Solar Inverter Underperformance: Conditions, DC Source, Conversion or Grid?

A rooftop system produces less than its yield model. An inverter’s calculated efficiency jumps from 97% to 80% and back during a single afternoon. The obvious conclusion is a faulty inverter — and very often it is the wrong one.

Comparing output against nameplate, or against a simple expectation, can mistake clouds, high module temperature, clipping or grid curtailment for a hardware fault. This guide shows how to define the measurement properly, keep the different performance quantities apart, and decide when a curve-based investigation is worth it.

The question to answer

Is lower output caused by operating conditions, the DC source, the conversion, or the grid?

Define the boundary first

Before anything is measured, write down the energy boundary: which DC inputs are included, where AC output is measured, and — for hybrid systems — where battery power enters and leaves. Most “impossible” efficiency figures come from a boundary that was never defined.

The measurement plan

1. Use appropriately rated voltage and current sensing on the DC and AC sides. Check polarity and zero offsets.
2. Align the averaging windows of AC and DC measurements — a one-second AC value divided by a one-minute DC value is meaningless during passing cloud.
3. Measure plane-of-array irradiance and a representative module temperature where you intend to normalise performance.
4. Record inverter state and curtailment through a validated interface or an external log.

Inspector’s Panel Mount Standard and Compact models add a DC power quality channel (0–630 V DC, sampled at 125 kS/s) alongside AC power quality, which suits solar, battery storage and EV charging installations within that range.

Keep the quantities separate

Three numbers are routinely confused. They have different inputs and answer different questions.

Plant yield is simply the energy delivered — and it falls on a cloudy week even when every component is perfect.

Reading the evidence

• Compare like conditions first — similar irradiance and module temperature.
• Investigate implausible efficiency at low power as a possible denominator, offset, timing or boundary problem before treating it as an inverter defect.
• Look for clipping: flat AC output at the inverter’s rating on bright days is a design choice, not a fault.
• Check curtailment: output capped by a grid or dispatch instruction will look like underperformance if the instruction isn’t logged.
• Include battery power in the balance for hybrid systems.

Acting on the evidence

Follow the evidence to the right branch:

• the measurement chain itself,
• a string inspection,
• cleaning or a shading assessment,
• an inverter service visit,
• or a grid or dispatch review.

Confirm the effect under comparable irradiance and temperature, or with an agreed normalisation method, and preserve the assumptions used by any expected-power model.

When an I-V curve helps

Sometimes monitoring identifies a recurring shortfall, but operating-point data can’t show why. A measured current-voltage (I-V) curve can help separate candidate explanations — shading, mismatch, soiling, connection problems or module condition — because each tends to change the curve’s shape differently.

Three cautions:

1. An I-V curve requires dedicated sweep hardware. Logging the DC voltage and current at the operating point is not an I-V trace.
2. Conditions must be stable and recorded. IEC 60891 describes how to correct measured curves for temperature and irradiance — but correction cannot rescue an unstable or unsuitable acquisition. A lower current may simply reflect lower irradiance.
3. Curve shapes are not uniquely diagnostic. A changed knee or a stepped shape motivates specific checks; it does not by itself name a defect.

Use the curve to guide inspection, confirm the cause physically, and repeat a controlled measurement after the action.

What to measure to show value

• valid-data coverage,
• condition-adjusted deviation from expected output,
• repeated availability losses,
• the duration of confirmed underperformance,
• for curve work, repeatability and the share of investigations with a confirmed cause.

Related reading

• What a power quality analyzer actually tells you
• Monitoring that gets a response: remote sites and fleets
• Harmonics and poor power factor