On Friday 15-MAR-2013 Devika showed JPlots from pedestal runs taken with a combination of beam and detector HV states. From all those plots I concentrate on the ones that give us, for the first time, an actual look at activity in the detectors other than ionization left by charged particles produced by collisions. The runs where there was no beams in the machine and the HV in the FGT was up to "Physics"
mode for more than 15 minutes are the ones I have looked in some detail. Fortunately, Akio's StFgtQAMaker can display the location of "point" in each disc. (Points are by definition the centroid of clusters.
The activity that we report here is such that can form clusters.)
The following figure shows such distributions for run 14066069:

Disc5 is the one that has the highest well localized activity. The  active spot on this disc correlates well with a low efficiency corner on the second quadrant as seen in the figure below.

The active spot in disc 1 is also present in the collision data and may not be of much consequence. Disc6 has similar weak activity.

The pulses that ones sees during this "No Beam" run are very similar to the ones from collisions. The following figure is used to support that statement;
The timing is, as expected not correct. As seen in the top left figure below, where the MPV from Landau fits to pulses is presented; the majority of the pulses were collected before the trigger that started the readout of the FGT. Some pulses also arrive late.   The width of the Landau shapes, shown on the top right figure is similar to the ones we find in collision data.

The bottom left shows the Integrated charge in the pulse, again with some similarity to actual collision pulses, Finally the difference data - Landau shown on the bottom right is telling us that the Landau distribution is an appropriate description of the pulses we are seeing during these "No Beam" runs.

The conclusion I would reach from what I have seen so far, is that the activity in disc 5 is not catasthophic breakdown but may be some continous avalanche in one or more GEM holes (this almost qualifies as "corona"). The other quadrants have less of a problem; that activity may be normal random avalanches but the fact that some appear well localized makes one think that they could evolve into features
like the one present in disc 5.