Test Beam Meeting April 24, 2000

1
Test Beam MeetingApril 24, 2000

• Spatial resolution studies
• Analysis cuts
• Cluster building
• Projection errors
• PFA comparison
• Detectors comparison
• Comparison Magnetic field dataand angle-scan
  data
• The threshold is about 4850 e-
• The B looks lower than 0.64T

2
Spatial Resolutionanalysis cuts
Cuts - Number of tracks - Reduction factor No
cuts 46375
100 Type1.eq.1 46375
100 Type2.eq.1 46375
100 Type5.eq.1 45261
97.6 Type6.eq.1 44222
97.7 Ntrks.eq.1
36534 82.6 Type7.eq.2
36513
99.94!!! 15ltav_row7lt50 24916
68.2 1ltav_col7lt9 16271
65.3 -------------------------------------
------------------------- Cluster building along
one column 15309
94.09 Width1.eq.2.or.Width3.eq.2 5786
37.8 Width1.eq.2.and.Width2.eq.2
4263 27.85

• Cut in red introduce potential statistical
  
  correlations
• Only charge-sharing along one column

3
Spatial ResolutionResidual width and cuts
Residual width for run number 6672
(FPIX0 p-stop at 10 degrees) --------------------
----------------------------------------- Black
cuts 5.37 ?m 5.282 ?m (
cluster building) 5.280 ?m (
ncols1) 4.720 ?m (
ncols1.and. xSSD cs2) All cuts 4.729 ?m
The cluster building deteriorates a little the
sigma
4
Spatial ResolutionCluster Building
Width7.eq.1 .and. abs(row7_1-av_row7_2)
.le.1 Width7.eq.2. .and. abs(row7_1-row7_2).le.2
.and. abs(col7_1-cols7_2).eq.0 So on up to
width7.eq.6.
I am demanding sorted hits in the Ntuple And
charge-sharing only along columns

• Sometimes q7 is not equal to the sum of
  
  elec7_1elec7_2
• If I demand q7 elec7_1elec7_2 the
  residual width improve a little
• I didnt ask q7 elec7_1elec7_2
• No correlations from the cluster building
  
  selection

5
Spatial ResolutionRMS sigma black cuts
Black cut
6
Spatial ResolutionProjection errors
Black cuts
All cuts
Demanding cs2 for x SSD is equivalent to cut on
the projection error
7
Spatial ResolutionProjection errors
No correlation introduced by demanding CS2 for x
SSD
8
Spatial ResolutionFPIX0 p-stop PFA comparison

• Used variable xf7 from ntuple to project locally
• Projection difference between Kalman fit and
  simple interpolation is about 0.87 ?m
• Not less than 4000 tracks used
• The simple interpolation gives 0.15 ?m better
  
  residual width
• The tuned generalized linear eta correction is
  quite good for 8 and 2 bits respects to the non
  linear.

9
Spatial ResolutionEta correction comparison

• Angle Ntuple FPIX0 p-stop FPIX0 p-spray
• 0 5.89 8.20
  7.5
• 5 12.31 15.34
  15.78
• 10 24.68 28.5
  28.45
• 28.97 37.7
  38.05
• Agreement with the Ntuple if
• 0 deg is 2.86 deg
• 5 deg is 6.3 deg
• 10 deg is 12.6 deg
• 15 deg never because Wmax29.3

10
Spatial ResolutionDetectors comparison

• Used generalized linear eta correction for FPIX0
• Used tuned 16 eta correction for FPIX1
• Discrepancy in FPIX0 p-spray due to charge
  losses?
  Partially yes.
• Discrepancy in FPIX1 p-stop due to thresholds?
  
  Partially yes.
• Geometry errors the z position is important.
• 15 degrees for FPIX0 p-spray still anomalous

11
Spatial ResolutionSharing along two columns
4
Sigma 4.65 ?m
The data looks good but very low statistics for
12
Comparison Magnetic field and angle-scan data

• The FPIX0 p-stop threshold is changed after the
  
  magnet was installed
• This can be seen by the ratio double/single in
  different setup
• Fpix0 p-stop 0 deg angle-scan
  r 0.457
• Fpix1 p-stop angle-scan
  r 0.474
• --------------------Magnet installed--------------
  -------
• Target
  r 0.330
• Fpix1 p-spray angle-scan
  r 0.316

13
Comparison Magnetic field and angle-scan data
0.64T

• Ratio double/total vs angle should be approx
  linear
• The extrapolated minimum ratio from B data
  
  used to find Qth
• The Lorentz angle slope is 2.65 smaller than
  
  angle-scan slope

B up to 0.64T 0.24T
14
Comparison Magnetic field and angle-scan data
0.24T
May be the approximation used (linearity) is not
valid?
15
Conclusions

• The cuts used do not introduce statistical
  correlations in the residual distribution
• The projection error is about 2.1 ?m in the
  Kalman fit for cs2 in x SSD ( I
  estimated 2.2 ?m for FPIX0
  angle_scan 2.4 ?m for FPIX1
  angle_scan 6.6 ?m for FPIX0 in B
• The digital algorithm generalized linear eta
  corrections for 1or2 CS values is a simple and
  good PFA for 2-8 bits, but I believe that a
  16 entry look-up table is the right thing to
  do for 2 bits
• The magnetic field data must be analyzed with
  a threshold of about 4850 e-, but still B must
  be reduced of about 2.65