Quality Control on the optical response

1
Quality Control on the optical response for the
ATLAS hadronic calorimeter1
Josep Flix Molina Universitat Autònoma de
Barcelona Departament de Física Institut de
Física d'Altes Energies Edifici Cn E-08193
Bellaterra (Barcelona) September 2000
1 Treball de Recerca de Tercer Cicle.
2
1
Contents.
?Introduction
The ATLAS detector for the LHC.
The ATLAS Tile Calorimeter.
Calibration of the Tile Calorimeter.
Assembly of the calorimeter at IFAE.
?Checking the calorimeter response
Optical Quality control system.
Results on the optical Quality Control for the
modules constructed
? How to identify and repair optical faults. ?
Example of a fiber fault.
Criteria for overall evaluation of modules.
A special problem variations of scintillator
light yield.
Conclusions.
3
2
The Large Hadron Collider (LHC).
LHC A pp collider to be operative in mid 2005
(CERN).
High-energy ECM 14 TeV High Luminosity
1034 cm-2s-1
Try to answer open questions about fundamental
physics!
Higgs particles?
SUSY particles?
etc...
CP violation?
New physics beyond LEP?
4
3
The ATLAS detector.
42 m
22 m
- The bulk of the hadronic calorimeter -
Extended Barrel modules
Barrel modules
64 modules
5
4
The ATLAS hadronic calorimeter TileCal.
Detection principle
Sampling calorimeter
Scintillators Active material. Iron Absorber
material.
Light yield by scintillators (tiles)
WLS fibers
PMTs
Front End electronics
Scintillators transparent granulated
polystyrene is used as the scintillation
matrix base. Dopants induce a
sequence of light absorptions
and emissions
PTP
POPOP
240-300 nm 320-400 nm
420nm
(Ionizing particles)
Fibers A pair of WLS optical fibers (? 1 mm)
collects light from the tiles at
both ? edges
WLS
420nm 480 nm
6
5
The ATLAS hadronic calorimeter TileCal.
The basic constructive element The period.
11 rows of tiles
Tile (3mm)
Spacer (4mm)
Master (5 mm)
Fiber routing Thin extruded plastic tubes, each
of which carries the
WLS fibers, are used to keep the fibers in
contact with tiles.
To PMTs
By this way, cells that point to the origin of
particles can be defined
7
6
?? x ?? 0.2 x 0.1
?? x ?? 0.1 x 0.1
The ATLAS hadronic calorimeter TileCal.
Example an Extended Barrel module
Geometry 3-D segmentation formed by
scintillator cell structures
3 longitudinal sub-divisions
8
7
Intercalibration and monitoring of the
calorimeter response with a radioactive source.
The concept The calorimeter is designed to
allow a Cs137 source to pass
through every scintillator.
Check the response quality and uniformity.
Equalize cell responses by adjusting the
PMT HVs.
Monitor in time the average current of each
cell during the detector lifetime.
Hidraulically system.
5 mCi Cs137 source.
E? 0.662 MeV.
? mean free path 20 mm.
9
8
Assembly of the TileCal hadronic calorimeter.
10
9
Assembly of the Extended Barrel modules at IFAE.
Mechanical assembly
8 standard sub-modules 1 slightly smaller
special-geometry sub-module (ITC)
Sub-modules assembled at IFAE and IFIC
(Valencia).
ITC assembled at the Univ. of Texas, in
Arlington.
Instrumentation
Insert the Tiles (1600).
Insert the profiles (600) which contain
the fibers (1750).
Sort fibers into bundles to be routed to
the PMTs.
Gluing the fibers into a permanent assembly.
Cutting and polishing the glued bundles.
- A very good team of technicians do all these
tasks! -
11
10
Quality Controls during module instrumentation.
QC0 ? Module preparation.
QC1 ? Tile Inspection.
QC2 ? Verify fiber lengths and sorting.
QC3 ? Verify routing with a light source.
QC4 ? Check for damaged fibers.
QC5 ? Inspect the fiber ends.
QC6 ? Optical QC test and certification.
? Repairing the optical faults. The intrinsic
non-uniformities of the calorimeter response
must not produce substantial effects in the
detected physics.
12
11
The optical Quality Control of a module.
Why optical QC and not source calibration?
? Modules cannot be equipped with its own
electronics (available in 2001).
? Modules have to be shipped ? possibility of
changes in response.
Final calibration can only be done from 2001
onwards, at CERN.
Using a blue LED to inspect the quality of a
module.
? The Blue Light Source simulates the blue light
caused by scintillator.
? A comparison with Cs137 showed good correlation
(rms value 7).
(No scintillation in Tiles)
? Advantages
1) Safer than the radioactive source. 2) It does
not require assembly activities to be
suspended (to avoid radiation exposure).
? Disadvantages
1) Does not explore the scintillation processes
in tiles.
Cs137 _at_ CERN ANL
Blue LED _at_ IFAE MSU
13
12
The environment.
14
13
QC6 Experimental Setup at IFAE.
The light Source
3 mm Window
Black Ink
Plastic Cylinder filled with Epoxy
Blue LED
? The light source consists of a Blue LED
(NSPB-310A from NICHIA) and a diffuser
mounted on the top of the LED. ?
Azimuthal uniformity of the emitted light ? (2.1
? 0.5) ? ?peak 470 nm ? No
Scintillation processes in tiles (only in WLS
fibers).
Scintillator
Spacer
Master
15
14
QC6 Experimental Setup at IFAE.
The Extender an automatic tile-scanning
machine.
? The machine responsible for the horizontal
motion of the LED into module and the
vertical motion between rows of tiles is called
Extender.
DC Motor
? Good Velocity stability ? ?VxLED/VxLED
(0.36 ? 0.07).
16
15
17
16
How Do We Flag Optical faults.
1? Make a complete Scan for the constructed
module.
2?To look for optical faults we defined a Drop

(1- Ij/ltIgt)100 Dr ()
3?We flag all cases with drops gt25 . All of them
are investigated and the optical faults are
repaired, that is bringing drops to values lt25.
4?Go to 1?, until the faults have been fully
repaired.
Ij
ltIgt
Repeatability measurements
? Repeatability for Drop Variable (0.6 ? 0.2)
18
17
How Do We Flag Optical faults.
The generated Faults list
?
? ? ? ?
This automatic diagnosis is based on a few
simple considerations
? - similar drops for tiles read by the same
fiber indicate a fiber fault ? - drops
correlated on left-right ITRs responses indicate
a possible problem in the corresponding
tiles ? - uncorrelated drops are related to bad
tile-fiber couplings.
These problems are investigated and repaired to
bring all drops to be lt25 (?5?).
19
18
Overall statistics.
We have already built 32 of our Extended
Barrel modules (21 modules).
? For the first 18 Extended Barrel modules
constructed at IFAE, the faults found were
related to
Bad tile-fiber couplings ? 70 of faults.
Profiles are not well introduced into the slot
between master plates.
Fiber nested between iron and scintillator.
Fiber not well centered.
Fiber faults ? 26 of faults.
Fiber does not reach the bundle end.
Short fibers.
Fiber damage.
Routing errors.
Rough fiber ends.
Scintillator faults ? 4 of faults.
Disuniformity of scintillator quality.
Scintillator Tyvek envelope displaced.
Chipped scintillators.
20
Fiber fault an example.
19
Fiber does not reach the bundle end
Description The curvature of fiber bundles
near the aspirin tube may make fibers slip away
from the end of the tube. After they are glued,
that results in fiber not reaching the end of
the bundle in this case, there is a substantial
loss of light transmission to the PMT.
Aspirin tube
fibers
Solution A thin (1.5 mm ?) teflon tube is
glued into a bundle. After the glue cures, the
teflon tube is removed, leaving a hole through
which a new fiber can be glued. After gluing and
polishing the new fibers, the drops are
completely recovered.
Polished fiber ends
A very good technician Idea!
Extract the Teflon tube
All Collaboration
Cut the fiber
New fiber re-polish bundle
Teflon tube
21
20
Criteria for overall evaluation of modules.
How the intrinsic non-uniformities of modules
could produce substantial effects in the detected
physics?
Overall module uniformity Monitoring the
quality of materials and work.
22
21
Criteria for overall evaluation of modules.
Quality of the 1rst and 2nd longitudinal
sub-divisions
? Repercussion of the disuniformity on the
measurement of the energy of isolated pions
(S. Bravo).
Conclusion A deterioration of the enertgy
resolution is observed when
the rms within a cell is 10. However, 5 is
entirely acceptable.
? The mean value of cell rms for all modules
constructed at IFAE is (4.3 ? 2.0).
?
- The result is compatible with the requirement -
23
22
Criteria for overall evaluation of modules.
Quality of the 3rd longitudinal sub-division
? The most stringent criteria for the uniformity
of the 3rd longitudinal sub-division comes
from muon detection, due to its relevance to be
used for the LVL2 trigger. ? Muon
simulations showed that the individual tile-fiber
pair response could affect muon detection in
the 3rd longitudinal sub-division
(O.Blanch). ? Therefore, the ITR uniformity in
the 3rd longitudinal sub-division needs
particular attention during optical QC.
24
23
Further studies of module uniformity and tile
light yield.
A potential problem
? Non-uniformity of scintillator light yield
(alarm given by ANL)
Sets of 20 scintillators are inserted sequentally
in modules serious source of response variation
within cells, if neighboring subsets are
different.
Tile edge 1
Tile edge 2
First step to learn more about the problem
? The optical quality of a module constructed at
IFAE was tested at CERN with the Cs137
radioactive source.
? Purpose variations in light output of tiles
delivered to IFAE might have
gone unnoticed with the LED-based test set-up.
25
24
Further studies of module uniformity and tile
light yield.
Group of tiles with different light yield
observed
- This module was chosen because the LED spot the
problem in one cell -
Such grouping is present because packs of 20
tiles were sequentially installed into this
module, and others.
26
25
Further studies of module uniformity and tile
light yield.
Tile light yield and estimators
? Two main properties measured for every pack
(Protvino)
I0 light yield I0/I1 transmission
Tile packs which are outside an
acceptance corridor are rejected.
Extreme cases could show differences up to
15-20, for the I0 and/or I0/I1 parameters
? Simple setup at IFAE to compare estimators
of tile light yield
?
- Based on simple assumptions. - Proposed by our
group.
27
26
Further studies of module uniformity and tile
light yield.
Decision adopted to reduce light yield
variations within cells
? For the latest part of 1rst batch of tiles A
time-consuming pack ordering process used at
IFAE (according to ).
? For the 2nd and 3rd batch of tiles
? By making these tiles divisions, for a given
tile size, the maximum variation in light
yield between the two most extreme packs is
(8-10) for CERN (2-4) for IFAE (6-9)
for ANL
This must be compared to the previous 15-20!
? Within each site, tile packs will be inserted
as they come.
28
27
Conclusions.
The procedure to test the calorimeter modules
produced at IFAE has been exposed.
The QC system is reliable enough to test the
optical quality of the constructed modules.
Good knowledge acquired about the origin of the
optical faults.
Good understanding of actions to be done to fix
a specific fault.
In most of the cases the faults are corrected
(drops lt25, over 3000 measurements/module)
In all cases, the instrumented modules are
within the criteria imposed by the physics
requirements.
The sorting of tiles into subsets, according to
, avoid significant local
disuniformities within cells.
29
Calibration of the Tile Calorimeter.
Motivations
Energy calibration factor conversion factor of
the energy released in
the calorimeter and the
digitized signal read.
Minimize variations this factor has to be
measured for every channel
and varations have to be
minimised (they affect
to the constant term of the
resolution).
Controlling the variations during the detector
lifetime.
Intercalibration and monitoring
CIS
Cs137
Combined calibration system - Monitor the
scintillators, fibers, PMTs and readout chain
responses. - Identify the source of any
variation and ensure cell-to-cell uniformity.
The calorimeter response can be monitored with a
precision of about 1
30
Bad Tile-fiber coupling an example.
Fiber nested between iron and scintillator
tile
Description When a profile is introduced into
the slot, the fiber may lose its center position
within the profile. In this case it may get
caught between iron and scintillator, remaining
trapped there.
Solution Take out the profile and put it back
gently pulling the fiber to keep it at the
center of the profile.
Statistics - 10 faults detected - 50 lt
Drop lt 100 - All cases repaired
Master plate
Spacer plate
fiber