Okay, I think it doesn't make much sense to normalize lumis with an uncalibrated BCM signal. So, for now, I just normalize them with the BCM average from the epics variable "hac_bcm_average". Here is the new plot.
I am not sure why channel 2 has different shape than others at low currents, but it looks like the pedestal subtracted lumi signals are stable above 4uA.
Figure 1