MKI Erratic: Beam Related Aspects

ABT/BTP-FPS-EC

Injection Scheme

Injected batch

Circulating LHC beam

MKI

TDI

Miss-kicked

Injected batch

- Kicked Circ. beam

- Over-kicked inj. batch

grazing

grazing

TDI Grazing with Injected Beam Nominal MKI kick = 0.85 mrad TDI grazing: 86% kicker strength (14% missing), 6s beam

envelope

Injected beam

Beam 1

MKI TDI

TC

DD

D1MQX

IP2

MQX

MQM

D2

TDI Grazing with Circulating Beam TDI grazing: 14% kicker strength, 6s beam envelope

Beam 1

MKI TDI

TC

DD

D1MQX

IP2

MQX

MQM

D2

Circulating beam

MKI Waveform

Rise time: 0.9 ms Fall time: 3 ms

Fall time injected Beam:

~ 90-45 bunches at TDI (~2 ms green) ~ 30-15 bunches escaping (~1 ms yellow)

Grazing

Rise time 0.9 ms 36-18 bunches (25-50 ns bunch spacing) less critical

Fall time 3 ms 120-60 bunches (25-50 ns bunch spacing) more critical

MKI Waveform

Rise time: 0.9 ms Fall time: 3 ms

Fall time circulating Beam: ~ 90-45 bunches at TDI (~2 ms yellow)~ 30-15 bunches escaping (~1 ms green)

Grazing

Rise time 0.9 ms 36-18 bunches (25-50 ns bunch spacing) less critical

Fall time 3 ms 120-60 bunches (25-50 ns bunch spacing) more critical

Number of Grazing Bunches

Nominal MKI flattop of 7.86 ms:

Full injected batch: 144/288 (for 25/50 ns bunch

spacing) kicked with 80-90% nominal MKI strength Worst Injection scenario

Erratic event:144/ 288 (for 25/50 ns bunch spacing) circulating bunches kicked with 10-20% nominal MKI strength.

Longer flattop more bunches can graze the TDI!!

Two erratic events with beam happened on July 28th

First Erratic: 28/7/2011 at 16:30:43

Normal (triggered) turn-on

Erratic (untriggered) turn-on of MKI2 MS3

Erratic (un-triggered) turn-on of MKI2 MS3 at full PFN voltage;

Interlocks detected erratic. Control (machine protection) philosophy is to trigger all MS and DS of system (within a delay of 1µs). Hence all 4 kicker magnets pulsed for up to 4.5µs.

Circulating beam was not in IP2 and therefore not disturbed.

Batch was extracted from SPS but saw no kick at MKI (already at zero current after the pulse) and went straight into the TDI.

M. Barnes

Loss PatternUpstream of IP2

Beam 1

Downstream of IP2

Beam 1

Very clean, TDI stopping the injected beam, no magnet quenched!

TDI

Well known loss pattern during injection

In comparison to flashover event of April 18th in P8 (see next slides), cleaner in arc (no quenches) – TDI stopping the beam + (maybe) retraction of TCLIB in between

Flashover in P8 on April 18th

Injection of 2 × 36 bunches spaced by 2.2 ms Breakdown after ~2 ms All 36b of 2nd batch were kicked with 75-90% 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) 12 magnets quenched

Breakdown after ~2 ms (from kick waveform)

Loss PatternUpstream of IP8

Beam 2

Downstream of IP8

Beam 2

Well known loss pattern during injection

Losses far in sector 7-8, 12 magnets quenched

TDI

Second Erratic: 28/7/2011 at 18:03:09

0

5000

10000

15000

20000

25000

30000

35000

40000

45000

50000

55000A

ppro

x PF

N V

olta

ge

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 (untriggered) turn-on of MKI2 MS3

Erratic (un-triggered) turn-on of MKI2 MS3 occurred during resonant charging – sending current to one of the four kicker magnets;

Interlocks did NOT detect erratic of MS3 (at 33kV): hence no immediate action was taken to turn-on other thyratrons. PFNs discharged via the DS after 4ms (no further magnet current). Interlocking problem to address.

The failure occurred early in the charging process, and the extraction from the SPS was inhibited;

The circulating beam which was swept over the aperture and protection elements (~17% of normal kick) for ~8-9µs 150-190 bunches

M. Barnes

Loss PatternUpstream of IP2

Beam 1

Downstream of IP2

Beam 1

Losses far in sector 2-3, 3 magnets quenched

TDI

Losses starting at TDI, no injection loss signature only circulating beam kicked by MKI

Beam lost: Not Dumped

Some beam miss

ing

Intensity before dump: 517 bunches - 5.81e13 p+ (from XPOC)

Intensity dumped: 344 bunches - 3.64e13 p+(from XPOC)

Beam lost: 173 bunches - 2.15e13 p+ (~186 bunches of 1.15e11 p+) in agreement with 150-190 bunches expected

For operation with 25 ns bunch spacing 300-380 bunches

How Many Protons at TDI?

Radiation monitors and BLM used to estimate the number of protons at the TDI

and (possibly) downstream elements (difficult for MQX since no calibration

factor is available.

First Erratic: Nb

1 =144 bunches at the TDI (upper jaw)

Second Erratic: Nb

2 =?? bunches at the TDI(lower jaw) Simple ratio:

Nb 2 = Nb

1 / 1.29 = 112 b

65% of the “lost” bunches

In agreement with BLM measurements (A. Nordt)

1.29

BLM analysis (A. Nordt)

MB MB MB MB MB MB MB MB MB MB

Number of protons lost at each element from Q6.L8 (Q6.R2 for B2) up to Q16.L8 (Q16.R2 for B2)

Combined results for: Q6 quench margin at injection MD Two erratic MKI events for B1 Flashover event for B2

Highest loss level recorded is indicated (blue: no quench, red: quench)

BLM analysis (A. Nordt)

1 2 3 4 5 60.00E+000

2.00E+001

4.00E+001

6.00E+001

8.00E+001

1.00E+002

1.20E+002

1.40E+002

1.60E+002

Losses on Triplets left and right of IPApril Failure Quench B2

Test No Quench B2

July Failure Quench B1

July Failure No Quench B1

Gy/

sec

IP

Beam

No calibration factor available for triplet magnets (not possible to transform Gy/s in number of protons)

Triplet magnets quench for losses higher than 21 Gy/s (BLM not in saturation)Q3 Q2 Q1 Q1 Q2 Q3

ALICE (from M.Lechman’s talk MPP)

The values of the threshold for the beam background were exceeded: 16:30 - 80 times 18:03 - 3559 times

9 out of 17 sub-detectors were affected during the second erratic event

All subsystems but Silicon Drift Detector (SDD) recovered using standard procedures

SDD calibration system - investigation ongoing

We are close to limits of safety

With actual MKI logic up to a factor of 2 higher intensity

(25 ns bunch spacing) expected in case of MKI erratic!

Related issues (from Rudiger)

Detecting erratic firing and trigger other kickers did not work. Are we sure that it would work for the beam dump kickers? Completely different logic:

MKI only retrigger when a fault detected MKD-MKB retrigger sent out at every dump and checked by IPOC

TDI aperture: Nominal aperture: 6.8 s = 4.8 mm for e =3.5 mm Actual e = 1.5 mm 4.8 mm = 10.4 s

Should TDI be closed to, for example, 3.9 mm = 8.5 ??s

Impedance, aperture, heating effects to be checked!

Summary Injection protection has worked (again) to prevent serious damage:

TDI intercepting miss-kicked beam (65%-100%) High losses on downstream elements in case of grazing magnets quench

ALICE: recorded losses up to a factor ~4000 above thresholds SDD did not recover using standard procedure Very close to safety limit BUT up to a factor of 2 higher losses might occur, in

case of missing retrigger, if MKI Voltage threshold logic unchanged To be understood in more detail why the retriggering did not work MKI new logic: remove voltage threshold retriggering always possible also for

low voltage (next TS), to be tested

Switch changed no more erratic events observed

First estimates of quench limit: MB: 2e9 p+ MQ: 5e9-4e10 p+ MQX: < 21 Gy/s (missing calibration factor)