A plant is down 8% this quarter, and three causes look identical at the inverter: soiling, shading, or genuine degradation. Each asks for a different, differently-priced response — and telling them apart needs measurement at the string and panel, referenced to irradiance.
Inverter-level telemetry is a lagging, aggregate signal. It records that the plant underperformed and attributes it to nothing. By the time it shows up, the cause is weeks old and averaged across the whole array. You can see that you lost energy; you cannot see why — and ‘why’ is the only part that sets the cost of the response.
Guessing is expensive in both directions: dispatching cleaning crews to plants that needed a warranty claim, or filing warranty disputes for panels that just needed a wash. The fix is to measure below the inverter and reference everything to the conditions the array actually saw.
Per-string current and voltage on the DC side localise a problem to a string instead of smearing it across the array. An irradiance reference — a reference cell or pyranometer on the modules’ plane — plus back-of-module temperature let you normalise output against the sunlight and heat present, so comparisons are like-for-like.
A soiling ratio, from a clean reference module measured against the working array, is the direct, unambiguous signal for whether cleaning will pay for itself. Add export and consumption metering at the point of connection, and inverter telemetry over Modbus, for the whole-site reconciliation.
Dirt shows up as a soiling ratio that diverges while individual strings stay healthy. Shade shows up as a predictable daily pattern on specific strings. Degradation shows up as output below the warranted curve after you have normalised for irradiance and temperature — which is exactly the evidence a warranty claim needs before the window closes.
So the soiling ratio crossing its break-even line means dispatch cleaning, and not before. A string dropping out means localise and fix it instead of walking the whole field. Irradiance-referenced degradation beyond the warranted curve means file the claim, with the data to win it.
The sensing, the irradiance reference, the edge gateway and the analytics have to work together, and they have to keep reporting from sites with patchy connectivity. Done right, monitoring stops being a dashboard you glance at and becomes the thing that tells you which cheque to write.
We treat it as that whole system: the string and environmental sensing, the edge gateway built for intermittent links, and the analytics that turn raw telemetry into the specific action that recovers the energy or the warranty claim.
Inverter telemetry is an aggregate, after-the-fact signal: it records that output dropped but not the cause. Soiling, shading, a string fault and genuine degradation all look the same at the inverter. Separating them needs per-string measurement referenced to an irradiance sensor, below the inverter.
Soiling shows as a rising gap between a clean reference module and the array (the soiling ratio) while strings stay otherwise healthy; it recovers after cleaning or rain. Degradation shows as output below the warranted curve even after normalising for irradiance and temperature, and it does not recover.
The ratio of a soiled array's output to a clean reference module's output under the same conditions. It is the direct signal for whether cleaning will recover more energy than the visit costs, so you clean when it pays rather than on a fixed schedule.
Because output only means something relative to the sunlight and temperature the array actually experienced. A reference cell or pyranometer on the modules' plane, plus back-of-module temperature, lets you normalise output so a shortfall can be attributed to a real cause rather than to weather.
Yes. The edge gateway is built to buffer and report from sites with intermittent links, so string-level data is not lost when the uplink degrades. Reconciliation and analytics run on what the gateway reliably delivers.
It is often what makes a claim winnable. Manufacturers warrant a degradation curve; proving you are below it requires output measured and normalised to irradiance and temperature over time. Panel-level monitoring produces exactly that evidence before the warranty window closes.
String-level sensing, an irradiance reference and analytics that tell dirt from shade from degradation — and report from sites with patchy coverage. Tell us the plant and the decisions you need to make.