DT PERFORMANCE AT THE END OF RUN 2

The performance of the CMS DT muon detector was evaluated at the end of the 2018 data taking, in terms of:

Hit Efficiency (dead channels, efficiency by chamber, grand summary), Segment Reconstruction Efficiency (efficiency by chamber, φ-𝛈 maps, grand summary), Trigger Primitive Efficiency (efficiency by chamber, φ-𝛈 maps), Hit Spatial Resolution, DT Time Resolution.
Hit Efficiency trend is also shown vs Integrated Luminosity.
Segment Efficiency and Trigger Primitive Efficiency trends are also shown vs Instantaneous Luminosity, time (run number) and muon transverse momentum (pT).

See also: CMS DP-2019/008

FRACTION OF ALIVE CHANNELS

A dead channel is defined as a channel with zero occupancy (within the considered dataset).

Dead channels are discarded in the analysis and don’t contribute to the computed efficiency.

On the contrary, channels with LOW occupancy (as are e.g. those with malfunctioning HV) are considered inefficient and affect the computed average efficiency of the chamber they belong to.

The overall fraction of working channels, according to the definition given above, is 99.7%. This fraction is somewhat larger than the figure resulting from online computation (99.5%), where channels with HV problems and quasi null efficiency are also considered dead.

As shown by the comparison of the following two plots, read-out problems were significantly reduced in 2018, thanks to a partial upgrade of the DT readout electronics, happened during the 2017/2018 winter shut-down and including the replacement of Read Out Server (ROS) boards with new “µROS”, less prone to data integrity errors.

SingleChamberAlivePerStationZMu2017F.png SingleChamberAlivePerStationRun2018D.png

Fraction of working wires, chamber by chamber, in 2017 (left) and in 2018 (right).
pdf version of the 2017 plot pdf version of the 2018 plot

DT HIT EFFICIENCY

The hit efficiency was computed as the ratio between detected and expected hits.

Expected hits were defined using the intersection of good quality reconstructed segments with the layer under study: the layer was considered efficient if a hit was found within the cell identified by the intersection.

Segments crossing known dead channels (see previous plot) were discarded in the analysis.

Hit Efficiency chamber by chamber

w.r.t to early 2018 results:

  • The problem in Wheel 2 Sector 7 MB2, which had the full Θ super-layer inefficient, was still observed.

  • A new HV problem appeared in YB+1/Sec/1/MB1 SL2, Layer3, ch 1-32

SingleChamberEfficiencyPerStationRun2018D.png

Hit Efficiency, chamber by chamber.
pdf version of this plot

Hit Efficiency: Grand Summary

Distribution of the efficiencies shown in previous plot: 1 entry per chamber. It is compatible with last year result.

ChamberEfficiencyRun2018D_Log.png

Distribution of chamber hit efficiency (1 entry per chamber).
pdf version of this plot

DT Hit Efficiency trends with Integrated Luminosity

Following plots show the trend of DT Hit Efficiency in time: they make visible any variations, either due to detector ageing or to changes in the detector configuration.
In order to protect the chambers most exposed to radiation from the risk of early ageing, High Voltages were lowered at the beginning of 2017 and again at the beginning of 2018 data taking, as shown in this sketch:


configurazioneHV.png


Sketch of the DT High Voltage configuration in 2017 and 2018: HV was lowered progressively in the regions with higher background.

Lower HV is expected to cause a slight loss of hit efficiency, which actually was observed last year.
Front End thresholds were also lowered in 2018 in order to balance the efficiency loss caused by lower HV.

MB1 chambers of external wheels are those most exposed to radiation, hence most at risk of early ageing.
In 2016 all chambers were operated at the nominal value of HV=3600 V: a very mild trend of efficiency loss (< 0.5% for 35 fb-1 collected) became visible in the external wheels.
In 2017 the HV of those chambers was lowered to 3550 V, causing a modest step of ~0.5 %, while the trend became flat (no further loss of efficiency was caused by accumulated radiation)
In 2018 the MB1 chambers of external wheels had their HV further reduced to 3500 V, while the 3 internal wheels were brought down to 3550 V. At the same time the FE thresholds were reduced from 30 to 20 mV in order to recover efficiency: in fact the overall reduction od efficiency was small (< 0.5%) and the trend stayed flat throughout the year.

In the following plots all the described transitions are clearly visible.

The origin of Integrated Luminosity is the start of LHC operation in 2011.

MB1PhiEffVsLInt_elaborata.png

Trend of Hit Efficiency in the φ Layers of MB1 chambers, wheel by wheel, as a function of LHC Integrated Luminosity.
(Corresponding time periods are also shown as reference)
pdf version of this plot

In every stations, Θ Layers of Wheel 0 show lower efficiency than other wheels: this is due to the loss of hits caused by the tube walls that act as a dead volume. This effect is larger for tracks impinging orthogonally to the wire layers, as shown in the following plot.

MB1TheEffVsLInt.png

Trend of Hit Efficiency in the Theta Layers of MB1 chambers, wheel by wheel, as a function of LHC Integrated Luminosity.
(Corresponding time periods are also shown as reference)
pdf version of this plot

Lowering HV to 3550 V at the beginning of 2017, in MB1 chambers of external wheels, caused a step of ~1 % in the efficiency of Θ Layers.
The further reduction to 3500 V in 2018 for MB1 chambers of external wheels, and to 3550 V for the three internal wheels, together with the lowering of FE thresholds from 30 to 20 mV, caused a small overall reduction of efficiency (< 0.5%) and the trend stayed flat throughout the year.
In the second half of 2018, the new problem, appeared in the MB1 chamber of Wheel+1 Sector 1, and already discussed above, caused a visible loss of 1.4% in the average efficiency of Θ layers of MB1 chambers in Wheel+1.

MB4 chambers of top Sectors are exposed to the background produced by interacting thermal neutrons which result from beam collisions and fill the cavern like a gas. Therefore these chambers are also at risk of early ageing.
In 2016 all chambers were operated at the nominal value of HV=3600 V: even though no trend of efficiency was observed so far, several tests were performed to evaluate the effects of HV reduction: the MB4 chamber in Sector 4 of Wheel-2 was the first one to be brought to 3550 V, at the end of 2016: this resulted in ~1% of efficiency loss.
In 2017 the HV was lowered to 3550 V also in the Top MB4 chambers of the other 4 wheels: the largest impact on efficiency was observed in Wheel 0 (~3% loss) where the muon tracks travel the shortest path within the drift tubes and ionise least charge. To recover efficiency in this chamber the FE threshold was lowered from 30 to 20 mV in the Fall: this change fully compensated the change of HV.
Finally in 2018 all the FE thresholds were lowered to 20 mV and the efficiency went up approximately to the initial levels. The trends were flat till the end of 2017. 2018 data show a possible loss of ~0.2 - 0.3 % in 70 fb-1 collected.

All these transitions are clearly visible in the following plot:

MB4TopPhiEffVsLint.png

Trend of Hit Efficiency in the φ Layers of MB4 chambers of top sectors, wheel by wheel, as a function of LHC Integrated Luminosity.
(Corresponding time periods are also shown as reference)
pdf version of this plot

MB2 and MB3 chambers receive much less background than the MB1 ones.
Nonetheless the MB2 chambers of Wheels ± 1 and ± 2 and the MB3 chambers of Wheels ± 2 were brought to 3550 V in 2018, while the FE thresholds were reduced to 20 mV.
The effects of these changes are visible in following the plots, but they are limited to < 0.5%.
The trends are flat everywhere.

MB2PhiEffVsLInt.png

Trend of Hit Efficiency in the φ Layers of MB2 chambers, wheel by wheel, as a function of LHC Integrated Luminosity.
(Corresponding time periods are also shown as reference)
pdf version of this plot

MB3PhiEffVsLInt.png

Trend of Hit Efficiency in the φ Layers of MB3 chambers, wheel by wheel, as a function of LHC Integrated Luminosity.
(Corresponding time periods are also shown as reference)
pdf version of this plot

DT SEGMENT RECONSTRUCTION EFFICIENCY

The segment efficiency is computed as the ratio between reconstructed and expected segments.
Expected segments are defined using a Tag&Probe method: events are selected containing 2 reconstructed muons, compatible with the Z mass.
The probe muon is only required to have ⦥ 1 associated segment in a muon chamber different from the one under study.
Only the information of the tracker is used to extrapolate the muon tracks to the DT chambers.
A chamber crossed by the probe muon track extrapolation is considered efficient if a segment is found within 15 cm of the expected position on the RΦ plane.

Segment Reconstruction Efficiency chamber by chamber

While variations of the DT hit efficiency of the order of few % don’t affect the segment reconstruction efficiency (thanks to the detector redundancy), groups of channels that fail to be read out have a visible effect, as shown by the comparison of the two plots below: in 2018 the implementation of µROS solved many readout problems that were present in 2017.

fig14.png eff_sec_vs_wh_no_zoom_all_two.png

Segment Reconstruction Efficiency, chamber by chamber, in 2017 (left) and in 2018 (right)
pdf version of the 2018 plot

Segment Reconstruction Efficiency: Grand Summary

Distribution of the efficiencies shown in previous plot: 1 entry per chamber.
In 2017 14 chambers had < 90% efficiency, while in 2018 just 1.

eff_chamb_all_log.png

Distribution of Chamber Segment Reconstruction Efficiency (1 entry per chamber).
pdf version of this plot

DT Segment Efficiency maps

The direction of the probe muon can be used to compute Segment Efficiency as a function of φ and 𝛈. The following four plots show it separately for each DT station.
Clearly visible is the effect of gaps between wheels and sectors (different for MB4, due to a more complex geometry that includes overlaps of top sectors), and also visible are the dead regions corresponding to “chimney chambers” in Wheel-1 Sector3 and Wheel+1 Sector4 (these chambers are by design shorter than standard in the z direction, in order to leave the necessary room for services).

eff_eta_vs_phi_MB1.png eff_eta_vs_phi_MB2.png
eff_eta_vs_phi_MB3.png eff_eta_vs_phi_MB4.png

Segment Reconstruction Efficiency as a function of Phi and Eta of the probe muon, station by station.
pdf version of the MB1 plot, pdf version of the MB2 plot, pdf version of the MB3 plot, pdf version of the MB4 plot

DT Segment Efficiency trends

DT segment efficiency is constant in time and does not depend on Instantaneous Luminosity within the range of available data, as shown by the next two plots.

Seg_vs_run_elaborati.png

Segment Reconstruction Efficiency as a function of the run number (i..e. of time), station by station.
pdf version of this plot

eff_vs_inst_lumi.png

Segment Reconstruction Efficiency as a function of LHC Instantaneous Luminsity, station by station.
pdf version of this plot

The next plot shows the segment efficiency as a function of transverse momentum of the probe muon.
It is ≳ 99% up to pT ≲ 100 GeV and keeps ≈ 98% up to 500 GeV (the mild trend being likely due to showering)

eff_vs_pt.png

Segment Reconstruction Efficiency as a function of transverse momentum, station by station.
pdf version of this plot

The average values are 1-2% larger than in 2017 thanks to the recovery of readout problems in 13 chambers, obtained with the implementation of µROS.

TRIGGER PRIMITIVE EFFICIENCY

The Trigger Primitive efficiency is computed as the ratio between detected and expected trigger primitives.
Expected trigger primitives are defined using the same Tag&Probe method applied to compute the segment reconstruction efficiency: events are selected containing 2 reconstructed muons, compatible with the Z mass.
The probe muon is only required to have ⦥ 1 associated segment in a muon chamber different from the one under study.
Only the information of the tracker is used to extrapolate the muon tracks to the DT chambers.
A chamber crossed by the probe muon track extrapolation is also required to have a reconstructed segment matched to the track and reconstructed in both Θ and Φ views.
The chamber is then considered efficient if a trigger primitive is found in the chamber at the correct BX.
Trigger primitives are read-out at the exit of TwinMux, so they also exploit the RPC information.

The main difference w.r.t. 2017 results is the addition of “RPC-only” primitives in the MB1 and MB3 stations, that have two layers of RPC chambers: this addition produced a ~3% efficiency increase in these stations.

Trigger Primitive Efficiency chamber by chamber

The Trigger Primitive efficiency could not be computed in Wheel 2 Sector 7 MB2, due to lack of information in the Θ view.
The red box in MB3 Wheel+1 Sector4 is caused by a faulty trigger board.

DTLTbyChamber.png

Trigger Primitive Efficiency chamber by chamber.
pdf version of this plot

Trigger Primitive Efficiency map

The direction of the probe muon can be used to compute Trigger Primitive efficiency as a function of φ and 𝛈. This is shown below for the four stations separately.
Clearly visible is the effect of not working Θ super-layer in Wheel 2 Sector 7 MB2 and those of the dead regions corresponding to “chimney chambers” in Wheel-1 Sector3 and Wheel+1 Sector4 (these chambers are by design shorter than standard in the z direction, in order to leave the necessary room for services).

DTLT_map.png

Trigger Primitive Efficiency as a function of the muon direction.
pdf version of this plot

Trigger Primitive Efficiency trends

The Trigger Primitive efficiency is constant in time and does not depend on Instantaneous Luminosity within the range of available data.
Clearly visible is the difference between internal (MB1, MB2) and external (MB3, MB4) stations: this is due to the contribution of RPC-only primitives in the internal stations. In these stations the trend of efficiency with transverse momentum is also milder.

DTLT_vs_run_elaborati.png

Trigger Primitive Efficiency as a function of the run number (i.e. time), station by station.
pdf version of this plot

DTLT_vs_lumi.png

Trigger Primitive Efficiency as a function of LHC Instantaneous Luminosity, station by station.
pdf version of this plot

DTLT_vs_pt.png

Trigger Primitive Efficiency as a function of transverse momentum, station by station.
pdf version of this plot

DT HIT SPATIAL RESOLUTION

The spatial resolution is obtained from the widths of residual distributions, i.e. of the distances between each hit and the segment it belongs to.
A comparison to previous public results show the effects of lower High Voltages:

  • In MB1, MB2 and MB3 the HVs were lowered more in the external than in the internal wheels: this brought the resolution of Φ layers in the external wheels to the same level of the central one, thus compensating the effect of the longer path traveled by muons within the tubes.

  • In MB4 the HV was lowered by the same amount in all wheels so the result was a slight worsening of resolution everywhere, preserving the same trend as a function of 𝛈 (wheel).

  • The resolution of Θ layers of external wheels also slightly worsened as a result of lowering the HV. In particular Wheel-2 MB1 and Wheel+2 MB1 now slightly exceed 1mm. However the resolution in these super-layers was poor also with nominal HV, due to track inclination (high eta) and to non uniform magnetic field.

Still no impact is expected on the muon reconstruction, even at high pT.

HitReso_phiTheta-2016-2018-2.png

Comparison between DT Hit Resolution in 2016 and 2018. The resolution is shown ring by ring. for Phi and Theta layers separately.
pdf version of this plot

DT TIME RESOLUTION

The time resolution is obtained from the time distribution of muons reconstructed in the barrel region (only DT time information is used).

Thanks to a very accurate calibration, it did not worsen as a result of lower HV (it was 1.57 ns in 2017)

When no matching with the tracker is required, the beam structure is evident in the secondary peaks.

dttime_2018D_loose.png dttime_2018D.png

DT time distribution of muons reconstructed in the barrel: all muons (left) and only muon tracks matched with tracker information (right).

pdf version of the left plot pdf version of the right plot

-- FrancescaCavallo - 2019-04-23

Edit | Attach | Watch | Print version | History: r3 < r2 < r1 | Backlinks | Raw View | WYSIWYG | More topic actions
Topic revision: r3 - 2019-10-21 - FrancescaCavallo
 
    • Cern Search Icon Cern Search
    • TWiki Search Icon TWiki Search
    • Google Search Icon Google Search

    CMSPublic All webs login

This site is powered by the TWiki collaboration platform Powered by PerlCopyright &© 2008-2021 by the contributing authors. All material on this collaboration platform is the property of the contributing authors.
or Ideas, requests, problems regarding TWiki? use Discourse or Send feedback