BABAR Risks and Mitigations David B. MacFarlane B Factory Operations Review April 26, 2006
Jan 31, 2016
BABAR Risks and Mitigations
David B. MacFarlaneB Factory Operations
ReviewApril 26, 2006
April 26, 2006 BABAR Risks and Mitigations 2
Sources of risk
Installation and schedule risk associated with replacing barrel RPCs with LST moduleso Extensive discussion of planning and issues in Bill
Wisniewski’s talk
Radiation damage and occupancy in detector hardware systemso Extensive discussion for SVT in Bill Wisniewski’s talk;
background studies in Matt Weaver’s talk
Ability to maintain an open trigger for full physics potentialo Discussed here
April 26, 2006 BABAR Risks and Mitigations 3
Risks to the Detector
Radiation Damageo SVT sensors, readout electronics (S/N degradation, shorts)
• Replacements available for horizontal modules, but will not be installed based on risk/benefit analysis
• Remainder of detector will remain operational through at least 2008
o DCH damage to wires (Malter effect): lifetime well beyond 2008
o EMC damage to crystals (color centers): lifetime well beyond 2008
o Continuing to monitor damage• Have performed extensive irradiation studies,
understand limits of SVT very well• Maintain tight control of beam abort and injection
inhibits (relaxed to improve data collection efficiency and machine performance once understood)
April 26, 2006 BABAR Risks and Mitigations 4
Risks to the Detector
Data-taking inefficiency & dead-time o Services: system backups (chillers) (continuing)o DAQ: upgrade to online farm (replacement funded and in
planning)o Trigger: upgrade for z info (successfully deployed and operational
since the end of Run 4)o DCH: data transmission bottleneck (successfully deployed and
operational since February)• Understanding other possible bottlenecks, e.g., SVT, EMC• Additional tools: shielding & restrictive trigger & understanding
Machine Detector Interface Groupo Existed during BABAR construction & commissioningo Re-established with added scope in 2003
• New mandate includes working with accelerator team on machine background simulations, beam parameters at the collision point, instrumentation & analysis
o Addresses issues of extrapolation and modeling of backgrounds that contribute to radiation damage and data-taking inefficiencies
April 26, 2006 BABAR Risks and Mitigations 5
Trigger Rate Projections From the Trigger Group: need <140us
Actual experience in
2005 and 2006 has proven better than
extrapolation
Extrapolation prior to DCZ
deployment: will revisit this spring
April 26, 2006 BABAR Risks and Mitigations 6
Fiber Transfer Bottleneck
Pro
cess
ing
Tim
e
ROM (only 2 for EMC: endcap/barrel)
Eliminated by DCH readout
upgrade
April 26, 2006 BABAR Risks and Mitigations 7
Deadti
me (
%)
Trigger Rate (Hz)
Front-end Readout
(4 buffers)
DCH Feature
Extraction Bottleneck
Deadtime problem was foreseen in DAQ projections
Installed for Run5a
Factor 2 improvement
Installed for Run5b but not activated in October
Present DAQ limit seen in October
Phase II DCH upgrade
Phase I DCH upgrade
FEX code now installed and operational on DCH
endplate
Actual soft rise in deadtime somewhat faster than model predicts: under
investigation
April 26, 2006 BABAR Risks and Mitigations 8
Behavior of Fiber Bottleneck
Deadtime behavior on this plot is worse if L1 more “bursty” than Poisson (current evidence suggests not Poisson).
Current performance
equivalent to ~3 FE buffers: under
investigation
April 26, 2006 BABAR Risks and Mitigations 9
Possible Plan of Attack for Fiber SVT:
o Readout specific bad modules from both left and right. Factor of 2 gain? Might not work everywhere (especially for damaged modules?)
o Mask out small regions with high occupancyo Try running system at 60MHz (requires substantial new
effort, may not work)o Reduce occupancy with thresholds
Plan: Investigate 60MHz clock rate for data acquisition, thresholds, and masking
techniquesImpact: Should allow L1 rates up to 5Khz
April 26, 2006 BABAR Risks and Mitigations 10
FEX Bottleneck
140us
Eliminated by DCH readout
upgrade
Can be improved Challenging
April 26, 2006 BABAR Risks and Mitigations 11
Plan of Attack for FEX EMC FEX hard! Already a lot of work on this, not likely
to be substantially improved.o New CPUs won't work easily: mechanical, electrical, software
issues. significant work and money ($300K).o May be possible to pass EMC data to a small secondary farm, but
substantial work would be required to investigate and deploy DRC and SVT FEX relatively easy (don’t “do” anything)
but still have to be validated carefully. EMT FEX: Event “prescaling” deployed, with expected
improvement Currently SVT, EMT, GLT deadtime behavior not as
predicted and under investigation
Plan: Removing EMC FEX limit may be prohibitive at this point, although some ideas are under
considerationImpact: L1 rates could be limited to 5Khz
April 26, 2006 BABAR Risks and Mitigations 12
VME Bottleneck
140us
OverestimatedWill re-split this crate
Plan: May be possible to speed up EMC
April 26, 2006 BABAR Risks and Mitigations 13
L1 Trigger System
April 26, 2006 BABAR Risks and Mitigations 14
L1 Trigger Primitives (I)
DCT primitives:o BLT coarse rf tracks with no Z/tan/Pt info
• A16: long track reaching SL10 (Pt>180MeV)
• B16: short track reaching SL5 (Pt>120MeV)
o ZPD 3D tracks with Z/tan/Pt info reaching SL7
• Z16: standard Z track (|Z|<12cm, |Pt|>200 MeV)
• Zt8: tight Z cut track (|Z|<10cm, |Pt|>200 MeV)
• Z’8: high Pt track (|Z|<15cm, |Pt|>800 MeV)
• Zk4: moderate Pt cut (|Z|<10cm, |Pt|>350 MeV)
April 26, 2006 BABAR Risks and Mitigations 15
L1 Trigger Primitives (II)
EMT primitives: o M20: strip energy sum MIP (>120MeV)
o G20: strip energy sum medium E (>300MeV)
o E20: strip energy sum high E (>800MeV)
o Y10: Backward barrel high E (>1 GeV)
IFT primitive: o U3: coded pattern number for various 2 muon and 1 muon
barrel/endcap hit topologies
April 26, 2006 BABAR Risks and Mitigations 16
This configuration is used throughout 2001-2004 runs.
The ‘Beam/beam’ contribution can also be due to low angle Bhabha debris.
Feb/02
April 26, 2006 BABAR Risks and Mitigations 17
Possible L1 configuration improvements
Present DCZ trigger
Additional 1Zn track for some
lines
April 26, 2006 BABAR Risks and Mitigations 18
Possible L1 configuration improvements
Example of tighter DCZ configuration: Add a requirement for a loose ZPD track = 1Zn, with either Pt>+0.8GeV/c or Pt<-0.25GeV/c
Current DCZ Test case BB generic 100.0% 100.0% B-> + B->X 99.85% 99.85% B-> + B->X 100.0% 100.0% cc 99.98% 99.96% uds 98.92% 98.78% Bhabha 99.94% 99.94% 99.74% 99.68% 98.58% 98.35%
(Hadronic final states: all events Leptonic final states: fiducial events)
Reduces L1 trigger rate by 13% with no impact on physics
acceptance
April 26, 2006 BABAR Risks and Mitigations 19
Conclusions
Radiation damage to hardware systems carefully monitored and controlledo Except for a small angular range in horizontal plane of the
SVT, all systems will continue to perform through and beyond 2008
o Occupancies also result in modest impact on efficiency, under investigation and study
Approaching a number limits for daq systemo Removing fiber limit for SVT next goal, as well as resolving
small discrepancies in model performance vs datao EMC FEX would be a substantial challenge, possibly
representing a limit for L1 trigger rate at ~5kHzo L1 trigger investigations ongoing; should be sufficient
handles with new DCZ trigger elements to keep rate below 5kHz with little or no physics impact