Um. Please take that back?
The SO and I had 9 kW of amorphous silicon cells installed on the roof in mid 2008. Total max output power (not normally hit, the sun has to be at just the right angle on a clear day) is 7.8 kW using two inverters.
There was some fump-de-dump with the bureaucracy, so the actual, "We're getting all the bells and whistles" on SRECs and such didn't kick in until early 2009.
It's now 2024. The panels are doing their things; on a non-completely-clear day I observed the larger inverter chugging along at 4.5 kW out of 4.8kW max and the smaller inverter (3kW) at 2.5 kW. Not bad.
The SO and I get about 12 MW-hr/year out of the things. In New Jersey, on the program we are grandfathered into, we get an SREC (Solar Renewable Energy Credit) for every MW-hr generated by the panels; at this time, they're worth $198 on the spot market. That'll end for us at the fifteen year mark in October. We'll then switch over to a different SREC system that gets us $25/MW-hr of energy generated.
Just so we're clear: The SRECs are for energy generated by grid-tied inverters; it's not how much is delivered to the power company.
Being connected to the grid sets us back about $4.95/month. On the other hand, we seem to have a surplus of around 2 MW-hr/year, some of which goes into moving the Teslas around. With Net Metering, a tariff that carries excess energy generation forward from month to month, we earn about $150-$250 a year as a generator, selling energy to PSE&G wholesale. With Net Metering, if the carried-forward amount goes to zero, then we pay retail, just like everybody else.
The net effect, though, is that we seriously haven't paid for electricity since fall of 2008. Really.
About once every couple of years I get up on the roof with one of those long-handled window washer sticks and a bucket and clean the whole business off.
Now, we're using amorphous silicon cells. When one looks at these, they're blue, but kind of mottled looking. At the time of install, these were far cheaper than crystalline panels, which look a flat, dead blue/black. Crystalline panels are also lots more efficient, so one needs fewer of them to reach a given energy generation goal.
Finally: Back when we did this, the approved technology method of the time had one hooking up panels in long strings, putting those strings in parallel, and then feeding the whole conglomeration into a DC->AC inverter that generates 240 VAC. So, for example, one part of the roof has three strings of seven panels each, with each of those strings wired in parallel, for a total of 21 panels. This goes to one inverter. The other set are two strings of 11 panels, also wired in parallel, which goes to the other inverter. So the inverters see incoming voltage and current that ranges from Zero/Zero to some maximum voltage and current, then back to 0/0 at night. Turns out that all of the above is relatively inefficient, since panels have manufacturing variations, resulting in some strings having higher voltages/power than other strings that they're in parallel with; as a result, with 9 kW on the roof, we only get 7.8 kW out, max, for an 86% efficiency.
Modern systems put a DC-DC converter on the back of each panel; one side of the DC-DC is hooked up to the actual solar panel. The other side is wired in series with, roughly, ten other panels. Software communicating between the converters in a string like this fixes the output voltage of a string to 300V. Since the secondaries of the DC-DC converters all have the same current (they're wired in series), those panels with more power than other panels end up having more than ~30V across the secondary, while those with less power have less than ~30V, but with the total assembly always having 300V. Additional strings of ten can then be wired in parallel with the first string so connected, so the whole assembly is at 300V; and that fixed 300V is then fed into a more-or-less fixed input voltage inverter, or inverter/battery box (this is what a Tesla PowerWall is), that converts to 240 VAC. Efficiency of this kind of system is over 90, more than making up for any losses in the DC-DC converters.
Main point, though: This system is still going strong at 14.8 years. When do I expect that the whole business, inverters and all, are going to die?
By the by: The whole business was paid off within six years of the install. At this point, the SRECs are just gravy.