L1TAU1_VTE200 trigger efficiency as a function of the energy sum of two egamma clusters. Efficiency is calculated as a ratio of events passing L1TAU1_VTE200 trigger to all events corresponding to γγ → e+e- process passing a supporting trigger requirement. Data points are compared with the fit to 2018 reference trigger efficiency derived for L1TAU1_TE4_VTE200 trigger https://arxiv.org/abs/2008.05355![]() |
![]() pdf png |
L1TAU1_TE3_VTE200 trigger efficiency as a function of the energy sum of two egamma clusters. Efficiency is calculated as a ratio of events passing L1TAU1_TE3_VTE200 trigger to all events corresponding to γγ → e+e- process passing a supporting trigger requirement. Data points are compared with the fit to 2018 reference trigger efficiency derived for L1TAU1_TE4_VTE200 trigger https://arxiv.org/abs/2008.05355![]() |
![]() pdf png |
L12TAU1_VTE200 trigger efficiency as a function of the energy sum of two egamma clusters.Efficiency is calculated as a ratio of events passing L12TAU1_VTE200 trigger to all events corresponding to γγ → e+e- process passing a supporting trigger requirement. Data points are compared with the fit to 2018 reference trigger efficiency derived for L1TAU1_TE4_VTE200 trigger https://arxiv.org/abs/2008.05355![]() |
![]() pdf png |
Simulated L1_TE4 trigger efficiency as a function of the energy sum of two egamma clusters. Efficiency is calculated as a ratio of events passing L1TAU1_VTE200 trigger to all events corresponding to γγ → e+e- process passing a supporting trigger requirement. Decision of L1_TE4 was simulated using a cut on a total energy distribution on events selected by L1TAU1_VTE200 trigger. Data points are compared with the fit to 2018 reference trigger efficiency derived for L1TAU1_TE4_VTE200 trigger https://arxiv.org/abs/2008.05355![]() |
![]() pdf png |
Correlation between transverse energy of two egamma clusters corresponding to the γγ → e+e- process. |
![]() pdf png |
Acoplanarity distribution for the γγ → e+e- process. Data events pass the L1TAU1_VTE200 trigger. Data points are compared with MC simulation scaled to the integrated luminosity, cross section and 2018 reference trigger efficiency. Error bars denote statistical uncertainties. |
![]() pdf png |
Track-eta distribution for events passing the γγ → e+e- selection. Data events pass the L1TAU1_VTE200 trigger. Data points are compared with MC simulation scaled to the integrated luminosity, cross section and 2018 reference trigger efficiency. Error bars denote statistical uncertainties. |
![]() pdf png |
L1Calo single electron trigger efficiencies in the inner calorimeter
barrel |η| < 0.8 for the legacy system (red) and the Phase-I system (blue) as
function of the electron pT . The Phase-I electron identification is provided by
the electron Feature Extractor (eFEX). The efficiencies are measured in data
recorded in ATLAS Run 438532 using electrons from Z → ee decays. The
event sample is triggered by the L1Calo legacy system using the L1_EM22VHI
trigger chain with both legacy and eFex readout enabled. Strict selection criteria
are applied on event and electron level to ensure clean events. Examples are that
the invariant mass of leading and subleading electron match the Z mass, and
that the electrons satisfy a likelihood-based ’tight’ identification. The efficiency
is computed using the Tag-and-Probe method with read out Trigger Objects
(TOBs) matched to leading and subleading electrons. The only requirement
for the efficiency is that there is a TOB with ET > 22 GeV matched to the
subleading electron in the η - φ plane, within a radius of ΔR < 0.15. No
isolation or hadronic veto bits are required.
ATL-COM-DAQ-2022-122![]() |
![]() pdf png |
Correlation of the L1Calo transverse energies measured by the legacy
Cluster Processor (CP) system and the Phase-1 electron Feature Extractor
(eFEX) for the inner ElectroMagnetic Barrel (EMB) |η| < 0.8. The correla-
tion is measured in data recorded in ATLAS Run 438532 using electrons from
Z → ee decays. The event sample is triggered by the L1Calo legacy system
using the L1_EM22VHI trigger chain with both legacy and eFex readout en-
abled. Strict selection criteria are applied on event and electron level to ensure
clean events. Examples are that the invariant mass of leading and subleading
electron match the Z mass, and that the electrons satisfy a likelihood-based
’tight’ identification. The Trigger Objects (TOBs), originating from electro-
magnetic (EM) calorimeter objects (electrons and photons), read out by both
systems are matched in the η - φ plane, within a radius of ΔR < 0.15, to the
leading and subleading electrons. Shown is the correlation of the transverse
energies for the TOBs matched to the leading electron.
ATL-COM-DAQ-2022-122![]() |
![]() pdf png |
Correlation of the L1Calo transverse energies measured by the legacy
Cluster Processor (CP) system and the Phase-1 electron Feature Extractor
(eFEX) for the inner ElectroMagnetic Barrel (EMB) |η| < 0.8. The correla-
tion is measured in data recorded in ATLAS Run 438532 using electrons from
Z → ee decays. The event sample is triggered by the L1Calo legacy system
using the L1_EM22VHI trigger chain with both legacy and eFex readout en-
abled. Strict selection criteria are applied on event and electron level to ensure
clean events. Examples are that the invariant mass of leading and subleading
electron match the Z mass, and that the electrons satisfy a likelihood-based
’tight’ identification. The Trigger Objects (TOBs), originating from electro-
magnetic (EM) calorimeter objects (electrons and photons), read out by both
systems are matched in the η - φ plane, within a radius of ΔR < 0.15, to the
leading and subleading electrons. Shown is the correlation of the transverse
energies for the TOBs matched to the subleading electron.
ATL-COM-DAQ-2022-122![]() |
![]() pdf png |
Electron-gamma output data comparison of the upgraded Level-1 calorimeter trigger system with respect to the legacy system. Phase-1 electron Feature EXtractor (eFEX) output data is shown in the Y -axis whereas the Cluster Processor Module (CPM) is on the X-axis. Trigger OBjects (TOBs) found using the new system are matched with the legacy system in the η − φ plane, within a radius of δR < 0.2. The region covered by this plot is limited due to ongoing commissioning work. Data for |η| > 0.8 and half of the φ coverage were unavailable at the time of this run.
ATL-COM-DAQ-2022-083![]() |
![]() pdf png |
Tau output data comparison of the upgraded Level-1 calorimeter trigger system with respect to the legacy system. Phase-1 electron Feature EXtractor (eFEX) output data is shown in the Y -axis whereas the Cluster Processor Module (CPM) is on the X-axis. Trigger OBjects (TOBs) found using the new system are matched with the legacy system in the η − φ plane, within a radius of δR < 0.2. The region covered by this plot is limited due to ongoing commissioning work. Data for |η| > 0.8 and half of the φ coverage were unavailable at the time of this run.
ATL-COM-DAQ-2022-083![]() |
![]() pdf png |
The raw energy correlation between Level-1 Calorimeter Trigger (L1Calo) ADC counts and Liquid Argon (LAr) transverse energies, ET , for trigger towers in the electromagnetic barrel (EMB), 0 < |η| < 1.4. The offline LAr ET is derived by summing the transverse energies of the individual calorimeter cells associated to a trigger tower. The L1Calo ADC counts are the digital values sampled at peak position of the analogue pulses within the L1Calo PreProcessor system. Streams of trigger tower ADC values, sampled at 40 MHz corresponding to the LHC bunch crossing frequency, are the inputs for the legacy L1Calo system. The calibration aims for a slope of 4 ADC counts per 1 GeV above the pedestal offset of 32 ADC counts. The figure shows the correlation in √s =13.6 TeV data for ATLAS runs 431178 to 431812 based on a loose Z→ e+e− event selection.
ATL-COM-DAQ-2022-083![]() |
![]() pdf png |
The L1Calo PreProcessor Module (PPM) input timing offset for the first Run 3 run (red), and for runs after a calibration update (blue) for trigger towers in the Liquid Argon (LAr) electromagnetic endcap (EMEC), 1.5 < |η| < 3.2. The misalignment at the beginning of Run 3 was caused by changes introduced with the Phase-I upgrade which also influences the analogue signal path to the L1Calo legacy system. The PPM input timing, marking the arrival time of analogue trigger tower signals in the L1Calo PreProcessor, was updated to re-align all trigger towers in order to guarantee the correct sampling of the analog pulse. The difference in time between the digitization point and the analogue signal peak, called the time offset, was derived by fitting functional forms to pulses in collision data. While the timing calibration update was based on special data acquisition runs with doubled digitisation rate of 80 MHz in the L1Calo PPMs, runs after the update could only be recorded with the standard digitization frequency of 40 MHz. This reduced resolution and hence increased imprecision of the fit results is responsible for the tails in the validation sample (blue curve). The red curve was derived from Run 3 collision data with √s =13.6 TeV recorded with 80 MHz sampling during ATLAS run 427394, while the blue curve is based on runs 431178 to 431812 with standard 40 MHz digitization frequency. A loose Z→ e+e− selection is applied in the analysis.
ATL-COM-DAQ-2022-083![]() |
![]() pdf png |
Eta-phi matched transverse energy deposits transmitted from the TREX and captured at the eFEX input. Both devices are read-out via the FELIX-SWROD path. The data matches between the eFEX input and the TREX.
ATL-COM-DAQ-2022-083![]() |
![]() pdf png |
TREX data readout is performed both via the legacy ROD and via the new FELIX-SWROD, which are two independent data paths. The figure shows the transverse energy results (LUT CP) read-out via the legacy and the FELIX path. The data agreement validates the new readout path.
ATL-COM-DAQ-2022-083![]() |
![]() pdf png |
Single-electron trigger efficiency computed from Z→ ee Monte Carlo simulation, comparing the performance of the existing electron trigger with the proposed Run-3 trigger ATLAS-TDR-023![]() ![]() |
![]() pdf png |
Di-tau trigger efficiency computed from Z→ τ τ Monte Carlo simulation, with a 20 (12) GeV threshold on the leading (subleading) tau, with respect to the offline reconstructed tau candidates. The energy threshold corresponds to the primary Run-2 di-tau trigger, without the additional topological selection applied. The Run-3 isolation thresholds were tuned to produce the same rate as the Run-2 trigger. Run-3 taus are reconstructed in eFEX, and the isolation requirement is computed from surrounding energy as seen in eFEX (grey) or jFEX (red). The availability of a larger surrounding area in jFEX, albeit with coarser granularity, significantly improves the performance of the isolation.
ATL-COM-DAQ-2022-021![]() |
![]() png pdf |
Efficiency of the new missing transverse momentum (ETmiss) algorithms proposed for the Run-3 jFEX and gFEX, compared to the Run-2 ETmiss trigger. The efficiency is computed from ZH → νν bb Monte Carlo simulation with respect to the offline ETmiss using the Tight working point ATLAS-CONF-2018-023![]() ![]() |
![]() png pdf |
The offsets from the ideal input timing delays in the L1Calo PreProcessor modules, measured with LHC beam splash events from beam 1 recorded during ATLAS run 405495. Shown are the results for the electromagnetic (left) and for the hadronic (right) calorimeter layer. The different time-of-flight between particles originating from beam splashes and from the nominal interaction point is corrected for in the measurement, in particular taking into account the opened detector position on the C side. The results in the very forward regions |η|>3.2 are affected by large systematic errors in the time-of-flight corrections. LHC beam 1 traverses the ATLAS detector from positive to negative η values.
ATL-COM-DAQ-2021-089![]() |
![]() png pdf |
![]() png pdf |
The offsets from the ideal input timing delays in the L1Calo PreProcessor modules, measured with LHC beam splash events from beam 2 recorded during ATLAS run 405495. Shown are the results for the electromagnetic (left) and for the hadronic (right) calorimeter layer. The different time-of-flight between particles originating from beam splashes and from the nominal interaction point is corrected for in the measurement, in particular taking into account the opened detector position on the C side. The results in the very forward regions |η|>3.2 are affected by large systematic errors in the time-of-flight corrections. LHC beam 2 traverses the ATLAS detector from negative to positive η values.
ATL-COM-DAQ-2021-089![]() |
![]() png pdf |
![]() png pdf |
The 10bit ADC value for the central time slice in the readout window of the L1Calo PreProcessor modules for all towers in the electromagnetic (left) and in the hadronic (right) calorimeter layer.
The total readout window spans 15 ADC slices sampled at 80 MHz. Shown is the η-φ map for the LHC beam 1 splash event number 50896 recorded during ATLAS run 405495. The display is provided by the L1Calo mapping tool, an essential application for monitoring the performance of the L1Calo trigger during ATLAS operation. LHC beam 1 traverses the ATLAS detector from positive to negative η values.
ATL-COM-DAQ-2021-089![]() |
![]() png pdf |
![]() png pdf |
The 10bit ADC values for 16 towers of the electromagnetic calorimeter end-cap on the C side, read out from the L1Calo PreProcessor modules. The total readout window spans 15 ADC slices sampled at 80 MHz. Shown is a measurement for the LHC beam 1 splash event number 50896 recorded during ATLAS run 405495. The display is provided by the L1Calo mapping tool, an essential application for monitoring the performance of the L1Calo trigger during ATLAS operation. LHC beam 1 traverses the ATLAS detector from positive to negative η values.
ATL-COM-DAQ-2021-089![]() |
![]() png pdf |
The signal delays introduced by the Phase-I upgrade of the LAr front-end electronics for the L1Calo trigger tower signals are shown. The delays of the arrival times in the L1Calo PreProcessor system with respect to Run 2 are determined from measurements of the peak position in L1Calo layer timing scans using pulses from the LAr charge injection system. Signals are collected for all trigger towers in the electromagnetic barrel calorimeter partition on the A side (EMBA). Signals from the area where the LAr Phase-I demonstrator electronics were installed during Run 2 (0 < η <1.5, 1.8 < φ <2.2) are excluded from the plot.
ATL-COM-DAQ-2021-035![]() |
![]() png pdf |
||
The delays of signals from the Presampler (PS) [left], front layer (FR) [middle] and back layer (BK) [right] with respect to the middle layer (MD) introduced by the Phase-I upgrade of the LAr front-end electronics for the L1Calo trigger tower signals are displayed. Shown are the differences in arrival time in the L1Calo PreProcessor system, determined from measurements of the peak position in L1Calo layer timing scans using pulses from the LAr charge injection system, for all trigger towers of the electromagnetic barrel calorimeter partition on the A side (EMBA). The plot compares the situation during Run 2, i.e. before the installation of the Phase-I LAr Trigger Digitizer Boards (LTDBs), after the installation of the LTDBs, and after recalibration of the LAr Tower Builder Board (TBB) delays which compensate for the introduced differences of the signal propagation times.
ATL-COM-DAQ-2021-035![]() |
![]() png pdf |
![]() png pdf |
![]() png pdf |
Per-jet efficiency computed from a HH→bb(bb) Monte Carlo simulation comparing the performance of the Run 2 trigger system with the proposed system after the Phase-1 upgrade, described in detail in ATLAS-TDR-023![]() ![]() ![]() |
![]() png pdf eps |
Per-jet efficiency for jets with nearby jets computed from a HH→bb(bb) Monte Carlo simulation comparing the performance of the Run 2 trigger system with the proposed system after the Phase-1 upgrade, described in detail in ATLAS-TDR-023![]() ![]() ![]() |
![]() png pdf eps |
Efficiency computed from a Z→ee Monte Carlo simulation comparing the performance of the existing electron trigger with the proposed trigger that will be implemented during the Phase-1 upgrade, described in detail ATLAS-TDR-023![]() ![]() ![]() ![]() ![]() ![]() |
![]() png pdf eps |
Efficiency of the new missing transverse energy (MET) algorithm proposed for the Run 3 jFEX ATLAS-TDR-023![]() ![]() |
![]() png pdf eps |
L1Topo Hardware-Simulation Mismatches:
Mismatch rates between L1Topo hardware and simulation. First row shows for each trigger chain the ratio
between number of events selected by the simulation but not by the hardware and the total number of events
accepted by the simulation. Second row shows the ratio of number of events selected by the hardware but not by the
simulation and the total number of events accepted by hardware. Masses and energies are expressed in GeV and
angles in radians. Sub-indexes “i, j” denote lists of trigger objects, “1” means leading in terms of transverse
momentum.
ATL-COM-DAQ-2018-170![]() |
![]() png pdf eps |
|
L1Topo Hardware-Simulation Mismatches:
Mismatch rates between L1Topo hardware and simulation. First row shows for each trigger chain the ratio
between number of events selected by the simulation but not by the hardware and the total number of events
accepted by the simulation. Second row shows the ratio of number of events selected by the hardware but not by the
simulation and the total number of events accepted by hardware. Masses and energies are expressed in GeV and
angles in radians. Sub-indexes “i, j” denote lists of trigger objects; bx+1 refers to the bunch crossing immediately after
the event. Super-indexes “C, F” for jets denote central (|η|<2.5) and full range (|η|<4.9), respectively.
ATL-COM-DAQ-2018-170![]() |
![]() png pdf eps |
Average Pedestal Shift (8b4e):
The plots show the average deviation of the digitised Level-1 Calorimeter Trigger input signals from the expected flat baseline as function of the bunch crossing (BC) number for a selected part of the LHC orbit. The pedestal baseline is the signal height in absence of energy depositions in a given trigger tower. One FADC count corresponds to a transverse energy deposit of approximately 250 MeV.
Shown is the mean pedestal shift taking into account all towers of the LAr Electromagnetic Calorimeter Barrel A partition (EMB-A) [left] and of the LAr Forward Calorimeter 1 A partition (FCAL1-A) [right] for lumi blocks 78 and 540 of ATLAS run 340368, corresponding to average mu values of 58.1 and 38.1, respectively. The corresponding LHC fill is 6370 in 8b4e bunch filling scheme. Only read out FADC values corresponding to filled LHC bunches are taken into account.
ATL-COM-DAQ-2018-004![]() |
![]() png pdf |
![]() png pdf |
Average Pedestal Shift (8b4e):
The plots show the average deviation of the digitised Level-1 Calorimeter Trigger input signals from the expected flat baseline as function of the bunch crossing (BC) number for a selected part of the LHC orbit. The pedestal baseline is the signal height in absence of energy depositions in a given trigger tower. One FADC count corresponds to a transverse energy deposit of approximately 250 MeV. Shown is the mean pedestal shift taking into account all towers of the LAr Electromagnetic Calorimeter Barrel A partition (EMB-A) [left] and of the LAr Forward Calorimeter 1 A partition (FCAL1-A) [right] for lumi blocks 78 and 540 of ATLAS run 340368, corresponding to average mu values of 58.1 and 38.1, respectively. The corresponding LHC fill is 6370 in 8b4e bunch filling scheme. The full readout window of 5 FADC counts centred around the triggered bunch crossing is taken into account. Filled bunches are indicated by grey background colour.
ATL-COM-DAQ-2018-004![]() |
![]() png pdf |
![]() png pdf |
Average Pedestal Correction (8b4e):
The plots show the pedestal correction as function of the bunch crossing (BC) number for a selected part of the LHC orbit. It is continuously calculated and applied by the firmware of the new Multichip Modules in the PreProcessor electronics of the Level-1 Calorimeter Trigger (L1Calo) in order to correct online the ET calculation result of each trigger tower for pile-up induced baseline shifts. To enhance the signal over noise ratio, L1Calo uses a finite impulse response filter operated on five consecutive values of the digitised input signal. The pedestal correction in units of FIR Counts (i.e. weighted FADC counts) is the difference between an average of each trigger tower's digital filter output over 65536 LHC orbits (approximately 6s) and a corresponding target value determined by the filter output in absence of any energy deposition. Shown is the mean correction for all towers of the LAr Electromagnetic Calorimeter Barrel A partition (EMB-A) [left] and of the LAr Forward Calorimeter 1 A partition (FCAL1-A) [right] for lumi blocks 78 and 540 of ATLAS run 340368, corresponding to average mu values of 58.1 and 38.1, respectively. The corresponding LHC fill is 6370 in 8b4e bunch filling scheme. A readout value of the pedestal correction is available only for filled bunches.
ATL-COM-DAQ-2018-004![]() |
![]() png pdf |
![]() png pdf |
Pedestal Correction:
The plot shows the Pedestal Correction as calculated by the firmware of the new Multichip Module (nMCM), a component on the PreProcessor of the Level-1 Calorimeter Trigger (L1Calo), for a selected part of the LHC orbit. The pedestal correction is computed for each bunch crossing, and is plotted in units of weighted ADC counts. The average pedestal correction is calculated over 65536 LHC orbits, which corresponds to a duration of approximately 6 seconds. Shown is the correction for the LAr calorimeter Electromagnetic Barrel A partition (EMB-A).
This plot is taken from the L1Calo offline monitoring that uses a dedicated data stream (express stream). The data was taken by ATLAS in run 327342 with =33.8. The corresponding LHC fill is 5849. ATL-COM-DAQ-2017-064 ![]() |
![]() png pdf |
|
Average Pedestal Shift:
The plot shows the average deviation from the expected flat baseline of the digitized L1Calo input signals as a function of the Bunch Crossing (BC) Number for a selected part of the LHC orbit. The baseline is the signal height in the absence of energy depositions in a given trigger tower. For this plot only input signals from the LAr calorimeter Electromagnetic Barrel A partition (EMB-A) are taken into account. The average is constructed over all trigger towers without significant energy depositions and over all considered events in the given ATLAS run. One ADC count corresponds to a transverse energy deposition of approximately 0.25 GeV.
This plot is taken from the L1Calo offline monitoring that uses a dedicated data stream (express stream). The data was taken by ATLAS in run 327342 with =33.8. The corresponding LHC fill is 5849.
ATL-COM-DAQ-2017-064![]() |
![]() png pdf |
Comparison of the L1Topo decision from hardware and from simulation for several L1Topo trigger
items. For this plot, 258k events have been analysed. Empty bins correspond to: < 3.8x10−4 [%]. Simulated
decisions are either from the ”non-bitwise” implementation or from the ”bitwise-correct” one. The fraction of
events for which both simulated outputs disagree with the hardware output is indicated in blue. The fraction of
events for which only the ”bitwise-correct” (”non-bitwise”) simulation differs from the hardware output is indicated
in red (green). The bitwise-correct implementation can significantly improve the accuracy of the simulation. Only
the L1Topo items that were used in active trigger in 2016 run are shown. Persistent mismatches (blue/red areas)
are mostly due to small rounding errors in the invariant mass calculation.
ATL-COM-DAQ-2017-008![]() |
![]() png |
|
Left:
Fraction of events for which the L1Topo ΔηΔϕ algorithm produced a different outcome in firmware and in simulation.
The allowed ranges of Δη and Δϕ are indicated, as well as the pT requirements used to select the input object used to compute these quantities. A small fraction of differences is currently expected because the firmware quantities are implemented as integers, while the ones in simulation are floating point values.
ATL-COM-DAQ-2016-143![]() Right: Fraction of events for which the L1Topo H_{T} algorithm produced a different outcome in firmware and in simulation. The minimum required H_{T} is indicated, as well as the requirements applied to the jets used to compute H_{T}. A small fraction of differences is currently expected because the firmware quantities are implemented as integers, while the ones in simulation are floating point values. ATL-COM-DAQ-2016-143 ![]() |
![]() png eps |
![]() png eps |
Left:
The trigger rate for the missing ET trigger with a threshold at 35 GeV per bunch is plotted as function of the inst. Lumi per bunch. The rates are shown for different settings with and w/o pedestal correction applied. The pedestal correction minimise pile-up effects and linearises the trigger
rate.
Right: The trigger rate for the missing ET trigger with a threshold at 50 GeV per bunch is plotted as function of the inst. Lumi per bunch. The rates are shown for different settings with and w/o pedestal correction applied. The pedestal correction minimise pile-up effects and linearises the trigger rate. |
![]() png pdf eps |
![]() png pdf eps |
Left:
The figure shows the trigger rates per bunch for various em triggers in 25 and 50ns operation as function of the inst. Lumi per bunch. A linear behaviour for all items is observed as well as a good agreement for the different bunch spacing schemes. EM12 and EM15 are triggers with thresholds at 12 and 15 GeV respectively. EM20VHI has additional requirements on isolation (electromagnetic and hadronic) applied.
Right: The trigger rate for the missing ET trigger with a threshold at 35 GeV per bunch is plotted as function of the inst. Lumi per bunch. The rates are shown for different settings with and w/o pedestal correction applied. The pedestal correction minimise pile-up effects and linearises the trigger rate. |
![]() png pdf eps |
![]() png pdf eps |
Left:
The figure shows the rate per bunch of the missing ET trigger with a threshold of 35 GeV (XE35) as function of the bunch position within a bunch train. Due to the interplay of in-time and out-of-time pile-up which leads to a higher level of the pedestal an increased rate at the beginning of the bunch train is observed.
Right: The figure shows the rate per bunch of the missing ET trigger with a threshold of 35 GeV (XE35) as function of the position within a bunch train. The interplay of in-time and out-of-time pile-up leads to an increased level of the pedestal at the beginning of the bunch train. A pedestal correction algorithm implemented in firmware compensates for this effect and results in stable rates over the full bunch train. |
![]() png |
![]() png | The figure shows the rate per bunch of the missing ET trigger with a threshold of 35 GeV (XE35) as function of the position within a bunch train. The interplay of in-time and out-of-time pile-up leads to an increased level of the pedestal at the beginning of the bunch train. A pedestal correction algorithm implemented in firmware compensates for this effect and results in stable rates over the full bunch train. |
![]() png |
Left:
The Figure shows the normalised coefficients of the Finite Impulse Response Filter (FIR) for the electromagnetic layer. The coefficients are shown for a matched filter which is given by the signal pulse shape assuming white noise and no correlations between the five input ADC slices. The x-axis indicates the different η-bins for which the filters are shown. The y-axis indicates the five coefficients per η-bin and the z-axis shows the normalised filter value.
Right: The Figure shows the normalised coefficients of the Finite Impulse Response Filter (FIR) for the electromagnetic layer. The coefficients are shown for an autocorrelation filter for 25ns bunch spacing. The filters are given by the signal pulse shape. They take correlations between different ADC input slices from out of time pile-up into account. The x-axis indicates the different η-bins for which the filters are shown. The y-axis indicates the five coefficients per η-bin and the z-axis shows the normalised filter value. |
![]() png pdf eps |
![]() png pdf eps |
Left:
The Figure shows the normalised coefficients of the Finite Impulse Response Filter (FIR) for the hadronic layer. The coefficients are shown for a matched filter which is given by the signal pulse shape assuming white noise and no correlations between the five input ADC slices. The x-axis indicates the different η-bins for which the filters are shown. The y-axis indicates the five coefficients per η-bin and the z-axis shows the normalised filter value.
Right: The Figure shows the normalised coefficients of the Finite Impulse Response Filter (FIR) for the hadronic layer. The coefficients are shown for an autocorrelation filter for 25ns bunch spacing. The filters are given by the signal pulse shape. They take correlations between different ADC input slices from out of time pile-up into account. The x-axis indicates the different η-bins for which the filters are shown. The y-axis indicates the five coefficients per η-bin and the z-axis shows the normalised filter value. |
![]() png pdf eps |
![]() png pdf eps |
Left:
The figure shows the efficiency that a calorimeter pulse is identified in the correct bunch crossing as function of its offline energy by the Trigger logic. The performance of a matched filter is compared to an autocorrelation filter for the electromagnetic barrel (EMB). Since the level of out of time pile-up is rather low, the filters are very similar to each other and consequently the performance is close.
Right: The figure shows the efficiency that a calorimeter pulse is identified in the correct bunch crossing as function of its offline energy by the Trigger logic. The performance of a matched filter is compared to an autocorrelation filter for the inner wheel of the electromagnetic endcap. Since the level of out of time pile-up is significant, the filters are different and consequently the performance is significantly better for autocorrelation filters. |
![]() png pdf eps |
![]() png pdf eps | The figure shows the efficiency that a calorimeter pulse is identified in the correct bunch crossing as function of its offline energy by the Trigger logic. The performance of a matched filter is compared to an autocorrelation filter for the forward calorimeter (FCAL). Since the level of out of time pile-up is significant, the filters are different and consequently the performance is significantly better for autocorrelation filters. |
![]() png pdf eps |
Left:
The figure shows the distribution of the transverse energy for em candidates identified within the cluster processor system of the Level-1 Calorimeter Trigger. The information is read out from the new Common Merger Module (CMX). The data were recorded during initial pp collisions in 2015 with protons colliding at centre of mass energy of √s=13TeV.
Right: The figure shows the distribution of the transverse energy for em candidates identified within the cluster processor system of the Level-1 Calorimeter Trigger which are transmitted to the Level-1 Topological Trigger. The information is read out from the new Common Merger Module (CMX). The data were recorded during initial pp collisions in 2015 with protons colliding at centre of mass energy of √s= 13TeV. |
![]() png pdf eps |
![]() png pdf eps |
Left:
The figure shows the distribution of the transverse energy for em candidates identified within the cluster processor system of the Level-1 Calorimeter Trigger. The information is read out from the new Common Merger Module (CMX). The data were recorded during initial pp collisions in 2015 with protons colliding at centre of mass energy of √s=13TeV.
Right: The figure shows the distribution of the transverse energy for em candidates identified within the cluster processor system of the Level-1 Calorimeter Trigger which are transmitted to the Level-1 Topological Trigger. The information is read out from the new Common Merger Module (CMX). The data were recorded during initial pp collisions in 2015 with protons colliding at centre of mass energy of √s= 13TeV. |
![]() png pdf eps |
![]() png pdf eps |
Left:
The estimated Level-1 trigger rate as a function of the missing ET (MET) threshold from 14 TeV minimum bias Monte Carlo for a <μ> = 54 and a 25 ns bunch spacing.
Shown are the operation scenarios with 2011 and 2012 noise cuts using matched FIR filters and two options for Run 2 with noise cuts optimised for a trigger tower
occupancy of 0.5% using autocorrelation FIR filters with and without a pedestal correction (p. c.) which are possible with the upgraded Level-1 calorimeter trigger system.
Right: The estimated Level-1 trigger rate as a function of the missing ET (MET) threshold from 14 TeV minimum bias Monte Carlo for a <μ> = 81 and a 25 ns bunch spacing. Shown are the operation scenarios with 2011 and 2012 noise cuts using matched FIR filters and two options for Run 2 with noise cuts optimised for a trigger tower occupancy of 1.0% using autocorrelation FIR filters with and without a pedestal correction (p. c.) which are possible with the upgraded Level-1 calorimeter trigger system. |
![]() png eps |
![]() png eps |
Left:
L1Calo trigger tower timing in ns as a function of η and φ for the electromagnetic (EM) calorimeter layer. The timing is derived by fitting the trigger tower ADC distributions using either a Gauss-Landau or Landau-Landau function, after
all timing corrections were applied in hardware. Precision of this method of timing determination
is estimated to be around 1 ns, also lowest step available to tune timing in L1Calo hardware (in PHOS4 chip) is
1ns. This plot shows the results using collision data from May 2012.
In ideal case of perfectly timed system all Trigger Towers would give entries at zero.
The plot shows that timing is within the target of +- 3ns for all Trigger Towers.
Right: The same for the hadronic (HAD) calorimeter layer. |
![]() png eps |
![]() png eps |
Left:
L1Calo trigger tower timing in ns as a function of η for the electromagnetic (EM) calorimeter layer. The timing is derived by fitting the trigger tower ADC distributions using either a Gauss-Landau or Landau-Landau function,
after all timing corrections were applied in hardware. Precision of this method of timing determination
is estimated to be around 1 ns, also lowest step available to tune timing in L1Calo hardware (in PHOS4 chip) is
1ns.
This plot shows the results using collision data from May 2012.
In ideal case of perfectly timed system all Trigger Towers would give entries at zero.
The plot shows that timing is within the target of +- 3ns.
Right: The same for the hadronic (HAD) calorimeter layer. |
![]() png eps |
![]() png eps |
Left:
L1Calo timing as a function of time since start of the run for the electromagnetic (EM) calorimeter layer.
The timing is an average and rms of the individual trigger tower timings which are derived using a simplified fitting method
based on the ADC peak position. This plot shows results obtained offline using collision data from
run 191426 (22 october 2011) compared with beam phase as measured by the Central Trigger. It is shown that during a run, timing
is stable to better than 1ns level.
Right: The same for the hadronic (HAD) calorimeter layer. |
![]() png eps |
![]() png eps |
Left:
The figure above concerns the trigger known within ATLAS as L1_EM18VHI, where ‘H’ denotes that a hadronic veto of 1GeV has been applied, ‘V’ denotes that the threshold energy is variable in η and ‘I’ denotes the use of electromagnetic isolation. This plot shows the efficiency turn on curve of this trigger after various levels of electromagnetic isolation have been applied.
The ET denoted on the above figure relates to offline reconstructed electron ET.
A subset of data identified as Z→ee candidates by standard offline reconstruction was selected from around
1 fb-1 taken around the end of November and early December, 2012.
The average number of interactions per bunch crossing in this data ranged from 10 to 40.
The efficiency is defined with respect to electrons from these Z candidates satisfying additional selection
criteria, among others:
- have |η|< 2.5 - are required to satisfy tight offline electron identification - are in active area of calorimeter - the invariant mass of the tag and probe must satisfy 80 < m < 100 GeV - are matched to L1 trigger EM object - are matched to High Level Trigger electron object - the tag electron must have an isolated track The error bars on the plots are statistical in nature. Right: The same for the trigger known within ATLAS as L1_EM25HI |
![]() png eps |
![]() png eps |
Left:
The figure above concerns the trigger known within ATLAS as L1_EM18VHI, where ‘H’ denotes that a hadronic veto of 1GeV has been applied,
‘V’ denotes that the threshold energy is variable in η and ‘I’ denotes the use of electromagnetic isolation. This plot shows the efficiency
at turn on plateau as a function of pile-up after various levels of electromagnetic isolation have been applied.
A subset of data identified as Z→ee candidates by standard offline reconstruction was selected from around 1 fb-1 taken around
the end of November and early December, 2012.
The average number of interactions per bunch crossing in this data ranged from 10 to 40.
The electron selection criteria are the same as those used for Figure 7.
The error bars on the plots are statistical in nature.
Right: The same for the trigger known within ATLAS as L1_EM25HI. |
![]() png eps |
![]() png eps |
The figure above investigates the triggers which, in ATLAS trigger nomenclature are notated as L1_16H, L1_EM18VH and L1_25H. Within this, the `H' in the name of these triggers denotes that they have already had a hadronic veto of ≤ 1 GeV applied and the letter `V' denotes that the threshold energy is variable in η. This plot shows the effectiveness of a veto isolation cut on the rates of the level 1 calorimeter triggers L1_EM16H, L1_EM18VH and L1_EM25H. The x-axis shows the relative rate reduction achieved by applying the extra isolation requirements and the y-axis is the relative reduction in rate. The labels denote isolation value was used for the corresponding trigger. Efficiency values are calculated using integrals between 30 and 100 GeV A subset of data identified as Z→ee candidates by standard offline reconstruction was selected from around 1 fb-1 taken around the end of November and early December, 2012. The average number of interactions per bunch crossing in this data ranged from 10 to 40. The cut on the energy of the probe electrons is set at 30 GeV when calculating the efficiencies for this plot. The electron selection criteria are the same as those used for Figure 7. The error bars are not included in this plot as they are viewed to be negligible. |
![]() png eps |
This plot shows the relative effectiveness of a fractional isolation cut on the rates of the level 1 calorimeter trigger L1_EM16 which is an electron trigger with a threshold of 16 GeV. The electromagnetic isolation considered here represents the total energy found in a ring of em trigger towers surrounding the area which caused the trigger. The three data series plotted are the trigger with an isolation veto as possible in the current firmware, the trigger with a fractional isolation where the isolation allowed is a fraction of the level 1 electron energy and the trigger with both the fractional isolation and a hadronic veto of ≤ 1 GeV. The x-axis shows the relative rate achieved by applying the extra isolation requirements and the y-axis is the relative efficiency. The labels denote which fraction (F) or isolation (I) was used for the corresponding trigger. Efficiency values are calculated using integrals between 25 and 100 GeV. The electron selection criteria are the same as those used for Figure 7. The statistical errors for the efficiencies are not shown as they are small compared to the points and systematic errors have not been considered at this point. |
![]() png eps |
Left:
This plot shows L1Calo receiver gains applied to signals in electromagnetic layer, as used at the end of 2012/13
data taking period.
Receiver gains are where L1Calo energy calibration is applied, ensuring correct energy scale on trigger tower level.
The gains are not uniform, because cables, carrying analog input signals from ATLAS front-end to L1Calo have different
length and attenuation. Another source of non-uniformities are differences in electronics response, corrections for
dead or noisy calorimeter cells and corrections for reduced high voltage.
Right: The same for L1Calo receiver gains applied to signals in hadronic LAr calorimeter. |
![]() png eps |
![]() png eps |
This plot shows L1Calo receiver gains applied to signals coming from Tile calorimeter, as used at the end of 2012/13 data taking period. Receiver gains are where L1Calo energy calibration is applied, ensuring correct energy scale on trigger tower level. The gains are not uniform, because cables, carrying analog input signals from ATLAS front-end to L1Calo have different length and attenuation. Another source of non-uniformities are differences in electronics response and corrections for reduced response in drawers in emergency mode. |
![]() png eps |
This plot shows relative change in L1Calo receiver gains, used by L1Calo to compensate for reduction of high voltage in Liquid Argon Presampler. The change happened on 28/9/2012 when HV was reduced to 1200 V. New receiver gains were calculated with offline script based on HV corrections for individual LAr cells and EM shower profile determined from analysis of collision data. Gains were updated for most of EM barrel, although for some areas the update was not necessary, as these were on reduced HV already. |
![]() png eps |
Left:
Fractional difference between L1Calo transverse energy and offline transverse
energy as a function of the offline transverse energy. The L1Calo energy is
calculated using two different methods; the energy based on the ADC peak
sample and the energy based on the result of the look-up-table (LUT). This
plot shows the distributions for the Liquid Argon electromagnetic barrel (EMB, -1.5< η < 1.5)
calorimeter using 2012 collision data, recorded on Oct. 21st. EMB-A and EMB-C represent different sides of the barrel, in terms of the pseudorapidity (A: η>0, C:η<0). The errors on the y-axis represent statistical errors, while the errors on the x-axis show the bin width. For most calorimeter partitions statistical errors are negligible.
Right: The same quantity for the Tile hadronic calorimeter (TILE, -1.5 < η < 1.5) using 2012 collision data, recorded on Oct. 21st. TILE-A and TILE-C represent different sides of the barrel, in terms of the pseudorapidity (A: η>0, C:η<0). |
![]() png eps |
![]() png eps |
Left:
Fractional difference between L1Calo transverse energy and offline transverse
energy as a function of the offline transverse energy. The L1Calo energy is
calculated using the energy based on the ADC peak sample. This plot shows the
distributions for all electromagnetic calorimeter partitions using 2012 collision
data, recorded on Oct. 21st. The partition names represent different parts of the detector in terms of the pseudorapidity (EMB/EMEC2/EMEC1/FCAL1: |η|< 1.5 /<1.8/<3.2/<4.9). The first bin for the FCAL1 (ET < 7 GeV) is subject to large pile-up effects and is therefore not shown. The errors on the y-axis represent statistical errors, while the errors on the x-axis represent the bin width.
For most calorimeter partitions statistical errors are negligible.
Right: The same quantity for all hadronic calorimeter partitions. The partition names represent different parts of the detector in terms of the pseudorapidity (TILE/HEC/FCAL23: |η| < 1.5/<3.2/<4.9). There was not enough data at high ET for the very forward regions (FCAL23), which is why those bins are empty. |
![]() png eps |
![]() png eps |
Left:
Ratio between offline transverse energy and L1Calo transverse energy as a
function of η. The L1Calo energy is calculated using the energy based on
the ADC peak sample. This plot shows the distribution for electromagnetic
calorimeter layer using 2012 collision data, recorded on Oct. 21st. The errors on the y-axis represent statistical errors, while the errors on the x-axis represent the bin width. For most calorimeter partitions statistical errors are negligible.
Righ: The same quantity for hadronic calorimeter layer. |
![]() png eps |
![]() png eps |
Level-1 Trigger cross-sections (rate/luminosity) for a selection of L1Calo-based trigger items. The left side of the figure corresponds to measurements from two 7TeV runs with 2011 nominal per-bunch luminosities, and colliding bunches delivered in bunch trains with 50ns spacing. The right side of the figure corresponds to a special high-luminosity 7TeV run with no bunch trains. The middle of the figure corresponds to an 8TeV run with 2012 nominal per-bunch luminosities and 50ns bunch trains. The falls in rate for XE50 and FJ75 triggers between 2011 and 2012 runs are due to trigger noise-cut increases in the forward regions of L1Calo. All other rate changes (increases) are due to the increased centre-of-mass energy.EM16 (EM30) is an electron-photon trigger with a threshold at 16 (30) GeV. EM16VH is an electron-photon trigger with an hadronic layer energy veto and varied thresholds across the calorimeter, with typically 16 GeV thresholds. TAU15 is an hadronically-decaying tau trigger with threshold at 15 GeV. XE50 is a trigger for missing ET above 50 GeV at the EM scale. XE50_BGRP7 is an XE50 trigger with a veto on the first 3 bunches of a bunch train. J75 is a trigger for a central jet (|η|<3.2) with ET above 75 GeV. FJ75 is a trigger for a jet in the forward region (|η|>3.2) with ET above 75 GeV. 4J10 is a trigger for four central jets with ET above 10 GeV. |
![]() png eps png (without fixed-rate lines) eps (without fixed-rate lines) |
Left:
Distribution of L1Calo Preprocessor ADC counts for four different regions of the EM FCAL calorimeter. The four bins represent divisions in |η| of the trigger towers, nominally: 3.1-3.2 (Bin 1), 3.2-3.5 (Bin 2), 3.5-4.2 (Bin 3), 4.2-4.9 (Bin 4). Zero-bias events from a single run were used for these distributions.
Right: Standard deviation (from RMS of distribution) of the L1Calo Preprocessor ADC distributions of four regions of the EM FCAL calorimeter, measured over a range of luminosities that have been quantified in terms of the interactions per bunch crossing. The four bins represent divisions in |η| of the trigger towers, nominally: 3.1-3.2 (Bin 1), 3.2-3.5 (Bin 2), 3.5-4.2 (Bin 3), 4.2-4.9 (Bin 4). Zero-bias events from a single run were used for these distributions. |
![]() png eps |
![]() png eps |
Left:
Distribution of L1Calo Preprocessor ADC counts for two different regions of the Hadronic FCAL2 calorimeter. The two bins represent divisions in |η| of the trigger towers, nominally: 3.1-3.5 (Bin 1), 3.5-4.9 (Bin 2). Zero-bias events from a single run were used for these distributions.
Right: Standard deviations (from RMS of distribution) of the L1Calo Preprocessor ADC distributions of two regions of the Hadronic FCAL2 calorimeter, measured over a range of luminosities that have been quantified in terms of the interactions per bunch crossing. The two bins represent divisions in |η| of the trigger towers, nominally: 3.1-3.5 (Bin 1), 3.5-4.9 (Bin 2). Zero-bias events from a single run were used for these distributions. |
![]() png eps |
![]() png eps |
Left:
Distribution of L1Calo Preprocessor ADC counts for two different regions of the Hadronic FCAL3 calorimeter. The two bins represent divisions in |η| of the trigger towers, nominally: 3.1-3.5 (Bin 1), 3.5-4.9 (Bin 2). Zero-bias events from a single run were used for these distributions.
Right: Standard deviations (from RMS of distribution) of the L1Calo Preprocessor ADC distributions of two regions of the Hadronic FCAL3 calorimeter, measured over a range of luminosities that have been quantified in terms of the interactions per bunch crossing. The two bins represent divisions in |η| of the trigger towers, nominally: 3.1-3.5 (Bin 1), 3.5-4.9 (Bin 2). Zero-bias events from a single run were used for these distributions. |
![]() png eps |
![]() png eps |
Left:
Distribution of L1Calo Preprocessor ADC counts for four different regions of the EM EndCap Inner Wheel calorimeter. The four bins represent divisions in |η| of the trigger towers, nominally: 2.5-2.7 (Bin 1), 2.7-2.9 (Bin 2), 2.9-3.1 (Bin 3), 3.1-3.2 (Bin 4). Zero-bias events from a single run were used for these distributions.
Right: Standard deviations (from RMS of distribution) of the L1Calo Preprocessor ADC distributions of four regions of the EM EndCap Inner Wheel calorimeter, measured over a range of luminosities that have been quantified in terms of the interactions per bunch crossing. The four bins represent divisions in |η| of the trigger towers, nominally: 2.5-2.7 (Bin 1), 2.7-2.9 (Bin 2), 2.9-3.1 (Bin 3), 3.1-3.2(Bin 4). Zero-bias events from a single run were used for these distributions. |
![]() png eps |
![]() png eps |
L1_XE50 (Missing ET Trigger) efficiency as a function of the offline topological cluster-based missing ET, for a sample of candidate W->e nu events (single electron with ET above 25 GeV, passing tight identification and other quality requirements, with the candidate W transverse mass greater than 40 GeV). Four different choices of trigger tower noise cuts were simulated offline, one with Forward Calorimeter (FCAL) noise cuts optimized to conditions with an average of 15 interactions per crossing, one with FCAL noise cuts optimized to 20 average interactions per crossing, one with both FCAL and Electromagnetic EndCap Inner Wheel (EMEC-IW) noise cuts optimized for 20 average interactions per crossing, and one with FCAL and EMEC-IW noise cuts optimized to 25 interactions per crossing. Data was taken from a single 2012 run, which had a peak luminosity of 23.1 average collisions per filled bunch crossing. |
![]() png eps |
FCAL/EMEC inner noise cuts and XE trigger Level-1 Missing Et (MET) rates as a function of threshold for several pile-up noise cut scenarios. The rates are estimated by applying noise cuts to ZeroBias events in run 191426 at a luminosity around 3.2x1033 and mu of 15. The 2011 configuration corresponds to noise cuts of approximately 1 GeV in all trigger towers. The loose forward noise cut applies cuts of 6.5, 5.5 and 2.5 GeV in the first FCAL layer and 4.5 in the second layer at |η| > 3.5. The tighter noise cuts raise these by 1 GeV, and also raise the noise cuts in all other towers beyond |η| = 2.5 by 0.5 GeV. The final case removes FCAL entirely from the Missing Et calculation. At this luminosity, the bulk of the fake MET rate reduction is achieved with the loose cuts. |
![]() png eps |
Timing monitoring 2010/11 Mean L1Calo timing as function of date for the electromagnetic (EM) and hadronic (Had) calorimeter partitions. The mean timing is an average of the individual trigger tower timings which are derived using a simplified fitting method based on the ADC peak position. The vertical lines indicate adjustments in the global CTP clock phase which synchronize the clock to the LHC radio-frequency system. |
![]() png eps |
![]() png eps |
Timing status March 2011 L1Calo trigger tower timing in ns as function of eta and phi for the electromagnetic (EM) and hadronic (HAD) calorimeter layer. The timing is derived by fitting the trigger tower ADC distributions using either a Gauss-Landau or Landau-Landau function. This plot shows the results using collision data from the initial 2011 running period. White bins have no measurement due to lack of statistics. |
![]() png eps |
![]() png eps |
Timing status April 2011 L1Calo trigger tower timing in ns as function of eta and phi for the electromagnetic (EM) and hadronic (HAD) calorimeter layer. The timing is derived by fitting the trigger tower ADC distributions using either a Gauss-Landau or Landau-Landau function. This plot shows the results using collision data after applying corrections to the L1Calo timing delays derived from figure 3 or 5, respectively. White bins have no measurement due to lack of statistics. |
![]() png eps |
![]() png eps |
Energy correlation (Beginning of 2011) L1Calo trigger tower transverse energy versus offline transverse energy. The offline transverse energy is derived by summing the individual calorimeter cells associated to a tower. These plots show the results for the electromagnetic (EM) and hadronic (HAD) calorimeter using 2011 collisions data. |
![]() png eps |
![]() png eps |
Energy resolution (2010) Fractional difference between L1Calo transverse energy and offline transverse energy as a function of the offline transverse energy. The L1Calo energy is calculated using two different methods; the energy based on the ADC peak sample and the energy based on the result of the look-up-table (LUT). These plots show the distributions for the electromagnetic barrel (EMB) and the Tile hadronic calorimeter using 2010 collision data. |
![]() png eps |
![]() png eps |
Energy resolution (Early 2011) Fractional difference between L1Calo transverse energy and offline transverse energy as a function of the offline transverse energy. The L1Calo energy is calculated using two different methods; the energy based on the ADC peak sample and the energy based on the result of the look-up-table (LUT). These plots show the distributions for the electromagnetic barrel (EMB) and the Tile hadronic calorimeter using 2011 collision data which include improvements in the LUT calculation. |
![]() png eps |
![]() png eps |
Correlation of FIR output and ADC peak sample The output of the finite-impulse-response (FIR) filter before drop bit truncation as a function of the peak ADC sample for an example trigger tower from the electromagnetic (EM) layer. A linear fit is applied to distribution for peak ADC values above 50, well above the pedestal value of about 32 counts. The fitted gradient is used to derive look-up-table (LUT) slopes which determine the energy calibration. The plot shows the results from the analysis of 2010 collision data. |
![]() png eps |
|
LUT calibration (2010) The fitted gradients as a function of eta and phi for the electromagnetic (EM) and hadronic (Had.) layer. The results are derived from linear fits to the distributions of the finite-impulse-response (FIR) filter output as a function of the peak ADC sample as shown in figure 13. This plot shows the result from the analysis of 2010 collision data. |
![]() png eps |
![]() png eps |
Normalised pulse shape (2010) The pedestal subtracted and normalized ADC pulse shape for an example L1Calo trigger tower as derived from the analysis of 2010 collision data. |
![]() png eps |
|
FIR calibration (2010) The sum (S_1+S_3) as a function of eta and phi for the electromagnetic (EM) and hadronic (Had.) calorimeter layer. The value S i is the normalized pulse height of the i-th ADC sample as illustrated in figure 16. This plot shows the result from the analysis of 2010 collision data. |
![]() png eps |
![]() png eps |
Figure 1: Figure 1(a) is a standard pulse as it is read out of the L1Calo system and Figure 1(b) is a reconstructed pulse with nanosecond time resolution which is derived from special pulser runs as described in the text. Both signals are fit with the hybrid Landau/Gaussian fit function described by Equation 1. The signals were taken from a Liquid Argon calibration run. |
![]() png eps |
![]() png eps |
Figure 2: This is an eta-phi plot of the peak time location with ns precision plotted on the z-axis. The peak times (t0) are measured by fitting each trigger tower signal with a Landau/Gaussian hybrid function. The timing reference was taken as 175 ns from Figure 1(a). The electromagnetic layer is shown in 2(a) with beam-1 approaching in the −h direction using event number 2166. The hadronic layer is shown in 2(b) with beam-2 approaching in the +h direction using event number 2666. Both events are taken from Run 140370. |
![]() png eps |
![]() png eps |
Figure 3: An eta-projection of the peak time location distribution in Figure 2 is shown. The electromagnetic layer is shown in 3(a) with beam-1 approaching in the −eta direction using event number 2166. The hadronic layer is shown in 3(b) with beam-2 approaching in the +eta direction using event number 2666. Both events are taken from Run 140370. |
![]() png eps |
![]() png eps |
Figure 5: The approximate time of flight from collision vertex to detector layer as a function of eta. The electromagnetic layer is shown in 5(a) and the hadronic layer in 5(b). |
![]() png eps |
![]() png eps |
Figure 6: The time of flight from collimator to detector layer as a function of eta is shown using eta = 0 as a reference. The time of flight from the interaction point to the detector, shown in Figure 5, is subtracted to get the total time of flight correction seen in Figure 7. The electromagnetic layer is shown in 6(a). The hadronic layer is shown in 6(b). The beam-1 (−eta) trajectory is used here with reflection across the eta = 0 axis representing the beam-2 trajectory. |
![]() png eps |
![]() png eps |
Figure 7: This is the total time of flight correction as a function of eta. The electromagnetic layer is shown in 7(a). The hadronic layer is shown in 7(b). The beam-1 (−eta) trajectory is used here with reflection across the eta = 0 axis representing the beam-2 trajectory. |
![]() png eps |
![]() png eps |
Figure 8: The corrected peak time location in nanoseconds for both calorimeter layers. The electromagnetic layer is shown in 8(a) with beam-1 (−eta trajectory) using event number 2166. The hadronic layer is shown in 8(b) with beam-2 (+eta trajectory) using event number 2666. Both events taken from Run 140370. |
![]() png eps |
![]() png eps |
Figure 9: The mean of the eta bins in Figure 8. The distributions would be flat for perfect timing, however, partition dependent offsets must be corrected. The electromagnetic layer is shown in 9(a) with beam 1 approaching in the −eta direction using event number 2166. The hadronic layer is shown in 9(b) with beam 2 approaching in the +eta direction using event number 2666. Both events taken from Run 140370. |
![]() png eps |
![]() png eps |
Figure 10: The peak position (t0), determined using Equation 1, as a function of the ATLAS event number. Figure 10(a) shows a non-uniform response to beam-1 and beam-2 events (PPM channel located in hadronic end cap), which was seen in a small number of channels. Figure 10(b) shows a typical TT with the expected uniform response to beam-1 and beam-2 splash events (PPM channel located in electromagnetic barrel). |
![]() png eps |
![]() png eps |
Figure 11: The mean correction to peak time location in nanoseconds for both calorimeter layers. The electromagnetic layer is shown in 11(a). The hadronic layer is shown in 11(b). Data taken from splash events in Run 140370. |
![]() png eps |
![]() png eps |
Figure 12: The final corrected timing delays for each TT in nanoseconds for both calorimeter layers. The electromagnetic layer is shown in 12(a). The hadronic layer is shown in 12(b). |
![]() png eps |
![]() png eps |
Energy ramp for one particular Trigger Tower from the Tile system Five different energies have been pulsed using the Tile charge injection system. The transverse energy measured in the Tile calorimeter is plotted vs. the energy measured by L1Calo. A linear fit is overlayed with the slope of the line corresponding to the calibration constant derived by this method. |
![]() png eps |
Energy ramp for one particular Trigger Tower from the LAr system Six different energies have been pulsed using the LAr pulser system. The transverse energy measured in the LAr calorimeter is plotted vs. the energy measured by L1Calo. A linear fit is overlayed with the slope of the line corresponding to the calibration constant derived by this method. |
![]() png eps |
Tile energy correlation plot Energy correlation plot from 7 TeV data where the transverse energy measured in the Tile Calorimeter is shown vs. the transverse energy measured by L1Calo. The transverse energy of all cells which correspond to a Trigger Tower is summed up and compared to the corresponding L1 Trigger Tower. The plot shows a very good correlation with only a very few outliers. |
![]() png eps |
LAr energy correlation plot Energy correlation plot from 7 TeV data where the transverse energy measured in the LAr Calorimeter is shown vs. the transverse energy measured by L1Calo. The transverse energy of all cells which correspond to a Trigger Tower is summed up and compared to the corresponding L1 Trigger Tower. The plot shows a very good correlation. |
![]() png eps |
Correlation between peak ADC input and LUT output L1Calo digitises trigger tower signals with a scale of 250 MeV per ADC. The digital pulses are passed through a filter to improve energy resolution, noise rejection and bunch crossing identification. The filter output is then passed into a look-up table (LUT) which performs pedestal subtraction, noise cuts, and the final ET calibration. The output from the LUT has a scale of 1 GeV/count. The initial LUT values were generated based on measurements of calibration pulses. The correlation between peak ADC input and LUT output for a single trigger tower is plotted. The correlation between peak ADC input and the LUT output for a single trigger tower in collision data is plotted. The expected gradient for a perfectly calibrated tower is 1/4 (250 MeV / 1 GeV). A straight line was fitted to the data and a gradient of 0.24 LUT/ADC was extracted. This shows that calibration and collision pulse shapes are comparable, and the initial calibration is already close to the final optimum LUT values required for collisions. |
![]() png eps |
LUT calibration L1Calo digitises trigger tower signals with a scale of 250 MeV per ADC. The digital pulses are passed through a filter to improve energy resolution, noise rejection and bunch crossing identification. The filter output is then passed into a look-up table (LUT) which performs pedestal subtraction, noise cuts, and the final ET calibration. The output from the LUT has a scale of 1 GeV/count. The initial LUT values were generated based on measurements of calibration pulses. Straight lines were fitted to ADC peak vs LUT distributions for each trigger tower and the gradients extracted. The expected gradient for a perfectly calibrated tower is 1/4 (250 MeV / 1 GeV). The distribution of fit gradients for towers in the LAr Barrel A-side show that the initial LUT calibration worked well, as most towers are already close to the optimum value. Future calibration will be based on offline reconstructed physics objects. |
![]() png eps |