How do I force some residues in a non-helix conformation?

I have a 3A data, and have the protein model refined. But in some region, 3~4 residues are in the helix conformation, and the density fit is good. Anyway, let me say this, I want to force these residues in a non-helix conformation. 

I tweaked the model a little bit and load them in VMD, and make sure they are not helix anymore.

I turn on secondary-structure-restraint in phenix, take out the helix remarks for these residues in the def file. but once the refinement is done, they are still in a helix region (phenix automatically secondary structure analysis, or viewed in VMD).

Is there a feasible way to keep them away from helix conformation?

Thanks!

Charles 

--