A new Highly Selective First Level ATLAS Muon Trigger With MDT Chamber
Data for HL-LHC
S. Nowak, on behalf of the ATLAS Muon Collaboration
Max-Planck-Institut für Physik, Munich
Frontier Detectors for Frontier Physics - 13th Pisa Meeting on Advanced Detectors
Abstract
Highly selective first level triggers are essential for the physics programme of the ATLAS experiment
at the HL-LHC where the instantaneous luminosity will exceed the LHC’s instantaneous luminosity
by almost an order of magnitude. The ATLAS first level muon trigger rate is dominated by low
momentum sub-trigger threshold muons due to the poor momentum resolution at trigger level caused
by the moderate spatial resolution of the resistive plate and thin gap trigger chambers. This limitation
can be overcome by including the data of the precision muon drift tube chambers in the first level
trigger decision. This requires the implementation of a fast MDT read-out chain and a fast MDT track
reconstruction. A hardware demonstrator of the fast read-out chain was successfully tested under
HL-LHC operating conditions at CERN’s Gamma Irradiation Facility. It could be shown that the data
provided by the demonstrator can be processed with a fast track reconstruction algorithm on an ARM
CPU within the 6 microseconds latency of the first level ATLAS trigger anticipated for the HL-LHC.
New front-end electronics which include the fast read-out
chain were designed for the demonstrator. The fast read-out
is implemented in an FPGA and operates with the 40
MHz LHC clock corresponding to a time resolution of 25
ns compared to 0.78 ns of the standard read-out. In
the next version, the time resolution of the fast read-out
will be improved to 12.5 ns, which brings a substantial
improvement in the spatial resolution as can be seen below.
Histogram based pattern recognition.
The algorithm projects the approximate
hit positions in the direction perpendicular
to the expected track.
μ
Histogram with
projected hit
positions
y
α
z
Motivation
The large rate of low energy muons and the poor momentum resolution of the existing muon trigger
system will lead to very high trigger rates for the low momentum trigger [1, 2].
Efficiency
dσ/dpT (µbarn/GeV)
10–1
10–2
Drift Time Spectra with Fast and Standard Read-Out
|ηµ| < 2.7
0.5
√s = 7 GeV
10–3
T
ATLAS
10–5
10
20
30
µ
pT (GeV)
40
0
0
50
100
20
50
0
40
60
p (GeV)
TGC
8
6
400
500
600
700
800
Drifttime [ns]
0.96
0.94
MDT
Standard read-out (no background occupancy)
0.92
Standard read-out (10% background occupancy)
0.9
μ
μ
Standard read-out
Fast read-out
Simulation (40 MHz TDC)
Simulation (80 MHz TDC)
0.45
0.4
0.35
0.3
0.25
0.2
Fast read-out (10% background occupancy)
0.86
TGC
0.5
Fast read-out (no background occupancy)
0.88
0.15
0.84
0.1
0.82
0.05
0.8
0
2
4
6
8
10
12
14
0
0
16
Drift radius [mm]
2
4
6
8
10
12
14
Drift radius [mm]
The spatial resolution of the fast MDT read-out in the sub-millimeter range is to be compared to the
spatial resolution of the trigger chambers which is on the order of few cm. This is sufficient to achieve
the desired improvement in the momentum resolution.
Barrel inner layer
4
300
0.98
Barrel middle layer
Small Wheel
200
1
include MDT chambers
to improve angular
resolution to 1 mrad
Barrel outer layer
10
Phase-II
Outer Wheel
Hit-Efficiency
The momentum resolution of the existing muon trigger system is defined by the spatial/angular
resolution of the trigger chambers. At the High-Luminosity LHC it will become possible to use in
addition information of the precision tracking chambers, the Monitored Drift Tube (MDT) chambers,
already in the first level trigger decision. This becomes possible due to an increased trigger latency.
Big Wheel
100
Comparison of the efficiency and spatial resolution between the fast read-out with 40/80 MHz TDC and
the standard read-out [3]:
Including the MDT Chambers in the Level-1 Trigger
12 m
0
Performance of the Fast Read-Out Chain
T
y
Fast read-out
150
MU11
MU20
MU40
Good agreement in shape.
I Rising edge with fast read-out slightly wider due to the limited
resolution.
I No discrete 25 ns time steps visible in fast read-out because
measurement was performed with cosmic ray muons which are
not correlated to the fixed 40 MHz clock.
I
Standard read-out
300
200
Hi-p trigger efficiency
0
HPTDC
Entries
104564
Mean
225.9
191.6
RMS
350
250
10–4
10–6
400
Resolution [mm]
1
Drift radius
1
Entries
c→µ
b→µ
t→µ
W→µ
Z/γ* → µ
π/K → µ
Shower muons
Punch-through
10
Tube in ROI
Hit positions (two−fold ambiguity)
End-cap
magnet
3 TGC layers provide
3 mrad angular resolution
Performance Study of the Fast Tracking Algorithm with Cosmic Ray Muon Tracks
z
2
4
6
8
10
12
14
End-cap MDT chamber
Barrel MDT chamber
End-cap TGC chamber
Barrel RPC chamber
16
18
20
End-cap CSC chamber
Technical Implementation
0.08
no background
occupancy
0 mrad seed
angle resolution
0.07
92.4±0.3%
0.06
[Simulation]
0.05
0.04
An additional fast read-out chain with reduced time resolution is needed in the MDT front-end electronics.
[92.9±0.1%]
Data within
3 mrad
Data outside
3 mrad
0.03
MDT chamber
ASD
24
R/O
part
TDC
Trigger
part
TGC chambers
GBT
FELIX
Trigger
logic
“Fast” R/O with moderate
time resolution (12.5 ns),
corresp. to ~ 0.12 mm
spatial resolution
0
iation Facility (GIF) - 2013/2014
-8
-6
-4
-2
0
2
4
8
10
αrec-αref [mrad]
Filters
-6
-4
-2
0
2
4
6
8
10
αrec-αref [mrad]
2
1.5
1
0.5
0
80 MHz TDC
data for track reconstruction simulated
0.35
0.3
No background occupancy (8715 events)
0.25
14.0% background occupancy (10098 events)
Find peak
Return 0
0.2
0.45
0.4
No background occupancy (40000 events)
0.35
10.0% background occupancy (40000 events)
0.3
20.0% background occupancy (40000 events)
0.25
0.15
Fit tracks
0.5
0.2
0.15
0.1
0.1
Return results
ARM Cortex-M4F
implementation
0.05
0.05
0
1400
1500
1600
1700
1800
1900
2000
2100
2200
0
1500
1600
1700
1800
Time [processor clock cycles]
1900
2000
2100
[1] ATLAS Collaboration, ATLAS muon spectrometer: Technical Design Report, CERN/LHCC/97-22,
May 1997.
Reference
chamber
[2] http://twiki.cern.ch/twiki/bin/view/AtlasPublic/MuonTriggerPlots
MDT chamber used for test
6 tube layers, 50 cm length
MDT chamber used for test
No muon beam in the GIF → use cosmic muons
6 tube layers, 50 cm length
Fast read-out and normal read-out are triggered by scintillators
[3] S. Nowak (ATLAS Muon Collaboration), A Muon Trigger with high pT-resolution for Phase-II of the
LHC Upgrade, based on the ATLAS Muon Drift Tube Chambers, Proceedings of Technology and
Instrumentation in Particle Physics 2014 Conference, PoS (TIPP2014) 205.
nowak@mppmu.mpg.de
2200
Time [processor clock cycles]
Bibliography
MDT chamber
MDT chamber
Lead
-8
Required processing time increases with the amount of cavern background.
I Required processing time <2200 clock cycles (11 µs; ∼6µs for ARM Cortex A9).
ers
Lead
-10
I
Scintillator
50 cm
Scintillator
[7.3±0.2%]
19.5% background occupancy (5517 events)
μ (cosmic)
Scintillator
Primary trigger
Calculate distance d
and fill histogram
An MDT chamber (photograph) equipped with new front-end electronics including the additional fast
read-out chain was used for track reconstruction of cosmic muons under photon irradiation in the CERN
μ (cosmic)
Goal:
of (GIF).
efficiency
and chamber
resolution
GammaMeasurement
Irradiation Facility
An sMDT
with half the tube diameter mounted below and
shielded
from the photon source served as reference tracker.
Scintillator
Lead
0.03
Data outside
3 mrad
40 MHz TDC
data for track reconstruction taken at GIF
Probability
FRO Input Buffer
(FIB)
urement
of
efficiency
and
resolution
CERN Gamma Irradiation Facility (GIF) - 2013/2014
500 GBq
Data within
3 mrad
The fast track reconstruction algorithm has been implemented in ARM Cortex-M4F (200 MHz)
assembler (processing time simulation with Keil µVision).
Demonstrator Setup Used for Cosmic Muon Track Reconstruction in the CERN
TestGamma
of new
hardware
Irradiation
Facilityand
(GIF) fast tracking algorithm
Reference
chamber
Lead
0.04
0
6
FRO Data stream
Cs
[80.9±0.3%]
Latency Estimation for the Fast Tracking Algorithm
ASD: Amplifier Shaper Discriminator → Analog read-out chip
TDC: Time to Digital Converter → Drift time measurement
GBT: Gigabit Transceiver → Optical link
FELIX: Front-End Link Interface eXchange → Interface to data processing
137
[Simulation]
0.01
2
1.5
1
0.5
0
Primary trigger candidates
Source
77.4±0.5%
0.05
7.7±0.3%
0.01
-10
0.06
0.02
[6.1±0.1%]
0.02
“Slow” R/O with high
time resolution (< 1 ns),
corresp. to ~ 0.08 mm
spatial resolution
14.0% background 0 mrad seed
occupancy
angle resolution
0.07
6.0±0.3%
Ratio
On-chamber data
concentrator
MDT Frontend Board
Counting Room
Ratio
Experimental Hall
Quality of reconstructed tracks
defined by the deviation of the
reconstructed slope (αrec ) from
the true one (αref )
I Efficiency: Successfully fitted
tracks (fast tracking algorithm)
I Simulation with Geant4
(numbers in brackets)
I Poor efficiency for high
background occupancy due to
preliminary long dead time
(1500 ns) of the fast read-out
chain (200 ns dead time due to
hit buffering already
implemented in the newest
demonstrator setup version)
I
Probability
0
Region of Interest
defined by TGC
Standardised events
0
Standardised events
2
w hardware and fast tracking algorithm
0 cm
Fast Track Finding Algorithm
New Front-End with Fast Read-Out Chain