I think Phenix can handle it without the need of specifying whether it will be A or B. These might be too "strong" constrains in a way from my point of view.

Your DNA may be mostly B if it is unmodified, with the exception of some part of the helix that has some C3´-endo puckers and therefore there "it would be A", say. By default the nucleic acids restraints are on, I would just keep working and refining, safest way to go indeed is to place the phosphates where they should be, Phenix will take care of assigning the puckers.

If you want to compare your dsDNA with an ideal A type of helix DNA or B type of helix DNA, you can just build yourself some withing coot. You find this option under calculate/other modelling tools/ideal RNA/DNA. 

I hope this helps,

Best wishes,

Almudena