Ssd simulation status : preparation for embedding Jonathan Bouchet Svt Meeting : 09/19
Ssd simulation chain SsdSimulationMaker SLS (slow simulator) Simulate charge over the strips SPA Signal+pedestal+noise Addition Pedestal supression Zero supression SsdClusterMaker SCF Scan the strips Form the clusters SCM Associate the clusters
Ssd Reconstruction Chain SsdDaqmaker Scan the strips and store the pedestals, id and noise into a ssdStripCalib table SsdPointMaker Do the reconstruction of the clusters Match the clusters to form a hit
To recall
Ssd soft : real data vs simu Many differences between the 2 codes Real Data : pick the ssdStripCalib (pedestal and noise) accorging to the time stamp in /Calibration/ssd. Reco : (s/n)>5 && noise > 0. Geometry : ssdWafersPosition.C (the geometry with alignment corrections) Simu : The actual calibration : pedestal =100 ADC, noise = 3 ADC in /starDb/svt/ssd/ Reco : (s/n)> 5 only Geometry : geom.C (old) To have a consistent simu with respect to the data, we must have makers and parameters (noise, pedestal, geometry) equivalent
New way for embedding StSsdDaqMaker DAQ File StSsdSimulationMaker Fzd File StSsdPointMaker St_scf_Maker St_scm_Maker Spa_strip old chain : StSsdSimulationMaker + St_scf_Maker + St_scm_Maker New chain : StSsdSimulationMaker + StSsdPointMaker
Noise of the clusters Simulation N-side P-side Real data
Noise : with the new calibration P-sideN-side
Charge of the clusters Pedestal = 100 ADC Noise = 3 ADC With real pedestal and noise values Remove the large tail for the fake clusters (noise)
Comparison between the 2 chains: signal to noise Old chainNew chain
Size of the clusters [strips] The 2 chains give the same results Old chainNew chain
Comparison with the efficiencies Within the bars errors, efficiencies look the same for the 2 chains Old chainNew chain
Efficiency The efficiency looked better (agree with the real data) But it was before a correction (use simu flag in StSsdDbMaker) After this correction, efficiency get back to 70 % Bands in residuals The difference between efficiencies in simu vs real data still here until now
Summary,outlook 1.Old Db vs ssdStripCalib Effect on the clusters properties (noise,size) Consistant with the real data 2.Old reco chain vs new reco chain The 2 chains give similars results (number of ssdHit reco,efficiency) Not entirely ready for embedding (see previous problem) 3. Check efficiency (if agree with real data) 4.Then eliminate old makers (status of StSsdEvalMaker)-->rename Ssd simu chain 5.Create StSsdMixer for embedding