If you know that none of your atoms are on top of each other, one thing to do is to double check your constraint definitions. If you accidentally repeated an atom in a constraint definition, then you’ll have a zero length bond from the atom onto itself, and that can lead to such an error.
The other thing to consider is if the bonds are not starting off as zero length, but are becoming zero length during sampling. Double check your ligand conformers to see if there are any odd or broken conformers.
.
