Some years ago, I downloaded SAM, the System Advisor Model, a tool from the National Renewable Energy Laboratory which allows modeling of, among other things, the performance of photovoltaic solar systems. At the time it was a bit daunting, and I didn't own my own roof so the project seemed a bit theoretical; I got busy doing other things and never used it. But now, I own a house with a roof surface pointing southwest in Wisconsin, well known as the sunnier-than-Germany state. I also have reasonably high electricity rates, and a relatively dirty (lots of coal-fired generation) local electricity generation mix, and I've been told that a PV system makes economic sense even this far north. A great incentive to go play with a new tool!
The new version of SAM has a ton of new features which makes it easier to use, and more accurate (at some expense to ease-of-use). In short, you tell it what your weather is like, how much electricity you use and when, what that costs, and what sort of PV solar system you want to model. Also tell it how you want to model the economics (outright purchase, loan, etc.) and what incentives you qualify for. Then it crunches some numbers and gives you the output: payback time, net present value, hourly statistics for generation, usage, shading, temperatures, etc.
I have an electric car (see my thoughts on the Nissan LEAF) which seems like a great combination with solar power, since it pushes a higher percentage of my electricity consumption into the off-peak hours that makes it easy to sign up for time-of-use billing that raises the rates the utility pays me for power I sell back to the grid when it's sunny.
In Wisconsin, in MG&E's territory, in 2016, with time-of-use billing (you want this!) and net metering (you need this one!), the payback period seems to be around 9 years for a variety of different system configurations and usage patterns. The payback is slightly quicker for a smaller system (~4.5kWDC) because of the way some local incentives are capped, or slightly longer if you oversize a few components to make future expansion easier. The installed cost is slightly above the national average, at almost $3.10/W before rebates (about $1.70/W after, for a small system). The forecasts I've seen for module prices call for a nice drop next year, which might save another 15-20 cents per watt, if the local rebates remain in force. If not, post-rebate prices will probably rise by about the same amount. A roughly 10% change in overall system cost will not significantly change the payback period, but a wider price differential in favor of local generation always helps.