Solar Inverter Sizing Guide: DC/AC Ratio and Clipping

Inverter sizing is usually reduced to one number, the DC/AC ratio. That number is the easy part. The ratio is the array’s size divided by the inverter’s size. Sizing by that ratio alone misses limits that matter. The ratio can look fine while the string voltage is too high on a cold morning, or while the current is too high for one input. Check each limit on its own, then compare cost and clipping. Clipping means the inverter cannot pass all the power the array makes, so some energy is lost.

Last updated: November 2026

This guide is for designers of string-inverter systems, where strings of panels feed one inverter. It also covers what changes with optimizers (small power devices on each panel), bifacial panels (which also collect light on the back), trackers (mounts that follow the sun) and batteries. The worked examples use made-up equipment unless a maker is named. They show the method. They are not approved equipment combinations.

Worked example: size a 10 kWp array

kWp is the array’s rated DC power. This example uses a made-up 555 W module and a made-up 8 kW inverter. It is the method in short. Later sections explain each step.

Made-up module: 555 W, Voc 49.5 V, Vmp 41.2 V, Imp 13.47 A, Isc 14.2 A, Voc coefficient −0.28 %/°C, Vmp coefficient −0.38 %/°C. Imp is the current at maximum power.

Made-up inverter: 8 kW AC rating, 600 V maximum input voltage, 200 to 550 V tracking range, two inputs (MPPTs, the trackers that find each string’s best operating point) with 16 A operating current and 22 A short-circuit current each, and 12 kWp maximum PV power.

Made-up site: coldest design temperature −10 °C, hottest cell temperature 70 °C.

The array is 18 modules in 2 strings of 9, one string per input.

Module voltages come from the formulas below. The cold Vmp uses the same Vmp coefficient: 41.2 × [1 + 0.0038 × 35] = 46.6796 V. The 1.25 factor on Isc is the one Fronius prints in its datasheet footnote, explained in the datasheet section. Your inverter manual and local code set the factor you must use.

Start with a complete inverter data record

Use the exact model, market version and document revision. Nearby inverter sizes can have different current ratings or inputs. The family name is not enough.

Real power and apparent power are not the same. A grid rule for reactive power can reduce the real power you get. Check the power factor and the maker’s limits.

The SAM (System Advisor Model) inverter documentation treats power limiting, voltage limits and temperature derating as separate effects. Pick the model that fits your question. Not every production calculator checks every limit.

Calculate the DC/AC ratio

The DC/AC ratio, also called the inverter loading ratio, is the array’s DC size divided by the inverter’s AC rating. Use the same unit for both. See the inverter loading ratio glossary entry for the short definition.

DC/AC ratio = array DC power ÷ inverter rated AC power

A 10 kWp array on an 8 kW inverter gives 10 ÷ 8 = 1.25. The ratio says nothing about string voltage, input current or yearly clipping. Those need their own checks.

Typical DC/AC ratios

Published figures come from large plants and from one inverter maker. Each row names its source.

The two Berkeley Lab editions use different statistics, a median in one and a capacity-weighted average in the other. Do not compare them as one trend. These sources do not split ratios by climate or by roof type.

The EIA explains why designers go above 1.0. Panels may reach their full DC output for only a few hours a year. An inverter sized for that peak may not be worth its cost.

Berkeley Lab links higher ratios to higher capacity factor measured in AC (the share of the year the plant runs at its full AC rating). Plants in the highest ratio quartile had about 2 percentage points higher capacity factor than the lowest quartile in the 2024 edition. The 2025 update puts that gap at about 1 point.

When to go lower or higher

• Lower: the inverter’s maximum PV power, cold-voltage limits or current limits can cap the ratio. The Fronius example above caps at 1.5.
• Higher: a DC-coupled battery can store power the inverter would clip. The Berkeley Lab 2025 update notes that DC-coupled plants often have a higher ratio for this reason.
• Compare, do not guess: model several AC sizes with the same array, weather, equipment and costs. A ratio below 1.3 is not a bad design by itself.

Do not assign one clipping percentage to a ratio. A shaded east-west roof and a clear single-direction array can clip differently at the same ratio. Bifacial gain, inverter temperature and export limits also change the result.

Check cold-condition open-circuit voltage

Open-circuit voltage (Voc) is the module’s voltage with no load. It is a limit check, not the normal operating voltage. It rises as the temperature falls. If your module’s datasheet gives a straight-line coefficient, a quick check is:

Voc(T) = Voc at reference temperature × [1 + betaVoc × (T − Tref)]

betaVoc is the coefficient as a decimal per degree Celsius. If the datasheet gives %/°C, divide by 100 first. Tref is the datasheet’s test reference temperature, usually 25 °C. Check it on your datasheet.

Sandia’s PV array model treats voltage and current as separate quantities that each change with temperature. This quick check leaves out irradiance effects, module tolerance and some code methods. For cold climates, see solar string sizing for extreme climates.

Round only at the end. Rounding the module voltage early can give the wrong module count. Use a documented source for the site’s minimum temperature. A general winter temperature for the region is not enough.

Check hot operating voltage and the rated-power range

Maximum-power voltage (Vmp) is the voltage at which the module makes the most power. It is not Voc. Use a Vmp temperature coefficient from the datasheet, or a module model. Do not use the Voc coefficient in its place. The two describe different points on the curve.

A quick check is Vmp(Tcell) = Vmp,ref × [1 + betaVmp × (Tcell − Tref)], where betaVmp is a decimal per °C. Cell temperature is not the same as air temperature. Mounting, sunlight and wind change it. Choose a model and a worst case that fit the site.

Take a made-up Vmp of 41.2 V, betaVmp of −0.38 %/°C, reference temperature of 25 °C and cell temperature of 70 °C. Module voltage is 41.2 × [1 − 0.0038 × 45] = 34.1548 V. Eleven modules give 375.7028 V, about 375.70 V.

Assume the inverter needs at least 200 V to keep tracking. The minimum string is ceil(200 ÷ 34.1548) = 6 modules. Assume it needs at least 300 V for rated power. The minimum is ceil(300 ÷ 34.1548) = 9 modules. These are two different questions. Neither replaces the other checks.

How to read an inverter datasheet

A real datasheet shows why the voltage ranges must stay separate. This table uses the Fronius Symo GEN24 10.0 technical page. It lists values for the 10.0 model only. Other GEN24 sizes list other values.

Starting at low voltage does not mean the inverter can deliver rated power from that voltage. Also check the upper operating voltage, the absolute open-circuit ceiling and the conditions that cause voltage losses.

Compare operating current and short-circuit current separately

Imp is the module’s current at maximum power. Isc is its short-circuit current. In a series string, every module carries the same current. At a parallel connection, the strings share the same voltage and their currents add. Sandia’s mismatch explanation says the same. Keep the wiring layout in view when you total each input.

Do not fold these four checks into one rule. Fronius’s footnote on its short-circuit rating reads: “Isc pv = Isc max ≥ Isc (STC) x 1,25 according to e.g.: IEC 60364-7-712, NEC 2020, AS/NZS 5033:2021”. That is where the 1.25 factor in this guide comes from. Your inverter manual and local code set the factor and the pass rule for your project.

Two-string current example

Take two matching strings in parallel. Each has Imp of 12 A and Isc of 13 A. Combined operating current at standard test conditions is 2 × 12 = 24 A. Combined short-circuit current is 2 × 13 = 26 A.

With the 1.25 factor, design Isc is 26 × 1.25 = 32.5 A. Take a made-up input rated 25 A operating and 40 A short-circuit. 24 A is under 25 A, and 32.5 A is under 40 A, so both checks pass.

A pass here does not cover higher temperatures, extra sunlight, bifacial rear gain or connector limits. Use the real module data and the inverter manual. Hitting a current limit also does not always damage equipment. Losing usable power is different from breaking an absolute limit.

Allocate MPPTs using the wiring layout and a model

An MPPT (maximum power point tracker) is the part of the inverter that finds each input’s best operating point. Group strings by how their voltages behave, what the manual allows and what your model shows. Different roof directions often suit separate trackers. “One direction per MPPT” is a good first question, not a rule for every inverter. For a regional example, see dual MPPT solar inverters in India.

Equal-length strings of the same module often run at similar voltages, even in different light. Shaded strings, unequal lengths and mixed modules can behave differently. Check your actual mix. Do not assume a fixed yearly mismatch loss.

The number of trackers does not equal the number of current ratings or connectors. Draw each string’s path to its input and tracker. Check shared-input limits, unequal tracker ratings and any manufacturer limits on parallel strings.

With optimizers, follow the maker’s rules for module, optimizer and inverter pairing. Plain string formulas do not validate optimizer systems. With microinverters, check the approved module pairing and branch-circuit limits. Do not invent a string length.

Use the panel stringing and wiring guide for related terms, and shadow analysis for modeled obstructions. A shade check and an electrical check answer different questions. See also common string design mistakes.

Model clipping as energy over time

Clipping happens when the array could deliver more AC power than the inverter can output. Voltage losses are a different problem. SAM models both. It describes power limiting losses, when AC output would exceed the inverter’s nameplate AC capacity, and MPPT voltage limit losses, when array voltage falls outside the tracking limits. In the second case, SAM moves the array voltage to the limit.

Snapshot comparison with made-up numbers

This table uses assumed DC input, a constant 97% conversion efficiency and an 8 kW AC limit. It only shows how power limiting works. It is not a measured day or a yearly simulation.

Each row uses AC equivalent = DC input × 0.97, delivered AC = the smaller of AC equivalent and 8 kW, and clipped = AC equivalent − delivered AC. The units are kW, not kWh. To get lost energy, add up power over time. These snapshots have no durations, so they cannot give a yearly percentage.

An hourly model can hide changes inside the hour. Test whether shorter time steps change your result, with weather data and a model that fit the question.

Compare the value of recovered energy with upgrade cost

Use clipping results from the same design assumptions for each inverter size. Weigh the recoverable energy, its value by time of day, the install cost and the running cost.

Take a simple example. An upgrade recovers 165 kWh a year. Each recovered kWh is worth €0.10. The extra install cost is €400. Annual benefit is 165 × 0.10 = €16.50 a year. Simple payback, with no discounting, is 400 ÷ 16.50 = 24.24 years.

These are made-up inputs, not a quote. The example leaves out degradation, financing, replacement cost, tax and tariff changes. If the recovered power would be exported at a lower price, use the export price, not the retail tariff.

Model bifacial arrays and trackers instead of adding a fixed correction

Bifacial modules need assumptions about rear light, mounting height, ground reflectance and electrical response. Rear gain can change current and the daily power curve. It is not a fixed percentage bonus, and it does not change the front-side label.

For trackers, set the geometry, direction, control and backtracking assumptions that fit your model. Backtracking means the rows turn to avoid shading each other. Compare AC sizes using their modeled power curves. Do not subtract a fixed amount from the DC/AC ratio just because the array tracks or uses bifacial modules.

If your software does not model rear light or trackers, say so. A basic rooftop layout tool may not give a bankable bifacial or tracking result.

Treat storage and backup as extra limits

A DC-coupled battery sits on the DC side of the inverter. It can store power that would be clipped, if the charging power, battery charge level and controls allow it. A full battery or a limited charging path stops this. Storage does not make unlimited array oversizing acceptable.

An AC-coupled battery receives power after the PV inverter. It does not generally recover power the PV inverter has already clipped. It can still change self-use and export. Describe those benefits separately from clipping recovery.

Check PV input limits, battery fit, charge and discharge power, usable energy, backup output and switching. A “hybrid” label does not prove whole-building backup or instant transfer. The battery coupling guide compares the two layouts.

Separate export limits from inverter output limits

An export limit applies at the grid connection point set by the utility. Inverter output and site export are not always equal. Loads and batteries on site can use energy before it reaches the grid.

Model the load, the battery, the inverter limit and the export control together. Do not add a clipping percentage and an export-loss percentage. They can cover the same hours and the same energy. A country’s past export rules are not the limit for every site.

Get the project’s connection offer and its allowed settings. Record the place, the effective date and any limits on zero-export or dynamic control. A qualified reviewer must confirm compliance. This guide does not give one maximum system voltage or export percentage for every country.

Issue a calculation record another designer can repeat

A sizing pack should hold the exact equipment records, the temperature basis, the string and MPPT diagram, the voltage checks, the current checks, the model settings, the loss results and the chosen layout. Match the pack to the layout and equipment schedule at the same revision.

Ask these questions before you issue it:

1. Do the module and inverter versions match what you will buy?
2. Are the absolute voltage, tracking range and rated-power range kept apart?
3. Did you use the right coefficient for Voc and for Vmp?
4. Are operating current and short-circuit current each checked against their own limit?
5. Are the connectors, parallel strings and MPPT mapping allowed?
6. Does the production model include the shading, temperature and limiting effects that matter here?
7. Are storage, export limits and grid support rules included where they apply?
8. Has the responsible engineer reviewed the design and the local rules?

The issued design should name who reviewed the equipment, the assumptions and the compliance. Software helps organize the work. It does not replace that review. If you want to see the sizing checks inside a design workflow, book a SurgePV demo.

Frequently Asked Questions

What is a good DC-to-AC ratio for solar?

Large US solar plants average about 1.3. Berkeley Lab reports a median of 1.34 for plants built in 2023 and a capacity-weighted average of about 1.32 for 2024. The EIA found 1.25 for large US plants at the end of 2016, and 1.17 for plants of 1 MW or less. Your best ratio depends on your site, your equipment and the value of the extra energy.

Can I oversize my inverter?

You can connect more DC array power than the inverter’s AC rating, up to the maker’s limit. The Fronius Symo GEN24 10.0 lists a maximum PV generator power of 15 kWpeak on a 10 kW AC nominal output, a ratio of 1.5. Voltage and current limits still apply. On sunny hours the inverter clips the extra power.

How do you calculate solar inverter size?

Divide the array’s DC power by the inverter’s AC rating to get the DC/AC ratio. Then check cold open-circuit voltage, hot operating voltage, operating current, short-circuit current and how strings map to inputs. Compare modeled energy and cost only for layouts that pass all those checks.

Is inverter clipping always bad?

Clipping loses energy in the hours it happens. But removing it can mean a bigger, more expensive inverter. Compare the value of the energy you would recover with the extra cost. Use an hour-by-hour model, because one clipping percentage does not fit every project.

Can I use the Voc temperature coefficient for Vmp?

No. Voc and Vmp are different points on the module’s curve, so they change with temperature at different rates. Use the Vmp coefficient from the datasheet, or a validated module model. If you do not have one, write that down as an open gap before you accept the hot-string voltage check.

Is an inverter’s minimum usable voltage enough for string sizing?

No. Start voltage, tracking range and the range for full power are three different things. A short string may start the inverter and be tracked, yet not deliver the power you expect. Check the maker’s manual and model the hot and cold conditions you expect.

Does DC-coupled storage eliminate clipping?

Not fully. A DC-coupled battery can store some power the inverter would otherwise clip, when it has free charging power and free capacity. How much it recovers depends on how full the battery is, the controls and the equipment ratings. It does not lift voltage, current or maximum array size limits.