[Ifeffit] High-r range EXAFS fitting by Larch

Dear All, I met a problem about the misfit of EXAFS at high-range (around 8 angstrom). My target is the B site EXAFS of perovskite structure (BaSnO3, a=b=c=4.105 angstrom) I want to study the displacement correlation between the Sn and the neighboring Sn atoms. For the first, second, and third Sn neighboring shells, the fitting is good, but the Sn-Sn4 is bad in both amplitude and phase. I try to include the third cumulant, not become better. Can anybody tell me what's the problem? I suspect the high frequency background signal of k space to produce non-structure contribution in the FT spectra. But the data is very clean, I think this should not be the problem. Best, Fan