preparation for physics: the compact muon solenoid (cms) detector greg rakness university of...
TRANSCRIPT
Preparation for Physics:The Compact Muon Solenoid
(CMS) Detector
Greg Rakness
University of California, Los Angeles
RNC Seminar
3 August 2009
Lawrence Berkeley National Lab
3 Aug 2009 G. Rakness (UCLA) 2/62
Outline• Description of CMS• Cosmic rays
– Running the experiment– Detector Performance:
Interaction Region Out
• CMS trigger and its synchronization
• What CMS saw during 10 minutes of beam in September 2008
… and how these things tell us that CMS is ready for physics…
3 Aug 2009 G. Rakness (UCLA) 3/62
CMS is an experiment in the Large Hadron Collider
LHC parameters: • Proton + proton
collisions with center-of-mass energy 14TeV
• Peak Luminosity ~ 1034/cm2/sec
~109 inelastic collisions/sec
~20 events per crossing (40MHz)
CMS
ATLAS
LHC-B
ALICE
3 Aug 2009 G. Rakness (UCLA) 4/62
CMS Design: Discover New Physics
SIMULATION
CMS Collab. 2007, J. Phys. G: Nucl. Part. Phys. 34 995–1579.
Some observables which must be detectable…– H Z + Z 4?– t + t b-jet + b-jet + 2
– Z e+ + e–?– H + ?– H + l + jet?
– Missing ET
– Di-jets– Heavy Ions– … and more…
3 Aug 2009 G. Rakness (UCLA) 5/62
How large is CMS? Compare its physical size with
other detectors…HERMES detector at HERA
(stopped taking data in 2007)
e+
3 Aug 2009 G. Rakness (UCLA) 7/62
CMS Detector at LHCCMS is the biggest detector that I have personally worked on…
p,Pbp,Pb
3 Aug 2009 G. Rakness (UCLA) 8/62
ATLAS Detector at LHC
Of course, there are bigger detectors out there…
p,Pb
3 Aug 2009 G. Rakness (UCLA) 9/62
The CMS Collaboration
2310 Scientific Authors 38 Countries 175 Institutions
CERN
France
Italy
UK
Switzerland
USA
Austria
Finland
GreeceHungary
Belgium
Poland
Portugal
SpainPakistan
Georgia
Armenia
Ukraine
Uzbekistan
CyprusCroatia
China, PR
Turkey
Belarus
EstoniaIndia
Germany
Korea
Russia
Bulgaria
China (Taiwan)
Iran
Serbia
New-Zealand
Brazil
Ireland
1084
503
723
2310
Member States
Non-Member States
Total
USA
# Scientific Authors
59
49
175
Member States
Total
USA
67Non-Member States
Number ofLaboratories
Associated Institutes
Number of ScientistsNumber of Laboratories
629
As of 3 Oct 2007
Mexico ColombiaLithuania
3 Aug 2009 G. Rakness (UCLA) 10/62
Yet another way to think of the magnitude of the CMS detector
• 468 chambers• >17,000 electronics boards
16 different types with firmware• ~400,000 readout channels• ~9000 High Voltage channels• 60 remote electronics crates• ~5,500 skew-clear cables
over a million shielded conductors• 1,400 gigabit optical fibers
All of these electronics must be operated and maintained in order just for the CSC’s to deliver physics…
Some numbers describing the CMS Cathode Strip Chambers (CSC)
3 Aug 2009 G. Rakness (UCLA) 11/62
Compact Muon Solenoid
Detector requirements to extract the desired physics are specified in the Technical Design Report: CERN-LHCC-2006-001
It’s been built. Does it meet the specifications?
From the Interaction Point out:
• Pixel detector
• Silicon strip tracker
• Electromagnetic calorimeter
• Hadronic calorimeter
• 4T superconducting magnet• Muon detectors interleaved with
magnet yokes: • Drift tubes• Resistive plate chambers• Cathode strip chambers
(exploded view)p
p
3 Aug 2009 G. Rakness (UCLA) 13/62
Cosmic Rays 100m Below GroundThe LHC (and CMS) is located under ~100m of rock
Muon spectra underground addressed in
L.N. Bogdanova, et al., Phys. Atom.
Nucl. 69 (2006) 1293;
arXiv:nucl-ex/0601019v1
Angular distribution of cosmic
ray muons is similar beneath
100m solid rock to that on the
surface…
Rate is smaller by factor of ~100
Cosmics are asynchronous
Many features of cosmic rays make them dissimilar to particles coming from the Interaction Point
3 Aug 2009 G. Rakness (UCLA) 14/62
Regarding Cosmic Ray Runs
… we are not collecting millions of cosmic ray triggers for their own sake…
Cosmic ray running prepares us to collect physics data when protons collide in the LHC later this year
Greater exposure to operation before beam, greater chance of finding problems before beam
Some CMS subdetectors have been operating 24/7 for months…
3 Aug 2009 G. Rakness (UCLA) 15/62
Chain of Command
• Shifters– Monitor all CSC activity from control room– If problems arise, call…
• CSC Expert Operators (CEOs)– Weekly rotating duty between proven experts– Mostly graduate students– If necessary, direct shifters’ questions to the appropriate…
• System experts – Experts on-call 24/7 available for each system involved in CSCs – For example, Slow Controls, Run Control, Data Quality
Monitoring, etc.
• CSC Operations Manager– Responsible for any shift logistical issues
Maintain high level of operation while avoiding personnel burnout
Flow chart for the Cathode Strip Chambers (CSC):
3 Aug 2009 G. Rakness (UCLA) 16/62
Dedicated run with Magnetic Field• Operated CMS for 3 weeks continuously
with magnetic field at 3.8T– October – November 2008
Num
ber
of t
rigge
rs (
x106 )
Day
Note: Months of cosmic ray running is
equivalent to a few minutes of
LHC collisions…
3 Aug 2009 G. Rakness (UCLA) 17/62
Event Display of Cosmic Ray with Magnetic Field OFF
+y
+x
+z
Cosmic ray enters CMS from
the top (+y), passes near the Interaction Point
(IP), exits CMS at the bottom (–y)
View of CMS looking down the beampipe
IP
3 Aug 2009 G. Rakness (UCLA) 18/62
Event Display of Cosmic Ray with Magnetic Field ON (3.8T)
+y
+x
+z
View of CMS looking down the beampipe
CMS Magnet: • 6m diameter• 13m long• 4T max field• Stored energy
= 2.5GJ
Solenoidal field in the z-
direction bends the cosmic ray in the r- plane
IP
3 Aug 2009 G. Rakness (UCLA) 19/62
Response of CMS to Cosmic Rays
Some morsels of detector performance, in the order that particles will see them when
they come from the IP
3 Aug 2009 G. Rakness (UCLA) 20/62
CMS Pixel Detector
66M pixels
>99% operational
At full luminosity the pixel occupancy will be
<10–4/crossing
Will be able to resolve individual tracks in a
high multiplicity environment
pp
beam pipe
Physics example: find displaced vertices from t + t bbWW 2 jet + 2
Hitmap of cosmic ray data in the pixel detector (B = 3.8T): ~75k events
3 Aug 2009 G. Rakness (UCLA) 21/62
Response of CMS Pixel Detector to Cosmic Rays
Data calibrated for gains
Simulation uses ideal gain calibration without charge smearing
Cluster charge associated with cosmic ray tracks:
Pixel response to muons well described by a Monte Carlo simulation
3 Aug 2009 G. Rakness (UCLA) 22/62
CMS Silicon Tracker
Silicon strip detectors surround the pixels
Pixels: ~ 1 m2 of silicon sensors, 65 M pixels Si strips: 223 m2 of silicon sensors; 10 M strips
r = 120 cm
z = 300 cm
Physics example: accurately measure the large momentum of muons from Z + + –
Quarter r-z view of tracker:
pixels
IP
3 Aug 2009 G. Rakness (UCLA) 23/62
CMS Silicon Tracker Alignment
not aligned
B=0 T
B=3.8T
Cosmic rays: median of residuals for Tracker Outer Barrel…
The quality of this cosmic ray alignment is comparable to what will be obtained with 10–50/pb of collision data
RMS = 10m
3 Aug 2009 G. Rakness (UCLA) 24/62
CMS Electromagnetic CalorimeterPhysics example: H + …
Energy loss per unit length for muons is well described by
Particle Data Group’s description of
stopping power in PbWO4
TotalCollision LossBremsstrahlung
Demonstrates reasonable understanding of…• Momentum scale of tracker • Energy scale of electromagnetic calorimeter
3 Aug 2009 G. Rakness (UCLA) 25/62
CMS Hadronic Calorimeter
Hadronic calorimeter response to muons well described by a
Monte Carlo simulation
Physics example: accurately measureSUSY particles jets + “missing” transverse energy
3 Aug 2009 G. Rakness (UCLA) 26/62
CMS Muon Systems (Barrel):Drift Tubes and Resistive Plate Chambers
Physics example: trigger on W (+ ) Measure efficiency with cosmic ray data (at 3.8T)…
Resistive Plate Chambers:
x (cm)
y (c
m)
Conclusion: overlap of technologies efficiency > 99%
Drift Tubes:
Efficiency (%
)
3 Aug 2009 G. Rakness (UCLA) 27/62
CMS Muon Systems (Endcap):Cathode Strip Chambers
Physics example: trigger on H + + + Use cosmic ray data to measure residuals…
Wid
th o
f ga
ussi
an c
ore
(cm
)
Resolution is better at the edges of the strips because the charge is shared between strips
3 Aug 2009 G. Rakness (UCLA) 28/62
CSC Residuals Distributions
position within the strip position within the strip
= 322 m 2= 566 m
Near the edges: |strip coordinate| > 0.25 Near the center: |strip coordinate| < 0.25
With the strip-staggering, the resolution of a chamber can be estimated by1/2 (chamber) = 3/1
2 + 3/22
= 161m
(c.f. value required in Technical Design Report = 150m)
per layer per layer
3 Aug 2009 G. Rakness (UCLA) 29/62
CMS Level-1 Trigger
It’s not a “detector,” it selects what physics we get out of the
detector…
3 Aug 2009 G. Rakness (UCLA) 30/62
Functionality of the trigger
Inelastic rate of 109 interactions/sec
Readout 105 events/sec
Write to disk 102 events/sec
Level-1 Trigger
High Level
Trigger
Detectors readout data when a Level-1 Accept (L1A) signal is sent by the trigger system
3 Aug 2009 G. Rakness (UCLA) 31/62
Level-1 Trigger
µ
e
n
p
Use prompt data (calorimetry and muons) to identify: High p
t electron, muon, jets,
missing ET
CALORIMETERs Cluster finding and energy deposition evaluation
MUON System Segment and track finding
New data every 25 ns Decision latency ~ µs
3 Aug 2009 G. Rakness (UCLA) 32/62
CMS Level-1 Trigger
On-detector
Calorimeter path Muon path
Final L1A decision L1A received at detectors 3.2sec after collision data readout
Off-detector
3 Aug 2009 G. Rakness (UCLA) 33/62
CMS Level-1 Trigger
Focus on the trigger of the Cathode Strip Chambers as an example of the detail in the
CMS L1 trigger…
3 Aug 2009 G. Rakness (UCLA) 34/62
Schematic of CSC Trigger Data Flow
Cathode Strip Front End Boards
Anode Wire Front End
Boards
FE
FE
Trigger Motherboard
Wire Local Charged
Track Board
ALCTTMB
CLCTCSC Track Finder
In counting house
In Peripheral Crate
On Chamber
Muon Port
Card
CSC TF
The CSC trigger is a complicated, fast tracking system
3 Aug 2009 G. Rakness (UCLA) 35/62
CSC Trigger Stubs Generated in Front End
Trigger stubs are formed by comparing hits to patterns…
Cathode Stub from strips• Comparator network rapidly determines hits
to ½-strip resolution per layer• Measures • Pattern Radius of curvature momentum
Anode Stub from wire groups • Defines trigger timing• Measures , bx• Pattern envelope pointing to Interaction
Point
3 Aug 2009 G. Rakness (UCLA) 36/62
CSC Trigger Stub Envelopes
oxxxxxxxxxxxoooooxxxxxooooooooooxooooooooooxxxxxooooooxxxxxxxxxoooxxxxxxxxxxxo
layers (z)
half strips (x)
OXXXOOOOOXXOOOOOOXOOOOOOXXOOOOOXXXOOOOXXXO
Anode collision pattern points at the Interaction Point:
wire groups (y)
layers (z)
A CSC trigger stub satisfies a local pattern of hits consistent with a high momentum track from the Interaction Point
x B-field
IP
Cathode envelope of patterns (see next slide):
Typical wiregroup width ~ 20mm Typical half-strip width ~ 10mm
Total chamber depth ~ 15cm
3 Aug 2009 G. Rakness (UCLA) 37/62
Cathode Trigger Stub Patterns
x B-field
Muon stubs which match patterns with higher momentum have higher priority to be passed to the CSC Track Finder
Some example patterns:
oooooxxxoooooooooooxooooooooooooxooooooooooooxoooooooooooxxxooooooooooxxxooooo
layers (z)
half strips (x)
ooooooxxxooooooooooxxoooooooooooxoooooooooooxxooooooooooxxxooooooooooxxxoooooo
oooooooxxxoooooooooxxoooooooooooxoooooooooooxxooooooooooxxooooooooooxxxooooooo
ooooooooxxxoooooooooxxooooooooooxooooooooooxxoooooooooxxxooooooooooxxxoooooooo
Increasing order of priority…
Increasing momentum…
3 Aug 2009 G. Rakness (UCLA) 38/62
CSC Trigger Stub Efficiency
CSC trigger primitive efficiency is >99.5% for high momentum muons (i.e., straight in dX/dZ) which point at
the Interaction Point (i.e., –1 < dY/dZ < 0)
3 Aug 2009 G. Rakness (UCLA) 39/62
CSC Track FinderCombine trigger stubs from 2, 3 or 4 stations to select
high pT muons
Efficiency: trigger matching a tracker track within |Δφ|<0.3
resolution: well reconstructed muons pointing at the Interaction Region
CSC Trigger selects muons with:
• High efficiency
• Good resolution
3 Aug 2009 G. Rakness (UCLA) 41/62
Trigger Synchronization is Critical
• CMS detectors store data in a pipelined readout architecture Data are selected from a position in the
pipeline according to the timing of the trigger
Unsynchronized trigger could turn…
… into…
Need both intra-detector and inter-detector synchronization
3 Aug 2009 G. Rakness (UCLA) 42/62
CSC Synchronization Problem
• 468 chambers, all with different cable lengths– 1 < < 2.4– - < <
…arranged to point at the Interaction Region…
• Cosmic rays are…– Asynchronous– Not coming from the Interaction
Region
Hit map of 1 station of CSC’s (there are 8 stations in total)
3 Aug 2009 G. Rakness (UCLA) 43/62
CSC Synchronization Solution
• Open up the coincidence windows
• Look at average timing differences for muons hitting multiple chambers
• Make a (big) matrix of pairwise differences for pairs with sufficient statistics
• Solve
Synchronizes simultaneous cosmic ray muons which ~point at the IP on average…
3 Aug 2009 G. Rakness (UCLA) 44/62
Relative Synchronization of CSC Trigger Stubs
0th order: cable lengths only… Triggers lined up within 1bx
Fine delay changes implemented Triggers lined up within 0.1bx
… the timing shift for each chamber is obtained by an analysis of the matrix of centroids of bx…
Before After
3 Aug 2009 G. Rakness (UCLA) 45/62
Electromagnetic Calorimeter Timing with respect to CSC Trigger
CSC
ECAL
• CSC timing muons passing simultaneously through any CSC are synchronous through trigger system
• Two peaks in ECAL distribution consistent with muons traversing:
1. CSC ECAL
2. ECAL CSC
• Timing difference approximately consistent with time-of-flight
Timing distribution of Electromagnetic Calorimeter (ECAL) data:
3 Aug 2009 G. Rakness (UCLA) 46/62
Trigger Synchronization with CosmicsTiming differences between different muon detectors in the barrel: Drift Tubes (DT) and Resistive Plate Chambers (RPC)
DT(Top) – DT(Bot) RPC(Top) – RPC(Bot) RPC – DT
CMSTriggers from different detectors are
well-synchronized to cosmic ray muons (on average)
3 Aug 2009 G. Rakness (UCLA) 47/62
The response of CMS to cosmic rays looks good.
How about something more substantial?
3 Aug 2009 G. Rakness (UCLA) 48/62
CMS response to LHC “Beam Splash” events
Beam splash = ~2 x 109 protons at 450 GeV/c incident on collimators ~150m upstream of CMS
CMSCMS
~150 m
p
(9 Sept. 2008)
3 Aug 2009 G. Rakness (UCLA) 49/62
Reconstructed segments in Cathode Strip Chambers…
BEAM SPLASH in CMS
Energy in Hadronic Calorimeter: Energy in Electromagnetic Calorimeter:
3 Aug 2009 G. Rakness (UCLA) 50/62
Significant Energy Deposition in CMS Calorimetry
Correlation between hadronic and electromagnetic energy
deposited in CMS from Beam Splash events
~106 GeV in the Hadronic Calorimeter…
(c.f. cosmic rays ~ 1 GeV)
3 Aug 2009 G. Rakness (UCLA) 51/62
Synchronization of the CMS Hadronic Calorimeter
Timing corrections from synchronous “Splash” events
Before corrections… After corrections…
Tre
cons
truc
ted
– T
pred
icte
d (n
sec)
Nanosecond-scale synchronization will be achieved with collision events
3 Aug 2009 G. Rakness (UCLA) 52/62
100 150 200 250 300 350 400 450 500
Bunch Crossing Number
Different CMS Detectors Triggering on LHC Beam Splash
Detector synchronization much easier with synchronous beam
“Beam Splash” events:
Hadronic Calorimeter
Cathode Strip Chambers
(Three pulses caused by
afterpulsing in photodiodes
filtered by CMS trigger rules)
Eve
nts
First synchronous signals ever seen from LHC
3 Aug 2009 G. Rakness (UCLA) 53/62
CMS response to LHC beam without collisions:
“Beam Halo”
pCMS
(11 Sept. 2008)
3 Aug 2009 G. Rakness (UCLA) 54/62
“Beam Halo” Muon Traversing CMS
“Beam halo” muon ~parallel to the beampipe passes through Cathode Strip Chambers on
both endcaps
3 Aug 2009 G. Rakness (UCLA) 55/62
Reasonable description of beam ON data: combination of • beam halo • cosmic rays
(Separate normalization of blue and black curves not quantitative…)
Cosmic rays
Beam halo
Orientation of beam halo eventsTracks reconstructed in the cathode strip chambers
3 Aug 2009 G. Rakness (UCLA) 56/62
Single Beam Illumination of CMS Cathode Strip Chambers
ME4ME3ME2ME1
ME+1 ME+2 ME+3 ME+4
3 Aug 2009 G. Rakness (UCLA) 57/62
Cosmic rays
Initial attempts to
capture beam
1st capture of LHC beam
Automatic beam abort
First LHC Beam Capture Seen at CMSTrigger rates for “Beam-halo” (i.e., “straight-through”) muons
in one endcap of the Cathode Strip Chambers
11 Sept. 2008
LHC beam control
established within minutes of first attempt
–45<<15
3 Aug 2009 G. Rakness (UCLA) 58/62
First beam in 2008 was incredibly exciting
It will only get better with collisions…
3 Aug 2009 G. Rakness (UCLA) 59/62
Early Physics Program of CMS (I)
10/pb: Commission the detector…
Detector synchronization and alignment In-situ calibration Commission trigger Start “physics commissioning”
Jet and lepton rates; observe W, Z, top And, first look at possible extraordinary signatures…
3 Aug 2009 G. Rakness (UCLA) 60/62
Early Physics Program of CMS (II)
106 W l + 105 Z l + l
104 t t + X
Improve understanding of…• Physics objects• Jet energy scale • Extensive use (and
understanding) of b-tagging• Backgrounds to SUSY and
Higgs searches
Initial Higgs sensitivity Early look for excesses from
SUSY, Z, di-jet resonances…
100/pb: (Re-)Measure the Standard Model…
3 Aug 2009 G. Rakness (UCLA) 61/62
Early Physics Program of CMS (III)
SIMULATION
CMS Collab. 2007, J. Phys. G: Nucl. Part. Phys. 34 995–1579.
1000/pb: Higgs discovery era?
In addition: explore large part of SUSY and resonances at ~few TeV…
L dt = 30/fb
3 Aug 2009 G. Rakness (UCLA) 62/62
Summary
CMS is ready for beam
We are eagerly anticipating proton collisions in the LHC...
3 Aug 2009 G. Rakness (UCLA) 64/62
High Level Trigger Boundary conditions:
Code runs in a single processor, which analyzes one event at a time HLT has access to full event data (full granularity and resolution) Only limitations:
CPU time Output selection rate (~102 Hz) Precision of calibration constants
Main requirements: Satisfy physics program: high efficiency Selection must be inclusive (to discover the unpredicted as well) Must not require precise knowledge of calibration/run conditions Efficiency must be measurable from data alone All algorithms/processors must be monitored closely
3 Aug 2009 G. Rakness (UCLA) 65/62
Tracking at LHC
Fluence over 10 years of
LHC Operation
Need factor 10 better momentum resolution than at LEP1000 particles emerging every crossing (25ns)
CMS SS09 tsv
3 Aug 2009 G. Rakness (UCLA) 66/62
Data vs. Simulation Comparisons of Cosmic Muons in Endcap
Cathode Strip Chamber response to cosmic ray muons are well described by a Monte Carlo
simulation
3 Aug 2009 G. Rakness (UCLA) 68/62
Before first data can be taken…Chamber-Peripheral Crate cables at Peripheral Crate:
• Chambers were checked • at assembly• upon arrival at CERN (ISR)• upon installation in SX5
• Peripheral Crate/chamber electronics were tested • at home institutions• upon arrival at CERN (ISR)• upon installation in SX5
• Low and High Voltage
• Gas
• Slow controls
• Safety systems
Only with all of this in place can we even begin to “commission” the system…
3 Aug 2009 G. Rakness (UCLA) 69/62
CSC ConfigurationCSC is highly configurable: ~100 parameters per CSC
Most CSC electronics are in experimental cavern (high radiation area):
• FPGAs are not radiation hard prone to single event upsets
• PROMs are radiation hard…
• FPGA program stored on one set of PROMs
• configuration parameters stored on another set of PROMs
CSC Configuration = TTC “hard reset”:
• FPGA program reloaded from PROMs
• FPGA reads its configuration from configuration PROMs
Solved issues:
1. Explicit configuration of TTCrq chip required
2. “Check configuration button” reads configuration from hardware and compares with configuration database values