I wrote a script that compares each Run 7 BTOW pedestal table with a corresponding L2ped printout. The script queries the RunLog_onl database for the runnumber immediately preceding the table timestamp, then it looks for a log file corresponding to that runnumber in the L2ped directory on online.star.bnl.gov. If it can’t find the precise runnumber, it compares the online pedestals to the next closest L2ped runnumbers both before and after, and then prints the comparison for the better match.

Here’s a summary log printing the number of channels per table where

1. the difference in pedestal value is greater than 4, or
2. the difference in pedestal sigma is greater than 1.5

There are a handful of channels per run with drastic differences. For example, emcOnline will declare a zero pedestal, but l2ped will say there’s a peak at 3500 ADCs. Channels like these are obviously bad. Of greater concern are the runnumbers where several hundred channels do not match. Most of the time it’s because the widths are different, but occasionally there are runs where a few hundred towers will have pedestals that differ by more than 4 ADCs. The full verbose output (listing every tower that fails the cut) can be found here.