...
The procedure for CAL gain intercalibration at PS and SPS is based on the CalTuple.
for the 4-range readout runs taken at different scanning positions. At SPS we checked the constants obtained at PS (towers 2 and 3), especially for
HEX8 for which the lever arm in energy deposit was not so large. We alsocalibrated also calibrated tower 1 for which LEX1/HEX8 was not well calibrated at PPS.
To intercalibrate gains of each given log end, we sum almost (see below) all
runs available and fit the profile histogram of s_i (signal for gain #i) vs
s_i+1 (signal for gain #i+1) to get the slope. This profile is limited to the
events where 1/ s_i does not saturate 2/ s_i+1 < saturation value of s_i and 3/
the ratio s_i/s_i+1 is close to 1 within 30% (or close to. In addition, for the logs at the
center of a module, we don't keep the runs corresponding to scan positions near
their end to avoid direct illumination of the diodes due to beam and shower
spread (in that case the scatter plots are a mess...). Despite these criteria,
some weird features were observed.
Let's indeed take the example of tower 2 and the 4+4=8 logs at the center in X
and Y direction (columns 4 to 7): for those we only keep the scan positions 5 to
8 (the scan position range from 1 to 12). Please have a look at pages 4 to 10
of the file in attachment:
- page 4 shows the plots for layer 6, col 5 and side 1 (L6-X5-S1). The , e.g. for the log end T2-L6-X5-S0 (i.e. tower 2, layer 6, column 5, side 0): in this file the 4 plots
at the top are the spectra (summed over all runs) for each range (R0=LEX8, ...
, R1=LEX1, R2=HEX8, R3=HEX1) . The while the 6 plots at the bottom are R0 vs R1 (scatter
plot on the left, with the linear fit of the profile histogram in red) and
R0/R1 (distribution on the right), R1 vs R2 and R1/R2, R2 vs R3 and R2/R3. Here
all plots are fine.- page 5: same plots for the other end of the log (L6-X5-S0): here 2 populations
are clearly visible in the R1 vs R2 plots
- page 6: L6-X5-S0 w/ Y scan runs only; still the same problem
- page 7: L6-X5-S0 w/o the Y scan position number 7 (run 700001796); now it's ok
- page 8: L6-X5-S0 for run 700001796 only: here is the 2nd population
(with a difference of ~7%), the "bad" population coming from one specific scanning position. Since other similar cases were found, new runs were taken at the same scanning positions, but the effect remains (and could not be attributed to a
Thus we decided to In the following, we investigated these discrepancies by looking at the dependency of the slope (from the linear fit of the profile histogram) as a function of run number, and the runs far from the mean were rejected, yielding a list of selected runs for each log end ratio and better results, see e.g. T2-L6-X5-S0.
This procedure was applied to each log end
We have found some other cases like this one. Note that the effect is of the
order of a few %, as you can see in page 10 which gives the fitted slope value
for each channel (left column, points in red): the plot on the left in the
middle show that LEX1/HEX8 slope is close to 1 (a bit less), somtimes as low as
~0.95 or even 0.93 for a few logs.I am investigating these small discrepancies by looking at the dependency of
each slope as a function of run number, however I dont see anything else that
such a systematic study to delimit them (this effect which is not geometric and
only seen on a few log ends like L6-X5-S0). Would you have any other idea ?
Update of CAL calibration constants in DB (Aug 7th) pdf resumé
...