Abstract--An extensive system test of the ATLAS muon
spectrometer has been performed in the H8 beam line at the
CERN SPS during the last four years. This spectrometer will use
pressurized Monitored Drift Tube (MDT) chambers and Cathode
Strip Chambers (CSC) for precision tracking, Resistive Plate
Chambers (RPCs) for triggering in the barrel and Thin Gap
Chambers (TGCs) for triggering in the end-cap region. The test
set-up emulates one projective tower of the barrel (six MDT
chambers and six RPCs) and one end-cap octant (six MDT
chambers, A CSC and three TGCs). The barrel and end-cap
stands have also been equipped with optical alignment systems,
aiming at a relative positioning of the precision chambers in each
tower to 30-40 micrometers.
In addition to the performance of the detectors and the
alignment scheme, many other systems aspects of the ATLAS
muon spectrometer have been tested and validated with this
setup, such as the mechanical detector integration and
installation, the detector control system, the data acquisition, high
level trigger software and off-line event reconstruction.
Measurements with muon energies ranging from 20 to 300 GeV
have allowed measuring the trigger and tracking performance of
this set-up, in a configuration very similar to the final
A special bunched muon beam with 25 ns bunch spacing,
emulating the LHC bunch structure, has been used to study the
timing resolution and bunch identification performance of the
Manuscript received October, 26, 2004. This work is supported by the
Israel Science Foundation and the German Israeli Foundation.
E. Etzion is with the School of Physics and Astronomy, Raymond and
Beverly Sackler Faculty of Excat Sciences, Tel Aviv University, Tel Aviv
69978, Israel (telephone: +41764870750, e-mail: firstname.lastname@example.org).
trigger chambers. The ATLAS first-level trigger chain has been
operated with muon trigger signals for the first time.
THE Large Hadron Collider (LHC) currently under
construction at the European Center for Nuclear Research
(CERN) is designed to collide proton-proton beams at the
energy of 7 TeV, the highest energy ever reached in a particle
accelerator. The ATLAS detector built for the LHC
experiment will collect events at a design luminosity of 1034
cm-2s-1 in a proton bunch crossing rate of 40 MHz (3-20
collisions every 25 ns).
The largest part in the ATLAS detector is the Muon
spectrometer . It has been designed to provide a standalone
trigger on single muons with transverse momentum of several
GeV as well as to measure final state muons with a momentum
resolution of about 3% over most of the expected momentum
range; a resolution of 10% is expected at transverse momenta
of 1 TeV. The ATLAS Muon spectrometer is a 4π detector
consists of four types of detector technologies.
Over most of the spectrometer acceptance, Monitored Drift
Tube (MDT) chambers are used for the precision measurement
of muon tracks. The MDTs are built from aluminum tubes
filled with Ar-CO2 at a pressure of 3 bars. Operated at a gas
gain of 2x104, each tube measures charged particle tracks with
an average spatial resolution better than 80 micrometers.
Cathode Strip Chambers (CSC) are used for muon momentum
measurement in the inner part of the end-cap regions where the
background rate is expected to reach 1 KHz/cm2.
15 Nov 2004
System Test of the ATLAS Muon Spectrometer
in the H8 Beam at the CERN SPS
On behalf of the ATLAS Muon Collaboration
Three stations of Resistive Plate Chambers (RPC) are
employed for triggering muons in the barrel region where
three Thin Gap Chambers (TGC) stations serve the same
purpose in the higher background region of the end-cap. Two
RPC stations are attached to the MDT middle station and
provide the low-Pt (transverse momentum) trigger. Similar
functionality is assigned to the two TGC double gap units
(Doublets) installed close to the end-cap MDT middle station.
For the high-Pt muon trigger the low-Pt signal is combined
with the one provided by the RPCs attached to the barrel MDT
outer station or the three gaps TGC units (Triplet) in the end-
cap region. The signal from the trigger chambers are amplified
discriminated and digitally shaped on the detector and sent to
an ASIC-based coincidence matrix board. These devices
perform the functions needed for the trigger algorithm and
apply the Pt cuts following preset thresholds. The trigger
chambers are also used to provide the coordinate along the
drift tubes that can not be measured by the MDT chambers.
For that purpose several additional TGC chambers are also
installed close to the end-cap inner MDT part to improve the
measurement precision of this coordinate.
A large scale test stand of the ATLAS detector including all
the Muon spectrometer components has been operated in the
CERN north area on the H8 beam line since the year 2000. A
beam of up to 320 GeV provided by the SPS accelerator was
used to study different aspects of the spectrometer. In this
paper we summarize the performance of the Muon
spectrometer and its different components during the 2004 H8
II. THE H8 MUON SETUP
The H8 Muon stand consists of two parts: a barrel stand and
an end-cap stand. The barrel setup is reproducing one ATLAS
barrel sector with its MDT and RPC stations. It consists of six
MDT chambers (two of each type: inner, middle and outer
chambers) fully instrumented with Front End electronics (FE)
read with a Muon Readout Driver (MROD) and fully equipped
with an alignment system. There are six RPC doublets: four
middle chambers (BML) and two outer chambers (BOL) in the
barrel set-up. Two additional barrel stations where used in the
test stand: one inner chamber (BIL) on a rotating support for
calibration studies and one outer (BOS) station (MDT+RPC)
upstream of the muon wall for noise and Combined Test Beam
In the end-cap stand which reproduces a muon spectrometer
end-cap sector there are 11 MDT chambers: two inner (EI),
two middle (EM) and two outer (EO). As in the barrel they are
fully instrumented with FE and read out through one MROD.
The chambers are equipped with the complete alignment
system and calibrated sensors for absolute alignment. One
CSC chamber has recently being installed and should be
integrated soon in the combined data taking. For triggering
there are three TGC units: one triplet and two doublets fully
instrumented with on-chamber electronics.
Figure 1 - A picture of the Muon spectrometer section setup in
the H8 area. On the right handside the barrel three stations
consisting of two inner, two middle and two outer MDTs.
Attached to the middle and outer barrel MDT are the partner
RPCs. On the left the end-cap chambers with six MDTs, two
doublets and one TGC triplet.
III. THE DETECTOR SLOW CONTROL, ONLINE SOFTWARE AND
The monitoring and operation on crucial parameters of the
spectrometer working conditions are controlled via the
Detector Control System (DCS) . The monitoring is done
via a commercial control environment, PVSS-II, which serves
all the LHC experiments. The readout, monitoring and analysis
initialization is distributed via DIM, a data transfer system .
The MDTs initialization and configuration, the temperature
monitoring, the control on the FE low voltage (LV) power
supplies and the barrel and end-cap alignment system were
controlled under the frame of the MDT DCS.
The TGC DCS system  was used to control the CAEN LV
and HV power supplies by the PVSS process communicating
via OPC server . The DCS was used to set the readout
threshold. The DCS was running a chamber charge monitoring
embedded in the DCS boards. The DCS board provided
monitoring of the analog charge of a wire in a chamber over
some time interval. The resulted histograms by the data
analysis performed in the ELMB  (see
Figure 2) were distributed through the CAN bus network .
The figure shows how the system allowed controlling the level
of noise and trigger loss by monitoring histograms at the PVSS
workstation and setting the threshold in a chamber. TGC-DCS
communicated directly with the TGC DAQ and used direct
SQL queries to store and retrieve information from/to the
conditions/configurations database (DB).
Figure 2 - Monitoring histogram of one analog TGC channel
read via the DCS. The blue histogram is collected with a
readout threshold of 180 mV and the pink and yellow are
collected with a threshold of 50 mV and 30 mV respectively.
RPC DCS which recently moved to PVSS was used to monitor
and control the conditions of: the gas status (flux of
components), manifold pressure, HV status (monitoring the
voltage and current), RPCs gap current and LV supplying the
We exploit the test beam date to test the ATLAS online
software. Two software packages were utilized for on-line
monitoring. One is the monitoring framework GNAM which
was co-developed with the ATLAS Tile Calorimeter and was
successfully used in the test beam . The other option which
was tested is monitoring within ATHENA, the ATLAS offline
software framework .
Figure 3 - A window of the online monitoring as used in the
The muon spectrometer used a MySQL DB server serving the
non-event data storage. Nearly a complete loop of
applications using the so called conditions DB has been
implemented. All the quantities needed were stored and read
out. High rate access for raw alignment image results was
used (every ~2 sec). Access from ATHENA is currently under
development, first prototype has been implemented for
Two approaches of communication to the DB were
implemented and tested. One way was connection to the DB
via a dedicated API and the second method was a direct SQL
connection from DAQ and the DCS systems. The second end-
up with a full relational DB functionality as required by
operation and maintenance of several component of the
IV. ALIGNMENT SYSTEM, OFFLINE SOFTWARE AND
In order to fully exploit the intrinsic resolution of the tracking
system, the signal wires must be positioned with an accuracy
of 20 micrometers with respect to the chamber reference
system. The precision of the mechanical construction is
monitored with elaborate X-ray techniques to an accuracy of 5
µm. Optical alignment systems have been developed which
allow to measure on-line internal deformations of chambers
under gravity, and the relative alignment of chambers
traversed by the same muon track, to a typical accuracy of 30-
40 µm . For that purpose a three dimensional grid of
alignment devices consists of over more than 10,000 sensors
optically connect groups of chambers together. In case of the
end-cap it is also connected to the alignment bars. These bars
which are up to 9.6 m long are nested in each layer of
chambers. They are self-aligning units measuring their own
positions and therefore can be considered precision rulers.
The performance of the barrel and the end-cap alignment
system has been validated in the H8 test beam performing
controlled physical movements and rotations of the chambers
or by inducing thermal expansion of the support structure. The
barrel and the end-cap alignment systems have been able to
track changes in muon sagita's to an accuracy of about 15
microns under normal temperature variation and controlled
movements. The system was found stable over a period of
months. Figure 4 demonstrates how the alignment correct
controlled movements as demonstrated with the sagita plotted
for the nominal setup and for controlled movement on the left
and with the alignment correction applied on the right. The
plots clearly show that once the alignment is taken into
account the three distributions perfectly match.
In parallel to some stand alone code written specifically to
analyze the test beam results, the ATHENA environment was
used as the offline monitoring, reconstruction and analysis tool
for the test beam data. Many aspects of the ATLAS offline
environment were used such as data converters reading the
raw data from all the Muon spectrometer technologies (MDT,
CSC, RPC, TGC, MUCTPI ) following the scheme
designed for the ATLAS.
Figure 4 - The two figures show the distribution of the sagita
plotted on the left before alignment and on the right after
alignment. The three different lines in the two plots correspond
to a plot without movement compared with controlled
movements of six and eight mrad. On the left before alignment
and on the right after alignment is applied.
Offline software was used for the three stages: second level
trigger, event filter and the offline analysis code. Similar
strategy for detector description and data analysis were
successfully implemented in the test beam as in the preparation
for the ATLAS data taking stage (the Data Challenge studies).
The combined test beam enabled combined tests of the Muon
spectrometer with other parts of ATLAS as well as among the
Muon sub-detectors. In Figure 5 depicted as an example the
correlation between projection in the horizontal plane of tracks
in the Muon system and the inner detector. A good agreement
1 0.99 0.08
) is demonstrated in the plot while the
shift of 80 mm resulted from a known shift from the original
position of that size.
Figure 5 - The correlation between the Muon system tracking
and the tracking in the inner detector (Pixels and TRT
The simulation of muons interacting with the H8 chambers and
the material has been performed with Geant4 (version 6) 
running within the ATHENA framework. The Geometry was
taken from an external ASCII database following the same
format that is used for the ATLAS Muon system description.
The comparison of reconstructed real data and simulated
events can provide at this point an important feedback to the
validation of both reconstruction and simulation software. A
comparison between the energy resolutions measured in the
simulation and in the data showed a similar dependence on the
actual test-beam beam energy (or simulated energy in the case
V. TRIGGER CHAMBER PERFORMANCE AND BEAM ENERGY
One of the main goals of this test was to study the trigger and
read-out of the trigger chambers. A performance of the trigger
chamber was studied: efficiency of wires and strips of the
TGC doublets and triplets in low and high Pt cuts, efficiency
of the RPCs as a function of the applied voltage, measurement
of the cluster sizes, and uniformity of the response. Another
goal set for the tests was to study the readout electronics and
the trigger using final electronics and cabling schemes for the
signals going from the FE electronics.
The special bunched muon beam with 25 ns bunch spacing,
emulating the LHC bunch structure, has been used to study the
timing resolution and bunch identification of the trigger
chamber. The ATLAS first level trigger chain  has been
operated with the muon trigger signals for the first time.
The TGC had some upgrades and modification since the
trigger was successfully tested in the 2003 test beam. The
TGC took data during the run with 25 ns bunched beam. It
provided a validation of the design of the end-cap Muon
Level-1 trigger scheme. The ATLAS Central Trigger
Processor (CTP) receives trigger information from the
calorimeter and Muon trigger processors and generates the
“Level-1 Accept” based on a trigger menu. The Muon Trigger
to CTP Interface (MUCTPI) is used to connect the TGC and
RPC information to the CTP. During the 2004 25 ns bunched
beam, TGC Level-1 trigger signals were sent through the
MUCTPI and a comparison between the TGC sector logic
output and the MUCTPI matched perfectly. TGC successfully
tested the integration with the condition DB. The TGC
performed successfully with the results on Level-1 trigger of
99.4% efficiency for low-Pt and 98.1% for high-Pt. Figure 6
demonstrates the correlation between track position as derived
from the MDT tracking and the groups of wires hit in the
TGC. This can be interpreted as a position correlation between
the two. On the second plot a breakdown of the efficiency
given separately for wires and strips, doublets and triplets. All
the values are above 99.5%.
During 2004 test beam RPC electronic was tested with the
final cabling scheme for the signals going from the front-end
to the read-out electronics. The performance of the RPC
chambers was studied: efficiency as a function of the applied
voltage, measurement of cluster sizes and uniformity of
Figure 6 - On the left a correlation between the MDT track
extrapolated to the TGC position and the TGC wire group
number, on the right the efficiency derived for the TGC
Triples wires (TW), Triplet strips (TS) Doublet wires (DW)
and doublet strips (DS).
Figure 7 presents the correlation between the positions as
derived from the MDTs and the one calculated from the RPC
strip hit position. The second plot shows the dependence of the
BML RPC efficiency as a function of the voltage for different
threshold values. One reaches efficiencies about 95% when the
voltage is above 9.4kV.
Figure 7 - On left a correlation between the RPC eta strip
position and the extrapolated position from the the MDT
tracks, on the right the efficiency derived for BML RPC
chamber as a function of the HV supplied to the chamber for
several values of readout threshold.
A large scale test stand of the ATLAS Muon spectrometer has
been operated in the H8 test beam at the CERN SPS. The setup
consisted of a barrel stand reproducing one barrel setup and a
stand reproducing an end-cap sector both fully instrumented
with electronics readout and complete alignment systems. A
muon beam with momenta ranging from 20 to 320 GeV was
used to study and integrate many aspects of the Muon
spectrometer. Special runs with 25 ns bunch spacing were
dedicated for trigger timing resolution and bunch identification
studies. The spectrometer has been extensively tested and
validated with this setup. The studies covered: mechanical
detector integration, integration between the spectrometer
different technologies (MDT, CSC, RPC and TGC), different
subsystems (readout, DCS, alignment) different spectrometer
tasks (trigger, tracking) and different software tools (data
acquisition, databases, high level trigger software, on-line and
offline monitoring and reconstruction, alignment and
calibration). Some of the measurements of quantities such as
track reconstruction, energy resolution, trigger performance,
alignment and comparison with simulation are presented in the
paper. Other than that the test beam has provided the ATLAS
Muon community with a unique opportunity to run a
significant part of the detector in a long scale experiment. It
gave a field to test the on-line and offline simulation and
analysis tools and to have a comparison of the simulation to
real data results.
The work reported here represents a joint effort of many
individuals in the ATLAS Muon collaboration and in the
ATLAS Trigger Data Acquisition community. We would like
to thank them all. A special thank goes to Stefano Rosati for
his assistance in preparing this summary. The support of the
CERN staff operating the SPS and the H8 beam line is
gratefully acknowledged. We thank the funding agencies for
the financial support.
 “ATLAS Muon Spectrometer Technical Design Report”,
 J. Rothberg, “Detector Control System in the LHC
experiment” 9’th Workshop on Electronics for LHC
Experiments, Amsterdam, Netherlands (2003).
 C. Gaspar et al., Internation Conference on Computing in
High Energy and Nuclear Physics, Padova, Italy (2000).
 S. Tarem et al., “Design, Prototype and Testing of the
Detector Control System for the ATLAS Endcap Muon
Trigger”, Colmar 2002, Electronics for LHC experiments,
 Open Process Control, an industrial-wide standard for data
 H. Boterenbrood et al., “The development of the
Embedded Local Monitor Board (ELMB)”, CERN-LHCC-
 International Users’ and Manufacturers group, “Controller
Area Network, CAN in Automation”,.
 P. Adragna et al., “The ATLAS Tile Calorimeter Test
Beam Monitoring Program”, Physics/0405154, (2004).
 ATHENA – The ATLAS Common Framework developers
guide, CERN, February 2004.
 For detailed description see: Jim Bensinger, “ATLAS Muon
Endcap Alignment System”, proceedings of 9th Topical Seminar on
Innovative Particle and Radiation Detectors, May 23-26, Siena, Italy.
Florian Bauer,“Test of the ATLAS alignment system performances at
the H8 test beam facility”, proceedings of IEEE NSS 2004, Rome
 N. Ellis et al., “The ATLAS Level-1 Muon to Central
Trigger Processor Interface (MUCTPI)”, 8’th Workshop on
Electronics for LHC experiment, Colmar, France (2002).
 A. Agostinelli et al., “Geant4 – a simulation toolkit ”, Nuncl.
Inst. Meth. A 506 (2003), 250-303.
 The ATLAS Level-1 Trigger group, “ATLAS Level-1
Trigger Technical design Report” CERN/LHCC/98-14 (1998).