modifying (tuncating) basis sets?

[Hannes Loeffler <Hannes.Loeffler.,at,.uibk.ac.at>]

Hello,
 I want to use the LANL2DZ for Cu(2+).  Now, this basis set contains some
 diffuse functions which should be unnecessary for a cation.  Some publications
 mention such truncated basis sets and some people in our lab also used to
 truncate basis sets.
 Well, I'm a little bit suspicious about such modifications.  Should such
 modifications be done, i.e. is it safe?  When should a function be considered
 as diffuse?  (LANL2DZ for instance has three uncontracted (1 x s, 2 x p)
 functions with exponents that "look" small.)  What about truncations
 in
 contracted functions?
 Hannes.