View
219
Download
0
Tags:
Embed Size (px)
Citation preview
Mark Thomson LCUK Meeting, Oxford
CALICE STATUSMark Thomson
University of Cambridge
Overview UK Hardware UK Simulation UK Reconstruction Conclusions
For the CALICE-UK groups: Birmingham, Cambridge, Imperial, Manchester, RAL, UCL
Mark Thomson LCUK Meeting, Oxford
Calorimetry at a Future LCMuch LC physics depends on reconstructing
invariant masses from jets in hadronic final statesKinematic fits don’t help – Beamstrahlung, ISRJet energy resolution is of vital importance
62 % charged particles : 27 % : 10 % KL,n : 2 %
The Energy Flow/Particle Flow Method
The energy in a jet is:
• Reconstruct momenta of individual particles avoiding double counting
Charged particles in tracking chambersPhotons in the ECALNeutral hadrons in the HCAL (and possibly ECAL)
need to separate energy deposits from different particles
Mark Thomson LCUK Meeting, Oxford
Calorimeter Requirements
ECAL
granularity more important than energy resolution, i.e. $$$
e
KL,n
Separation of energy deposits from
individual particles
Discrimination between EM and
hadronic showers
• small X0 and RMoliere : compact showers
• small X0/had
• high lateral granularity : O(RMoliere)
• longitudanal segmentationContainment of EM showers in ECAL
Energy flow drives calorimeter design:
Mark Thomson LCUK Meeting, Oxford
Calorimeter Concept ECAL and HCAL inside coil Better performance – but impacts cost
ECAL: silicon-tungsten (SiW) calorimeter:• Tungsten : X0 /had = 1/25, RMoliere ~ 9mm (gaps between Tungsten increase effective RMoliere)• Lateral segmentation: 1cm2 matched to RMoliere
• Longitudinal segmentation: 40 layers (24 X0, 0.9had)
HCAL: digital vs. analogue (major open question):• Tile HCAL (Analogue readout) Steel/Scintillator sandwich Lower lateral segmentation 5x5 cm2 (motivated by cost)• Digital HCAL High lateral segmentation 1x1 cm2 but digital readout RPCs, GEMS…
Mark Thomson LCUK Meeting, Oxford
CALICE Collaboration
Study calorimetry for a future linear colliderProposed high-granularity ECAL/HCAL $$$ need to fully justify/optimize the calorimetry for FLCTestbeam studies of ECAL and HCAL ECAL studies of Si-W calorimeter HCAL studies of both analogue and digital options
GOALS:Demonstrate technical feasibility of ECALValidate MC simulation (particularly hadronic showers ) vital for optimisation of final design Study digital vs analogue HCAL
PEOPLE: 177 people, 27 institutes (including DESY) 23 UK collaborators !
AIMS
Mark Thomson LCUK Meeting, Oxford
UK ContributionReadout and DAQ for test beam prototype Provide readout electronics for the ECAL (Possibly use UK boards for some HCAL options) DAQ for entire system
Simulation studies ECAL cost/performance optimisation Impact of hadronic/electromagnetic interaction modelling on design. Comparisons of Geant4/Geant3/Fluka
Reconstruction/Energy Flow Started work towards ECAL/HCAL reconstruction Ultimate goal – UK Energy flow algorithm
Luminosity spectrum from Bhabha acolinearity (UCL)
Mark Thomson LCUK Meeting, Oxford
HCAL
ECAL1m
Beam monitor
DAQ
Test Beam and Prototype
Moveable table
Combined ECAL & HCAL Engineering Run late 2004 in e- beam at DESY (ECAL only) Physics Run in 2005 p/ beam at FNAL (TBC) HCAL: 38 layers Fe Insert combinations of:
“digital” pads (350k, 1x1cm2 pads)
GEM RPC
“analogue” tiles (8k, 5x5cm2)
Scintillator tiles
Mark Thomson LCUK Meeting, Oxford
Prototype ECAL
3x10 layers, Si-W 0.4X0, 0.8X0, 1.2X0
Each layer 3x3 wafers Each wafer 6x6 pads 9720 channels total
External Readout (VFE) Wafers Si/W/Si Sandwich
Carbon Fibre/Tungsten
Mark Thomson LCUK Meeting, Oxford
Readout OverviewCALICE ECAL has 9720 channelsEach gives analogue signal, 14-bit dynamic rangeVery-front-end (VFE) ASIC (Orsay) multiplexes 18 channels to
one output lineVFE-PCB handles up to 12 VFEs (216 channels)Cables from VFE-PCBs go directly to UK VME readout boards,
called Calice ECAL Readout Cards (CERCs)Based heavily on CMS tracker readout
• Rutherford Laboratory– Adam Baird, Rob Halsall, Ed Freeman
• Imperial College London– Osman Zorba, Paul Dauncey
• University College London– Matt Warren, Martin Postranecky
• Manchester University– Dave Mercer
Mark Thomson LCUK Meeting, Oxford
•Prototype design completed last summer•Two prototype boards fabricated last year
•Arrived on November 21 at Rutherford Laboratory
CERC status
•Currently under stand-alone tests in the UK•Aim to test with a
VFE-PCB in the UK very soon
•Move UK hardware to Paris (Ecole Polytechnique) for cosmic tests with fully populated VFE-PCB with Si wafers in Feburary
Front End FPGAs
Back End FPGA
Mark Thomson LCUK Meeting, Oxford
Outstanding IssuesFinal path for data has several complex steps
FE digitises ADC data for each triggerAutomatically transferred to 8MByte memoryMemory read from VME when bandwidth available
Needs data transfer, memory control and VME interfaceBE FPGA firmware not yet functionalMemory components delayed in delivery; not yet mounted on
CERCsAiming for end of March for all this to be working !
Backup for VFE testsImplement simple RS232 interface from PC to BE and hence to
FEsRS232 reads FIFO one word at a time directly to PC8MByte memories bypassed, must read each event before next
triggerRate is slow ~1Hz for events; sufficient for cosmics
Mark Thomson LCUK Meeting, Oxford
ScheduleVFE tests in Paris in February
Essential test of prototypes before moving to production
Possible AHCAL test in AprilNeed more information on what is required; number of
channels, interface specification for VFE-PCB equivalent,…
Finalise redesign by end MarchRe-layout/fabricate 9 production CERCs in April-May
Simple fixes for the few known problems may be possible
If so, maybe no need to re-layout; save a month
Only have components for nine boards; need to know early if more wanted for HCAL
Will need non-UK funds for HCAL readout
Full ECAL system tests from July onwardsOn schedule for DESY ECAL test beam in Oct/Nov
Mark Thomson LCUK Meeting, Oxford
Test Beam Requirements
Use MC studies to study what data would be most useful in validating MC models (David Ward)
e.g. Compare samples of
5 GeV + in Geant3 (histohisto) and Geant4 (points)
Significant differences seen at the level of 104 events
HCAL shows greatest discrepancies
5 GeV +
What Data ? Proton/pion/muon ?How much data ?
Mark Thomson LCUK Meeting, Oxford
Differences depend on Energy
1 GeV + 50 GeV +
Therefore scan over energies
Mark Thomson LCUK Meeting, Oxford
Protons vs Pions 5 GeV p 5 GeV +
Need to understand beam ! i.e. pion/proton ratioFind protons/neutrons v. similar (at least in MC)Greater differences for Scintillator HCAL vs. RPC
Mark Thomson LCUK Meeting, Oxford
Test Beam : Conclusions
1% precision suggests >104 events per particle type and energy.
Would like energies from 1-80 GeV (~10-15 energy points?).
Pions and protons desirable (Čerenkov needed). +Electrons (+ muons?) for calibration.
Need to understand beam Both RPC and Scintillator HCAL needed. Position scan – aim for 106 events/energy point? Also some data at 30-45o incidence.
Mark Thomson LCUK Meeting, Oxford
Study of hadronic models (G Mavromanolakis, N. Watson)
Compare: (G Mavromanolakis)• Geant 3 with Gheisha• Geant 3 / Gheisha (SLAC
version)• Geant 3 / Fluka• Geant 3 / Fluka / Micap (used
for n < 20 MeV)
• Geant 4 / Mokka
Also Studying:Variations of Geant 3/Geant 4
cutoffs (G Mavromanolakis)Geant 4 FLUKA (N.Watson) - Geant 3 version deprecated - Geant 4 implementation extremely interesting - tricky to get working, but making excellent progress
Mark Thomson LCUK Meeting, Oxford
Calorimeter Reconstruction High granularity calorimeter – very different from
previous detectors
`Tracking calorimeter’
• Requires new approach to reconstruction
• Already a lot of good work on powerful energy flow algorithms
• Still room for new ideas/ approaches
• Current codes : inflexible
UK Effort just starting (Chris Ainsley)• Important for future analysis and `energy flow’ studies/detector optimisation
Mark Thomson LCUK Meeting, Oxford
ECAL Clustering Aim – to produce a flexible algorithm, not tied to
specific geometry/MC program.
• Algorithm needs to cope with tracks and clusters
• Sum hits within cell; apply threshold of ⅓ MIP
• Form clusters in layer 1 of ECAL.
• Associate each hit in layer 2 with nearest hit in layer 1 within cone of angle . If none, initiate new cluster.
• Track onwards layer by layer through ECAL and HCAL, looking back up to 2 layers to find nearest neighbour, if any.
Mark Thomson LCUK Meeting, Oxford
Example Events 15 GeV - 15 GeV e-
(Reconstructed clusters are colour-coded, black = highest energy cluster)
Handles CLUSTERS and TRACKS
Mark Thomson LCUK Meeting, Oxford
Some more difficult examples 15 GeV 15 GeV -
Separates nearby ECAL clusters
So far things look good, but this is just the first stage
Mark Thomson LCUK Meeting, Oxford
Conclusions CALICE ECAL prototype progressing well - test beam before end of 2004 ! Confident that UK Electronics/DAQ will be ready Work on Digitization simulation starting
(D.Bowerman, C.Fry) UK contributing significantly to understanding FNAL
test beam requirements On-going studies of hadronic models UK reconstruction effort starting - important for analysis of test beam data - important for optimisation of ECAL design Next 2 years are going to be very interesting UK groups well placed to participate in analysis of test
beam data
Mark Thomson LCUK Meeting, Oxford
Mark Thomson LCUK Meeting, Oxford
RPC vs. Scintillator HCAL Scintillator RPC
Mark Thomson LCUK Meeting, Oxford
Neutrons vs Protons
5 GeV p 5 GeV n
Mark Thomson LCUK Meeting, Oxford
Eight Front End (FE) FPGAs control all signals to front end electronics via front panel input connectors
Back End (BE) FPGA gathers and buffers all event data from FE and provides interface to VME
Trigger logic in BE for timing and backplane distribution; only active in one board
Each input is one full or two half-full VFE-PCBs; need 45 inputs = 6 CERCs
Based on CMS tracker readout (FED)
CERC overview
Mark Thomson LCUK Meeting, Oxford
Readout DetailsBased on CMS silicon tracker readout (FED)
Will “borrow” a lot of firmware from themUnfortunately not yet as well-developed as hoped
Dual 16-bit ADCs and 16-bit DACDAC fed back for internal as well as front end calibrationADC 500kHz; takes ~80ms to read and digitise event data from
VFE-PCBNo data reduction in readout board
ECAL event size: 3.5 kBytes per board, 20 kBytes total per eventOn-board buffer memory; 8 MBytes
No buffering available in ECAL front end; receive data for every trigger
Memory allows up to ~2k event buffer on readout board during beam spill
VME readout speed ~20 MBytes/s; several seconds readout after spill
Large amount of unused I/O from BE FPGA to backplaneWill implement trigger logic and control/readout interface to VME
in BE