The permit issues in real time and never asks about August: desert solar and the 25 degree nameplate

The permit is the fast part now. SB 379 put residential solar permits on a real-time clock. Nothing in that path asks what the array makes in August, and in the Coachella Valley that's the number the customer counts.

Start with the datasheet, because that's where the number enters the job. Qcells publishes the Q.TRON BLK M-G2.4+ series, 405 to 430 W, on a Rev04 sheet dated March 2024. The nameplate comes out of a test that fixes irradiance at 1,000 W/m2, holds the cell at 25 plus or minus 2 degrees C, sets air mass at 1.5 and reports the result under IEC 60904-3, which is how a room-temperature cell measurement ends up on a module bound for a desert roof deck. Read the 25 degrees carefully. That's cell temperature, not air temperature.

The same sheet carries a second condition, and almost nobody quotes it. NMOT assumes 800 W/m2 and a nominal module operating temperature of 43 plus or minus 3 degrees C. Minimum power for the 430 W module at NMOT is 325.0 W, 24.4 percent under the nameplate.

Most of that gap is the 800 W/m2, though. The 18 K of module temperature above STC is worth about 5 points, at the -0.30 %/K the same sheet prints for Pmpp. The desert hasn't entered the conversation yet.

The agency that publishes the rating tells you not to use it

The California Energy Commission runs the PV Module List. For each record it publishes nameplate Pmax, PTC, average NOCT and the temperature coefficients. So gamma-Pmax is already sitting in a state database.

PTC is a different laboratory. It fixes 1,000 W/m2, 20 degrees C air temperature, and 1 m/s wind measured 10 m up. STC fixes the cell. PTC fixes the air.

For the Q.TRON BLK M-G2+ 420 the list shows nameplate 420 W and PTC 397.3 W, a gap of 5.4 percent. If you were taught PTC runs 10 to 15 percent under STC, that was true of older modules. It isn't true of this one.

The CEC then says, in its own list notes, that neither figure is a production number. It names the losses. One of them is "high-temperature losses for arrays mounted close to or integrated within a roofline." Listing isn't required by the state at all, and some utilities and local governments may use the lists during interconnection or permit review. So the number is advisory twice over.

Two coefficients exist inside one product family on the state's own list. Worth knowing before you defend an estimate. On the CEC full data export, listing date October 11, 2023, the Q.TRON BLK M-G2+ 405 and 410 carry -0.314 %/degree C for gamma-Pmax, while the 415 on up carry -0.276. One line item apart. Whichever one went into the estimate, somebody can pull the other off the same page.

Two derates on one line, and only one of them is in the code

Sandia's model SAND2004-3535 gives module temperature as irradiance times an exponential of two published coefficients, plus ambient. For glass on glass the coefficients are a = -2.98, b = -0.0471 close roof mounted, and a = -3.47, b = -0.0594 on an open rack. Run both at 1,000 W/m2 and 1 m/s wind. The rise over ambient comes out at 48.5 degrees C flush and 29.3 on the rack.

Same irradiance, same wind, about 19 degrees C apart at the module. That is a differential, not an operating temperature.

Nobody can tell you the operating temperature from the drawings. It depends on the standoff, the roof surface, and the wind at that address. At the -0.30 %/K on the Qcells sheet, 19 degrees is worth roughly 5.7 points of Pmax, and the only thing that buys those points is air under the array. Nothing in the permit set specifies that air.

NOAA's 1991 to 2020 normals for station USW00093138, Palm Springs Regional AP, put the July normal daily maximum at 108.6 F, which is 42.6 degrees C. August normal max is 108.1 F. The station normally spends 115 days a year at or above 100 F.

The second derate is the one the code actually governs. Table 310.15(B)(1)(1) in the California Electrical Code is based on 30 degrees C ambient. Round that 42.6 up to a 43 degree design ambient and a 90 degree C conductor lands in the 41 to 45 row at 0.87.

Section 310.15(B)(2) then handles raceway exposed to sunlight on a rooftop: with less than 3/4 inch of clearance to the bottom of the raceway you add 33 degrees C before you go to the table, which puts that same 43 degree ambient at 76, drops it into the 76 to 80 row, and takes the identical wire from 0.87 to 0.41. The 2020 edition set that threshold at 7/8 inch, the 2023 edition tightened it to 3/4, and the 2025 California Electrical Code, published July 1, 2025 and effective January 1, 2026, is built on that 2023 NEC.

Which design ambient goes into the table is not a code question. The table doesn't name a source for the outdoor temperature. Round 42.6 down instead of up and the adder puts you at 75, in the 71 to 75 row, at 0.50. The plan checker either takes the number or doesn't.

At the array itself the only temperature correction the permit set applies runs the other way: Section 690.7 sizes maximum system voltage off open-circuit voltage corrected for the lowest expected ambient. Section 690.8 sizes conductors from STC short-circuit current, 125 percent, then 125 percent again. Cold.

The whole heat problem sits outside the drawing set. That's why the shortfall never gets caught, and why nobody in the approval chain is on the hook for it.

The month the desert loses

PVGIS v5.2, running the NSRDB radiation database over 2005 to 2015, models a 1 kWp building-mounted array at 20 degrees of tilt, due south, 14 percent system loss. At 33.830 N, 116.507 W, the Palm Springs Regional AP coordinate, it makes 1,718.4 kWh a year. At Santa Monica the same array makes 1,590.4. Palm Springs wins the year by 8.0 percent, and the temperature loss line reads 13.58 percent against 11.65.

So the heat doesn't erase the irradiance advantage. It narrows it.

The month is where it breaks. Palm Springs runs 176.6 kWh per kWp in May, 167.5 in June, 155.9 in July, 155.8 in August. Santa Monica runs 160.8, 149.7, 155.5, and 159.9. The two series cross between July and August, and by August the modelled coastal array is 2.6 percent ahead per installed kW.

August at Palm Springs sits 11.8 percent under its own May peak, while plane of array irradiation over the same step falls from 247.4 to 226.9 kWh/m2. That is 8.3 percent. So most of the August shortfall is sun angle, and roughly three points of it is heat. Three points is the whole argument, because the coastal array only finishes 2.6 percent ahead.

The valley isn't one number. Indio comes to 1,680.1 kWh at a 16.25 percent temperature loss. Palm Desert, 1,700.5 and 15.52 percent.

Who eats it, and when

Not the plan checker. He verifies ampacity, voltage under the lowest expected ambient, standoff and labelling. A production estimate is not a code document. Under SB 379, non-exempt cities and counties have to issue that residential permit in real time to a licensed contractor, up to 38.4 kW AC with paired storage. So the check that was never going to catch this now happens in seconds.

Not the utility either. SCE wants final electrical permit approval from the building department in the interconnection package, runs Rule 21 initial review inside 15 business days once the package is complete and valid, delivers a draft agreement 15 business days after that where nothing needs upgrading, and issues Permission to Operate within 10 business days on a simple, complete, solar-only request. Not one of those steps asks for an August figure.

The estimator eats it. He priced off the nameplate, or off a template built somewhere cooler. Then he put a summer kWh number in front of a homeowner whose air conditioning peaks in the month PVGIS models the array falling behind the coast.

Same as the reroof-and-array sequencing problem, the decision gets made early and the consequence arrives late. Whoever picked the standoff height made the call that matters, and no plan check in California asked him to justify it. That's true of nail-on solar assemblies too, where the module sits about as close to the deck as it can get.

That reading is the comfortable one. The missing check isn't always the story. Sometimes the mounting detail is just a bad detail, and no reviewer was ever going to save it.

Nothing on the approval calendar prices that 19 degree gap at the module. Nothing on that calendar takes more than ten business days to not price it, either.

Interconnection runs on business days and the estimate runs on months. NEM billing setup runs a few cycles behind Permission to Operate. Whoever is selling desert solar into that calendar closes the file in the fall. The bill that tests the number shows up the following August.