As we have discussed, the problem with the old calibration originates in the low statistics of pi0s reaching the corners of bigcal. A calibration was completed removing the corners by placing a "cross cut" on bigcal. After calibrating the cross in isolation, the corners are calibrated by events with one cluster in the cross. This cluster's calibration is fixed while the corner's is being adjust. Only events that have a large opening angle reach the corners. The distributions of events can be easily seen below.
It is worth noting that the net calibration coefficient change is towards lower values.
Although I did not make use of these as the starting point for my latest calibration from scratch, the gain matching parameters are in the text file below.
All the plots of the fits can be found in this directory There are a few outliers. They can be seen in the plot of all the fits to the ADC below.
It appears to do a good job. Looking at the gain matched ADCs for a row (actually only half a row) you can see the cross section anglular dependence.
The energy distributions look much better!
|
old |
new |
|
|
|
|
|
|
The plots above are not with the latest calibrations of the corners. There are a few outliers and RCS seems to have some problematic rows. However, this could just be a script problem. I am looking into this.
|
old |
new |
|
|
The asymmetries over all look much better, however, there remains a difference between RCS and Protvino.
My previous attempts to see the eta mass in the 2 photon invariant mass spectrum were unsuccessful. There was a possible hint, but the kinematics tend to make things difficult.
I think this is worth trying to get right because it gives an absolute check that RCS and Protvino are in agreement.
Considering the Protvino difference in asymmetry and the RCS energy distributions there seems room for improvement.
By arranging only certain parts of bigcal to participate in the two photon mass reconstruction, we eliminate the pion "background". Of course statistics are low.
|
|
|
|
|
|
|
|
|
The last two plots are for the "parallel" calibration but use all data.
I have yet to do a thorough check of the parallel/perpendicular difference, but I believe it has gone away with the new calibrations.
I want to do one more calibration which starts with the gain matched coefficients. It will include all data.
![[ICO]](/icons/blank.gif){kind=icon}
![[PARENTDIR]](/icons/back.gif){kind=icon}
![[IMG]](/icons/image2.gif){kind=icon}
![[ ]](/icons/unknown.gif){kind=icon}
![[TXT]](/icons/text.gif){kind=icon}
![[DIR]](/icons/folder.gif){kind=icon}
