I took a look at run 20133015.

The issue was that out of 1513 events in the event pool (averaging a few hundred Kb) in size, there was a single monster event of 115Mb.   I do not currently understand the cause of this event, whether it is real or an artifact of something else.   The event was #743 occurred at 6:17:29, about 50 seconds after the run started and about 1 minute and a half before the run ended.

Here are the online plot for a single sector as the with all events:

With the monster event (#743) removed we recover a "typical" plot:

With just the monster event, we get something very near the original plot.   Note that this is for a single event.   Later I will show the charge step which is relatively uniform for the TPX, which means that the average ADC value is about 100e3/400= 250, which is similar to the charge in a large hit. 

If you look at the sizes for each detector, the both itpc and tpx get blasted.   For this event:

    itpc size = 42.2 MB
    TPX size = 77.3 MB




The charge steps for the inner and outer sectors are significantly different, but I think that is because of differences in the handling of ADC data when the event is completely black.

Inner:

And outer: