Study of Elementary Particle Physics at the LHC

Task A – ATLAS

The University of Michigan, Ann Arbor, MI 48105

UM ATLAS Personnel name list

1 Introduction

1.1 ATLAS at the University of Michigan

The Michigan ATLAS group was formed in 1998 and rapidly developed into a world-class pro-

gram. Our objectives for the program have been to accept major responsibilities and to play

leading roles in many important areas within the ATLAS experiment. Among these assumed

responsibilites we have designed and built large precision muon chambers, designed and built

muon detector frond-end electronics, have led the US detector integration, pre-commissioning,

and system tests at CERN both for H8 and X5 beam facilities and for detector preparations in

building 184. We have contributed broadly to the muon software development including cali-

bration algorithms and strategies for combining other detector data with muon tracking, GRID

computing R&D, and collaborative tools development. We have assumed a new leadership role

within the ATLAS physics community with the establishment of a muon calibration center in

Ann Arbor. Michigan is one of the strongest institutes in the ATLAS experiment.

This document presents a short perespective on the challenges and opportunities for ATLAS

in the three years covered by this proposal. It then describes the university community we

bring to this endeavor. the individuals working under the grant and the university facilities

that are available. We then turn to the work accomplished in the past three year grant period

by the University of Michigan ATLAS group. In the final sections we describe our plans and

commitments to ATLAS and the support we require to accomplish these tasks.

1.2 The ATLAS Experiment

The ATLAS experiment is well into the commissioning phase with most detector components

in place in the ATLAS cavern. Figure. 1 shows a picture of the last major component of the

ATLAS endcap MDT, the A-side small wheen, as it was lowered into the cavern February 29,

2008. This installation culminates the surface activities during which 240 muon chambers were

built, instrumented, and tested. With the first collisions due at the LHC on July 1, 2008, our

highest priority has to be the delivery of a top quality detector to ATLAS in the limited time

remaining. The first of May 2008 has become the target date, our focus, and our challenge. The

Michigan ATLAS group has fully committed and will continue to shoulder the primary load for

ATLAS endcap muon detector commissioning and calibration.

LHC is the first new energy frontier in a generation. The precise discoveries that will emerge

from the LHC are unknown. In spite of these uncertainites, we must be ready to fully exploit

the opportunity with top quality hardware and software. To do so requires that our attention

remain directed toward detector commissioning, trigger, calibration/data preparation, software

preparation for data acquision monitoring, event reconstruction, and for physics analysis. The

software must be fully developed and mature if we are to understand what is present in the

early data. We must also understand the SM physics landscape, as seen by ATLAS, prior to

LHC collisions. Only when those foundations are built and well understood, can we address,

with confidence, the most exciting and compelling issues in high energy physics:

• The mystery of the origin of mass and the nature of electroweak symmetry breaking;

• Whether supersymmetry existing, and at which energy scale such symmetry breaking?

• Whether quarks and leptons have further structure at much smaller distance scale?

• Whether nature indeed has extra spatial dimensions?

• Whether a new interaction force, such as Technicolor, or a new gauge bosons exist?

Figure 1: A picture of the last major part of the muon MDT, the A-Side Small Wheel as it was

lowered into the ATLAS cavern on Feb. 29, 2008.

1.3 Michigan’s Commitments

To extract answers to the above profound questions with LHC data, our group is committed

to continued strong leadership in muon detection focusing on the physics potential for new

discoveries with muon final states. Since we have established a large and highly experience team

working on muon detector construction, commissioning, calibration, and database development

which also possesses considerable expertise in computing and muon software, we propose to

merge these activities into a team sufficient to tackle the task of muon “object” qualification.

To optimize the quality of the “objects” we call muons, we must understand the detector, the

signal muons, the signals we conclude are backgrounds, and calibrate the measurements so that

the precision meets the specification required by the physics goals. These “objects” will be the

input to our physics analysis and to the analysis of others. We will take this service role seriously

including the need to document thoroughly algorithms and procedures developed.

1.4 Overview of the group

The Michigan ATLAS group comprises 25-30 people, including 6 faculty, 4 Research Scientists,

3 postdoctoral fellows, 2 engineers, and 5 students. Four faculty, Chapman, Neal, Thun and

Zhou are primarily working on the ATLAS Project. Chapman has been the project leader for

the ATLAS muon front-end readout certification and for the design of the readout multiplexer

(CSM), he led the endcap Big-Wheel commissioning at CERN; Neal has been the project leader

for US ATLAS collaborative tools development, he also leads the ATLAS GRID computing effort

at Michigan; Thun has been the muon detector construction project leader and the chamber

constrution and quality assurance officer. He is leading the muon detector calibration effort for

endcap muon system. Zhou is the US ATLAS MDT project leader, she led the US ATLAS

endcap work at CERN for single chamber integration and certification. She continues to play

a leading role for ATLAS endcap commissioning. Two additional UM ATLAS faculty, Amidei

and Qian are currently focusing on the Tevatron experiments (CDF and D0), and are currently

in transition from Tevatron to LHC activity.

Four Research Scientists, Diehl, Goldfarb, McKee, and Levin are supported by DoE base

grant and are working full time on the ATLAS project. Levin led the US H8 test beam effort at

CERN, the muon combined performance working group, and a muon data quality assessment

activity in coordination with the calibration center at Michigan; McKee has been the endcap

MDT production and certification database coordinator and is also the US ATLAS network R&D

project leader. He continues to play a leading role at Michigan for ATLAS GRID computing

and for high speed data transfers between the calibration center and CERN; Goldfarb served as

the overall ATLAS muon software coordinator (2002 - 2007), and muon configuration database

manager along with Tiesheng Dai (2008- now); Diehl manages the calibration center in Ann

Arbor. He has also taken the responsibility for the development of programs for endcap muon

calibration, particularly for handling the long tube wire sag RT function corrections.

Three Michigan postdoctoral fellows are resident at CERN, Liu, Strandberg, and Ferretti.

They are supported by base and non-base funds. Jianbei Liu is our liasion to the CERN gas

group for all MDT chambers. He has been extremely valuable for maintaining safety and quality

control over the gas system, making sure that chambers are appropriately outfitted with gas lines,

maintained at the nominal 3 bar pressure, and flushed adequately prior to high voltage testing.

Jonas Strandberg has contributed to the muon system running under control of the ATLAS

daq. He has evaluated the on-line monitoring software and written an operations manual for

shift personnel; Claudio Ferretti has contributed a broad spectrum of software and information

handling software, both for data storage, retreival, and display.

Our group has a very outstanding engineering staff, mechanical engineer Weaverdyck and

electrical engineers, Ball and Dai. They are partially supported by our base program and par-

tially supported by Project funds. These engineers represents one of the strongest assets we have

provided the ATLAS experiment over the past years. They have handled many complex techni-

cal issues. Weaverdyck is majority supported by another activity but serves as an outstanding

reference for ATLAS mechanical designs. Dai remains in residence at CERN and continues to

provide crucial technical support for commissioning, and detector calibration activities. Ball

has assumed management of the Great Lakes Tier-II Center but continues to support the MDT

front-end electronics to which he contributed much of the design. All three engineers have truly

remarkable records over past 15-25 years in experimental high energy physics.

2 MDT Commissioning

2.1 Readout Chain and the Chamber Service Module (CSM)

The readout for the muon MDT chambers is shown in Figure 9 and is composed of passive cards

called hedgehog boards that deliver the high voltage at one end of the tubes and channel the

readout signals from the opposite end to the front-end amplifier-shaper-digitizer cards at the

other end. Digitized data from these cards is delivered to the on-chamber multiplexer (CSM)

located nearby. The multiplexer assembles data from up to 18 front-end cards and sends it along

an optical fiber to event builder modules (MROD) approximately 100 meters away. Michigan

has contributed the design and programming of the on-chamber multiplexer and has coordinated

the certification of the overall design. Our responsibilities extend to the full muon system since

all MDT chambers are instrumented with the same US built front-end electronics.

**Some text about continuing responsibilities for the front-end electronics.**

Figure 2: The Front-end Design of the MDT readout. Tubes connect through passive hedgehog

cards to amplifier-shaper-discriminator chips (ASD) and then to time-digital-converter chips

(AMT) on mezzanine cards that each process 24 tubes. Up to 18 mezzanine cards connect to a

chamber-service-module (CSM), a time-division-multiplexer that sequences the data to muon-

readout-driver modules (MROD) through a Gigabit optical fiber. Initialization of the multiplexer

and mezzanine cards is provided through JTAG from an external-local-monitor-board (ELMB).

Detector control, monitoring, and JTAG are all received by the ELMB from data sent on the

CANbus.

2.2 Chamber Preparations, On-surface and In-cavern Testing

2.2.1 On-surface Chamber Preparations

Commissioning of the ATLAS MDT has proceeded in several major steps or phases first for

the Big Wheel chambers and then for the Small Wheel chambers. In Phase-I single chambers

were integrated and commissioned at CERN building 184 and later in building 191. In this step

chambers were outfitted with electronics and monitoring devices followed by thorough testing.

These single-chamber tests provide the primary chamber QA/QC parameters and database

entry information. In Phase-II chambers and services including global alignment devices were

added. In this phase, fully integrated and commissioned BW Sectors (32), SWs (2), as well

as 16 individual EIL4 chambers were prepared and made ready for installation in the ATLAS

cavern. Phase-II preparations started in October 2004 for BW sector assembly and completed

in the December 2007 with the mounting of the SW chambers.

During the many days of Phase II & III work, Michigan served as qualtiy control agents for

much of the endcap muon effort, discovering problems and developing solutions. A partial list

of issues is below.

The Russian built EO chambers that mount directly to the cavern wall conform to the same

design as US built chambers and use US/Japanese built electronics. For this reason these 192

chambers have made use of the same procedures and daq stations provided by Michigan. In

addition, the EO chamber component IDs and test results conform to the same format and are

also entered into the commissioning database by Michigan. This completely unified approach

to test software and information handling for all endcap chambers has led to a smooth and

successful transfer of configuration data to the ATLAS data acquisition system. Responsibility

for the configuration data for all of the MDT and CSC chambers is one of Michigan ongoing

service responsibilities.

• Discovery of Hedgehog board manutacturing faults (design of MECCA card tester)

• Discovery of a mezzanine testing error (sorting and return of cards)

• Application of Negative HV applied to tubes with high dark current

• Location/repair/replacement of all defective electronic cards, T/B-sensors, cables, etc.

• Discovery of a front-end chip fault (doubling of the readout multiplexer input and output

speeds to sustain ATLAS specification)

• Repair/disconnection of all physically damaged tubes

• Repair of all broken fibers

• Repair of all shorted tubes, disconnecting them from the HV line

2.3 In-cavern Commissioning

In Phase-III the Big Wheel and EIL4 chambers were tested in the cavern with the same small

Michigan daq system used for surface testing five to eight chambers at time. The small daq

station, high voltage, and low voltage supplies required moving between floors of the 9 story

cavern. Testing was deemed absolutely essential after the sectors were trucked from the surface

building, lowered into the cavern, formed into a single wheel, and mated to TGC wheels at front

and back since repairs are essentially impossible now that the wheels are closed.

Later, as components became available, the Big Wheels were connected and tested with

the final readout electronics. The Small Wheel chambers have been connected directly to the

final electronics and tested in the ATLAS cavern. Commissioning with cosmic rays began in

November 2007, first with the Big Wheel Side-C, later with Side-A, and finally with the two

Small Wheels. In-pit commissioning will continue with cosmic-rays until collisions from the

LHC become available. Michigan has led the Big Wheel commissioning effort and has continued

in the Small Wheel effort providing test procedures, checklist, and daq stations developed for

the Big Wheel work. All chamber characterizations and test results have been stored in the

commissioning database loaded and managed by Michigan personnel.

Over the past three years, the Michigan group has also played a leading role in the devel-

opment of the software and database infrastructure essential to the commissioning of ATLAS

Muon Spectrometer. The contributions have covered key aspects of the data chain, from detector

running and monitoring, to data simulation, reconstruction and storage, to physics and perfor-

mance studies. These efforts have been complementary to and supportive of our construction

and installation tasks, and are essential for a thorough understanding of the detector. Steven

Goldfarb coordinated the offline software development, Manuela Cirilli was the Muon Database

Task Leader, and Dan Levin was a major contributor to the muon physics performance studies,

as Muon Combined Performance Working Group Convener.

In Phase-IV testing in the ATLAS cavern, the muon endcap commissioning team has ex-

panded to include individuals from many institutions. Installation and operation of the MDT

chambers in the ATLAS cavern has required immense coordination and many hours of work.

All 432 chambers required connection to the services supplied by CERN but not attached to the

detector elements by them. Chamber gas, low voltage, high voltage, slow controls, alignment

readout, high speed fibers for readout, and trigger/timing control were all connected by the end-

cap team. Michigan has been particularly central to most of these tasks. The challenge in all

of these efforts is heightened by the very tight schedule requiring US team to be ready with the

parts and manpower to step-in when a window opens. Handling the scale of these last minute

tasks in the available time has only been possible with a significant on-site team. Michigan has

provided the largest contingent of this team.

2.4 Old Roles and New Roles

With the broadening activities and increasing size of the operating detector, the Michigan team

has focused on some new tasks and expanded their role in supporting tasks where they were

already prime contributors. Jianbei Liu and student Aaron Ambruster have the primary respon-

sibility for in-pit gas and HV issues, stopping leaks and certifying that the chambers operate

successfully at the nominal HV. J. Chapman and Tiesheng Dai continue to examine the per-

formance of the initialization and readout of the MDT, addressing any issue of reliability and

functionality. Claudio Ferretti provides the code and primary off-line monitoring of the detec-

tor’s operating state and environmental conditions. Manuela Cerilli, Jonas Strandberg, Andrew

Eppig, J. Chapman, and Daniel Levin have all played key roles in the evaluation of data taken

in runs since in-pit readout has commenced. Manuela and Jonas are responsible for the coor-

dination of shifters and the DQMF developers along with the preparation of the Web log of

run quality. Tiesheng Dai has assumed responsibility with Steven Goldfarb for the MDT and

CSC configuration database, insuring that the database has correct entries for all of the nearly

1/2 million channels of the detector. Each of the newly adopted tasks as well as the continuing

responsibilities of the Michigan group are described in detail in sections below. To insure good

coordination of Michigan activities internally and with others, Bing Zhou and J. Chapman have

maintained a strong presence at CERN through the commissioning period.

2.5 Services

2.5.1 Chamber Gas

The MDT gas system must be leak tight to ATLAS specifications throughout the AAAAA

tubelets, BBB gas bars, CCC lines, and DDD valves. The system must provide balanced flow of

91% Argon 9% CO

at 3 bar to all 240 chambers such that 1 or 2 volumes are exchanged each

24 hours. Gas services up to the connection to the chamber frames is provided and managed by

CERN staff. Competing for the attention of this staff has been a major challenge for both the

endcap MDT community and our competition.

Gas tightness is a major factor in the MDT operation. Our success in maintaining the

ATLAS specifications in this regard is illustrated in the plot of leak rate shown in Figure ??

where the scale of the y-axis is the leak rate in multiples of the ATLAS specification, shown

separately for the Big Wheel sectors and the Small Wheels, EE millibar/day. Quality control

and repair work on the MDT gas system has primarily been a Michigan responsibility.

2.5.2 Chamber High Voltage

Providing High Voltage to the MDT in the ATLAS cavern is also a CERN responsibility up

the sector or wheel edge. The contract for supplies is with CAEN (Italy). Delivery of these

commercial supplies has been the most severe bottleneck inhibiting our cosmic-ray testing of

the detector. Once the cables reach the edge of the detector, responsibiltiy shifts to the US teams,

Michigan for the Big Wheels and Harvard/BU for the Small Wheels. Quality control clearly

falls onto the US team for damage repair, leakage current testing for the cables, and chamber

current draw at working voltage. In order to meet the aggressive schedule for installation while

symultaneously being held back to permit access to other detectors, we have had to contribute

to the connection and testing of the High Voltage power supplies from CAEN. A test fixture

has been established at CERN to examine and certify High Voltage modules. Unfortunately,

far too many, X%, of the units do not pass these initial tests. In-use failures are also far too

common. In spite of the difficulties we have certified the wheels for operation within the ATLAS

specifictions at the nominal chamber high voltage, 3080V. A plot of the leakage current is shown

separately for a Big Wheel and Small Wheel chamber in Figure ??.

2.5.3 DCS, LV, and Fibers

Low Voltage services are also provided to the MDT by CERN, again under a contract with

CAEN. Slow delivery of modules has also produced delays in testing with the final readout

electronics. Cabling of the low voltage power up to the detector edge has also been provided by

CERN. For the Big Wheel the commissioning has only been troubled by the shortage of final

readout electronics and power supplies. The compressed commissioning schedule that resulted

in the capture of the MDT Big Wheel between TGC wheels left us unable to discover and repair

a few faults before they were inaccessible. This is particularly true for the fibers. Many of the

Big Wheel trigger and readout fibers are radiation soft due to a material acquision problem at

the supplier of the fibers. This sourcing problem has been acknowledged by the supplier and

new radiation-hard fibers have been provided but access to install them is no longer possible.

An early connectorization problem with the fibers was discovered and correction measures taken

by the manufacturer. Any remaining connector separations of the fiber from the connectors will

also be removed when the replacement fibers are installed at the first major shutdown. The

Small Wheel has not experienced either problem.

Figure 3: Endcap Temperature Sensors Distribution.

3 Data Collection, Monitoring, and Evaluation

3.1 Detector Conditions Monitoring

The health of the detector is critical both for avoidance of problems and for sustaining resolution

by monitoring environmental conditions that impact detector operation. These conditions are

examined on-line for exceptions to set alarms for shifters when limits are exceeded. In extreme

situations detector components will automatically shutdown. On-line alarm systems are still

under development and since Michigan is not responsbile for these systems, our descriptions will

be confined to the off-line monitoring of the recorded measurements of temperature, voltage, B-

field. The recorded values are maintained in a conditions database for association with data taken

mfor muon tracking. One example displayed in Figure 3 exhibits the location of the temperature

sensors in each of the stations of the endcap. From the sensors located at these positions, the

temperature profile within the detector can be determined. It provides the necessary calibration

information since the drift time depends on temperature.

3.2 Muon Spectrometer Commissioning with Cosmic Rays

A series of commissioning runs starting in late 2007 has enabled a evaluation of the detector

characteristics from the basic drift tube functioning, to chamber resolution and calibration sta-

bility. These runs have verified the detector connectivity, module identity, and proper database

representation of the configuration and alignment of chambers. Overall performance and cali-

bration characteristics have met expectations at this early stage of commissioning. These early

tests are essential for the muon spectromter to be ready for physics on Day One of LHC running.

A few faults have been identified and repaired when the faulty component is accessible. Our

contribution to these efforts is guided by expert knowledge of the chamber performance and

electronics design, gained over the years of fabrication and particiaption in the H8 test beam.

Cosmic ray commissioning milestone runs, commenced in the Autumn of 2007 and are con-

tinuing through the Spring and summer of 2008. Two types of the commissioning runs are

conducted: combined runs involving multiple ATLAS detector subsystems, and muon system

only runs. These latter runs are intended to provide high statistics datasets wherein the data

are collected in overnight periods with minimal human intervention. Data from these runs are

very useful for MDT evaluations. While known problems exists in the chambers—e.g.: discon-

nected tubes—these commissioning studies reveal problems hitherto unknown. It is critical that

such faults be found quickly during the commissioning period—to allow some opportunity for

remediation.

The commissioning run program is analyzed by two complimentary means. The first is via

the online data monitoring system which has been overseen by UM’s Manuela Cirilli. Jonas

Strandberg, Andew Eppig, Aaron Armbruster and others from our group resident at CERN

have been extensively involved in the data taking and analysis using this online monitoring

framework. The online monitoring is an essential tool that will be used by shifters to insure

data quality during actual LHC running. The commissiong runs provide the ideal environment

to test this framework.

3.2.1 Online Database and Data Quality Monitoring

From 2005-2007, Manuela served, along with Joe Rothberg of the University of Washington, as

the Muon Database Task Leader. In that role, she helped to plan and manage the development

of a variety of databases specific to the Muon detectors, including certification, commissioning,

configuration, and calibration, and reported directly to the Muon Software and ATLAS Online

Database coordination. Manuela’s work typically involved much than project management, as

several of the necessary databases either did not exist or were in an unsuitable state for long-

term maintenance, when she started. She was thus responsible for designing developing tools

and aiding in the creation of a well-integrated, coherent, and maintanable set of databases,

accessible by a variety of online and offline interfaces.

Some of Manuela’s specific efforts included: design of common barrel and endcap tables

for the MDT certification database, definition of data schema and access in the commissioning

database, adaption of interfaces to endcap commissioning data, and development of the MDT

calibration database using the LCG COOL framework. It should be noted that the MDT

calibration database was the first major implementation of a large conditions database, based on

COOL, and interfaced to Athena. This work has expanded to support storage and retrieval of all

conditions data, and will be a key component of the Muon Calibration Centers located at ATLAS

Tier 2 Sites, such as Ann Arbor. Manuela’s proposed design of the database infrastructure for

the Tier 2 sites, including data streaming and replication at CERN, has been tested and appears

to satisfy the difficult 24-hour data turn-around constraints imposed for LHC data taking, an

extremely important milestone in preparation for start up.

3.2.2 Muon Performance

In 2005, Dan Levin was asked to convene the Muon Combined Performance (MCP) working

group, a part of the ATLAS Physics Coordination, steering all activities related to the identifi-

cation and reconstruction of muons in ATLAS. For several years prior to his term, the MCP had

been dormant, so Dan faced the additional challenge of identifying contributors and creating a

coherent project. During his term, Dan organized and presided over a large number of sessions

through which significant contributions to the muon performance have been commissioned and

presented. A solid body of contributors now regularly participates and is engaged in a number

of specific studies of physics performance under realistic conditions. These studies are intended

to quantify features of a large array of reconstruction algorithms, outline performance require-

ments, establish standard metrics and specify the baseline reconstruction pathways to be used

by ATLAS for LHC running. With each software release the MCP group evaluates the new

performance with repect to previous releases and identifies potential faults in reconstruction

that degrade muon sensitivy and momentum measurment. These studies are largely predicated

upon physics processes and address the specific physics reach issues. Specific studies have ad-

dressed misalignment impacts on Z and Higgs reconstruction, the effects of off-centered wires

in drift tubes, evaluations of different inner detector tracking algorithms combined with muon

spectrometer tracking.

In the MCP Dan organized a sequence of Computing System Commissioning (CSC) Data

Challenge based performance notes and appointed editors, and generally coordinated the ac-

tivities of a widely dispersed community of physicists. He provided muon related support to

external (non-Muon system experts) parties and actively encouraged participation of students

in the activities of our group. This support includes documentation of the entire taxonomy of

muon track reconstruction software in an ATLAS website: MuonRecoPedia. This site receives

regular visits and updates from contributors. Dan also contributed to the authorship of the

muon performance section of the new ATLAS Performance paper.

3.2.3 Offline Software

In 2007, Steve stepped down after completing two 2-year elected terms as the Muon Software

Coordinator. Since beginning this task, the muon software effort evolved from a small group of

somewhat independent tasks, to a large and coherent project, with more than 30 full-time devel-

opers contributing to the software. Steve introduced structured project management, including

detailed task breakdowns, personnel planning, and a relatively complete set of documentation.

He also participated on the ATLAS Software and Muon Project Management Boards, as well

as several steering groups, helping to drive the development in a direction that would best take

advantage of the hard work we contributed to the construction of the detector.

At the time of Steve’s departure, the software had successfully proven its ability not only to

simulate physics in one of the most complex detectors ever constructed, but to reconstruct that

data under realistic conditions, including the effects of detector imperfection, mis-alignment,

mis-calibration, cavern background and event pile-up, with high precision and efficiency. Over

the past few years, we have successfully used that same software to reconstruct cosmic data

both in standalone mode and integrated with the rest of the ATLAS detector, as part of our

detector commissioning effort. These capabilities were critical milestones in our preparation for

LHC start up.

3.2.4 Offline analysis

For an alternative and immediate feedback on detector performance we also rely on standalone

algorithms developed by Dan Levin, Tiesheng Dai, J. Chapman, and Andrew Eppig. This code

currently is used to provide a detailed performance evaluation. The topics covered pertain to

low-level and high-level performance diagnostics. The low level analyses focus on the fundamental

and most critical aspects of MDT functioning, such as:

• check that all enabled components report data reliably.

• determination of live, fully efficient tubes and inefficient tubes.

• identification of isolated faulty tube and electronics channels

• identification of the morphology of electronics related faults at various levels of chamber

granularity. The morphology of a fault helps determine if the source is an isolated elec-

tronics channel, a high voltage problem, a gas problem, or related to an ASD chip, a CSM

module or MROD channel.

• visual determination of the integrity of the TDC spectra.

• chamber electronics noise frequency and hit occupancy.

The high-level analysis extends beyond specific measurements of individual tube and elec-

tronic channels and includes measurements of integrated chamber performance. The emphasis

is to evaluate the uniformity of response of multiple chambers. This uniformity is determined

here by a measurement of the maximum drift time distributions of all chambers with adequate

hit statistics. A secondary of objective of the high level analysis is to evaluate the uniformity of

track reconstruction by virtue of the hit residuals when a single time-to-space (RT) function is

applied multiple chambers.

A few examples from this analysis follow: We have prepared polar hit distribution plots

where the chamber, multilayer, layer, and tube tube positions are shown approximately in their

physical locations. This representation allows the pattern of hits to assist in the identification

of the source specific failure modes. Tubes are symbolized by a rectangle color-coded for hit

density. As an example: Figure 4 shows the hit distributions for the Endcap big Wheel Side

A, Eta sector 5. In this run, half of the wheel was powered with high voltage. The annotation

on the plot shows an inefficient chamber multlayer. This problem was linked to a gas flow fault

and quickly remedied.

As an example of a mid-level performance analysis, the maximum drfit time distributions for

all chambers is measured. The maximum drift time is very sensitive to the gas mixture and to

contaminants. So this measurement determines the gas homogeneity in the detector chambers.

Figures 5 and 6 show the maximum drift time vs chamber and run number. The large scatter

of the points is due the low statistics in many of the fits. However, the typical maximum drift

time is near 700 ns, as expected.

The success of the calibration depends on having a stable gas flow and gas conditions.

Here we investigate the application of a single RT function obtained from the UM calibration

chamber. This calibration is assessed from the track residuals. A typical endcap track is shown

in Figure 7. These endcap chamber tracks are fairly oblique to the chamber multilayer plane

and produce many tube hits. Residual distributions are found for those 15 chambers having

adequate number of tracks: Figure 8, shows an example of these distributions with Gaussian fits

around the peaks. The distribution of residuals suggests that the universal RT function yields

reasonable track ”resolution” across all the endcap chambers. This initial attempt to apply a

universal RT is very encouraging. Residual distributions for 15 chambers are fairly uniform and

point to reasonable resolutions on the order of 100 microns.

3.3 Ongoing Work

A special focus of this work is to evaluate performance and calibrations of the MDT chambers.

A single universal RT function, derived from the UM gas monitor shows encouraging results

for endcap. Continuing commissioning runs will expand the number of chambers included in

Figure 4: Hit distribution in Endcap “Big Wheel”, Side A, Eta Section 5, in overnight cosmic ray

Run 33182. Tube numbers increase outward radially. Counting of chamber Φ sector, Multilayers

and layers increases counter clockwise.

Figure 5: Maximum drift time for MDT wheel EMA vs chamber and run number.

Figure 6: Maximum drift time for MDT wheel EMC vs chamber and run number.

these analysis. In a few months we expect that the entire muon precision spectrometer will be

well-tested and calibrated for initial LHC running.

3.4 Configuration Database

4 Online Data Quality (DQ) Introduction

UM has been a driving force in setting up an organization for fast feedback on the data quality

during the cosmic-ray data taking in 2007 and 2008. The shifters are integral to the commis-

sioning to verify that the MDT chambers are performing at the required level. The online DQ

shifters collect fast information about status of Low Voltage (LV), High Voltage (HV), dead

or noisy tubes and the quality of the data. This is then communicated to the muon operation

shifter and to the hardware experts.

5 Monitorig Tools

The online DQ shifter can access the data from the Read Out Drivers (RODs) using the Online

Histogram Display (OHD). The data is collected from the RODs, see Figure 9, and presented

to the OHD with an application called GNAM. The histograms in GNAM contain chambers

level distributions such as TDC and ADC spectra, hit multiplicity and error reports. There are

also summary histograms showing the status of all chambers in a sector or a collection of four

sectors.

The number of histograms available are too numerous for the shifter to monitor and a large

part of the initial DQ work has been to establish procedures and producing macros which can

reliably catch problems with the data.

The muon DQ shifters have pioneered the use of the Data Quality Monitoring Framework

(DQMF). DQMF displays the histograms from GNAM and performs automatic tests for each

chamber by comparing it to a set of reference distributions collected in previous runs. This is

still in a development stage, but is forseen to be the main tool to be used by the DQ shifters in

the future. Producing a set of reference histograms is problematic at this stage of the experiment

as the operation of the detector is still in a state of flux due to, for example, changes in the

trigger, in the readout configuration, in the LV and HV configuration, and so on. However,

effective training of shifters can only be accomplished by having shifters actively using the tools

and providing feedback to the developers.

Figure 7: Example of a tracks in EML5C13 chamber Run26464

Figure 8: Track residuals as seen in Sector 13 of Run 26464

Figure 9: Cartoon showing the muon calibration tools. For quick online monitoring of the data

GNAM and DQMF are used, which accesses the data directly on the RODs. Analysis of the

express stream data and the calibration ntuples allow for a more sophisticated monitoring of the

data quality with a time delay of the order of a few days.

6 Online DQ for Cosmic Data Taking

The current DQ organization has been established since the fall of 2007 and has provided

feedback during the last five cosmic data taking weeks (first in october of 2007 and last in

march 2008). Since the start, shifters have been responsible for collecting information about the

good runs and making it public on web pages such as the one shown in Figure 10. The runs are

graded for quality by the shifter sent to the Tier0 for off-line processing and to the calibration

centers for in-depth data quality monitoring and for determination of calibration constants.

The sheer number of chambers to monitor proved challenging in the beginning and problems

existed for days without being noticed. This spurred the development of more automatic tools

to monitor the features of each run. The organization and working procedures have progressed

to the point that all known problems in the latest data taking week (late february 2008) were

spotted within hours and communicated to the detector experts.

7 UM Contributions

Manuela Cirilli is the current Muon Online DQ coordinator for Atlas and have been the driving

force behind getting the data quality work started. She promoted efficient connections between

the shifters and the hardware experts present at CERN and set up web pages which link infor-

mation to the people stationed at the muon calibration centers. Jonas Strandberg is currently

acting as the Muon Online DQ shift coordinator. For the graduate students stationed at CERN,

DQ monitoring shifts provide an excellent introduction to the raw data coming from the Atlas

muon chambers. All of the UM students at CERN have participated in the online DQ shift

effort.

Monitoring the data will be increasingly important as more and more chambers are being

Figure 10: The on-line DQ shifter grades the cosmics runs and provides run information and

plots in a table of runs available on the web. The runs graded as good are then used at the

calibration centers and are also sent to the Tier0 farm for offline processing.

read out through the final readout chain. UM has had a leading role in the online data quality

effort since the start of cosmic data taking and our expertize is vital to this effort. We foresee a

continued role as the major institution responsible for the fast feedback from the DQ shifters.