With generous incentives, many houses around here now have rooftop solar. In an emergency, however, these panels would not be able to provide electricity: we've regrettably chosen to set things up where almost all residential solar only provides power when the grid is up.

But this does have me wondering: if we had an unprecedented long-term blackout, how practical would it be to jury-rig standard residential solar into doing something useful?

Now, before I go any further, I want to be clear that I'm not suggesting homeowners start messing around with their systems. Panels can provide a lot of power, start fires, and kill you. They're not toys. If there are emergency configurations that would make sense, we should produce clear documentation and distribute hard copies for electricians to use in partially restoring power.

There are two main configurations for solar:

• String inverters: panels are connected in one or more "strings" (series). Power comes down from your roof as DC, and is converted to AC in a stand-alone inverter. Previously very common because you can have a bunch of dumb panels with a single smart inverter, but after the 2017 NEC and 2020 NEC changes in most states you now need module-level electronics to support rapid shutdown, removing a lot of the advantages of this system.

• Microinverters: each panel has its own independent inverter. Power comes down from your roof as AC. More efficient in partial shading, because different panels can be running at different voltages (though you can do this with a string inverter and module-level optimizers). New systems generally use microinverters.

With a string inverter you could potentially disconnect the panels from the inverter and connect them to an independent device. This device could be a standard charge controller and a battery, but it would need to be a pretty powerful version. Generally the charge controller does Maximum Power Point Tracking, where it varies the load it places on the solar cells to maximize the power they produce given the current conditions. This makes sense if you want to get the most out of a single panel, but if instead you have a big array on a roof and are worried about getting too much power it's not the right approach. I think you could make a device that worked without a battery and used this property of solar cells, where power varies dramatically based on load, to make the array produce exactly the amount of power you were currently consuming. Possibly this is what my SPS is doing?

A big downside of this approach, however, is that it only works for string inverter systems, and these will likely become progressively rarer over time.

A different idea would be to disconnect the inverter from the grid and hook it up to something that pretends to be the grid. That means it, as far as I understand the setup, can:

1. Supply a little power. Just enough that the inverter can power its own electronics on startup and have a sine wave to match.
2. Sink a lot of power. The inverter expects it can send out its nameplate capacity and the grid will cheerfully take it.

The first one is easy: you just need a small battery and 240V inverter, and you can look just like the grid. Until the second one kicks in and you get lots of power flowing out, possibly more than you're able to use. For example, our system peaked at about 3.25kW today, and in the middle of summer might get up to about 4.5kW. A bigger system might produce more like 10kW or 15kW. At 240V these would be 14A, 19A, 42A, and 63A respectively.

The simplest option might be a big fat toaster. For $116 you can buy a big resistor in a nice metal case that will take up to 2.5kW at 240V and turn it into heat. A bit of electronics and you make a bank of outlets which will let you draw up to the current capacity of the solar array while sending the excess into the heater. Depending on how it performs if you send it low voltage you might need a few different sizes so you can mix to match the load.

A better sink for large amounts of power, though, might be electric car charging. The cars are nice and portable, are useful to be able to recharge in an emergency, and are generally capable of charging very quickly. The main question here is what the protocol between the charger and the car looks like: can the charger ask the car to draw a specific current, or modulate its output by varying the provided voltage? Or does the charger just commit to providing a specific current at a specific voltage and the car is free to draw or not?

Probably the ideal system here is a battery, a toaster for once the battery is fully charged, and some spiffy electronics? I wonder how much the simplest version would cost if manufactured in bulk? And how difficult hard it would be for an electrician to safely switch an inverter over from a grid connection to this system?