KEK Beam Test Analysis

1
KEK Beam Test Analysis

• Hideyuki Sakamoto
• MICE Collaboration Meeting, Osaka, 2006/03/01

2
CONTENTS

• Data Analysis
• VLPC stability study
• Mapping
• Simulation
• Light yield
• Efficiency
• Residual cut

3
VLPC stability study

• Checking items
• Pedestal mean
• Pedestal width
• Gain
• Noise rate

4
Checking Scheme

• For stable RUN (1257 - 1433)
• First 8 channels at each AFE board are checked.
• Each components are determined by fitting with
  triple Gaussian
• Pedestal peak P4 -0.0370.03
• Pedestal width P6 3.80.03
• Gain P5 P4 17.80.3
• Noise rate P2 / P1 0.030.002
• For low gain AFE (104, 107),
• only pedestal is checked.

P1
P2
P6
P4
P5
5
Common Mode

• In order to suppress the electric noise on AFE
  board, each channel is subtracted by common
  mode defined at each ADC chip.
• 2048 channels are divided by an unit of ADC chip
  (16 channels).
• Pedestal is defined at every ADC chip (so
  2048/16128 channels), as common mode.
• Common mode is an average of a summation of 12
  channels. (Here, two maximum and two minimum
  channels are rejected)

6
Pedestal distribution after common mode
subtraction
AFE 106 Channel 1
Common mode
RMS 1.5 counts
RAW DATA common mode
RAW DATA
RMS 6.8 counts
RMS 6.4 counts
Fitting this distribution by triple Gaussian
7
Mean of pedestal
Fluctuation is less than 1 count for all AFE.
8
Width of pedestal
RMS/MEAN 0.02
RMS/MEAN 0.02
RMS/MEAN 0.01
RMS/MEAN 0.01
Fluctuation is less than 2 for all AFE
9
Gain
AFE106
MEAN 12.1 RMS/MEAN 0.02
AFE105
MEAN 17.5 RMS/MEAN 0.02
Gain is stable within 2
10
Noise Rate
AFE106
MEAN0.12 RMS/MEAN0.05
AFE105
MEAN 0.03 RMS/MEAN0.08
Noise Rate is stable within 10
11
Lack of VLPC data

• We had 2048 channels data from VLPC per event,
  but data is less than 2048 at several runs.
• The reason why data is not 2048 is considered
  that connection of LVSB link cable was not stable

2000
Number of VLPC data
0
RUN NUMBER
RUN NUMBER
12
Summary of stability study

• The stabilities of pedestal peak, pedestal width,
  noise rate and gain about fist 8 channels of each
  AFE board are checked throughout RUN.
• All components are stable for almost RUN. The
  maximum values of fluctuation of each component
  are summarized as follows.
• There found the lack of VLPC data for several
  runs. This is considered that VLSB link cable was
  not unstable.

13
Mapping

• Checking items
• Checking scheme
• Original mapping
• Modified mapping
• Results by Arons check

14
Checking Scheme

• Using beam data
• 3 GeV/c, 0 deg-rotated (RUN1255,1299)
• Total number of events is 142,352.
• Mapping is checked by difference in fiber number
  between view of another station as a function of
  fiber number.

15
Original mapping
These fiber at these connector is assigned to
those of station A The confirmation like this way
is done for X view of all another station.
Station B X view
Fiber _at_Station A X view - Fiber _at_Station B X
view
Fiber _at_Station B X view
16
Modified mapping - version 12
Station B X view
Fiber _at_Station A X view - Fiber _at_Station B X
view
For in the same way
Fiber _at_Station B X view
17
Unknown
Station D X view
?
Residual (cm)
Fiber (bundle)
18
List of modification
Clear Fiber
SciFi

• Reversed order
• Connection of scifi into a MICE is reversed.
• Connectors for station D were newly
• designed to solve this problem
• Skipped by 1
• One (multiplexed) bundle of scifi is not
• connecting into a MICE connector hole.
• Swapped by 1
• One bundle of scifi is connecting into the next
• MCE connector hole.
• Swapped by 2
• Same as 3) but into two next hole.

19
Table of alignment error
Item number shown at slide17
Reversed order for these old stations is
consistent with Log book
Clockwise order whereas others are anticlockwise
20
Results by Arons check
21
Summary of Mapping

• Mapping check is done by beam data of 3GeV/c, 0
  deg-rotated RUN.
• There found four kinds of miss connection,
  reversed, 1-skipped, 1- and 2- swapped
  connection.
• Fixing the connection error at a level of 3HF is
  hard (impossible) even using beam data or
  cosmic-ray test.
• So, At the time of connecting 3HF bundle into
  MICE connector
• Careful Attachment and
• Correct Recoding
• are necessary !!

22
Simulation

• Calculating items
• Set-Up (Design value)
• Deposit Energy _at_ SciFi
• Residual Distribution _at_ Each Station
• Hit Efficiency per View
• Track Reconstruction
• Residual Of Reconstructed Momentum

23
Set-Up Detectors
1Tesla
24
Set-Up Fiber sheet

• Scifi.
• Diameter350um
• gt()core350umx88
• gt()clad350umx6(one side)
• Pitch420um
• Distance280um
• Materialscinti.
• gt()corescinti.(Polythylene)
• gt()cladPMMA
• () from Kuraray catalogue
• View
• Distance0.63mm
• MaterialAir
• gt polyurethane adhesive(PSILRA1000) (temporary)
• Mylar
• Thickness25um
• Material(Polyethylene terephthalate) C10H8n

25
Deposit Energy
Station B X-VIEW

• Deposit energy in a view is calculated
• There are two peaks in the distribution according
  to times passing through 3HF fiber

One bundle
One bundle
26
Light Yield Distribution MC

• Light yield is calculated by distributing deposit
  energy with Poisson and Gaussian
• Two parameters are assuming
• Production energy of one photon (MeV/photon)
• Gain

p.e.
27
Clustering
Without Clustering

• Clustering is adding the light yield of
  neighboring fiber bundle
• The ratio of crossing two bundle is
• about 1/3 to crossing only one bundle of
  fiber
• Gaussian mean
• 9.4 0.1 p.e.
• 10.1 0.3 p.e.
• Clustering contributes to light yield about by 1
• From next slide on, clustering is applied

With Clustering
28
Light Yield Distribution DATA
Station B X-VIEW

• DATA
• Momentum1.0 GeV/c
• Rotation0 degree
• B-Field 0 Tesla
• Run1224-1228
• Event selection
• Timing cut by L1 Acceptance
• Residual cut
• PID by TOF of D1 D2 counter

Pion
Proton
29
Comparison with DATA Proton
Station B X-VIEW
Agreement with DATA
Data is saturated around 50 p.e. , so events over
40 p.e. are unified by one bin.
30
Comparison with DATA Pion
Station B X-VIEW
Agreement with DATA
MC distribution is reproducing DATA
31
Efficiency vs Light Yield

• Estimating efficiency as a function of light
  yield at several threshold.
• Light yield is fitted mean of MC distribution.
• Efficiency is determined by the ratio of /event
  with the threshold to /event without threshold.

32
Residual Distribution MC
BEAM

• Residual at each station is calculated
• in the setup of vacuum and including
  materials
• Input momentum is 3 GeV/c

Vacuum
Including materials
33
Cut value for residual

• Determining cut values of residual for each
  station
• Hit events rejecting by residual cut are
  considered as noise hits.
• Or Mapping error
• Cut value
• Station B 0.2 cm
• Station A,C,D 0.1 cm

90
Vacuum
Including materials
34
Residual Distribution DATA
X view
Station A
Station B

• DATA
• Momentum-3.0 GeV/c
• Rotation0 degree
• B-Field 0 Tesla
• Run1255,1259
• Efficiency per view
• Station B 84
• Station A 94
• Station C 92
• Station D 88
• The reason why the efficiency is such small
  considered as
• Mapping error and/or
• Fiber or VLPC channel is dead
• Under checking

Station D
Station C
35
Track Reconstruction

• Assuming Uniform B-Field and NO scattering
• Chi2 to be minimized
• Input xi,yi, i1,4
• Output x0,y0,p0,R,L
• Degree of freedom 8-53

36
MC v.s. reconstructed pt pz
MC
Reconst.

• Why is Reconst. longitudinal momentum so broad?
• Bug?
• Effect of scattering?
• NOW checking this.

37
Example of failed
Failure due to small turning angle
True
Pz416MeV/c
? Dq(Dz95cm)39 deg
38
Broadness of fitted Pz
1

• This problem occurs when turning angle, Dq
  becomes small
• Dq Dz L -1 Dz 0.3 B Pz -1
• Small Dq ? Small B or Large Pz
• In MICE
• B4T,Pz200MeV/c ? L17cm gt Dq (Dz1m) 337deg
  lt NO Problem
• In T585
• B1T,Pz400MeV/c..etc.. ? 1/8 times as MICE42deg
• For analysis, using TOF for Pz reconstruction

2
3
4
Dq
39
chi2

• Chi2 assuming helical fitting
• Chi2 with TOF for pz determination

Only Pt is estimated by this chi2 Pz is
determined by TOF
40
Fitted Pz distribution
BLACKINPUT
Pz
REDchi2
GREENchi2 with TOF
41
Summary of Simulation

• Simulation for KEK beam test was held
• Light yield is calculated from distribution of
  energy deposit
• It is good agreement with DATA
• Efficiency is estimated as a function of light
  yield with several threshold
• Efficiency is smaller than that estimated by
  Poisson distribution (e.g. 97.0 lt 99.4 )
• Cut value for residual distribution is calculated
• For station B 0.2 cm
• For other station 0.1 cm
• In DATA, efficiency with residual cut is smaller
  than the expectation. One of the reason is
  considered as Mapping error and/or dead channels.
  Now under estimation.