Introduction

My previous work with the barrel preshower and shower max timing scans is documented here:

http://drupal.star.bnl.gov/STAR/blog-entry/aliceb/2009/feb/02/emc-meeting-2-february-2009

http://drupal.star.bnl.gov/STAR/blog-entry/aliceb/2009/feb/09/emc-update-february-9-2009

The suggestions from the group on how to procede are summarized in this email on hypernews:

http://www.star.bnl.gov/HyperNews-star/get/emc2/2975/2/2/1/2.html

For completeness, the runs used are given in the table below.

Procedure

I take the raw ADC spectrum for the preshower and shower max detectors and pedestal subtract by hand.  I produce an ADC-ped spectrum for the 8 crates of the shower max in each plane and the 4 crates of the preshower.  I do not include channels that have extremely low or high pedestal values or where the pedestal was poorly fit. 

For each ADC-ped spectrum I fit the slope on the right hand side of the Gaussian peak.  These fits are shown in the attachments: smde_fits, smdp_fits and pre_fits for each run, in the order of the table above.  There is still some jitter in the spectrum for the shower max eta and phi, but I was not able to identify the guilty channels.  However I don't think this affects the fit to the slope and can probably be ignored.  The preshower ADC-ped distributions, however, look bad with lots of bumps.  I will try to investigate the bad channels here - a more doable task with 4100 channels instead of 18000.

Next I look at the dependence of 1/slope on the timing.  These are shown for the shower max planes for each crate in: smde_slopes and smdp_slopes.  I also look at the "gain deviation" in smde_dev and smdp_dev.  I did not even attempt this with the preshower given the quality of the fits above.

I also added plots where, instead of fitting a slope, I take the ratio of counts in an ADC region high above pedestal (100,200) to the total number of events.  These are posted as smde_sum and sudp_sum.

Comments

Using the sum instead of the 1/slope doesn't help with showing a timing dependence.  It's possible something was wrong with the data taking and these timings were not, in fact, changed.  I will continue to investigate QA to be done on the preshower, but don't have much hope for seeing a time dependence there.