MERIT analysis - Beam spot size

Goran Skoro

More details: http://hepunx.rl.ac.uk/uknf/wp3/shocksims/mermar/

UKNF Meeting, Oxford, 16 September 2008

MERcury Intense Target

Purpose: To provide a proof-of-principle for the NF/MC 4-MW target concept. To study the effects ofhigh-magnetic fields on the beam/jet interaction.

1234

Syringe PumpSecondaryContainment

Jet Chamber

ProtonBeam

Solenoid

“Each beam pulse is a separate experiment”

Estimate of energy density is based on the beam optics calculations

Beam size has to be measured for each pulse

The MERIT Experiment

Beam monitoring

MERITCamera 484

Camera 454 Camera 414

Beam monitoring (example)

CERN software for online (and offline) beam monitoring

X, Y projections

At the beginning (May, 2008), the goals of this analysis were:

To make a better fitting algorithmTo find x,y positions of the beamTo find ratios of the x,y sigmas

Problem: saturation

For most shots the light intensity is saturated

Find a shadow Fit a tail2nd approach:

Devils Tower Wyoming

Distribution looks like this

To extract ASCII from SDDS

How to extract a beam size?

z(x,y) distribution is in a saturation here

1st approach: To fit projections*

Mean = 0.31(7) mmSigma = 6.42(12) mm

X

Y

X

Y

2nd approach: To fit shadows**

Mean = -4.64(3) mm

Mean = 0.16(4) mm

Mean = -4.71(3) mmSigma = 2.21(3) mmSigma = 2.22(4) mm

Sigma = 4.82(5) mm

We have 3 beam ‘cameras’ -> 3 images for each beam pulse

Shot from Camera 484

* Projection for X is

yn

ii

y

yxzn

xP1

),(1

)(

similarly for Y.

, ** Shadow for X is ),1()],,(max[)( yi niyxzxS similarly for Y.

,

Fitting: Procedure

Simple fitting function: Gaussian + ‘background’

Fitting algorithm (how to avoid gaps; how to choose initial value of the ‘background’ term, etc…) was based on the analysis of the 15-20 randomly selected images (after this, completely ‘blind’ analysis -> no parameters tuning)

In total: 520 beam pulses* x 3 cameras x 2 projections = 3120 distributions have been fitted

Result: Table – ntuple (part of it shown below)

Fitting: ProcedureCamera

414Camera

454Camera

484TARGET

BEAM

Camera 414

Camera 454

Camera 484

Date(ddmmyyyy)

Time(hhmmss)

Xmean

(mm)Xmean

(mm)Ymean

(mm)Ymean

(mm)Sigmax

(mm)Sigmax

(mm)Sigmay

(mm)…………………

……………………

……………………

……………………

……………………

……………………

…This will be used to reconstruct the Run number and to attach this table to the ‘global’ table with experimental results.

This will be used to recognize a shot with the ‘suspicious’ fitting result and to fit it ‘manually’.

* Period: 23 Oct 2007 – 11 Nov 2007

Xmean (mm) Xmean (mm) Xmean (mm)

Ymean (mm) Ymean (mm) Ymean (mm)

Rela

tive

inte

nsit

y

Camera 414

Camera 414

Camera 484

Camera 454

Camera 454

Camera 484

454 484

TARGET

BEAM

414

Results: Shadows Distributions of the Gaussian means

Distributions of the Gaussian sigmas

x (mm)

Rela

tive

inte

nsit

y

Camera 414

Camera 414

Camera 484

Camera 454

Camera 454

Camera 484

x (mm) x (mm)

y (mm) y (mm) y (mm)

(empty shots, beam on the edge

of the ‘visible field’, etc…)

-Suspicious results

Find the corresponding

event in the table (Slide 6) and fit it

manually (if possible)

454 484

TARGET

BEAM

414

Results: Projections

Xmean(shadow)/Xmean(projection)

Rela

tive

inte

nsit

y

Camera 414

Camera 414

Camera 484

Camera 454

Camera 454

Camera 484

454 484

TARGET

BEAM

414

Cross-checking: Projections vs Shadows

Distributions of the ratios (shadow/projection) of the Gaussian

means

Xmean(shadow)/Xmean(projection)Xmean(shadow)/Xmean(projection)

Ymean(shadow)/Ymean(projection)Ymean(shadow)/Ymean(projection)Ymean(shadow)/Ymean(projection)

Everything is (more or less) symmetrical

around 1. As expected, both approaches

return similar values of x/y means.

Rela

tive

inte

nsit

y

Camera 414

Camera 414

Camera 484

Camera 454

Camera 454

Camera 484

x(shadow)/x(projection) x(shadow)/x(projection) x(shadow)/x(projection)

454 484

TARGET

BEAM

414

y(shadow)/y(projection) y(shadow)/y(projection) y(shadow)/y(projection)

Distributions of the ratios (shadow/projection) of the Gaussian

sigmas

Distributions are not symmetrical around 1 (shifted towards left). It means that sigmas for projections are, in general, bigger than sigmas for shadows.

Cross-checking: Projections vs Shadows

Distributions of the ratios of the Gaussian sigmas

Rela

tive

inte

nsit

y

454 484

TARGET

BEAM

414

jy

iy

jx

ix

,

454

414

x

x

484

454

x

x

454

414

y

y

484

454

y

y

Nice agreement with ‘Beam Optics’

values

Results: Shadows

BO ~ 1

BO ~ 1.33

Rela

tive

inte

nsit

y

454 484

TARGET

BEAM

414

iy

ix

414

414

y

x

454

454

y

x

484

484

y

x

Distributions of the ratios of the Gaussian sigmas

Results: Shadows

‘Beam Optics’ value = 1.3

‘Beam Optics’ value = 1.7

Beam position on target

* From ‘Beam Spot Information’ talk, I. Efthymiopoulos, VRVS Meeting, November 30, 2007

454 484

TARGET

-5.77 m -4.17 m 0.0

454 484

TARGET

BEAM

414

EXP

Taken online (estimated by the eye from the screen data)

FIT

Calculated by using:

1) the fitted beam positions for Camera454 and Camera484 (see Slide 4, for example);

2) the Camera454, Camera484 and target positions*

EXP

FIT

EXP

FIT

454 484

TARGET

BEAM

414R

ela

tive

inte

nsit

yDistributions of the ratios

of the Gaussian sigmas

484

454

x

x

484

454

y

y

484

484

y

x

Mean = 1.07

Mean = 1.41

Mean = 1.80

‘Beam optics’ ~ 1

‘Beam optics’ ~ 1.33

‘Beam optics’ = 1.7

Beam position on target

So far, everything (sigmas ratios, beam positions) looks nice…

… except the absolute values of the beam widths!!! Beam optics calculations: beam sigmas are much smaller

Projections

If we assume that the ‘light intensity’ (from the screens) is proportional to beam intensity (before we reach a ‘saturation intensity’) we can, at least, estimate the correction factor when fitting the projections.

Similar results have been obtained by fitting of shadows.

Objections:

1) Saturation is a problem (‘we could have many sigmas hidden here’)2) Shadows approach looks problematic for the highest beam intensities (only a few points left to fit tails)

This is not a problem (intensity is below the saturation level) and a projection approach will give us correct value of beam width(s)

This is a problem (intensity is 10x higher than the saturation level)

Results: Beam size vs beam intensity

Simulation: Saturation effects

(1)

(2)

(3)

Intensity = 10x ‘saturation intensity’

Intensity = 100x ‘saturation intensity’

Intensity <= ‘saturation intensity’

x = y = 2 mm

x = y = 2 mm

x = y = 2 mm

It is obvious that an extraction of projections from (2) and (3) will not give us gaussians

Projections

X, Y (mm)

(1) x,y = 2.00 (2)

(2) x,y = 2.97 (3)

(3) x,y = 3.78 (4)

Inte

nsi

ty

But, what will happen if we try to fit corresponding projections by using gaussian(s)?

Simulation: Saturation effects

Intensity = 10x ‘saturation intensity’

Intensity = 100x ‘saturation intensity’

Intensity <= ‘saturation intensity’

x = 3 mmy = 1.5 mm

x = 3 mmy = 1.5 mm

x = 3 mmy = 1.5 mm

- In our case, expected value of sigma_x/sigma_y ~ 2

‘Symmetry’ between x and y is broken

Sig

ma_fi

t/sig

ma_i

np

ut

Intensity / ’Saturation intensity’

- Previous slide is for sigma_x = sigma_y

- By plotting sigma_output/sigma_input as a function of intensity we can estimate a correction function

Next step: To find a value of ‘saturation intensity’ in our case

xy

Results: Beam intensity below 0.2 Tp454 484

TARGET

BEAM

414

There are 10 -15 shots (23 Oct 2007) where beam intensity is below 0.2 Tp. The distributions (few examples are shown below) look like perfect double-Gaussians for all shots.

Camera 484

No saturation here BUT sigmas are ~2x bigger than expected from BO calculations!!!

Results of fitting of the shadows

X (mm)

Y (mm)

(1) x = 2.93 (2)

(2) x = 3.26 (2)

(3) x = 3.22 (3)

(1) y = 1.90 (1)

(2) y = 2.05 (2)

(3) y = 1.66 (2)

454 484

TARGET

BEAM

414

For low beam intensity shots, in around 50% of the cases the situation is similar to (1) and (2). Even when we have a beam shot similar to case (3) the x/y widths ratio is close to 2. The plot above shows the results of the fitting of these 3 distributions.

(1)

(2)

(3)

Saturation

Saturation (but not so clear)

‘Proper’ Gaussian

Camera 484

It looks that saturation starts somewhere here

Rela

tive

inte

nsit

y

Results: Beam intensity around 0.3 Tp

Results: ‘Correction’ functionS

igm

a_fi

t/sig

ma_i

np

ut

Intensity / ’Saturation intensity’

~ 0.3 Tp

’Saturation intensity’ ~ 0.3 Tp

0.3 Tp 30 Tp

We can use these ‘functions’ to correct the data

x y

~ 30 Tp

454 484

TARGET

BEAM

414

Beam size vs beam intensity (after correction)

Dots: from beam monitors dataLines: from beam optics calculations

Summary

Analysis of the MERIT beam monitors data has been completed

Two approaches: cross-checked

Measurements vs. calculations: controversy remains

http://hepunx.rl.ac.uk/uknf/wp3/shocksims/mermar/

More details: