Steps in calibration of Endcap with MIP's using UxV
Updated September, 2005
- the code:
cvs co StRoot/StEEmcPool/StEzSmdCal/
cons
STEP 1 : no calib is avaliable,
Goal: produce relative SMD gains
- root4star -b -q rdEztMuSmdCal.C
Disable all masks, calibration is irrelevant, need enough events to have reasonable slopes in SMD, 700K CuCu minB events was enough. Make sure you do not cut on pre/post/smd energy - does not exist.
LSF: bigSort.sh
- Fit SMD slopes
Use: doSmdGains.C
to fit 'expo' to just one plain, set internal switch to do that.
Use :bigSectloop.sh to loop over plains.
Pick absolute normalization scale to get abs gains close to 1.
STEP 2a : rel SMD calib is avaliable,
Goal: Determine corrections for absolute SMD gains.
Re-play all events with SMD gains from slopes. Now you need all events you can afford.
to find corrections (smooth function vs. strip ID) for all 24 SMD plains use:
doSmdGains.C again, butch change internal switch to sum,average, and fit sums of SMD pairs.
- Apply gain corrections to SMD slopes, use: reCalSmd.C
- Display UxV points distribution: plSmdCal.C, plPresMap.C ?
STEP 2b : rel SMD calib is avaliable,
Goal: Determine absolute pre/post/tower calibration
- Fit pre/post spectra with gauss(ped)+Landau(MIP). Use: fitPrePost.C .
You need to play with fit params. Note, automatic Web-page with plots and some QA is generated in WWW/tmp area.
- Fit tower spectra with gauss&Landau(MIP). Use: fitTower.C . Ped resudua are subtracted.
You need to play with fit params. Note, automatic Web-page with plots and some QA is generated in WWW/tmp area.
- Display tower spectra vs. eta bin: plTwCal.C
STEP 3 : ~all is absolutely calibrated
Goal: slopes needs to be used for some tiles
- run script: fitSlope5.C , called by: fixGainUsingSlope.tcl to do relative calibration based on slopes of neighbours at the same eta bin.