[phenixbb] bulk solvent mask question

I notice that during refinement of a MR solution for a protein (search model is 100% homologous to target), a very large ligand-binding cavity (as much as 20 A wide in some parts) in my protein gives weak positive difference density for a ligand soaked into the crystals, unless I turn off bulk solvent modeling. Then the fo-fc difference density is quite strong (2.6 Angstrom data). I think this is due to part of the cavity being included in bulk solvent? Can I edit the mask easily, or flag parts of the protein as NOT exposed to solvent? Thanks Laurie Betts