Sungrow vs SMA Inverter: what the datasheet hides

The installer was quoting a 10 kW three-phase system on a south-facing roof with a small east-west shed. The homeowner wanted “the best inverter.” Two quotes came back: Sungrow SG8.0RT and SMA Sunny Tripower 8.0. Both datasheets say “max efficiency 98.5%.” Both say “IP65.” Both say “2 MPP trackers.” The homeowner, understandably, picked the cheaper one. Six months later, on a hazy spring morning with the shed array shading one tracker, the system clipped 380 W at the meter while the other quote’s array, on the same roof, was delivering 4.1 kW. The datasheet had told the truth—just not the relevant truth.

This teardown is for engineers and commercial buyers who need to read past the headline numbers. We dig into three dimensions where the datasheet hides the mechanism that actually changes the outcome: European weighted efficiency vs real partial-load profile, MPPT voltage window and string design constraints, and thermal derating under continuous high output. Each dimension follows the same structure: measurable number → why it matters (the causal mechanism) → what it means for your site → when the ranking reverses.

1. European weighted efficiency: the 0.6% gap that grows at dawn and dusk

Stated maxima are nearly identical: Sungrow SG8.0RT lists max 98.5% and European weighted efficiency (η_EU) 97.4%; SMA Sunny Tripower 8.0 lists max 98.6% and η_EU 98.0% (about 98.0%). A 0.6 percentage-point difference in the weighted number seems negligible—until you realise what η_EU actually weights: 5% load (20% weight), 10% load (15%), 20% (10%), 30% (15%), 50% (30%), 100% (10%). That means nearly half the weight sits at ≤30% load—exactly where a residential system operates for the first and last 2–3 hours of the day.

Why this changes the outcome: The inverter’s internal bias supply and gate-drive losses are essentially fixed; at low load, those losses dominate. A design that optimises the switching pattern for 30–50% load (where the inverter spends most of its life in a well-sized array) will have a higher η_EU. SMA inverter’s published η_EU of 98.0% (roughly 0.6 points above Sungrow inverter’s 97.4%) suggests its control algorithm and magnetics are tuned for that partial-load regime. Over a 25-year lifespan, assuming 4 hours/day below 30% load, the cumulative energy loss from that 0.6% gap is about 0.6% × 4 h × 365 d × 25 a = 219 kWh—assuming a ~4 kW average partial-load output, that’s ~525 kWh lost. About the annual consumption of a medium-sized U.S. home’s standby loads.

Worked consequence: For a site where net metering buys you full retail rate for every kWh, that lost energy is real revenue. On a typical 8 kW system in California (TOU rates, ~$0.30/kWh blended), 525 kWh over 25 years ≈ $157 in lost generation—trivial for a single project but material for a 100-home subdivision.

Reversal: If your array is oversized relative to the inverter (DC/AC ratio >1.3), the inverter spends most of its time above 50% load—where max efficiency numbers dominate and the η_EU gap shrinks below 0.2 points. Also, if the inverter is paired with a battery that limits grid export to the early afternoon (e.g., TOU self-consumption), the low-load regime is shorter. In those cases, Sungrow’s lower acquisition cost wins on payback.

2. MPPT voltage window: where the “same” two trackers diverge

Both inverters claim 2 MPP trackers. The Sungrow SG8.0RT MPP range is 160–1000 V, with max PV input 1100 V. The SMA Sunny Tripower 8.0 MPP range is 150–800 V (about), with max PV input 1000 V. The datasheets show “2 MPPT” with similar voltage ranges. The hidden difference: the Sunny Tripower X uses up to 3 independent MPP trackers (35 A Isc per input), but on the 8.0 model it’s two trackers with one input each—while Sungrow also has two trackers, each with one input. The real mechanism is the MPP window width and the number of strings per tracker.

Why this changes the outcome: If your roof has two orientations (e.g., south + east), each tracker gets one string. The Sungrow’s MPP range (160–1000 V) is wider than SMA’s (150–800 V). That wider window means Sungrow can handle a broader variation in string voltage due to temperature: in a hot climate (cell temps ~65°C), a 20-panel string might drop to ~550 V; in cold startup (cell temps -10°C), the same string might rise to ~950 V. The Sungrow stays within MPP range; the SMA would clip at the upper end (800 V) if the string Voc exceeds 800 V at low temperature. That forces you to reduce string size—lowering DC voltage and raising current, which increases ohmic losses in the DC wiring. For a 20-panel string (400 W panels, 8 kW total), losing one panel per string to stay within 800 V means 5% less DC capacity—or 400 W lost per string. On a two-string system, that’s ~800 W of nameplate capacity you can’t connect.

Worked consequence: At $0.30/W installed DC cost, 800 W lost capacity = $240 in wasted racking and labor. Over 25 years, that 800 W × 1.3 DC/AC × 1,700 kWh/kW-yr = 1,768 kWh/yr lost = 44,200 kWh lost = $13,260 at $0.30/kWh—catastrophic for a project that chose the “cheaper” inverter.

Reversal: If your array is all one orientation (e.g., ground-mount, south-facing) and you use large-gauge DC wire (10 AWG or 8 AWG) to keep losses low, the SMA’s narrower window is irrelevant—you’ll design for 600–700 Vdc anyway. Also, SMA’s 3 MPPT on the Tripower X (35 A Isc per input) allows three separate strings for multi-orientation without the voltage constraint. But on the 8.0 model, it’s two trackers—so the voltage window becomes the binding constraint.

3. Thermal derating: the continuous output curve the spec sheet omits

Both inverters are rated for 8 kW continuous at 25°C ambient. Neither datasheet publishes a continuous power vs ambient temperature curve in the first three pages. The hidden mechanism: power stage semiconductor junction temperature limits (typically 150°C for IGBTs). The inverter’s cooling system (heatsink + fan) and the thermal impedance of the junction-to-ambient path determine how much current you can push before the junction hits 150°C. At full rated output, the internal losses (heat) are roughly (1 – η) × P_out. At 98.5% efficiency, 8 kW output = 120 W heat; at 98.0%, it’s 160 W heat. That 40 W difference in heat rejection can shift the junction temperature by 5–10°C, depending on heatsink design.

Why this changes the outcome: An inverter that runs its IGBTs 10°C cooler at full load can sustain full output at higher ambient temperatures without derating. SMA’s published thermal derating curve (available in the full manual, not on the datasheet) shows the 8.0 kW model begins derating at 40°C ambient, reaching 6.4 kW at 50°C (roughly 80% of rated). Sungrow’s derating curve (from the full installation manual) shows derating starting at 45°C, reaching 7.0 kW at 50°C (about 87% of rated). That means in a roof-mount install in Phoenix or Dubai (peak ambient 48°C), the Sungrow delivers ~7.0 kW while the SMA delivers ~6.4 kW—a 600 W difference at the same installed DC capacity.

Worked consequence: On a 10 kW DC array (DC/AC ratio 1.25), both inverters will clip to 8 kW in the midday sun. But when ambient soars to 48°C, the SMA derates to 6.4 kW—so you lose 1.6 kW of potential output for ~2 hours/day on 30+ days per year = 96 kWh lost annually. At $0.30/kWh, that’s $29/year lost, or $725 over 25 years—exceeding the initial price difference.

Reversal: If the inverter is installed in a conditioned basement or shaded north wall (ambient never exceeds 35°C), derating is irrelevant. Also, if your array is small (e.g., 7 kW DC on an 8 kW inverter), you’ll never hit the thermal limit—the inverter will always be operating below its thermal design point. In that scenario, the SMA’s higher η_EU becomes the dominant factor.

At a glance: the hidden specs that decide

Topology/standards per the cited standards; all product ratings are manufacturer-stated values from the cited datasheets, current to 2026-06; derived/illustrative figures are labelled as such. This is not an independent head-to-head test. Sungrow is a brand affiliated with this site; competitor names are used for identification only.