I replayed the data with my script and the PRL look OK to me.
But when I checked the data base and compared with the map Julie gave this afternoon, I did found there is an error: Last four ADC channels are actually 16, 17, 33, 34 but in the data base they are assigned to 31,32,33,34. I corrected them in all data base, as well as their calibration facotrs.
Although I still don't know why you see the big offset, Pion Rejectors work fine.
Figure 1