Intrinsically disordered proteins (IDPs) or intrinsically disordered regions have not fixed tertiary structure, but play key roles in signal regulation, molecule recognition, and drug target. However it is difficult to study the structure and function of IDPs by traditional experimental methods because of their diverse conformations. Limitations of current generic protein force fields and solvent models were reported in the previous simulations of IDPs. We have also explored to overcome these limitations by developing ff99IDPs and ff14IDPs force fields to correct the dihedral distribution for eight disordered promoting residues often observed in IDPs and found encouraging improvements. Here, we extend our correction of backbone dihedral terms to all 20 naturally occurring amino acids in the IDP-specific force field ( ff14IDPSFF) to further improve the quality in the modeling of IDPs. Extensive tests of seven IDPs and 14 unstructured short peptides show that the simulated Cα chemical shifts with the ff14IDPSFF force field are in quantitative agreement with those from NMR experiment and are more accurate than the base generic force field and also our previous ff14IDPs that only corrects the eight disorder-promoting amino acids. The influences of solvent models were also investigated and found to be less important. Finally our explicit solvent MD simulations further show that ff14IDPSFF can still be used to model structural and dynamical properties of two tested folded proteins, with a slightly better agreement in the loop regions for both structural and dynamical properties. These findings confirm that the newly developed IDP-specific force field ff14IDPSFF can improve the conformer sampling of intrinsically disordered proteins.