AtlasPublicTopicHeader.png

ApprovedPlotsTileReconstruction

Introduction

Signal Reconstruction

Time Reconstruction

Energy Reconstruction

Absolute difference between the amplitude reconstructed online (DSP) and offline (OFL) as a function of amplitude (OFL) for Charge Injection data in High Gain (top) and Low Gain (bottom).


Contact: alberto.valero@cern.ch
Reference: Approval Meeting
Date: May 2010
Fig1_Run146031_CISramp_EdiffHG.png Fig3_Run146031_CISramp_EdiffHG.png
Absolute difference between the amplitude reconstructed online (DSP) and offline using the DSP emulation (OFL-DSP-E) as a function of amplitude offline (OFL-DSP-E) for Charge Injection data in High Gain (top) and Low Gain (bottom). The residual difference represent the different precision in the online and offline calibration constants.


Contact: alberto.valero@cern.ch
Reference: Approval Meeting
Date: May 2010
Fig2_Run146031_CISramp_EmulateDSP_EdiffHG.png Fig4_Run146031_CISramp_EmulateDSP_EdiffLG.png
Absolute difference between the amplitude reconstructed online (DSP) and offline using the DSP emulation (OFL-DSP-E) as a function of amplitude offline (OFL-DSP-E) for Charge Injection data in High Gain. The calibration to pC units (top) is performed offline to the DSP result in order to avoid differences due to calibration constants precision.


Contact: alberto.valero@cern.ch
Reference: Approval Meeting
Date: May 2010
Fig5_Run146027_CISramp_EmulateDSPEdiff.png Fig6_CISmono_146027_Energy_EmulateDSPEdiff.png
Difference between the signal amplitude calculated on pseudodata with the Non-Iterative Optimal Filtering Algorithm online, using the Digital Signal Processor (EDSP), and offline (EOFL NI). The signal amplitudes are measured in ADC counts. The dashed lines indicate the maximum expected difference due to fixed point arithmetic. The data shown correspond to about 55000 events generated in the whole ADC range with a phase scan from -30 ns to +30 ns.


Contact: alberto.valero@cern.ch
Reference: Approval Meeting
Date: Aug 2010
Ediff_adc_lab.png
Difference between the reconstructed offline energy, , and the energy given by the DSP relative to and as a function of (in GeV), extracted from cosmic muon runs.
Contact:
Reference: CDS
Edsp_ofl_comp.eps
[eps] [png]

Quality factor as a function of reconstructed amplitude in TileCal pulse simulator with ideal pulse shape, but timing and noise effects emulated. No out-of-time pile-up. One can observe that with ideal pulse shapes the quality factor is not energy dependent. The x-axis shows the amplitude in ADC counts before cell-dependent calibration constants are applied. The calibration factor is approximately 12 MeV per ADC count.


Contact: pawel.klimek@cern.ch
Reference: Public Tile Calorimeter Plots for Collision Data and Link to CDS record, plots and Link to CDS record, poster and proceedings of IEEE NSS, Valencia, Spain, 2011
Date: November 2011
QFvsAopt_MC_hi_IdealPulse.png
Quality factor as a function of reconstructed amplitude in TileCal pulse simulator with non-ideal pulse shapes, timing and noise effects emulated. No out-of-time pile-up. There is amplitude dependence of quality factor. The x-axis shows the amplitude in ADC counts before cell-dependent calibration constants are applied. The calibration factor is approximately 12 MeV per ADC count.


Contact: pawel.klimek@cern.ch
Reference: Public Tile Calorimeter Plots for Collision Data and Link to CDS record, plots and Link to CDS record, poster and proceedings of IEEE NSS, Valencia, Spain, 2011
Date: November 2011
QFvsAopt_MC_hi.png
Illustration of out-of-time pile-up (+50 ns) in ATLAS Tile Calorimeter with pulse shapes similar to those in the real detector.


Contact: pawel.klimek@cern.ch
Reference: Public Tile Calorimeter Plots for Collision Data and Link to CDS record, plots and Link to CDS record, poster and proceedings of IEEE NSS, Valencia, Spain, 2011
Date: November 2011
Pileup_Pulses.png
Quality factor as a function of the amplitude in different pile-up scenarios in TileCal pulse simulator with ideal pulse shape. Ain (Aout) is the amplitude of the in-time (out-of-time). Here neither noise nor timing effects are emulated. The x-axis shows the amplitude in ADC counts before cell-dependent calibration constants are applied. The calibration factor is approximately 12 MeV per ADC count.


Contact: pawel.klimek@cern.ch
Reference: Public Tile Calorimeter Plots for Collision Data and Link to CDS record, plots and Link to CDS record, poster and proceedings of IEEE NSS, Valencia, Spain, 2011
Date: November 2011
QF_vs_Amp_MC_hi.png
Normalized distributions of quality factor in TileCal simulator with ideal pulse shape. Generated amplitude of in-time pulse is 12 GeV. Amplitude of the out-of-time pulse follows the distribution of the ZeroBias trigger with a cut on 34 ADC counts (deviation between reconstructed and true amplitude is ~1% in a presence of out-of time-pulse with amplitude 34 ADC counts). Three cases: No out-of-time pile-up (black), with out-of-time pile-up (-50 ns), with out-of-time pile-up (+50 ns). The Zero Bias data was recorded while the LHC was running with only 2 bunches per beam separated by at least 2.5 μs and corresponds to the runs: 177531, 177539, 177540, 177593, 177682 from March 2011.


Contact: pawel.klimek@cern.ch
Reference: Public Tile Calorimeter Plots for Collision Data and Link to CDS record, plots and Link to CDS record, poster and proceedings of IEEE NSS, Valencia, Spain, 2011
Date: November 2011
Pileup_IdealPulse_MC_hi.png
Normalized distributions of quality factor in TileCal simulator with non-ideal pulse shape. Generated amplitude of in-time pulse is 12 GeV. Amplitude of the out-of-time pulse follows the distribution in ZeroBias stream with a cut on 34 ADC counts (deviation between reconstructed and true amplitude is ~1% in a presence of out-of time-pulse with amplitude 34 ADC counts). Three cases: No out-of-time pile-up (black), with out-of-time pile-up (-50 ns), with out-of-time pile-up (+50 ns). The Zero Bias data was recorded while the LHC was running with only 2 bunches per beam separated by at least 2.5 μs and corresponds to the runs: 177531, 177539, 177540, 177593, 177682 from March 2011.


Contact: pawel.klimek@cern.ch
Reference: Public Tile Calorimeter Plots for Collision Data and Link to CDS record, plots and Link to CDS record, poster and proceedings of IEEE NSS, Valencia, Spain, 2011
Date: November 2011
Pileup_NonIdealPulse_MC_hi.png
The plot in the top shows the absolute difference between the time reconstructed online (DSP) and offline (OFL) as a function of amplitude (OFL) for pseudo-data. The reconstructed time in the DSP is limited by the fixed point arithmetic and the use of a look up table for the divisions. The plot in the bottom shows the absolute difference between the time reconstructed online (DSP) and offline (OFL) as a function of the time offline (OFL) for pseudo-data. The difference between DSP and OFL increases with the time of the pulse.


Contact: alberto.valero@cern.ch
Reference: Approval Meeting
Date: May 2010
Fig7_LABADC_tdiffvsE.png Fig8_LABADC_tdiffvst.png
Absolute difference between the time reconstructed online (DSP) and offline (OFL) as a function of amplitude (OFL) for Charge Injection data in High Gain (top) and Low Gain (bottom).


Contact: alberto.valero@cern.ch
Reference: Approval Meeting
Date: May 2010
Fig9_Run146027_CISramp_tdiffHG.png Fig10_Run146027_CISramp_tdiffLG.png
Absolute difference between the time reconstructed online (DSP) and offline with the DSP emulation (OFL-DSP-E) as a function of amplitude (OFL-DSP-E) for Charge Injection data in High Gain (top) and Low Gain (bottom).


Contact: alberto.valero@cern.ch
Reference: Approval Meeting
Date: May 2010
Fig11_Run146027_CISramp_EmulateDSP_tdiffHG.png Fig12_Run146027_CISramp_EmulateDSP_tdiffLG.png

Signal-to-noise ratio with an alternative method

Signal and noise distributions for both the OF2 without iterations and Matched Filter (MF) methods for the channel 15 of LBA47. The dataset used in this preliminary study was simulated and it comprises noise signals taken from pedestal run and event signals generated by convoluting noise signals with the Tilecal reference pulse shape with a known amplitude distribution. The MF method is very similar to the OF2 without iterations but it estimates the pedestal instead of being immune against it, as the OF2 without iterations is. The pedestal value used was the mean value of the first sample of the incoming signals (both signal and noise). The aim of this study is to evaluate the impact of signal detection, and therefore amplitude estimation, when using an estimated value of the pedestal.
Contact: Bernardo Sotto-Maior Peralva bernardo@cernNOSPAMPLEASE.ch
Reference: CDS record
Date: Sep 2011
OF.gif MF.gif
Signal and noise distributions for both the OF2 without iterations and Matched Filter (MF) methods for the cell A4 of LBA47, i.e the channels' signals were summed up before estimating the amplitude. The dataset used in this preliminary study was simulated and it comprises noise signals taken from pedestal run and event signals generated by convoluting noise signals with the Tilecal reference pulse shape with a known amplitude distribution. The MF method is very similar to the OF2 without iterations but it estimates the pedestal instead of being immune against it, as the OF2 without iterations is. The pedestal value used was the mean value of the first sample of the incoming signals (both signal and noise).
Contact: Bernardo Sotto-Maior Peralva bernardo@cernNOSPAMPLEASE.ch
Reference: CDS record
Date: Sep 2011
OF2.gif MF2.gif
Amplitude relative error for both the OF2 without iterations and Matched Filter (MF) methods for both single channel (channel 15) and cell (A4) analysis of LBA47. The dataset used in this preliminary study was simulated and it comprises noise signals taken from pedestal run and event signals generated by convoluting noise signals with the Tilecal reference pulse shape with a known amplitude distribution. The MF method is very similar to the OF2 without iterations but it estimates the pedestal instead of being immune against it, as the OF2 without iterations is. The pedestal value used was the mean value of the first sample of the incoming signals (both signal and noise).
Contact: Bernardo Sotto-Maior Peralva bernardo@cernNOSPAMPLEASE.ch
Reference: CDS record
Date: Sep 2011
errorSingleChan.gif errorCell.gif
Receiver Operating Characteristics (ROC) curves for the OF2 without iterations and Matched Filter (MF) methods for both analysis, i.e single channel (channel 15) and cell A4 (sum of channels' signals 15 and 18 before estimating the amplitude). The dataset used in this preliminary study was simulated and it comprises noise signals taken from pedestal run and event signals generated by convoluting noise signals with the Tilecal reference pulse shape with a known amplitude distribution. The MF method is very similar to the OF2 without iterations but it estimates the pedestal instead of being immune against it, as the OF2 without iterations is. The pedestal value used was the mean value of the first sample of the incoming signals (both signal and noise).
Contact: Bernardo Sotto-Maior Peralva bernardo@cernNOSPAMPLEASE.ch
Reference: CDS record
Date: Sep 2011
rocDetection.gif


Major updates:
-- HenricWilkens - 08-Aug-2013

Responsible: HenricWilkens
Subject: public

Edit | Attach | Watch | Print version | History: r3 < r2 < r1 | Backlinks | Raw View | Raw edit | More topic actions...
Topic revision: r1 - 2013-08-08 - HenricWilkens
 
    • Cern Search Icon Cern Search
    • TWiki Search Icon TWiki Search
    • Google Search Icon Google Search

    Atlas All webs login

  • Edit
  • Attach
This site is powered by the TWiki collaboration platform Powered by PerlCopyright & 2008-2019 by the contributing authors. All material on this collaboration platform is the property of the contributing authors.
Ideas, requests, problems regarding TWiki? Send feedback