ST Alignment Validation M. Needham EPFL

ST Alignment ValidationM. Needham

EPFL

IntroductionDetailed comparision of alignment databases

• October TED Overlaps: magnet on/ magnet off alignment

• Residuals November TED/ Beam off: magnet on/off alignment

• Residuals Long tracks magnet on: magnet on/off alignment

• IT Error tuning

Definitions + Databases:Magnet off: head-20100119Magnet on: slice VeloOTTxTyModulesTxITTxTyRzTTModulesTxRz20100119.dbmagnet on++: magnet on + IT ladder alignment TxRz (with health warning) VeloOTTxTyModulesTxITTxTyRzLaddersTxRzTTModulesTxRz20100122.db

IntroductionDatasets:

Magnet off: November TED, all Collision data with magnet off (including data with VELO moving) Magnet on: 63801, 63806, 63807, 63807, 63809, 63811, 63815, 63849

Magnet on selection cuts

The ST selection: p > 6 GeV, 2/dof < 7

Overlaps• Take Wouters database, transform to October TED and look at overlaps

• Pros: Large high quality dataset that is well understood

• Cons: October to December is a long time, IT + OT opened in October

Box T1 T2

B -800 -1025

A -800 -1042

T None -60

C None -80

Transformation October- November

• Based on studies November TED

• ‘pre-alignment’ in x only

• Quality: dominated by systematics

• Uncertainty: 50 microns ?

OverlapsOverlaps in y decrease with z(beampipe hole bigger)T1: 19 mmT2: 12 mmT3: 5 mm

A-Side C-Side

Top

Bottom

1 mm

1 mm

3 mm

5 mmIT3

No physical overlap IT3C and IT3 Bottom

Survey + correctionfor OT closing, boxAlignment changes by ~ mm

Other stations similarshifts

A-Side Overlaps

T1 T2 T3

B1 B2

A-Side OverlapsOverlap to Residual Magnet

on/ mmResidual Magnet off/mm

Bottom1 -0.17 0.05

Bottom2 0.17 0.09

Top1 0.2 0.1

Top2 0.45 0.32

Top3 0.6 0.43

C-Side Overlaps

T1

B1

T2 T3

Bias’s : 100- 200 micron level

Better than anything we saw before….

Large biases but comparing October TED

with December beam running and including

constrainsts from VELO + OT

C-Side OverlapsOverlap to Residual Magnet

on/ mmResidual Magnet off/mm

Bottom1 -0.32 -0.19

Top1 0.690 0.12

Top2 0.02 0.02

Top3 -0.46 -0.12

Overlap Summary• Magnet off data has biases of 200 microns in TED overlaps

• Magnet on data has larger biases

• But the movements are all at the 200 micron level [apart from Top T1] compared to magnet off

• Constants are not so inconsistant magnet off/on

• Pick one false minimum magnet off, another magnet on ?

Look at residuals in T2

• Dataset: November TED + magnet off

• Databases: magnet on , magnet off

Box Magnet on/mm Magnet off/mm

Top -0.03 -0.02

Bottom -0.02 -0.01

C-side 0.09 -0.01

A-side -0.15 -0.03

T1 T2 T3

dx

Differences magnet on/off at 100 micron level

x Residuals Test

Residual Biases: Tracks

Bias @ 15micron level

Magnet-off dataMagnet off alignment

Magnet-off dataMagnet on alignment Bias @ 18

micron level

IT1

IT2

IT3

A C

Residual Biases: Tracks

RMS 85 mMagnet-off dataMagnet on++ alignment

But this database is good for magnet on

Magnet++Magnetoff

Unbiased residualalso degrades

Residual Biases: Tracks

RMS 62 mMagnet-on dataMagnet on alignment

RMS 64 mMagnet-on dataMagnet off alignment

Residual Biases: Tracks

RMS 15 m

Magnet-on dataMagnet on++ alignment

• Looks good and unbiased residual width improves by ~ 30 %

• But this database is very bad for magnet off

Magnet++Magnet

Unbiased residual

Future StrategyFinding the correct minimum is not trivial

• Is there a real difference magnet off/on ?

Get back to what we did on the MC, what we learnt was good with TED

• Tune clusters errors [next slide]

• Combining datasets [Wouter]

• Track selection

• Evolving cut and strong isolation criteria as in TED/MC studies

• Momentum cut

IT Error TuningAs in the TED, tune unbiased residuals for 1, 2, 3 ,4 strip clusters

• Reasonable track selection: 2/dof < 7, p > 20 GeV

• Assume 1 strip has a binary contribution

• Unfolding this gives the misalignment / multiple scattering

• Unfold this from the 2,3,4 cluster unbiased residualCluster Size Unbiased

residual/mmMC value/pitch Data tune/pitch

1 77 0.19 0.29

2 68 0.12 0.22

3 87 0.17 0.35

4 No Statistics 0.04 put 0.35 ?

itClusterPosition = STOfflinePosition('ToolSvc.ITClusterPosition')itClusterPosition.ErrorVec = [0.28, 0.22, 0.35. 0.35]itClusterPosition.APE = 0.1

Early data tune:

Summary• Getting the IT boxes inter-aligned is tough

• Best database: good to 100 micron

• Databases with magnet off/on differ

• Differences @ the level of 100 -200 micron

• But general trend is reasonably same ?

• Magnet++ is different

• Fixes magnet on by LARGE movements of ladders at price of worse magnet off

Backup

Go to TT

A-Side C-Side

Project to TT

Better than Anything I have seen

Go to TTIT Track Type Residual/mm (Beam) Residual/mm (Oct)

A-side 0.3 0.6

Bottom -0.4 0

C-side -0.3 -0.8

Top -0.2 -0.7

Compare residuals extrapolating IT tracks to TT in TED + beam data

Again discrepancies at the level of hundreds of microns

Can partially be explained by different illumination [at least in Bottom case]

Go to TT

Examine the badGuy in TTaX

TED dataIt’s a rotation

Propagate IT generic tracks to TT

Beam smaller y ]

-5 mrad inLocal frame

Go to TT-0.93 -1.02 -0.52

0.24 -1.88 -0.35

-0.84 -0.84 -0.52

0.1 -0.7 -0.3

-0.79 -0.76 -0.43

0.2 -0.62 0.0

-0.73 -1.16 -0.61

0.12 0.87 -0.08

x

TTaX

TTaU

TTbV

TTbX

Columns:TTa 7, 8, 9TTb 8, 9, 10

TEDdata

Bad luck

Go to TT• Bottom is consistant with Velo-TT studies of C. Salzmann

• Very good cross-check !

• Top: Seems to be global offset. Coming from IT alignment ?

• Subtracting offset my results consistant with C. Salzmann

Comments• Inner Tracker consists of four loosely coupled systems

• Lightweight frames+ boxes: twisting + distortions (ie rotations) important

• Inter-alignment is a challenge: seen already June TED

• Clear we did not take enough magnet off data (especially with Velo not moving)

• But on the otherhand, we have many good things

• A reasonable survey

• Varied track sample: beam-gas, collisions, halo tracks, ….

• Can use November TED simulataneously

• Easy to use large October TED as cross-check (good to see IT overlaps)

Comments• Thinking ahead…

• Many of these advantages we lose if the detector is opened

• OT is already opened. Movements of IT at 50 micron level

CommentsEnsure complementary + cross-checks @ same level as 2009

• Survey: what survey is needed, what is possible ?

• TED data: ~ 50 shots to allow pre-alignment/cross-checks

• Similar mix of beam-gas as 2009 ?

Improve on 2009

• Larger sample with magnet off (100k ?) to see IT overlaps clearly

• Anything else ? in TAE mode beam2 gas, displaced bunch collisions…

• Magnet off: give me the hits of particles from anywhere I will give you the tracks

October TED presented in December

October TED presented in December

Residual Biases: Tracks

RMS 41 m ~ 18 m

Magnet-on dataMagnet on alignment

RMS 42 m ~ 21 m

Magnet-on dataMagnet on alignment

T-tracks