Double Stack Tracking Trigger Strawman

Transcription

1 Double Stack Tracking Trigger Strawman

2 Scope of this Discussion: Outer Tracker The region of the inner-most Pixel Layers is fundamentally challenging g at the SLHC, especially for the Sensor Technology One may speculate as to the most promising way forward B-tagging, e/γ discrimination remain Very Important Assume 4 Layers of Fine-Pitch Pixels To be better defined Here focus on Outer Tracker Assume boundary between inner-most Pixel Layers and Outer Tracker is somewhere between 20 ~ 40cm In any future baseline layout, Outer Tracker and inner-most Pixel Layers will have to make a coherent Tracking System

3 CMS from LHC to SLHC cm -2 s At SLHC CMS faces new challenges, in particular for both Tracking and Triggering I. Osborne

4

5 An L1 Track Trigger for SLHC is not an Elective Project Joel

6 Required Functionality L1 Trigger Confirmation of Isolated High-pt μ Candidates Fast, Efficient & Clean Tracking Excellent Pt resolution Isolation Increased Rejection of fake e/γ Candidates Match with Track (nb conversions ) Isolation Tau Jet trigger Low Multiplicity, Isolation MET? Clean up High Pile-up environment Rejection of Uncorrelated Combinations, from different primary vertex? Match with Tracks at Vertex? Factor ~ 100 reduction For same Pt threshold

7 Required Functionality L1 Trigger Confirmation of High Pt Track Candidates Tracks with Pt above 15 ~ 20 GeV Excellent Efficiency Good Pt resolution Isolation Tracks with Pt above ~ 2 GeV Good Efficiency Longitudinal Vertex association Tracks with Pt above ~ 2 GeV Good Z Vertex resolution

8 Local Occupancy Reduction Cannot possibly transfer all Tracker data at 40MHz! Crossing Frequency / Event Read-Out ~ 40MHz / 100kHz ~ 1 / 400 L1 Data reduction by a factor of 100 ~ 200 is a reasonable target For L1 Trigger propose to transfer only hits from tracks with Pt > ~ 2 GeV The aim is to provide useful Isolation information Tracks with Pt > ~ 2 Gev are less than 1% of the Tracks inside acceptance This corresponds to the maximum plausibly manageable L1 data rate In addition, must provide means of rapidly & reliably identifying high Pt (isolated) tracks ( Pt > 15 ~ 25 GeV)

9 Local Occupancy Reduction Tracks with Pt > 1GeV < 10% of Tracks in acceptance Tracks with Pt > 2.5 GeV < 10% of the remaining Tracks

10 Local Occupancy Reduction with Local Track Vectors J. Jones (~2005) CMS Tracker SLHC Upgrade Workshops α

11 Local Occupancy Reduction with Local Track Vectors Pairs of Sensor Planes, for local Pt measurement High Pt tracks point towards the origin, low Pt tracks point away from the origin Use a Pair of Sensor Planes, at ~mmdistance Pairs of Hits provide Vector, that measure angle of track with respect to the origin Note: angle proportional to hit pair radius J. Jones (~2005) CMS Tracker SLHC Upgrade Workshops α Keep only Vectors corresponding to high Pt Tracks

12 Local Occupancy Reduction a Hierarchical scheme with Double Stacks Collect Pairs of Hits from each sensor doublet & match into Track Stub Pass onto L1 Trigger Local Information Gathering, and Processing Hierarchy ~2mm ~40mm Collect hits from each sensor & match Hit Pairs Collect hits from each sensor & match Hit Pairs Within a Stacked-Sensor Module Collect Hits from each Sensor Match into Hit Pairs & Reject Hit Pairs from Very low Pt Tracks: Pt < ~ 1GeV Nb one datum / Hit Pair Within a Double Stack Collect Hit Pairs from each Sensor Doublet Module Match into Track Vectors & Reject Track Vectors with Pt < ~ 2GeV Transmit to USC for High Pt & Isolation L1 Track Trigger Primitives

13 Recent results for a Stack of closely spaced sensors: pitch ~ 100um*2.4mm (M. Pesaresi) High rejection factors possible Mark Pesaresi Much Sharper Threshold For Low Threshold Value

14 Recent results for a Stack of closely spaced sensors: pitch ~ 100um*2.4mm (M. Pesaresi) Mark Pesaresi No useful discrimination at Pt ~ 20 GeV

15 Recent results for a pair of Double Stacks spaced ~ 10cm apart (M. Pesaresi) Excellent discrimination up to Pt ~ 20 GeV Mark Pesaresi

16 CMS SLHC Tracker Straw Man Layout Illustrations R-Phi Hermitic Double Stacks: get all 4 hits in one ROD or in the neighbor No communication across r-phi stacks ~ few cm ~mm

17 Full Double Stack Trigger Tracker Straw Man Layout Basic L1 Tracker Trigger concept: Local Data Reduction based on Track Vectors An r-phi hermetic Double Stack arrangement of BEAMs is proposed Rapid L1 High Pt Track identification (10~25 GeV), in hermetic r-phi sectors Isolation criteria with lowest possible Pt threshold (~ 2 GeV) The Double Stack layers will also provide Tracking Track Reconstruction for the HLT & Off-line should be very fast Track Parameters should be of high quality (to be verified in detail) The use of ~mm long Pixels provides opportunity for primary vertex association of Track Trigger Primitives The BEAMs provide opportunities for Material Budget Reduction

18 Stacked Tracking Trigger Straw Man This Simple Concept drives all aspects of the System, and Defines Requirements and Challenges throughout the System Module Sensors; Alignment; On Module Connectivity, Data Transmission & Reduction; Module I/O and Interface to ROD; Power & Cooling BEAM Module Alignment; On BEAM Data Transmission & Reduction; Power Distribution; Mechanics & Cooling Off-Detector BEAMt to USCD Data Transmission; i Tracking Ti Trigger Primitives; Pi iti Event tread- Out; CTRL System; Power System; Cooling System

19 Stacked Tracking Trigger Straw Man This Simple Concept drives all aspects of the System, and Defines Requirements and Challenges throughout the System Module Sensors; Alignment; On Module Connectivity, Data Transmission & Reduction; Module I/O and Interface to ROD; Power & Cooling BEAM Module Alignment; On BEAM Data Transmission & Reduction; Power Distribution; Mechanics & Cooling Off-Detector BEAM to USC Data Transmission; i Tracking Trigger Primitives; iti Event Read- Out; CTRL System; Power System; Cooling System

20 Some Numbers Basic Input: Occupancy at at tr ~ 35cm (TIBL2R Radius) Typical ~ 2 hits / cm 2 / 25ns Maximum < 10 * 2 = 20 hits / cm 2 / 25ns Strip Occupancy ~ 120MHz / cm 2 at R = 25cm Strip Occupancy ~ 80MHz / cm 2 at R = 34cm Strip Occupancy ~ 40MHz / cm 2 at R = 50cm 1/2 Strip Occupancy ~ 20MHz / cm 2 at R = 60cm 1/2 (Geoff Hall, compilation of full simulation results from Ian Tomalin) Nb these occupancy are for 320um~500um thick sensors: 2 ~ 4 hits/cluster Assume Reduction Factor ~ 2 from clustering To be verified Crossing Frequency / Event Read-Out ~ 40MHz / 100kHz ~ 1 / 400 L1 Data reduction by a factor of 100 ~ is a reasonable target

21 Some Numbers Material Budget ~ Material / cm 2 Consider rates and power / cm 2 Nb normalize to cm 2 of Silicon 1 module = 2 sensitive layers = 2 * x*y cm 2 (eg 2 * 100cm 2 ) Present CMS Tracker Event Read-Out ~ 4 channels / cm 100KHz Data Rate ~ 4MHz / cm 2 (analogue info ~ 10bits equivalent) Present CMS Tracker Power Inside Volume ~ 33kW over ~ 210m 2 Power Density ~ 16mW /cm 2 inside Tracking volume 6 Single-Sided + 4 Double-Sided = 14 Sensitive Layers

22 Data Transmission, Reduction, Power Density In the following Assume Zero Suppressed Read-Out Data rates ~ driven by Occupancy, NOT by Channel Count De-randomized Read-Out from Module to USC Available Bandwidth ~ Average Bandwidth, with * 2 safety margin Non De-randomized within Module: Available Bandwidth ~ 10 * Average Reduce Output Data Rates from Module by ~ 2 * 10 1 pair of accepted clusters = 1 datum per Hit Pair Output from Module Accept ~ 1 / 10 Hit Pairs: Pt Threshold 1 ~ 2 GeV Reduce Output Data Rates from ROD by ~ 5 2 accepted cluster pairs = 2 data per Track Vector Output from ROD Accept ~ 1 / 5 Track Vectors: Pt Threshold ~ 2GeV

23 Data Transmission, Reduction, Power Density In the following Assume Pixel Dimension ~ 100um * 1mm Pixels /cm 2 (more on this later) ~ 18 bits / L1 hit Address & Time Stamp info within Module Assume no analogue information for L1 ~ 24 bits / L1 hit Address & Time Stamp info from Module 32 bits / Read-Out hit info inside Tracker Assume ~ 8 bits analogue information for Read-Out Nb if Short Strips ~ 32bit address field is reduced by ~ 5bits ~ 20% reduction in Address Information for ~ 32 fewer channels / cm 2

24 Data Transmission, Reduction, Power Density Within a Doublet-Sensor Module: Un-terminated Lines Only transmit from one sensor plane to the other Transmission distance ~ few mm Input * Output Data reduction ~ 2 * 2 * 10 Power driven by by Actual Usage Available ~ 10 * Average Energy/bit of Link over ~ few mm < 2pJ/bit (1pJJ/bit possible?) Transmission rate ~ 320Mb/s (1Gb/s possible?) Data Rates / cm 2 Average Bandwidth Available Bandwidth L1 ~ 400Mb/s < 4Gb/s Read-Out ~ 6Mb/s < 60Mb/s Power / cm 2 Average Bandwidth Available Bandwidth L1 ~ 1mW < 10mW Read-Out

25 Data Transmission, Reduction, Power Density Within a Doublet-Sensor Module: Un-terminated Lines Only transmit from one sensor plane to the other Transmission distance ~ few mm Input * Output Data reduction ~ 2 * 2 * 10 Power driven by by Actual Usage Available ~ 10 * Average Energy/bit of Link over ~ few mm < 2pJ/bit (1pJJ/bit possible?) Transmission rate ~ 320Mb/s (1Gb/s possible?) Data Rates / cm 2 Average Bandwidth Available Bandwidth L1 ~ 400Mb/s < 4Gb/s Read-Out ~ 6Mb/s < 60Mb/s Links/Chip ~ 6cm 2 Average Bandwidth Available Bandwidth L1 ~ 6 ~ 60 Read-Out

26 Data Transmission, Reduction, Power Density To the End of a ROD ~ PP1: Transmission Line Transmission distance 3 ~ 10m Input * Output Data reduction ~ 200 Power driven by Available Bandwidth (~ 2 * Average) Energy/bit for Link over ~ 10m < 20pJ/bit (10pJ/bit over ~ 1m) Transmission Rate ~ 1Gb/s (is 1Gb/s possible?) Includes Clock & Error Recovery Data Rates / cm 2 Average Bandwidth Available Bandwidth L1 ~ 50Mb/s ~ 100Mb/s Read-Out ~ 6Mb/s ~ 12Mb/s Power / cm 2 Average Bandwidth Available Bandwidth L1 ~ 1mW ~ 2mW Read-Out <0.1mW ~ 0.1mW

27 Data Transmission, Reduction, Power Density To the End of a ROD ~ PP1: Transmission Line Transmission distance 3 ~ 10m Input * Output Data reduction ~ 200 Power driven by Available Bandwidth (~ 2 * Average) Energy/bit for Link over ~ 10m < 20pJ/bit (10pJ/bit over ~ 1m) Transmission Rate ~ 1Gb/s (is 1Gb/s possible?) Includes Clock & Error Recovery Data Rates / cm 2 Average Bandwidth Available Bandwidth L1 ~ 50Mb/s ~ 100Mb/s Read-Out ~ 6Mb/s ~ 12Mb/s Links/Module ~ 200cm 2 Average Bandwidth Available Bandwidth L1 ~ 10 ~ 20 Read-Out ~ 1 ~ 2

28 Data Transmission, Reduction, Power Density To USC: Optical Link Transmission distance ~ 100m Input Data Reduction ~ 200 Power driven by Available Bandwidth (~ 2 * Average) Energy/bit for Link over < 200pJ/bit (100pJ/bit possible?) Transmission Rate = 10Gb/s Includes Clock & Error Recovery Data Rates / cm 2 Average Bandwidth Available Bandwidth L1 ~ 10Mb/s ~ 20Mb/s Read-Out ~ 6Mb/s ~ 12Mb/s Power / cm 2 Average Bandwidth Available Bandwidth L1 ~ 2mW ~ 4mW Read-Out ~ 15mW 1.5mW ~ 3mW

29 Data Transmission, Reduction, Power Density To USC: Optical Link Transmission distance ~ 100m Input Data Reduction ~ 200 Power driven by Available Bandwidth (~ 2 * Average) Energy/bit for Link over < 200pJ/bit (100pJ/bit possible?) Transmission Rate = 10Gb/s Includes Clock & Error Recovery Data Rates / cm 2 Average Bandwidth Available Bandwidth L1 ~ 10Mb/s ~ 20Mb/s Read-Out ~ 6Mb/s ~ 12Mb/s Links/Module ~ 200cm 2 Average Bandwidth Available Bandwidth L1 ~ 1/4 ~ 1/2 Read-Out ~ 1/8 ~ 1/4

30 Data Transmission, Reduction, Power Density At R ~ 35cm Based on 1/2*10 off Module * 1/10 off ROD data rate reduction Power for Data Transmission within Module L1@ 40MHz ~ 1mW/cm 2 100kHz < 0.1mW/cm 2 Power for Data Transmission i To the End of a ROD 40MHz ~ 2mW/cm 2 100kHz ~ 0.2mW/cm 2 Power for L1 Trigger Info Transmission To USC (at Bulk head) 40MHz ~ 4mW/cm 2 100kHz ~ 3mW/cm 2 Total Power Budget L1 & Read-Out Data R ~ 35cm Inside Tracking Volume: ~ 3mW/cm 2 At Bulkhead: ~ 7mW/cm 2

31 Data Transmission, Reduction, Power Density At R ~ 35cm Based on 1 / 2*10 off Module * 1 / 10 off BEAM data rate reduction Total Power Budget L1 & Read-Out Data R ~ 35cm Inside Tracking Volume: ~ 3 mw/cm 2 Compares with ~ 16mW /cm 2 inside present Strip Tracker volume At Bulkhead: ~ 7mW/cm 2 A L1 Track Trigger based on the scheme presented here is NOT ruled out by the Power requirements for the L1 Data Transfer Challenges for Data Transmission & Reduction include: Module interconnect technology High rate (1Gb/s) Low Mass Link over length of BEAM De-randomized L1 data transfer protocol Hit Doublet & Track Vector Logic

32 Granularity: Short Strips vs Long Pixels The CMS Silicon Strip Tracker is extremely effective because: Excellent Quality of Pixel Seeds Fine strip pitch, from 80um to 200um each hit has high resolution and track parameters are rapidly constrained Strip length, from 10cm to 20cm results in cell size ~ 0.5mm 2 occupancy ~ 2% or less at Pattern recognition converges ~ unambiguously with first few hits => fast At SLHC occupancy 10~20 times higher Short Strips Strip length in range 1 ~ 2cm to maintain low occupancy Long Pixels es Pixel length in range 1 ~ 2mm => reduce occupancy to ~ Inner Pixel like 3D info => 3D Tracking without Stereo Layers Sufficient Z resolution at L1 to sort Trigger Primitives by Interaction Vertex

33 Granularity: Short Strips vs Long Pixels Comparative Performance Studies are Important Guidance Rejection of tracks from different interaction vertices at L1? Cost and Manufacturability are a Key Input Implications on System, Read-Out Architecture etc. Reliable projections of Power Dissipation/cm 2 are a Fundamental Input Short Strips vs Long Pixels Extrapolate from Strip Tracker APV25 to reduced capacitance short strips Extrapolate from Pixel ROC to larger capacitance long pixel Compare: Power, Material, Cost, Feasibility, Performance Pursue both approaches until these points are sufficiently well understood to draw some conclusions

34 Front-End Power for Long Pixel Tracker Power of present CMS Pixel ROC ~ 30uW / channel 100um * 150um Pixel, Power Density ~ 200mW / cm 2 Pixel Front-End Read-Out chip Power Density ~ 16 * Strips Pixel Channel Density ~ * Strips! Assume 20 ~ 30uW / channel for 100um * 1 ~ 2mm Long Pixels Private communication from Roland Results in ~ 20mW / cm 2 Compares with ~ 16mW /cm 2 inside present Strip Tracker volume Compares with ~ 3mW/cm 2 for Data Transmission inside TK Volume Long Pixel Channel Density 100 ~ 200 * Strips Long Pixels not ruled out by Front-End Power requirements

35 Straw Man Layout The Function of the Straw Man is to Illustrate the Underlying Ideas, for a CMS SLHC Tracker with L1 Trigger capability It is intended to highlight the Pros and Cons of these Ideas, to allow informed decisions down the line And to Provide a Framework to help Direct and Focus different Lines of Activity Performance Studies Sensors / Front-End Read-Out / Interconnects Module Functionality & Design Mechanics / Cooling and Services Integration Data Reduction and Data Transmission Improved Power Distribution Scheme, Local Voltage Regulation etc Material Budget Reduction and Optimization Off-Detector Data Processing Etc. On the way to a Base-Line Layout

36 Straw Man Layout The Function of the Straw Man is to Illustrate the Underlying Ideas, for a CMS SLHC Tracker with L1 Trigger capability It is intended to highlight the Pros and Cons of these Ideas, to allow informed decisions down the line And to Provide a Framework to help Direct and Focus different Lines of Activity An Effective L1 Track Trigger is a Major Challenge: A Straw-Man is Required in order to make Effective Progress On the way to a Base-Line Layout

37 Straw Man Layout: Double Stack Layers η 1040 Each Double Stack Layer requires 4/4 hits Minimal potentially viable configuration is 2 Double Stack Layers Require 1 OR the Other Double Stack L1 & Tracking Layers, with full acceptance up to η ~ 2.5: Each Layer provides 2 * 2 = 4 hits 2 Layers = 8 hits 2700

38 Straw Man Layout: 2 Double Stack Layers + Outer Tracker η Double Stack L1 & Tracking Layers, with full acceptance up to η ~ 2.5: Each Layer provides 2 * 2 = 4 hits 2 Layers = 8 hits 2700 Outer Tracker: Optimized for Tracking No L1 functionality Introduces 3 rd System, in two flavors

39 Straw Man Layout: 3 Double Stack Layers η 1040 Build on Minimal, Potentially Viable Parts Kit Focus the Effort Add complexity only if / when Needed Double Stack L1 & Tracking Layers, Each Layer provides 2 * 2 = 4 hits 3 Layers = 12 hits Single System provides Full L1 & Tracking functionality 2700

40 Build on Minimal, Potentially Viable Parts Kit Focus the Effort Add complexity only if / when Needed Hole above η ~ 1.7 Straw Man Layout: 3 Double Stack Layers η Double Stack L1 & Tracking Layers, with full acceptance up to η ~ 1.7: Each Layer provides 2 * 2 = 4 hits 3 Layers = 12 hits Single System provides Full L1 & Tracking functionality 2700

41 Build on Minimal, Potentially Viable Parts Kit Focus the Effort Add complexity only if / when Needed Use Short Forward Cylinders to Avoid Hole above η ~ 1.7 Straw Man Layout: 3 Double Stack Layers η Double Stack L1 & Tracking Layers, with full acceptance up to η ~ 2.1: Each Layer provides 2 * 2 = 4 hits 3 Layers = 12 hits 2700 Single System provides Full L1 & Tracking functionality Short FWD Cylinders close acceptance Total Silicon Surface ~ 375m 2 Present Tracker ~ 210m 2

42 Full Stacked Trigger Tracker Straw Man Layout Propose that a Full Stacked Trigger Tracker Straw Man be studied A a Potentially Viable Single Concept providing Both Tracking and Triggering Functions, and to establish its performance potential and possible shortcomings As a means of providing a focus for the System Design & defining sets of work-packages for each subsystem in the Upgraded Tracker As a Benchmark for alternative Stacked Trigger + Outer Tracker Layouts There are Many Challenges BUT CMS needs a viable Trigger for SLHC Robust L1 Track Trigger primitives are a Must An all Pixel Stacked Trigger Tracker will be Game Changing detector Just as the present CMS Tracker is a Game Changing g detector

43 Scope of the Tracking Trigger Project On Module Processing Off Detector Data Transmission On Rod Data Transmission To Trigger Power, Cooling, Mechanical Structures Sensors On Rod Processing Front End Electronics Interconnection Technology Off Detector Processing

44 Basic Units for Tracking Trigger Project Module Unit 100 * 100 * 2mm Sensors, FE Electronics, Interconnections, On module Processing Mechanical Structure 100 mm 40 mm 2mm 2700 mm On BEAM Data Transmission 100 mm 100 mm Power Unit ~12V in ~1.8V out C0 2 Cooling System

45 Basic Units for Tracking Trigger Project On Rod Processing Off Detector Processing Trigger Off Detector Data Transmission FED DAQ Cooling System Power System

46 Module Design Basic Units for Tracking Trigger Project Sensor, FEE, Interconnection, On Module Processing Construction On BEAM Data Transmission i On BEAM Processing Simulation Off Detector Data Transmission and Off Detector Processing System FEDs and DAQ system Engineering Powering and DCS System (Grounding and Shielding) and Cooling System Optimization Mechanical Layout and Structures

47 FNAL Tracking Trigger Summary The FNAL Workshop was an Opportunity to sign up additional US groups to the CMS SLHC Tracker Upgrade, and strengthen the present Tracker Collaboration This was very successful About 40 ~ 50 people attended the Tracking Trigger Sessions Many new groups signed on to work on various aspects of the proposed system There is a substantial re-focusing of people and developments, originally projected towards the ILC, on the CMS SLHC Tracker Upgrade Strong FNAL groups working on Vertical Integration (maps onto doublet module) and Very Low Mass Mechanics Next steps: consolidate newcomers into existing working groups Define goals, plan of work, milestones, etc.

48 Proposed Project Time Scales 2009 Define and prove the Viability of All Systems within the New Straw Man and Review Progress 2010 Optimize the Layout and System Developments in the light of the work in Prepare Demonstrators of all the Systems 2012 Prepare the TDR for an Upgraded Tracker Double January Stack 2009 Tracking Trigger Strawman

49 Straw Man Design Study: V0 Starting Point Strawman Design Study V0 Starting Point

50 Straw Man Design Study: V0 Starting Point Agree on basic inputs to layout Module size, Tiling strategy & Overlaps, Stacked Barrel layer radii & length, Agree on parameters to vary Pixel dimensions, Module Separation within Doublet, Doublet Separation within ROC, Cluster Finding Efficiency Agree on basic inputs to Structures & Materials in Tracking volume Module: Sensors, ROC, PCB, Connector, Mechanics, Power Dissipation ROD: Mechanics, Cooling, Data Transmission, Power Distribution, Controls & monitoring Mechanical Supports & End-Flanges, Manifolds & Connections Propose to start with day 0 set of educated guesses Then input more realistic estimates/targets Use Layout Task-Force Tools for consistent comparative studies etc

51 Straw Man Design Study: V0 Starting Point Progress since FNAL Workshop Working towards detailed V0 Starting Point Design Study Progressing on First pass through the system, from Front-End to Back-End See presentation by Bill Cooper on Layout and Mechanics

52 Straw Man Design Study: V0 Starting Point Sensor Characteristics Agree on basic inputs to layout Sensor Active Area Dimensions 9.4cm * 9.4cm * 160um Sensor Physical Dimensions 9.6cm * 9.6cm * 320um Tiling strategy in phi Propose High-Low as in TOB with no Lorentz Angle Tilt (Presently with Lorentz AngleTilt) Tiling strategy in Z Modules are butted end-to-end Overlaps in Phi Stacked doublets are hermetic in phi for tracks from IP Min. projective overlap between Stacked Doublets (4/4 hits) 2 ~ 4mm Provide overlap with up to 5mm radial displacement of IP Overlaps in Z No overlaps in Z: modules are butted end-to-end Clearances All sensor clearances > 1mm

53 Straw Man Design Study: V0 Starting Point Agree on basic inputs to layout Stacked-Doublet Barrel Super-Layer radii & length Layer 1 Radius ~ 30cm Length to η ~ 2.5 Layer 2 Radius ~ 50cm Length ~ 270cm to η ~ 2.5 Layer 3 Radius ~ 100cm Length ~ 270cm Forward Layers 1 & 2 Radius & Length to optimize η coverage

54 Straw Man Design Study: V0 Starting Point Agree on basic inputs to layout Stacked-Doublet Barrel Super-Layer Z overlap at η = 0 The 3 main stations are each ½ the length of the Tracker Ensure all hits over full length of IP contained in one or other ½ Barrel Guesstimate is Z Interlocking overlap 1 ~ 2cm; ΔR ~ 4mm Set Z overlap to to minimum required to ensure coverage η = 0 ~ 4mm 1 ~ 2cm

55 Straw Man Design Study: V0 Starting Point Agree on basic inputs to Structures & Materials in Tracking volume Module Sensors: Silicon Physical Dimensions: 9.6cm * 9.6cm * 320um ROC: Silicon 9.4cm * 9.4cm total coverage; 25um thickness Interposer/Interconnect: Silicon 11cm * 9.4cm * module separation within Doublet Remove ~ 50% of the material Cable & Connector 1 * 60 pin Flex Kapton Cables with low mass connectors Mechanics Protective Kapton envelope covering sensors: 50um thick High TC Carbon Composite heat spreader plates: 2 * 11cm * 10cm * 300~500um Remove ~ 50% of the material Power Dissipation: < 4W / Module + 0.8W / DC-DC ~ 5W Module + DC-DC <15mW/cm 2 ROC; < 3mW/cm 2 Data Transmission; i < 7mW/cm 2 Digitalit

56 Straw Man Design Study: V0 Starting Point Sensor to Read-Out Chip (ROC) Connections pitch ~ 100um * 1mm ROC to Interposer Connections density 40 ~ 100 / cm 2 Use ROC through -vias ROC 1 to ROC 2 Connections density ~ 20 ~ 80 / cm 2 Use Interposer through-vias Carbon Fiber heat spreader ~ 320um thick ~ 12um thick ROC 1 Sensor 1 0,5mm ~ 2.mm thick Interposer ROC 2 Kapton Sensor protection Sensor 2

57 Straw Man Design Study: V0 Starting Point Agree on basic inputs to Structures & Materials in Tracking volume ROD Mechanics Box-like structure supporting doublet modules on inner and outer faces 1000um thick high modulus CFC, with 50% coverage (50% material removed) Cooling Assume a cooling circuit above & one below doublet => 2 circuits for a ROD Not required for power rating May be needed for temperature homogeneity Assume CO 2 cooling, with 1.5/1.3mm outer/inner diameter pipes p Assume Copper-Nickel (TOB) or Titanium (TEC) Assume pipe embedded in Al heat spreader ~ 1.25mm * 4mm * 90mm

58 Straw Man Design Study: V0 Starting Point Agree on basic inputs to Structures & Materials in Tracking volume Cables Twisted pair with each strand = 150um Al with 20um Cu cladding (Roland) Used for Data Transmission, Power Distribution, Controls & Monitoring Data Transmission Assumes 1Gbps data electrical data transmission 24 * twisted pair cables / doublet module (1/ chip) Power Distribution Assume 10% worst case power loss on 6m long cable, from PP1 to module Assumes doublet module power dissipation ~ 4W Assumes DC-DC with 6 / 1 Voltage step-down close to module, 80% Efficiency 4 * twisted pair cables / doublet module Controls 4 * twisted pair cables / doublet module

59 Straw Man Design Study: V0 Starting Point Agree on basic inputs to Structures & Materials in Tracking volume DC-DC converters: One pair of converters / Doublet Module One for Analogue and one for Digital power Each converter consists of An IC with a high voltage power transistor Silicon, 500um thick * 1cm * 1cm A Toroid FR4 PCB, 3mm thick, 1cm * 1cm, 50% of material removed Two Copper layers, each 20um thick, 1cm * 1 cm The pair of converters is housed on a PCB & shielded FR4 PCB, 500um thick, 3cm * 3cm Cu circuit, 4 layers, 10um thick, 3cm * 3cm, 50% coverage Cu shield, 1 layer, 20um thick, 2* (2cm * 2cm)

60 Straw Man Design Study: V0 Starting Point Agree on basic inputs to Structures & Materials in Tracking volume Mechanical Supports & End-Flanges For V0 assume ~ TOB 2 CF skins + spacers: 2mm thick CF, ~ 20% of material removed Manifolds & Connections Outside of Tracking volume

61 Straw Man Design Study: V0 Starting Point In the present Tracker, pipes & cable are light within Barrels They blow up when exit barrels: connectors, manifolds, routing There is a lot of Electronics : Inter-Connect Boards, Opto-hybrids, CCU etc

62 Straw Man Design Study: V0 Starting Point Preliminary Straw Man Material Budget indications based on initial set of assumptions Supports the potential advantage of long barrels: look forward to improved estimates Nb Red points are edges of short fwd barrels ~ 110% ~ 40% ~ 20%

63 Backup Slides

64 Straw Man Design Study: V0 Starting Point Agree on parameters to vary Pixel dimensions: phi pitch 80 ~ 100um ~ 120 Pixel dimensions: Z Length 1.0 ~ 1.5mm ~ 2.0 Module Separation within Doublet: 0.5 ~ 0.75mm ~ 1.0 Doublet Separation within ROD: 4.0 ~ 6.0cm ~ 8.0 Cluster Finding Efficiency: 90 ~ 98% ~ 99

65 Straw Man Design Study: Performance Potential Strawman Design Study Performance Potential

66 Straw Man Design Study: Performance Potential Tracking Trigger with Stacked-Doublet Tracker Single mu, pion, e, tau Inclusive Tracks in Min Bias SLHC Luminosity Environment L1 Trigger performance for Benchmark Channels

67 Straw Man Design Study: Performance Potential Hit Pairs within Stacks: Target data Reduction: 2 * 6 ~ 10 ~ 16 Get factor of 2 by transmitting single combined info for the cluster pair Range of Nominal Thresholds: ~ 1GeV ~ 1.5 Efficiency vs Pt over range of Nominal Thresholds Rate & reduction versus Nominal Threshold Purity versus Nominal Threshold Compare performance with/without cluster centroid interpolation

68 Straw Man Design Study: Performance Potential Track Vectors within Double Stacks Target data Reduction: 6 ~ 10 ~ 16 Range of Nominal Thresholds: 1.5 ~ 2GeV ~ 2.5 Efficiency vs Pt over range of Nominal Thresholds Rate & reduction versus Nominal Threshold Purity versus Nominal Threshold Pt Resolution Longitudinal Vertex Resolution vs Pt Compare performance with/without cluster centroid interpolation

69 Straw Man Design Study: Performance Potential Trigger Primitives from combined Double Stack Layers For 2 Inner Stacked Doublet Stations: require 1/2 Stations For all 3 Stacked Doublet Stations: compare 1/3 and 2/3 Stations Range of Nominal Low Pt Thresholds: 1.5 ~ 2GeV ~ 2.5 Range of Nominal High Pt Thresholds: 15 ~ 20GeV ~ 25 Efficiency vs Pt over range of Nominal Low & High Thresholds Rate versus Nominal Low & High Thresholds Purity versus Nominal Low & High Thresholds Pt Resolution Longitudinal Vertex Resolution vs Pt

70 Straw Man Design Study: Performance Potential Tracking with the Double Stack Straw Man Assume 4 Inner Pixel Layers, with 50um * 150um Pixels Single mu, pion, e, tau Inclusive Tracks in Min Bias SLHC Luminosity Environment Tracking Performance for Benchmark Channels Timing Performance Efficiency vs Pt & η Fake Rate vs Pt & η Pt Resolution vs Pt & η Impact Parameter & Longitudinal Vertex Resolution vs Pt & η

71 Straw Man Design Study: Performance Potential Using Kalman Filter CTF: Tracking with the Double Stack Straw Man Standard Tracking, with Pt > 1 GeV Iterative Tracking: 1 st iteration: Use only hits passing Low Pt Stack & Double Stack L1 cuts & reconstruct tracks with Pt > 1 GeV 2 nd iteration: add singlet hits to tracks already identified, refit etc. 3 rd iteration: Use remaining singlet hits, reconstruct tracks with Pt > 1GeV 4 th iteration: Use remaining singlet hits, reconstruct tracks with Pt > 0.5GeV

72 Straw Man Design Study: Performance Potential Tracking with the Stacked-Doublet Straw Man Develop algorithms adapted to Stacked-Doublet Layout Kalman Filter CTF starts from a seed, and combines seed/track hypothesis with compatible hit Could think of Kalman Filter Combinatorial Vector Track Finder (CTFV)? Treat Inner Pixels as Seed Generator Treat Double Stack Layers as Vector Generators Combine seed/track hypothesis with compatible Vector Iterative implementation, so finally use also singlet hits & extend to lowest Pt

73 Strawman Proposal Common usage Strawman Proposal Common usage

74 Straw Man Proposal from Wikipedia A "straw-man proposal", also known as an Aunt Sally, is a brainstormed simple business proposal intended to generate discussion of its disadvantages and to provoke the generation of new and better proposals. p Often, a straw man document will be prepared by one or two people prior to kicking off a larger project. In this way, the team can jump start their discussions with a document that is likely to contain many, but not all the key aspects to be discussed. As the document is revised, it may be given other edition names such as the more solid-sounding "stone-man", "iron-man", and so on, etc.

75 Straw Man Proposal from Mind Tools In a culture that values being right, the notion of constructing a straw man is difficult to embrace. Why spend time drafting something that, ultimately, isn't going to be used? If you can get past this perception you will be surprised at how useful the technique can be. One of its main advantages is that it forces you to do something. Taking too long to deliberate the merits of an idea or hypothesis can be costly, as you risk never making a decision at all. With a straw man, you force an early, if incomplete, decision. This ensures that a final decision will be reached because doing nothing means accepting a poor plan by default. Tip: Be very careful when you're using a Straw Man approach that people p understand what you're doing: The last thing you want is to develop a reputation for "coming up with half-baked ideas." Make sure that your document is clearly labeled as such, and that the people receive it understand what it is.

76 Straw Man Proposal from Mind Tools A straw man is also useful in ensuring that everyone involved has a tangible concept to work from. Otherwise, there is a risk that people are working with different pieces of the whole, different perceptions, and different, unstated assumptions, as they continue to research and discuss aspects of the idea or solution. The risk of using a straw man proposal is that, by definition, you are jumping to conclusions. Providing you are aware of this risk, you'll challenge, test, and retest the real solution and so use "jumping to a conclusion" as a vehicle to find a better conclusion.

77

78 Straw Man Layout: 3 Stacked Doublet Layers + Outer Tracker η Stacked Doublet L1 & Tracking Layers, with full acceptance up to η ~ 2.5: Each Layer provides 2 * 2 = 4 hits 3 Layers = 12 hits Minimize Stacked Doublet Layer Surface 2700 Outer Tracker: Optimized for Tracking No L1 functionality Introduces 3 rd System, in two flavors More Material in the Tracker Volume

79 Straw Man Layout: 2 Stacked Doublet + 3 Stacked Single Layers η Stacked Doublet L1 & Tracking Layers, 3St Stacked dsingle L1&T Tracking Layers: with full acceptance up to η ~ 2.5: Use Cluster Width Each Layer provides 2 * 2 = 4 hits Each Layer provides 2 * 1 = 2 hits 2 Layers = 8 hits 3 Layers = 6 hits Minimize Stacked Doublet Layer Surface Introduces 2 nd System flavor More Material in the Tracker Volume 2700

80 Full Stacked Trigger Tracker Straw Man Layout 12 Measurement Layers Organized in 3 Super-Layers Each Super-Layer consists of a Stack of Doublet Sensor Modules (4 measurement layers / Super-Layer) Inner Super-Layer ~ 30cm (Geometry of Inner Vtx layers?) Middle Super-Layer ~ 50cm Outer Super-Layer ~ 100cm

81 Full Stacked Trigger Tracker Straw Man Layout 12 Measurement Layers Organized in 3 Super-Layers Each Super-Layer consists of a Stack of Doublet Sensor Modules (4 measurement layers / Super-Layer) Can search for high Pt Track Stubs Independently in each Super-Layer Can Combine Super-Layers to ensure High Efficiency & Low Fake rate Can use for L1 Trigger And for Prompt Tracking at HLT

82 Full Stacked Trigger Tracker Straw Man Layout Material Budget Reduction Stack of Sensor Pairs provide opportunity for shared mechanics and services A Double-Sided ROD = 2 hits For 1.5 * X0 of Single-Sided ROD 6L Layers of fdouble Modules = 12hit hits For 9 * X0 of Single Module layer Current Tracker = 14 hits For 12 * X0 of Single Module layer (If all TOB - Like ) Stacking Doublets onto Beams could allow to further reduce X0 with respect to RODs?

83 Simulation and Performance Issues Basic Things to Check Hit Pair Pt Resolution & Discrimination Rate vs threshold Track Stub Pt Resolution & Discrimination Rate vs threshold Track Quality Combinatorial Complexity & Calculational Efficiency: L1 & HLT Fake Rate & Efficiency if require Single Hit Efficiency: 95%~99.5% 4/4 hits in sensor pair 1/2 vs 2/3 Track Stubs All the above varying the design parameters over the Plausible Range

84 Straw Man Design Study: V0 Starting Point In the present Tracker, pipes & cable are light within Barrels They blow up when exit barrels: connectors, manifolds, routing There is a lot of Electronics : Inter-Connect Boards, Opto-hybrids, CCU etc

85 Straw Man Design Study: V0 Starting Point Straw Man Material Budget estimate based on above assumptions: Supports the potential advantage of long barrels Straw Man design also does without Interconnects etc inside Tracker volume Nb Red points are edges of short fwd barrels ~ 160% No Pixels ~ 105% With present Pixels ~ 75% ~ 40% ~ 20% ~ 25%