roberto chierici - cern

31
Roberto Chierici Roberto Chierici - CERN - CERN Preliminary results from Preliminary results from test beam data test beam data Aim and experimental setup Aim and experimental setup Event reconstruction Event reconstruction Pedestal, common mode and noise evaluation Pedestal, common mode and noise evaluation Cluster finding Cluster finding Module performance Module performance S/N, stability, cluster characteristics S/N, stability, cluster characteristics Latency scans Latency scans Peak and deconvolution modes: features and Peak and deconvolution modes: features and observations observations Efficiencies and delay curves Efficiencies and delay curves Conclusions (still preliminary) Conclusions (still preliminary) On behalf of the CMS tracker collaboration On behalf of the CMS tracker collaboration

Upload: uri

Post on 19-Jan-2016

46 views

Category:

Documents


0 download

DESCRIPTION

Roberto Chierici - CERN. Preliminary results from test beam data. On behalf of the CMS tracker collaboration. Aim and experimental setup Event reconstruction Pedestal, common mode and noise evaluation Cluster finding Module performance S/N, stability, cluster characteristics - PowerPoint PPT Presentation

TRANSCRIPT

Page 1: Roberto Chierici - CERN

Roberto ChiericiRoberto Chierici - CERN- CERN

Preliminary results from Preliminary results from test beam datatest beam data

Aim and experimental setup Aim and experimental setup Event reconstructionEvent reconstruction

Pedestal, common mode and noise evaluationPedestal, common mode and noise evaluationCluster findingCluster finding

Module performanceModule performanceS/N, stability, cluster characteristicsS/N, stability, cluster characteristics

Latency scansLatency scansPeak and deconvolution modes: features and Peak and deconvolution modes: features and

observations observations Efficiencies and delay curves Efficiencies and delay curves

Conclusions (still preliminary)Conclusions (still preliminary)

On behalf of the CMS tracker collaborationOn behalf of the CMS tracker collaboration

Page 2: Roberto Chierici - CERN

Roberto Chierici 2

Experimental setup Experimental setup

Pitch=183 Pitch=183 mm

w/p~0.25w/p~0.25

Sensor width=500 Sensor width=500 mm

V=300 VV=300 V

25 ns bunch spacing

1 2 3 4 5 6

100 mrad

Non irradiated TOB modules

120 GeV ,

Almost final DAQ setup

25-Oct-2001/3-Nov-200125-Oct-2001/3-Nov-2001

Page 3: Roberto Chierici - CERN

Roberto Chierici 3

Event ReconstructionEvent Reconstruction Rough pedestal pedRough pedestal ped00 (n (n00 events) events)

<ADC<ADCii> in time > in time

Refined pedestals and first common mode noise (nRefined pedestals and first common mode noise (n11 events)events)

<ADC<ADCii>, >, ADCADC in time removing strips with pol×(ADC in time removing strips with pol×(ADCii-ped-ped00)>threshold)>threshold

CMNCMN00=<ADC=<ADCii-ped-pedii>> over stripsover strips

Noise determination and better pedestals/CMN (nNoise determination and better pedestals/CMN (n22 events) events) Exclude those strips for which pol×(ADCExclude those strips for which pol×(ADCii-ped-pedii)>K)>KADCADC

CMN=<ADCCMN=<ADCii-ped-pedii>> over strips; pedover strips; pedii= <ADC= <ADCii-CMN> in time-CMN> in time nnii

22= <ADC= <ADCii-ped-pedii-CMN>-CMN>22 in time in time

Loop over eventsLoop over events Remove bad events, determine noisy/dead stripsRemove bad events, determine noisy/dead strips Recalculate CMN; update pedestals and noise after nRecalculate CMN; update pedestals and noise after n00+n+n11+n+n22 events events

Cluster findingCluster finding ssii=ADC=ADCii-ped-pedii-CMN-CMNii consider only those for which s consider only those for which sii/n/nii>2>2 Good clusters if nGood clusters if nstripstrip>0, S>0, Sclcl

11/N/Nclcl11>5>5

Page 4: Roberto Chierici - CERN

Roberto Chierici 4

PedestalsPedestalsModule 1

Module 2

Module 3

Module 4

Module 5

Module 6

Plots from G. PásztorPlots from G. PásztorAPV 1APV 1 APV 2APV 2 APV 3APV 3 APV 4APV 4

Module 2

d

eco

nvolu

tion

deco

nvolu

tion

Page 5: Roberto Chierici - CERN

Roberto Chierici 5

NoiseNoise

Noise < 3 ADC countsNoise < 3 ADC counts

N N depends upondepends upon

the updatingthe updating

Very stable in Very stable in

space space and and timetime

Module #2: noise for twoModule #2: noise for two

different updating windowsdifferent updating windows

Page 6: Roberto Chierici - CERN

Roberto Chierici 6

Corrected dataCorrected data ssii=ADC=ADCii-ped-pedii-CMN-CMNi i ( (CMNCMN~0.3~0.3noise)noise) pedestal runs tell us we correctly estimate our noisepedestal runs tell us we correctly estimate our noise

deviation of a factor 2 only outside 4deviation of a factor 2 only outside 4 regionregion

Pedestal runPedestal run

=1.04=1.04

Page 7: Roberto Chierici - CERN

Roberto Chierici 7

Cluster characteristicsCluster characteristics

Unexpectedly large number of strips per cluster !Unexpectedly large number of strips per cluster ! Confirmed by different analyses (Pisa)Confirmed by different analyses (Pisa) Effect not present in July 2000 beam test Effect not present in July 2000 beam test

(but very different settings)(but very different settings)

APV 1APV 1 APV 2APV 2 APV 3APV 3 APV 4APV 4

Module #2Module #2

From Pisa groupFrom Pisa group

S/NS/Nclustercluster~20~20

Page 8: Roberto Chierici - CERN

Roberto Chierici 8

Latency scans Latency scans

Detectors 1-2-5-6 kept at optimal latency valueDetectors 1-2-5-6 kept at optimal latency value Latency scans for detectors 3-4 Latency scans for detectors 3-4

Excellent way for studying delay curves and efficiencyExcellent way for studying delay curves and efficiency

Preliminary 1D trackingPreliminary 1D tracking Use 15k muons for alignment of modules (fits to residuals)Use 15k muons for alignment of modules (fits to residuals) Modules 3-5 (4-6) aligned with respect to 1 (2)Modules 3-5 (4-6) aligned with respect to 1 (2) Look for coincident clusters in modules 1-5 (2-6) and build a Look for coincident clusters in modules 1-5 (2-6) and build a

“track” :)“track” :) Look what happens around the intercept (Look what happens around the intercept (10 strips) in modules 3 10 strips) in modules 3

and 4and 4

Averaging over eventsAveraging over events

Scans of all APVs.Scans of all APVs. Alignment procedure as aboveAlignment procedure as above Tracking gives worse performance: use the highest strip in moduleTracking gives worse performance: use the highest strip in module

Runs in deconvolution mode

Runs in peak mode

Page 9: Roberto Chierici - CERN

Roberto Chierici 9

Delay curvesDelay curves

deconvolution

undershoot...

Non-ideal APV parameters(VFS=70, Isha=90) asymmetry too efficient 25 ns off peak undershoots

q in 10 strips from theintercepts in modules 3,4 no cluster finding

dependence

Page 10: Roberto Chierici - CERN

Roberto Chierici 10

Amplitudes in timeAmplitudes in timeDeconvolution Peak

Page 11: Roberto Chierici - CERN

Roberto Chierici 11

Strips in clusterStrips in cluster

Peak

Q

q

q

Cb

Cint

Cint

Cb

Cb

CAC

preampl. shaper

C

L

R

Mode

latency

Symmetric charge sharing (Y)

Asymmetric diffusion (X)

q1 q2

Diffusive component ~ 20% of the total

Page 12: Roberto Chierici - CERN

Roberto Chierici 12

Delay curves per strip Delay curves per strip (peak)(peak)

Test beam data ‘faster’ curve for adjacent strips signal propagating on 2 closest strips

X-checked by lab calibration very similar peak ratios

|L-R|/(L+R+T)<0.2

Cal. channel

1

2

(L. Mirabito)(L. Mirabito)

Page 13: Roberto Chierici - CERN

Roberto Chierici 13

Delay curve per strip Delay curve per strip (deco)(deco)

Reasonable shape of the hit strip deconvolution nicely tuned

Different output for the neighboring: result of the different pulse shape(input to the deco not anymore a CRRC 50 ns) deconvolution enhances q1/q0

|L-R|/(L+R+T)<0.2

Possible explanation given by the behaviour of the amplifier as R at high frequency: +Cint=h.p. filter to adjacent strips

The shape is expected to be APVparameter dependent !

Consequences for the tracker: position resolution cluster reconstruction two track separation data volume studies going on...

=delay curve

Delay (ns)Delay (ns)

Page 14: Roberto Chierici - CERN

Roberto Chierici 14

Cluster finding efficiencyCluster finding efficiency

Excellent efficiency at 75 ns (mod. 3) not too sensitive to the position of the maximum cluster finding much more efficient than charge integral

~99.5%

Still too efficient at 25 ns better tuning of APV parameters

Cluster finding cuts to be optimized… efficiency curve can be adjusted the efficiency ‘plateau’ can be considerably smaller

Page 15: Roberto Chierici - CERN

Roberto Chierici 15

Response functionResponse function

Diffusive regionsDiffusive regions

The response function can be The response function can be determined by assuming a uniform determined by assuming a uniform beam intensity beam intensity over the strip:over the strip: diffusion regiondiffusion region position resolutionposition resolution(work is going on…)(work is going on…)

Charge sharingCharge sharing

Page 16: Roberto Chierici - CERN

Roberto Chierici 16

ConclusionsConclusions The 25 Oct - 3 Nov test beam on 25 ns beam was a successThe 25 Oct - 3 Nov test beam on 25 ns beam was a success

6 TOB modules tested on a 25 ns beam with the next-to-final DAQ setup6 TOB modules tested on a 25 ns beam with the next-to-final DAQ setup excellent quality of the collected dataexcellent quality of the collected data

Several configurations triedSeveral configurations tried latency scans in peak and deconvolutionlatency scans in peak and deconvolution latency scans with different APV parameterslatency scans with different APV parameters special triggersspecial triggers

Preliminary results very interestingPreliminary results very interesting S/N of clusters ~ 20 (deco, non irradiated detectors), noise as expectedS/N of clusters ~ 20 (deco, non irradiated detectors), noise as expected work going on for optimizing delay curves. Excellent track efficiencywork going on for optimizing delay curves. Excellent track efficiency CCintintAPV at high frequency may cause undesired features (not dramatic)APV at high frequency may cause undesired features (not dramatic)

A lot of things to do...A lot of things to do... Huge amount of data (340 GB on castor) to analyzeHuge amount of data (340 GB on castor) to analyze optimize/distribute common tools for data analysisoptimize/distribute common tools for data analysis further investigations + lab tests to be continuedfurther investigations + lab tests to be continued

Everyone very welcome to join !

Page 17: Roberto Chierici - CERN

Roberto Chierici 17

Further infoFurther info

Module #2Module #2

Page 18: Roberto Chierici - CERN

Roberto Chierici 18

Two eventsTwo events Zero suppressed infoZero suppressed info

Module tilting visibleModule tilting visible

Page 19: Roberto Chierici - CERN

Roberto Chierici 19

Deconvolution...Deconvolution...

22=100 ns=100 ns

22=50 ns=50 ns

22=25 ns=25 ns

CR-RC(CR-RC(11))RC(RC(22)) deconvolutiondeconvolution

Page 20: Roberto Chierici - CERN

Roberto Chierici 20

DAQ setupDAQ setup

Page 21: Roberto Chierici - CERN

Roberto Chierici 21

Shape vs amplitudeShape vs amplitude Pulse height ratios are stable for different input amplitudes.Pulse height ratios are stable for different input amplitudes.

AmplitudeAmplitude

Page 22: Roberto Chierici - CERN

Roberto Chierici 22

Diffusion and sharingDiffusion and sharingDeconvolution Peak

Page 23: Roberto Chierici - CERN

Roberto Chierici 23

Some (nice) picture...Some (nice) picture...

Page 24: Roberto Chierici - CERN

Roberto Chierici 24

Charge asymmetry in Charge asymmetry in timetime

Deconvolution Peak

Page 25: Roberto Chierici - CERN

Roberto Chierici 25

Charge sharing in timeCharge sharing in timeDeconvolution Peak

Page 26: Roberto Chierici - CERN

Roberto Chierici 26

AimAim Experimental setup (hardware and software)Experimental setup (hardware and software) Event reconstructionEvent reconstruction

Pedestal and common modePedestal and common mode Noise evaluationNoise evaluation Cluster findingCluster finding

Module performanceModule performance S/N and stabilityS/N and stability Cluster characteristicsCluster characteristics

Latency scansLatency scans Peak and deconvolution mode: features Peak and deconvolution mode: features Efficiency and delay curves per strip Efficiency and delay curves per strip New observationsNew observations Resolution curveResolution curve

Conclusions (preliminary!)Conclusions (preliminary!)

Page 27: Roberto Chierici - CERN

Roberto Chierici 27

From Pisa groupFrom Pisa group

Page 28: Roberto Chierici - CERN

Roberto Chierici 28

7 entries per track7 entries per track

From Pisa groupFrom Pisa group

Page 29: Roberto Chierici - CERN

Roberto Chierici 29

Special runsSpecial runs Trigger changed from 1001 to Trigger changed from 1001 to 00110011

another way to study efficiency off peakanother way to study efficiency off peak

<n>~1.86 <n>~1.552nd cluster

Page 30: Roberto Chierici - CERN

Roberto Chierici 30

Further crosschecks Further crosschecks (Pisa)(Pisa)

July 2000July 2000

November 2001November 2001

Page 31: Roberto Chierici - CERN

Roberto Chierici 31

S/N in clusterS/N in cluster Pisa resultsPisa results