Howto check for run-to-run variations in tower pedestals

This document describes a simple tool/procedure for examining run-to-run variations in tower ADC spectra.

First, one needs to generate root files containing adc spectra for each tower in each run. This is described here. One really only needs to run the accumTwPed.C script as described, but make sure the trigger selection in the script makes sense.

With root files in hand, change to the directory in which they are stored (you're probably already there) and grab a copy of checkRuns.C. Executing

$ root -q -b checkRuns.C

will result in lots of output for each ".hist.root" file in the directory, except for the first file which is used (by default) as the reference to compare with all of the other runs.

First, a comma-separated list of towers is printed out, followed by an entry for each run/root file. This entry will be a comma-separated list of four characters, with the following meaning:

• NULL string -- the towers were consistent
• X -- the towers failed a chi² test ( chi²/ndf greater than 2.0 by default), which may be a problem if one of the following are present...
• P -- the pedestal has shifted (default by 1 full channel)
• W -- the width of the ADC spectra disagree (default by 30%)
• 0 -- the mean of the ADC spectrum is below 2 (signals many 0 entries)

One may change the "cuts" used to trigger the different error messages by passing in arguements to checkRuns.C:

$ root -q -b 'checkRuns.C("reference.hist.root",3.0,0.75,0.5,200)'

Sets "reference.hist.root" to be the file containing the reference histograms, sets a chi²/ndf cut of 3.0, warns whenever pedestals shift by 0.75 channels and when widths change by 50% or more. The last arguement tells the code to ignore ADC values above 200 (by default it is 100).