1. HyCal Energy Calibration using dedicated Compton runs
Yelena Prok
PrimEx Collaboration Meeting
July 29, 2005

2. Objectives Check the quality of calibration
Monitor gain drift with time
Comparison with LMS monitoring
Obtain gains for every run (possibly)
Compton runs
first calibration
second calibration
10/09
11/22

3. Compton Runs Carbon target, beam energy 4.9<E<5.5 GeV
7 groups of runs, in chronological order
Data taken on the following dates:
Oct 10, 2004
Oct 19, 2004
Nov 2, 2004
Nov 3, 2004
Nov 9, 2004
Nov 14, 2004
Nov 19, 2004
Radiation
Incident
Run 5050 1 2 3 4 5 6 7
Calibration constants change in the database

4. Selecting pairs of clusters
|((e1+e2)/ebeam-1.0)|<0.2
emin > 0.5 GeV
Area around the central hole (§ 6 cm), and the area most affected by the pair contamination are cut out
|Ebeam-EC|<0.4 GeV, EC=me*sin(/[2*sin2(/2)*tan(e)]-me
Ebeam – EC (GeV)

5. Selecting electrons and photons
Detector Match package is used
to select electrons/photons
(based on the HyCal-VETO match in x)
Some distributions for selected pairs:
|1 - 2| (deg)

6. Calibration Procedure
“Expected” energy of the photons/electrons can be calculated from the Compton kinematics:
E(el)=Ebeam/[1+2Ebeam/me*sin2((el)/2)]
For each HyCal module, divide the measured cluster energy by the “expected” value: Gain_Corr is defined: EHyCal/E(el)
Module 1498
Using photons
Gain_Corr = 1.038
 » 3.8 %

7. Results: Gain_Corr vs Modules
photons
Following the outlined procedure, gain correction factors (Gain_Corr) are found for modules, using selected photons, electrons and a mixed sample of both.
7 colors correspond to 7 groups of Compton runs
Depending on the module and the group #, Gain_Corr varies from 0.95 to 1.11
In order to apply this correction, gain factor from the calibration database would have to be DIVIDED by Gain_Corr
electrons

8. Photons vs Electrons Group 2
Modules (lost files during first calib.)
Group 7, closest to the second calib.
Following the radiation incident
Group 5
Gain_Corr factors,
found with an electron sample
are consistently higher (1-3%)
that the ones found using photons

9. Photons vs Electrons
Mean gain correction factors (+/- RMS) vs group number are plotted .
Green: electrons
Blue: photons
Black: mixed sample
Top plot: x >0
Bottom plot x<0
Electron gains are consistently higher
Possible explanation: The shower depth correction was derived for the photons
Group number

10. Left(x>0) vs Right (x<0)
A spatial asymmetry in the gain correction factors is observed between the left and right sides of the calorimeter
Modules with x<0 have larger gain correction factors, with a larger gain drift in time
Gain_Corr
x<0
x>0
module

11. Left (x>0) vs Right (x<0)
Distributions of gain correction factors, plotted separately for each group of runs
The asymmetry is particularly strong in group 5 (following the radiation incident); it is not seen in group 7 (closest to the last calibration)
gr 2
gr 2
gr 3
gr 3
gr 5
gr 4
gr 6
gr 7

12. Left (x>0) vs Right (x<0)
Mean gain correction factors (+/- RMS) vs group number are plotted for the photon sample
Gain correction factors are larger for the modules with x<0 by 1-5%
Possible explanation: due to the initial beam angle, one side of the detector is more affected by the radiation damage than the other one
x<0
x>0
group number

13. Beam position matters
x variation
Gain correction factors from two sides of the detector will converge into one value if the center of HyCal is moved artificially (in x or y)
Evaluation of the gain correction factors is sensitive to the beam offset (position on the target) and the beam angle.
The alignment procedure corrects for the beam offset in x and y, but gains remain sensitive to the beam angle
Offset of 1 mm in x results in 1% change of the gain correction value
Need to have information from the BPMs and the photon profiler
x<0
Gain_corr
x>0
Shift in x0 (in arb units)
y variation
y>0
Gain_corr
y<0
Shift in y0 (arb units)

14. Gain drift with time
Selected 5 modules on each side of the calorimeter
x>0:1487,1488,1489,1490,1491
x<0:1496,1497,1498,1499,1500
Evaluated gain correction factors are with respect to the last calibration for all groups
Colors correspond to different modules
Due to the incident, gain factors of the modules with x<0, have changed by »15-20%
x>0
x<0
group number
radiation incident

15. Comparison with LMS
Gain correction factors for the five selected modules with x<0 are plotted vs group number, along with the LMS gains for these modules
The two gains follow the same pattern
More work needs to be done to understand how to use the LMS gains along with the gain correction factors
Gain_corr
LMS gains
Modules more spread out
LMS_gain*Gain_corr

16. Calibration by Iteration
After the gain correction factors (gi) are found for the central modules of clusters, the corrections are applied, and the procedure is repeated:
New gi=[ icluster members Ei/gi ]/Ec
To check the effect of this procedure, the quantity (e1+e2)/ebeam is plotted, before and after the application of gain correction factors
Results converge after 3-4 iterations
Group 3
after
before

17. More Results
Group 4
Group 2
Group 7
Group 5

18. Summary of Results Group # Mean before Mean after  before  after
 (%)
1 (low stat)
1.018
1.011
0.0157
0.0147 6 2
0.989
1.002
0.0184
0.0151 18 3
1.029
1.001
0.0193
0.0154 20 4
1.028
0.0188
0.0153 19 5
0.983
0.0488 70
1.006
0.0168
0.0144 14 7
1.013
0.0148 3

19. Conclusions/Outlook What remains to be done What we learned
Gains drift with time
We are sensitive to the beam position
Gain correction factors found using the Compton data are compatible with LMS patterns
Radiation incident has caused a dramatic change in the gain factors in some modules
This calibration method has resulted in the energy resolution theoretically expected for the detector
What remains to be done
Include low energy events to increase the number of modules that can be analyzed
Need to have beam position information from the beam line devices
Check the linearity of energy response
Understand how to use the LMS data
Check results with the matrix inversion method
Single arm Compton?