Simulation of the CSC Track Finder
D.Acosta, A.Madorsky, S.M.Wang
University of Florida
A.Atamanchuk, V.Golovstov, B.Razmyslovich
PNPI
CSC Sector Processor Review
July 1999
Report
on the results from the studies of the CSC
Track Finder simulation
CSC Track Finder simulation is written in Fortran
Interface with the ntuple produced by CMSIM
Emulate as close as possible to the design of the hardware
Most results from the Endcap Track Finder simulation. Some new results from the Overlap+Endcap
Track Finder simulation
CSC Muon Trigger Scheme
Sector Processor / Muon Track Finder
handle
the track primitives (LCTs) in a 60 sector
link LCTs into tracks
Measure Pt, and Send 3 best track candidates to the Muon Sorter
Sector Processor Block Diagram
Two Bunch
Crossing Analyzer
Input Data &
CCB Interface
Input Data
15x32
Clock
Bx 0
Reset
3x26
3x26
3x26
3x26
6x26
Data Line
Control Line
Downloading /
Readout Line
3x26
6x26
Input Data
15x18
1x9
1x9
32-4
32-2
32-1
EU3
2–3
1x9
31-4
31-2
31-1
EU2
1–3
2x18
23-4
23-3
23-1
EU1
1–2
2x18
EU4
2–4
TAU2
2 Bx
14 Bx
Total
2 Bx
Track Assembler Units
Link 33
ME4
ME3
Link 32
Link 31
TAU1
Link 23
22-4
22-3
22-1
Link22
21-4
21-3
21-1
Link21
ME2
9+5
9+5
9+5
9+5
9+5
9+5
ME1
FSU
Global
Buffer
(FIFO)
Track-Finder Processor. Block Diagram.
Final Selection Unit
Select
MUX
VME
BUS
Input
3x26
9U Custom
Backplane
Figure 7: Block diagram of the Sector Processor architecture.
3x26
EU5
3–4
33-4
33-2
33-1
3 Bx
PC
Serial Port
3x26
1 Bx
Output
Connector
Extrapolation Units
3 Bx
Downloading/
Readout
Interface
2 Bx
Pt
assignment
unit
1 Bx
Output
Assignment Unit
Two Bunch Crossing Analyzer : Analyze LCTs received in dierent bunch crossings (NOT in simulation)
Extrapolation Unit (EU) : Links LCTs in two CSC
stations together
Track Assembler Unit (TAU) : Use the extrapolation
results to form tracks
Final Selection Unit (FSU) : Selects 3 best track candidates
Assignment Unit : Determines the Pt of the selected
track candidates
Extrapolation Unit
(For Endcap Track Finder)
ME4
41
4
42
3
ME3
31
32
33
ME2
21
22
23
ME1
11
12
13
Perform all combinations of
1i $ 2k , 1i $ 3k , 2i $ 3k ,
2i $ 4k , 3i $ 4k ,
no 1i $ 4k
14
15
16
extrapolations :
Extrapolation Unit in Detail
η1
6
SM
η2
6
η1−η2
7
LUT
128 x 1
η1
6
&
LUT
Match η
6
64 x 1
LUT
7
1
η2
6
LUT
6
64 x 1
CMP
128 x 7
7
LUT
128 x 7
7
CMP
7
1
η∗,∆φ
PRE
3-2
η∗∗,∆φ
LUT
2
16 x 2
2
&
2 Bits
Input Data
52 Bits
η1
6
6
LUT
128 x 7
10
SM
φ2
10
φ1−φ2
η∗∗∗,∆φ
7
1
φ1
CMP
7
11
3
LUT
Match φ
&
8x1
6
LUT
64 x 6
6
SM
∆φ−ψ1
ψ1
6
ψ2
6
6
SM
Q1
3
Q2
3
AMB1
1
AMB2
1
7
6
LUT
64 x 2
∆φ−ψ2
LUT
Match ψ1
128 x 1
6
7
LUT
Quality
2 Q
Match ψ2
128 x 1
2
NAND
Track primitives are matched in Coarse Pt is assigned based on the
angle of the two track primitives
FIG.2. EXTRAPOLATION UNIT. BLOCK DIAGRAM.
dierence in the
(classied as either : Low Pt, Medium Pt, High Pt)
Only for extrapolations ME1-ME2, ME1-ME3. (Pt
resolution is poor for the other extrapolations)
A
simple version of the road nder is simulated.
Only looks at the relation between the bending angle
and the sign of Accelerator Muon bits are used to reject halo muons
Results of the extrapolation are in the form of Quality
codes
Data Streams to Track Assembler Units
33
33– 4
33 – 2
33 – 1
3 – 1, 3 – 2, 3 – 4
Extrapolations
32
32 – 4
32 – 2
Track Assembler
32 – 1
Unit
(TAU2)
Track types:
Stream 2
Extrapolation
3–4
31
3–2
31 – 4
3–1
31 – 2
Units
2–3–4
1–3–4
31 – 1
1–2–3
1–2–3–4
Track Assembler
23 – 4
Unit
(TAU1)
23 – 3
23
Track types:
23 – 1
2–4
Stream 1
2–3
2–1
2–3–4
1–2–4
22 – 4
1–2–3
22 – 3
2 – 1, 2 – 3, 2 – 4
Extrapolations
22
1–2–3–4
22 – 1
21 – 4
21 – 3
21
21 – 1
Data streams structure
Results
Streams
from extrapolations are sent to TAUs in 2
{ Stream 1 : 1
{ Stream 2 : 1
$ 2, 2 $ 3, 2 $ 4
$ 3, 2 $ 3, 3 $ 4
) TAU 1
) TAU 2
Track Assembler Units (TAU 1 and 2)
Block Diagram of the Track Assembler Unit
3
ME21 – ME3
3
ME21 – ME4
From Extrapolation Units
12 ME22 – ME1
3
ME22 – ME3
3
ME22 – ME4
12 ME23 – ME1
3
ME23 – ME3
3
ME23 – ME4
12 ME31 – ME1
3
ME31 – ME2
3
ME31 – ME4
12 ME32 – ME1
3
ME32 – ME2
3
ME32 – ME4
12 ME33 – ME1
3
ME33 – ME2
3
ME33 – ME4
LINK
21
9
LINK
22
9
LINK
23
9
LINK
31
9
LINK
32
9
LINK
33
9
SRAM
256Kx16
IDT
5
5
5
5
To Final Selection Unit
12 ME21 – ME1
5
5
5 bit absolute quality &
9 bit hit i.d. :
3 bits for ME1
2 bits for ME2
2 bits for ME3
2 bits for ME4
Quality
of the extrapolations are sent to LINK units
In TAU 1 (TAU 2) each LINK unit handles all the
extrapolations to a single LCT in Station 2
(Station 3)
Possible extrapolations going into one LINK unit
11 ! 21 31 ! 21 41 ! 21
12 ! 21 32 ! 21 42 ! 21
13 ! 21 33 ! 21 43 ! 21
14 ! 21
15 ! 21
16 ! 21
LINK unit
1
Q24
3
Q
1
1
1
LINK
1
2
21
1
Q
12
2
Q23
2
4
1
1
1
23
2
Q24
3
4
2
2
Q12
Q
3
12
3
unit is programmed to select extrapolations
with highest qualities to form tracks
Example for the above case :
Q112 > Q212 , Q312
Q123 > Q223
Q124 > Q224
) form a track using LCTs 11, 21, 31 and 41
Each LINK unit will form one track
Three possible tracks from each TAU
Final Selection Unit (FSU)
8 bits:
1st track segment number – 4 bits;
2nd track segment number – 4 bits.
9
Track 5
9
Track 4
9
MUX
8
Track 3
9
Track 2
9
Track 1
9
STREAM 1
8
Sel1 Sel2 Sel3
We should compare:
Track1-Track4; Track1-Track5;
Track1-Track6; Track2-Track4;
Track2-Track5; Track2-Track6;
Track3-Track4; Track3-Track5;
Track3-Track6 (9 bits as total)
10
10
10
9
9
9
9
9
Hit Number
Comparators
(9 Units)
9
9
Final
Decision
Unit
LUT
256Kx32
Track 6
Track 5
Track 4
Track 3
Track 2
Track 1
5
5
5
5
5
5
Extrapolations
Quality Comparators
(9 Units)
Stream1 Stream 2
From Track Assemling Unit
(Extrapolations Quality Part)
LCT ID Comparator
9
Final
Decision
Unit
LCT Quality Comparator
Final Selection Unit
To Data Extraction Multiplexer
Stream 2
Track 6
(if we need only 2 track segments
for Pt calculation)
8
Stream 1
From Track Assemling Unit
(Hit Number Part)
STREAM 2
Each track consists of 4 track
segments as maximum
⇓
6 Tracks has 24 track
segments
⇓
We need 10 (5+5)bits to
describe all possible
combinations
In the FSU of the TF simulation :
LCT Quality Comparator : compares the qualities of
the tracks found in STREAM 1 to the tracks found
in STREAM 2
LCT ID Comparator : compares the LCT IDs of the
tracks found in STREAM 1 to the tracks found in
STREAM 2
4i
Stream 1
3i
2i
1
Stream 2
i
4j
3j
2j
1j
ID Comparator is programmed to consider two tracks
are \identical" if both tracks share at least one
common LCT
Results from the comparators are sent to
Final Decision Unit to select 3 unique tracks of highest
quality. (These 3 track candidates should then be
forwarded to the Muon Sorter)
NOTE : In the actual hardware design both Quality and ID Comparators compare the qualities and
the LCT IDs of each track found in STREAM 1 and
STREAM 2 to all the other tracks found in both
STREAMS.
) 15 bits instead of 9 bits going from each comparator to the Final Decision Unit.
Pt Assignment Unit
A
simple two-station Pt assignment unit
Only compute the Pt for two modes
Pt from measured between ME1-ME2
Pt from measured between ME1-ME3
Pt
has nonlinear scale
eta
4
delta phi
12
mode
2
(1 = ME1 - ME2
2 = ME1 - ME3 )
Pt
Assignment
Unit
Pt
5
Results from the Simulation
The results were obtained from simulations using old
CSC LCT simulation
Extrapolation Eciency for Single Muon Events
%
Pt = 3 GeV
ME1-ME2 ME1-ME3
successful extrapolation
= 93.4
81.3
Low Pt
= 77.1
58.1
Medium Pt
= 14.5
19.5
High Pt
= 1.8
3.7
Pt = 5 GeV
successful extrapolation
Low Pt
Medium Pt
High Pt
=
=
=
=
99.1
44.8
53.2
1.2
98.1
46.4
46.8
4.9
Pt = 50 GeV
successful extrapolation
Low Pt
Medium Pt
High Pt
= 99.4
= 0.02
= 0.9
= 98.4
99.4
0.01
1.1
98.2
(Note :The CSC chamber eciency was not taken into
account)
Extrapolation Eciency of Single halo- events :
P = 100 GeV (Total # Evts = 9142)
%
successful extrapolation
= 0.01
Low Pt
= 0.00
Medium Pt
= 0.00
High Pt
= 0.01
Final Selection Unit
Eciency of FSU on single muon events
High eciency for high Pt muons ( 100%)
0:5% of events FSU found > 1 track.
bremsstrahlung,
delta rays
0.35
Y x10 (m)
{ Extra LCTs due to
0.3
0.25
0.2
0.15
0.1
0.1
0.15
0.2
0.25
0.3
X x10 (m)
{ Broken track due to sector boundary, failed extrapY x10 (m)
Y x10 (m)
olation
0.065
0.06
0.28
0.26
0.055
0.24
0.05
0.045
0.22
0.04
0.2
0.035
0.18
0.03
0.16
0.025
0.14
0.02
0.08
0.1
0.12
0.14
0.16
0.18
0.2
0.22
X x10 (m)
0.12
0.5
0.6
0.7
0.8
0.9
1
1.1
Z x10 (m)
Eff
Single Muon Track Finding Efficiency
1
0.75
Pt = 3 GeV/c
0.5
0.25
Eff
0
1
1.2
1.4
1.6
1.8
2
2.2
1.4
1.6
1.8
2
2.2
1.4
1.6
1.8
2
2.2
2.4
ηgen
1
0.75
Pt = 5 GeV/c
0.5
0.25
Eff
0
1
1.2
2.4
ηgen
1
0.75
Pt = 50 GeV/c
0.5
0.25
0
1
1.2
2.4
ηgen
Eff
For > 1:2
1
0.8
0.6
Pt = 5 GeV/c
0.4
0.2
0
0
50
100
150
200
250
Eff
(Pt 5 GeV η gt 1.2) FSU track (Phi Gen)
300
350
300
350
Φgen (deg)
1
0.8
0.6
Pt = 50 GeV/c
0.4
0.2
0
0
50
100
150
200
250
(Pt 50 GeV η gt 1.2) FSU track (Phi Gen)
Φgen (deg)
Contribution to fake tracks due to bremsstrahlung or
delta rays may be lesser than expected :
Old LCT simulation (Je/Darin) allows > 1 LCT
within 16 cathode strips or anode wires
Look at the dierence in of two LCTs in a single
station
For 10 chambers with 80 cathode strips,
) 16 strips = 2 in For 20 chambers with 80 cathode strips,
) 16 strips = 4 in Only a small fractions of single muon events have two
LCTS that are separated by more than 16 cathode
strips
Preliminary studies using new LCT simulation (Benn)
indicates only 0:04% single muon events have > 1
reconstructed tracks in the FSU
Pt from Pt Assignment Unit
Pt = 3 GeV/c
Pt = 10 GeV/c
Pt = 50 GeV/c
Pt = 100 GeV/c
Overlap + Endcap Track Finder
Include
extrapolations MB1-ME2, MB2-ME2
4 LCTs in MB1, MB2
3 LCTs in ME2 (LCT < 1:2)
Add in the third data stream, STREAM 3
Overlap
EU
B1-E2
B2-E2
TAU 3
Stream 3
Link 7
Link 8
Overlap
FSU
Id
Quality
Comp.
Link 9
6
Choose
Key off from ME2
3
Endcap
FSU
Extrapolation E. for Single Muon Events in Overlap
Pt = 5 GeV
successful extrapolation
Low Pt
Medium Pt
High Pt
%
MB1-ME2 MB2-ME2
= 100
100
= 48.9
70.9
= 39.9
3.1
= 11.2
26.0
Pt = 50 GeV
successful extrapolation
Low Pt
Medium Pt
High Pt
=
=
=
=
99.8
0.3
0.6
98.9
100
1.8
0.8
97.5
(Note :The chamber eciency was not taken into account)
Eff
Eciency in the Overlap + Endcap regions
1
(A)
0.8
0.6
Pt = 50 GeV/c
Old CSC LCT sim. CMSIM 114
0.4
0.2
0
1
1.2
1.4
1.6
1.8
2
2.2
Eff
FSU track Pt=50 GeV
1
2.4
ηgen
(B)
0.8
0.6
Pt = 50 GeV/c
New CSC LCT sim. CMSIM 116
0.4
0.2
0
1
1.2
1.4
1.6
1.8
2
(CMSIM116 old LCT) FSU track Pt=50 GeV
Lower eciency
2.2
2.4
ηgen
for (B) compare to (A), due to more
realistic simulation of the magnetic eld eect on the
charged particles in the Barrel muon chambers in CMSIM 116 (the overlap region), and lower eciency for
the new CSC LCT nder.
Eff
Eciency vs (1:05 < < 1:15)
1
0.8
0.6
0.4
0.2
0
0
10
20
30
FSU track (Phi Gen)
MC
40
50
60
Mod(Φgen,60) (deg)
sample, single muon at Pt = 50 GeV/c,
Old CSC LCT sim. CMSIM 114
The two dips at about 30 apart are due to the gaps
between the Barrel muon chambers
Summary/Plans
Modify the Track Finder simulation so that it is much
closer to the real hardware design
Implement the 3-stations Pt assignment