APEX Software Seamus Riordanhallaweb.jlab.org/experiment/APEX/collab15/riordan_apexsoft.pdf · APEX Software Seamus Riordan Stony Brook University [email protected] April

APEX Software

Seamus RiordanStony Brook University

[email protected]

April 19, 2015

Seamus Riordan — APEX 2015 APEX Soft 1/25

Outline

HRS VDCs and APEX

High Rate Tracking Performance

Optics Going Forward


Basic VDC Operation

Tracks enter nominally 45◦, produce signals on 3-7 wires

Drift time patterns among several wires matched to construct“cluster”

2 U-plane and 2 V-plane clusters fit to recreate full 3D track

Ci+4

Ci+3

Ci+2

Ci

Ci+1

di+2

crossover

point

id

di+3 di+4

D=13mm

wire plane

4.24mm

"pivot wire"

τslope

di+1

(HV) plane

cathode

(HV) plane

cathode

particle trajectory

Drift Time (ns)50 0 50 100 150 200 250 300 3500

200

400

600

800

1000

1200

1400

1600

1800

2000

2200

Drift Time Spectrum, U1


VDC Modifications

Modifications for performance up to 5 MHz (full experimentluminosity).

Standard APEX High RateHV −4.0 kV −3.5 kVDisc. LeCroy (Ith = 8 µA) JLab Custom (Ith = 1 µA)Gas 60-40 Ar/CH2 60-40 Ar/CH2

Max Rate 500 kHz 5 MHzGain 20× 103 25× 103

Max VDC current draw I/wire/cm ∼ 5 nA

For APEX, QVDC < 0.1 C (no serious aging)


Tracking Algorithm - Clustering

Ci+4

Ci+3

Ci+2

Ci

Ci+1

di+2

crossover

point

id

di+3 di+4

D=13mm

wire plane

4.24mm

"pivot wire"

τslope

di+1

(HV) plane

cathode

(HV) plane

cathode

particle trajectory

→

Wire Number160 162 164 166 168 170

Dri

ft T

ime (

ns)

50

0

50

100

150

200

VDC U1

Algorithm scans for ’V’ shaped clusters in time

Hits in each cluster must be within reasonable time constraints

Allow for gaps of 1 wire, must have 3 ≤ wires in cluster ≤ 7

Time of the cluster is offset, calculated through fit based ontime-to-drift distance mapping

Time resolution from fit on σt ≈ 15 ns


Advanced Scanning

Wire Number0 50 100 150 200 250 300 350

Dri

ft T

ime (

ns)

100

0

100

200

300

400

500

VDC U1

Multihit TDC information used since rates are high

Earliest hits used to fit clusters

Several passes over data taken to maximize clusters foundwhen separated in time, but not space


Tracking Algorithm - UV Association

Cut on U cluster, V cluster time difference w/ trigger, ±40 ns

Cluster positions must be in chamber active area

Time Difference (ns)100 80 60 40 20 0 20 40 60 80 1000

500

1000

1500

2000

2500

VDC UV Cluster Time Difference

Select clusters

near in time

VDC UV Cluster Time Difference

Ambiguity prevents uniqueassociation between U-V clusters

If ambiguity in UV association, no tracks are returned


Tracking Algorithm - Track Fitting

All chamber 1 - chamber 2 UV clustersbuilt

Sort by χ2 based on angularinformation from drift time fit

Accept as many χ2 clusters untilmaximum found


Tracking Efficiency

Tracking efficiency found in left arm for:

Left arm s2m scintillator triggerHigh preshower+shower calorimeter signal (e−)

Low rate, 0.3 MHz

Cluster Size

2 3 4 5 6 7 80

2000

4000

6000

8000

10000

12000

14000

16000

18000

u1 Cluster size 2uA Tantalum

Highest rate, 5 MHz

Cluster Size2 3 4 5 6 7 8

0

2000

4000

6000

8000

10000

u1 Cluster size High Current Lead

Average wires in clusters become smaller at high rate due toefficiency


Tracking Efficiency

Can take low rate as base and calculate how it shifts for hitinefficiencies

2 3 4 5 6 7 80

5000

10000

15000

20000

2 3 4 5 6 7 80

2000

4000

6000

8000

10000

12000

14000

16000

Have ∼ 92% hit efficiency

Expect 2% loss of tracking efficiency with hits < 3


Tracking Efficiency

Can take low rate as base and calculate how it shifts for hitinefficiencies

Cluster Size2 3 4 5 6 7 8

0

2000

4000

6000

8000

10000

u1 Cluster size High Current Lead

2 3 4 5 6 7 80

2000

4000

6000

8000

10000

12000

14000

16000

Have ∼ 92% hit efficiency

Expect 2% loss of tracking efficiency with hits < 3


Tracking Efficiency - II

From Eric’s thesis:

Required coincident event and then frequency of how often atrack was found in an arm99.0% for LHRS, 98.2% for RHRSClose enough with simple cluster argument?


Tracking Efficiency - Losses

Losses come from:UV association ambiguityNo clusters found (bad timing structure, overlapping, hitinefficiency)

LHRS, Single Arm Trigger:

Trigger Rate (MHz)0 1 2 3 4 5 6

Fra

cti

on

al L

oss

0

0.1

0.2

0.3

0.4

0.5 Total Loss

Loss from Ambig.

Loss from No Clusters

Event Loss

Loss goes up to 40% for highest current running


Efficiency Improvements

Tracking Work To Do:

UV ambiguity may be broken through use of other detectors,χ2 fitting, geometry considerations, event distributionconsiderations

Some clusters from “no cluster” events may be recoveredthrough better cluster searching code

Try to understand how spurious clusters are generated - G4simulation?


Overview and Strategy

APEX has unique optics requirements

FOM is driven by scattering angle resolution rather thanmomentum resolutionSeptum needed to go from 6◦ to 12.5◦

Requires careful examination of HRS tune to optimize

John LeRose is gone so we have to think through much of thisourselves

Have produced 2nd order transport matrices for ideal case asstarting test ground

Confirmation in existing ray tracing required for final result


Optics Overview

1.80

1.80

4.42

3.57

1.50

1.25

0.80

1.69

45°30°

30°

8.40

10.37

3.05

HRS Design Layout(design magnet effective lengths displayed)

Q1Q2

Dipole

Q3

20.76

1st VDC Plane

Dimensions in meters

1 m

θ0

φtg y

tg

ysieve

z

Sieve plane

Beam

Scattered

electron

L

D

Hall centerzreact

ytg

tg

central ray

Spectrometer

Using idealized representation of HRS first and secondorder elements can be calculated

Excellent primer for framework

http://cds.cern.ch/record/283218/files/SLAC-75.pdf

http://github.com/seamusriordan/hrstrans


APEX Tune - Simple Optimization Attempt

t [m]0 5 10 15 20 25

4−

2−

0

2

4

6

8

10

12

14

)tg

(x|x)tgθ(x|

)δ(x|

x Couplings

t [m]0 5 10 15 20 25

2−

0

2

4

6

8 )tg

(y|y)

tgφ(y|

y Couplings

Using first order code allows for rapidcalculation of acceptance and matrixelements

A fitter is used which

Preserve acceptance for targetlength (as seen at 6◦)Produce first order reconstructionmatrix by inverting transportMinimize reconstruction elementsof φtg with VDC spatial/angularresolutions

Immediately drove (y |ytg) coupling tozero and increased (y |φ)

About the same acceptance - butserious losses in momentum and vertexresolution (> factors 2?)


APEX Tune - Simple Optimization Attempt

θ0.2− 0.15− 0.1− 0.05− 0 0.05 0.1 0.15 0.2

0

0.05

0.1

0.15

0.2

0.25

0.3

Origin - Acceptedθ

y [m]0.2− 0.15− 0.1− 0.05− 0 0.05 0.1 0.15 0.2

0

0.05

0.1

0.15

0.2

0.25

0.3

Y Origin - Accepted

φ0.1− 0.05− 0 0.05 0.1

0

0.1

0.2

0.3

0.4

0.5

0.6

Origin - Acceptedφ

δ0.1− 0.05− 0 0.05 0.1

0

0.05

0.1

0.15

0.2

0.25

0.3

Origin - Acceptedδ

Using first order code allows for rapidcalculation of acceptance and matrixelements

A fitter is used which

Preserve acceptance for targetlength (as seen at 6◦)Produce first order reconstructionmatrix by inverting transportMinimize reconstruction elementsof φtg with VDC spatial/angularresolutions

Immediately drove (y |ytg) coupling tozero and increased (y |φ)

About the same acceptance - butserious losses in momentum and vertexresolution (> factors 2?)


Summary and Work To Do

Tracking generation for trying to match all clusters is about60% at 5 MHz trigger, can this be understood?

John LeRose’s files were recovered and are available - I haveaccess to SNAKE and the files used for standard tune, PREX,and have been using them

Need to include our new septum, produce full polynomialtransport, verify acceptance and reconstruction

I believe this is a project which we can handle on our own andthis is at a point where a grad students could get involved


BACKUP SLIDES


HRS Vertical Drift Chambers

45O

Side view

Top view

nominal particle trajectory

Upper VDC

Lower VDC

2118 mm

288 mm

nominal particle trajectory

d =335mmv

d =335mmu

v2

d =26mmuv1

d =26mmuv2

v1

u1

u2

U1

V1

U2

V2

Nominal Characteristics

U/V angle ±45◦

Sense wires/plane 368Wire Spacing 4.24 mm

Pos. Res ∼ 100 µmAng. Res ∼ 0.5 mrad


Pre-test run PAC Requirements

Requested for test run by PAC:Prove that the vertical drift chambers (VDCs) can operate ata rate higher that 20 kHz/wire (that, according to the TACreport, is the maximum Hall A has operated till now).

VDCs had not been run at such high rate (for extended periodof time)

Required to go to ∼ 5 MHz (75 kHz/wire)

Requires hardware modifications to run efficiently withoutsevere aging


Timing Offset Calibration

VDC requires software offsets for drift time

Calibrated in groups of 16 wires (discriminator inputs)

500 600 700 800 900 1000 1100 1200 1300 14000

20

40

60

80

100

120

140

u1 TDC grp 8u1 TDC grp 8

Calibration done by fitting time dist. peak and fixed at 1.4 σearlier from peak (arbitrary)


Timing Offset Calibration Results

Calibration is done to ∼ ns level

Offsets may be different for different triggers

Minimized in hardware to ∼ 10 ns level, fully corrected insoftware

wire group number0 2 4 6 8 10 12 14 16 18 20 22 24

Mean

Peak P

osit

ion

(n

s)

15

16

17

18

19

20

21

22

23

24

25

time (ns)-50 0 50 100 150 200

No

rma

lize

d e

ntr

ies

0

50

100

150

200

250

300

350

400Trigger1

Trigger2

Trigger3

Trigger4

Trigger5

Trigger6


Drift Time-to-Distance

Drift time-to-distance conversion follows form:

Theta dependence:

v2t < 0 : d = v2t

0 < v2t < a1 : d = v1t = v2t

(1 +

a2

a1

)a1 < v2t : d = v2t + a2

a1 and a2 carry tan θ = ∆z∆r dependence (r = u or v)

a1 =3∑

i=0

a1,i tani θ

a2 =3∑

i=0

a2,i tani θ


TTD Calibration

No serious differences between high and low rate data

Restricting to slice in incident angle θ:

Low Rate, 0.4 MHz

Drift Time (ns)50 0 50 100 150 200 250 300

Dri

ft D

ista

nce (

m)

0

0.002

0.004

0.006

0.008

0.01

0.012

0.014

TTD Function

|<5 mradfp

θTTD, low current, |

High Rate, 4.6 MHz

Drift Time (ns)50 0 50 100 150 200 250 300

Dri

ft D

ista

nce (

m)

0

0.002

0.004

0.006

0.008

0.01

0.012

0.014

TTD Function

|<5 mradfp

θTTD, high current, |

Small recalibrations for θ dependence are necessary


TTD Calibration

Requires some θ dependence re-fitting

Discrepancies are in the tail on the level of 20 µm

(rad)fp

θ

0.1 0.05 0 0.05 0.1

Resid

ual (m

)

0.002

0.0015

0.001

0.0005

0

0.0005

0.001

0.0015

0.002

fpθU1 Residual vs.


Tracking Resolutions

Track resolutions found through residuals of full χ2 fit

Residual (m)0.003 0.002 0.001 0 0.001 0.002 0.003

0

2000

4000

6000

8000

10000

12000

14000

mµRes. = 354.0

U1 Residual

Second, broader Gaussian distribution appears at high rate

Should be fitting to students t distribution with mult scatt

Corresponds to 100 µm, 0.3 mrad detector resolution


Tracking Efficiency

Tracking efficiency found in left arm for:

Left arm s2m scintillator triggerHigh preshower+shower calorimeter signal (e−)

Low rate, 0.3 MHz

N Clusters0 2 4 6 8 10

0

2000

4000

6000

8000

10000

12000

+/ 40ns

No ambiguity

Used in track

u1 Clusters in 40 ns 2uA Tantalum

Highest rate, 5 MHz

N Clusters0 2 4 6 8 10

0

2000

4000

6000

8000

10000

12000+/ 40ns

No ambiguity

Used in track

u1 Clusters in 40 ns Highest Current Lead

Average wires in clusters become smaller at high rate due toefficiency


Event distribution

Event distribution has small distortions due to non-uniformefficiency

How does this affect the acceptance?

Xdet

(m)det

x1 0.8 0.6 0.4 0.2 0 0.2 0.4 0.6 0.8 1

0

50

100

150

200

250

300

350

310×

0.4 MHz, Lead

4.6 MHz Lead

Detector X Distribution, S2m trigger (T1), Normalized to charge and prescales

Ydet

(m)det

y0.06 0.04 0.02 0 0.02 0.04 0.06

0

100

200

300

400

500

600

700

800

900

310×

0.4 MHz, Lead

4.6 MHz Lead

Detector Y Distribution, S2m trigger (T1), Normalized to charge and prescales


Standard Tune Comparison

x-plane

z0 5 10 15 20

-6

-4

-2

0

2

4

6

8

10

12

<x|θtg>

y-plane

z

0 5 10 15 20

-4

-2

0

2

4

6

8

Q1 Q2 Q3Dipole

Q1 Q2 Dipole Q3

<x|xtg>

<x|δ>

<y|φtg>

<y|ytg>

[m]

[m]

[m]

[m]

t [m]0 5 10 15 20

4−

2−

0

2

4

6

8

10

12

)tg

(x|x)tgθ(x|

)δ(x|

x Couplings

t [m]0 5 10 15 20

2−

1−

0

1

2

3

4

5

6

7)

tg(y|y

)tg

φ(y|

y Couplings

Framework does a reasonably good job of recreating firstorder elements

Acceptance ∼ 6 msr using simple aperture cuts


APEX Tune

APEX test run ran with special tune which nominally reproducesstandard tune optics and acceptance but with pure (y |φ) coupling

Standard Optics

t [m]0 5 10 15 20

4−

2−

0

2

4

6

8

10

12

)tg

(x|x)tgθ(x|

)δ(x|

x Couplings

t [m]0 5 10 15 20

2−

1−

0

1

2

3

4

5

6

7)

tg(y|y

)tg

φ(y|

y Couplings

→

APEX Test Run Configuration

t [m]0 5 10 15 20 25

4−

2−

0

2

4

6

8

10

12

14

)tg

(x|x)tgθ(x|

)δ(x|

x Couplings

t [m]0 5 10 15 20 25

2−

0

2

4

6

8

)tg

(y|y)

tgφ(y|

y Couplings


APEX Tune

APEX test run ran with special tune which nominally reproducesstandard tune optics and acceptance but with pure (y |φ) coupling

Standard Optics

θ0.2− 0.15− 0.1− 0.05− 0 0.05 0.1 0.15 0.2

0

0.05

0.1

0.15

0.2

0.25

0.3

Origin - Acceptedθ

y [m]0.2− 0.15− 0.1− 0.05− 0 0.05 0.1 0.15 0.2

0

0.05

0.1

0.15

0.2

0.25

0.3

Y Origin - Accepted

φ0.1− 0.05− 0 0.05 0.1

0

0.1

0.2

0.3

0.4

0.5

0.6

Origin - Acceptedφ

δ0.1− 0.05− 0 0.05 0.1

0

0.05

0.1

0.15

0.2

0.25

0.3

Origin - Acceptedδ

→

APEX Test Run Configuration

θ0.2− 0.15− 0.1− 0.05− 0 0.05 0.1 0.15 0.2

0

0.05

0.1

0.15

0.2

0.25

0.3

Origin - Acceptedθ

y [m]0.2− 0.15− 0.1− 0.05− 0 0.05 0.1 0.15 0.2

0

0.05

0.1

0.15

0.2

0.25

0.3

Y Origin - Accepted

φ0.1− 0.05− 0 0.05 0.1

0

0.1

0.2

0.3

0.4

0.5

0.6

Origin - Acceptedφ

δ0.1− 0.05− 0 0.05 0.1

0

0.05

0.1

0.15

0.2

0.25

0.3

Origin - Acceptedδ


Documents

APEX Software Seamus Riordanhallaweb.jlab.org/experiment/APEX/collab15/riordan_apexsoft.pdf · APEX Software Seamus Riordan Stony Brook University [email protected] April