Differences in the total ECAL energy between HLT1 and HLT2
During the PbPb run, a difference in the total ECAL energy was observed from the lumi counters between HLT1 and HLT2 for data. Specifically, the HLT1 distribution shows negative values, while the HLT2 distribution does not (Blue is HLT1, red is HLT2).
After investigations from @jmarchan , the reason was traced back to be due to this zero-suppression logic in the HLT2 decoding that is missing in HLT1. Attached plots after applying (1) the 1-D zero suppression and (2) the 2-D zero suppression (identical to HLT2) to the HLT1 total ecal energy algorithm. Blue is HLT1, red is HLT2.
(1) 
(2) 
While the HLT1 clustering algorithm has ADC cuts already applied for both seed and neighboring cells: https://gitlab.cern.ch/lhcb/Allen/-/blob/master/configuration/python/AllenConf/calo_reconstruction.py#L143 it is not ensured that they are aligned with the HLT2 version. Additionally, the total ECAL energy calculation and the electronID in HLT1 seem to not have such cuts applied.
One way to harmonize the different cuts and ensure the best agreement between HLT1 and HLT2 would be to apply the 2-D suppression directly on the HLT1 decoding level, which would require an additional loop over the digits - the throughput impact of such a development will need to be studied. Alternatively, a 1-D suppression, while not giving identical results, already brings a good agreement in the total ECAL energy.