MIT
TOF detector lecture, 19. august 2004
1
Time of Flight @ CDF (II)
reconstruction/simulation group
J. Beringer, A. Deisher, Ch. Doerr, M. Jones, E. Lipeles,
S. Menzemer, M. Shapiro, R. Snider, D. Usynin
calibration group
E. Palencia, J. Piedra, L. Scodellaro, I. Vila, R. Vilar
Outline:
• TOF Offline Calibration
• TOF Reconstruction
• Validation/TOF Performance
• Physics with TOF
S. Menzemer
MIT
TOF detector lecture, 19. august 2004
2
The TOF-Detector
279.5 cm
4 cm
4 cm
140 cm
z
• 216 scintillator bars (≈ 4 × 4 × 279.5 cm3 ),
as cylinder directly outside of the COT
• 1 photomultiplier-tube at each end
• PMT + readout electronics measure time
when collected charge is above threshold
• integrates the charge within a time window
(≈ 10 ns)
S. Menzemer
MIT
TOF detector lecture, 19. august 2004
3
Contribution to Time Measurement in the TOF System
tspeed = d /s
scintillator bar PMT
tof
t cable
TAC stop
−−> C s
TDC
t0
discriminator
+z
d = L/2 −
(L=280cm)
TAC start
Cs : common stop
t walk−correction = α /sqrt(Q)
TDC = Cs - (t0 + tof + ts + tc + tw )
S. Menzemer
MIT
TOF detector lecture, 19. august 2004
4
TOF Calibration
• Online Calibration
Pedestal subtraction and TAC calibration
• Time Walk Effect (slewing correction)
Due to dependence of discriminator time on pulse height
(≈ 2 ns)
• Speed of Signal Propagation in Bars
Time depend on z postion of hit along the bar (≈ 20 ns)
• Channel-to-Channel Time Offset
Includes cable length and clock signals (≈ 1 ns)
• Residual Correction in z Position along the bar
(≈ 200 ps)
(Design TOF resolution: 100 ps)
S. Menzemer
MIT
TOF detector lecture, 19. august 2004
5
TOF Calibration: Effective Speed of Light
• Time difference East - West vs. z
tE − tW = t0E − t0W − (twE (Q) − twW (Q)) + tcE − tcW + 2z/s
s: speed of signal propagation
teast - twest (ns)
Time walk dependence almost cancels in time difference of east/west.
The effective speed of light comes from the fitted slope.
Channel-by-Channel offset doesn’t affect the slope, just causes a shift.
The effective speed of light is fitted again after the time walk correction.
20
Run number range: 151509-151517
10
0
-10
C z = 14.85 ± 0.02 cm/ns
-20
-100
-50
0
Bar number 45 (East)
50
100
z (cm)
S. Menzemer
MIT
TOF detector lecture, 19. august 2004
6
• Time Walk Correction
Due to leading edge discriminator:
Larger pulses fire the discriminator
at earlier time than smaller pulses
During calibration event t0 is
not yet known.
Effect studied
with tracks passing through two
neighbouring bars (A,B)
αA
;
tA = t0A + tsA (z) + tcA − √
Q
αB
;
tB = t0B + tsB (z) + tcB − √
Q
αA ≈ α B
tA − tB ∝ α ∗ ( √ 1 − √ 1 );
QA
QB
t45 - t46 (ns)
TOF Calibration: Time Walk Correction
10
Run number range: 151509-151517
5
0
-5
α45 = 79.20 ± 0.83
-10
-0.04
-0.02
0
0.02
Pair bar numbers 45-46 (East)
0.04
θ45-46
S. Menzemer
MIT
TOF detector lecture, 19. august 2004
7
TOF-Occupancy
• TOF occupancy is about 30% (average b-event)
• Two thirds of the TOF signals are not related to a
reconstructed track → significant background even for isolated
tracks
Occupancy is present in data and MC
→ It is related to real particles in the detector
S. Menzemer
MIT
TOF detector lecture, 19. august 2004
8
TOF-Track Matching
• extrapolate track into tof bar
• calibrate time according to z position of intersection
• measured charge and time on east and west side after
calibration should be the same
2
• cut on χ =
(tE −tW )2
2 +σ 2
σE
W
+
(QE −QW )2
2
σ∆Q
• charge has to be calibrated for attenuation length
• single track matching efficiency: 74% (average b-event)
• single track matching purity: 96% (average b-event)
S. Menzemer
MIT
TOF detector lecture, 19. august 2004
9
Arrival-Time Resolution
Like to study arrival time resolution and track-matching independent of t0.
Isolate calibration and matching effects from t0 computation.
→ Study time difference of two muon legs of J/Ψ event (KS → ππ, ...)
∆t
=
=
(tofµ1 − tofexp.µ1 ) − (tofµ2 − tofexp.µ2 )
s
s
!
!
2
2
mµ
mµ
s1
s2
1+ 2
1+ 2
− t µ2 − t 0 −
t µ1 − t 0 −
c
p1
c
p2
arrival
arrival
Entries
Mean
Errors on momentum (p) and arclength
(s) are negligible compared to timing
resolution.
Arrival time resolution:
√
σt = σ ∆t / 2
√
σt = 156/ 2 ps = 110 ps
104
RMS
169632
-0.005212
0.2092
Underflow
817
Overflow
703
Integral
1.681e+05
χ2 / ndf
305.9 / 94
Prob
3
10
0
p0
7410 ± 26.1
p1
-0.002361 ± 0.000472
p2
0.1563 ± 0.0004
p3
412.5 ± 5.3
p4
-0.02875 ± 0.00354
p5
0.45 ± 0.00
102
-1 -0.8 -0.6 -0.4 -0.2 -0 0.2 0.4 0.6 0.8 1
(tofmu1-exp.tofmu1)-(tofmu2-exp.tofmu2)
S. Menzemer
MIT
TOF detector lecture, 19. august 2004
10
Efficiency-Purity Definition
• Efficiency: all J/Ψs where both muon fulfill matching cut
divided by number of J/Ψs with both muons fiducial tof tracks
• Combined χ2 of time and
charge works best
2
• Default cut: χ ≤ 14
• Single track efficiency: 74%
• Single track purity: 96%
Efficiency
• Purity: J/Ψ in narrow gaussian divided by all J/Ψs which pass
matching cuts
1
χ2
0.8
0.6
0.4
∆Τ
∆Q
0.2
0
0
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Purity
70% of all tracks are fiducial → absolute tof efficiency about 52%
S. Menzemer
MIT
TOF detector lecture, 19. august 2004
11
A Closer Look at the Tails (1)
Bad Muons:
|(T ofµ1 − T ofexp.µ1 ) − (T ofµ2 − T ofexp.µ2 )| ≥ 500ps &
|T ofµ1/2 − T ofexp.µ1/2 | ≥ |T ofµ2/1 − T ofexp.µ2/1 |
70
60
late arrival times
50
40
early arrival times
30
20
10
0
−2
−1.5
−1
−0.5
0
0.5
1
1.5
tof −exp. tof [ns]
2
We measure more often too late than too early times
S. Menzemer
MIT
TOF detector lecture, 19. august 2004
12
A Closer Look at the Tails (2)
Early tails:
• Another particle crosses the bar before the track we are interested in
→ time measurement is triggered too early
Late tails:
• Another particle crosses the bar while the
gate is still open. Additional charge makes
signal look larger as it is.
→ Time Walk Correction too small
→ corrected time is too late
Inefficiency:
• Another particle has triggered or polluted either the time or the charge
measurement
S. Menzemer
MIT
TOF detector lecture, 19. august 2004
13
t0 Computation
The bunches @ CDF are long:
σz ≈ 30 cm → ∆t = σz /c ≈ 1 ns >> σtof (≈ 110 ps)
25000
20000
15000
10000
5000
0
−10 −8
−6
−4
−2
0
2
4
6
8 10
t0 [ns]
Need to calculate t0 event by event (vertex by vertex) to achieve
good resolution.
Need at least two tof tracks excluding the track of interest in order
to compute unbiased t0 .
S. Menzemer
MIT
TOF detector lecture, 19. august 2004
14
t0 Likelihood-Fit
L = P(π)*L(π) + P(K)*L(K) + P(p)*L(p)
Likelihood fit function L(π/K/p) consists of two gaussian
distributions:
• mean: t0 = arrival time - expected time (π/K/p)
• narrow gaussian corresponding to tof resolution (≈ 110 ps)
• broad gaussian to take misreconstructed/mismatched tracks
into account (600 ps)
• ratio narrow/broad gaussian: 0.9:0.1
• weight Likelihood with particle probability obtained by COT
dE/dx and apriori probability (0.8:0.1:0.1)
S. Menzemer
MIT
TOF detector lecture, 19. august 2004
15
Define a Metric for t0 Studies
• Split number of tracks per vertex into two subsets with n/2 tracks
• Compute t0 for both subsets
• Plot the difference, fit the distribution with double gaussian
104
• resolution: size of narrow gaussian
√
divided by 2
103
• purity: square-root of the fraction
of vertices where |t01 − t02 | ≤ 3σ
102
−0.6
−0.4
−0.2
0
0.2
0.4
0.6
(t01−t02) [ns] (all tracks)
S. Menzemer
MIT
TOF detector lecture, 19. august 2004
16
t0 Resolution/Purity
Resolution [ps]
25000
75
20000
70
15000
65
10000
60
5000
55
0
5
10
15
20
number of tracks
50
2
Purity
0
0.99
3
4
5
6
7
8
9
number of tracks
The t0 resolution is a strong function
of number of tracks used in the fit
√
(statistical ∝ 1/ N )
0.98
0.97
0.96
Average t0 resolution is about 50 ps
0.95
2
3
4
5
6
7
8
9
number of tracks
Average t0 purity is about 99%.
S. Menzemer
MIT
TOF detector lecture, 19. august 2004
17
TOF Resolution
tof contains contributions from arrival time and t0 resolution
unbiased
Entries
4
10
Mean
RMS
χ 2 / ndf
3
10
early arrival time
or late t0
K looks like
π
291484
−0.002899
0.1808
539.9 / 94
p0
7139 ± 28.2
p1
−0.01259 ± 0.00040
p2
0.1186 ± 0.0004
p3
213.9 ± 6.6
p4
−0.0189 ± 0.0051
p5
0.4792 ± 0.0075
∆ time of flight [ns]
TOF measured − expected , unbiased
2
1.8
1.6
1.4
π/p
1.2
1
102
late arrival time
or early t0
π looks like K
−1 −0.8 −0.6 −0.4 −0.2 −0 0.2 0.4 0.6 0.8 1
[ns]
• σ = 119 ps
(design value: 100 ps)
K/p
0.8
0.6
0.4
Κ/π
2σ
0.2
0
0
0.5
1
1.5
2
2.5
3
3.5
4 4.5 5
pT [GeV/c]
• 4.5% in tails (|tof-exp.tof| ≥ 0.3 ns)
Separation of different particle species is a function of tof resolution BUT tails
cause background e.g. for kaon-tagging or CHAMPS search
S. Menzemer
MIT
TOF detector lecture, 19. august 2004
18
Protons: Λ → pπ
2000
450
pull_Proton
1800
400
pull_Pion
1600
350
pull_Kaon
1400
300
1200
2GeV ≤ pT ≤ 2.5GeV
250
200
150
400
50
400
350
300
200
−4
−3
−2
−1
0
1
2
3
4
5
0
1.1
pull_Pion
2.5GeV ≤ pT ≤ 3.5GeV
150
100
1.115
1.12
1.125
1.13
2
mass [GeV/c ]
1400
1200
50
1000
−4
−3
−2
−1
0
1
2
3
4
5
800
300
200
1.11
Λ → pπ candidates, mass[GeV]
pull_Kaon
200
250
1.105
pull_Proton
250
0
−5
800
600
100
0
−5
450
1000
pull_Proton
600
pull_Pion
400
pull_Kaon
200
pT ≥ 2.5GeV
150
100
2.5
3
3.5
4
4.5
5
5.5
6
6.5
7
p t [GeV/c]
proton pt[GeV/c]
50
0
−5
0
2
−4
−3
−2
−1
0
1
2
3
4
5
Armin Scheurer (Karlsruhe)
S. Menzemer
MIT
TOF detector lecture, 19. august 2004
19
Soft Pions: D ∗ → D 0 (→ Kπ)πs
π hyp.
K hyp.
1000
800
Events/ 1.00 MeV
D*→ D0 (→ K π) πs
π hyp.
K hyp.
1000
800
700
600
500
400
600
600
400
400
200
200
300
200
100
0
-30
-25
-20
-15
-10
-5
0
5
0
1.82 1.83 1.84 1.85 1.86 1.87 1.88 1.89 1.9
D0 Mass [GeV]
0
-30
10
-25
-20
-15
-10
-5
0
5
pull
10
pull
pt ≤ 0.5GeV
0.5GeV ≤ pt ≤ 0.75GeV
1800
180
300
π hyp.
K hyp.
250
160
140
200
150
100
1600
π hyp.
K hyp.
1400
1200
120
1000
100
800
80
600
60
400
200
40
0
0
50
20
0
-30
-25
-20
-15
-10
-5
0
D ∗ candidate, mass[GeV]
5
10
0
-30
-25
-20
-15
-10
-5
0
pull
0.75 ≤ pt ≤ 1.0GeV
5
1
1.5
2
2.5
3
3.5
pt [GeV]
10
pull
pt ≥ 1.0GeV
0.5
soft pion pt[GeV/c]
Kurt Rinnert (Karlsruhe)
S. Menzemer
MIT
TOF detector lecture, 19. august 2004
20
Combining TOF and dE/dx PID
separation of protons/pions (Λ → pπ)
Λ0b → Λc (→ pK − π − )π +
mass_all
N / 10 MeV/c 2
TOF works better for low momentum
tracks, COT dE/dx works for very low
or high momentum.
L~220pb
120
-1
2
M=5,612.3 ± 2.7 MeV/c
N = 332 ± 35
2
σ = 25 ± 3 MeV/c
100
80
60
40
20
0
5.0
5.2
5.4
5.6
5.8
M(Λcπ-)
6.0
2
[GeV/c ]
Mass of potential Λ0
b candidates
N / 10 MeV/c 2
mass_pid
L~220pb
70
-1
2
60
50
M=5,612.1± 1.9 MeV/c
N = 253 ± 21
2
σ=22.7±2.1 MeV/c
LHp>0.3
40
30
20
10
0
5.0
5.1
5.2
5.3
5.4
5.5
5.6
M(Λcπ-)
5.7
5.8
5.9
6.0
2
[GeV/c ]
Mass with combined PID for proton
Paula Squillacioti (Pisa)
Dmitry Litvintsev (FNAL)
S. Menzemer
MIT
TOF detector lecture, 19. august 2004
21
PID used @ CDF
• Background Rejection : e.g. penta quark search (proton)
• CHAMPS search (looking for very massive and slow particles)
• Monopole search (highly ionizing particles, curving along the B field (z)),
TOF trigger
• B-Flavour Tagging
• Opposite-Side-Kaon Tagging:
It is more likely that a B̄ meson contains a K − than a K + in the final
state
• Same-Side-Tagging (Bs )
Bs is likely to be accompanied close by a K + from fragmentation
b
B
b
q
c
s
u
K
B
b
q
c
s
u
K
b
Bs
s
s
+
Κ
u
u
S. Menzemer
MIT
TOF detector lecture, 19. august 2004
22
Summary
• TOF calibration and reconstruction are interwoven
• TOF inefficiency and impurity are caused by high occupancy in
the detector (30%, 10% track related)
• t0 has to be computed event-by-event
• TOF separation power does not only depend on resolution but
on tails of the measured arrival time and the t0 computation:
σ ≈ 120 ps, fraction of tails (|tof-tofexp | ≤ 300 ps) about 4.5%.
• TOF works well for low momentum (up to about 1.6 GeV) and
is complemented by dE/dx for higher momentum.
• PID is started to be used @ CDF
• Further development of TOF reconstruction code is going on
S. Menzemer