GPS receivers must be able to see *four* satellites, not three, to
produce accurate positions in spacetime. The reason will become
evident if you set up and solve the equations for determining
position. One can get an approximate position using three satellites
because one coordinate, the distance from Earth's center, is fairly
accurately known and will provide a substitute fourth equation.
I bought my Garmin 38 almost two years ago, when they cost US$200. It
wedges beautifully between the dash cushion and the windshield on my
Dodge Caravan (it looks like it was made to do so) and is a marvelous
navigational aid when one is traversing terra incognita.
The handheld GPS units are the only consumer products I know which
rely for their operation on applied general relativity in more than a
trivial manner. That alone makes them nice demonstration items. I
take my students outside, and the next time I teach special
relativity I have a dandy way to introduce it with space and time
entering symmetrically into the solution of the GPS problem.