Embed Size (px)
Citation preview
CMS Tracking Detector case study This lecture is based on a PG lecture generously provided by Prof G Hall of Imperial College, London Updated in December 2016
Prof Peter R Hobson C.Phys M.Inst.P. Department of Electronic and Computer Engineering Brunel University London, Uxbridge
Development of a tracking detector for physics at
the Large Hadron Collider
Geoff Hall
October 2002 G Hall 3
CMS = Compact Muon Solenoid detector • missing element in current theoretical framework - mass
Total weight 12,500 tons Diameter 15m Length 21.6m Magnetic field 4T
Tracking system 10 million microstrips Diameter 2.6m Length 7m Power ~50kW
14000 t
October 2002 G Hall 4
LHC parameters (CMS)
• Consequences High speed signal processing Signal pile-up High (low) radiation exposure High (low) B field operation Very large data volumes New technologies
pp Pb-PbLuminosity 1034 cm-2.s-1 1027 cm-2.s-1
Annual integrated L 5x1040 cm-2 ?
CM energy 14 TeV 5.5 TeV/ Nσ inelastic ~70mb ~6.5 b
interactions/bunch ~20 0.001tracks/unit rapidity ~140 3000-8000beam diameter 20µm 20µmbunch length 75mm 75mmbeam crossing rate 40MHz 8MHzLevel 1 trigger delay - 3.2µsec - 3.2µsecL1 (average) trigger rate Š100kHz < 8kHz
October 2002 G Hall 5
Design philosophy • Large solenoidal (4T) magnet
iron yoke - returns B field, absorbs particles technically challenging but
smaller detector, p resolution, trigger, cost • Muon detection
high pT lepton signatures for new physics • Electromagnetic calorimeter
high (∆E) resolution, for H => γγ (low mass mode) • Tracking system
momentum measurements of charged particles pattern recognition & efficiency
complex, multi-particle events complement muon & ECAL measurements
improved p measurement (high p) E/p for e/γ identification
October 2002 G Hall 6
Parameters for hadronic collider physics • E, p, cosθ, φ prefer variables which easily Lorentz transform e.g E, pT, pL, φ • pT divergences from simple behaviour could imply new physics
eg heavy particle decay => high pT lepton (or hadron)
• rapidity Lorentz boost
• pseudorapidity
y =12
ln(E + pL
E − pL
) dy =dpL
E
y → ′ y = y +12
ln(1 + β1 − β
)dNdy
invariant =>
d2 Nch arg ed
dη.dpT
≈ H. f ( pT ) H~ 6 |η| < 2.5 LHC
y =12
ln(cos2 (θ 2) +
m2
4 p2 + ...
sin2 (θ 2) + m2
4p2 + ...) ≈ − ln tan(θ 2) ≡ η
October 2002 G Hall 7
Physics requirements (I) • Mass peak - one means of discovery
=> small σ(pT) eg H => ZZ or ZZ* => 4l±
typical pT(µ) ~ 5-50GeV/c
• Background suppression measure lepton charges good geometrical acceptance - 4 leptons background channel t => b => l
require m(l+l-) = mZ ΓZ ~ 2.5GeV precise vertex measurement identify b decays, or reduce fraction in data
m 2 = Ei2
i∑ − pi2
October 2002 G Hall 8
Physics requirements (II)
∆pT
pT
≈ 0.15pT (TeV) ⊕ 0.5%
• p resolution large B and L
• high precision space points
detector with small intrinsic σmeas • well separated particles
good time resolution low occupancy => many channels good pattern recognition
• minimise multiple scattering • minimal bremsstrahlung, photon conversions
material in tracker most precise points close to beam
σ( pT )pT
~ pTσ meas
B.L2 Npts
October 2002 G Hall 9
Silicon diodes as position detectors
+V bias
~25µm
~300
µm
~1pF/cm
~0.1pF/cm
• Spatial measurement precision defined by strip dimensions
ultimately limited by charge diffusion
σ ~ 5-10µm
October 2002 G Hall 10
Vertex detector ~1990
October 2002 G Hall 11
Interactions in CMS
7 TeV p
7 TeV p
October 2002 G Hall 12
2.4m
~ 6m
Microstrip tracker system
~10M detector channels
October 2002 G Hall 13
Event in the tracker
October 2002 G Hall 14
• Constraints on tracker minimal material high spatial precision sensitive detectors requiring low noise readout power dissipation ~50kW in 4T magnetic field radiation hard Budget
• Requirements large number of channels limited energy resolution limited dynamic range
Silicon detector modules
October 2002 G Hall 15
Radiation environment • Particle fluxes
Charged and neutral particles from interactions ~ 1/r2 Neutrons from calorimeter
nuclear backsplash + thermalisation ≈ more uniform gas only E > 100keV damaging
• Dose energy deposit per unit volume Gray = 1Joule/kg = 100rad mostly due to charged particles
October 2002 G Hall 16
Imperial College contributions to Tracker
FED
APV25 APVMUX/PLL
• Hardware development • Hardware construction • Beam tests & studies • Preparation for physics
October 2002 G Hall 17
programmable
gain
APV25 0.25µm CMOS
Low noise charge
preamplifier
50 ns CR-RC shaper
192-cell analogue pipeline
1 of the 128 channels
SF SF
Analogue unity gain inverter
S/H
APSP
128:1 MUX
Differential current O/P
APV25-S0 (Oct 1999)
APV25-S1 (Aug 2000)
Chip Size 7.1 x 8.1 mm
Final
amplifiers pipeline memory
signal processing
& MUX
control logic
October 2002 G Hall 18
0.001
2
3
4
56789
0.01
Noise
[µV/
Hz1/
2 ]
1042 4 6
1052 4 6
1062 4 6
107
Frequency [Hz]
pre-rad 10 Mrad 20 Mrad 30 Mrad 50 Mrad anneal
10-10
10-9
10-8
10-7
10-6
10-5
10-4
10-3
Id [A
]
-0.6 -0.4 -0.2 0.0 0.2Vg [V]
Pre-rad 50 Mrad Anneal
Irradiations of 0.25µm technology • Extensive studies
CMS tracker data from IC, Padova, CERN ALL POSITIVE and well beyond LHC range
• CMOS hard against bulk damage
Qualify chips from wafers with ionising sources
• Typical irradiation conditions 50kV X-ray source Dose rate ~ 0.5Mrad/Hour
to 10, 20, 30 & 50Mrad
PMOS
PMOS 2000/0. 36
400µA
October 2002 G Hall 19
APV25 irradiations (IC & Padova) • IC x-ray source
Normal operational bias during irradiation clocked & triggered also 1 0 MeV linac electrons(80Mrad) and 2. 1 x1 01 4 reactor n. cm- 2
Post irradiation noise change insignificant
10 Mrad pre-rad APV25-S1
October 2002 G Hall 20
CMS Silicon Strip Tracker Front End Driver
VME-FPGA
TTCrx
BE-FPGA Event Builder
Buffers
FPGA Configuration
Power DC-DC
DAQ Interface
12
12
12
12
12
12
12
12
Front-End Modules x 8 Double-sided board
CERN Opto-
Rx Analogue/Digital
96 Tracker Opto Fibres
VME Interface
Xilinx Virtex-II FPGA
Data Rates 9U VME64x Form Factor
Modularity matched Opto Links
Analogue: 96 ADC channels (10-bit @ 40 MHz )
@ L1 Trigger : processes 25K MUXed silicon strips / FED
Raw Input: 3 Gbytes/sec* after Zero Suppression...
DAQ Output: ~ 200 MBytes/sec
~440 FEDs required for entire SST Readout System
*(@ L1 max rate = 100 kHz)
FE-FPGA Cluster Finder
TTC
TCS
Temp Monitor
JTAG
TCS : Trigger Control System
9U VME64x
October 2002 G Hall 21
CMS Silicon Strip Tracker FED Front-End FPGA Logic
ADC 1 10 10 sy
nc
11
trig2
Ped
sub
11
trig3
Re-
orde
r cm
sub
8
Hit
findi
ng
s-data s-addr 8
16
hit
Pack
etis
er
4
averages 8 header control
DPM
16
No hits
Sequ
ence
r-m
ux
8 8 a
d
a
d
ADC 12 10 10
trig1
sync
11
trig2
Ped
sub
11
trig3 R
e-or
der
cm s
ub
8
Hit
findi
ng
s-data s-addr 8
16
256 cycles 256 cycles
hit DPM
16
No hits
Sequ
ence
r-m
ux
8 8 a
d
a
d
status
averages 8 header status
nx256x16
trig4
Synch in Synch out
Sync
h
emulator in
mux
Serial I/O
Seria
l Int
B’Scan
Loca
l IO
Config
Cluster Finding FPGA VERILOG Firmware
Full flags
data
Global reset
Con
trol
Sub resets
10
10
Phas
e R
egis
ters
Ph
ase
Reg
iste
rs
2 x 256 cycles 256 cycles nx256x16
trig1 Synch error
4x
Temp Sensor
Delay Line
Opto Rx
Clock 40 MHz D
LL 1x
2x 4x
per adc channel phase compensation required to bring
data into step
+ Raw Data mode, Scope mode, Test modes...
160 MHz
October 2002 G Hall 22
The CMS Tracking Strategy
Radius ~ 110cm, Length/2 ~ 270cm
3 disks TID
6 layers TOB
4 layers TIB
9 disks TEC
Number of hits by tracks: Total number of hits Double-side hits Double-side hits in thin detectors Double-side hits in thick detectors
• Rely on “few” measurement layers, each able to provide robust (clean) and precise coordinate determination
2-3 Silicon Pixel 10 - 14 Silicon Strip Layers
October 2002 G Hall 23
Vertex Reconstruction
At high luminosity, the trigger primary vertex is found in >95% of the events
Primary vertices: use pixels!
October 2002 G Hall 24
High Level Trigger & Tracker
In High Level trigger reconstruction only 0.1%
of the events should survive. “How can I kill these events using the least CPU time?” This can be interpreted as: o The fastest (most approx.) reconstruction o The minimal amount of precise reconstruction o A mixture of the two
DAQ
40 MHZ
100 KHz
100 Hz
Same SW would be use in HLT and off-line :
algorithms should be high quality
algorithms should be fast enough
Events rejected at HLT are irrecoverably lost!
HLT Track finding
October 2002 G Hall 25
References (updated by P Hobson) N. Ellis & T. Virdee. Experimental Challenges in High Luminosity Collider Physics. Ann. Rev. Nucl. Part. Sci 44
(1994) 609-653.
• G.Hall Modern charged particle detectors Contemporary Physics 33 (1992) 1-14 & refs therein
• G.Hall Semiconductor particle tracking detectors Reports on Progress in Physics.57 (1994) 481-531
• A. Schwarz 1993 Heavy Flavour Physics at Colliders with silicon strip vertex detectors. Physics Reports
238 (1994) 1-133.
• C. Damerell Vertex detectors: The state of the art and future prospects. Rutherford Appleton Laboratory
report RAL-P-95-008 A pdf version is available on the CERN library Web site.(Search preprints)
• The CMS experiment at the CERN LHC, Journal of Instrumentation, 3 (2008) S08004
• Performance studies of the CMS Strip Tracker before installation , Journal of Instrumentation, 4 (2009)
P06009
• Stand-alone cosmic muon reconstruction before installation of the CMS silicon strip tracker , Journal of
Instrumentation, 4 (2009) P05004
2010 P Hobson 26
How does it perform at the LHC?
2010 P Hobson 27
How does it perform at the LHC?
2010 P Hobson 28
How does it perform at the LHC?
2010 P Hobson 29
How does it perform at the LHC?
2010 P Hobson 30
How does it perform at the LHC?
Kaons, Protons,
Deuterons
Published results • VERTEX 2012
December 2015 P Hobson 31
Efficiency
November 2015 P Hobson 32
dE/dx • Using dE/dx data to fit the KK invariant mass distribution to detect the φ(1020).
October 2002 P Hobson 33
13 TeV data
Photon conversions in the pixel layers
December 2015 P Hobson 34
• Reconstructed photon conversions (photon “radiography”)
Finding the cooling pipes!
October 2002 G Hall 35
IP within jets
October 2002 G Hall 36
Secondary vertex b-tagging
October 2002 G Hall 37
Leakage Currents in Strips
December 2015 P Hobson 38
Leakage current (top) and simulated 1 MeV neutron equivalent dose (bottom)
Leakage current vs radius
Even more from CMS …
December 2016 P Hobson 39
From JINST 9 (2014) P1009 https://cms-results.web.cern.ch/cms-results/public-results/publications/TRK-11-001/index.html Tracker performance plots (public) https://twiki.cern.ch/twiki/bin/view/CMSPublic/DPGResultsTRK
