Upload
others
View
1
Download
0
Embed Size (px)
Citation preview
1
presented by Bert Koene
October 3, 2001 /1
2
Topics
Introduction
Tracker layout
Straw tube drift cells
Detector modules
Performance
Stations
Electronics
HV, gas, alignment
Project organization
Summary
October 3, 2001 /2
3
INTRODUCTION
Momentum spectrometer T2 – T9Track slopes in RICH T1 - T2
T6 - T9Link vertex⇐⇒ CAL/muon
October 3, 2001 /3
4
Tracker = Inner Tracker + Outer Trackerfine grained coarse grained
surface 2 % 98 %
October 3, 2001 /4
5
Boundary IT - OT:cell occupancy←− track density ⊗ granularity
x position (cm)
y po
sitio
n (c
m)
October 3, 2001 /5
6
TRACKER LAYOUT
Detection elements5 mm diameter straw tubes packed in double-layeredmodules
October 3, 2001 /6
7
Module cross section
October 3, 2001 /7
8
Station = 4 planes of modules arranged as X U V Xviews
active area
station zmax xmax ymax channels
(mm) (mm) (mm)
2 2275 704 581 4.3 k
3 3635 1125 929 6.8 k
4 6035 1867 1541 11.4 k
5 7035 2176 1796 13.1 k
6 8038 2486 2052 15.1 k
7 8497 2628 2170 16.0 k
8 8956 2770 2287 16.8 k
9 9415 2912 2404 17.5 k
101.0 k
Stereo angle ±5o optimised for
εseeding ⊗ εfollowing⊗ RICH slopes
October 3, 2001 /8
9
z (cm)0 200 400 600 800 1000 1200
B (
Tes
la)
-1
-0.8
-0.6
-0.4
-0.2
0
yB
z and BxB
=0yθ=xθ
R1← following → seeds R2
October 3, 2001 /9
10
Economised since Techn. Proposal:
• No station behind RICH 2
• No Y planes at RICHes
• One magnet station less
Without T11, extended RICH
With T11
Momentum (GeV)
π -K
sep
arat
ion
(si
gma)
π-K separation for true pions
October 3, 2001 /10
11
STRAW TUBES
Cell size limited by:occupancysignal collection time
clean electrostatics (tube)fast drift gas
Tmax < 25 ns idealTmax < 50 ns realistic
consequence: time overlap with hits fromneighbouring BX
Detector sample
25ns 25ns 25ns 25ns 25ns 25ns
October 3, 2001 /11
12
Use Ar / CF4 / CO2 for fast driftLarge electron capture c.s. of CF4 at few eV:
• degrades drift resolution insignificant
• breakup to free fluorine radicals
Cathode surface of noble metal or carbon loaded polymer(ATLAS, HERA-B, COMPASS)
Technical Proposal: “drift cells of straw tube geometry builtwith honeycomb chamber technique”
HC proto’s of C-doped polycarbonate foil (Pokalon)Switched to real straw tubes of C-doped Kapton because:
1. Pokalon-C very expensive
2. doubts about Pokalon-C ‘aging’ in Ar / CF4 / X
3. slow drift in HC cell corners
4. better precision of cell shape
Channel density increase: 5 + 1 mm HC cell pitch5.25 mm tube pitch
Less planes =⇒ stayed at 100 k channels
October 3, 2001 /12
13
Inner layer = 40 µm Kapton XC with 30 % volume doping
Tested in Ar / CF4 / CO2 to1.2 C/cm (realistic: p beam)
several C/cm (extreme: glow discharge)
Estimate hottest OT regions: 0.25 C/cm per year
Module aging tests in HERA-B and with X-ray source
realism accelerated
October 3, 2001 /13
14
Outer layer = 25 µm Al⇐= X-talk suppression
Tested 1. Kapton XC2. aluminium3. pure Kapton4. aluminised Mylar
1 & 2 cathode to HV GND via outer layer
October 3, 2001 /14
15
X talk by amplitude:0.5 % aluminium1.8 % Kapton XC
stays worse with shielding foils
Pair of Al-XC tubes acts as 1/4λ resonator if onlysees GND at preampL = O (2 m) =⇒ ν = O (40 MHz), inside preampbandwidth
Remedy: tube grounding at short intervals
October 3, 2001 /15
16
STRAW TUBE MODULES
64 cells widesandwich plates of HC core + 100 µm carbon facings + 20µm Al foil
no module overlap =⇒ loose 1/2 cell at border
October 3, 2001 /16
17
drift gas flows through and around the tubes
gas distribution into tubes
October 3, 2001 /17
18
Electrical split near y = 0 to limit cell occupancy andsignal propagation time
Wire centering supports every 80 cm
October 3, 2001 /18
19
midway splitterboost far-end amplitude: no terminating impedance
October 3, 2001 /19
20
signal round trip ≤ 20 ns ' shaping time
October 3, 2001 /20
21
10 mm long wire support =⇒ 12 mm insensitiveregion
October 3, 2001 /21
22
assembly of end piece and front end boards
October 3, 2001 /22
23
MATERIAL BUDGET
Module elements: 0.67 % X0
2 tube layers 0.276 %4 x 0.1 mm facing 0.14 %2 x 10 mm HC core 0.11 %2 x 20 µm Al foil 0.045 %gluings (est.) 0.1 %
Break down for tube layers:
aluminium 0.165 %glue 0.02 %Kapton XC 0.091 %
Add localised materials =⇒ 0.75 % per module4 x 0.75 % = 3.0 % X0 per stationSum for Outer Tracker:
24 % X0
10 % λI
October 3, 2001 /23
24
TEST BEAM RESULTS
October 3, 2001 /24
25
plateau curves and cell resolution at different CF4
concentrations
October 3, 2001 /25
26
Tmax vs B field: 15 % CF4 is enough
October 3, 2001 /26
27
smallish effect of CF4 concentration on resolution
October 3, 2001 /27
28
homogeneous response along 3.3 metre long and fullyrealistic prototype
October 3, 2001 /28
29
PERFORMANCE STUDIES
OT configuration optimization
Track reconstruction on simulated bb̄ events + min.bias bkg events
pp interactions at√
s = 14 TeV Pythiaparticle decays QQ packagefollow through LHCb GEANT based SICBOuter Tracker response GAUDI
tuned to test data, incl. dead time etc.track reconstruction TRAIL package
October 3, 2001 /29
30
Background mixed into bb̄ events :
PILE UP: min. bias in same BX (average 0.53)
SPILL OVER: hits due to different BX
1. overlap of 50 ns OT time windows per BX
2. long-lived curling tracks
October 3, 2001 /30
31
quiet Bd → π+π− event
October 3, 2001 /31
32
busy Bd → π+π− event
October 3, 2001 /32
33
OCCUPANCY
Def: fraction of channels hit within the 50 ns eventwindowimportance:
• pattern recognition efficiency & ghost raterequires mean occupancy
seeding ≤ 10 %
other ≤ 15 %
• f.e. dead time ' 35 ns / hitoccupancy =⇒ inefficiency
0 2000 4000 6000 8000 10000
100
95
90
85
80
Hit multiplicity in OT
OT
see
ding
eff
icie
ncy
(%
)
seeding eff. vs number of hits in T6 – T9
October 3, 2001 /33
34
Tuning of IT - OT boundary
x position (mm)
aver
. occ
up. (
%)
average occupancy vs cell position in T7
October 3, 2001 /34
35
occupancy
# E
vent
s
event-to-event occupancy fluctuation in hottestregion of T3
October 3, 2001 /35
36
Station
Occ
upan
cy
TO
P
CO
RN
ER
SID
E
Average occupancies in the hottest regions of each stationat L = 2× 10
32cm−2s−1
October 3, 2001 /36
37
Track reconstructionOT ⊕ IT
Upstream tracking
• find track seeds in T6 – T9
• assume origin in vertex⇒ δp/p to O(2 %) from pt kick
• follow into the gently rising B field
Fast variant for L2 trigger
• match track seed with VELO track, p from kink
δp/p = 1.3 %matching eff. = 92 % for p > 5 GeV/c
Downstream tracking
• follow VELO track into the magnet (T4)δp/p ' 0.6 %
ε = 80 % – 90 % depending on p
October 3, 2001 /37
38
Upstream tracking
0
1
2
3
4
5
6
7
8
0 20 40 60 80 100 120 140 160 180 200
p (GeV)
δp/p
(pe
r m
ille)
m.s. term (constant) and resolution term (∝ p) areequal at 100 GeV/c
< δp/p > = 0.39 % for p > 1 GeV/c
October 3, 2001 /38
39
Upstream tracking efficiencies
0 20 40 60 80 100
100
95
90
85
80
Momentum (GeV/c)
Seed
ing
effi
cien
cy (
%)
seeding
momentum(GeV/c)0 20 40 60 80 100
Fo
llow
ing
eff
icie
ncy
(%)
80
85
90
95
100
following
requiring > 70 % purity and efficiency of hit assignment
October 3, 2001 /39
40
momentum (GeV/c)0 20 40 60 80 100
com
bin
ed e
ffic
ien
cy (
%)
80
82
84
86
88
90
92
94
96
98
100
October 3, 2001 /40
41
Stereo angle
track following ≤ 5o
track seeding ≤ 10o
RICH slopes not too small, 5o still OK(⇒ 0.09 mrad on trackvs 0.58 mrad per photon)
October 3, 2001 /41
42
Stereo angle (degrees)
Seed
ing
effi
cien
cy (
%)
Processing time (seconds/event)
seeding eff. vs stereo angle
Stereo angle(degrees)0 2 4 6 8 10 12 14 16
Fo
llow
ing
eff
icie
ncy
(%)
90
95
100
Stereo angle(degrees)0 2 4 6 8 10 12 14 16
Gh
ost
rat
e(%
)
0.1
0.15
0.2
following eff. and ghost rate vs stereo angle
October 3, 2001 /42
43
Physics performance illustration
Tracking efficiency
Bd → ππ 94 % (⇐ 97 % per track)
Bs → Ds(KKπ)K 80 % (⇐ 95 % per track)
Mass resolution
(GeV)0Bm
5.2 5.40
10
20
30
40
50 σ =21.8 MeV
(GeV)sBm
5.3 5.35 5.4 5.450
10
20
30
40
50 σ =10.7 MeV
(GeV)sDm
1.94 1.96 1.98 20
10
20
30
40
50 σ =4.3 MeV
Bd → ππ Bs → DsK Ds → KKπ
October 3, 2001 /43
44
comment to mass resolution
Tracker =⇒ track momentaVELO =⇒ track angles
δM/M ∝ δp/p and ∝ δα/α
B → ππ hard tracks, large opening anglemomentum error dominates
Ds → KKπ soft tracks, small opening angleerrors on momenta and angles contributeequally
cp. Techn. Proposal:assumed δp/p = 0.3 %δM(Bs → DsK): unchangedδM(Bd → ππ): 17 MeV/c2 × (0.39 % / 0.3 %) = 22MeV/c2
October 3, 2001 /44
45
STATION DESIGN
seeding stations T6 – T9 are identical
October 3, 2001 /45
46
L/R station halvesmodules mounted on top/bottom precision bars
October 3, 2001 /46
47
10 cm between pole faces and ± 250 mradacceptance limit
October 3, 2001 /47
48
3734
3082
3924
340 170
200
581
(end of active region)
so cut sawtooth of stereo planes away
October 3, 2001 /48
49
T3 – T5 mounted on rails fixed to magnet yokeL/R halves separate when moving out
October 3, 2001 /49
50
ELECTRONICS
32x32x
2x
TTCrxECSslave
Powerregulator
L0 derandomizerL0 derandomizer
level 0bufferlevel 0buffer
serviceboards
~ 40
3250front−endboards
815 signaldistributionboards
102level 1boards
104000 channels
815 optical links
204 optical links
DLLDLL
DACsDACs
triggerlogic
triggerlogic
ECSslave
L1 Yesclock
I2CJTAG
level 1 logic &header generator
level 1 buffer(FIFO)
TTCrx
output FIFO
L1 derandomizer
straw tube signals
clock
L0−yes
8 I2C
ASDBLR
4xV threshold
32channels
LH
Cb
cav
ern
fine timefine time
zero−suppressor32:1 multiplexor
Co
un
tin
g r
oo
m
Front−End links to DAQ
Multiplexor 4:1
GOL
OTISTDC
Power
TTCECS
October 3, 2001 /50
51
Front end:
• Hit rates locally up to 4 MHz atL = 5× 1032cm−2s−1
• radiation dose O(10 krad/yr) =⇒ rad. tolerant /hard
Preamp: 8-channel ASDBLR chip
DMILL→ radiation hardpeaking time 8 nsshaping time 20 ns
October 3, 2001 /51
52
Custom designed TDCOTIS chip, ASIC lab Heidelberg
• rad hard 0.25 µm CMOS
• clock driven (via TTC system from LHC clock)
• DLL adds 5 bits fine time
• 32 channels / chip
• each channel has its own clock-synchronous L0buffer
• on L0 accept two buckets to derandomizer
OTIS development steps
submits of TDC core and memory tested July 2001first submit of full chip April 2002final evaluation May 2003
October 3, 2001 /52
53
fall-back solution: already existing HPTDC chip
• 32 channels share 1 L0 buffer⇒ in hot regionsuse only 16 channels/chip
• limited radiation tolerance⇒ not on f.e. boards(racks near detector)
• FPGAs convert data to OTIS format⇒ same L1electronics
Can do the job at added cost of crates and cabling
October 3, 2001 /53
54
HV DISTRIBUTION
2 × 64 ch per HV branch
October 3, 2001 /54
55
October 3, 2001 /55
56
Drift gas system
recirculating⇐ CF4 cost
90 % regeneration per cycle
flow rate O(1 vol/ 2 h)
October 3, 2001 /56
57
Hardware alignment monitoring(if it fits in)
RASNIK system logs movements within station tripletsMASK (1) → LENS (2) → CCD CAMERA (3)
October 3, 2001 /57
58
PROJECT ORGANISATION
Milestone Date
Mechanics
Engineering design of modules completed 12/2001
Engineering design of frames completed 6/2002
Start of module production 6/2002
Start of frame production 3/2003
All frame parts at IP8; start of station assembly 6/2004
Last modules to IP8 4/2005
Last station installed 6/2005
Electronics
TDC choice 5/2003
Start of electronics production 6/2003
All electronics produced and tested 3/2005
Project
Commissioning completed 3/2006
October 3, 2001 /58
59
October 3, 2001 /59
60
Task Institutes
Mechanics
T2 – T5 Heidelberg, Krakow
T6 – T9 Beijing, NIKHEF, Warsaw
Electronics
Before TDC NIKHEF
From TDC to DAQ
Baseline design Dresden, Heidelberg, NIKHEF
Fall-back design Krakow, NIKHEF
Production & testing Beijing, Dresden,
Heidelberg, NIKHEF
Services
Drift gas CERN, Krakow
High voltage CERN, NIKHEF
Alignment Warsaw
Controls NIKHEF
October 3, 2001 /60
61
Comments
620 modules includes 10 % spares
assembly of large stations at CERN
tracking software and performance studies:common for OT + IT by Tracking Task Force
October 3, 2001 /61
62
PROJECT COST
detector modules 2720 kCHFelectronics 5570frames and installation 560services 450
TOTAL 9300 kCHF
October 3, 2001 /62
63
SUMMARY
• Module design: ready and proven
• Known how to assemble into stations
• Electronics is designed; TDC still underdevelopment, fall-back is on the shelf
• Tracker layout simplified: two stations less thanin Technical Proposal
• Track reconstruction efficiency through entirespectrometer ≥ 90 %
• Momentum resolution 0.39 %(T.P. assumed 0.3 %)
• Outer Tracker represents 0.24 X0
(T.P. idealized design 0.18 X0)
• Resources of people, time and money match thepresent project size
Further LHCb optimization: trend to ‘less is evenbetter’May drop another station, but no effect on technicaldesign
October 3, 2001 /63