The MPPT loop in 30 seconds
A photovoltaic inverter's MPPT (Maximum Power Point Tracking) algorithm continuously adjusts the operating voltage of the PV array to extract maximum power. The control loop has three measurements:
- DC string voltage (V_pv)
- DC string current (I_pv)
- Calculated power (P = V × I)
If your current sensor has a 1 % offset error, your inverter's calculated power is also off by 1 % — and the MPPT algorithm seeks the wrong operating point.
Why current accuracy matters more than you think
The error compounds in three ways:
- Wrong MPP location — losing 0.3-1 % of harvest at every operating point
- Algorithm hunting — when sensors drift, the MPPT tracks an incorrect curve and oscillates around it, adding switching losses
- Gradual lifetime degradation — sensor offset drift over 20+ years can shift the MPP location enough to lose ~0.5 % yield by year 25
Worked example: 100 kW solar inverter
Assume a 100 kW PV string, 1,800 sun-hours/year, 25-year service life.
| Sensor accuracy | Annual yield loss | 25-year energy lost | At ₹4/kWh |
|---|---|---|---|
| 1.0 % (industrial-grade) | 0.7 % | 31,500 kWh | ₹1,26,000 |
| 0.5 % (MPPT-grade) | 0.35 % | 15,750 kWh | ₹63,000 |
| 0.2 % (premium) | 0.15 % | 6,750 kWh | ₹27,000 |
The premium sensor pays back ₹99,000 over 25 years. Sensor cost difference is typically <₹2,000. ROI ≈ 50×.
For a 100 MW utility-scale farm, those numbers multiply by 1,000.
Where the loss actually hides
It rarely shows up as "the inverter is broken." Instead, two patterns:
Pattern 1: Yield underperformance vs PVsyst
The plant simulation predicted 1,650 kWh/kWp/year. After 5 years, you're getting 1,580 — and you can't pin down which component is wrong. Often it's accumulated sensor drift.
Pattern 2: O&M chasing ghosts
Field teams swap MPPT trackers and inverters that are healthy, because the data points to "low harvest" when the underlying issue is sensor accuracy.
Specification recipe for MPPT sensors
For each MPPT input on a string or central inverter:
- Accuracy at nominal current: ≤ 0.5 %
- Accuracy at 10 % of rated (low-irradiance morning): ≤ 1 %
- Offset drift over lifetime: < 0.5 % of full-scale, over 25 years
- Bandwidth: ≥ 50 kHz (for fast irradiance step response)
- Operating temperature: -40°C to +85°C (outdoor cabinets get hot)
- Isolation: per IEC 62109-1 — ≥ 6 kV reinforced for 1500 V DC systems
- Residual current: separate RCMU sensor required by IEC 62109-2
The RCMU you can't skip
IEC 62109-2 mandates Residual Current Monitoring Unit for transformerless inverters. This is not the same as your MPPT current sensor — it's a separate, sensitive sensor measuring the algebraic sum of DC currents to detect ground faults.
Typical specs:
- Sensitivity: 30 mA (Type B AC+DC sensitivity)
- Response time: < 100 ms for 30 mA fault
- Fault discrimination: distinguishes leakage from genuine faults
Skipping this is one of the most common reasons new inverters fail certification.
Decision tree
Is your inverter < 5 kW residential?
└─ Use 1 % open-loop. Acceptable budget.
Is your inverter 5-100 kW commercial?
└─ Use 0.5 % closed-loop on MPPT input.
Add separate 30 mA RCMU per IEC 62109-2.
Is your inverter > 100 kW utility-scale?
└─ Use 0.2 % premium closed-loop.
The lifetime energy difference dwarfs the sensor cost.