injection and lessons for 2012 - espace.cern.ch · 18:03:09 interlocks detected an erratic of the...

Injection and Lessons for 2012

M.J. Barnes, W. Bartmann, C. Bracco, K. Cornelis, L. Drøsdal,

B. Goddard, V. Kain, M. Meddahi, V. Mertens, J. Uythoven

Special thanks to: Collimation, BLM, MPP,OP teams

Outline

2011 Operation

Injection: How far did we go? 144 and 288 bunches

Mitigation measures

Transfer line stability:

Shot-by-shot and bunch-by-bunch variations

Source of instabilities

Steering

Improvements

MKI Failures

Other issues

Conclusions and 2012 Operation

06/02/2012 LHC Performance Workshop - Chamonix 2012

Injection Losses and Intensity Limitations

Loss type Losses in % of dump threshold B1/B2

8b 16b 24b 32b 48b 96b 144b

TCDI shower 1/2 3/5 4/6 5/8 23/24 <50? <75?

Uncaptured beam 4/2 12/3 12/5 16/8 20/8 <40? <60?

Linear extrapolation for

2011 operation

2010 2011 From Chamonix 2011:

Not optimised


B1, max loss

4% dump

Injection of 144 bunches became

fully operational in 2011!

B1, max loss

62% dump

Injection of 288 Bunches During MD


288 bunches (1.05e11 ppb, 2.5-2.7 mm) injected at 30% of

thresholds

Beam 1

288 bunches < 27%




thresholds

Beam 1

288 bunches < 27%

Beam 2

288 bunches < 32%




thresholds

Beam 1

288 bunches < 27%

Beam 2

288 bunches < 32%

Still work to be done to optimize beam in the

SPS/injector chain and accumulation in the LHC

ring and lifetime (RF, transverse damper,

chromaticity…..)

Vacuum activity observed at the MKI during

injection

More statistics needed

Very promising in view of operation with

nominal intensity

Mitigation Techniques

TL showers: Local shielding between TCDIs and LHC (×2-3)

Beam scraping in SPS (×2)

Opening TCDIs (×4) Moving/adding TCDIs (under study)

BLM sunglasses (Little Ionisation Chamber LICs)

Improve transfer line stability

Uncaptured beam Local shielding at TDI (under study)

Injection and abort gap cleaning (×10)

Carefully monitoring beam quality in injectors (transverse beam size and shape, bunch length, satellites,..)

BLM sunglasses


Scraping in the SPS


Scraping studies showed the

importance of the scraping position. The

scraper has to cast a shadow on the

transfer line collimators. In this way even

large emittances can be transferred.

Just increase the scraping when there

are losses at LHC injection is rarely the

best solution

K. Cornelis, Evian 2011

L. Drøsdal

%

Deeper

scra

pin

g Too high losses

in the SPS

B1

TDI

MKI

Q8

Triplet

factor 4

Open TCDI Gap: Loss Shower on LHC BLMs for B1


W. Bartmann, Evian 2011

BLM Sunglasses


Why sunglasses: Unnecessary beam dumps due to TCDI and unbunched beam showers from outside

Aim of sunglasses: Factor 5 margin between injection losses and BLM thresholds for comfortable operation

Options:

replace certain monitors by LICs and relax thresholds at 450 GeV no HW modification, certain thresholds constantly higher for 450 GeV (This year)

blind out BIS input from certain BLMs while injection via deploying additional energy level, regrouping crates, new monitor flag need external signal, threshold change only at injection, severe changes to SIS or BLM system (longer term???)


BLM Sunglasses – LIC Solution


Potential to increase thresholds by factor 60 by replacing

present ICs by LICs

No change to BIS system nor BLM firmware needed

List of monitors to be replaced for B1 and B2 identified

Procedure:

Replace certain monitors: 7 LICs will be installed before start-up

Start with present thresholds (be sure to be aware of bad

injections)

Define maximum thresholds for LICs

Apply increase of thresholds via monitor factor in case

unnecessary dumps limit operation


TL Stability: Observation


L. Drøsdal, Evian 2011



• Tight transfer line collimators – high losses if trajectory not centered (17

dumps for B1 and 10 for B2)

• Injection oscillations have to be below 1.5 mm to respect available aperture

in the LHC

5 h later – same super cycle

composition Trajectory has changed

Necessary to re-steer the line




5 h later – same super cycle

composition Trajectory has changed

Necessary to re-steer the line


• Beginning of run: ~ twice a week

• End of run: every second day

• Frequently the same corrector proposed (TI 2: RCIBH.20804, in phase with MSE/MST)

• Offsets drifting back and forth

• Dependence on SPS supercycle?

Dedicated periods of repeated

extractions on downstream TEDs

with 12 bunches for study of

shot-by-shot variations

TI 2: 82 Shots, 19 June

• Horizontal plane: max 760 mm

• Vertical plane: max 260 mm

TI 8: 117 Shots, 2 November

• Horizontal plane: max 770 mm

• Vertical plane: max 260 mm

Shot-by-shot Stability Studies

used data in Model Independent Analysis to find

strongest Eigenmodes of oscillations

760 mm


Shot-by-shot Variation Sources


TI2H- instability sources TI8H- instability sources


Shot-by-shot Variation Sources


TI2H- instability sources TI8H- instability sources


Observed variations in the magnet currents: • Observed MSE variations are large enough to produce this oscillation

• MST is not strong enough

• MKE variations still to be investigated

MSE:

• Low inductance 20 kA circuit

• Power converter ripple was reduced from 18A to 9A 30% improvement

• Improvement by factor 2 in H still necessary for both lines (work ongoing)

Bunch-by-Bunch Variations


Large bunch-by-bunch trajectory variations observed for TI 8 H –

seen on the bunch-by-bunch injection oscillation amplitudes


Plot from the IQC

B2, 144 bunches

Horizontal plane

Bunch ID

Inje

cti

on

Oscilla

tion

am

plitu

de [m

m]

• Suspected too large ripple of the horizontal extraction kicker (MKE)

waveform

Variation > 1mm (max)

MKE4 Waveform Scan



The MKE4 waveform shows a ripple varying up to 2.5% of kick (max. to min.) at the

flattop (4% at initial overshoot) - specification: 1 % flattop ripple!

Need to improve the MKE4 flat-top ripple (PFN)? - Only possible in LS1

Immediate action: change the delay to move the beam to a flatter part of the waveform

-12-11-10-9-8-7-6-5-4-3-2-101234

44454647484950515253545556

% o

f vo

tage

Kick delay usec

SPS clock + 4.1us

MKE4 waveform scan

4% 2.5%

Impact on Operation


Steering is complicated due to several effects:

Shot-by-shot variations Need to average over several shots to correct trajectory TI2 H

TI8 H

Bunch-by-bunch variations When steering on 12 bunches (intermediate intensity) we only

sample a small part of the waveform – not representative for 144

bunches!

TI8 H

LHC orbit and transfer line

trajectory drifting apart

Difficult to optimize injection oscillations and trajectory at TCDIs at

the same time

TI2 H

TI8 H

→In 2011: ~ 30 min – 2 h to steer (excluding some big outliers)

Estimate 2012 if stability is not improved:

1h steering × 0.5/days × 120 days = 60h!

Can we improve?


Impact on Operation


Steering is complicated due to several effects:

Shot-by-shot variations Need to average over several shots to correct trajectory TI2 H

TI8 H

Bunch-by-bunch variations When steering on 12 bunches (intermediate intensity) we only

sample a small part of the waveform – not representative for 144

bunches!

TI8 H

LHC orbit and transfer line

trajectory drifting apart

Difficult to optimize injection oscillations and trajectory at TCDIs at

the same time

TI2 H

TI8 H


Reduce sources of instabilities (MSE ripple, adjust beam delay wrt MKE waveform)

Time for setup during commissioning new reference trajectory Carefully monitoring beam quality in injectors better detection

of bad beam quality early in the chain (longitudinally already well covered by BQM, transv. can be improved)

Improve IQC references

Limits at TL BPM (MD to define them) easier steering

Clear warning level at critical BLMs depending on number of bunches injected (i.e. MSI ≥ 5% with 12 bunches steering needed )

MKI Flashover 18/04/2011

Evian, 12/12/2011 LHC Beam Operation Workshop

Injection of 2 × 36 bunches spaced by 2.2 ms

Breakdown after ~2 ms All 36b of 2nd batch

were kicked with 110-125% nominal MKI deflection

Beam was on LOWER TDI jaw and over-kicked, i.e.

breakdown in second half of magnet (LHCb signals

support this)

Nearly all p+ of the 36b impacted on the

TDI/TCLIB (grazing) 11 magnets quenched

B1 1

2b

B2 1

2b

B1 7

2b

B2 7

2b

B1 7

2b

B2 7

2b

B1 7

2b

B2 7

2b

B1 is causing the

vacuum pressure

rise in the B2

kicker – this is

not new

MKI HW vacuum interlock reduced from 5e-8 mbar to 2e-8 mbar

New SIS to prevent injection if MKI pressure >2e-9 mbar for 50 ns beam (temporary >2.5e-9 mbar for scrubbing with 25 ns beam)

New interlock implemented (Xmas stop)

Checked carefully TDI angular alignments in IR2 and IR8

TCLI openings TCLIB to 8.3 s

Solenoids between MKI and Q4/A5 switched ON

ALICE

Polarity

B1 left

[mrad]

B1 right

[mrad]

B2 left

[mrad]

B2 right

[mrad]

+ -70 -750 -190 -110

- +86 -1035 -190 -110

dtP

IR2 MKI Erratics on 28/7/2011


4.5µs 2µs

MS-C

erratic

Current in kicker magnet

0

5000

10000

15000

20000

25000

30000

35000

40000

45000

50000

55000

App

rox

PFN

Vol

tage

Sample Number

MKI2. 28 July 2011, 6:03:09 PM

7/28/2011 6:03:09 PM PFN1

7/28/2011 6:03:09 PM PFN2

7/28/2011 6:03:09 PM PFN3

7/28/2011 6:03:09 PM PFN4

Erratic turn-on

of MS-C

16:30:43

18:03:09

Interlocks detected an erratic of the Main Switch of

MKI-C (MS-C) and correctly triggered MS’s and DS’s

of system (within 2µs), emptying PFN via both ends.

Hence kicker-C pulsed for 6.5ms and 3 other

kicker magnets pulsed for up to 4.5µs, emptying

PFNs of energy.

Circulating beam was not in IP2 and therefore

not disturbed.

Batch was extracted from SPS but saw no kick at

MKI2 (current already back to zero in all 4 MKI

magnets) and went straight into the TDI upper jaw.

Erratic of MKI2 MSC at 33kV during resonant

charging – sending current to one of the four

kicker magnets

Interlocks did NOT detect erratic of MS-C:

hence no immediate action was taken to turn-on

other thyratrons full 9 ms PFN pulse length

to kicker C.

Failure ~500µs into charging process:

extraction from SPS correctly inhibited;

Circulating beam was swept over aperture and

grazed TDI (~17% of normal kick) for ~8-9µs

150-190 bunches

M.J. Barnes



4.5µs 2µs


0

5000

10000

15000

20000

25000

30000

35000

40000

45000

50000

55000

App

rox

PFN

Vol

tage

Sample Number

MKI2. 28 July 2011, 6:03:09 PM

7/28/2011 6:03:09 PM PFN1

7/28/2011 6:03:09 PM PFN2

7/28/2011 6:03:09 PM PFN3

7/28/2011 6:03:09 PM PFN4

Erratic turn-on

of MKI2 MS-C

16:30:43

18:03:09






PFNs of energy.


not disturbed.






kicker magnets

Interlocks did NOT detect erratic of MS3:



to kicker C.





150-190 bunches

173 bunches - 2.15e13 p+ lost (not dumped)

3 magnets quenched

ALICE: permanent effects on the Silicon Drift

Detector

MS-C

erratic

M.J. Barnes



4.5µs 2µs


0

5000

10000

15000

20000

25000

30000

35000

40000

45000

50000

55000

App

rox

PFN

Vol

tage

Sample Number

MKI2. 28 July 2011, 6:03:09 PM

7/28/2011 6:03:09 PM PFN1

7/28/2011 6:03:09 PM PFN2

7/28/2011 6:03:09 PM PFN3

7/28/2011 6:03:09 PM PFN4

Erratic turn-on

of MKI2 MS-C

16:30:43

18:03:09






PFNs of energy.


not disturbed.






kicker magnets




to kicker C.





150-190 bunches

173 bunches - 2.15e13 p+ lost (not dumped)

3 magnets quenched

ALICE: permanent effects on the Silicon Drift

Detector

MS-C

erratic

Hardware problem faulty components exchanged

+ additional diagnostic + faster detection

electronics with lower voltage threshold no other

events in 2011 BUT erratics can occur several

times per year!!

M.J. Barnes



4.5µs 2µs


0

5000

10000

15000

20000

25000

30000

35000

40000

45000

50000

55000

App

rox

PFN

Vol

tage

Sample Number

MKI2. 28 July 2011, 6:03:09 PM

7/28/2011 6:03:09 PM PFN1

7/28/2011 6:03:09 PM PFN2

7/28/2011 6:03:09 PM PFN3

7/28/2011 6:03:09 PM PFN4

Erratic turn-on

of MKI2 MS-C

16:30:43

18:03:09






PFNs of energy.


not disturbed.






kicker magnets




to kicker C.





150-190 bunches

Up to a factor of 2-4 higher losses might occur

in future years (higher intensity and # bunches,

worse impact parameter at TDI)

OK for machine, but main concern is safety for

ALICE/LHCb.

Detectors must be sufficiently off during

injection!

MS-C

erratic

M.J. Barnes

y = 7.39E-10x + 2.04E-07R² = 9.98E-01

y = 3.79E-09x + 3.25E-07R² = 1.00E+00

6.00E-07

6.10E-07

6.20E-07

6.30E-07

6.40E-07

6.50E-07

6.60E-07

6.70E-07

6.80E-07

6.90E-07

7.00E-07

7.10E-07

2.580E-07

2.600E-07

2.620E-07

2.640E-07

2.660E-07

2.680E-07

2.700E-07

2.720E-07

2.740E-07

2.760E-07

2.780E-07

2.800E-07

75

77

.5

80

82

.5

85

87

.5

90

92

.5

95

97

.5

10

0

50%

De

lay

(s)

TMR

Vo

ltag

e R

ise

Tim

e (s

)

Magnet Inductance Scale Value (%)

Predicted 5% to 95% rise-time versus magnet Inductance

RISE_TIMEC(V(MagOut)*50kV/250,1.25k,23.75kV)

XVALUE_AT_YV(V(MagOut)*50k/250,12.5k)-XVALUE_AT_YV(V(MagIn)*50k/250,12.5k)

MKI Temperature Interlock


M.J. Barnes

Magnet inductance decreases when reducing

ferrite permeability magnet strength decreases

Rise-time decreases with reducing inductance

and/or capacitance

Delay decreases with reducing inductance

and/or capacitance

Softstart (no beam) measure rise-time

indirect measurement of inductance

(temperature)

During Xmas stop: new diagnostic to measure

delay (more sensitive, better time resolution of

kicker waveform)

y = -1.325E-05x + 7.069E-01R² = 1.501E-02

y = -7.324E-05x + 7.041E-01R² = 7.792E-01

y = -2.065E-05x + 6.970E-01R² = 3.514E-01

y = -3.563E-05x + 7.004E-01R² = 8.375E-01

y = -2.828E-04x + 7.153E-01R² = 6.809E-01

0.695

0.696

0.697

0.698

0.699

0.7

0.701

0.702

0.703

0.704

0.705

0.706

0.707

0.708

0.709

20.00 25.00 30.00 35.00 40.00 45.00 50.00 55.00 60.00 65.00 70.00

Ris

e T

ime

(µ

s)

MKI8 Measured Temperature (˚C)

MKI.UA87.IPOC.?B2:T_RISETIME (October 2011)

MKI.UA87.IPOC.AB2:T_RISETIME

MKI.UA87.IPOC.BB2:T_RISETIME

MKI.UA87.IPOC.CB2:T_RISETIME

MKI.UA87.IPOC.DB2:T_RISETIME

SIS interlock presently 62 ˚C (originally

55 ˚C ) cannot be further increased !

MKI Temperature of up to 68 ˚C measured

during physics (10 hours time-constant for

ferrite heating and cooling)

Soft-start: OK injection

not Ok wait for MKI cooling

24 stripes under investigation for

installation during TS in August 2012.

M.J. Barnes

Other Issues Related to Injection System


UFOs at MKI several beam dumps at the beginning of the

run (T. Baer’s talk, S07 Thursday )

TDI:

Controls problem TDI in IR2: settings different from measured motor

position and drift of the position reading triggered by electromagnetic noise

Heating Vacuum pressure increase at TDI in IR2 and IR8 when at parking

position (±20 mm) High background in ALICE (shielding?) new

parking position ±55 mm (E. Metral’s talk, S02 Monday)

Beam screen deformation discovered during Xmas Stop.

25 mm 38 mm

TDI IR8

Investigation on cause

and solutions ongoing!

Conclusions


Injection of 144 bunches fully operational: consistent/better than predictions (some mitigations applied: scraping, TCDI at 5s, shielding, injection and abort gap cleaning )

Successful injection of 288 bunches for both beams with losses at ~ 30% from thresholds promising (further mitigations available: new TCDI locations, BLM sunglasses)

TL stability caused a lot of problems:

Steering: tradeoff between minimum transverse losses (beam position at TCDI) and injection oscillations time consuming

big shot-by-shot (MSE) and bunch-by-bunch (MKE) variations

How to improve TL stability for 2012 operation:

Reduce MSE ripple and optimise delay beam/MKE

Dedicated time for setup during commissioning

Clearer references in IQC

Injection Failures:

MKI flashover and erratics: magnets quenched, experiments affected…

Replaced faulty components, improved diagnostics, safer interlock limits.

MKI failures can occur several times per year (according to specifications) respect safety instruction, time for cooling and reconditioning, interlock limits and experiment off

TDI heating and beam screen deformation: investigation ongoing to find cause and short (before LS1) and long term solutions

injection and lessons for 2012 - espace.cern.ch · 18:03:09 interlocks detected an erratic of the...

Documents