Level1 Calorimeter Trigger Stephen Hillier University of Birmingham
Level-1 Calorimeter Trigger Stephen Hillier, University of Birmingham ATLAS Level-1 Calorimeter Trigger Collaboration: University of Birmingham, University of Heidelberg, University of Mainz, Queen Mary University of London, CCLRC Rutherford Appleton Laboratory, University of Stockholm Brief System Overview o Recent Highlights o o Cabling, Calibration Signals, Full Crate Tests Schedule o Summary o February 22 nd, 2006 Level-1 Calorimeter Trigger 1
Calorimeter Trigger Architecture Digitized Energies Features: Realtime Path: fixed latency pipeline Calorimeter Signals Massive parallelism Multi-purpose modules Heavily FPGA based Real-time Data Path Jet/Energy Processor 32 modules Merging 4 modules Readout Driver (ROD) 14 modules Readout Data Merged Results To CTP Region of Interest ROD 6 modules Region of Interest Data Five Main Types of Custom 9 U Modules February 22 nd, 2006 Merging 8 modules Preprocessor 124 modules Latency <1 µs for trigger electronics Many processing stages Cluster Processor 56 modules PPM CPM JEM CMM Level-1 Calorimeter Trigger ROD 2
Real-Time Modules Cluster Processor 56 modules Merging 8 modules Jet/Energy Processor 32 modules Merging 4 modules Preprocessor 124 modules Real-time Data Path February 22 nd, 2006 Level-1 Calorimeter Trigger 3
Real-Time Modules 4 Crates PPM Full specification prototypes fully tested o CPM CMM High density custom backplane (>1000 pins/slot) o High speed serial data: o JEM 8 Crates February 22 nd, 2006 2 Crates o o Level-1 Calorimeter Trigger 400 MBit/s differential 160 MBit/s single-ended 4
Analogue Input Chain 256 Tile Calorimeter Patch Panels 360 128 360 Analogue Trigger Tower Receivers Long Cables: Cavern to USA 15 Typically 40 -70 m Detector Responsibility 360 LAr, 256 Tile. Cal February 22 nd, 2006 152 Receiver To PPM Patch Panels Preprocessor Modules 136 Short Cables: Internal to Trigger Racks Typically 5 -8 m L 1 Calo Responsibility 776 in total Level-1 Calorimeter Trigger 5
Cabling: Problems and Solutions o Cable Quality o o Original LAr Nexans cable fine Tile Calorimeter cables poor o o o o Required quality time-consuming Lack of technician manpower Cable Routing o New tile cable ordered from Nexans Low impedance, differential skew, speed ‘Connectorization’ o o Thick, inflexible cables Tight rack and cable tray space Access problems for rack turbines, power supply units etc February 22 nd, 2006 Move to crimping method quicker than soldering o o o Delay in installation, but quality fine Still only 2/person/day Use outside company for 1000+ connectors Manpower freed for this effort Flexible ‘stock’ supports Access improved o Rear mountings, cable loops Level-1 Calorimeter Trigger 6
Cabling Example: top crates February 22 nd, 2006 Level-1 Calorimeter Trigger 7
Cabling Example: bottom crates February 22 nd, 2006 Level-1 Calorimeter Trigger 8
Calibration signals at Point 1 o Commissioning of calibration techniques using: o o o Prototype Preprocessor Modules Tile Calorimeter (Barrel) calibration signals Full patch panel and receiver connectivity Liquid Argon calibration signals (to do) First real signals we’ve seen since October 2004 (Combined Test Beam) February 22 nd, 2006 Level-1 Calorimeter Trigger 9
Calibration Software Development o Prototype of calibration runs being developed o o o Within standard T/DAQ framework In collaboration with detectors Essential functionality already present o Unpopulated Channels Calibration Pulses Requires refinement o o Eg Database access Detector communication February 22 nd, 2006 Level-1 Calorimeter Trigger 10
Calibration Runs and Tests o Timing calibration o o o Energy calibration o o o Scan in 1 ns steps Determine optimum sampling phase (40 MHz FADC) Scan over energy range Check response to saturated signals Connectivity testing o Confirm signals appear where they are expected February 22 nd, 2006 Level-1 Calorimeter Trigger 11
Full-Crate Production Tests o Cluster Processor Module production: first stage o o o Testing for 2 weeks now, no problems seen so far o o o 220 high-speed serial input data links run error free Zero errors in 1012 bits Finish testing soon (1 -2 weeks) o o o 12 already manufactured Performing full crate test at Birmingham Inter-module data transfer Outputs to merger modules Continue to full production run (~50 more modules) February 22 nd, 2006 Level-1 Calorimeter Trigger 12
Full Crate test infrastructure February 22 nd, 2006 Level-1 Calorimeter Trigger 13
Production Status o o o Custom Back-plane pre-production underway Cluster Processor Module production well advanced Common Merger Module pre-production finished o o Jet-Energy Module in pre-production o o o 4 modules due in mid-March Solved high-speed back-plane signal cross-talk problem Other infrastructure modules already manufactured or ready for full production o o No major problems in manufacture Timing modules, patch panels But Preprocessor Module and Readout Driver Modules are running late February 22 nd, 2006 Level-1 Calorimeter Trigger 14
Preprocessor (PPM) schedule o PPM components schedule OK o o PPM motherboard is late o o o All ASICs produced and tested MCM production and testing almost finished with encouraging high yield (90%) Problem with LVDS fan-out daughter-card now fixed Carrier for above components Large number required (124) Plan for first full crate in May/June o Progressively more until end 2006 February 22 nd, 2006 Level-1 Calorimeter Trigger 15
Readout Driver (ROD) schedule o Two full-scale RODs tested for some time o o o No real hardware problems seen after extensive testing Two further RODs manufactured However, firmware still requires work o o Multi-purpose module, many firmware variants (13 data formats) Flow control has been problematical o o Recent testing at RAL is looking better Plan for FDR/PRR in March or April February 22 nd, 2006 Level-1 Calorimeter Trigger 16
Summary o Much encouraging progress over last few months o o o Still a long way to go o Initial installation experience in USA 15 Cabling taking shape Solutions found to remaining hardware problems Major production of components and modules Challenging production and testing schedule ahead Two big items (PPM and ROD) on critical path Emphasis moving from outside institutes to CERN o Now three residents at CERN, soon to be five February 22 nd, 2006 Level-1 Calorimeter Trigger 17
- Slides: 17