Moeller Polarimeter: Results of the Runs with Unpolarized Beam, March 1998

Last updated: Tue, 24 Mar 1998 moller@jlab.org

Detector Tuning

ADC Pedestals

The pedestals were measured using pulser signals without beam.

ADC pedestals
ADC 1 2 3 4 5 6 7 8 9 10 11 12
pedestal 52. 78. 63. 79. 75. 79. 57. 49. 71. 70. 80. 86.

HV Tuning

High voltage of the two aperture counters was set to 1750V. For the Lead Glass detectors ADC data were taken at various HV. The plots of amplitudes versus HV are linear in the double logarithmic scale. After the initial adjustment 3 measurements were done at 4.045 GeV with 3 different high voltage values, with a step of 50V for each channel. The left aperture (AL) counter was used for the trigger. The run 1091 was taken with the optimized voltages which provided the Moeller peak positions at the bin 300 in the ADC after pedestal subtraction, for all LG modules.

HV for the test runs at 4.045 GeV
LG # 1089 1090 1091
1 1730 1780 1680
2 1700 1750 1650
3 1715 1765 1665
4 1633 1683 1583
5 1664 1714 1614
6 1694 1744 1644
7 1731 1781 1681
8 1684 1734 1634

The Moeller peaks were fit and the dependence of the amplitudes on HV is shown on a plot. Using these plots we calculated the voltages which should move the peaks to the ADC bin 300 for other energies. Such a setting is convenient since it allows the same discriminator level independently on the hight of the hit in the lead glass column. We should keep in mind that the dipole provides a dispersion of about 16% for the module size.

Calculated HV for various energies
LG # 3.355 0.845 1.645 2.445 3.245 4.045
1 1696. 1842. 1770. 1728. 1699. 1677.
2 1664. 1819. 1743. 1699. 1668. 1645.
3 1680. 1821. 1751. 1711. 1683. 1661.
4 1599. 1747. 1674. 1632. 1603. 1580.
5 1631. 1792. 1712. 1666. 1634. 1610.
6 1660. 1793. 1727. 1689. 1663. 1642.
7 1700. 1848. 1775. 1733. 1703. 1680.
8 1652. 1786. 1720. 1682. 1655. 1634.

HV and Timing

The timing of the LG left and right sums slightly depend on the HV with a slope of about -2 ns for 100V (see a plot).

Energy Calibration

Runs 1227 and 1228 were used. Events with a coincidence of all 4 signals in +/-10ns window were selected. Calibration was performed in 2 steps:
  1. Calibration of the modules in the left and right columns individually. Distributions of ADC-pedestal for all the modules were filled. The peaks were fitted. The electron energy was assumed to be E=0.5*(Y_center+Y_symm)/(Y_center+Y_module), where: Calibration factors CALIB were found such that CALIB*ADC=E.
  2. A fine tuning was performed. Histograms for the full energy were filled for the cases when the maximum energy release happened in the given modules of the left and right arms. Such a histogram was fitted and the calibration factors of both modules involved were moved in order to shift the full energy to 1. Several iterations were done. This fine tuning did not change the calibration factors but for 1-2%. The histograms used for the fit are presented on a plot where the top row shows combination of Left 1 with Right 1-4, the second row shows combinations of Left 2 ... and so on. The energy distributions in 2 arms along with the total energy are also presented.
Calibration factors 1./ADC_bin
ADC 1 2 3 4 5 6 7 8
calib 0.002014 0.001571 0.001382 0.001180 0.001889 0.001601 0.001373 0.001243

The energy resolution is 5.5% at 4 GeV. Typical lead glass calorimeters would provide something about 3-4%. In our case, however, a part of the showers should escape to the sides making the resolution worse.

Position Measurement

One can not expect a good position resolution because of a large width of the lead glass modules and of considerable shower leakage to the sides. Nevertheless, it should be possible to make a better estimate of the hit position in the vertical projection than just the center of the module with maximum energy. The following procedure was used:
  1. Baricenter positions were found. Naturally, the baricenter distributions have strong peaks at the centers of the modules (see the plots). It is also seen on this picture that the acceptances of two arms are slightly different - it looks like the detector is tilted in a way that the top is shifted to the left (looking downstream the beam).
  2. Assuming that the coming particles have a flat distribution along the vertical axis Y (which is not really true) one can find a function f(Yb) to correct the observed baricenter value Yb, such as Y=f(Yb) by integrating the observed distribution for Yb. It was done separately for the top (1,5) and the bottom modules (4,8) with the distributions for L and R columns added, and for the modules 2,6 with the L ditribution added to an inverted R distribution (in order to minimize the effect from a non-flat initial Y distribution). The distributions obtained were integrated and the hesulting histograms smoothed using his/oper/smooth in PAW. The smoothed functions were used for f(Yb). The results are shown on (see the plots). The resulting corrected distributions for the left and right arm positions are better than the baricenters' distributions, though display certain systematic shifts (see the plots). The two bottom plots present a position correlation of the hits in both arms the left plot was obtained with no coincidence condition while the right one - with 4-coincidence. A line on the right plots shows the correlation expected from geometry of the spectrometer.