The big question for the calibration right now is whether or not to use the mean to characterize each p slice, and if not, what to use instead.  Closely related to this question is the question of the origin of the peak at ADC sum<100 for large momenta (see Figures 1+2), as this peak makes people reluctant to trust the mean.  Also, if we attempt to switch to an MPV or a fit the peak at 0 will still have to be taken into account for the larger momentum bins.  In the case of an MPV, there would have to be a good reason for picking the MPV of all the bins but the first 1 or 2; in the case of a fit, the peak at 0 would certainly screw up the fit.  So it would be nice to be able to simply discard the first bin (or two), but before that it would be good to have some idea of what the peak consists of.

Figure 1: BSMDE ADC Sum vs. momentum (for |eta|<.25 only, but peak is visible for all eta bins): ignore the overlaid means for now.

Figure 2: Momentum of tracks with BSMDE clusters summing to less than 100 ADC.  Some part of the peak at ~9 is probably due to trigger turn-on, but is all of it?

Figure 3: Eta distribution of BSMDE clusters summing to less than 100 ADC from tracks with p>5.

One possibility that struck me was just that for whatever reason these were tracks where the strip with maximum ADC was not the central strip, but after I discarded all clusters in which the hit strip was not the maximal one the peak remained.  The eta distribution suggests either a pileup effect or late-showering hadrons.

Update 3/24: More evidence for the hadron theory.

Comparing Figure 4, which reproduces Figure 3 for particle momentum between 5 and 10 GeV, with Figure 5, which reproduces Figure 3 for particle momentum > 10 GeV, shows a much more pronounced slope for the higher momentum bin, as would be expected.  Additionally, Figure 6, a plot of the average of the BSMDE clusters with ADC summing to less than 100 shows that the average increases towards higher eta, again as would be expected on the assumption that we're seeing the increased likelihood of hadronic showers at higher eta.

Figure 4: Eta distribution of BSMDE clusters summing to less than 100 ADC for tracks with 5<p<10.

Figure 5: Eta distribution of BSMDE clusters summing to less than 100 ADC for tracks with p>10.

Figure 6: Average of the BSMDE clusters summing to less than 100 ADC vs. eta.