Title: Trigger/DAQ/DCS
1Trigger/DAQ/DCS
- TDMT on behalf of the Trigger/DAQ System
2LVL1 Introduction
O(1M) RPC/TGC channels
7000 calorimeter trigger towers
Muon trigger
Calorimeter trigger
Muon Barrel Trigger
Muon End-cap Trigger
Pre-Processor (analogue ? ET)
Muon-CTP Interface (MUCTPI)
Cluster Processor (e/g, t/h)
Jet / energy-sum Processor
Central Trigger Processor (CTP)
Timing, Trigger, Control (TTC)
ROD_BUSY
LTP
Timing, Trigger, Control (TTC)
ROD_BUSY
LTP
3LVL1 Calorimeter Trigger
- Preprocessor
- Preprocessor Module is being assembled and will
be debugged in March - essential for slice tests
- enough existing ASICs and MCMs work well enough
to do slice tests - ASIC resubmitted with minor design faults
corrected - final version expected back imminently
- MCM yield is currently lower than expected, due
to surface finish and faulty ASICs - must understand surface finish problem
- must test ASIC dies more thoroughly before
mounting them
4LVL1 Calorimeter Trigger
- Cluster Processor
- Cluster Processor Module design updated to
improve timing margins now being made - problems with badly made boards now hopefully
solved by finding "one-stop shop" firms with
better instrumentation and QA - Jet/Energy-sum Processor
- Jet/Energy Module re-designed to use Virtex-II
FPGA now being made - Common Modules
- Common Merger Module design is essentially final
no big changes needed - Readout Driver full-specification design is being
laid out - Handles readout to DAQ from all types of trigger
modules - Also handles RoIs to level-2
- Each ROD reads out a full crate of trigger modules
5LVL1 Calorimeter Trigger
- Have tested analogue receiver/monitor modules
from Pittsburgh and been in close contact with
them - Simulated LAr and TileCal signals sent via
receiver to Preprocessor analogue input circuit - TileCal signals sent to receiver in test-beam
- Further tests have been done by TileCal group
- Receiver circuit has been modified to handle
saturated pulses better - Discussions continue on use of LAr and TileCal
calibration signals for energy and timing
calibration of calorimeter trigger
6LVL1 Muon Trigger
Barrel and endcap systems successfully operated
in 25 ns test-beam last September endcap chain
included MUCTPI and CTPD
RPC
Off-detector
7LVL1 Muon Trigger
MDT
Off-detector
TGC
Near-detector
On-detector electronics
HPT
SSW
8LVL1 Muon Trigger
- Work going on to produce final version of all
parts of on-detector system in time for test-beam
this summer - Slave Board ASIC for endcap trigger
- Revised version recently submitted
- Coincidence Matrix ASIC for barrel trigger
- Design revisions ongoing hope to submit in
April - Final version of PAD boards for barrel trigger
- Prototypes of revised design being assembled
- Much design work still to be completed,
including - Many variants of PS boards for endcap trigger
(connectivity depends on detector region) - Off-detector electronics
- Has been given lower priority than on-detector
electronics since it is not critical for the
detector integration/installation schedule
9LVL1 Muon Trigger
- Production quantities of on-detector barrel
electronics needed as soon as possible for
integration with detectors - Pre-production of barrel Splitter boards
already available - Main production in preparation (orders for
components, etc ongoing) - Production of barrel Pad boards will start as
soon as possible - After completion of tendering process
- After checking of prototypes
- Schedule for production (and QA testing) of
on-detector electronics for both barrel and
endcap triggers is tight
10LVL1 Central Trigger Processor
CTP crate
- COM_backplane produced
- PIT/CAL_backplane produced
- CTP_MON tested successfully
- CTP_MI produced recently (see photo)
- CTP_CORE under design
- CTP_IN under design
- CTP_OUT under design
- CTP_CAL to be designed later
Plan
- Prototype with one module each
- no CTP_CAL
- CTP_CORE w/ reduced functionality
- Lab-tests Jul-Aug 2004
- Testbeam 25 ns period Sep 2004
- Final CTP beginning of 2005
CTP_MI
11LVL1 Central Trigger Processor
Trigger Menu the collection of the (256)
Trigger Items, formed from combinations of (256)
Trigger Conditions on the (160) Trigger Inputs.
LUT
CAM
Mask Prescaling Priority
256
256
160
Trigger Conditions
TriggerInputs
Trigger Items
programmable
programmable
programmable
160 inputs at any time selected from gt 160 on
input boards
Decode incoming multiplicities, map onto 256
Trigger Conditions
Each trigger item can be a combination of ALL 256
trigger conditions
- more trigger items (TDR limit was only 96)
- programmable LUTs/CAM (no FPGA re-configuration)
- larger flexibility to construct Trigger items
from all inputs
Compared to TDR
12HLT/DAQ/Controls
- Some highlights and examples of ongoing work
13HLT, DAQ and Controls TDR
Several meetings took place with the
LHCC referees since the last Plenary,
presentations and summaries are on the Web A
very major milestone for ATLAS is the positive
conclusion of the HLT/DAQ/Controls TDR review
process at the end of 2003
CERN/LHCC 2003-069/G-067 The LHCC finds both
the technology adopted and the procedures
proposed for the ATLAS HLT/DAQ/DCS to be adequate
to achieve the physics goals stated in the
Technical Proposal, and congratulates the ATLAS
Collaboration on the quality of the work
presented in the TDR. The LHCC therefore
recommends general approval of the ATLAS
HLT/DAQ/DCS TDR.
14Network overview
15Plans for large-scale test (control)
16Example of work in PESA group
mFast
- mFast latency reduced by 60 ms thanks to the new
data-access schema - fewer MDT hits to handle and process
- RPC and MDT data preparation improved
- total data prep time 800 ms
- data preparation takes the same amount of CPU
time of mFast
Total
The muon selection latency at Level-2 is now ok!
17Unified Monitoring Scheme (for test-beam)
18HLT Development Integration Testbeds
- LVL2 alone
- no algs ?Jan04 DF-06-00-gtDF 06-02 , Online 20
- EF alone
- no algs ?Jan-Feb04 nightly-gtDF 06-02 , Online 20
- HLT (LVL2EF) Integrated
- no algs ? Jan-Feb04 nightly-gtDF 06-02 , Online
20 - Large Scale tests
- Starts 1st March 04
- No algorithms
- Testbeam preparations
- In progress
- Testbeds with algorithms
- Begin in March with Offline R8
19I/O path for Read-Out System
- In TDR (30 June 2003)
- The optimisation of the ROS architecture will be
the subject of post-TDR studies using a
Read-Out Buffer (ROBIN) prototype implementing
bus-based (PCI) and switched-based (GEth) I/O
paths - Schedule and milestones to match ATLAS
commissioning - ROBIN Final Design Review completed
LHCC 31.05.04 - Final decision on ROS Input/Output path EB
31.12.03
20I/O path for Read-Out System
- On 11 Dec 2003, TDAQ decision was madeBus-Based
(BB) with Switch-Based (SB) as option for
increased scalability - ROBIN 3 Slink input - PCI and GEth output
- Max input retaining max output functionality
- High potential for scalability and upgrades
- The baseline implementation of the ROS would be
BB with an upgrade path to combined
BB and SB I/O or only SB I/O for future
upgrades. Full assessment of the potential
of the SB Read-Out will be done as soon
as possible (planning in preparation),
so as to be ready with a viable
upgrade if and when needed.
21ROBin schedule
- Measurements on existing prototype continuing
- Meeting required performance for final system
- Schedule for final module
- FDR schedule detailed on time for LHCC
milestone May 2004 - Final Protoypes Sep 2004
- Tendering starts Sep 2004
- PRR Oct 2004
- Production starts Feb 2005
- Production completed May 2005
22Software releases and testbeam
- DAQ release (targeted to TB-2004)
- LHCC milestone (see TDR) 30.04.04
- Online Software release 29.02.04
- Dataflow release 31.03.04
- Status
- Online Software release 00-21-00 done
on 09.02.04 - Dataflow release (including now ROD-Crate DAQ
and EF Dataflow) 00-07-00 01.03.04 - New HLT Release 13.02.04
- Release and software integration testing
- Continuously done in our HLT/DAQ test-beds
- Major testing step large scale scalability test
on 250 PCs starting next Monday for 4 weeks - Detector integration for TB-2004
- Global planning being finalized
- Pre-testbeam lab setup in Bdg 40, 5th floor -
Integration at RCD started
23DCS ?? Subdetectors
- Individual meetings with each subdetector have
been held to - Agree on deliverables provided by the central DCS
team to subdetector groups with the timescale - Define and document a DCS Baseline
- The point of connection between the subdetector
DCS and the central DCS is the Subdetector
Control Station (SCS) - Subdetectors are responsible for everything
towards the detector - Central DCS is responsible for the items towards
the control room
24Central DCS Deliverables
- SCADA software PVSS and JCOP Framework
- Connection software to DAQ (DDC)
- Tools and libraries for front-end read-out of
standard devices - All functions needed for operations from the
control room - Supervision of common experimental infrastructure
- Connection to Detector Safety System (DSS)
- Information exchange with the CERN services
- Information exchange with the LHC machine
25Subdetector responsibilities
- The subdetector groups are responsible for all
applications in the SCS and for all HW and SW in
the SCS and the Local Control Stations (LCS)
below - The CAN/ELMB hardware will be procured centrally
(on subdetector budget) - CAN/PCI interface (commercial, selected)
- ELMB (in production)
- CAN Power Supervisor (being prototyped)
- CAN cables (to be defined by subdetector)
26Issues
- Conditions DB ATLAS plans?
- For testbeam 2004 Lisbon API
- Configuration DB ATLAS plans?
- Changes in interaction with DAQ Run Control?
- Workspace for subdetector groups
- Commissioning and debugging in US(A)15 and
(where) on surface? - Where are the SCS located?
- DCS Network
- Security ? ? (remote) Accessibility
- Summary
- No major problems discovered
- Some issues to tackle
- DCS Baseline is defined and documented
- See http//agenda.cern.ch/displayLevel.php?fid3l8
9