Hi Manoj,
At a resolution of 3.3 A map_to_model is supposed to be able to build regular DNA structures reasonably well, so it is worth trying a few things to see if it can work better.
If you want only the DNA to be built by map_to_model then you do need to tell map_to_model the solvent_content:
Usually it figures this out from the sequence file, but yours will only represent a part of the model so you should specify it directly. This could have an effect but perhaps not huge.
Another thing to try is to cut out the density for a part of your DNA chains and just work with that. Map_to_model can work with a part of a map just fine. So try this: find your best double-stranded DNA regular helix and cut it out with map_box to make a small map. Now try to build with that map. If it doesn't work, you can send me this small map and I'll have a look. If it does work, you can make a few overlapping boxes and build up overlapping parts of the molecule. Then you can combine them all with phenix.combine_models and a map that covers all of them.
Let me know if these things don't help!
All the best,
Tom T