I took a look at the MPS and Hel plots for their 792 vs. 1881 values. Most yielded similar results to run#1090. I looked at other runs, from 1089 to 1096. MPS "good" corresponds to the expected plot of DL.bbMPS_792 vs. DL.bbMPS_1881, while "no good" corresponds to plots similar to run#1090. Number in parenthesis is the number of events seen after the cut has been made, for the graph Draw("DL.bbMPS_792:fEvtHdr.fEvtNum","DL.bbMPS_792<500&&DL.bbMPS_1881>15500")>
run# buffer(y/n) MPS(good/not good)
1089 y good
1090 n no good (2)
1091 n no good (2)
1092 y good
1093 y good
1094 y no good (1)
1095 n no good (1)
1096 n no good (1)