This guide is written for European installers, EPCs and distributors specifying hybrid inverters. Every worked value below is arithmetic you can reproduce — the method is the point, not the example module.
A PV module's open-circuit voltage (Voc) has a negative temperature coefficient. Colder cell = higher voltage. So the highest voltage your array will ever produce is on the coldest, brightest morning — not at midday in summer.
That single fact creates two opposing constraints:
Failure | When it happens | What it costs you |
Over-voltage | Coldest morning of the year, highest irradiance | Inverter shuts down on DC over-voltage; in the worst case, damaged input stage |
MPPT drop-out | Hottest afternoon of the year, cells at 60–70 °C | Array voltage falls below the inverter's MPPT floor; production collapses without any error code |
The window between them is what you are actually designing into. IEC 62548 (PV array design requirements) is the reference standard for this exercise; your inverter datasheet supplies the numbers.
The corrected open-circuit voltage at the lowest expected cell temperature is:
Voc(Tmin) = Voc(STC) × [1 + β × (Tmin − 25)]
where β is the module's Voc temperature coefficient (a negative number, expressed per °C) and Tmin is the lowest cell temperature, not the lowest air temperature.
Tmin deserves its own comment. A module mounted 10 cm above a dark roof on a clear, still winter night can sit several degrees below ambient air temperature. Using the local record-low air temperature as your design value is the safer habit.
The maximum module count per string is then:
N_max = floor( V_dc,max,inverter ÷ Voc(Tmin) )
Always round down. Half a module is not a module, and rounding up is how over-voltage events are created.
Worked example. Using a 550 Wp module with Voc(STC) = 49.5 V and β = −0.25 %/°C (illustrative values — substitute your own module datasheet), on an inverter with a 1000 V maximum DC input voltage (the DF-HVHHxxK3P-EU platform):
● At Tmin = −10 °C:Voc(−10) = 49.5 × [1 + (−0.0025 × −35)] = 49.5 × 1.0875 = 53.8 V
N_max = floor(1000 ÷ 53.8) = floor(18.6) = 18 modules
● Eighteen modules is the ceiling for that string. It is not automatically the right number — that comes from Checks 2 and 3.
Now re-run the arithmetic hot. At a cell temperature of about 65 °C, both Voc and Vmp have fallen appreciably. Two things must hold:
N × Vmp(hot) ≥ lower bound of the full-load DC voltage range
N × Vmp(STC) ≤ upper bound of the full-load DC voltage range
On the 20K model of the DF-HVHHxxK3P-EU platform, the full-load DC voltage range is 150–850 V with a 600 V rated DC input voltage and a 180 V start-up voltage. Eighteen modules at Vmp(STC) = 41.2 V give 18 × 41.2 = 741.6 V — inside the window, with roughly 108 V of headroom to the ceiling.
The start-up voltage matters more than installers expect. A string that only reaches 180 V at high cell temperature may refuse to start on a warm morning until the array is fully lit. If a site has a persistent low-voltage complaint with no fault code, a string close to the start-up threshold is the first thing to check.
Inverter datasheets give two different current figures per MPPT tracker, and they are not interchangeable:
Datasheet field | What it limits | How to use it |
Max. DC short-circuit current | The tracker's fault withstand | Isc × strings_per_MPPT ≤ this value — this is the safety limit |
Input current (rated) | Continuous operating current | Impp × strings ≤ this value — this is the normal-operation limit |
A common specification error is checking only the short-circuit figure and then running more strings than the operating current rating supports, which shows up as clipping or unexplained current limiting rather than a clean fault.
String count per tracker is a datasheet field of its own. On the DF-HVHHxxK3P-EU platform: 2 trackers at 8–12K, 3 trackers at 15–30K and 4 trackers at 40–60K, with 2 strings per tracker on the 20K model and 1 string per tracker on the smallest models. Two strings of different orientation on one tracker is a legitimate design choice — but only where the tracker is rated for it, and only when you accept that the tracker will optimise for the average of the two, not for either one individually.
With a hybrid inverter there is a second voltage window to respect — the battery side — and it interacts with your array design in ways a pure grid-tied inverter never does:
● The DF-HVHHxxK3P-EU platform accepts a 160–700 V battery voltage range, with charge and discharge current of 37 A on 8–15K models and 50+50 A on 20–60K models.
● The low-voltage single-phase platform (DF-LVHHxxK1P-EU) works on a 40–60 V battery bus with much higher currents — 135 A to 250 A depending on model — which is why cable sizing and fuse selection on LV battery systems deserve as much attention as the PV string.
● Both platforms charge lithium batteries by self-adaption to the BMS over RS485 or CAN. The BMS is the authority on the battery-side operating window; the inverter follows. If a third-party battery cannot communicate its limits, the inverter cannot respect them.
The practical consequence for string sizing: the PV side and the battery side are sized against independent windows. A string that is perfect for the array can still be useless if the battery voltage window and the array voltage window cannot both be satisfied over the site's temperature range — which is exactly why hybrid inverter selection belongs in the same conversation as string sizing, not after it.
Pull these fields off the inverter datasheet and write the actual values into the design note:
● Maximum DC input voltage (V) — the Check 1 ceiling
● Start-up voltage (V) — the warm-morning risk
● Full-load DC voltage range (V) — the Check 2 window
● Rated DC input voltage (V) — the efficiency sweet spot
● Input current per MPPT (A) and max DC short-circuit current per MPPT (A) — the two Check 3 limits
● Number of MPPT trackers and strings per tracker
● Battery voltage range (V) and max charge/discharge current (A) — the second window
● Operating temperature range and the derating threshold
For reference, both DMUDA hybrid platforms publish these fields in full. The DF-HVHHxxK3P-EU three-phase range (8–60 kW) lists IEC 61727 / IEC 62116 and EN 50549-1 under grid regulation; the DF-LVHHxxK1P-EU single-phase range (6–12 kW) lists IEC 61727 / EN 50549-1 and IEC 62116 / IEC 61727 / EN 50549-1, with IEC/EN 61000-6, IEC/EN 62109-1 and IEC/EN 62109-2 under safety and EMC on both. Both are IP65, wall-mounted, and derate above 45 °C — which is the thermal fact behind Check 2.
Can I exceed the inverter's rated DC input power with more modules? DC input power rating and DC voltage limits are separate constraints. Modest DC oversizing (a DC/AC ratio above 1.0) is normal practice for improving low-light yield, and both DMUDA platforms publish a Max. DC Input Power above their AC rating — 26,000 W against a 20,000 W AC rating on the 20K model. What is not negotiable is the voltage ceiling in Check 1.
Should I use the record-low air temperature or the record-low cell temperature? Use a cell temperature. Mounting, roof material and wind all move cell temperature away from air temperature, and on still clear nights the module runs below ambient. If you only have air-temperature records, subtract a margin and state the assumption.
Why does my array underperform in summer with no fault code? Check the MPPT floor first. As cells heat, string voltage falls; if it drops below the full-load lower bound, the inverter works outside its tracking window and output falls without an error. It is the single most common silent loss in hot-climate retrofit work.
Does a hybrid inverter change the string sizing method? The PV-side method is unchanged. What changes is that you now have a second, independent window (the battery bus) that must also be satisfied over the same temperature range, and a BMS that must be able to communicate its limits over CAN or RS485.
What happens at −25 °C? The three-phase platform is rated to −25 °C on 8–15K models and −40 °C on 20–60K models. The panel's cold Voc rises as described regardless
