In the possibility to rerun year 7 data, an improvement of SSD data can be added.

here

The goal is to fill the WaferConfiguration table : this table allows to do (or not) the reconstruction in a given wafer. (By default, this table has only one entry, which is to do the reconstruction in all wafers)

For this, we have to rerun the reconstruction in the PointMaker and look at the quality of data (strips signal --> clusters signal -- > hits) and take a decision of which wafer(s) has(ve) to be removed.

I started by taking runs at the beginning of the run7 : 

/star/data03/daq/2007/091/8091018/st_physics_8091018_raw_1040040.daq

/star/data03/daq/2007/091/8091095/st_physics_8091095_raw_1020001.daq

/star/data03/daq/2007/092/8092002/st_physics_8092002_raw_1020013.daq

/star/data03/daq/2007/092/8092119/st_physics_8092119_raw_1020003.daq

/star/data03/daq/2007/093/8093002/st_physics_8093002_raw_1020005.daq

/star/data03/daq/2007/093/8093017/st_physics_8093017_raw_1040001.daq

/star/data03/daq/2007/094/8094008/st_physics_8094008_raw_1030011.daq

/star/data03/daq/2007/095/8095065/st_physics_8095065_raw_1040020.daq

/star/data03/daq/2007/096/8096018/st_physics_8096018_raw_1030002.daq

/star/data03/daq/2007/097/8097009/st_physics_8097009_raw_1030001.daq

1. 092/8092002/st_physics_8092002_raw_1020013.daq

The first thing to check is the number of strips fired per wafer, as the number of reco. clusters is conditionned by it.

• Fig 1 : number of strips fired (for all the events in the daq file) per wafer on side P

We clearly see hot spots meaning that strips in those wafers are always fired.

In principle, the offline signal/noise cut should keep only meaningful strips (ie strips fired by particle).

But it can also happens, due to lower noise, that this cut is always fullfilled.

I looked then in particulary at ladder 9, where wafer 6 seems hot.

• Fig 2 : adc values of strips, as a function of wafer, for ladder 9

 We see that the signal for strips in wafer 6 seems to be ~ 20-30 adc for most of them, whereas for the others, no particular adc value is prefered (note to me : have to look at wafer 10)

• Fig 3 : signal distribution for wafer 6 (in red) (and compared to wafer 5(in blue), which seems normal)

For wafer 6 : <signal> = 28 +/- 13 adc ; number of entries = 60726

For wafer 5 : <signal> = 26 +/- 67 adc ; number of entries = 5004

These values come from the histogram (not a fit)

For comparison, the number of strips of the entire ladder 9 is 139415 strips ==> it means that almost the half of the strips fired for ladder 9 comes from the wafer 6

==> This is the kind of case where I want to remove the wafer : a possibilty to discriminate this case is to look at the RMS of the histogram : for a normal wafer, the RMS ~ 50-60 adc because of the tail.

For case where it seems to be an electronic issue, the RMS is much lower.

• Fig 4 : number of strips fired vs event.

Another interesting quantity is the number of strips fired as a function of event (for this daq file)

in black : number of strips fired for ladder 9

in red : number of strips fired for ladder 9 wafer 6

in blue : number of strips fired for ladder 9 wafer 5

Comments :

1. we see the previous behavior, that ~ 50 % of the strips fired come from wafer 6 : this number is constant as a function of event
2. there seem to have particular events where a large fraction of strips are all fired : this is a typical case of common mode shift, where all the strips in a given ladder (SSD ? ) are all fired in the same time.

Looking at the red line, it seems that the mean level is between 100 and 150 strips ; this could be due to a bad A128C chip. (leakage capacitor)

This is in fact confirmed by the next plot, that shows the id of strips in this wafer (to recall, we have 6 A128C chips, each of them are reading the signal from 128 strips)

Clearly the chip 2 (strips id 129 to 256) is always fired.

• Fig 5 : id of strips for wafer 6