I have been trying to estimate the LED intensity variation on a run by run basis. The basic idea is to divide cells into different groups in which they are connected to a single LED pulser.

Data: all of the runs (8) on day85, 9 runs on day80,

1 .average LED adc over the run v.s. run#

http://www.star.bnl.gov/protected/spin/yuxip/celladcled080.pdf

http://www.star.bnl.gov/protected/spin/yuxip/celladcled085.pdf

2. group of cells fed by same pulser

LED map: www.star.bnl.gov/protected/spin/yuxip/ledmap07202010.gif

LED ch1, 19 cells: www.star.bnl.gov/protected/spin/yuxip/CelladcLED_080_1.pdf

                                  www.star.bnl.gov/protected/spin/yuxip/CelladcLED_085_1.pdf

LED ch2, 17 cells: www.star.bnl.gov/protected/spin/yuxip/CelladcLED_080_2.pdf

                                  www.star.bnl.gov/protected/spin/yuxip/CelladcLED_085_2.pdf

the green points in the last frame are calculated by sum( adc(i)/adc(1))/Ncells, where adc(i) is the mean adc of run i, then the ratio was summed over all the cells in this group. Ncell is number of cells in the group.

Only 2 groups for comparison are available up to now since I can't find the actual text file that Jingguo used to generate the map.

On the other hand, analysis on a day-by-day basis is also necessary to look at even slower variations, if there is no strong correlation within the group.