The ATLAS Liquid Argon Calorimeter:
Construction, Integration, Commissioning
Ph. Schwemling on behalf of the ATLAS LAr Group
• Introduction
• Construction,
Integration and
commissioning on the
surface
• Installation and
commissioning after
installation in the
cavern
• Selected performance
results from test-beams
• Some results from
cosmics data analysis
The ATLAS Detector
The ATLAS Liquid Argon Calorimeters
4.5 m
46 m
• LAr Calorimeters:
–
–
–
–
13 m
endcap A
EM Barrel : (||<1.475) [Pb-LAr]
EM End-caps : 1.4<||<3.2 [Pb-LAr]
Hadronic End-cap: 1.5<||<3.2 [Cu-LAr]
Forward Calorimeter: 3.2<||<4.9 [Cu,W-LAr]
• ~190K readout channels
barrel
endcap C
The ATLAS Electromagnetic (EM)
Calorimeter
tdrift =450 ns
•
•
•
E
2 wheels (16 modules) in the barrel
and 1 wheel (8 modules) per endcap
Accordion shape in EM barrel and end
cap calorimeters (>22X0)
Main advantages
–
–
–
–
–
–
LAr as act. material inherently linear
Inherently radiation hard
Hermetic coverage (no cracks)
Longitudinal segmentation
High granularity (Cu etching)
Fast readout possible
The ATLAS Forward and Hadronic
Calorimeters
FCAL End View
• Forward Calorimeter (FCal)
– 3 wheels per endcap (10)
• Cu matrix for the first wheel (2.6, 28X0)
• W matrix for the other two wheels (2x 3.7)
• Hadronic Endcap Calorimeter (HEC)
– 2 wheels per endcap (10), 32 modules each
• Cu absorbers (25mm/50mm thick)
• Each gap consists of 4 sub gaps of 1.85mm
LAr gap FCAl 1/2/3: 250/375/500 µm
HEC module
FCAL1 Anode Structure
Physics requirements
•
•
•
•
•
•
•
•
•
•
•
Discovery potential of Higgs determines most of the
performance requirements for the EM calorimetry
Largest possible acceptance
Large dynamic range : 20 MeV…2TeV ( 3 gains, 12bits)
Energy resolution (e±γ): E/E ~ 10%/√E  0.7%
–  precise mechanics & electronics calibration (<0.25%)…
Linearity : 0.1 % (W-mass precision measurement)
–  presampler (correct for dead material), layer
weighting, electronics calibration
Particle id: e±-jets , γ/π0 (>3 for 50 GeV pt)
–  fine granularity
Position and angular measurements: 50 mrad/√E
–  Fine strips, lateral/longitudinal segmentation
Hadronic – Et miss (for SUSY)
– Almost full 4π acceptance (η<4.9)
Jet resolution
– E/E ~ 50%/√E  3% η<3,
– E/E ~ 100%/√E  10% 3<η<5
Speed of response (signal peaking time ~40ns) to suppress
pile-up
Non-compensating calorimeter  granularity and
longitudinal segmentation very important to apply software
weighting techniques
Jets
Jets,
Missing Et
Electrons Photons
Construction
HV and electrical tests
after module assembly
Repair of abnormal
channels
Mechanical measurements
during production/assembly:
•Stability at 1% level
•Resulting constant term 0.5%
Integration
•
•
•
•
•
Module construction in the labs
Delivery of 3 cryostats and preparations of
cryostats at CERN starts in 2002
Module production in institutes and delivery to
CERN from 2001 – 2004
Wheel assembly in clean rooms from 2002 – 2004
Wheel insertion into cryostats from 2003 – 2004
Cold tests on the surface in 2004 – 2005
Barrel wheel inside cryostat
EM EndCap A wheel after its insertion, cabling finished, July 2004
Cold Commissioning on the
Surface
• All three cryostats were commissioned at LAr temperatures
(88K) filled with LAr on the surface (cold commissioning)
• The goal of the cold commissioning was to :
•
•
•
•
– Check the connectivity and integrity of the connections and all
the detector channels and calibration lines at LAr temperatures
– Test the integrity of all HV channels at nominal voltage at LAr
temperatures
– Measure electronics parameters needed for signal
reconstruction
– Measure the noise and coherent noise (with ATLAS electronics
boards)
The barrel commissioning lasted 10 weeks in summer 2004
The end cap C commissioning lasted 8 weeks in winter 2005
The end cap A commissioning lasted 6 weeks in summer 2005
Number of channels to be tested on all three subdetectors:
– 190304 read out channels
– 14592 calibration lines
– 4248 HV channels
Cold Commissioning – Tests
•
Applying High Voltage
•
HV continuity tests (EM and FCal only):
•
•
•
–
–
Slow ramp up (measure current draw)
Stability test during several weeks
–
–
AC signal applied on the HV line
Via the detector capacity a signal is induced on
the signal cables
Checks connectivity of the HV and signal lines
–
Pulsing all lines with the
calibration board, and
reading pulses back with
the front end boards
Reflectometry
measurements
LC, Rcal measurements (EM)
–
–
•
Measure these parameters
at cold
Used in the calibration run
analysis
Tests of the final FrontEnd-Crate electronics
–
Noise measurements
0.1-0.2 % amplitude
stability over one
month
Few (<1‰)
abnormal
channels
Rcal
Calib. signal
To FE elec.
Cold Commissioning
Test Summary
>»
99.7%
99.9%
ofof
the
detector
channels work!
Correction
needed
(HV#|acce
ptance%)
Dead
(HV#|acce
ptance%)
EMEC C
25 | 5.00
0 | 0.00
3 | 0.11
HEC C
12 | 7.50
0 | 0.00
0 | 0.00
0 | 0.00
FCal C
8 | 1.80
0 | 0.00
20 | 0.06
4 | 0.16
8 | 0.03
EMEC A
35 | 8.75
1 | 0.25
HEC A
0 | 0.00
1/3 | 0.8
3 | 0.11
HEC A
13 | 8.10
0 | 0.00
FCal A
9 | 0.63
0 | 0.00
0 | 0.00
FCal A
11 | 4.19
0 | 0.00
49 | 0.04
1 | 0.01
31 | 0.03
EM Barrel
8.5 | 1.90
0 | 0.00
Bad
channels
(#|%)
Bad
calibration
lines
(#|acc.%)
Dead
channels
(#|%)
40 | 0.13
1 | 0.04
6 | 0.02
HEC C
3 | 0.11
3 | 0.37
FCal C
10 | 0.70
EMEC A
Signal
&
Calibration
EMEC C
EM Barrel
HV
Situation in April 2006
•Since then, some of the HV problems repaired by « burning » shorts
•ECC had to be emptied. When refilling, changes may appear
•Final list of problematic channels being prepared now.
Commissioning
After Installation
As soon as the front end boards are installed in the
cavern and connected to the ATLAS read out system
regular calibration runs are recorded
ADC
–
–
(ADC-linear fit)/600
DAC counts
Linearity of one
calibration channel
DAC counts
Injection of a calibration pulse (via the Rcal) on the
detector module inside the cryostat, and reading it back
with the whole read-out chain
Allows to check the integrity of all connections, the
status of the detector cells and the functioning of the
read out chain and the calibration system Time (ns)
Amplitude (ADC counts)
•
Ramp run pulsing
3 barrel modules
Calibration pulse shapes for
some FEB (one plot per front
end board, each 128
channel)
Gaining practical experience :
cryogenics
Small temperature variations
understood by cryo team.
Gaining practical experience :
Read-out stability
• Comparison of 3 runs (1 per month)
• Observe excellent stability of the maximum
amplitude (calibration runs) and the baseline
– Maximum amplitude stable to 0.14%
– Pedestal variations below 0.05 ADC counts (<0.1MeV)
Production testbeam
•Located at CERN, H8 (barrel) and H6 (end-cap)
beam-lines.
Electron beams from 10 to 300 GeV
•4 Barrel modules (out of 32 total),
3 End-Cap modules (out of 16 total)
tested between 2000 and 2002.
•Topics studied
•Position resolution, γ/π0 separation
•Response to muons
•Time resolution
•Uniformity, linearity, energy resolution
Calorimeter response to muons
Middle
Strips
Time resolution of the
LArg calorimeter
E beam
Linearity for electrons
Erec=λ(off+w0E0+E1+E2+w3E3)
•4 parameters,  dependant
•Simple method
χ
2=Σ(Ei
2i
rec
-Ei
2
true)
Resolution for electrons
End-cap combined testbeam 2004
Understand high  (2.5-4.9) region
Cosmic ray data
Cosmics commissioning goals
• Exercise Front-End
electronics and readout
chain
• Look for dead/bad cells
• Check/adjust detector
timing at ns level
• Check cell response
uniformity at 1-2% level
Commissioning with cosmics
A
4 SD
1
2
(1  4)
3
8 trigger cables, more
if LAr technicians
4
4 SD
B
Typically 15 k events over
one week
2% events projective
S/B about 10 in middle sampling
Individual cell
signal
Projective muons
energy deposit
(middle sampling)
Predicted
physics
pulses
vs
data
Predicted physic pulses vs data
25 samples
 Typical pulses with GeV energy
deposit (same as for 5 samples)
FT7; Slot3; Ch72
Run 24606
E = 3.8 GeV
Front, η ~ 0.23
 Fair qualitative agreement in
the peak and in the undershoot for
every layer
FT1; Slot11; Ch5
Run 24606
E = 17.0 GeV
Middle, η ~ - 0.04
FT1; Slot9; Ch64
Run 24606
E = 6.3 GeV
Back, η ~ - 0.03
Correlating LAr Time and Tile Time
Correlating LAr time and Tile time
• Tile time is the time at the y = 0
crossing. A time-of-flight to the
LAr cell correction is made.
• Slope consistent with 1 to ~5%
with intercept values ~ 2ns. After
fixing slope, intercept decreased
to just under ~ 1ns.
FEB: Side A (1), FT 23, Slot 12
Timing resolution with cosmics
• Spread of data about linear fit in
previous slide represents
combined LAr-Tile resolution.
Width of
TLAr - TTile distribution quantifies
the resolution.
•Resolution goes as the inverse of
the energy - separate into bins of
LAr cell energy.
• Fit to quadrature sum of E-1 term
and constant term. Res agrees
with test beam, Const
substantially larger (TB, 0.65 ns).
 to be understood
Muon response dependance over 
•Plot MPV of muon Landau
distribution as a function of
.
•Shape of various clustering
methods agree well with MC
and the expected cell depth
at the ~2% level in 0.1 
bins. The results use ~ 10k
clusters.
• A naïve statistical
extrapolation suggests
~160k clusters necessary to
probe depth in cell bins at
1% level.
* All normalized to 0.3 <  < 0.4 bin
Conclusions
All LAr calorimeters successfully integrated in their cryostats
and in ATLAS, after extensive test and quality control
during construction.
Excellent condition of the calorimeter (more than 99.9% of
channels work), stability very good.
Cosmic muons are routinely recorded together with the
hadronic Tile calorimeter, muon detector and ID system.
Data used to improve understanding of the detector and
prepare real data taking
The ATLAS calorimeter system will be ready in spring 2008
for the first physics data, expected in the second half of
2008 !
Spares
Calorimeters