After the initial L1 TRT alignment the plan was to understand the L1 barrel alignment.
This was given higher priority b/c the barrel was better understood both from the hardware side (in terms of timing) and the software side (in terms of tracking and TRT calibration)
From the large number of available cosmic runs "good runs" were selected that are listed here:
At the time there was no official sign-off of good runs.
This was the main source of information about the possible runs.
Running on different events. Event list
- Problems with the running on different events. B On vs Off
Here are the L1 barrel results before Lorentz angle fix.
In parallel to this it was found the residuals in the Si were different for B-field on vs off runs.
It was then determined the an incorrect correction for the Lorentz angle was being applied.
Silicon Lorentz angle saga.
The main problem with the first Si alignment was that the wrong Geometry was used.
* Geometry version:
from Grant ...
"People should use the ATLAS-GEO-xxx geometries.
Most of the cosmics have been processed with
ATLAS-GEO-03-00-00 : With field
ATLAS-GEONF-04-00-00: No field
The older Atlas-Comm geometries have a bug such that the Lorentz shift is in the wrong direction and should not be used. "
However this was used for the first Si alignment, leading to corrupt constants.
Release differences. (from Grant)
Note, the Lorentz correction is about -28 um (in local x) in the Pixel.
* Use of DCS. In recent releases, use of DCS (for temperature and high voltage) was introduced. It turned out that the HV from DCS was not reliable however but this was then later fixed.
Shifts are wrt to the correct shift.
1. <= AtlasTier0-14.2.22.2 (or AtlasProduction-14.2.22.1)
Old default values of HV and temperature for pixel and SCT.
pixel: ~ 6um shift (default of ~100 V rather than 150V)
This is a consistent shift for all barrel modules.
SCT: ~ 1um shift (default of -10 C rather than +10 C)
(opposite direction for each side)
2. AtlasTier0-14.2.22.2 to AtlasTier0-14.2.23.1
(or AtlasProduction-14.2.22.1 to AtlasProduction-14.2.23.2)
Use of DCS introduced but problem with reliability of HV.
Pixel: Wrong HV for some modules.
Some OK, Others wrong by ~ 26um.
SCT: OK. (not from DCS but reasonable defaults used.)
3. >= AtlasTier0-14.2.23.2 (or >= AtlasProduction-14.2.23.3)
Pixel HV fixed to 150V (it is very stable at that voltage)
Lorentz correction OK for both pixel and SCT.
* Cluster error corrections: The cluster error is dependent on the track angle and this must be corrected for the Lorentz angle. The sign for this correction was inconsistent for the latest geometry giving the wrong effective angle. This should however not introduce any bias. This is fixed in SiClusterOnTrackTool-01-14-00 and that tag is requested for the next AtlasTier0-14.2.24.x
New set of Si constants.
- After the Si alignment with the correct lorentz angle. Here for details
The TRT alignment was done repeated with the new 'InDetAlign_Cosmics_2008_03' alignment.
Athena release '-tag=14.5.0.4,AtlasProduction' was used.
The inconsistency of the L1 constants with field on and off then went away.
The new TRT alignment became 'TRTAlign_Cosmics_2008_03'
See Here for details
Here are the L1 barrel results after Lorentz angle fix. (with Si constants InDetAlign_Cosmics_2008_03)
