Since you have posted this to the Accelrys forum my guess is that by "CHARMM FF method" you refer to the "CHARMm FF" (AKA "Momany-Rone"). This was created by Accelrys, whereas "charmm27, "charmm36", etc. generally refer to the academic set of force fields. Mixing them would be ill advised.
If your goal is to run your final simulations using "CHARMM 36 ff" presumably top_all36_carb and par_all36_carb along with the most recent protein FF then I would not recommend using partial charges for the nonstandard residues based on CHARMm FF (Momany-Rone). Discovery studio can indeed auto-generate residue topologies and charges, and parameter records for a variety of force fields, but nothing more recent than charmm 27 at present. Therefore, if you wish to use charmm36 and have non-standard residues you might obtain charges and other parameters using QM calculations. There is a wealth of information on this topic in Dr. MacKerrell's website:
http://mackerell.umaryland.edu/Empirical_FF_Dev.html