It depends on what you are using to drive the relay, if you are driving it with a semiconductor, then yes, you should include a diode across the coil, in reverse bias connection.
This diode will act as a current path when the switch opens, it should be rated for the coil current X 2 at least. Without the diode, the voltage will rise until *something* provides a current path for the inductive spike, usually that *something* will be your semiconductor blowing up.
Relay coil suppression is a more complex issue than is commonly supposed. A simple diode is said to be the worse solution as it slows the relay release, encouraging arcing, and can shorten it's operational life. If the relay is controlled by another relay or a simple switch don't bother. The low impedance of am automobile supply line will soak up any back emf. Of course if it is to be switched by a solid state device, in spite of the drawbacks, the reverse bias diode is a common solution.
There are several informative articles on the net about it. Goggle "relay coil suppression" or similar. here's one: http://www.te.com/commerce/Document...v&DocNm=13C3264_AppNote&DocType=CS&DocLang=EN
But probably you can just go for the single diode like everyone else does. If the relay contacts are not switching several amps it won't be a problem anyway
I was going to link to the article below but will do so anyway with a query about the relevant diagrams on page 12 and 13. Surely the mysterious "R" should be in the base circuit !! As it is drawn the base will surely blow ?
Probably not needed if it is not being switched, i.e. Energized all the time.
Although presumably you will remove power at some point?
If so there will still be a spike at power down.
Max.