A Quick Look at the PrimEx Dec. 2003 Luminosity Test Data

D. Lawrence -- Dec. 30, 2003

Here are some plots from a quick analysis I did of the dec. 2003 PrimEx luminosity test data. The main purposes here are 1.) to try and identify major problems with the data and 2.) to try and ensure the database and analysis software are working properly.
Here's a discussion of some of the more major problems with the test data.
Triggers
For the runs I looked at (TAC runs with TAC "collimator" in), The trigger configuration was more or less the same with the primary source of events coming from the MOR trigger. There were also so prescaled TAC triggers, and the always present Clock trigger.
TAC Timing
The TDC spectra are still one of the big mysteries of this data set. The plot shown here is discussed in the problems page. One possibility for the above plot was that it due mainly to the low, 35mV threshold if it was somehow very noisy. Here is a 2D plot showing the TDC spectra as a function of TDC threshold. You can see the noise does seem to lessen with threshold, but only slightly. Note that the 35mV timing is shifted because it came from the first level linear fan-out. The 45mV, 55mV, and 65mV thresholds were from a second linear fan-out. The cabling paths were different which causes the descrepancy. Here's the projections of the above plot onto the X axis. This just shows that relatively few events are lost from the "real" peaks due to the TAC thresholds. The secondary peaks seem to be more sensitive to it. The root file with the 2D histogram (2nd plot up from here) can be found here. The macro used to produce the 1D plot from the 2D histo can be found here
Tagger Timing
This plot from run 1530 shows the difference (in TDC counts) between the TAC and the individual T-counters. I don't understand the structure beyond 350 on the x-axis. The band at about 300, though, is just what we would expect. This plot shows the same data as above, only the values have been converted to ns and shifted into alignment using Eric Clinton's calibration constants. Here's one last plot showing that the tagger calibration has aligned the E and T counters nicely. This plot is of the E-T timing difference after calibration.
TAC ADC
This plot shows the correlation between TAC ADC(x-axis) and Tagger T-counter (y-axis). Eugene Pasyuk pointed out that the gate timing must have been very close since T-counter #9 (one he knew was slightly out of time with the others) was not integrating the same fraction of the signal that events triggered by other T-counters were (hence the shift).
Tagging Ratios
I hesitate to even put these up since I believe a lot can probably be done to clean them up. Nevertheless, here they are:
This first one is from the 100pA runs 1529,1530, and 1532 (combined). It represents the tagging ratio as a function of T-counter. Events in this (and the next plot) used the following cuts:
  • MOR was the only active trigger for the event (TAC was out of time so there were no MOR & TAC triggers)
  • T-counter time was within +/-20ns of zero.
  • T-counter did not fire in coincidence with his neighbor (i.e. electron did hit in the overlap region of the T-counters.)
The numerator of the ratio was made using the same cuts as listed above with the additional requirement that the TAC be within +/-20ns of the T-counter hit. This second one is from the 300pA runs 1516 and 1517 (combined). The tagging ratio is much lower (maxes out at around 90%). I suspect it is due to two things:
  1. The TAC rate is high enough (a few hundred kHz) that the TDC deadtime is causing hits to be missed. (This is easily handled in a slightly more detailed analysis.)
  2. The higher beam current causes more background. The +/-20ns cuts used are too wide. They were chosen, however, because there are a couple of T-counters who look to be out of time with the TAC. An improved calibration should easily allow +/- 4ns cuts which will remove background in the tagger.
I'm optimistic that these plots will agree with each other and come much closer to a ratio of 1 once a more detailed analysis is performed implementing the suggests I'm made above.

David Lawrence
davidl@jlab.org
Last Updated
(click image to enlarge)
EPS