Central Fiber Tracker Alignment Calibration

Please refer to the CFT homepage for a general description of the Central Fiber Tracker project. The CFT Trigger homepage also has a lot of information.

skip to:

• Introduction
• Method
• Inter-detector Calibration
• References
• Misc. Details

back to:

• Bryan's homepage
• Bryan's CFT homepage

Introduction

No interesting physics will come out of D0 until the CFT detector is properly aligned and calibrated.

The alignment calibration of the CFT would be a trivial task if one were to assume that:

• The CFT was a perfectly rigid body (i.e. no temperature or humidity dependencies).
• Each of the 80,000 fibers was perfectly straight.
• Each fiber was unambiguously referenced to the same reference coordinate system.

However, an idealized CFT is a bad assumption, for we know that:

• Unexpected time dependencies may arise.
• The CFT is sensitive to temperature and humidity.
• The fibers are not straight (although, statistically, the fiber straightness is within design specs).
• Since the detector was transported and installed in the D0 detector, neither a detailed in-situ alignment nor a cosmic ray run has been made because of lack of access and lack of AFEB boards, respectively.

At present (Feb 2001, i.e. before I started working on the problem):

• Only ideal geometries can only be used in reco.
• The CFT CMM (Coordinate Measurement Machine) measurements are not yet in the D0 database.
• There is no automatic way of dealing with the residual histograms in reco.

In other words, the bulk of the work remains to be done.

back to the top

Method

Here is the pilot method:

1. Obtain very large event sets with the magnetic field on, off, and reversed.
2. Write out the CFT data to a stand alone file.
3. Ask Jadzia to provide the gain and pedestal data.
4. Fit tracks assuming ideal geometry and measure the residuals.
5. Update and iterate (using my "sly-dog median" method for speed).
6. Once converged, write aligned fiber coordinates out to a database.
7. Make sure reco can read the database.

Once the stand-alone method works, integrate it with the rest of the D0 code.

Discussion on the number of fit parameters

If one assumes that each of the 80,000 fibers is independent and wiggles (that is, deviates severely from a straight line) along the length of the fiber, then conservatively, 2.4 million alignment fit parameters are needed: 80000 * 3 * 10, where it is assumed that a binning of 10 along the length of the ribbon is enough to describe the wiggles, and that r,z, and phi are needed for each fiber segment. (For any track candidate of a 3D fit, one would just use simple interpolation to get the best aligned coordinate estimate. Note: it would of course be necessary to iterative.)

Initially, I will assume that 608 parameters are enough and work from there (r-phi for each of the 304 ribbons). Of course, one has to be careful with the angle of the stereo ribbon versus z.

If I assume that 1.0Hz of events with nice tracks in the rapidity range of the CFT are written to tape, then about 200 hours of running would be necessary to get the zeroth order calibration constants (7040 fibers in outer cylinder * 100 events). To do a good job of the calibration to first order would require ten times as much data, or, in other words, 83 straight days of 24 hours running at 1Hz of nice CFT data. (It is not clear to me how much time we would get to run with the magnet off or polarity reversed.)

Discussion on the size of the stand-alone calibration file

The size of the minimal calibration data file (200 hours at 1Hz) will be about 70Megabytes:

((8bit ADC + 17bit ID)*(16 hits per track)*(2 noise hits)*7040fibers*100events)/8/1000000

The number of noise hits and the number of hits per track are both conservative estimates.

back to the top

Inter-Detector Calibration

Once the stand-alone CFT calibration is working, the fully aligned/instrumented/calibrated SMT will be used as an external reference coordinate, and recalibration will begin. Then, one iterates until convergence.

I anticipate quite a few problems will arise.....

back to the top

References

Gaston Gutierrez, Thomas Nunneman, Alan Bross, and many others made the Lab 3 CMM map of the CFT.

back to the top

Misc. Details

To do list:

• Figure out how to get at the data quickly, i.e. without all of the D0 overhead, so the calibrations can be made in a timely manner.
• Try to understand the history of the existing code (a.k.a. do some code archeology).

back to the top

Send comments and suggestion to brsmith@fnal.gov

Last modified: March 14, 2001