Looks somewhat reasonable to me.
I don’t think you’d actually need the virtual atom, though. I think (though haven’t confirmed) that you should be fine with just the central carbon and having the hydrogens off it.
As far as rotamers, you don’t need a separate rotamers file if you don’t need/want to do any conformational sampling. The ICOOR records should be sufficient for specifying the geometry. (molfile_to_params.py should work for ligands with only one conformation. If you had issues, it was likely due to the small size, rather than having only one conformation.)