It's definitely a good tactic. It forces you to really imagine the depth of the shapes you're trying to draw underneath the colors so you can adequately figure things out. :)

I'd suggest practicing more with four colors instead of three. Get yourself a colorwheel and set up a palette with the following:

* A primary color

* A secondary, complimentary or overall contrasting color

* A highlighting color

* A shadowing color

The reason for using four colors specifically is because of the Four Color Theorem: http://en.wikipedia.org/wiki/Four\_color\_theorem Essentially, given a set of regions, you only need a maximum of four colors to separate all the regions individually. This way, you can make incredibly dynamic shapes with a very limited colorscheme.