The TPC SpaceCharge & GridLeak calibration which was done for the P12ic preview physics production of Run 12 UU193 used the ZDC sum (zdce+zdcw). I looked a little more carefully at these scalers using my scaler ntuples, and found there are some issues with these scalers...

First is a plot of zdce/zdcw (blue) and zdcenk/zdcwnk (red) vs. day ("nk" means "no-killer-bit", and I used (t-1.303e9)/(24*3600)-258.4 as an approximation of day in eastern time...it's close enough for this purpose):

Next is the three ZDC scalers using killer bits divided by no-killer bits: zdcx/zdcxnk + 2 (black), zdce/zdcenk + 1 (red), zdcw/zdcwnk (blue):

While I can be confident the problems with the east and west are in zdce and zdcw (killer bit versions), I wasn't sure about the coincidence rate until I compared it to bbcx. The next plot is zdcx/bbcx (blue) and zdcxnk/bbcx (red). The latter shows the step in the middle of day 122 which is seen above in zdcx/zdcxnk, implicating zdcxnk.

_______

One additional important check: the above were all made using the 60-second-averaged scalers stored in the DB every 30 seconds. Do the 1-second readings which are stored with the data stream show the same issues? Here is zdce/zdcw from a random sampling of P12ic MuDsts, showing that the problem exists in that data as well:

-Gene

p.s. Jamie pointed out a relevant note in the ShiftLog for day 130 at 15:21: "splitters from ZDC PMT for all 6 channels with 80/20 split; both (80/20) go to summer separately before sent to QT"