What I have started doing is working on a vastly reduced version of bigger projects that I want to accomplish. You could call them prototypes, but I try not to saddle myself with the baggage of the larger project...

I use the larger project as a starting point for deriving the smaller project, but once I've got the small project defined, I do my best not to be hindered by the larger project ideas and just work the smaller project by itself.

I try to make it small enough to finish in a weekend.