Gerrit suggested the following method of common mode noise subtraction.  I wanted to write it up carefully, as it is easy to write, but harder to explain.  Attached is a more careful, mathematical write up.  The basic idea,though, is to subtract off, per event, the average ADC value (per APV chip) from each channel in the APV chip.

Correlation plots from the cosmic test stand due not indicate that the correlations between strips on a single APV are higher than between strips on different APVs.  However, one could still try the method and see if the pedistal widths decrease.