I am having an issue refining 2 superimposed MES molecules A and B (morpholino ethane sulfonic acid, buffer seen in the density). phenix.refine works fine up to the end where it crashes because of the following error:

Duplicate dihedral restraints:
  tor id: var_10
  residue: MES, conformer "A"
  atoms:
    "HETATM 3993  O1 AMES T   1 .*.     O  "
    "HETATM 3994  C2 AMES T   1 .*.     C  "
    "HETATM 3995  C3 AMES T   1 .*.     C  "
    "HETATM 3996  N4 AMES T   1 .*.     N+1"


I can't see where is the conflict since all atoms have distinct labels and I provided the CIF file generated with phenix.elbow. What should I change to circumvent this problem?

Thanks

Ben