Figure 1:
Test Setup
(Top View)
Figure 2:
Test Setup
(Front View)
Included error bars on the subtracted plots
Revised the conclusion to point out that the major contribution of the
scattering was due to backscattering from a fixture behind the source
(based on a suggestion made by Jack and a calculation on my
part)
At Jack's request, I included a plot of occulter vs no occulter
spectra (figure 11) (Rev. B: 9 April 2002).
At Ann's suggestion, modified fig. 2 so it shows the setup all the
way down to the bottom of the aluminum box. (Rev. A: 8 April 2002)
The purpose of this experiment was to determine the effect of placing a segment of the fringe shield next to a Detector Module (DM) when taking spectra. A similar attempt was made on March 16, but not enough counts were accumulated to discern a difference.
A Cobalt-57 source was placed approximately 8 inches above DM no. 118, centered on the DM. The DM was inside an aluminum housing at a GSE test station with the top of the housing removed and a black felt cloth over the top of the box to prevent light leakage. The top of the housing had to be removed so that the shield segment could be placed next to the DM. A bias voltage of 200 V was used. When the shield segment was present, it was placed next to the DM inside the housing as shown in the figures below.
In three of the measurements, an occulter (in the form of a 1 mm-thick lead tile) was taped to the underside of the source. The occulter was 2.3 mm long and 1.9 mm wide, resulting in an occulting angles of 36° in the long dimension and 24° in the short dimension (giving an occulted area of 3.8 inches x 5.4 inches at the detector plane).
Six measurements were taken, each for 60 minutes (true time):
These measurements were saved in the data4 directory of GSE Test Station 4 as:
Shown below are spectra from a detector near the middle of the DM (A-73).
Figures 3 through 6 show spectra taken without the occulter present.
Two extra line features are present when the shield is included (one near channel 220 and one near channel 230). The positions of these lines are consistent with the most prominent fluorescence line from tantalum (58 keV) and from lead (75 keV).
When the copper side of the shield is facing the DM, the tantalum and lead lines are still present, but the lead line (in particular) is much less prominent, as it is absorbed by the layers of tantalum, tin, and copper.
Figures 7 through 10 show spectra taken with the occulter present.
The tantalum and lead lines are prominent in these spectra, too. Note that the occulter data shows a (less prominent) lead line even when the shield is absent. This is due to the occulter itself.
Figure 11 shows spectra with and without the occulter present, and figure 12 shows the difference between the two. For both, the shield segment was in place (lead side facing the DM).
The photon rate was so high that the dead time was non-negligible. The dead time was also much higher when the occulter was not present than when the occulter was present. To compare the two, I had to compensate for the differing dead times. I did this by estimating the amount of dead time for each measurement (by multiplying the number of counts accumulated by 100 µs — assuming the dead time for each event is 100 µs):
subtracting these from the 60 minute true time gives the live time:
live time (no occulter present):
15.87 min.
live time (occulter present):
44.23 min.
The "occulter present" spectrum is adjusted by multiplying the number of counts in each channel by the ratio of the "occulter not present" live time to the "occulter present" live time:
15.87 ÷ 44.23 = 0.3588
The features that appear when the shield are present consist of a prominent lead line and a prominent tantalum line.
It is also possible that some of the scattering shown in figures 6 and 7 is due to the shield (When the shield is absent, the aluminum housing behind it would contribute to the scattering). However, by far the largest contribution to the scattering was an aluminum fixture to which the source was attached.
I couldn't distinguish any obvious differences between detectors near the shield and those far from the shield for this setup.
During the calibration (when the mock fringe shield is placed next to the array) when a control measurement is made without the fringe shield, it may be important to not have any support structure or other material where the fringe shield had been, so that the scattering component from the shield can be accurately determined.