After the L2 alignment of the barrel was seen to bring large improvements the focuos turned to aligning the TRT endcaps.
The initail attempt to run the L1 endcap alignment is described here.
It was seen that the L1 alignment seemed to be hahaving resonably well.
There was convergence of the parameters and an imporvement in the validation plots.
It was found that the r-t calibration was offset considerably in the endcaps and that this was responsible for most of the residual degradation.
Endcap Calibration
From talking to sasa, we decided to correct the endcap calibration by adding a constant to the t0s.
It was determined that we should add ~5ns to the endcap t0s.
With this fix, the resolution using the global Barrel rt for the Endcaps, gives results as good or better that was seen above from changing the rt in the endcap and iterating.
With this fix, the global rt used in the Barrel is a resonable fit for the endcaps (as seen below), this gives results as good or better that was seen above from changing the rt in the endcap and iterating.
Endcap L1 Alignment
The L1 alignment of the TRT endcaps has been done, using the L1 barrel alignment and the L2 TRT barrel alignment found using TRT Only Tracks(geometry tag: TRT_Cosmic_2008_07), with respect to Si grometry tag InDet_Cosmics_2008_6.
This Si alignment was done is consistent with Bon data, and this was used to get the TRT L1 endcap constants.
Need to put some results here.
These constants are in the db as TRT_Cosmic_2008_11.
The L1 alignment of the TRT endcaps has been done, using the L1 barrel alignment and the L2 TRT barrel alignment found using TRT Only Tracks(geometry tag: TRT_Cosmic_2008_07), with respect to Si grometry tag InDet_Cosmics_2008_05.
This Si alignment was done is consistent with Boff data, and this was used to get the TRT L1 endcap constants.
Need to put some results here.
These constants are in the db as TRT_Cosmic_2008_12.
Endcap L1 Alignment With different runs
The TRT L1 Endcap alignment described above was repeated for the 5 golden runs.
The results were seen to be consistent.
put a demonstration of this here.