1. The dE/dx resolution does look reasonable, but
 2. still there are strange dependences on pad row and Z.  
I guess these dependences are coming from distrotion correction and 
I try to figure out how to eliminate them.