Download - Calo Calibration meeting 10/19/2010 Tools & software O. Deschamps (LPC)

Transcript
Page 1: Calo Calibration meeting 10/19/2010 Tools & software O. Deschamps (LPC)

Calo Calibration meeting

10/19/2010

Tools & software

O. Deschamps (LPC)

Page 2: Calo Calibration meeting 10/19/2010 Tools & software O. Deschamps (LPC)

Outline

• Online monitoring & automated analysis : new features (for piquets)

• Online monitoring and Data Quality

• New stripping line for calorimeter ‘validation’ : D0Kππ0

• Calibration, reconstruction tuning & condDB status

Page 3: Calo Calibration meeting 10/19/2010 Tools & software O. Deschamps (LPC)

Online monitoring & automated analysis : tools

• New features in online monitoring tools• improvement in cfTrend tool

• alarm level printed in the output log

• new command syntax allowed :• cfTrend [Xcal]

• produce cfTrend for a given Calo (Xcal=Ecal,Hcal,Prs,Spd)

• cfTrend [depth]• extend the trending to older fmclog files (default 2 last files)• Xcal is then prompted• can manage older zipped files (thanks to Sergey F. and Iouri)

Page 4: Calo Calibration meeting 10/19/2010 Tools & software O. Deschamps (LPC)

Online monitoring & automated analysis : tools

• New features in online monitoring tools (cont’)

• improvement in caloAnalysis tool (thanks to Benoît V.)

• possibility to perform trending over long period, sampling 1 or 2 savesets per day

Page 5: Calo Calibration meeting 10/19/2010 Tools & software O. Deschamps (LPC)

Online monitoring & automated analysis : alarms

Alarm message log

Related histogram

Alarm browser

• New features in automated analysis :

• alarm publishing : alarms & warning appear in the presenter

• activated only for ‘stable’ monitored quantity so far : Ecal and Hcal pedestal shift/noise/...• when the problem disapear the alarm is archieved• clicking on the alarm message in the presenter browser provide more info and display a related histogram.

Page 6: Calo Calibration meeting 10/19/2010 Tools & software O. Deschamps (LPC)

Online monitoring & automated analysis : alarms

• more info on the monitor page (here Ecal/PedestalShiftOverNoise) && trending

Page 7: Calo Calibration meeting 10/19/2010 Tools & software O. Deschamps (LPC)

Online monitoring & automated analysis : global info

• New features in automated analysis :

• add min-to-max for the distribution of the monitored quantity in the page comment

New formatoutput

• Next to come : trending of the RMS and min-to-max over channels(trending of the average over channels already available)

Page 8: Calo Calibration meeting 10/19/2010 Tools & software O. Deschamps (LPC)

Online monitoring & automated analysis : Prs pedestal

• New features in automated analysis :

• Prs semi-automatic pedestal follower

• automated analysis produces configuration file for Prs pedestal offset.• Store at : /group/calo/sw/PrsPedestalAutoUpdate/Offline update/

• can be uploaded to hardware when pedestal shiftis observed

• Corrected offset are automatically monitored

Page 9: Calo Calibration meeting 10/19/2010 Tools & software O. Deschamps (LPC)

Online monitoring & automated analysis

• New features in automated analysis (next to come)

• production of condDB table from automated analysis • gain change, channel status, average (and RMS) of LED signal on PMT&PIN

• code prepared by Benoît Viaud : to be implemented/tested asap at pit

• work the same way as for Prs pedestal offset configuration file• condDB table to be stored as text file • then managed by the PVSS interface J.L. Panazol is producing.• actual condDB update to be triggered by calo-piquet expert when needed

• progress in this area expected soon

Page 10: Calo Calibration meeting 10/19/2010 Tools & software O. Deschamps (LPC)

Online monitoring & automated analysis : Spd rate

• Spd monitoring in calibration farm is different from Ecal/Hcal/Prs

• based on CaloEfficiency algorithm :

• monitor the rate of Fake signal (« 1 » when no LED flash)• and rate of Missing signal (« 0 » when LED flash)

• expected ‘range’ for Fake/Missing signal was [0,0]

• Without collision the measured ‘fake signal’ rate averaged over SPD channel is of order of 10-6

for odd BX and 10-9 for even BX • Only few channel see few ‘fake signal’

• but lot of warning/error/alarm recently with bunch train configuration during collisions• few thousend channels see few fake signal during the ~40k calibration trigger (over 15mn)• average fake rate ~5 10-5

Page 11: Calo Calibration meeting 10/19/2010 Tools & software O. Deschamps (LPC)

rate of Spd=1 when no LED flash

rate of Spd=1 (x,y) 2Dview

location of channelsoutside the expected rate range

Text log

Online monitoring & automated analysis : Spd rate

Page 12: Calo Calibration meeting 10/19/2010 Tools & software O. Deschamps (LPC)

start ofcollisions

end ofcollisions

Online monitoring & automated analysis : Spd rate

expected ‘range’ for Fake/Missing signal changed to [0,10-3] to avoid too much alarms during collision

Saveset timestamp

Page 13: Calo Calibration meeting 10/19/2010 Tools & software O. Deschamps (LPC)

Online monitoring & data quality

• Strange peak of photon momentum and Pt in reconstructed/stripped data• Problem observed by Fatima Soomro in B->K*gamma analysis

• after investigation it is found to be due to a single middle channel with ADC stuck

Page 14: Calo Calibration meeting 10/19/2010 Tools & software O. Deschamps (LPC)

• mostly concern fill 1298 (run 77969-77979) of 23rd of august – just after the power cut

• problem was identified online – documented in Elog – tagged as ‘severe’ in problemDBhttp://lbproblems.cern.ch/problems/469/

• but flagged ‘OK’ by data quality group (although the problem shows up also in DQ histograms)

Online monitoring & data quality

• after discussion with M. Adinolfi the fill (~0.2/pb) will be removed from next reprocessing• DQ detector-flag is being put in place (fill will be declared CALO-bad)• Possibility to mask single channel offline to be implemented in software – will allow to resurrect this fill

• Need a mechanism online to explicitely tag as BAD a single channel by hand in condDB • could be integrated inside the PVSS interface managing the automated analysis• still possible to provide hand-made condDB a posteriori

Page 15: Calo Calibration meeting 10/19/2010 Tools & software O. Deschamps (LPC)

D0Kππ0 : Calorimeter ‘validation’ stripping line

• New stripping line for D0Kππ0 with both resolved and merged pi0 implemented by Régis

• D0 from primary vertices – (almost) unbiased wrt neutral-ID• dedicated to calorimeter ‘validation’ (Charm stream)

• useful to check reconstruction tuning, alignment, calibration, ...

• available with Reco06Stripping10 data

Note : another D0Kππ0 stripping line exists (Andrea Contu) for D from B (with strong cut on photon-ID). Also accepted for Stripping10 (Bhadron stream)

Page 16: Calo Calibration meeting 10/19/2010 Tools & software O. Deschamps (LPC)

D0Kππ0 : Calorimeter ‘validation’ stripping line

D0Kππ0(resolved) D0Kππ0(Merged)

• Directly obtained from the stripping line output • ~8/pb of Reco06Stripping10 data (3M stripped events - sept/oct. high-mu data)

Page 17: Calo Calibration meeting 10/19/2010 Tools & software O. Deschamps (LPC)

D0Kππ0(resolved)

D0Kππ0(Merged)

NPV=1 NPV=2 NPV=3

NPV>3

NPV=1 NPV=2 NPV=3

NPV>3

D0Kππ0 : Calorimeter ‘validation’ stripping line

• looking at nPV sharing

Page 18: Calo Calibration meeting 10/19/2010 Tools & software O. Deschamps (LPC)

#PV

Resolved Merged Sharing of nPV in

preselected sample

m σ N m σ N

All 1876±1 23±1 3700±160 1834±3 65±4 2800±210

1 1872±1 22±1 (39±2)% 1832±5 67±8 (35±5)% 35%

2 1879±2 26±2 (42±3)% 1831±4 61±7 (34±4)% 40%

3 1878±2 19±2 (15±1)% 1828±5 74±10 (24±4)% 19%

>3 1878±4 23±3 (5±1)% 1837±7 68±13 (7±2)% 6%

D0Kππ0 : Calorimeter ‘validation’ stripping line

• Preselected D0 looks (surprisingly) not too much affected by the high-mu condition

Page 19: Calo Calibration meeting 10/19/2010 Tools & software O. Deschamps (LPC)

Calibration, reconstruction tuning & condDB status

• No progress so far with calibration/tuning/alignment/condDB stuff on my side• some work ongoing :

• pi0 mass peak trending versus nPV over sept/oct run• pi0 mass peak for channel with large LED-measured gain change at beg. of run• implementing pile-up correction as a function of nPV in CaloCorrection) • validate Irina/Iouri LED-measured gain change on data • Ecal-tracking alignment with electron&Brem (P. Ghez interested to contribute)

• Deadline for condDB release for winter reprocessing is next friday• what could be in ?

• Aurelien new Eflow+Pi0 : should be feasible (fast validation on pi0 to be done)• Irina LED-measured gain change : unlikely to be in place and validated by Friday• Pile-up substraction / nPV : not possible (request soft update + validation)• Prs alignment : possible ?• Ecal alignment : not ready• Electron dependency with incidence : not ready• Merged Pi0 energy tuning (to be validated on D0) : should be feasible• S-shape, L-correction : not ready

• Anyway everything can be applied a posteriori within DaVinci• just one line to be added in options (+updated condDB tag)