Trigger and DAQ at LHC Trigger and DAQ at LHC C.Schwick Contents - PowerPoint PPT Presentation

Trigger and DAQ at LHC Trigger and DAQ at LHC C.Schwick

Contents Contents INTRODUCTION The context: LHC & experiments PART1: Trigger at LHC Requirements & Concepts Muon and Calorimeter triggers (CMS and ATLAS) Specific solutions (ALICE, LHCb) Hardware implementation Part2: Readout Links, Data Flow, and Event Building Data Readout (Interface to DAQ) Data Flow of the 4 LHC experiments Event Building: CMS as an example Software: Some techniques used in online Thanks to my colleagues of ALICE, ATLAS, CMS, LHCB for the help they gave me during the preparation of these lectures. C. Schwick (CERN/CMS) 2

Introduction: LHC and the Experiments Introduction: LHC and the Experiments C. Schwick (CERN/CMS) 3

LHC: a “ “discovery discovery” ” machine machine LHC: a CMS Aleph LHC startup p p 14 TeV 10 33 LEP - LHC Opal Alice L3 SPS Delphi ATLAS PS LHCb C. Schwick (CERN/CMS) 4

p-p interactions at LHC interactions at LHC p-p σ tot = + + ≈ 100mb diffractive diffractive ≈ 10mb ≈ 10mb + + double elastic diffractive ≈ 10mb ≈ ”small” Interesting Physics inelastic ≈ 70mb C. Schwick (CERN/CMS) 5

Interesting Physics at LHC Interesting Physics at LHC σ tot ≈ 100 mb Events / s (L = 10 34 cm -2 s -1 ) 1 : 100 000 000 000 σ pp σ H(500GeV) ≈ 1 pb C. Schwick (CERN/CMS) 6

LHC: experimental environment LHC: experimental environment L=10 34 cm -2 s -1 • σ inel (pp) ≈ 70 mb σ inel (pp) ≈ 70mb • event rate = 7 x 10 8 Hz • Δ t = 25ns events / 25ns = 17.5 Not all bunches full (2835/3564) • events/crossing = 23 C. Schwick (CERN/CMS) 7

Collisions at LHC Collisions at LHC 7x10 12 eV Beam Energy 10 34 cm -2 s -1 Luminosity 2835 Bunches/Beam 10 11 Protons/Bunch 7.5 m (25 ns) 7 TeV Proton Proton colliding beams Bunch Crossing 4 10 7 Hz Proton Collisions 10 9 Hz ν e e- Parton Collisions q µ + χ 1 - µ - ~ q q Z ~ p g H p p p New Particle Production 10 -5 Hz ~ σ ≈ 0.001pb q (Higgs, SUSY, ....) Z µ+ µ + q µ − ~ χ 2 0 µ - ~ χ 1 0 Selection of 1 event in 10,000,000,000,000 C. Schwick (CERN/CMS) 8

LHC Detector: main principle LHC Detector: main principle Hermetic calorimetry Materials with high number of • Missing Et measurements protons + Active material Heavy materials Electromagnetic and Hadron Muon detector calorimeters n n • µ identification � e � e • Particle identification µ µ (e, � Jets, Missing E T ) p p • Energy measurement � � Heavy materials Light materials (Iron or Copper + Active material) Central detector • Tracking, p T , MIP • Em. shower position • Topology • Vertex Each layer identifies and enables the measurement of the momentum or energy of the particles produced in a collision C. Schwick (CERN/CMS) 9

CMS : study of pp CMS : study of pp SUPERCONDUCTING CALORIMETERS COIL ECAL Scintillating PbWO 4 HCAL Plastic scintillator Crystals brass sandwich Total weight : 12,500 t Overall diameter : 15 m Overall length : 21.6 m IRON YOKE Magnetic field : 4 Tesla TRACKERs MUON ENDCAPS MUON BARREL Silicon Microstrips Pixels Drift Tube Resistive Plate Cathode Strip Chambers ( CSC ) Chambers ( DT ) Resistive Plate Chambers ( RPC ) Chambers ( RPC ) C. Schwick (CERN/CMS) 10

Atlas : study of pp Atlas : study of pp C. Schwick (CERN/CMS) 11

ALICE : study of heavy ion collisions ALICE : study of heavy ion collisions TRD C. Schwick (CERN/CMS) 12

ALICE: Magnet ALICE: Magnet C. Schwick (CERN/CMS) 13

LHCb : study of B-decays (CP) : study of B-decays (CP) LHCb beam interaction point C. Schwick (CERN/CMS) 14

LHCb: Dipole put in place : Dipole put in place LHCb C. Schwick (CERN/CMS) 15

LHCb: : Rhich Rhich Mirror Mirror LHCb C. Schwick (CERN/CMS) 16

First Level Trigger First Level Trigger C. Schwick (CERN/CMS) 17

Why choosing? I want it all !!! Why choosing? I want it all !!! Every 25ns interactions occur and produce 1MB data – 40 Mhz * 1 MB = 40 TB/sec (200 harddisks per second) – Would need 40000 Gigabit Ethernet links to transfer this amount of data – Assuming you need 300ms to analyze and event, a computer would need 140 days to analyze 1 second of data. Compare LEP (e+/e-): Essentially triggering on any (significant) activity in the detector: Trigger rates around 20Hz C. Schwick (CERN/CMS) 18

The 1st level trigger at LHC experiments The 1st level trigger at LHC experiments Requirement: Do not introduce (a lot of) dead-time – O(1%) is tolerated – Introduced by trigger rules : not more than n triggers in m BX pipeline – Needed by FE electronics Trigger Need to implement pipelines – Need to store data of all BX for latency of 1st level trigger 3 µ s (exactly known) – Typical : 10 7 channels / detector some GB pipeline memory – Also the trigger itself is “pipelined” no yes Trigger must have low latency (2-3 µ s) – Otherwise pipelines would have to be DAQ-system very long C. Schwick (CERN/CMS) 19

Imagine you had to choose… … Imagine you had to choose How to decide ? Trigger DAQ at LHC C. Schwick (CERN/CMS) 20

Imagine you had to choose… … Imagine you had to choose How to decide ? Trigger DAQ at LHC INTRODUCTION The context: LHC & experiments PART1: Trigger at LHC Requirements & Concepts Muon and Calorimeter triggers (CMS and ATLAS) Specific solutions (ALICE, LHCb) Look at Hardware implementation Part2: Table of Contents Data Flow, Event Building and higher trigger levels Data Readout (Interface to DAQ) Data Flow of the 4 LHC experiments Event Building: CMS as an example C. Schwick (CERN/CMS) 21

“Typical event Typical event” ” “ No track reconstruction for trigger (2-3 µ s) possible Prepare an “event - TOC” with today’s electronics – Data must be available fast (I.e. shortly after the interaction) H -> Z 0 Z 0 -> 4 µ – Use dedicated sub-detectors – Prepare data with low resolution and low latency in sub-detectors Reconstructed tracks Therefore at LHC: with pt > 25 GeV – Use only calorimeter and muon data C. Schwick (CERN/CMS) 22

Issue: synchronization Issue: synchronization Synchronization: Signals/Data from the same BX need to be processed together But: Particle TOF >> 25ns Cable delay >> 25ns Electronic delays Need to: 25m • Synchronize signals with programmable delays. • Provide tools to perform synchronization (TDCs, pulsers…) C. Schwick (CERN/CMS) 23

Signal path during trigger Signal path during trigger TIME ~3µ Level-1 Accept/Reject Synchronization delay Level-1 signal distribution Global Trigger Processor 1st level trigger Regional Trigger Processors Trigger Primitive Generation Synchronization delay Light cone Data transportation to Control Room Detector FrontEnd Digitizer Particle Time of Fligth SPACE Control Room Experiment C. Schwick (CERN/CMS) 24

Triggering at LHC Triggering at LHC The trigger dilemma: • – Achieve highest efficiency for interesting events – Keep trigger rate as low as possible • Most of the interactions (called minimum bias events) are not interesting • DAQ system has limited capacity • Need to study event properties – Find differences between minimum bias events and interesting events – Use these to do the trigger selection Triggering wrongly is dangerous: Once you throw away data in the trigger it is lost for ever • Offline you can only study events which the trigger has accepted! • Important: must determine the trigger efficiency (which enters in the formulas for the physics quantities you want to measure) • A small rate of events is taken “at random” in order to verify the trigger algorithms (“what would the trigger have done with this event”) • R edundancy in the trigger system is used to measure inefficiencies C. Schwick (CERN/CMS) 25

Triggering at LHC : what info can be used Triggering at LHC : what info can be used Measurements with Calorimeters and Muon chamber system • – Momentum • Measurement of muon p t in magnetic field • p t is the interesting quantity: – Total p t is 0 before parton collision (p t conservation) – High p t is indication of hard scattering process (i.e. decay of heavy particle) – Detectors can measure precisely p t – Energy • Electromagnetic energy for electrons and photons • Hadronic energy for jet measurements, jet counting, tau identification • Like for momentum measurement: E t is the interesting quantity • Missing E t can be determined (important for new physics) Trade off: trigger thresholds versus trigger rate The lower the thresholds the higher the trigger efficiency (good for physics) The lower the thresholds the higher the trigger rate (conflict with DAQ system) C. Schwick (CERN/CMS) 26

First Level Trigger of ATLAS and CMS First Level Trigger of ATLAS and CMS C. Schwick (CERN/CMS) 27