I'm not familiar with the T100 Evaluation software but have you tried analyzing the single curves (i.e. not a global fit with all 4 conc) ? I've gotten screwy nonsensical numbers in the past doing a global fit and it has turned out that only one of the concentrations was throwing the whole fit out of line. When I discovered which concentration was askew, I was able to adjust the region selected for fitting and get a more reasonable global fit.
ProfPlum