SciFi Trigger Review: Physics Goals and Algorithms

1
SciFi Trigger ReviewPhysics Goals and Algorithms

• Kin Yip, Fermilab
• L1 Trigger Review, Jan. 22, 1999

• Physics Ingredients
• Min. PT
• CFT Efficiencies and Occupancies
• Simulation involved
• Algorithms (VHDL coding etc.)

2
The Fiber Tracker L1 Trigger Group

• LAFEX/CBPF, Rio de Janeiro, Brazil
• Mario Vaz - CBPF/LAFEX DEL/UFRJ
• University of California/Davis
• Sudhindra Mani
• Fermilab
• John Anderson,
• Linda Bagby,
• Fred Borcherding,
• Stefan Grünendahl,
• Dave Huffman,
• Marvin Johnson,
• Jamieson Olsen,
• Manuel Martin,
• Mike Matulik,
• Pat Sheahan,
• Kin Yip

3
Tevatron Impact

• Impact on DØ
• Integrated L Þ rad damage in tracking chambers Þ
  replace tracker
• Shorter bunch crossing interval Þ avoid pileup Þ
  electronics pipeline, faster trigger
• higher rate Þ need higher rejections (at every
  trigger level)

4
Physics Goals

• Precision Tests of the Standard Model
• precision measurements (top mass, W mass, ...)
• Constrain the Higgs Boson Mass
• improvements to B physics capabilities
• (Bs mixing, CP violation, )
• QCD with W, Z, photon
• Extend the Discovery Reach
• More data samples allow significant increases in
  the reach for SUSY, leptoquarks, W,Z,anomalous
  couplings, intermediate mass Higgs

• All these will benefit from the CFT tracking
  trigger to help trigger on electrons/muons etc.
  
• STT and Muon Trigger rely on the CFT track
  seeds
• PreShower uses CFT tracks to identify
  electrons/photons.

5
Minimum PT Number of Equations

• Minimum PT is set to be 1.5 GeV/c
• to reasonably cover B physics (etc.) spectrum
• J/? good for calibration
• tracks with lower PT would not pass thro
  calorimeter.
• ? Plot
• Due to 1.5 GeV and the fact that we use 8 CFT
  layers for trigger, the total no. of equations
  (allowed trajectories) in each CFT sector ?16000
  ( with the present geometry )
• ? Plot

6
Typical B physics spectrum
7
of Equations vs PT/offset
? NPT 1/PT
8
Tracking Trigger Overview
VLPC

• (1) Fiber light signals electronic signals
• (2) Feed all axial fibers into logic gates/cells
  in Programmable Logical Devices
• (3) Fiber hit pattern recognition to look for
  tracks consistent with momentum PT gt 1.5 GeV/c
• (4) Send out the track information to outside L1
  CFT

Trigger response for Z ee with 4 min.bias
9
Tracking Algorithm

• There are 80 sectors in CFT, each subtending
  4.5?
• Seamless tracking requires fiber sharing between
  nearest sectors
• Tracks with PT?1.5 GeV are contained within 2
  neighbor sectors
• Fiber hits are transmitted from a sector to
  either side for track matching.

Sector boundary
Track
CFT Sector 1
CFT Sector 2
No crack in tracking
10
Tracking Algorithm (cont.)
j
j -1

• Doublet Algorithm
• Form doublet bin by combining individual fiber
  hits in the inner and outer singlet layers.
• Or combines inner and outer layers AND with
  the NOT makes the bins non-overlapping.

Outer Singlet
Inner Singlet
j
f
f 0
Doublet
Doubletj NOT(Outerj) AND INNERj
OR Outerj1
It has been shown that doublet layer efficiency gt
99.5.
11
Tracking Algorithm (cont.)

• Basic algorithm matching hit patterns in all 8
  layers with a pre-programmed set of equations
• Compute trajectories ( equations ) analytically
  for all possible tracks for momentum PT ? 1.5 GeV
  and download them into PLDs on the FE board
• based on the fact PT ? magnetic field strength
  (2T) ? radius of curvature
• equation - a set of 8 fiber indices
• There are about gt16000 equations for each sector
  
• Algorithm uses 8 out of 8 doublet layers
• with an option to require only 7 out of 8 layers
  at highest PT later in the run
• Use the outermost layer (8th layer, H layer) as
  the anchor layer (reference layer) where there
  are 44 fibers in each sector.

12
Monte Carlo Simulation

• Monte Carlo simulation studies using the D?
  upgrade configuration have been done in various
  physics samples.
• Single electron/muon samples are used to tune
  the efficiency of the trigger algorithm. For PT
  gt3 GeV, efficiencies
• gt97 for muons and
• 95 for electrons,
• limited by multiple scattering and various
  radiation effects.
• A working (though preliminary) C code in Run
  II framework is available.
• Plot ? shows how the CFT trigger efficiencies
  when different sets of
• equations (belonging to certain PT thresholds)
  are used.

13
Track Binning

• PT binning yields sharper turn-on than offset
  binning
• offset (projection of H layer fiber hit on A
  layer) - A layer hit fiber in units of fibers

Eqn
15
15
20
50
14
Tracking Algorithm (cont.)

• From Monte Carlo simulation studies, we can
  limit ourselves by allowing only 2 tracks in each
  H layer fiber and 6 tracks in each of the 4 PT
  thresholds in each sector virtually without
  losing any tracking efficiency.
• Need 2 tracks because of extra hits at high
  luminosity which create a fake track (7 points on
  original track and 1 fake).
• Fake track can be higher or lower in PT than the
  real one.
• 90 only 1 track passes through a fiber
• 10 2 tracks pass through a fiber.
• Only 48 tracks per broadcaster.

Inefficiencies
15
Occupancy
Using samples with 8 interactions, average no. of
tracks 15 per event ? 0.19 / sector per
event non-overlapped only 8.8 per event or
0.11/sector per event
For samples with 2 interactions, applying (7
out of 8 ) would increase the number of tracks by
67.
16
Tracking Algorithm (cont.)

• Basic Algorithm is to match 8 fiber hits in 8
  layers (A,B,H) with the
• pre-programmed equations in PLDs
• Use PLDs (each with 100,000 logic gates) to
  handle the trigger logic
• Use HDL (Hardware Description Language) to
  implement the tracking logic like
• T1013172227323945 A10 AND B13 AND C17 AND
  D22 AND
• E27 AND F32 AND G39 AND H45
• ( There are gt16000 of them in each sector ! )

17
Tracking Algorithm (cont.)

• Put all equations in 4 PLDs according to the
  PT threshold
• The trigger codes in the PLD can be reprogrammed
  many times but the I/O pins to each PLDs are
  fixed
• All fiber signals in a sector ( with some from
  the neighbor sectors ) would be feeded into every
  PLD
• so that each PLD can handle tracks with the
  entire PT range
• gives us flexibility to change equations with
  different PT in a PLD
• We allow each PLD to have at most 2 threshold
  ranges of PT only.

18
Tracking Algorithm (cont.)

• Input fiber patterns are matched with the
  equations ? tracks at certain (PT,?) bin
• A matrix of PT ? ? (H fiber position)
• PT
• ?
• Scan the matrix horizontally in groups and put
  in priority encoder which outputs the indices of
  PT bins with the highest priority
• 1-D list of indices and concatenated in a binary
  tree structure down to a list of 6 (tracks with
  the highest PT) in each PT threshold
• A mixture of parallel/serial modes to reduce
  latency while keeping resources low.

19
Test Result

• Testing Result
• Algorithm and timing have been tested in the
  vendor software simulation and implemented in a
  trigger test board with PLDs
• The measured timing in the real PLDs agrees
  very well with the simulation and the result of
  the trigger logic is what is expected.

simulation result
Tracking logic completed
lt 85 ns
20
Output
Output CFT track information are formatted in
the FE board, combining cluster information from
Central Preshower (L1) 17 terms -- number of
tracks in each of 4 PT thresholds, track/cluster
matches, total number of hits etc. are sent via a
fast serial link to L1 trigger framework for a
final L1 decision A max. of 6 highest PT tracks
are sent to L1 muon trigger (within 800 ns) Up
to 24 tracks are pipelined for later readout to
L2 trigger Leaving FE boards, all available CFT
tracks are selected and combined into 6 global
lists of up to 48 tracks each, which are track
seeds for L2 trigger.
21
Trigger terms
22
Summary

• Reasonable algorithms/restrictions on the L1 CFT
  trigger system
• New technology such as VLPC trigger logic in
  PLDs meet
• Timing constraints
• allows feeding of L1 info to other detectors
  (muon preshower)
• Flexibility
• allows for adaptable thresholds/PT bins and
  changing equations
• Monte Carlo simulation in various physics samples
  have been performed to verify the trigger
  algorithm.

23
The End !!!
24
Trigger Schematic
25
The Detector

• calorimeter replacement of preamps/shapers
• muon system
• replacement of muon chamber readout electronics
• Iarocci drift tubes replace forward muon chambers
• central and forward scintillator pixel layers
  enhance trigger capability.
• DAQ trigger add track and vertex triggering,
  add buffering, add processing power
• central tracker
• 2 T supraconducting coil inside r70 cm
  calorimeter bore
• lead/scintillator preshower detector with
  fiber/VLPC readout
• 16 layer SciFi/VLPC tracker (80k channels)
• 4 barrel / 16 disk Silicon tracker (1M channels)
• forward tracker/preshower scintillator cells
  with fiber/VLPC readout

The DØ upgrade builds upon the strengths of the
existing detector (excellent calorimetry, muon
coverage) and augments it with a high resolution
Silicon/Scintillating Fiber tracker.
26
Visible Light Photon Counters

• Passage of charged tracks produces light in
  scintillating fibers
• VLPC converts the light produced to electronic
  signals which are then transferred to the Front
  End (FE) board
• On the FE board, signals are digitized and
  discriminated by the Silicon Vertex and custom
  (SIFT) chips
• Signals are further latched into the PLDs
  (Programmable Logical Devices) on the FE board to
  perform track finding logic.

Optical Connector
Scintillating Fiber
Mirror
Waveguide Fiber
Electrical Signal Out
Photodetector Cassette
VLPC
Cr
y
o
s
t
at
27
System Overview

• L1 System overview
• Front End Board (on cassette)
• VLPC signal discrimination
• track pattern matching
• track sorting list building per 4.5? sector
• Receiver Concentrator system (two crates)
• list building per octant (L1)
• list building per sextant (L2)
• L1-CFTTM and L2pp links
• L1 CFT Trigger Manager
• forms trigger (and/or) terms for L1

28
Trigger Rates
Using minbias events of 2 interactions 7 out 8
in (A B layers) ? 5 increase in the trigger
rate.
8 out of 8 ( from Bornali )
29
Truncation schemes
30
DØ Tracking
calorimeter cryostat

• Solenoid
• 2 Tesla superconducting
• Central Fiber Tracker (CFT)
• 16 doublet layers of Sci-Fi ribbon
• 8 axial (parallel to the z-axis) ? TRIGGER
• 8 stereo(2o pitch), not used in TRIGGER
• 76,800 830 ?m fibers (multiclad)
• coverage 20ltrlt52cm, polar angle to 22?
• In the radial plane, CFT is divided into 80
• sectors (4.5?)
• Silicon Tracker
• Preshowers
• Central
• Forward

1.1
1.7
50 cm
1.3 m
z-axis