Inverter Clipping: How to Tell If It's Actually Costing You Anything

You’ve bolted 5.4 kWp of panels onto a 3.6 kW inverter and now your monitoring app shows a generation curve with a lawn-mowered top. Cue the forum posts saying you’ve been sold a dud. Before you spend anything, work out what the flat top is actually costing. For most UK homes the honest answer is: a small number of kWh, worth a small number of pounds.

This post shows you how to spot solar inverter clipping losses in your own data, put a figure on them over a year, and compare that figure with the price of fixing it.

What clipping is, in one paragraph

An inverter converts DC from your panels into AC for your house. It has a maximum AC output, say 3,600 W. If the panels could supply 4,200 W at that moment, the inverter simply refuses the extra 600 W. Output sits flat at 3,600 W until the sun drops or a cloud arrives. That refused energy is clipped. It isn’t damaged or stored anywhere. It’s just never harvested.

Oversizing the array relative to the inverter (a DC/AC ratio above 1) is normal and often deliberate. In the UK, 1.3 to 1.5 is common, and installers do it because panels almost never hit their rated output here.

Step 1: Look for the flat top

Open the day view in whatever you use: the SolarEdge or Enphase Enlighten app, GivEnergy Cloud, Solis Cloud, the Huawei FusionSolar portal, Octopus’s smart meter portal, or Home Assistant with the Energy dashboard. Pick a clear, cold, bright day in April or May. That’s your best candidate, because cool panels run more efficiently and the sun is high but not yet baking them.

A clipped day looks like this in a 5-minute export:

time,ac_power_w
10:30,2890
10:35,3120
10:40,3388
10:45,3600
10:50,3600
10:55,3600
11:00,3599
...
13:25,3600
13:30,3600
13:35,3548
13:40,3410

The giveaway is a run of near-identical values at the same ceiling, across several days, on the clear ones. A single cloud-edge spike isn’t clipping. Real clipping repeats at the same number, give or take a few watts.

Two traps:

• Battery systems. If you have a DC-coupled battery (GivEnergy hybrid, Solis RAI, Tesla Powerwall 3), the battery can soak up what the inverter would otherwise clip. Look at PV input power, not just AC export, or the flat top may disappear and you’ll miss it.
• Export limits. A G98 or G99 export cap, or a 3.68 kW limit set in the inverter, produces the same flat top for a different reason. Check the settings page before blaming the hardware.

Step 2: Find your ceiling and count the clipped intervals

You need interval data. Most dashboards let you export CSV; Home Assistant stores it in its database; SolarEdge’s API gives 15-minute data. Then ask an AI tool to do the tedious part. Paste a few hundred rows into ChatGPT or Claude (or upload the CSV) with a prompt like:

This is a year of 15-minute AC output in watts from a 3.6 kW inverter. Find the clipping ceiling (the most common maximum value). Flag every interval within 2% of it. Then estimate the energy lost by fitting a smooth bell curve to each clipped day and integrating the area above the ceiling. Return a monthly table in kWh.

If you prefer a spreadsheet, this works fine in Google Sheets or Excel:

1. Column A: timestamp. Column B: power in W.
2. Ceiling cell: =MODE(B:B) restricted to values above 3,000, or just =MAX(B:B) if you trust it.
3. Clipped flag: =IF(B2>=0.98*$E$1,1,0).
4. Count clipped intervals per day with a pivot table.

Be sceptical of any figure that comes out suspiciously large. AI tools will happily invent a lost-energy number if you don’t make them show the working. Ask for the fitted curve parameters for three sample days and check them against the raw rows.

Step 3: Quantify the lost kWh

You can’t measure the energy that never happened, so you reconstruct it. Three methods, in order of effort:

1. Rectangle estimate. Take a clipped day. Assume the unclipped curve would have peaked about 10 to 15% above the ceiling for the flat section’s length. Crude, but fine for a first pass.
2. Parabola fit. Fit a curve to the shoulders either side of the flat top and integrate the part above the ceiling. This is what the AI prompt above does.
3. Model comparison. Run your array through PVGIS (the EU’s free tool, which still covers the UK) or PVWatts. Set the system loss to about 14% and compare the modelled hourly output with the real thing. The gap at peak hours is your clipping.

A worked example

A 5.4 kWp south-facing array in the Midlands (12 × 450 W panels, 35° tilt), with a 3.6 kW inverter. DC/AC ratio: 1.5.

Total clipped intervals found across the year: 1,180 fifteen-minute slots, roughly 295 hours. Average shortfall in those slots, from the curve fit: 420 W.

295 h × 0.42 kW ≈ 124 kWh/year clipped

Annual generation measured: 4,900 kWh. So clipping is about 2.5% of output.

Monthly breakdown from the fit:

Now price it. At 0.15 per kWh export (Octopus Outgoing Fixed is around that, and SEG rates vary between roughly 4p and 15p), those 124 kWh are worth about £19 a year. If you’d have self-consumed them at a 26p import rate, the ceiling is £32. Call it £20 to £30 a year, the whole prize.

Step 4: Price the fix

What would stop the clipping? A bigger inverter.

Take the 5 kW swap at £1,100 against a £25 annual recovery: 44 years. The inverter warranty runs 10 to 12 years. A bigger unit also has a slightly lower efficiency at low output, which in the UK winter, where you spend most of your days at 5 to 15% load, eats a few kWh back. The numbers don’t close.

The rule of thumb that falls out of this: if clipping is under about 5% of annual generation, an upgrade is almost never economic. At 2.5%, you’re nowhere near it.

When it would be worth caring

Be honest about the exceptions, because they exist:

• DC/AC ratio above about 1.8. Clipping then climbs towards 8 to 10% of the year. Still rarely worth a swap, but worth checking whether a cheaper micro-fix exists, like re-stringing.
• West or east split arrays. These spread their output and clip less, so if yours clips hard something else is off.
• Inverter that’s derating, not clipping. A unit that cuts power because it’s too hot (look for the output sagging when it’s warm, rather than sitting flat) is a ventilation problem, not a sizing one. Moving it out of a sun-trapped loft hatch or garage corner costs nothing.
• Falling output on the plateau. If the “flat” top is slowly sliding downwards through the day, or sits lower than it did last year, that’s a fault signature rather than clipping. This is the kind of thing a proper alerting setup catches, and the monitoring and anomaly detection guide covers how to set the thresholds.

Set it up once, then leave it alone

Create a monthly check, not a daily worry. In Home Assistant, a template sensor that counts intervals above 98% of your ceiling does the job. In a spreadsheet, refresh the pivot table once a month. Ask your AI tool for one line: “clipped kWh this month vs. last year’s same month.”

If the annual figure stays below 3% and the pound value stays in double digits, close the tab. The flat top on your graph looks alarming, but it is the cheapest loss on your system to live with.