Transcript Daniela Fabris

Test beam data analysis at
Padova/LNL
F. Antinori1, D. Fabris1, M. Lunardon1,2, S. Moretto1,2,
M. Cinausero3
1 INFN Sezione di Padova
2 Dep.of Physics of the University of Padova
3 INFN Laboratori Nazionali di Legnaro
D.Fabris - SPD Meeting May 2004
Contents
• Introduction
– Ideal precision of binary detectors
– Track incidence angles
•
•
•
•
•
Analysis program
Study of cluster size variations at 0º
A first look at the 20º data
Conclusions
Outlook
D.Fabris - SPD Meeting May 2004
Ideal precision of binary detectors
• In the simplest model of a position detector with a channel size
d and binary read-out, assuming that:
– each channel outputs is 1 if a particle has passed within the channel
boundaries, 0 otherwise
– the hit distribution inside each channel is uniform
the expected precision is given by the rms of a flat distribution:
 
d
12
• The precision can be improved if the charge is shared between
neighbouring pixels (clusters)
• For instance, in an ideal situation where the cluster size is only
determined by geometry and diffusion, neglecting dE/dx
fluctuations, for normal incidence, the minimum (best) precision,
obtained when the numbers of clusters of size 2 and 1 are the
same, is as much as a factor 2 better:

d
2 12
D.Fabris - SPD Meeting May 2004
• Numerically, for the ALICE SPD, with a pitch of 50 µm, we
have:
R = N(cl. size 2) / N(cl. size 1)
– for R = 0   ~ 14.4 µm
– for R = 1   ~ 7.2 µm
– simulation   ~ 12 µm
 Eventually, we should be able to extract a measurement of the
detector precision from the track residuals on the test plane
(plane 2) once:
– the alignment of the reference planes is optimized
– the effect of the tracking precision on the track residuals has been
evaluated
 As a preliminary step, we are studying the variation of the
cluster size distributions with the threshold, in order to see if
it is worthwhile to attempt an optimization
D.Fabris - SPD Meeting May 2004
Track incidence angles
 In the following, we evaluate incidence angles expected due to
the SPD geometry and the magnetic field
• from the geometry, with field off:
<j> = 0º (layer 1)
<j> = 18º (layer 2, due to the turbo layout)
– with a maximum deviation from the mean due to the finite rj width
of the ladder of about:
• 9º (layer 1)
• 5º (layer 2)
• the angles due to the deviation in the magnetic field are much
smaller:
j ≈ y’ ≈ x/r, where r(m) = p(GeV/c)/0.3/B(T) is the curvature;
for p = 1 GeV/c, B = 0.5 T, the deviation is about:
• 0.34º @ 4 cm (~ layer 1)
• 0.6º @ 7 cm (~ layer 2)
D.Fabris - SPD Meeting May 2004
 Summing up, we should consider incidence angles with a
spread of a few degrees around:
j = 0º (layer 1)
j = 18º (layer 2)
 For layer 1, at incidence around normal, the precision, as
we saw, is expected to be maximum when N(cl. size 1) ~
N(cl. size 2)
 For layer 2, given:
– the values of the incidence angles (tan j ~ 0.3)
– the geometry of the detector (pitch = ¼ thickness)
we should also have a look at N(cl. size 3) vs N(cl. size 2)
D.Fabris - SPD Meeting May 2004
Analysis program
 The aim is to study the variations of the
cluster size with the threshold
 For this study, we have implemented:
– masking of noisy pixels
– a simple, rough tracking program, in order to:
• reject events if no beam is reconstructed in the
reference planes
• control the detector efficiency on plane 2
D.Fabris - SPD Meeting May 2004
Proton/pion beam at 120 GeV
Setup # 3
D.Fabris - SPD Meeting May 2004
Definition of a mask for noisy pixels
Plane 1
Plane 2
With mask
D.Fabris - SPD Meeting May 2004
Plane 4
Analysis program ( I )
 Approximate alignment of the planes using the
centroids of the beam distributions
 Calibration in mm
0
1
1
Ch
mm
1
3
2
4
mm
Ch
D.Fabris - SPD Meeting May 2004
Analysis program ( II )
Considering the planes 0 and 3
as reference for the tracking
 Events with only 1 cluster on the
planes 0,1 and 3,4 and
with 0 or 1 clusters on plane 2
 Relative distances of the tracks in
planes 1, 3, 4 w.r.t. plane 0 within 3σ
of the distributions
Distance planes 0-3
mm
 Relative distances of the tracks in the
plane 0 and 2 within 5σ
 The beam must hit plane 2 away from
the plane borders (at least 10 rows,
2 cols)
D.Fabris - SPD Meeting May 2004
mm
Distance planes 0-2
Cluster Size variation with the Threshold
at 0o
Vth 80
Vth 200
Vth 120
Vth 210
D.Fabris - SPD Meeting May 2004
Vth 170
Vth 214
N(cl. size 2) / N(cl. size 1)
vs Threshold
2
1.5
1
Best resolution expected
around ~ Vth 210
0.5
0
0
50
100
150
200
250
Threshold (VTH)
1.2
Efficiency on plane 2 vs
Threshold
Max. efficiency for
Vth ≥ 130
efficiency
ratio cluster size 2/1
Cluster size variation at 0o
1
0.8
0.6
0.4
0.2
0
0
50
D.Fabris - SPD Meeting May 2004
100
150
Threshold (VTH)
200
250
First look at the 20o data
• Preliminary look at cluster size distributions
(without tracking conditions)
D.Fabris - SPD Meeting May 2004
Cluster Size variation with the Threshold at
20o
Vth 100
Vth 160
Vth 120
Vth 180
D.Fabris - SPD Meeting May 2004
Vth 140
Vth 210
N(cl. Size 2)/N(cl.Size 1)
vs Threshold
ratio clustre size 2/1
Cluster size variation at 20o
2
1.5
1
0.5
0
0
50
100
150
200
250
ratio cluster size 3/2
Threshold (VTH)
6
5
4
N(cl. Size 3)/N(cl. Size 2)
vs Threshold
3
2
1
0
0
50
100
150
200
250
Threshold (VTH)
D.Fabris - SPD Meeting May 2004
Conclusions
• The data show substantial variations of the cluster size with the
threshold, in both the 0º and the 20º settings. Therefore, there
may be room for optimization...
• This should be studied further, since it may be possible to
improve the precision to better than the simulation values
• We may eventually want run at different threshold settings on
layer 1 and layer 2
• During the next beam test, we may want to collect some more
data around 20º for various Vth settings around 180 mV
D.Fabris - SPD Meeting May 2004
Outlook
• In the sharing of the upcoming work, we
propose to take care of the analysis of the
20º data
D.Fabris - SPD Meeting May 2004