Inverter sizing looks like a simple division – array kilowatts divided by something – and it is not. Two arrays of the same rating can legitimately need different inverters, because the numbers that decide it are the voltage at the coldest hour of the year, the current at the hottest, and what you want the system to do when the sun is at its best.
This guide follows the order we actually use when a customer sends an array layout and asks which inverter to buy. It is written for installers, EPCs and facility teams specifying commercial and industrial projects.
Why the inverter is sized last, not first
The array decides the inverter. Working the other way – picking a round inverter size and then filling it with modules – produces a design that works on paper and underperforms on the roof.
The practical consequence is ordering. Fix the module, the string length and the layout first, then run the electrical window, then choose the inverter. Sizing in that order also means the DC-to-AC ratio is a decision you make rather than one that happens to you.
The DC-to-AC ratio: the decision that shapes everything
The DC-to-AC ratio is the array rating divided by the inverter rating. A 120 kW array on a 100 kW inverter is a ratio of 1.2, and that is a deliberate choice rather than an error.
- Below about 1.1 the inverter is larger than it needs to be for most of the year. You paid for capacity that is used for a few hundred hours.
- Between 1.2 and 1.4 is the usual commercial range. Production shifts into the morning and evening, and inverter utilisation improves.
- Above about 1.5 the clipping becomes significant and has to be modelled rather than assumed. It still makes sense where module cost is low relative to inverter cost, and where the irradiance profile is broad rather than peaky.
The right ratio comes out of the production model for the specific site. It is not a rule of thumb, and it is worth asking to see the modelled clipping loss before you accept a number.
String voltage at the coldest hour, not at 25 degrees
Open-circuit voltage rises as the cell gets colder. The datasheet figure is measured at 25 degrees Celsius, and a commercial roof in January is not at 25 degrees.
The check is straightforward and it is the one that destroys equipment when it is skipped. Take the module open-circuit voltage, apply the temperature coefficient of Voc over the range from 25 degrees down to the record low for the site, multiply by the number of modules in series, and confirm the result sits below the inverter maximum DC voltage with margin.
On a 1 500 V platform the headroom is larger and the string can be longer, which is where the cable and combiner savings come from. On a 1 100 V platform the string is shorter and the balance-of-system cost is higher. Both work; only one of them is right for a given site and module.
Current per MPPT: the check that decides module compatibility
Every MPPT input has a maximum DC current, and the array has to sit inside it. The number to compare is the module short-circuit current at 1.25 times, which is how the string fuse is sized.
This is the check that stops a large-format module from dropping into an existing design. A 210 mm cell carries meaningfully more current than a 182 mm one, so moving up a module format without re-running the current window is the most common way an inverter specification turns out to be wrong.
If you are specifying both at once, check it once. If you are adding to an existing array, check it against the inverter that is already on the wall, and get the answer in writing before the order.
Clipping is a design outcome, not a fault
On any design with a ratio above one, the inverter will limit output on the best days. The production curve flattens at the top instead of peaking. That is the design working as intended, and the energy it costs is usually far less than the cost of the larger inverter that would avoid it.
The reason this matters operationally is that clipping looks like a fault in the monitoring. It is the most common reason a healthy plant gets investigated. The fix is to know the modelled clipping before handover so the flat top to the midday curve is recognised rather than escalated.
When storage is in the picture, size for the load
Once a battery is involved, the inverter rating stops being a function of the array alone. It now has to serve the array and the load, and those peak at different times.
- AC-coupled storage sits on the AC side and does not change the inverter sizing for the array. It is the usual retrofit route.
- DC-coupled or hybrid puts the battery and the array through the same unit, so the rating has to cover the combined duty.
- Backup requirement often decides the rating outright, because the inverter has to carry the backed-up circuits at once regardless of what the array is doing.
Decide the architecture before the electrical design is frozen. Changing it later is expensive in a way that sizing decisions usually are not.
Derating: what the nameplate does not tell you
An inverter rated at 100 kW will not deliver 100 kW in every condition. Two things reduce it, and both are on the datasheet if you look past the headline.
- Ambient temperature. Most units hold full output to around 40 or 45 degrees Celsius and derate above that. A unit in direct afternoon sun on a dark wall will derate on exactly the days you want full output.
- Altitude and ventilation. Above roughly 2 000 metres the cooling medium is thinner and the rating drops. Enclosed spaces need the ventilation the manual specifies, not the space that was available.
Where you put the unit is therefore part of the sizing decision, not an afterthought of the installation.
One large unit or two smaller ones
At the same total rating, one larger unit is usually cheaper per watt and takes less wall space. Two smaller units give redundancy and a finer granularity of MPPT inputs.
The deciding question is the consequence of an outage. If losing the whole array for a week costs more than the premium for the second unit, buy the redundancy. If it does not, buy the single unit and hold a spare. Where the roof has several orientations, the MPPT count usually decides it before the redundancy argument does.
Specification checklist
Reference table
Swipe horizontally to view all columns →
| Item | Confirm before ordering | What goes wrong if skipped |
|---|---|---|
| DC/AC ratio | Modelled for the site, clipping loss stated | Either wasted inverter capacity or unmodelled clipping |
| Maximum DC voltage | String Voc at the record low temperature | Equipment destruction and a void warranty |
| Current per MPPT | Module Isc at 1.25 times, per input | Incompatible module, redesign on site |
| MPPT count | Matched to roof orientations and shade groups | Mismatch loss across the whole array |
| Architecture | Grid-tied, hybrid or off-grid decided first | No backup path without replacing the unit |
| Derating | Ambient range and siting checked against the curve | Summer output shortfall |
| Grid code | Certified to the standard your utility lists | Refused interconnection |
| Monitoring | String-level data and an open interface | Faults that cannot be diagnosed remotely |
Before you place the order
Send the module model and quantity, the string layout if you have it, the site location for the temperature range, and the utility and grid code. We run the voltage and current window against your array and confirm the configuration in writing before you commit.
Equipment options are listed under inverters and controllers, including the 100 kW string inverter for single-plane roofs and the 125 kW unit where the roof has several orientations. Grid-side requirements are covered in the grid-connection guide, and module electrical data in the 600 W module page. For anything site-specific, use the contact page.
BUYER CHECKLIST
Buyer considerations for this scope
Has anyone checked string current against the MPPT input limit?
Take the module short-circuit current at the lowest expected cell temperature and apply the safety factor, then compare it with the MPPT input current limit. Skipping the temperature correction is the most common way a design passes on paper and trips in the first cold clear week.
Does the service voltage decide which inverter you can use?
It does. Split-phase 120/240 V residential service, 208 V and 277/480 V commercial service each narrow the field, and the utility interconnection position should be obtained in writing before the inverter is ordered rather than after it arrives.
One large unit or several smaller ones?
Decided by MPPT count and roof shading, service panel rating, which circuits must run on backup, and whether losing one unit should take down the whole array. Redundancy and partial shading usually argue for more than one.
PRACTICAL ANSWERS
Questions buyers ask about this
What DC-to-AC ratio should we design to?
Commonly 1.2 to 1.4 on commercial roofs. It lifts production in the morning and evening and improves inverter utilisation, at the cost of some clipping on the best days. It is a modelled trade-off for the specific site rather than a default.
Can we move to a higher wattage module with the inverter we already have?
Only if the string current still fits the MPPT input. A 210 mm cell carries more current than a 182 mm one, so the short-circuit current at 1.25 times has to be re-checked against the maximum DC current per input before you change module format.
Will the inverter clip on the best days?
On any design with a ratio above one, yes, and that is intended. The modelled clipping loss should be stated before you accept a ratio, so the flat top to the midday curve is recognised in the monitoring rather than investigated as a fault.
One large inverter or two smaller ones?
One is usually cheaper per watt; two give redundancy and more MPPT inputs. If losing the whole array for a week costs more than the premium, take the redundancy. Where the roof has several orientations, the MPPT count usually decides it first.
EXPLORE FURTHER