Status of TPC experiment ---Online & Offline
M. Niiyama
H. Fujimura
D.S. Ahn
W.C. Chang
Results from cosmic ray test
•
•
From cosmic ray test, 1 track resolution is
- 300 um in pad plane w/ magnet
- 700 um in drift direction w/o magnet.
If we apply magnetic field, residual in drift direction
shifts layer by layer. Possible reason are
- circle fitting is not suitable for our solenoid magnet.
-> runge kutta is needed
- E x B effect disturbs drift property of electrons in TPC.
Bx at z=350mm is 0.5T
-> simulation of drift property is needed.
Bx at z=350mm
E&M Calculation
E // B  Cycloid Motion
Install & Geometry
Top view
Shield
Collimator
target
20cm
solenoid
Pb
g
Dipole magnet
TPC
32.4cm
151.6 cm
40.7cm
37.5 cm
February run
ＥＸＰ．ＨＵＴＣＨ（ February ）
dipole magnet
Side
scintillators
Ｓｈｉｅｌｄ
Ｃｏｌｌｉｍａｔｏｒ
Ｕｐｖｅｔｏ
Downstream
scintillators
March run
ＥＸＰ．ＨＵＴＣＨ（Ｍａｒｃｈ）
dipole magnet
Side
scintillators
Ｖｅｔｏ ＠ＴＯＦ
Ｓｔａｒｔ ｃｏｕｎｔｅｒ
Ｓｈｉｅｌｄ
Ｃｏｌｌｉｍａｔｏｒ
Ｕｐｖｅｔｏ
Downstream
scintillators
Trigger condition
•
Run in Feb, to know performance of TPC
tag x (up-veto) x (8scinti M>=2) x (TPC side scinti M>=1)
95Hz at 400kHz tagger
• Run in Mar, Particle ID by spectrometer calibration of TPC.
gamma-> pi+ pi- p, trigger is similar as K0 exp.
- tag x (ATG) x (TOF M>=2) x (veto after TOF)
x (TPC side scinti M>=1)
Eye scanning
Scanning criteria
• At least one particle comes from target holder
• Reject if two tracks overlap each other
Analysis
• dE is measured by pad rows
• By correction by dip angle between pad plane and track was
applied.
• Truncated mean was used to suppress Landau distribution.
- for each track, to get truncated mean,
- cut the Landau tail of the distribution
- keep 60% of the sample, which have the most
lowest amplitude
- get the mean value of the remaining distribution
 truncated mean
• Circle fitting in pad plane, line fitting in arclength z plane
• dE/dx value of given tracks
• Cut condition ; chi2 for pad plane and arclength z plane
1000 events eyescan
(dE/dx Truncated mean)*cos(lambda)
• Calibration of gain is not
finished.
• Runge-Kutta tracking is
not yet used. Just circle
fitting for p_t determination
proton
Pions or electrons
Momentum/charge
DAQ and Offline Analysis
1200-channel FADC Modules
(10 bits, 40 MHz, pedestal-subtraction)
for TPC
DAQ Speed

Goal: 100 Hz.


Feb run:15Hz.



# of words read =10000-30000 word, for 1 IRQ( =
4events), 1 module
Unstable pedestal: drifting with temperatures. (Aircondition is desirable.)
March Run: 20Hz




In total 3 SUN SPARC (collector) to read out data in
parallel.
On-board width-cut for discrete pulses (<4 time bins).
Tag*\bar{up}*ATG*\bar{e+e-}*(TOF M>=1) * (TPC side
M>=2)
# of words = 3000, 1 IRQ( = 4events) for 1 module
Where is the bottle neck of DAQ speed?



Reduce the number of sampling to 600.
Reduce the CLOCK from 40 MHz to 20 MHz.
Look for the other possible source in collector or builder.
Determination of
Particle Trajectory
TPC analyzer 1
In one layer
Pad
1
Pad
2
Pad
3
Pad#
Time
pulse1
ADC
TDC
pulse1
ADC
TDC
pulse1
ADC
TDC
Pulse2
ADC
TDC
Pulse2
ADC
TDC
Pulse2
ADC
TDC
……….
If overlap in time
direction was
……….
found
in adjacent pad
they are include
……….
in a cluster.
Continue in all pads in layer while
overlap is found
TPC analyzer 2
Under construction
In one layer
Cluster
Hit
x, y, z coordinate
ADC
To reduce # of
hits to be
searched by
find track.
Select hits of
enough large
ADC.
BoneHit
x, y, z coordinate
ADC
TrackFinder
TrackFinder

Track pattern recognition:




Track Parameters for RK fitting:




Segment the acceptance into sub-volumes in
(r,,z).
Nose-finding algorithm: pick up hits on the
most outer layer and search the nearest hit in
the neighboring sub-volume inwards.
Construct track-candidates: after conformal
mapping, make the cut on the chi2 values of a
straight line fit.
Assuming uniform B field.
Circle fitting on the transverse plane: (x0,y0)
and (Px0,Py0)
Slope of arclength vs z: (z0) and Pz0.
Runge-Kutta Fit with real B field map.
TPC Analyzer




Input file: MC ntuples or real data.
Output ntuples: hits information
and track parameters.
Test with MC ntuples.
Test with real data.
Conclusion





TPC has started operation in this Feb.
We are taking data now with triggers from
both tracks in TPC and spectrometer.
Struggling with large data size and slow
DAQ speed. Need to improve it
significantly.
Need to demonstrate the ability of trackfinding and PID.
We need helps from other LEPS members
for data-taking and offline analysis.