“It’s 98.5% efficient, so a noisy generator feed won’t matter” –

If you’ve ever sized a string inverter for a site that relies on a diesel or LPG generator as a primary or backup feed — think off-grid cabins with weekly genny runs, telecom shelters with poor fuel quality, or agri-pump controllers tied to a genset — you’ve heard some version of the statement above. It feels physically intuitive: an inverter is just a power converter, and as long as the AC voltage and frequency stay within the window, the MPPT will find the DC power, the IGBTs will switch, and efficiency will be close to the nameplate. The myth is that generator-induced distortion (voltage harmonics, frequency wobble, phase imbalance) is a second-order effect that a modern inverter can absorb with no meaningful consequence.

Why the myth survives — and the one number that changes everything

Both Growatt and Sungrow inverters are certified to UL 1741 / IEEE 1547, which means they must disconnect from a distorted grid if voltage or frequency drifts too far. But inside that envelope, the inverter’s control loops — especially the MPPT algorithm and the current regulator — interact with generator impedance (typically 3–8% subtransient reactance, much higher than a utility transformer) and voltage distortion. The consequence is a decision threshold: above a certain generator THD_v, the inverter’s own output current harmonics increase, the generator voltage becomes more distorted, and a positive-feedback loop can trigger overcurrent protection or MPPT oscillation that reduces capture by 10–25% — even though nominal efficiency remains 98%+.

Dimension 1: Voltage harmonic withstand – the real de‑rating threshold

The Sungrow SG8.0RT datasheet quotes 98.5% max efficiency and 2 MPPTs, with MPP range 160–1000 V and max PV input 1100 V. The Growatt MIN 8000–10000TL-XH (in the US lineup) has a similar peak efficiency of ~98.4% and dual MPPT. On a stiff utility grid (THD_v

But connect either to a typical 10–20 kW diesel generator with a THD_v of 6–10% at no-load, and the dynamic changes. The inverter’s grid-tied current controller tries to synthesise a sinusoidal current waveform that is in-phase with the voltage fundamental. When the voltage waveform is flat-topped (characteristic of a saturated generator core), the inverter’s current regulator can saturate its voltage reference, causing the actual current to have a ~15–25% third harmonic content. That harmonic current flows through the generator’s stator windings, causing additional I²R losses, and the generator’s AVR sees a distorted voltage — the loop gain changes, and the voltage THD can jump from 6% to 12% within a few seconds.

Worked consequence: At a site with a 12 kW generator and a Growatt MIN 10000TL-XH running at 5 kW DC (50% load), if the generator THD_v at the inverter terminals exceeds ~8%, the inverter’s internal over‑current protection (or anti‑islanding algorithm, which monitors dV/dt) can reduce power output by repeatedly disconnecting and reconnecting — a phenomenon installers call “genny hunting.” On a 10‑hour generator run, this can lose 1–2 kWh of capture — easily 5–10% of daily yield — even though the inverter nameplate says 98.4% efficient. The Sungrow SG8.0RT is not immune: its datasheet does not list a THD_v tolerance for generator feed, but UL 1741 testing uses a stiff grid with

Dimension 2: MPPT stability under DC‑link ripple caused by generator harmonics

Both inverters use a DC‑link capacitor bank to smooth the rectified AC ripple. On a stiff grid, the DC‑link voltage ripple is typically

Growatt’s MOD series claims “up to ~99.9% MPPT tracking efficiency”, but that figure is measured under static irradiance and a clean grid — not under generator harmonic distortion. The Sungrow SG‑RT line uses a standard dual‑MPPT topology with no published MPPT ripple rejection spec. The practical effect: under distorted generator voltage, both inverters can experience a 3–8% reduction in MPPT tracking efficiency, meaning the array never delivers its full DC power to the inverter. The effect is worse at low irradiance (morning/evening) because the MPPT step size is smaller relative to the ripple.

Worked consequence: On a 6 kW array (about 20 panels) with a Growatt MIN 8000TL‑XH, if the generator THD_v is 9% at the inverter terminals, the MPPT can oscillate with a period of ~3–5 seconds, reducing average DC power capture by roughly 4% — equivalent to losing 240 W of potential output for that hour. Over a 200‑hour annual generator run (common for a telecom shelter with weekly genny tests), that’s about 48 kWh lost per year — not massive, but real, and not accounted for in the efficiency spec. The Sungrow unit would show similar losses under the same conditions because neither brand has published a “generator‑tolerant” MPPT algorithm.

Dimension 3: Harmonic current injection back onto the generator – thermal stress on windings

This dimension is seldom discussed because datasheets only show output current THD (which is v (mostly 3rd and 5th), the inverter’s input current can have a THD of 15–30%. That harmonic current flows into the generator’s stator windings, causing additional heating proportional to the square of the RMS harmonic current. A 20% current THD at 70% load means the stator copper losses can increase by about 15–20% beyond the fundamental loss.

Worked consequence: For a 10 kW generator feeding a Growatt MIN 10000TL‑XH at 7 kW load (70% of inverter rating), if the input current THD is 20%, the additional stator I²R loss is about 150–200 W — which might seem small, but on a generator that already runs at 85–90°C ambient in a shelter, a 20°C rise in winding temperature reduces insulation life by roughly half for each 10°C increase (Arrhenius rule). The same thermal stress applies to the Sungrow inverter’s input stage, though neither manufacturer publishes a generator‑specific input THD spec.

Dimension 4: Control‑loop instability under rapid frequency variation (genset droop)

Generators under light load often have a slight frequency droop (e.g., 61 Hz at no load down to 59 Hz at full load). IEEE 1547 requires inverters to operate within 59.3–60.5 Hz (for 60 Hz systems). The inverter’s phase‑locked loop (PLL) must track frequency variations. On a generator with poor governor performance, the frequency can change by 0.5–1 Hz within a few seconds. Both Growatt and Sungrow inverters use standard PLL architectures that are designed for slow utility frequency variation (0.01–0.1 Hz/s). A 1 Hz/s ramp can cause the PLL to lose lock momentarily, forcing the inverter to disconnect and reconnect after a 5‑minute wait (UL 1741 requirement). The result is a “cycling” behaviour: the inverter runs for 2–3 minutes, then disconnects for 5 minutes, reducing effective capture to 40–50% of the nameplate.

Worked consequence: On a site with a 15 kW generator that has mechanical governor (droop set to 3%), the frequency can vary by ±0.5 Hz under steady load, and by ±1 Hz during load transients (e.g., a refrigerator compressor starting). If the inverter’s PLL has a fast loop bandwidth (typical for high‑efficiency inverters), it can overshoot and lose lock. Neither brand’s datasheet specifies PLL bandwidth, but field reports from off‑grid installations suggest that both brands are susceptible when generator frequency variation exceeds 0.8 Hz/s.

The decision threshold – a rule you can execute

Failure mode: what happens if you ignore the threshold

If you install a Growatt or Sungrow inverter on a generator with THD_v >10% and undersized (generator kVA = inverter kVA), the most common outcome is not a catastrophic failure but repeated nuisance tripping — the inverter disconnects multiple times per hour, the generator runs unloaded (which can cause wet‑stacking on diesel units), and the battery (if present) cycles deeper than intended. After a few months, the inverter’s DC‑link capacitors can degrade faster due to the high ripple current, reducing lifespan from ~15 years to maybe 8–10 years. The same applies to both brands because the capacitor bank is sized based on grid‑tied ripple, not generator‑sourced ripple.

Bottom line: when the myth costs you real yield

The myth that a 98.5% efficient inverter is immune to generator feed quality is dangerous only because it leads to under‑spec’d balance‑of‑system. Both Growatt and Sungrow inverters are excellent grid‑tied machines. On a noisy generator feed, they are equally vulnerable — the decision threshold is generator THD_v and governor quality, not brand. Spend your budget on measuring the generator’s actual power quality at the inverter terminals, and add a line reactor if needed. That $200 part will do more for uptime than any brand swap.