At least for run 8911, if I try to use the Device_Error_Code==0 cut on the dslumi sum, I only get a handful of events. dslumi3 looks bad in all three cases, for run's 8906 and 8909 dslumi's 1 and 5 also look non gaussian. dslumi sum widths if at all meaningful without regression are
8911 (3.9 FR): 1695 8906 (3.7 FR): 1942 8909 (3.3 FR): 2314
run 8911
run 8906
run 8909