I quite like this one:http://www.amazon.co.uk/Practical-Electronics-Inventors-Paul-Scherz/dp/0070580782
...it's good for getting your head round the simple stuff, but also has some breadth and depth in it; plus lots of nice diagrammes ...some of the more academic books can be a bit too wordy and skip through things a bit too cavalier-ly.
...it also depends very much on whether you prefer the American or the British style of book ...what your level of familiarity with EE-theory is (no offence, but there's a lot more to it); and what kind of things you want to design.
I would imagine the best approach is to start with a very general block diagramme of your entire intended system, and then produce block diagrammes of each block, being more specific about what your inputs and outputs are in human language (this becomes your requirements capture); then researching available components (for your specifications capture); before you even begin thumbing through schematics that may not necessarily be suitable for your system.
You might be best off designing your system as though it were made only of off-the-shelf modules; and then look at what you might be able to substitute with your own electronics. You may well find lots of schematics and instructions for the various bits you want to DIY, but you may have trouble making it all work as well if you haven't spent at least a couple of years learning EE theory.
In other words, I don't think "learning electronics" is as cheap an option as it might seem... it'll cost time; and if you rush it, it might cost you money if you make a mistake. A good book or two might cost as much as a decent kit!
I've got one of those "spy books", and they're rubbish really ...trying to take short cuts might lead to short circuits.