Pedestal drift problem

This problem is considered different that the pedestal noise problem, which is described here. There are two problems with the pedestals,
1) The ADC pedestal depends on the AMU cell number ( look here for some sample plots)
2) The pedestals drift over a period of hours.

Here are some links that Hubert collected related to the pedestal problem. Here are some of my notes about the noise problem. Another set of links on a wider variety of noise-related problems is here.

It is not certain that it is always true, but the pedestal shifts seem to occur suddenly. We only know that "suddenly" means less than a few minutes. It may happen in much shorter time periods. One piece of evidence in favor of the sudden pedestal shift occurred in some test runs we did not try to set the level-1 timing. I can not find the relevant plot right now. However, during this set of runs, the beam was accidentally dumped in the PHENIX IR (someone "pushed the wrong button"). This happened between two of our test runs. The pedestals shifted downwards by 25-50 channels (I do not remember it more accurately) where ~30 channels = 1 mip. Over a series of runs taken in the next ~30 minutes, the pedestals gradually drifted back to within about 5 channels of their original location. I take this as evidence of sudden pedestal shifts which may be caused by beam.

There is also a tendency for the pedestals to slowly drift with time, as shown in this work by Heonsoo Koo.

This pedestal drift is not seen in a set of events taken over a period of about 5 hours on the bench -- in a MCM test setup which is a lot simpler than the real setup. That test also shows the AMU cell number variations, but with a much smaller amplitude than in typically seen in the deal data.

Another interesting test was to look at the noise with and without the photon converter. This was a cylinder of brass wrapped around the MVD for a series of runs. There was no obvious difference in the noise.