Yield Is Down — What Do You Check First? Sequencing Data, IR, I-V and EL, and Knowing When to Stop

ⓘ This article is currently published in Chinese. An English version is being prepared — read the Chinese original.

When a plant's output slips, most O&M teams reach for an instrument before they reach for the data they already own. That gets the order backwards and wastes field days. This article lays out a four-stage troubleshooting sequence and, more importantly, the stop-loss point of each stage — the moment the question is answered and the next instrument becomes unnecessary. Stage one is free: monitoring history, string-to-string comparison, irradiance normalisation and inverter logs will retire a large share of 'low yield' tickets without anyone walking the array, because soiling, shading drift, clipping, tracker faults and MPPT problems all leave signatures in data. Stage two is infrared, which is fast and area-wide but answers only 'where is it hot' — it finds hotspots, bypass-diode conduction and gross cell failure, and goes blind to cracks that have not yet turned resistive. Stage three is I-V, which quantifies how much power is actually missing and separates current-type from voltage-type loss, but localises nothing inside the laminate. Stage four is EL, the only method that images the module's interior and turns a vague deficit into a named, located defect with evidentiary weight for warranty claims. The article also argues the unfashionable case for *not* shooting EL — when the data stage has already closed the question, EL adds cost without adding an answer — and covers two scenarios where the sequence is deliberately inverted: commissioning acceptance and post-transport claims, where EL comes first because the goal is a baseline rather than a diagnosis. It closes on why daytime EL changes the economics of the final stage.