# Introduction

public plots from the Tilecal noise activity

# Tilecal noise public plots

## Noise and LVPS performance ApprovedPlotsTileNoise

### Noise characteristics (2014)

Electronic noise (defined as the RMS of the pedestal) as a function of the PMT number, after the LS1 maintenance campaign. The different partitions are shown separately, EBA, LBA, LBC and EBC. Each point represents the average of the noise RMS of the 64 modules in the partition, for High Gain and Low Gain separately. The plot includes PMTs connected to all TileCal cells, gap/crack scintillators and MBTS. The PMTs closer to the patch panel (with the higher PMT number) tend to have larger noise (especially PMT48 in HG). This feature was present also in the previous version of the LVPS used in Run1, and it's largely reduced with the newest LVPSs.
Contact: valerio.rossetti@cern.ch
Reference: https://cdsweb.cern.ch/record/1757209

eps pdf png

Electronic noise (defined as the RMS of the pedestal) as a function of drawer number, after the LS1 maintenance campaign. The different partitions are shown separately, EBA, LBA, LBC and EBC. Each point represents the average of the noise RMS of all the PMTs in a module, for High Gain (HG) and Low Gain (LG) separately. The plot includes PMTs connected to all TileCal cells, gap/crack scintillators and MBTS. LBC16 has all HG channels of digitizers 4 and 5 with larger noise (50% higher with respect to the others), with no additional features.
Contact: valerio.rossetti@cern.ch
Reference: https://cdsweb.cern.ch/record/1757209

eps pdf png

\phi-averaged electronic cell noise (defined as the RMS of the pedestal) as a function of \eta of the cell, with both readout channels in High Gain (HG). Values are measured after the LS1 maintenance campaign. For each cell, the average value over all drawers is taken. The different cell types are shown separately, A, BC, D, and E (gap/crack). The transition between the long and extended barrels can be seen in the range 0.7 < |\eta| < 1.0 . HGHG combination is relevant when the energy deposition in the cell is \lesssim 15 GeV. The D cells are slightly lower because of the calibration constants. E cells have typically smaller noise RMS because they are associated to only one readout channel, instead of two. The cell with the largest noise is A10, which corresponds to PMT47 and 48.
Contact: valerio.rossetti@cern.ch
Reference: https://cdsweb.cern.ch/record/1757209

eps pdf png

\eta-averaged electronic cell noise (defined as the RMS of the pedestal) as a function of \phi of the cell, with both readout channels in High Gain (HG). Values are measured after the LS1 maintenance campaign. For each cell, the average value over all cells in a drawer is taken. The different cell types are shown separately, A, BC, D. HGHG combination is relevant when the energy deposition in the cell is \lesssim 15 GeV.
Contact: valerio.rossetti@cern.ch
Reference: https://cdsweb.cern.ch/record/1757209

eps pdf png

Electronic noise (defined as the RMS of the pedestal) in ADC counts as a function of PMT number. The two periods in comparison are October 2011, when an older version of the LVPSs was present in the drawers, and September 2014, after the LS1 maintenance campaign and installation of newer LVPSs. Each point represents the average over 256 modules (all TileCal modules), for High Gain and Low Gain separately. The plot includes PMTs connected to all TileCal cells, gap/crack scintillators and MBTS. With the newest version of the LVPSs, a significant reduction of the electronic noise is observed.
Contact: valerio.rossetti@cern.ch
Reference: https://cdsweb.cern.ch/record/1757209

eps pdf png

Electronic noise (defined as the RMS of the pedestal) in MeV as a function of cell \eta. The two periods in comparison are October 2011, when an older version of the LVPSs was present in the drawers, and September 2014, after the LS1 maintenance campaign and installation of newer LVPSs. Only cells of the layer A are shown. Each point represents the average over \phi, for HighGain-HighGain only. With the newest version of the LVPSs, a significant reduction of the electronic noise is observed.
Contact: valerio.rossetti@cern.ch
Reference: https://cdsweb.cern.ch/record/1757209

eps pdf png

### Noise characteristics (2012)

This plot is a comparison the RMS of the electronic cell noise reconstructed from the pedestal run 196309 with the cell noise that was used for MC12. The MC noise was produced using the cell noise 2-Gaussian parameters to define 2-Gaussian digital noise for each readout channel. This 2-Gaussian digital noise was then used to produce digits for a full simulation of a pedestal run. The output of this run has then been used to produce cell noise constants for MC. The plot shows good agreement (<1% difference) between the noise for data and MC. For more information on the process of making cell noise constants for MC see the internal note https://cdsweb.cern.ch/record/1317318
Contact: olof.lundberg@cern.ch
Reference: https://cdsweb.cern.ch/record/1453801

eps version of the figure
Values have been extracted from high statistics (100k) pedestal run 195843 (2012) which have been fitted using the non-iterative optimal filtering method. The plots show the eta-dependence of the phi-averaged RMS of the noise in these runs in the Long Barrel. Each point is an average for a given cell over all modules containing this certain cell type. The cells are divided into the three different sample layers. This is done for the High Gain - High Gain combination of the readout channels. In this plot only the 39 modules in Long Barrel where the LVPS was changed from version 6.5.4 to version 7.5 in winter 2011/2012. One cell has been excluded, this is the A7 cell from LBA62 for which there were known problems in this run.
Contact: olof.lundberg@cern.ch
Reference: https://cdsweb.cern.ch/record/1442284

eps version of the figure
A comparison between the reconstructed energy in all 100k events in two high statistics pedestal runs, for module LBC41, channel 47, a previously very noisy channel. Reconstruction of energy in the events is performed using the Non-iterative Optimal Filtering method. LBC41 had its LVPS changed from version 6.5.4 to version 7.5 in winter 2011-2012. The RMS of the distribution goes down by almost a factor 2 in the channel after changing LVPS. Values taken from ped run 192130 (2011) and run 195843 (2012).
Contact: olof.lundberg@cern.ch
Reference: https://cdsweb.cern.ch/record/1442284

A comparison between cell noise rms in in 2011 and 2012 in all cells in the 40 modules that had their LVPS changed from version 6.5.4 to version 7.5 during winter 2011-2012. Values taken from ped run 192130 (2011) and run 195843 (2012).
Contact: olof.lundberg@cern.ch
Reference: https://cdsweb.cern.ch/record/1414789

eps version of the figure
Electronic noise of Tilecal channels is measured in pedestal calibration runs as the RMS of the amplitude distribution and is compared with the sigma of the Gaussian fit of the the amplitude distribution. The LVPSs of 40 modules in the ATLAS cavern have been replaced with a new generation of LVPS. The plots show the average ratio RMS/sigma per channel for the 40 modules before and after the replacement of the current production LVPS (v6.5.X) with the new production LVPS (v7.5). For Gaussian noise the ratio RMS/s should be close to 1.
Contact: olof.lundberg@cern.ch
Reference: https://cdsweb.cern.ch/record/1414789

eps version of the figure

### Noise characteristics (2011)

Electronics noise of Tilecal channels is measured in pedestal calibration runs as the RMS of the amplitude distribution and is compared with the sigma of the Gaussian fit of the the amplitude distribution. The LVPSs of 5 modules in the ATLAS cavern have been replaced with a new generation of LVPS. The plots show the ratio RMS/sigma for the 5 modules before and after the replacement of the current production LVPS (v6.5.X) with the new production LVPS (v7.3.1). For Gaussian noise the ratio RMS/s should be close to 1. The ratio differers significantly from 1 in several channels for the current production of LVPS. The typical RMS value of the channels is 25 MeV for the current LVPS and 20 MeV for the new production of LVPS.
Contact: Luca.Fiorini@cern.ch

eps version of the figure

eps version of the figure eps version of the figure

eps version of the figure eps version of the figure
Values have been extracted using high statistics pedestal run 192130 (November 9th 2011), which have been fitted using the non-iterative optimal filtering method. The plots show the eta-dependence of the phi-averaged RMS of the noise in these runs. Each point is an average for a given cell over all modules containing this certain cell type. This is done for a given gain combination of the readout channels – High Gain-High Gain (HGHG), Low Gain-Low Gain (LGLG) or HGLG. This third is an average of when channel 1 is read out in HG and channel 2 in LG and vice versa. Plotted is only cells in modules with Low Voltage Power Supplies of version 6.5.4 (all modules except 5 at this time).
Contact: olof.lundberg@cern.ch
Reference: https://cdsweb.cern.ch/record/1414789

### Noise characteristics (2009)

190 runs of randomly triggered events were used to evaluate the stability in time of the Tile Calorimeter ADC noise. The ADC Noise (RMS of the digital samples) for the Tile Calorimeter is shown as a function of time from the start of ATLAS continuous running in Aug 2008 to the opening of the detector in November 2008 and then from the closure of the detector in June 2009 till November 2009. The variation in time of the ADC noise for an individual channel is also shown. The relative variation of the detector average noise is within the +-1% green bars. The detector average value is of 1.44 ADC counts and it is compatible with both 2008 and 2009 averages. The RMS/mean is of 0.32% for the detector average and 1.1% for an individual channel.
Contact: Luca.Fiorini@cern.ch
Reference: http://indico.cern.ch/conferenceDisplay.py?confId=73518

Random triggers of a physics run taken in September (run 127453) have been used to study the accuracy of the noise description in the Tile Calorimeter. The energy deposited in the Tile cells divided by the noise constant in the DataBase is supposed to correspond to the residual probability of being compatible with noise energy deposits in units of gaussian sigmas (the value Ecell/DB constant=1 contains 68% of the noise ampitudes, etc.). The black points correspond to Randomized data: the cells energy values are substituted with random values from a gaussian distribution of mean=0 and sigma=DB noise contant of the cell. The distribution follows a gaussian of mean=0 and sigma=1, as expected. The red points are the measured Ecell/DB constant for the Tile simple gaussian description. The tails of the distribution don't follow the expected gaussian distribution. The blue points are the measured Ecell/DB constant for the Tile double gaussian noise description. Such description restore the expected behavior of the noise significance for the Tile cells. The black line is the result of a gaussian fit of the double gaussian significance distribution.
Contact: Luca.Fiorini@cern.ch
Reference: http://indico.cern.ch/conferenceDisplay.py?confId=61346

### LVPS trips 2012

This is a plot of the delivered luminosity vs the number of LVPS trips. The red points are data and the blue line is the linear regression. The tips/pb^-1 is 0.6 approximately. The data taking period used is between March 13th and December 16th 2012.By S. Norberg, J. Proudfoot, S. Chekanov.
Contact: norberg@ou.edu

[png]
The graph shows the number of LB modules that have tripped a specific number of times. This is for the period between March 13th 2012 and December 16th 2012. The red area our the version 6 power supplies. The green are the v7.5, the newest modules to be put in; one of them has tripped this year. The blue are the v7.3 or the power supplies put in in 2011, three v7.3 power supplies have tripped twice this year, and one has tripped once this year. The green or v7.5 have 38 modules that have not tripped this year, and one module that has tripped in the LB region. By S. Norberg, J. Proudfoot, S. Chekanov.
Contact: norberg@ou.edu

[png]
The graph shows the relationship between trip rate and peak luminosity. The red points are the total number of trips divided by the integrated luminosity. The red line going through is just the average total trip rate divided by integrated luminosity for 2011. The line is there to show that there is very little deviation from a central value showing that there is no correlation between peak luminosity and number of trips. The data taken period used is between March 13th and October 30th 2011. The blue points are the total number of trips divided by the integrated luminosity for March 13th till December 16th 2012. The blue line going through is just the average total trip rate divided by integrated luminosity for the period of March 13th till December 16th 2012. The trips/pb-1 has gone down because less modules have tripped in 2012 then 2011, because in 2012 40 new fLVPS were installed. By S. Norberg, J. Proudfoot, S. Chekanov.
Contact: norberg@ou.edu

[png]

### LVPS trips 2011

This is a plot of the delivered luminosity vs the number of LVPS trips. The red points are data and the blue line is the linear regression with a fit. The tips/pb^-1 is 0.8 approximately. The data taking period used is between March 13th and October 30th 2011.By S. Norberg, J. Proudfoot, S. Chekanov.
Contact: norberg@ou.edu

[png]
The graph shows the number of LB modules that have tripped a specific number of times. This is for the period between March 13th 2011 and October 30th 2011. The red shows the new power supplies which consist of LBA47, LBC33, LBC36, LBC40. Only one new module has tripped so far, all of the other modules have tripped at least 35 times. By S. Norberg, J. Proudfoot, S. Chekanov.
Contact: norberg@ou.edu

[png]
The graph is showing the relationship between trip rate and peak luminosity. The red points are the total number of trips divided by the integrated luminosity. The line going through is just the average total trip rate divided by integrated luminosity. The line is there to show that there is very little deviation from a central value showing that there is no correlation between peak luminosity and number of trips. The data taken period used is between March 13th and October 30th 2011. By S. Norberg, J. Proudfoot, S. Chekanov.
Contact: norberg@ou.edu

[png]

### Correlated Noise

The correlation matrices are presented in high-gain for the Atlas.Tilecal module LBA9 before (top) and after (below) applying the correlated noise unfolding. 10,000 events were used from the run 125204, a standalone bi-gain pedestal run taken with the standard final front-end Tile electronics, final finger LVPS, on 16th August 2009 during cosmics data taking. Regions of high and low correlation values are visible reflecting the configuration of the Atlas.Tilecal hardware with clear clusters of neighbour channels determining the PMT signal responses, this behavior in magnitude and shape is typical for Atlas.Tilecal drawers. The correlated noise component is significantly reduced after applying the method.
Reference: ATL-COM-TILECAL-2010-023

[eps]
[eps]
Reconstructed energy from channel 19 (top) and channel 47 (below) of the Atlas.Tilecal LBA23 module are shown in ADC counts before (red dots) and after (blue line) applying the unfolding. 10,000 events were used from the run 125204, a standalone bi-gain pedestal run taken with the standard final front-end Tile electronics, final finger LVPS, on 16th August 2009 during cosmics data taking. Channel 19 is an example of a non-correlated channel and the signal remains uncorrelated after applying the method. Channel 47 is an example of a highly correlated channel and the tails are significantly reduced after applying the method.
Reference: ATL-TILECAL-INT-2010-004

[eps]
[eps]
Reconstructed energy from channel 46 is plotted against the one from channel 47 before (top) and after (below) unfolding the correlated noise component with the method. 10,000 events were from the run 125204, a standalone bi-gain pedestal run taken with the standard final front-end Tile electronics, final finger LVPS, on 16th August 2009 during cosmics data taking. A clear improvement is observed i.e., the correlation between both readout channels are very much reduced after applying the unfolding.
Reference: ATL-TILECAL-INT-2010-004

[eps]
[eps]
Correlation matrices calculated for Sample 3 are shown before (top) and after (below) applying the correlated noise unfolding, using calibration run 146627 where a charge of 100 pC is injected, per event, by the Charge Injection System in each read-out channel of the drawer. A total of 1000 events per channel were analyzed when channel 3 was being fired. The information from Sample 0 (only sensitive to pedestal noise) is successfully used to unfold the noise correlations in the presence of physics signals. The correlated noise component is significantly reduced for Sample 3. The unfolding was applied to all Samples with similar results to validate the performance of the unfolding method in the presence of a controlled injected signal.
Reference: ATL-COM-TILECAL-2010-023

[eps]
[eps]
These two plots show the sample correlation coefficient between channel amplitudes in an individual superdrawer for two different pedestal runs. The first run was taken in December 2010 when the module was powered by a standard power supply. Areas of high correlated noise are noticeable especially in the lower half of the drawer. During the winter 2010 maintenance period, the power supply was replaced with a new production prototype without any other changes made to the drawer. The result is that the correlated noise that was previously observed has been drastically reduced. The average coefficient for instrumented channels in the lower half of the drawer was 0.22 with the old power supply and 0.06 with the new model which is nearly a factor of 4 in reduction. The other four superdrawers that received new power supplies showed similar improvements.

Contact: Rob Calkins

[eps]
[eps]
Reconstructed energy from channel 47 of the Tilecal LBA47 module is shown in MeV before (red dots) and after (blue line) the replacement of the low voltage power suply (LVPS) in this module. In 5 modules of the Tilecal, the current LVPSs (v.6.5.x) were replaced with a new generation (v.7.3.1). Data was used from standalone high gain (HG) pedestal runs 170713 and 171950, respectively before and after the LVPS change. For non-correlated channels, the RMS is smaller and the distribution shows smaller tails. Channel 47 is an example of a highly correlated channel, where the tails are significantly reduced with the new LVPS, having a RMS value of 25.687 MeV before the LVPS change and 17.897 MeV for the new production of LVPS.
Reference: ATL-COM-TILECAL-2011-022

[eps]
The correlation matrices between two channels are presented in high-gain for the Tilecal module LBA47 before (top) and after (below) the replacement of LVPSs with a new generation. The LVPSs (v.6.5.x) were replaced with a new generation (v.7.3.1) in 5 modules of the Tilecal. The pattern of high and low correlation values, typical of Tilecal drawers, reflects the configuration of Tilecal hardware, with clear clusters of neighbour channels determining the PMT responses. The correlated noise component is significantly reduced with the new generation of LVPSs, with a mean correlation of 0.044 before the replacement and 0.031 after.
Reference: ATL-COM-TILECAL-2011-022

[eps]
[eps]

## Pile-up Noise Performance

### Pile-up Noise (2012)

The noise distribution in different Tilecal cells is represented as a function of || of the cells for zero bias run 208781 of 2012 at a centre-of-mass energy of 8 TeV with a bunch spacing of 50 ns and an average number of interactions < > = 15.7 per bunch crossing). The Monte Carlo was reweighted to the average number of interactions in data, according to the weight given by the PileUpReweighting tool. The noise was estimated as the standard deviation of the measured cell energy distribution. The top left (right) histogram shows the results obtained for the cells of Layer A (Layer BC). The bottom left (right) histogram shows the results obtained for the cells of Layer D (Special Cells).

Reference: ATL-COM-TILECAL-2011-022

[pdf] [eps]
The noise distribution in different Tilecal cells is represented as a function of || of the cells for zero bias run 216416 of 2012 at a centre-of-mass energy of 8 TeV with a bunch spacing of 25 ns and an average number of interactions < > = 10.0 per bunch crossing). The Monte Carlo was reweighted to the average number of interactions in data, according to the weight given by the PileUpReweighting tool. The noise was estimated as the standard deviation of the measured cell energy distribution. The top left (right) histogram shows the results obtained for the cells of Layer A (Layer BC). The bottom left (right) histogram shows the results obtained for the cells of Layer D (Special Cells).

Reference: ATL-COM-TILECAL-2011-022

[pdf] [eps]

### Pile-up Noise (2011)

The noise distribution in different Tilecal cells is represented as a function of || of the cells for zero bias run 182424 of period F2 of 2011 data (at a centre-of-mass energy of 7 TeV with a bunch spacing of 50 ns and an average number of interactions < > = 4.8 per bunch crossing). The Monte Carlo was reweighted to the average number of interactions in data, according to the weight given by the PileUpReweighting tool. The noise was estimated as the standard deviation of the measured cell energy distribution. The top left (right) histogram shows the results obtained for the cells of Layer A (Layer BC). The bottom left (right) histogram shows the results obtained for the cells of Layer D (Special Cells).

Reference: ATL-COM-TILECAL-2011-022

[pdf] [eps]
The noise distribution in different Tilecal cells is represented as a function of || of the cells for zero bias run 190617 of period M2 of 2011 data (at a centre-of-mass energy of 7 TeV with a bunch spacing of 50 ns and an average number of interactions < > = 11.3 per bunch crossing). The Monte Carlo was reweighted to the average number of interactions in data, according to the weight given by the PileUpReweighting tool. The noise was estimated as the standard deviation of the measured cell energy distribution. The top left (right) histogram shows the results obtained for the cells of Layer A (Layer BC). The bottom left (right) histogram shows the results obtained for the cells of Layer D (Special Cells).

Reference: ATL-COM-TILECAL-2011-022

[pdf] [eps]
The Tilecal noise dependence with the actual number of interactions, , is represented, for several zero bias data runs from different periods of 2011. The noise was estimated as the standard deviation of the mean energy value per cell. The black markers represent data and the red points Monte Carlo. The different marker styles correspond to different cell layers in the Tilecal.

Reference: ATL-COM-TILECAL-2011-022

[pdf] [eps]