Transcript ssd_status_enrico_fragiacomo - Indico

SSD offline status
Offline week
CERN, 7 July 2008
Enrico Fragiacomo
on behalf of ALICE SSD
Run statistics
The SSD was one of the detectors that participated in the majority of the global
runs (Panos). Calibration runs were extensively tested and used.
Many standalone runs to check:
1. The DDL mapping
2. The hardware/software adc map
3. The output of the pedestal run
Global runs:
1. In the same partition with the SPD, later with layer 3 and recently with part
of layer 4 of the SDD triggered by the SPD FastOR when SPD was able to
provide it.
2. In the same partition with the TPC triggered by the combination
ACORDE.OR.SCINTILLATOR
3. In the same partition with the rest of the ITS subsystems and TPC triggered
by the combination FastOR.OR.ACORDE.OR.SCINTILLATOR
See https://twiki.cern.ch/twiki/bin/view/AliceSSD/WebHome
Top three debated items
Wrong coordinates on layer 6
Overlaps in geometry
Memory consumption of calibration objects
Wrong coordinates on lay6
 The problem has shown up after reconstructing
the cosmic data. No problem with simulations.
 Recpoints coordinates on layer 6 were wrong!
 It looked like a local x->-x inversion but it was in
fact more entangled
 Due to a wrong assumption in the
segmentation (swapped stereo angles between
lay5 and lay6) (see next slide)
How modules are installed in hw
LAY6 MODULE
1. Stereo angles (unlike
in the sw) are the
same but lay6 module
is up-side-down!
2. Moreover, readout
from positive x!
LAY5 MODULE
Marco Bregant
 AliRoot simulation modified according to the new local
coordinates
 Two classes modified:
 AliITSsimulationSSD, where the actual response simulation is
implemented: small changes -> stereo angles no longer
swapped between lay5 and lay6
 AliITSsegmentationSSD, where the details (e.g. stereo angle,
width, pitch, ...) about the SSD module are plugged in
GetPadTxz (which returns P and N strips for given local
coordinates) has changed
 Effective since trunk 26/06/08 and ported back to the
v4-13-Release
 Simulation and reconstruction are again compatible (see
next slides)
Global x-y projection
Simulated cosmic rays (muons) via AliGenCosmicParam class
Straight line fit through SPD points via AliEveCosmicRayFitter
Scale in cm
y
x
Only SPD+SSD
geometry shown
Reconstructed point
Geant hits
Note the many hits in y
Global x-z projection
z
x
Global z-y projection
y
z
Problems in the geometry
 Volumes breaking the simulation! Due to
volumes created with TGeoXTRU. Too many
vertices gave unpredictable results (negative
volumes)!
 Overlaps (149) obtained with the “s” checking
option. Many were repeated due to the
hierarchy. Almost all of them have been fixed
(see next slides), 13 remaining (see also Mario’s
talk)
 Many of the overlaps (39) due to a request of
infinite precision 
[email protected]
Overlap Fixing
 Overlap between ITSssdSensor e Ladder
Segment
Replacement of Physical Mother Volumes with
TGeoVolumeAssemblies (TGeoXtru Problem)
Overlap Fixing
 Overlaps between adiacent Ladder Segments:
fgkSSDTorelance variable has been introduced
Carbon Fiber Overlaps
Cooling Tube Overlaps
Overlap Fixing
 Overlaps between adiacent Ladder Segments and
EndLadder Segment: fgkSSDTorelance variable
has been introduced
Carbon Fiber Overlaps
Cooling Tube Overlaps
Overlap Fixing
 Overlaps between adiacent ladders
Cooling Tube Overlaps
What Remains?
Overlap between Ladder Cables
and Hybrid Flex
What Remains?
Overlap between LadderSupport
And Layer
Top three debated
 Wrong coordinates on layer 6
 Overlaps in geometry
 Memory consumption of calibration
objects
Next-todo item
There are 1698 AliITSCalibrationSSD objects
(one for each SSD module)
Each AliITSCalibrationSSD has pointers to the
calibration objects AliITSNoiseSSD,
AliITSGainSSD and AliITSBadChannelsSSD
Each of these objects has 2 TArrays of 768
values (as many as the strips on each side)
Simple implementation of the calibration scheme
but memory consumption!
TArray to be replaced by arrays and one object
for all the modules (in place of 1698)