ALICE Calibration Requirements  
 
 
TPC  July 18
Par # Parameter Data format/size per channel Data size (Total) Bytes Update freq Source Confirmed  Run type / Trigger type # of required events/sampling rate Processing level: sub-event or event Results: FEE/Archive Accessible by offline Calib. Procedure in AliRoot[1] use case #
      on FES in OCDB reference                    
Pad by pad calibration
1 Pedestal ALTRO 1K (Time bin) x10b value per pad   0 7.00E+08[2] run DAQ or HLT (+Offline check) yes ? (Roland, Marian) Dedicated run (with pedestal)  100 black events Configure FEE through DIM/DDL. Stored on pedestal file server(?) migrated to CASTOR. The pointer to the Castor file is stored in DAQ Logbook. - no  -  
2 Pedestal  4 By - one value per pad    2.40E+06[3] 4.46E+08[4] run DAQ or HLT (+Offline check) yes ? (Roland, Marian) Dedicated run (with pedestal)  101 black events subevent (DAQ) DAQ or HLT FES / OCDB yes algorithm - rec usage - rec  
3 Noise  4 By - one value per pad    2.40E+06[5] run DAQ or HLT (+Offline check) yes ? (Roland, Marian) Dedicated run (with pedestal)  102 black events   DAQ or HLT FES / OCDB yes algorithm - rec usage - rec  
4 Gain calibration of electronics 4 B/channel   2.40E+06   LHC period   yes ? (Roland, Marian) Pulser run - subevent (DAQ) DAQ FES/ OCDB yes    
5 Timing alignment of electronics 4B/channel   2.40E+06   DAQ/HLT(Offline check of algorithm)   Dedicated run (with pedestals) 100 black events sub-event  DAQ/HLT FES/ OCDB yes usaage -rec 01.09.06  
6 Gain calibration-per pad or per group  4 B/channel   2.40E+06 4.46E+08[6] month Offline  yes ? (Roland, Marian) physics run 1e6 pp events (minimal)  1e8 pp events   - OCDB yes usage: sim and rec  
    2.40E+06   Offline   physisc run 1e4 pp events - OCDB yes usage: sim and rec  
7 Tail cancellation (CoG change)  4Bx6(Parameters)     [7] 1.30E+07[8] Rare Offline   Physics run (Black events) selected pulses out of 10M  pp events (100TB)   OCDB yes Not needed for reconstruction. CoG might be needed, but not yet clear   
DCS values                            
                           
6 Temperature probes  8B/channel   1.00E+06[9] DCS update frequency ~0.1 Hz  minute ( defined by deadband in compression ) DCS 30/10/06 - - - DCS ArchiveDB/OCDB yes 01.05.07 4
7 Temperature probes (FEC) 8B/channel   1.00E+06[10] DCS update frequency ~0.1 Hz  10 minutes ( defined by deadband in compression ) DCS 30/10/07 - - - DCS ArchiveDB/OCDB yes 01.05.07 4
8 Pressure 4By atmospheric pressure + 4By overpresure   1.00E+04[11] DCS update frequency ~0.1 Hz  10 s ( defined by deadband in compression ) DCS 30/10/08 - - - DCS ArchiveDB/OCDB yes 01.05.07 4
9 Drift velocity monitor (Goofie) 8By    200[12] - 20 minutes DCS 30/10/09 - - - DCS ArchiveDB/OCDB yes 01.05.07 4
10 Gas gain 0, 1 (Goofie) 8By    400[13] - DCS 30/10/10 - - - DCS ArchiveDB/OCDB yes 01.05.07 4
11 Gas composition (Goofie) 8By (CO2 and N)   400[14] - DCS 30/10/11 - - - DCS ArchiveDB/OCDB yes 01.05.07 4
12 Gas composition (Analyzer) 8By (O2, H2O and CO2)   600[15] - DCS 30/10/12 - - - DCS ArchiveDB/OCDB yes 01.05.07 4
13 Chamber - HV (Current) 8By     5.00E+05[16]   DCS 30/10/13 - - - DCS ArchiveDB/OCDB yes 01.05.07 4                            
14 Chamber - HV (Voltages) 8By     5.00E+04[17]   DCS 30/10/14 - - - DCS ArchiveDB/OCDB yes 01.05.07 4                            
15 Drift - VHV (Current) 8By     2.00E+03[18]   DCS 30/10/15 - - - DCS ArchiveDB/OCDB yes 01.05.07 4                            
16 Drift - VHV (Voltages) 8By     200[19]   DCS 30/10/16 - - - DCS ArchiveDB/OCDB yes 01.05.07 4                            
Data stream and event trailer  
1 Moving average & zero suppression  state ON/OFF     1 - Rare Event trailer   Dedicated run  1-10 black events sub-event  - yes   -
2 ALTRO frequency 1 value/detector   - - Rare Event trailer   - - - - yes   -
3 ALTRO acquisition window      - - pp/HI Event trailer   - - - - yes   -
state pp/HI 1 value/detector   1 - pp/HI Event trailer   - - - - yes   -
4 Active channel list -   - - - Data stream   - - - - yes   -
Space point corrections
1 Drift velocity map     1.00E+06   Rare Offline   - - - -      
2 ExB map     1.00E+06   Rare Offline   Dedicated runs with laser for various value of B       
3 Space charge map     2.00E+06     Offline                
2.04E+07 1.61E+09
TPC calibration
Raw DATA volume:
Black event :
Number of pads x number of time bins x 10 bit
557568x1Kx10b  ==> 700 MB
Zero suppressed event:
1 MBy  (pile-up of 5 events – ALICE Computing TDR)
….
Events per year ~ 10^9 pp 
Size of histograms pad x pad calibration :
557568 pads x 100 bins (non equidistant binning) x 2By ~ 111 MBy
Histograms stored in CDB
Calibration per pad:
Nevents x Ntracks x Nrows/Npads = X
Nevents – Number of events to get statistic X for pad calibration
Ntracks – number of tracks per event – for pp ~ 8
Nrows   - number of rows per track ~ 150
X           - required statistic per pad calibration
Nevents = X x (Npads/ (Ntracks*Nrows)) ~ X x 500
---------------------------------------------------------------
Nevents x Ntracks /(Npadsmax*NSector) = X
Npadsmax – maximal number of pads per row ~ 140
NSectors   - 36
Nevents = X x (Npadsmax*NSector/Ntracks) ~ X x 630
Calibration per bin (pad – pad row grouping):
Nevents x Ntracks x Nrows / (Nbinspad x Nbinspadrow x Nsectors) = X
Nbinspad   ~ 10
Nbinsrow  ~ 10
Nsectors        72
Nevents = X x (Nbinspad x Nbinspadrow x Nsectors)/  (Ntracks x Nrows) ~ X x 6
Gain Calibration (pad by pad):
Needed statistic determined by required precision of gain calibration ~ relative uncertainty  3 sigma ~ 1 %
Lwidth - Relative width (sigma) of landau distribution ~ 30 %
W = Lwidth/sqrt(X)
Number of points in histogram X ~ 1000 Landau distribution of amplitudes
Minimal  Required number of events  ~   0.5 x10^6 pp events
Source :  OFFLINE,   HLT (?) 
Algrithm:
                 OFFLINE    histograming of normalized amplitude spectra (Normalization to the track length). Postprocessing of ESD files after reconstruction.
Gain calibration with Kr, another possibility.
Calibration of parameters for tail cancellation:
Needed statistic determined by required precision of gain calibration ~ 5 % -W
Pwidth - Relative width of Gaussian distribution ~ 20 %  but + outliers
W = Pwidth/sqrt(X)
Number of points in histogram X ~ 50 (Gaussian distribution of values)
Cuttof factor   -  only signal over threshold  ~ 150 ADC ~ 20
Pad by pad:
Required number of events ~  0.5 x10^6 pp black events
Data volume   to process                     ~ 10^3 TBy
Pad grouping:
Required number of events ~  0.5 x10^4 pp black events
Data volume   to procces                     ~ 10 TBy
Source : OFFLINE
               Neccesity of pad- by pad calibration will be tested during TPC commisioing with laser pulse.
               Defualt approach – calibration for group of pads
Algorithm:
Update frequency: year (voltage setting, gas mixture)
Sim rec -  NO
DCS values:
Array of pair of time stamps, values
Maximal writing rate : 1 Hz, precision of time stamp value -1 s 
Reading rate from DCS database: ~ 1000 values per second
Data size:  8 Bytes per entry pair (time stamp, value)
Update frequency: Defined by interpolation precision – deadband  in DCS, highest frequency ~ 1 Hz.
Deadband defined as critical change of the current measured value in respect with previously stored one.
CDB storage: Graph of DCS values (value, time stamp), respectively fitted values (linear or parabolic fit for a range of time) 
                                 
Temperature probes:
Deadband value: Required precision ~ 0.05 C, The accuracy of temperature measurement ~ 0.05 C
Update frequency: 
10 s  update frequency in DCS (minimum is 5 second)
1 minutes for offline
500 channels -           No deadband used.  Old/New comparison (exact matching).
                                    5 sensors for chamber  x 36
                                    6x 18 – outer cylinder
                                    6 x  2 +  inner cylinder
                                    2 x 12    - conical section
44          - inside conical volume
4500 FEC channels: Deadband precision ~ 0.1 C
Depends on the resolution of temperature sensors and on the variation of temperature during day.  Day variation, switch on-off detector electronic (~ 1 degree ?). To be compressed (fitted) on minute level in Shuttle.
Source: (DCS Archive DB) temperature probes
Number of channels (sensors): ~ 4500 FEC and ~ 500 distributed on the surface of detector.
Expected data volume per run: 5000(sensors) x 4 (hours) x 60 (minutes) x 8 Bytes (9MBy)
Reading time from DCS: 20 min
Pressure probes:
Deadband value: Defined by required precision of drift velocity correction, and gas gain correction ~ 0.1 mbar
Wd ~  E/p
Townsend coefficient alpha = f(E/p)
Typical variation of gain ~ -0.3 % per mbar
Deadband value:  0.1 mbar
Update frequency: Defined by deadband value and by variation during day (~ 15 mbar),  minimum  ~ 10 s
DCS  - defined only by deadband value 0.1 mbar
Offline – combination of deadband around linear fit
Source: DCS, pressure probes
Number of channels: 2
                                     1  – atmospheric pressure
                                     2. - overpressure
Expected data volume: ~ kBy per run
Reading time from DCS: bellow 1 s
GAS composition  & Drift velocity monitor:
Deadband value:
Analysers: CO2: 0.1% deadband, update frequency according to this.
O2, H2O: 1 ppm deadband
Goofie: Update frequency: ~20 min
 Values: Vd, Gain1, Gain2, CO2, N2, Pressure, Temp, Time stamp.
No deadband –
                          Vd -  precision       ~0.05%
  Gain1 and Gain2  ~ 0.1 %
   Co2 and Nitrogen content calculated form vd and Gain (corrected for pressure and temperature) precission ~ 0.1 %  
DCS  - not yet connected 
Number of channels: 1  - CO2 content – CO2 analyzer
                                    
Deadband value:  relative 10^-4 – too optimistic – what is the resolution?
                               I will put there +-3 sigma of resolution 
Expected data volume: ~ kBy
Expected data volume: ~ kBy
High voltage (HV - ISEG):
Source: DCS
OFFLINE usage:  monitoring 
Number of channels: three per sector 
Dead band value: 
                        New value on change, Frequency 1 Hz
            Required precision  
Voltage: minimum step of power supply ~  0.5 Volt 
            Current: ~  resolution 4 nA 
Update frequency: voltage ~ run
                                 Current ~ ?
Expected data volume: 
                                 Number of samples * Nchannels 
Very High voltage (VHV):
Source: DCS (Joachim) - Heinzinger
Number of channels: 1
Deadband value:
Measured precision ~ 10 bits ? 
            Required precision  
Voltage: relative change (reproducibility)  ~ 10 ^-3
                Stability ~ 10^-5
                Mesurement ~ 10^-4
            Current: relative change ~ 0.01
Update frequency: voltage ~ run
                                 Current 
Expected data volume: 
                                 Number of samples * Nchannels 
[1]
Yves Schutz:
Implemented and tested end of september
[2]
Yves Schutz:
557568x1Kx10b ~ 700 MB
[3]
Yves Schutz:
4 By x 557568 pads  ~ 2.4  Mby
[4]
Yves Schutz:
557568 histograms x 100 bins 
[5]
Yves Schutz:
4 By x 557568  ~ 2.4 Mby
[6]
Yves Schutz:
557568 histograms x 100 bins 
[7]
Yves Schutz:
4By x 6 (values) x 557586 pads ~ 13 MB
[8]
Yves Schutz:
4By x 6 (values) x 557586 pads ~ 13 MB
[9]
Yves Schutz:
8By x 500x (probes) x Run time(4h) xFrequency( 1/minute)  ~ 1MBy(4hours)
[10]
Yves Schutz:
8By x 4500x (probes) x Run time(4h) xFrequency( 1/10minute)  ~ 1MBy
[11]
Yves Schutz:
8By x 2 (probes) x Run time(4h) xFrequency( 1/10s)  ~ 10kBy
[12]
Yves Schutz:
8By x Run time(4h) xFrequency(1/20 minute)  ~ 200By
[13]
Yves Schutz:
8By x 2 channels x Run time(4h) xFrequency(1/20 minute)  ~ 400By
[14]
Yves Schutz:
8By x 2 channels x Run time(4h) xFrequency(1/20 minute)  ~ 400By
[15]
Yves Schutz:
8By x 3 channels x Run time(4h) xFrequency(1/20 minute)  ~ 600By
[16]
Yves Schutz:
8By x 3 channels x 72 sectorsxRunTime(4h)xFrequency(1/minutes) ~0.5 MBy 
[17]
Yves Schutz:
8By x 3 channels x 72 sectorsxRunTime(4h)xFrequency(1/10 minutes) ~0.05 MBy 
[18]
Yves Schutz:
8By x 1 channels x xRunTime(4h)xFrequency(1/minutes) ~2 KBy 
[19]
Yves Schutz:
8By x 1 channels xRunTime(4h)xFrequency(1/10 minutes) ~200 By