Central Fiber Tracker Analysis

Please refer to the CFT homepage for a general description of the Central Fiber Tracker project.

skip to:

back to:

Introduction

During early prototype work with the fiber tracker, it was claimed that each MIP generates on average 12 photo-electrons per scintillating fiber. As of January, 2000, the cosmic ray test of actual cylinders to be installed in the D0 detector sees only 7 photo-electrons per MIP, i.e. 40% less than was expected from the prototype work. Where does the discrepancy come from?

The purpose of this study is to examine the X-ray test stand data versus time in order to check for a large decrease in light output. Several batches of fiber were used during the construction. Perhaps the cylinder on the cosmic ray test stand being tested, cylinder 3, was made with an isolated batch of fibers with lower light output? (NOTE: a new cylinder 3 is currently being built.)

To the extent we are limited by the quality of the X-ray test stand data, there is evidence that the light output has slightly changed for different cylinders during the year long production of ribbons, see the analysis plots below and judge for yourself.

NOTE: The current understanding of the CFT group is that the waveguide fibers, and not the scintillating fibers analyzed below, are the origin of the decreased light yield. The outer cladding of the production waveguide fibers does not seem to transmit light as well as the prototype fibers.

back to the top

Analysis Plots

Here is a plot of the first photomultiplier's output during a X-ray scan across stereo ribbon #22 of cylinder #5 (file "b") at the scan point closest to the PMT (near side of the ribbon). Notice how there are 16 peaks per PMT, as expected by the multiplexing scheme (see a sketch for clarification). The light blue curve is a 6th order polynominal fit to the background. The dark blue curve is a Gaussian fit to the fiber signal, sequentially done to each peak; the area of the fit is recorded. Occasionally, as seen in this plot, the edge fibers are slightly damaged and have a lower counting rate.

back to the top

The fits seen in the previous plot are made on all 16 PMTs at each of the scan positions and the results are processed, as seen in the next plot (cylinder 3, ribbon 32, file b). The RMS of the residuals to a line fit across the fiber indicate that the fibers have been very precisely glued in the ribbons. The attenuation and the reflection coefficients extracted from the fits are reasonable; however, the limited number of data points caused large fluctuations in the fit values (with respect to other, more precis ways of measuring these coefficients). Note: points on the right plot are ribbon averaged count rates at each scan point along the ribbon.

back to the top

In the next plots, "day 100" corresponds to June 18, 1999 (The first three months worth of ribbons were discarded, i.e. never mounted on a cylinder.) and the most recent ribbon data was taken May 8, 2000. The data plotted span an 11 month period. Only ribbons that were eventually mounted on a cylinder were analyzed. Each point is one ribbon's averaged or fit value of that quantity. The count rate has been normalized by the ⁵⁷Co X-ray source lifetime (An exponential was fit to the data, and then that function divided into the data; hence, the average value is one in the upper left plot). It must be noted that for each series of a few ribbons which are placed in the setup, new waveguides and connectors are used; therefore, variations in addition to that of light yield are to be expected. Furthermore, we assume that the photomultiplier's gain has remained constant over the year. In light of these considerations, there is evidence of real changes in light yield over the year long production of ribbons, with cylinder 4 ribbons showing the highest light yield. The attenuation length of the cylinder 2 ribbons shows a drop of about 1 meter. The light yield for the cylinder 2 and 1 ribbons is in fact worse than is shown in the plot because the 2.52 versus 1.66 meter length correction has not been made, i.e. one expects the cylinder 1 and 2 ribbons to be "brighter" by about 10 to 15 percent because they are shorter.

back to the top

Misc. Details

The fibers are 835µm in diameter, but the active diameter is 775µm. With a simple convolution integral, one could extract the radius from the X-ray data; however, it would be much less precise than a microscope measurement!
There are 256 fibers per ribbon, stacked in two layers of 128 fibers; however, there are a few half ribbons which are used to fill in the gaps in the cylinders.
In the X-ray test stand, the fibers on the ribbons are multiplexed 16 to 1 and sent into 16 photomultipliers.
The scans along the ribbon occur in different directions for the axial and stereo fibers, the stereo "distances" have to be flipped in order to fit the attenuation and reflectivity.
The X-ray source being used is ⁵⁷Co, which decays via electron capture with a half-life of 0.744 years. 122KeV photons are emitted in the transition 86% of the time. Because of the photomultiplier multiplexing scheme, I think I can also see the double Compton Scattering in the data because of the shape of the background!
The stepping motor is not moving in precise 100µm steps. There is a 50µm spread in the steps; however, it has been claimed that the position encoder is very accurate.
The length of the outer 6 layers is 2.52 meters; the length of the inner 2 layers is 1.66 meters.
The waveguides used to read out the fibers were changed quite often, so that is the origin of the occasional multiplexing errors in the data (cables plugged in upside-down)
General algorithm has to be used in order to account for the improperly mounted waveguides or broken fibers.
I've assumed that the start and the stop of the scan occur 21cm from the ends of the ribbons.
Gaston's database can be found at d0server4/users/gaston/cylinder_8b_7a_6 (or a slight variation of this file name).

back to the top

Send comments and suggestion to brsmith@fnal.gov

Last modified: May 23, 2000