The GTPC Package: Tracking and Analysis Software for GEM TPCs Linear Collider TPC R&D Meeting LBNL, Berkeley, California (USA) 18-19 October, 003 Steffen Kappler Institut für Experimentelle Kernphysik, Karlsruhe University (Germany) CERN, EP Division, Geneva (Switzerland)
Outline 4 The GEM-TPC with STAR Electronics 4 The GTPC Package 4 Overview 4 Pedestal & Noise Analysis 4 Zero-Suppression 4 Cluster Finding & Space-Point Determination 4 Space-Point based Tracking 4 Measurement Methods: 4 Efficiency 4 Spatial Resolution 4 Summary
The GEM-TPC with STAR Electronics Overview Drift cylinder + Inner diameter d=0cm + Length l=5cm Detector readout + Double-GEM, 10x10cm + 56 micro-pads, 1.7x1.5mm Readout electronics + Modified version of the STAR-TPC electronics
Front-end Electronics and DAQ Characteristics STAR-TPC Front-end electronics - + low noise (1075 e rms for 5pF) + Pseudo-Gaussian pulse shape, 90-190ns peak time, 180ns fwhm + Variable sampling clock: 10-40MHz + Here: 180ns peak time 19.7 MHz signal sampling rate 50.86ns per time slice TPC FEE cards RDO controller opt. fibre VME-crate Data Acquisition + Signal digitized in the FEE cards + via flat cable to the RDO controller + via optical fibre to the VME module + via ethernet to the DAQ & Online PC DAQ & Online PC ethernet external trigger data files in.cpt - format
Data Analysis Software General overview Features + Visualization + DAQ online monitoring + Batch analysis capability Functionality + Pedestal & noise analysis + Zero suppression + Space-point based tracking: + Combinatorial track finder + c based track fit + Provides typical histograms Realization + OO software design + Realized in Microsoft VB (due to historical reasons) + Works only under Microsoft Windows Java version currently being developed by J. Kaminski
Data Analysis Software General overview Example event + Three projections, spacepoints and track fit for a MIP in Ar-CO (70:30). 3 at G = 4 10 + Time evolution of one pad: Undershoot due to hardwired ion tail correction
Data Analysis Software Pedestal & Noise Analysis Signal example + Response to a cluster after 0cm drift in Ar-CH 4 90:10 a) before and b) after signal inversion and pedestal subtraction a) Noise analysis (avg. noise <1.5 ADC) Test pulse resp. (induced via GEM) b)
Data Analysis Software Zero Suppression Method + Applied after signal inversion and pedestal subtraction + Only for voxels of each pad with. - Signal \S\ >.5 N (pad noise). - Signal \S\ >.5 n (avg. noise of all pads) time slice n and signal S are stored + Additionally, the noise N of each pad is stored Suppression efficiency + Data from the beam tests at CERN could be suppressed from 50kB down to 0kB per event (8%).
Data Analysis Software Zero Suppression Method + Applied after signal inversion and pedestal subtraction + Only for voxels of each pad with. - Signal \S\ >.5 N (pad noise). - Signal \S\ >.5 n (avg. noise of all pads) time slice n and signal S are stored + Additionally, the noise N of each pad is stored Suppression efficiency + Data from the beam tests at CERN could be suppressed from 50kB down to 0kB per event (8%). OFFLINE process, so that the original data isn t lost...
Data Analysis Method Cluster Finding & Space-Point Determination Cluster finding + Search each pad row for clusters + Sum all cluster voxels with S/N > 3 + Accept clusters with: - Cluster S/N > 4 - Cluster size: # pads fwhm > 1 # pads over noise > z (t) p =50 ns + Determine the space-point by Center-Of-Gravity or Gaussian fit in one pad row x p = 1.7mm. 3 Example event: MIP in Ar-CH (95:5) at G = 3 10 4
Data Analysis Method Tracking I Tracking + Accept only space-points from clusters with S/N > 5 + Combinatorial track finder: 1) Seed search in row -5, seed length limit of 5mm. Example event: MIP in Ar-CH (95:5) at G = 3 10 3 4
Data Analysis Method Tracking I Tracking + Accept only space-points from clusters with S/N > 5 + Combinatorial track finder: 1) Seed search ) For each seed add space-points a) within a tube of 10s b) within a cone of 6 c) Update track parameters each time a space point is added. Example event: MIP in Ar-CH (95:5) at G = 3 10 3 4
Data Analysis Method Tracking I Tracking + Accept only space-points from clusters with S/N > 5 + Combinatorial track finder: 1) Seed search ) For each seed add space-points 3) Likelihood L() based decision: a) The track with the best L() value is preliminarily accepted b) c based track fit is performed c) All contributing space-points are flagged out for the next iteration. Example event: MIP in Ar-CH (95:5) at G = 3 10 3 4
Data Analysis Method Tracking I Tracking + Accept only space-points from clusters with S/N > 5 + Combinatorial track finder: 1) Seed search ) For each seed add space-points 3) Likelihood L() based decision: a) The track with the best L() value is preliminarily accepted b) c based track fit is performed c) All contributing space-points are flagged out for the next iteration PS: Possibility to make geom. constraints for speeding up the combinatorial track finding. Example event: MIP in Ar-CH (95:5) at G = 3 10 3 4
Data Analysis Method Tracking II Track Analysis + Tracks with 5 or more spacepoints are finally accepted c < 1.5 8000 6000 + Tracks with c < 5.0 are accepted for general analysis 000 c < 5.0 counts 4000 000 + Tracks with c < 1.5 are accepted for efficiency and spatial resolution analysis counts 1500 1000 0 0 4 6 8 10 # space-points per track PS: The s ij used in the c calculation are set to the actual residual - width of each pad row 500 0 0 4 6 reduced c # points > 5
Data Analysis Method Tracking III Multi-track events + ~50ms reconstruction time per event (zero-suppressed data, 1GHz Pentium CPU, 51MB RAM). 3 Example event: MIP in TDR gas at G = 3 10
Data Analysis Method Tracking III Multi-track events + ~50ms reconstruction time per event (zero-suppressed data, 1GHz Pentium CPU, 51MB RAM) + Reasonable performance in multitrack events counts 000 1500 1000 500 0 0 4 6 reduced c. 3 Example event: MIP in TDR gas at G = 3 10
Data Analysis Method Tracking III Multi-track events + ~50ms reconstruction time per event (zero-suppressed data, 1GHz Pentium CPU, 51MB RAM) + Reasonable performance in multitrack events Resolution of overlapping events + Important for the study of multi-track separation + Track finding: yes + Cluster finding: not yet. 3 Example event: MIP in TDR gas at G = 3 10
Data Analysis Method Measurement of Efficiency & Spatial Resolution Example event: MIP in Ar-CH4 (95:5) at G = 3. 103 In good tracks (c < 1.5, well isolated) efficiency and spatial resolution are measured for each pad row i (0 to 7) independently by: + Repeating the track fits while excluding any space points from the target row i + Searching for each track the space point pi in row i with the shortest distance di to the track Single-pad-row spatial resolution measurement: + Residuals given by the components of di Single-pad-row efficiency measurement: j j + Is there a space point with di <4 s? (in each component)
Summary 4 Tracking and Analysis Software for GEM TPCs: The GTPC Package 4 OO software design, currently realized in Microsoft VB 4 Java Version is being developed by J. Kaminski 4 Functionality: 4 Pedestal & noise analysis 4 Zero suppression 4 Space-point based tracking: 4 Combinatorial track finder 4 c based track fit 4 Provides typical histograms 4 Analysis of single-pad-row efficiency and spatial resolution