Solar String Sizing: From Voc to Wire Gauge

Quick Answer

Why This Matters for EPCs

String sizing is the calculation every other part of a design depends on. Get the module count per string wrong, and the failure isn't cosmetic: an inverter that receives more voltage than its rated maximum can trip its input protection, shut down, or in a worst case suffer lasting damage. Get the conductor sizing wrong, and the failure mode is a fire risk, not a performance shortfall.

Most content on this topic covers one half of the problem. General "wiring basics" explainers cover series-versus-parallel concepts and stop before showing the actual math. Deep technical pieces on the NEC 690.8 conductor rule rarely connect back to how the string length was chosen in the first place. An EPC actually running this calculation for a real project ends up reading three or four different sources to get from panel count to wire gauge. This piece walks the whole sequence in one place, one project, one set of numbers, start to finish.

1. Series and Parallel, Briefly

Two panels wired in series (positive terminal of one to the negative terminal of the next) add their voltages while current stays the same as a single panel. Two panels wired in parallel (positive terminals joined together, negative terminals joined together) add their currents while voltage stays the same as a single panel.

A "string" is a series-connected chain of panels. Most grid-tied commercial and residential systems using string inverters are built from multiple strings, each sized to a target voltage, connected in parallel to reach the target current the inverter (or a given MPPT input) can accept.

This part is genuinely basic and not this piece's differentiator, it's covered well elsewhere. What follows is where most content stops short.

2. Voltage Sizing: The Real Formula, Not a Screenshot

Every crystalline silicon panel carries a temperature coefficient of Voc (%/°C, always negative), the same coefficient covered in the temperature section of a real-world efficiency analysis. As cell temperature drops below the 25°C STC reference, Voc rises. As it climbs above 25°C, Voc falls.

The formula:

Voc(corrected) = N × Voc(STC) × (1 + (γ ÷ 100) × (T − 25))

Where N is the module count per string, Voc(STC) is the panel's rated open-circuit voltage, γ is the Voc temperature coefficient in %/°C, and T is the relevant design temperature (a minimum for the cold-weather maximum-voltage check, a maximum for the hot-weather minimum-voltage check).

Worked example: the cold-weather maximum-voltage check

• Panel (illustrative spec, representative of a current commercial monocrystalline module): Voc(STC) = 49.5V, Vmp(STC) = 41.8V, Isc(STC) = 13.92A, Voc temperature coefficient = -0.26%/°C.
• Inverter (illustrative commercial string inverter): Max DC input voltage 1000V, MPPT voltage window 200-850V, max input current per MPPT 30A.

A 20-module string at STC: 20 × 49.5 = 990V. Under the 1000V inverter maximum, this looks safe.

Now apply the actual minimum design temperature for a cold-climate site, using an illustrative -20°C (a real project must use the actual ASHRAE 99.6% extreme minimum design dry-bulb temperature for the specific site, not this placeholder):

Voc(corrected) = 20 × 49.5 × (1 + (-0.26/100) × (-20 - 25)) = 20 × 49.5 × 1.117 = 1,105.8V

That's 106V over the inverter's 1000V maximum, a real violation that a bare STC calculation completely misses. This is exactly the failure mode a plain "multiply Voc by module count" approach produces, and exactly why a temperature-uncorrected calculation isn't a shortcut, it's wrong.

Solving for the maximum safe string length at this site's temperature, same panel and inverter:

18 modules technically passes, with almost no margin, 0.5% is not a real safety buffer against a colder-than-design night, a slightly optimistic ambient design temperature, or manufacturing tolerance on the panel's actual Voc. 17 modules, with a 6% margin, is the more defensible design choice here. This is a genuine judgment call, not just a pass/fail number, and it's the kind of margin decision that rarely makes it into a generic wiring explainer.

The hot-weather check: does the string stay above the MPPT minimum?

The same 17-module string, checked at the opposite extreme, using an illustrative 70°C cell temperature (a realistic summer reading for a hot-climate rooftop, the same figure used in the temperature-loss worked examples for panel efficiency):

Vmp(corrected) = 17 × 41.8 × (1 + (-0.26/100) × (70 - 25)) = 17 × 41.8 × 0.883 = 627.5V

Well above the inverter's 200V MPPT minimum. In this particular design, the cold-weather maximum-voltage check is the binding constraint, not the hot-weather minimum. That won't always be true. A shorter string or a site with a milder winter but extreme summer heat can flip which check actually governs the design. Both checks are required, every time, because either one can be the limiting factor depending on the specific site and string length.

3. Current and Wire Sizing: The NEC 690.8 Sequence

Once the string voltage is settled, the current side determines wire gauge and overcurrent protection, governed by NEC Article 690.8.

The 156% rule. NEC 690.8 applies two separate 125% factors to a string's rated short-circuit current (Isc):

• First 125%, because a panel's actual short-circuit current under real irradiance can exceed its STC rating (irradiance above 1000 W/m² occurs in the real world, briefly, under certain cloud-edge and reflective conditions).
• Second 125%, because a solar array is treated as a continuous load (operating 3 or more hours at a stretch), which under standard NEC continuous-load rules requires conductors and overcurrent protection sized 125% above the calculated maximum current.

Combined: 1.25 × 1.25 = 1.56. Conductor ampacity and the overcurrent protective device must both be rated for at least 1.56 times the panel's Isc, before any further derating is applied.

Worked example, same panel (Isc = 13.92A):

Minimum conductor ampacity = 13.92 × 1.56 = 21.72A, before derating.

Then derating stacks on top. Two further adjustments commonly apply and can meaningfully reduce a conductor's usable ampacity relative to its table rating:

• Ambient temperature correction, per NEC Table 310.15(B)(1)(1), applied based on the actual ambient temperature the conductor will see.
• Rooftop conduit temperature adder, per NEC 310.15(B)(3)(c), an additional ambient-temperature adjustment required when conduit runs close to a rooftop surface, since rooftop surfaces run significantly hotter than open-air ambient temperature.

The combined effect in a genuinely hot-climate rooftop installation can be substantial, real installer-community discussion of this exact calculation shows derated ampacity dropping well below what the base 1.56 multiplier alone would suggest. The exact correction factor depends on the specific ambient design temperature, the conduit's proximity to the roof, and the current NEC edition adopted by the local jurisdiction, so treat the 21.72A floor above as the starting point, not the final answer. The real, final wire gauge for any given project has to be checked against the actual NEC temperature correction tables for that site's conditions and the locally adopted code edition, not assumed from a generic example.

For parallel strings feeding a shared MPPT or combiner, the same 156% logic applies again at the combined-current level, not just at the single-string level. Two 17-module strings in parallel, each at 13.92A Isc, combine to 27.84A into a single MPPT rated for 30A max, a 7.2% margin. The output-circuit conductor after the combination point must independently satisfy the 156% rule on the combined current, not just inherit the single-string conductor size.

4. MPPT Matching: Why Mismatched Strings Need Separate Inputs

An inverter's Maximum Power Point Tracker finds the voltage and current combination that extracts the most power from whatever is connected to it, but it does this per MPPT input, not per individual panel. Every string sharing one MPPT input is forced to operate at a single, shared operating point.

If two strings on the same MPPT have different orientation, tilt, or shading, one will be pulled away from its own true maximum power point to match the other, and total output drops below what either string could deliver independently. This is a distinct loss mechanism from panel-level shading mismatch inside a single string, it happens at the string level, between strings, not between individual panels.

The fix is straightforward where the inverter has multiple MPPT inputs: put strings with matching conditions (same roof face, same tilt, same expected shading pattern) on the same MPPT, and route strings with genuinely different conditions to a separate MPPT channel. On a single-MPPT inverter, or when MPPT channels are already allocated elsewhere, mismatched strings either need microinverters or module-level power electronics, or the mismatch loss has to be accepted as a real, quantifiable design tradeoff rather than an oversight discovered after the system underperforms.

5. Inverter Loading Ratio: Oversizing the DC Side on Purpose

A DC:AC ratio (or inverter loading ratio, ILR) above 1.0, more DC panel capacity than the inverter's rated AC output, is a deliberate, common, and usually economical design choice, not a mistake. Because panels rarely all operate at their peak rating simultaneously across an entire array, and because inverter capacity costs more per kW than panel capacity, oversizing the DC side and accepting some "clipping" loss during the rare true-peak moments is often the better economic design than matching DC and AC capacity one-to-one.

Published research on typical inverter loading ratios and the nonlinear relationship between ILR and clipping loss is covered in the real-world efficiency analysis referenced above; the same DC:AC interaction applies directly here, since the string sizing decisions in this piece determine the array's actual DC capacity feeding that ratio. A string design that changes the array's effective DC output, more modules per string, a different panel wattage, should trigger a re-check of the inverter's loading ratio, not just a voltage and current compliance check in isolation.

What EPCs Should Do Now

Common Mistakes to Avoid

• Don't size a string using STC Voc multiplied by module count without temperature correction, a string that looks safe at STC can fail once corrected for the site's real minimum temperature.
• Don't treat a string that barely clears the inverter's maximum voltage as an acceptable final design, leave real margin.
• Don't stop at the base 156% conductor calculation without checking ambient and rooftop temperature derating for the actual site.
• Don't put strings with different orientation, tilt, or shading on the same MPPT input when a separate channel is available.
• Don't treat inverter clipping as an error to eliminate entirely, a deliberately oversized DC array is often the better economic design, provided the clipping loss is modeled and accepted.

Where Reslink Fits for EPCs on String and Wire Sizing

Every check in this piece, cold-weather Voc, hot-weather Vmp against the MPPT window, the 156% conductor rule, MPPT matching by roof condition, depends on knowing the real site conditions and the real array geometry, not a generic assumption plugged into a formula after the fact.

Reslink's string and inverter configuration is auto-assigned directly from the completed 3D layout, so a string count that would exceed the inverter's cold-weather maximum voltage gets caught against the actual design, not discovered after equipment is on order. Cable lengths for the Bills of Electrical are generated from the actual string run geometry in the 3D model, not estimated, which is what the wire-sizing side of this piece depends on getting right. NEC-compliant single-line diagram generation runs from the same design, so the voltage and current values that go into a string-sizing and conductor calculation match what's actually on the bankable document set, not a separate manual recalculation.

See how string sizing, cable lengths, and SLD generation run from one 3D design instead of a separate spreadsheet → Book a demo

Frequently Asked Questions

Q1. Why does a string that measures safely under the inverter's maximum at STC still fail in the field?

Because STC Voc is measured at 25°C, and real string voltage rises as temperature drops below that reference. A string sized only on the STC number, without applying the site's actual minimum design temperature through the temperature-corrected Voc formula, can exceed the inverter's real maximum voltage on a cold, clear day, exactly when Voc peaks.

Q2. Why does the 156% rule use two separate 125% factors instead of just one?

The first 125% accounts for real irradiance sometimes exceeding the 1000 W/m² STC rating, which can push actual current above the panel's rated Isc. The second 125% is the standard NEC continuous-load factor, since a solar array runs at or near peak output for 3+ hours at a stretch. The two stack multiplicatively (1.25 × 1.25), not additively, which is why the combined figure is 156%, not 150%.

Q3. Does the 156% conductor rule apply to every part of the circuit, or just the string itself?

It applies at each level where current is calculated: the individual string (source circuit) and again at the combined output after strings are paralleled together. A conductor sized correctly for one string's current doesn't automatically cover the combined current after multiple strings converge into a shared output circuit; each stage needs its own 156%-based check.

Q4. If two strings have slightly different orientation, is it always necessary to use separate MPPT inputs?

Not always, it depends on how different the conditions actually are and how much output loss is acceptable. A minor tilt or azimuth difference may cost very little. A meaningful difference in shading pattern or orientation, sharing one MPPT, can cost real, measurable output. Where a spare MPPT channel exists, separating mismatched strings is close to a free fix; where it doesn't, the tradeoff needs to be quantified rather than assumed away.

Q5. Is a higher DC:AC ratio ever a design mistake rather than a deliberate choice?

It becomes a mistake when the resulting clipping loss wasn't modeled and simply gets discovered after the system is producing less than the DC nameplate capacity would suggest. The ratio itself, oversizing DC relative to AC, is a normal and often economical design decision; the mistake is skipping the step where that tradeoff gets checked and accepted deliberately.

Q6. Does string sizing math change for bifacial panels?

The core Voc and Isc temperature-correction formulas are the same, since they're based on the panel's front-side rated specifications. What changes is the string's real-world delivered current, which can run above the front-side Isc rating once rear-side bifacial gain is accounted for, a factor covered in the bifacial gain and DC:AC interaction sections of the real-world efficiency analysis referenced above. A string sizing exercise for a bifacial array should account for that effective current increase when checking against MPPT current limits, not just the front-side nameplate Isc.

Sources

• NFPA 70, National Electrical Code, Article 690 (Solar Photovoltaic Systems): governing standard for maximum PV system voltage (690.7) and conductor/overcurrent protection sizing (690.8); requirements cross-corroborated across multiple independent technical explainers of the current code text, not directly quoted here due to NFPA copyright on the code text itself
• NEC 310.15(B)(1)(1) and 310.15(B)(3)(c): ambient temperature correction and rooftop conduit temperature adder for conductor ampacity
• National Renewable Energy Laboratory (NREL), inverter loading ratio (ILR) benchmarking research: referenced for the DC:AC ratio and clipping-loss discussion, full citation in the real-world efficiency analysis this piece cross-references