I have looked at Run 12075028 for time dependence.

Look at the following plot (long pdf file)

 http://www.star.bnl.gov/protected/spin/heppelmann/tmp/ADCLed75_28.pdf

The plots are organized with a pair of plots for each cell (side by side) and 8 pairs per page.

The first of the pair is the adc distribution for this run in blace and the led distribution in blue.

The Second of the pair is the led distribution organized in 40 led event points.

The ave led-adc value with error bar is plotted as a function of something proportinal to event number.

The set of led-adc points are fit to a constant line. If chi/dof > 2 I fit these points to a 3rd order polynomial.

We can scan through this file and see how many events need special attention by looking for non-horizontal  lines.