Having an electroneutral system is not strictly necessary (execpet perhaps when you use Ewald summation?), but since it is assumed to keep referees from asking why the system is not neutral one often adds enough counter ions to get a zero total charge...
Set up your system as usual, protein+ligand+water, then decide on a number of water molecules to be replaced by ions; you can pick these waters at random, or use some procdure to compute the Coulomb energy of each water oxygen (inte sele type oh2 .and. resi xxxx end sele .not. resi xxxx end) and replace those with the highest energy (if you have to add positive counter ions), or any other method that you may find suitable in your case.
Once you know which waters to replace there is a short script (add-ions.str) that can be used for the actual replacement. See also run-sdb-md.inp for an example of how to use this script.
Best regards from -5C and snowing,