200 likes | 298 Views
Pixel Results from CRAFT09. U. Langenegger (PSI), G. Giurgiu (JHU) Pixel General Meeting November 10, 2009. CRAFT09 Analyses. Gain calibration + Validation – R. Rougny (Antwerpen), U. Langenegger (PSI) Pixel hit efficiency - L. Mucibello, R. Rougny, N. van Remortel (Antwerpen)
E N D
Pixel Results from CRAFT09 U. Langenegger (PSI), G. Giurgiu (JHU) Pixel General Meeting November 10, 2009
CRAFT09 Analyses • Gain calibration + Validation – R. Rougny (Antwerpen), U. Langenegger (PSI) • Pixel hit efficiency - L. Mucibello, R. Rougny, N. van Remortel (Antwerpen) • Pixel hit resolution – K. Ulmer (Colorado) • BPIX Lorentz angle – M. Ivova, V. Chiochia (Zurich) • FPIX Lorentz angle – A. Kumar, A. Godshalk, A. Kharchilava (Buffalo) • Data/MC comparison – A. Jaeger, V. Chiochia (Zurich), M.Swartz (JHU) • All analyses done with most recent CRAFT09 reprocessing • /Cosmics/CRAFT09-TrackingPointing-CRAFT09_R_V4_CosmicsSeq_v1/RAW-RECO
Gain Calibration R. Rougny (Antwerpen), U. Langenegger (PSI) • Pixel thresholds minimized before CRAFT09 • → procedure led to significant fraction of negative BPIX pedestals • After CRAFT09, detector settings changed to fix negative pedestals (Ben Kreis, D. Kotlinski) • → new gain calibration taken (run 117680) • → fraction of negative pedestals indeed negligible in new calibration: FPIX BPIX
Gain Calibration ValidationU. Langenegger (PSI), R.Rougny (Antwerpen) • Analyze post-CRAFT09 data to validate new gain calibration CRAFT09 data Runs 119226 119094 119090 119088 119079 119022 119017 118969 118878 118762 118621 - Fraction of post-CRAFT09 data - No problems seen
Gain Calibration ValidationU. Langenegger (PSI) In http://www.phys.ethz.ch/~ursl/cms/091110/ you can find a higher statistics version of the CRAFT09-II gain calibration validation. This is still without the SP skim. CRAFT09-II ---------- cluster charge: http://www.phys.ethz.ch/~ursl/cms/091110/craft09-II-clusterCharge.ps BPIX Landau+Gauss: http://www.phys.ethz.ch/~ursl/cms/091110/craft09-II-bpix.ps FPIX Landau+Gauss: http://www.phys.ethz.ch/~ursl/cms/091110/craft09-II-fpix.ps reprocessed CRAFT09 SP skim --------------------------- cluster charge: http://www.phys.ethz.ch/~ursl/cms/091110/reprocessed-sp-data-clusterCharge.ps BPIX Landau+Gauss: http://www.phys.ethz.ch/~ursl/cms/091110/reprocessed-sp-data-bpix.ps FPIX Landau+Gauss: http://www.phys.ethz.ch/~ursl/cms/091110/reprocessed-sp-data-fpix.ps The MPV have come down a bit in both BPIX and FPIX, the effect of the slightly lowered gains, visible in the overlayed offline payload comparison for the new and previous gain calibration runs http://www.phys.ethz.ch/~ursl/cms/091110/gains-108062-117680.png The relative width has decreased (increased) slightly for the BPIX (FPIX). The gain calibration looks good
Data / MC Comparison – Cluster Charge A. Jaeger, V. Chiochia (Zurich) • Fair agreement between data and MC • Disagreement at low charge could be explained by lower thresholds in MC than in data • Andreas will produce MC with higher thresholds which match data barrel barrel
Data / MC Comparison – Pixel Charrge A. Jaeger, V. Chiochia (Zurich) barrel barrel
Data / MC Comparison – Pixel Hit Probability A. Jaeger, V. Chiochia (Zurich) • Pixel hit probability calculated as chi2 probability of the matching between the observed • cluster shape and the expected template • Fair qualitative agreement between data and MC • → important test since we plan to use pixel probability to improve tracking (remove bad • hits, split merged clusters…) barrel • Work in progress • re-digitize MC with higher thresholds to match MC • optimize phase space to emulate collisions better (~4000 electrons)
Pixel Hit EfficiencyL. Mucibello, R. Rougny, N. van Remortel (Antwerpen)
Pixel Hit EfficiencyL. Mucibello, R. Rougny, N. van Remortel (Antwerpen) Eff = Nvalid / (Nvalid + Nmissing) Pixel sensor efficiency ~ 98.3 – 98.5% White boxes = known bad modules Layer 1 Layer 2 Layer 3 low statistics - Will investigate pixel efficiency with strip seeded tracks to avoid biases in efficiency
Pixel Hit ResolutionK. Ulmer (Colorado) • Pixel hit resolution measured in CRAFT09 using the “double difference” method • - Compare measured resolution with predicted errors from template based cluster parameter • estimator (CPETemplate) - Resolution in microns : reprocessed original processing measured predicted measured predicted X 18 ± 1 15 ± 1 19 ± 3 15 ± 1 Y 26 ± 1 23 ± 1 32 ± 2 25 ± 2 • Measured resolution about 10-20% worse than predicted by CPE • Will repeat measurements with much better statistics with collisions • Will correct CPE predicted errors to match observed resolution
E BPIX Lorentz Angle – Cluster Size MethodM. Ivova, V. Chiochia (Zurich) - Good agreement LA measured in latest and first processing as well as with the PIXELAV Simulation cot(a)min = -0.408 +/- 0.002 – latest CRAFT09 reprocessing cot(a)min = -0.405 +/- 0.003 – first CRAFT09 processing cot(a)min = -0.397 +/- 0.003 – PIXELAV simulation (M. Swartz) B field OFF B field ON LA consistent with zero within 2s qLorentz ≈ 22°
FPIX Lorentz Angle – Cluster Size Method A. Kumar, A. Godshalk, A. Kharchilava (Buffalo) - Good agreement LA measured in latest and first processing as well as with the PIXELAV Simulation cot(a)min = -0.084 +/- 0.008 – first CRAFT09 reprocessing cot(a)min = -0.080 +/- 0.005 – latest CRAFT09 processing cot(a)min = -0.081 +/- 0.003 – PIXELAV simulation (M. Swartz) B field OFF B field ON LA consistent with zero qLorentz ≈ 5°
Avg. Drift (mm) Depth (mm) FPIX Lorentz Angle – Grazing Angle Method A. Kumar, A. Godshalk, A. Kharchilava (Buffalo) • Buffalo group also measured FPIX LA using grazing angle method in CRAFT08 and got • good agreement with cluster size method • → nice proof of principle • 3.75° ± 0.41° - grazing angle • 3.95° ± 0.39° - cluster size • This is important since with collisions • cluster size method is not adequate • With collisions grazing angle method • will be used for BPIX • For FPIX neither grazing angle nor • cluster size are optimal with collision tracks • → might have to rely on cosmics • only
Conclusion • Gain calibration in place and validated with post CRAFT09 data • Lorentz angles measured and in agreement with previous processing • and with PIXELAV simulation • Pixel sensor efficiency ~98.5% • - Next step: use strip seeded tracks to avoid biases • Reasonable data/MC agreement when collision like cosmic tracks • selected but some cosmic specific effects not simulated • Pixel hit resolution within 10-20% from expectation • Will repeat with collisions and adjust predicted errors if needed
Gain Calibration ValidationU. Langenegger (PSI), R.Rougny (Antverpen)
Gain Calibration ValidationU. Langenegger (PSI), R.Rougny (Antverpen) CRAFT09 data - Fraction of post-CRAFT09 data - No problems seen