Frontemare di Trieste
Istituto Nazionale di Fisica Nucleare
Sezione di Trieste

Detector Commissioning - Part 6

Tool to check high-occupancy modules

Basically, there are two macros SSDRawNoiseMap.C and showBadModules.C. The first one runs over (rootified) raw data and makes histograms. You probably want to compile this macro when running it. To do that, make sure you have:

ACLiC.IncludePaths: -I$(ALICE_ROOT) -I$(ALICE_ROOT)/ITS -I$(ALICE_ROOT)/include 

in your .rootrc.

Then simply run from AliRoot:

root [2] .x SSDRawNoiseMap.C+("/data/local_rec/r55633/08000055633013.10.root",1000) 

The output file is called SSD_mod_occ.root. The summary histogram is called hModuleID, which looks like this for run 55633:occ_map_r55633.png

The next macro showBadModules.C runs over it and displays modules with many hits.

From the same macro: a distribution of hits per strip. There were 1000 events, so the max occupancy is something like 3%.hits_per_strip_r55633.png

It looks like getting more than 80 hits in a strip is 'a lot'. If you select modules with strips that have > 80 hits, you get ~50 modules, most have a few strips (1, 2 or 3) with more than 80 hits. A few modules (less than 10 or so) have many strips > 80 hits.

Investigation into low-Q noise clusters at the edge of modules

This is all based on run 55633.

Here is a distribution of golden clusters per module

  • black all
  • red golden clusters with local-x > 3.3 cm
  • blue golden clusters with local-x < -3.3 cm

golden_clust_edge.png

There are two modules with many clusters at < -3.3, module 736 and 1082.

Here is the strip distribution for module 1082, with two zooms at low strips numbers on P and N side:

mod_1082_r55633.png

mod_1082_r55633_zoom1.png

mod_1082_r55633_zoom2.png

Conclusions

  • The edge-clusters are a significant effect, but mostly limited to a few modules. Not sure how 'special' the edges are.
  • Large numbers of golden clusters appear when one side is 'extremely' noisy and the other side normal or a little more than normal.
  • In module 1082, masking out one strip (9,10 or 11 on the P side) would likely remove the golden noise clusters. There would still be a lot of noise on the N-side at small strip numbers

Golden noise clusters are clearly the most visible off-line, but to reduce the noise, we may should probably look at individual strips/chips first

COMMUNITY

Calendar

NOTE! This site uses cookies and similar technologies.

If you do not change your browser settings, you agree. Learn more

I agree