TOB rod analysis

Index:

Files: where to find them, and what they do
Numbering convention for modules
Reference runs for signal studies
Clustering algorithm
Bad strip search algorithm
Noise plots for a reference run
Talks
Technical documentation
Useful links

Hostname: pcepcmt17.cern.ch
Username: control
(To have the password, ask me or Laurent)
Directory: ~/anal

Directory structure:

~/anal/src: fortran source files (see Useful links for infos about fortran programming)
~/anal/include: include files
~/anal/job: job files
~/anal/anal: PAW macros (see Useful links for infos about PAW)
~/anal/data: the raw data are recorded here
~/anal/out: "ntuples" are recorded here

The job file is ~/anal/job/test1_andrea.job. It compiles, making use of the makefile ~/anal/src/andrea.mk, the source code ~/anal/src/templ_main_andrea.F, which calls the following files:

templ_card_andrea.F: creates the variables in the cards, and reads the cards
templ_init_andrea.F: initializes and books the histograms
templ_read_andrea.F: reads the events and calls templ_anal_andrea.F
templ_end_andrea.F: ends and prints some statistics

File ~/anal/src/templ_anal_andrea.F analyzes the events, for example calculating the noise and executing a clustering algorithm (look here for the description).
The run number to be entered in the job file has to be taken from the log book.
After the end of the job, the ntuples can be analyzed in PAW by ~/anal/anal/showntp_andrea.kumac, which plots (see latest talk, apr.03):

highest cluster in module
number of clusters found
signal to noise ratio of highest cluster
strips per highest cluster
beam profile
misalignment between modules 7 and 1, 2 and 8, 7 and 8 (see the numbering convention)
"tracking" with two hits (7 and 1)
"tracking" with four hits (7,1,2,8)

Using option minimal=1 you can plot only S/N and beam profile for one only reference module (useful for presentations...)
Other options (center, overlap, cosmic, bonding, peak, off) select specific runs (see the list).

Macro ~/anal/anal/showp3.kumac plots the noise (defined in several ways) for the chosen module.
The very simple file ~/anal/anal/x0y0.f computes the source position (x,y) from the parameters of the linear best fit in "tracking".

Numbering convention for modules:

Warning: in the ntuples, a different numbering is followed (the same of the FED).

FED numbering: Our numbering:

1 5

2 11

3 6

4 12

5 7

6 1

7 8

8 2

9 9

10 3

11 10

12 4

Reference runs for the signal analysis (when not otherwise specified, DECONVOLUTION MODE is assumed):

65770 (option "center"): source over the center of a sensor;
65768 (option "peak"): as above, in PEAK MODE;
65752 (option "bonding"): source over the bonding between two sensors (to plot a comparison with run 65770: ~/anal/anal/geom.kumac);
65766: as above, in PEAK MODE;
65796 (option "overlap"): source over the overlap region between modules 7,1,2,8;
65801 (option "cosmic"): cosmic run;
65803 (option "off"): random triggers run (source and scintillators far from modules).
For the "delay scan" (source over the center of a sensor), use runs from 65778 to 65774 (use ~/anal/anal/delay.kumac to plot S/N for all the runs, ~/anal/anal/pulseshape_andrea.kumac to plot efficiency vs delay - the values here have to be written by hand).

Clustering:

The clustering algorithm is implemented in routine FINDCLUSTERS in ~/anal/src/templ_anal_andrea.F.
To find a cluster:

loop on the APV strips; any set of consecutive strips with S/N>2 is a cluster candidate

a bad strip (see next section) is included in the cluster candidate only if the preceding strip and the following strip are both good strips with S/N>2

a cluster candidate is a good cluster only if its highest signal strip (the "seed") is a good strip and has S/N>5

The signal of a cluster is the sum of the signals of its (good) strips; its noise is the noise of the "seed" (this can be changed by varying the value of the flag NOISEDEF); its position is the baricenter of the signals of its (good) strips.

Bad strips:

If the card BADS is set to TRUE, the routine BADSTRIPS (in ~/anal/src/templ_anal_andrea.F) is called only once, after the first INIPED events (=500 by default) which are used to calculate pedestals and noise. No updating is performed (at least at the moment).
It fills the GOODSTRIP logical matrix, with the following rejection criteria:

Noisy strip: noise greater than 5 sigma of the truncated mean noise in the APV.
Dead strip: noise less than 50% of the truncated mean noise in the APV.

The truncated average and sigma of the noise in the APV are obtained excluding the 5% (i.e. 6 out of 128) higher and lower noises in the APV.

Reference noise run 20690:

Detailed noise study for all APV chips (done with ~/anal/anal/showpfast.kumac).

Synthetic view of the noise figures for modules (done with ~/anal/anal/showp3.kumac):

(warning: this is the FED numbering; see numbering convention)

Talks:

11/4/2003, Tracker Week general meeting.
Previous talks (from Duccio's page).
Livia's talk (summer student, july 2003)
Livia's report (final report for the summer students program 2003)

Technical documentation: