CALORIMETER CELL MONITORING TOOL

1
CALORIMETER CELL MONITORING TOOL

• Viacheslav Shary

2
INTRODUCTION

• Goals
• Detect the problems which were not found in the
  previous stages of the data quality check.
• Provide the data quality information for further
  use e.g. lists of cells, BLSs with unusual large
  or small signal. This information is of interest
  for the reconstruction and simulation programs
  and, of course, for the data analysis.

3
THE SCHEME
Raw calorimeter data from event
distributor. Only ZERO BIAS events.
Thumbnail calorimeter data. Only ZERO BIAS
events.
on-line rcp file off-line rcp file dq_calo
package
text report file (for shifter)
input file for the database
file with root trees
4
DATA QUALITY (DQ) CHECK
EVENT DQ CHECK N cells, total energy, total
squared energy MET, SET
CELL DQ CHECK (per run) N entries, total energy,
total squared energy
Fraction of bad events i.e. events with
parameters out of limits
bad cells and BLSs, i.e. channels with energy
and/or RMS out of limits
Run data quality GOOD, REASONABLE, BAD Lists
bad cells, bad BLSs, BLSs without signal
5
BLS QUALITY CRITERIA

• BLS is BAD if
• mean energy Elt -1.5 GeV or E gt 2.0 GeV
• RMS lt 0.1 GeV or RMS gt 2.5 GeV
• more than 5 bad cells in the BLS
• bad cell have a mean energy E lt -4 GeV
  or E gt 5 GeV
• All limits can be changed via rcp file.

6
RUN QUALITY CRITERIA

• If percentage of BAD events gt 10 the run quality
  is BAD.
• Else if number of cells with mean energy out of
  limits gt 24 the run quality is BAD.
• Else if number of ZERO_BIAS events in the run lt
  100 the run quality is UNKNOWN.
• Else if number of BLSs without signal gt 1 the run
  quality is BAD.
• Else if number of BLSs without signal gt 0 the
  run quality is REASONABLE.
• Else if number of BAD BLSs gt 0 the run quality is
  REASONABLE.
• Else if percentage of BAD events gt 2 the run
  quality is REASONABLE.
• Else the run quality is GOOD.

7
OUTPUT EXAMPLES

• WWW
• http//d0-france.in2p3.fr/WORKING_GROUPS/DQ/
• At this page the following information is
  available.
• Run list with corresponding qualities
• Run report with bad cells BLS list
• 2d histogram for many runs

8
An example of 2d (?-?) histograms. The Z axis
corresponds to the mean energy per entry in each
cell (GeV).
9
An example of 2d (?-?) histograms. The Z axis
corresponds to the mean energy per entry in each
cell (GeV).
10
An example of electronics 2d histogram. X is ADC
number. Y is serial channel number. Z is the
mean energy per entry in the cell (GeV).
11
An example of electronics 2d histogram. X is ADC
number. Y is serial channel number. Z is the
mean energy per entry in the cell (GeV).
12
An example of 2d (?-?) histograms. The Z axis
corresponds to the mean energy per entry in each
cell (GeV).
13
RESULTS
RUNS TOTAL GOOD REAS. BAD UNKN.
160000 - 169999 (August December) 601 123 413 23 42
170000 - 177681 (January May) 485 238 179 54 14
14
TYPICAL PROBLEMS (1)

• Usually runs have 0-2 of bad events, but some
  runs have near 25. These runs have 1 or 2 BLSs
  with a cell(s) with a huge signal. These runs
  should be declared BAD, because it can affect
  physics quantities.
• Some runs have no signal from several BLSs.
  Quality - BAD.
• Some runs have no signal from one BLS. Quality -
  REASONABLE.

15
TYPICAL PROBLEMS (2)

• There are runs with large signal in all BLSs in
  one crate during some time. Quality - BAD.
• Runs have frequently the following problem 1 or
  2 BLSs with a large signal (average signal per
  entry 2-10 GeV and RMS near 2-10 GeV). Usually
  these BLSs are the same in several runs. This
  problem affects missing ET and probably other
  physics quantities. Quality - REASONABLE.

16
Missing ET (Ecellgt0, without CH)

• PEAK 2 GeV,
• FWHM 3 GeV
• FWHM/2.36 1.3 GeV

17
MISSING ET FOR ZERO BIAS
PEAK, GEV FWHM FWHM/2.36 all cells Egt0 Egt100 MeV
all cells 4.5 7.5 3.2 3.5 5.5 2.3 3.5 5.5 2.3
without CH 2.5 4.5 1.9 2 3.0 1.3 2 3.0 1.3
18
MET for good reasonable runs
GOOD QUALITY FWHM 3 GeV
REASONABLE QUALITY FWHM 3.5 GeV
19
MET for reasonable runs
CORRECTION BAD CELLS AND CELLS IN BAD BLS ARE
NOT USED FOR MET CALCULATION
AFTER CORRECTION
BEFORE CORRECTION
20
MET for good correc. runs
CORRECTED REASONABLE QUALITY
GOOD QUALITY
21
MET for good reasonable runs
REASONABLE GOOD QUALITY
REASONABLE QUALITY
GOOD QUALITY
22
MET for reasonable runs
CORRECTION BAD CELLS AND CELLS IN BAD BLS ARE
NOT USED FOR MET CALCULATION
AFTER CORRECTION
BEFORE CORRECTION
23
MET for good correc. runs
CORRECTED REASONABLE QUALITY
GOOD QUALITY
24
ISSUES PERSPECTIVES (1)

• How to organize the permanent monitoring and
  feedback with shifters ?
• The package was tested online by Laurent. How to
  proceed further ?
• In the output text file shifter can find an
  information about run problems (lists of bad
  cells and BLSs).
• How to make the data quality information
  available ?
• The runs quality database can be used to save
  lists of bad cells and BLSs.
• The bad cells can be removed from calorimeter
  data chunk.

25
ISSUES PERSPECTIVES (2)

• Now its time to try this tools!
• i.e. one can compare analysis results for good
  and reasonable runs.
• What would the next step for calorimeter data
  quality be ?
• Are there other hardware failure that we can
  detect ? Shared energy problem ? Do we need to
  look at physics trigger to detect some problem ?
• Possibly, one could use clusters (CellNN) from
  reco to assert the impact of hardware problems on
  physics. It could also help discover warm zones,
  local inefficiencies and so on.