, for the opposite reason

Separately, confirm the string's operating voltage does not fall below the inverter's minimum MPPT tracking voltage on a hot day, using the module's temperature coefficient of Vmp rather than Voc, and using cell temperature rather than ambient air temperature as the input. A module in full sun runs well above ambient, commonly twenty to twenty-five degrees Celsius above ambient on a well-vented rack mount and more on a flush, low-profile roof mount with poor rear ventilation, and it is the hot cell that drags Vmp down toward the inverter's floor, not the hot air around it. This is the mirror image of the cold-voltage check: cold weather threatens the ceiling, hot cell temperature threatens the floor, and a string sized only against one of the two extremes has only been half verified. A string that lands under the floor on the hot check needs more modules in series, or the array split across more strings on a separate MPPT input if the count cannot grow, again a design change rather than a field workaround at the connector.

Confirm current against the inverter's input rating

Check the string's short-circuit current against the inverter's maximum input current per MPPT input, and where multiple strings land on one MPPT input in parallel, confirm every paralleled string matches in Voc and Imp class; a mismatched string dragged into a parallel pair does not average out, it drags the whole input toward the weaker string's operating point. Separate a mismatched string onto its own MPPT input where the inverter has one free, or replace the outlier module so the pair matches, rather than leaving two unlike strings paralleled together and calling the input compliant because the total current happens to fit. Where a string reads voltage low against its design figure for reasons other than temperature, that is a wiring or module fault, not a sizing question, and belongs to the string-voltage troubleshooting path rather than this check.

Measure it, with the hazard treated in the same clause as the meter

Take the field Voc reading at the combiner or inverter DC input with the string open circuit, using a meter and leads rated at or above the corrected maximum voltage just calculated and gloves rated for that voltage class, because the string is live in any daylight and this measurement is taken on exactly the conductors the whole check is about. Follow the full boundary, PPE and proving practice in the energized DC array safety standard for this measurement; this step names the minimum in-clause action, not the complete practice. Record back-of-module temperature at the same moment, correct the reading to standard test conditions with the same coefficient used above, and compare the corrected reading against nameplate Voc times module count. A match confirms the wiring is what the design assumed; a mismatch by roughly one module's worth of voltage points at a missing, reversed, or bypassed module before it points at anything else.

A worked pass, two strings against the same gate

Same module both times: nameplate Voc 49.5 volts, temperature coefficient of Voc 0.27 percent per degree Celsius, inverter and code ceiling both set at 600 volts for this installation.

String A, ten modules, site design-low ambient of 2 degrees Celsius. The correction spans 25 minus 2, or 23 degrees Celsius below the 25-degree rating condition, so voltage rises 0.27 times 23, or 6.21 percent. Corrected Voc per module: 49.5 times 1.0621, or 52.57 volts. String total: 52.57 times 10, or 525.7 volts, against the 600-volt ceiling. Passes, with a margin of about 74 volts, roughly 12 percent.

String B, fourteen modules, a colder site with a design-low ambient of negative 18 degrees Celsius. The correction spans 25 minus negative 18, or 43 degrees Celsius, so voltage rises 0.27 times 43, or 11.61 percent. Corrected Voc per module: 49.5 times 1.1161, or 55.25 volts. String total: 55.25 times 14, or 773.5 volts, against the same 600-volt ceiling. Fails by 173.5 volts, close to 29 percent over.

String B's failure is too wide for a field fix at the connector. Ten modules at this coefficient and this site's design-low temperature clears the ceiling with a margin similar to String A; fourteen does not, and there is no embedment or torque check to redo here, only a design change: fewer modules in series, more strings, back through the permit set the AHJ approved, not a call made on the roof.

References

• See related: the energized DC array safety standard for the boundary, PPE and proving practice this measurement is taken under.
• See related: the string voltage low after a module replacement troubleshooting path for a mismatch found by wiring or fault rather than by sizing.
• NEC Article 690, including 690.7 for maximum PV DC circuit voltage and its low-temperature correction, in the edition your AHJ has adopted.
• Module datasheet for Voc, Vmp, and their temperature coefficients; inverter installation manual for rated maximum input voltage, MPPT window and maximum input current.