calibration data taking

1
calibration data taking

• pedestals ok
• pulsers first tests done sucessfully by Taka,
  Robert and Stephanie
• problem writing to the database takes a long
  time
• Taka is working on improvement
• trigger files pulser parameters are downloaded
  automatically
• missing ICD (preamp) pulser download, TC
  parameters, PIB reset
• news from Ulla?
• l3fCalCalibTool small changes
• remove 1st event in each ramp step
• define ramp_step as step_number step_size
• free gain runs
• Robert working on this
• validation code link between calibration_manager
  _gui and validation
• finale cuts for pedestal values
• test for pulsers
• assembles corresponding runs in a collection
• produces ascii file with mean/sigmas for all
  steps and channels

2
New ERROR codes
type hard 1 invalid 2
cut type peds/pref 1 mean 8 2
ratio 1/8 4 sigma 8 8 diff mean
8 16 diff sigma 8 32 drift mean 8
64 drift sigma 8 128 hkill 256
warn mean 8 512 warn ratio 1/8 1024 warn
sigma 8 2048 warn diff mean 8 4096 warn
diff sigma 8 8194 warn drift mean 8 16384
warn drift sigma 8 32768 mod cell
type gain 1 type time 1
type puls 1 type corr 1
some changes in offline database design necessary
3
Calibration Validation

• distributions to determine cuts for pedestal
  validation
• pref pedestal reference file with 10k events
• peds pedestal monitor file with 5k events

mdrift
mdiff
sdrift
sdiff
using ?cell0.3
4
pref-10k difference in mean values
mdrift
mdrift
mdiff
mdiff

• weak dependence of mdiff with sigma
• tail of mdrift up to 20

5
pref-10kdifference in sigma value
sdrift
sdrift
sdiff
sdiff

• clear dependence of sdiff with sigma
• cut on sdrift at /- 10 instead?

6
error correlation for pref -10k
total 79 channels rejected with mdrift8/sdrift8
7
error correlation peds-500
in green are the channels that have also been
rejected in the 10k sample
72 channels

• cuts on mean, ratio and sigma reject in total
  same channels than in ref sample (26 channels)
• changes in cut on mean/sigma difference

8
peds-500difference in mean values
mdrift
mdrift
mdiff
mdiff

• strong correlation of mdiff with sigma, dominate
  by error on 500 evts sample
• cut on mdrift /-5 instead?

9
peds-500 difference in sigma
sdrift
sdrift
sdiff
sdiff

• similar correlation of sdiff with sigma,
  dominated by error on 500 evts sample
• cut on sdrift /-5 instead?

10
error correlation
87 channels rejected with mdrift8/sdrift8
11
Hot Cell Killer

• select channels to be killed
• save them in a list
• commit to database and create pickle download
  files
• shows index of pedestal set and run numbers of
  calibration runs used
• possiblity for experts to choose?
• shows the list of all channels to be killed
  (saved channels)
• possiblity to deselect channels?
• displays on screen various selection of bad
  channels
• write rather in file?

12
Hot Cell UnKiller

• shows hot-cell-killed channels
• select channels from list
• save them
• commit to database and create pickle download
  files
• or unkill all hot-cells

13
Hot Cell Expert

• shows 0-suppression parameters for specified
  channels
• allows the values to be modified
• save them in list
• commit to database and create new pickle download
  file
• GUI ready to be tested
• Instructions provided in HELP-menu
• (even balloons!)
• comments, corrections welcome!
• improve error recovery procedures