A number of new cuts have been added for the BSMD absolute gain calibration.

1. No BSMDP cluster can now start within any of the four strips closest to the module phi edges (two on either side).

2. A cluster in one plane is only accepted if the track it is associated with also has a cluster in the opposite plane.

3. All clusters (in both planes) must have at least 3 strips with ADC>3*sigma.

4. A cluster is accepted only if the ratio of the ADCs in the cluster is greater than 80% of the total energy in the cluster and the six strips surrounding it (3 on either side).

These cuts clean up the sample considerably: for instance, the peak at low ADC that was noticed earlier is no longer seen (mostly thanks to the cut on the number of strips).  They also make direct comparisons between the two planes much more meaningful.  However, this comes at the price of some loss of statistics, especially at mid-rapidity.

Below, cluster ADC vs. track momentum is shown in four eta bins, for the BSMDE plane first and then the BSMDP plane.  Note that the peak at ~p=10 is due to trigger effects.

Additionally, I'm working on creating a private version of the calibration trees to include information like the x,y,z position of the track at each SMD plane and the SMD pressure.