I've just started using DMol3 and I'd like to gather everyone's collective wisdom on orbital cutoff distances in DMol3. I'm working with organometallic (d5 high spin distorted and undistorted octahedral systems) systems that can be 15 angstroms end to end, and I wonder if a 5 angstrom cutoff is long enough to capture all the physics.

I'm also having some convergence issues (most of which are fixed by smearing to 0.05 hartrees). I wonder if these issues might be related to my cutoff distance.

I appreciate any input I can get from more experienced DMol3 users.

Thanks.