DOC

NSS-MIC 2005 Conference Record Template

By Glen Nichols,2014-01-20 02:28
11 views 0
NSS-MIC 2005 Conference Record Template

    F. Petrucci

    On behalf of the Atlas Muon collaboration

    This software structure is being exercised both on test beam AbstractHigh pressure Monitored Drift Tube (MDT)

    chambers, will be used as precision tracking detectors in the and GEANT4 simulated data. Validation of the full calibration Muon Spectrometer of the ATLAS experiment at the LHC (Large chain is based on the analysis of the tracking residuals and Hadron Collider) at CERN. An accurate knowledge of the space-pool distributions. On simulation, direct comparison of the time relation is needed to reach the design average resolution of calibration results to the parameters entering in the digitization 80 microns. The MDT calibration software has been designed to is also possible. extract the space-time relation from the data themselves, through Current plans are to apply te calibration chain to the cosmic the auto-calibration procedure, to store and retrieve the relevant ray events acquired during the detector commissioning in the information from the conditions database, and to properly apply

    cavern and to perform a calibration challenge on simulated data. it to calibrate the hits to be used by the reconstruction algorithms,

    taking into account corrections for known effects like temperature The latter is aimed mainly at validating the handling of a and magnetic field. realistic number of calibration regions and the organization of We review the design of the MDT calibration software for ATLAS the calibrtion data in the data base. and present performance results obtained with detailed

    GEANT4-based simulation and real data from the recent II. CALIBRATION MODEL combined test beam.

     The basic response of the MDT chambers is the threshold

    crossing time (tdc time) and the collected charge. Additional

    I. INTRODUCTION information which are necessary for MDT calibration are the

    trigger time, the bunch crossing identification and the igh pressure minitored drift tube (MDT) chambers will be H measurement of the coordinate along the wire provided by the used as precision tracking detectors in the Muon

    Spectrometer of the Atlas experiment at the LHC (Large trigger chambers and the measurement of the temperature, of Hadron Collider) at CERN. An accurate knowledge of the the magnetic field and of the gas composition provided by the space-time relation is needed to reach the design average Detector Control System (DCS). resolution of 80 microns. To reconstruct tracks in MDT chambers we need to compute This document describes the software developed both to the drift time, to obtain the impact parameter appling the space extract the calibration from the data (calibration framework) time relation and then to fit the track to the drift cicles. and to properly apply it to calibrate the hits to be used by the The calibration software is responsible of computing the reconstruction algorithms (calibration service). required quantities and of providing them to the reconstruction. The MDT calibration procedures, and in particular the The first step is the drift time computation. It is obtained, as possibility of autocalibration, have been extensively studied in (1), from the measured raw tdc time subtracting the t0 and within the Atlas muon communty. The experience gained from applying a set of corrections listed in Table I. the analysis of test beam data, simulation and cosmic rays has been used to develop the software running on hits associated to t? t(ns) - t0TDCselected muon track segments. The autocalibration requires

     (1) multiple iteration over the same set of data which are t ? t ???t(t)performed within a dedicated C++ framework interfaced to the driftii

    Atlas offline framework (ATHENA). The framework also allows the application of the autocalibration procedure separately to different regions (calibration regions) of the Corrected knowing particle Time of flight ?t fMDT spectrometer and provides some flexibility in the choice trajectory. of the calibration regions. Flexibility is needed since size and Correction of the propagation time number of the calibrtion regions are not yet defined. Position along along the wire. The second ?t xthe wire coordinate is provided by the trigger

     chambers.

    Manuscript received November 10, 2005. Correction of the the tdc F. Petrucci is with the INFN - Sezione di Roma III , Via della Vasca Navale , Time slewing ?t qmeasurement knowing signal 84 - 00146 Rome - Italy (telephone: +390655177206, e-mail: petrucci @roma3.infn.it).

amplitude. Up to know, three different algoritms [x],[x],[x] have been developed and at a later stage, after some experience has been Table I. Correction to the measured tdc time.

    gained, the relative quality of the algorithms will be judged and

    the appropriate choise will be made. Another set of corrections, listed in Table II, are due to the

     difference between the nominal value of a parameter for which

    The computed constants have to be stored in the database the r-t relation is defined, and the measured value. These

    toghether with the parametrization used for the corrections. corrections are applied knowing the parametrization of the

    The outcome of the calibration procedure are the t0 and the dependence of the drift dime upon variations of these

    average measured charge for each tube and the parametrization parameters.

    of the r-t relation and of the resolution for each calibration

    region. Temperature ?t TIn principle the various effect changing the t0 and the r-t B field ?t?Brelation could be parametrized and a single r-t could be used Background ?t?bgfor the whole detector. In practice the required precision will Gas composition ?t gnot probably be obtained this way as the correction are not Wire sag ?t scompletely known and environmental conditions may not be correctly measured. Therefore we plan to divide the detector in Table II. Corrections to the the measured tdc time due to variations in the operating conditions. The correction is based on a parametrization of their a set of Calibration Regions that are described by the same r-t effect. relation. About 10000 regions are foreseen; this number depends on conflicting requirements as the statistical and Tdrift is used as the nominal measured drift time. The quantity systematical error on one side and the time spent to collect the t0 that appears in (1) is the relative delay between the different data, the time variation of the environmental conditions, the channels and is obtained with a fit to the drift time distribution, time reqiored to process the data and the size of the data base as shown in Fig. 1, using the ad-hoc function in (2). on the other.

    ???Pt5ALIBRATION SOFTWARE III. C???PPexp23??P (2) dN4??The calibration software has been designed having in mind ??P1dt????????some basics guidelines. First is to fully exploit the facilities ??PttP56???????1exp1exp????????provided by the Atlas offline framework (ATHENA) in terms PP78????????of data decoding, pattern recognition and tracking and database access. Computing algorithms run within Athena but

    do not ave any dependence on it. The calibration should be

    independent from the particular reconstruction implementation

    and thus both reconstruction packages available in the Atlas

    software can be used. Last, we want to easly switch between

    different calibration algoritms.

    The calibration software is composed of 4 different part

    which are in principle disconnected one to another.

    A. Calibration Service

    The calibration service has to properly handle the appplication

    of all the corrections to the measured drift time and to convert it into a drift radius and an associated error. It must provide all Fig. 1. Tipical distribution of the drift time (ns). The result of the fit with the informations to the reconstruction and to the other function (2) and the definitions of t0 and of the aximum drift time are also shown. calibration tasks.

    The second step of the calibration procedure is the B. Calibration Event Data Model calculation of the r-t relation. The Atlas collaboration has The Calibration EDM is the collection of all the data objects developed a procedure (autocalibration) that uses the data needed by the software to comunicate with its clients and to themselves to determine the r-t relation. The procedure is based store informations in the intermediate steps. on an iterative fit, which requires some thousands of muon

    tracks distributed over a wide angular range. The r-t relation is C. Calibration Framework

    modified , until the quality of the track fit is satisfactory. The Calibration Framework computes the calibration

    constants (t0) and functions (r-t relation, resolution) and stores

    them into the transient data store (StoreGate) and on ascii files. The full chain described is almost complete and we are It is a client of the reconstruction and is made of different finalizing a preliminari version to be ready for the cosmic runs elements as can be seen from Fig. 2. in the cavern at the end of 2005.

     EST WITH SIMULATED DATA IV. T

    To perform accurate studies of the developed software

    framework we started with single muons generated in a small

    eta-phi range. A dedicated high statistics simulation of 100

    GeV single muons in one Atlas barrel secor (eta=1, phi=2) has

    been performed. At the moment 500000 events are available.

    100000 of these have been used for the following analysis. Fig.

    4 and Fig. 5 show the distribution of the drift time and of the

    collected charge for all the tubes of the BIL chamber.

     Fig. 2. Diagram showing the different parts of the calibration framwork and their interplay with the rest of the software..

The FMW Calibratin Algorithm is the steering Athena

    algorithm that retrieves data from StoreGate and sends them to

    te tool which splits data between the different calibration

    regions and initializes one EventLoop for each region. In this

    stage the track segments are loaded into memory. The tool is also responsible of the writeout of the computed calibration constants. The EventLoop calls the algorithms that actually Fig. 4. Distribution of the drift time (ns) for all the tubes of the BIL chamber. compute the constants. D. Calibration Database

    The technology choosen for the dtabase is the standard LCG

    COOL interface to Oracle. The model is based on two different

    database; a block diagram is shown in Fig. 3.

     Fig. 5. Distribution of the collected charge (adc counts) for all the tubes of the BIL chamber. Fig. 3. Diagram of the data base model. The distribution of the t0 fitted in each single tube of the chamber is shown in Fig. 6 (left). We have about 2000 hits in All the informations coming from the calibration alghoritms each tube, the mean value and the spread of the distribution are written in a private DB (CalibrationDB). A this level an are of 816 ns and 2.5 ns respectively. This is in agreement with off-line monitoring of this quantities can be done to check the the expectations for the limited examined statistics as can be correct behaviour of the detectors. After a validation procedure seen from Fig 6 (right) that shows measurements done with test the informations are formatted and only the quantities needed beam data. by reconstruction are passed to the condtionDB. This schema has the major advantages that we are free to

    change the private DB that is completely decoupled from the

    clients of the calibration and that the DB does not need to be

    shared if the calibration will be performed on different farms.

     Fig. 6. Distribution of the t0 fitted in te BIL chamber from simulation (left) and dependence of the t0 distribution spread upon sample size (right) as obtained from test beam data. Fig. 8. Difference w.r.t time between the r-t relations computed with two samples of 10000 and 1000000 tracks respectively. The computation of the r-t relation is deeply related to tracking ans thus suffers of the same problematics. The In simulation all chambers have the same r-t relation. In Fig. 9 following results have been obtained tracking on both the difference between the computed r-t for the BIL and BML multilayers of each chamber. This can be done as we are chambers is shown with respect to time and a good agreement dealing with 100 GeV muons for wich the sagitta within a is found. single chamber is of the order of xxx microns and can be neglected. We required 6 hits per chamber and a loose cut on

    the chi2 of the fitted track because the input r-t relation to the

    calibration procedure was different from the real one as can be

    seen in Fig. 7. Fig. 7 shows the distributions of the residuals

    with respect to drift time for tree different iterations. The first

    one shows that the difference input and real r-t relations is

    large (of the order of 50 microns) while the distributions for

    the following iterations proof the convergence of the method.

     Fig. 9. Difference w.r.t time between the r-t relations computed in a BIL and a BML chamber.

    In Fig. 10 the difference between the computed r-t for the BIL

    chamber and the r-t relation obtained in the digitization phase

    of the simulation is shown with respect to the drift time.

    Except for the region near the wire (where an exact definition

    of the r-t relation is a complicate point that we are still

    addressing) we are able to compute the r-t relation within the

    required systematics. The mean value of the difference is 4 Fig. 7. Distributions of the residuals(mm) with respect to time (ns) for the microns with an r.m.s of 15 microns. strdth1 (black), the 3 (red) and the 10 (blue) iteration.

    Fig. 8 shows the difference between the r-t relation computed

    in the BIL chamber with two samples of 10000 and 100000

    events. As can be seen, the difference is within the required

    systematic error and we do not need the large statistics.

     Fig. 11. Difference w.r.t time between the r-t relations computed with the described calibration software and the Calib package. Fig. 10. Difference w.r.t time between the computed r-t relations and the one extracted from the digitization.

    ACKNOWLEDGMENT

    V. TEST WITH TEST BEAM DATA

    TestBeam data In the last years (2000-2004) several tests

    with high energy muons at the CERN H8 muon beam have REFERENCES been performed on different setups of increasing complexity. [1] ATLAS, “Letter of intent”, CERN/LHCC/92-4, LHCC/I-2, October 1992. [2] ATLAS Muon Collaboration, “ATLAS muon spectrometer technical In 2003/2004 a full sector of the Atlas Barrel and EndCap dsign report”, CERN/LHCC 97-22, June 1997. Muon Spectrometer were arranged. H8 data have been used for some test of compatibility of the new developed Calibration

    Software with the standard tools used since many years for

    calibration in smaller setups and in some production sites. The

    results, shown in Fig. 11, are in good agreement.

Report this document

For any questions or suggestions please email
cust-service@docsford.com