@sdobbs tracked this down to the merging of hits in the same channel which are within the 25 ns TOF_TWO_HIT_RESOL parameter. The previous algorithm took an energy-weighted average of the hit time which lead to the long tail in the timing distribution in MC that is not observed in data. This PR uses the earliest hit to set the time to match the leading edge time we use for these detectors in the electronics.
TOF hit times in MC showed a much longer tail of late arrivals than in data, as discussed in the halld_recon issue https://github.com/JeffersonLab/halld_recon/issues/780.
@sdobbs tracked this down to the merging of hits in the same channel which are within the 25 ns TOF_TWO_HIT_RESOL parameter. The previous algorithm took an energy-weighted average of the hit time which lead to the long tail in the timing distribution in MC that is not observed in data. This PR uses the earliest hit to set the time to match the leading edge time we use for these detectors in the electronics.