Solar Production Normal in Spring But Drops Only in Midsummer: Clipping vs Heat Decision Tree

Why this matters

A homeowner who watches their monitoring app sees April and May production hit the modeled peak, then July and August daily kWh sag below spring numbers despite longer days and clearer skies. The instinct is to call a failure, but two normal physics behaviors mimic a fault: inverter power clipping when the DC array oversizes the AC rating, and module temperature derate that bleeds voltage as cell temperature climbs above the 25 C STC rating. Misdiagnosing this leads to needless inverter swaps or warranty claims that get denied because the array is performing exactly as designed. The split matters because clipping is a design tradeoff that costs only a few percent annually, while excessive heat losses can point to a real airflow or racking problem. Knowing which one you are looking at tells you whether to do nothing, advise the customer, or actually fix something.

Symptom presentation

The classic pattern is a daily production curve that goes flat-topped at the inverter AC limit during peak sun hours in summer, then recovers normal arch shape in spring and fall. In monitoring you see midday power pinned at exactly the inverter nameplate AC watts for two to four hours. Heat-driven loss looks different: the whole curve scales down a few percent with no flat top, and the loss tracks ambient temperature. Both can coexist. The customer reports "less energy on the hottest, sunniest days," which is counterintuitive and drives the call.

Quick checks

Pull the inverter spec sheet and find the rated AC output watts. Compare to the DC array size (module count times module STC watts). A DC-to-AC ratio above about 1.2 invites clipping. Look at the live power graph at solar noon on a clear July day. If power sits dead flat at the AC nameplate, that is clipping. Check the module datasheet temperature coefficient of Pmax, typically near -0.34 to -0.40 percent per degree C. On a 65 C cell day (about 40 degrees above STC) that is a 14 to 16 percent power loss, fully expected.

Isolation tree

Start at the midday power line. Is it flat at the inverter AC rating? Yes branch: this is clipping. Confirm the DC-to-AC ratio is over 1.15 and stop. No flat top, just a lower arch? Go to temperature. Pull cell temperature from monitoring or estimate it as ambient plus 25 to 30 C in full sun on a roof mount. Multiply the delta over 25 C by the Pmax coefficient. Does the calculated loss match the observed shortfall versus the spring baseline at equal irradiance? Yes branch: normal heat derate, no defect. If the observed loss exceeds the calculated heat derate by more than a few percent, branch to airflow and soiling: check standoff height (a flush roof mount runs hotter than a raised rack), confirm no debris dam under the array trapping heat, and inspect for module hot spots with a clamp meter comparing string currents. A single shaded or failing module drags a whole series string and shows up worse under high irradiance.

Confirming diagnosis

For clipping, overlay a clear spring day and a clear summer day at the same irradiance. The summer curve will show the same instantaneous power until it hits the ceiling, then flatten while spring keeps rising. The clipped energy is the area above the flat line. This is by design and recoverable only by adding inverter capacity, which rarely pencils out. For heat derate, log module backsheet temperature with a contact thermocouple or IR thermometer at peak sun. Plug the measured temperature into the coefficient. If string voltage has dropped proportionally and current is unchanged, the loss is pure thermal. Confirm string voltages with a meter at the combiner or inverter MPPT inputs and compare strings against each other; matched strings drooping together is thermal, one string low is a module or connection fault.

Remediation

Clipping needs no repair. Document the DC-to-AC ratio and show the customer the annual clipping loss is typically 1 to 3 percent, a deliberate tradeoff that boosts morning and evening harvest. If they want it back, the only fix is a larger inverter, which usually does not justify the cost. Heat derate within spec needs no repair either, just an explanation. If measured losses exceed calculated derate, address the real cause: raise module standoff height to improve back-of-module airflow per the racking manufacturer spec, clear any debris dam, clean soiled modules, and chase any single low string for a failed bypass diode, cracked cell, or loose MC4 connector showing voltage drop under load.

References

• IEC 61853-1 Photovoltaic module performance testing and energy rating, temperature and irradiance conditions
• UL 1741 / IEEE 1547 Inverter interconnection and output rating requirements
• NEC 2023 Article 690.8 Circuit sizing and current, including the 690.7 maximum voltage temperature correction tables
• Module manufacturer datasheet temperature coefficient of Pmax and NOCT rating (cite the specific module model on the install)