MKP-I Kicker System Design and Feasibility

MKP-I Kicker System Design and Feasibility

M. Barnes, L. Ducimetiere, T. Kramer , L. Sermeus

E. Carlier, B. Goddard, W. Höfle, R. Noulibos, G. Kotzian, B. Salvant, J. Uythoven, F. Velotti, C. Zannini

LIU-SPS 50 ns Injection System for Pb Ions Review, October 4TH 2013

LIU-SPS 50 ns Injection System for Pb Ions Review 2

Outline• Overview of present system• Design requirements• Rise time studies • Circuit and magnet design • Overview of variants studied • Costs • Conclusions


Present Installation

Photos: J.Uythoven

MKP-IMKPS MKPSMKPS MKPL

(2-98%) 145

4LIU-SPS 50 ns Injection System for Pb Ions Review


RP Survey 2013 • Current injection

kicker are immediately after the internal dump

• To be considered when discussing modifications.


MKP-I Initial Design Requirements

Magnet type Open “C”

Rise time 50 ns (10%-90%)

Fall time up to 6 µs

Flat top 160 ns

Flat top ripple +/- 1%

B*r ions (p+) 60 (90) T.m

Nominal system kick (100%)

1.333 mrad

Aperture (h/v) 100 mm / 44 mm

Total physical length 4.25 m

hv


“Our” Constraints & Assumptions• Would like to avoid SF6 gas filled cables.

Voltage below 40 kV. • Would like to avoid costs of moving the generators

into a gallery close to the magnets. 200 m long Tx cables.

• Magnet cell length should not be much shorter than 28mm to ease assembly.

• Reuse of existing designs as far as possible.• Short flat top will allow for PFL. • It was suggested to start the design with a low

number of cells per magnet.


Verification of Initial Design Parameters: Rise Time

50ns 100nsCirculating bunch

injected bunches

45ns

Waveform shown for 50, 5 cell magnet

50 ns (10-90%) feasible


Rise Time Definition: 50ns (10% to 90%)

5 ns

~2.7mT vs. ~4.7mT∆=2mT ~ 50%

Head & tail of the bunch will see different B-field:


5 ns

32.0mT vs. 34.2mT ∆=2.2mT ~6.6%

Rise time jitter 5 ns (Thyratron + Trigger electronics)

?

+/- 1% Ripple


Rise Time • Assumed 50 ns (10% to 90%) from preliminary studies

could just be met however:• Considering a 5ns bucket length a 50 ns bunch spacing means

in fact 45 ns rise time;• Considering a 5 ns jitter will already require a 40 ns rise time.• Different kick on head/tail to be studied (cannot be damped by

the transverse damper!) (see presentation from Gerd) • Transverse damper cannot damp a kick amplitude mismatch of

10% without substantial emittance blow-up (see presentation from Gerd)

• Actual rise time (10-90%) would need to be in the range of <30 ns

• New definition required!• Relaxation to 75ns but 2% to 98% and 2% ripple?


• Considered longitudinal separation of modules and tanks.

⟹ Total physical length increases compared to previous drafts.

• Terminated 50 Ω magnet does not deliver the required kick strength per meter for 4.25 m total length and 40 kV.

⟹ Decision to go for system in short circuited mode. Advantage: no regular exchange/inspection of termination resistor -> less dose)

Early Drafts

Parameters for 50/25 Ω SC Impedance [Ω] 50 50 25 25Voltage [kV] 40 65 40 40Current [A] 800 1300 1600 1600vertical gap [m] 0.044 0.044 0.045 0.045Total physical length [m] 6.21 4.13 4.71 4.5

Total magnetic length [m] 3.7 2.3 1.8 1.8Number of tanks 3 2 2 2Tank length [m] 1.92 1.84 2.13 2.025Number of Magnets 11 7 10 8Magnet length [m] 0.333 0.333 0.18 0.225

Number of magnet cells 11 11 5 5

Magnet cell length [m] 0.0303 0.0303 0.036 0.045Magnet filling time [ns] 40 40 44 53Magnet Inductance [nH] 999 999 540 660

25Ω Magnet to avoid SF6 cables?


Filling time (25 Ω): Shorter magnets needed to compensate for lower impedance

Can gain 21cm by having longer (&less) magnets.


Circuit and Rise time Analysis • Modified existing Booster EK PSpice Model* • 2 magnets per PFL

* very detailed EK system model done by L. Sermeus


PSpice Model of Magnet Module• Modified from existing* Booster EK-magnet model:

* very detailed EK magnet model done by L. Sermeus


Results 50Ω SCImpedance Z0 50 ΩVoltage U 42 kVCurrent I 842 AVertical gap 0.044 mTotal physical length 5.73 mTotal magnetic length 3.3 mNumber of tanks 3Tank length 1.76 mNumber of Magnets 10Magnet length 0.333 mNumber of magnet cells 5Magnet cell length 0.0666 mMagnet filling time 40 nsMagnet Inductance 999 nHInductance per cell 199.7 nH

50 ns (10-90%)62 ns (5%-95%)72 ns (2%-98%)Ripple within ±2%

2 magnets per PFL (25Ω thus 2 x RG220)


Results 25Ω SC

Impedance Z0 25 25 ΩVoltage U 40 40 kVCurrent I 1600 1600 Avertical gap 0.045 0.045 mTotal physical length 4.71 4.5 mTotal magnetic length 1.8 1.8 mNumber of tanks 2 2Tank length 2.13 2.025 mNumber of Magnets 10 8

Magnet length 0.18 0.225 m

Number of magnet cells 5 5Magnet cell length 0.036 0.045 mCapacitance per cell 173 211 pFInductance per cell 108 132 nHMagnet filling time 43.2 52.8 nsMagnet Inductance 540 660 nH

61.5 ns (10-90%)70 ns (5%-95%)80 ns (2%-98%)Ripple within ±2%

55.2 ns (10-90%)65 ns (5%-95%)74 ns (2%-98%)Ripple within ±2%

2 magnets per PFL (12.5Ω thus 4 x RG220)


Magnet Design

5-cell magnet module assemblyHV-plate

Ferrite

Earth-plate

Magnet Cell

Magnet Module


Module / Tank Separation

Distance between magnets 0.15 m Distance before 1st mag. 0.15 m Distance following last mag. 0.15 m Distance between tanks 0.15 m Distance before 1st & after last tank

0.15 m

Max. Tank length 2.5 m

MKP: 139 (272.3) mm

MKP: 140 (229) mm

MKP: 137 mm

Ferrite cross section

Ferrite

HV- Conductor

Return- Conductor

45 105

40

40

44

125

dumped p+ beam

Inj. Beam

50

Shielding box (not necessary for dumped beam)

xx

yy


Yoke Flux Density

Flux in back leg ~110 mT Ferrite: 45 mm (MKPS: 60mm) (Cross section can be made smaller during detailed design.)

I=1600 A


Good field region • ±0.5%• Horizontally: -40 to +30 mm• Vertically: +/- 17 mm

5 sigma beam envelopes*:FT: 75.73 mmLHC: 32.72 mm Ions: 24.41 mm

* Courtesy: F.Velotti

Static simulation only, no end effects; (first guess). By(0/0)= 0.0414 T


HV/Earth-Plates Total cell length 0.036mHV-conductor thickness 0.005mHV-conductor height 0.044mHV-plate thickness 0.005mEarthplate thickness 0.008mEarth-conductor thickness 0.005mEarth-conductor height 0.04mFerrite Thickness (long.) 0.031mFerrite top/bottom leg height 0.04mFerrite back leg thickness 0.045mSpace between HV- and E-plate 0.0115mCalculated Cell Capacitance (without end effects) 172.7pFplate area: 0.224m²active plate length (square) 0.474mHV plate length 0.634mHV plate height 0.474mhor. separation distance 0.010mEarth plate length 0.484mEarth plate height 0.474m

Very similar sizes to current MKP magnets


Switch

CX1175

Peak forward/inverse anode voltage

70kV

Peak forward anode current

10kA

Rate of rise of anode current

100kA/µs (10-90%)

Jitter Typical 1ns, max 5ns

Possible choice, (variant used for PSB EK):


Cable• RG220U• Up to 40 kV • Rather cheap

(~50 CHF/m)

• Other option: ~25 Ω, 80kV, SF6 gas filled cable • considered to be purchased for PS, however very difficult to obtain. • Considered to be much more expensive, detailed costs currently

unknown. • Would need whole infrastructure for handling the gas as well.• Gas is definitely not halogen free -> IS23


Beam Impedance??• See presentation from Carlo and Benoit for

details. • It is already a known possible issue for the

present MKP magnets and HL beams.• Additional magnets (similar construction) will

definitely not improve the situation.• Thus a new magnet design should take any

possibility for improvement into account.• Without having the detailed data an enlarged

vertical aperture was already studied.


Additional Options Studied

a) Enlarged Apertureb) MKPS+PFL


Enlarged Vertical Aperture• To improve beam impedance issues and to

provide aperture for a possible beam screen• 56 mm (provides 6mm on each side)• Several ideas:

• Wires. • Ceramic chamber/plates with coating. • Ceramic plates with stripes.

Lets look into the magnet design first:

Ferrite

HV- Conductor

Earth- Conductor

45105

56136 Inj. Beam

50

44 48

4

757 7 10

Silver painted stripes?

30 nm Ti- coating?

Ceramic plate


25Ω SC Enlarged Aperture (I)Impedance Z0 25 ΩVoltage U 40 kVCurrent I 1600 Avertical gap 0.056 mTotal physical length 5.25 mTotal magnetic length 2.25 mNumber of tanks 2Tank length 2.4 mNumber of Magnets 10Magnet length 0.225 mNumber of magnet cells 5Magnet cell length 0.045 mCapacitance per cell 177.72 pFInductance per cell 111.08 nHMagnet filling time 44.43 nsMagnet Inductance 555.4 nH

Larger aperture (less field) has to be compensated by magnetic length.

Not such a big change as h/v ratio is beneficial.


25Ω SC Enlarged Aperture (II)• 56ns (10-90%)• 65 ns (5%-95%)• 74.9 ns (2%-98%)• Ripple within ±2%

Without any beam screen


Ferrite Yoke• BY = 35.5 mT (Opera) vs. 35.9 mT (calculation) • Back leg: ~100 mT • L’=2.65 µH/m

Static simulation only (first guess).


Beam screen

92 94 96 98 100 102 104 106 108 110 112 114 116 118 120 122 124 126 128 130 132 134 136 138 1400.95

0.955

0.96

0.965

0.97

0.975

0.98

0.985

0.99

0.995

1

d= 10nm, Bd= 30nm, Bd=100nm, Bcurrent

Time [ns]

No

rmal

ized

fie

ld

• Is it really necessary? • If yes, it will increase the field rise-time. • Several ideas - ceramic plate /chamber with coating/stripes seems easiest.• Once decided, detailed transient simulations need to be done.

Courtesy: T. Stadlbauer

For indication only:

Studies done for a 100ns kicker system (Ti-coated racetrack chamber) show substantial field delays above 30 nm Ti-coating thickness


What about connecting a PFL to the existing kickers?

• Connect present MKPS kicker to PFL as planned in 2001 (115ns).

• Can’t do the 50ns nor 75ns but possibly 100ns (2-98%).

• PFL is rather simple. • Connections to present system / switches is a

bit tricky/costly.


Theoretical parameters MKPS+PFL

Impedance Z0 16.67 Ω

Termination terminatedVoltage U 40 kVCurrent I 1440 Avertical gap 0.061 mKick angle (12 magnets) 2.6 mradMagnet length 0.542 mNo. of magnet cells 17Capacitance per cell 236.4 pFInductance per cell 65.7 nHMagnet filling time 67 nsMagnet Inductance 1116.6 nH

• 10%-90%: 62ns • 5%-95%: 76ns• 2%-98%: 100ns• However undershoot of ~5% (to be optimized)

Will not work without additional angle from septa (MSI-V) or auxiliary kicker.


MKPS and PFL • Magnets and Tx Cables already in place. • 6 PFL per magnet pair + switch and filter circuits needed. • PFL rather simple, connection box and switches is more challenging and

needs to be studied in more detail.


Overview (I)Option Advantage Disadvantage

50 Ω SC / 65 kV • Within system length specification (4.5m).• 50ns (10-90%) can be met.

• New magnets• Expensive SF6 cables needed

50 Ω SC / 40 kV • 50ns (10-90%) can be met.• Using conventional cables.

• New magnets• System longer than specified.• Possible beam blow up as damping might not

be efficient. • Highest risk of building a new expensive

system without getting the envisaged overall benefit in the LHC.

25 Ω SC / 40 kV • Using conventional cables.• System length specification can be met.

• New magnets• 50ns (10-90%) cannot be met (61ns).

25 Ω SC / 40 kV Enlarged aperture

• Using conventional cables.• Advantageous for beam impedance / future

measures

• New magnets• Increased magnetic length• 50ns (10-90%) cannot be met (56ns)• Possibility to improve rise-time if additional

length is approved (more but shorter magnets).

Current MKPS (16.67Ω term.) + PFL

• Cheapest solution studied• No new magnets • No dose taken by personnel for installation

AND maintenance (dump region) • No additional SPS impedance • lower risk as tests with spare MKPS can be

organized.

• Only 100ns (2-98%). • Switch between systems needs to be studied. • Operation is more complicated. • Limited to max. 40 kV


Overview (II)

Impedance Z0 [Ω] 50 50 25 25 25 16.67Voltage U [kV] 65 42.1 40 40 40 40 Current I [A] 1300 842 1600 1600 1600 1440 Vertical gap [m] 0.044 0.044 0.044 0.045 0.056 0.061Total physical length [m] 4.131 5.73 4.71 4.5 5.25Total magnetic length [m] 2.331 3.33 1.8 1.8 2.25Number of tanks 2 3 2 2 2 3Tank length [m] 1.84 1.76 2.13 2.025 2.4Number of Magnets 7 10 10 8 10 12Magnet length [m] 0.333 0.333 0.18 0.225 0.225 0.542Number of magnet cells 11 5 5 5 5 17Magnet cell length [m] 0.030 0.067 0.036 0.045 0.045Magnet filling time [ns] 39.9 39.9 43.2 52.8 44.4Magnet Inductance [nH] 998.6 998.6 539.8 659.7 555.4Inductance per cell [nH] 90.78 199.7 108.0 131.9 111.1 65.7Capacitance per cell [pF] 36.31 79.9 172.7 211.1 177.7 236.4Rise time 10%-90% [ns] 46 50 55.2 61.5 56 62Rise time 5%-95% [ns] 55 62 65 70 65.0 76Rise time 2%-98% [ns] 66 72 74 80 74.9 100Ripple [%] < 2 <2 <2 <2 <2 <5HV-conductor thickness [m] 0.005 0.005 0.005 0.005 0.005HV-conductor height [m] 0.044 0.044 0.044 0.044 0.056HV-plate thickness [m] 0.005 0.005 0.005 0.005 0.005Earthplate thickness [m] 0.008 0.008 0.008 0.008 0.008Earth-conductor thickness [m] 0.005 0.005 0.005 0.005 0.005Earth-conductor height [m] 0.04 0.04 0.04 0.04 0.04Ferrite Thickness (long.) [m] 0.0253 0.0616 0.031 0.04 0.04Ferrite top/bottom leg height [m] 0.04 0.04 0.04 0.04 0.04Ferrite back leg thickness [m] 0.045 0.045 0.045 0.045 0.045Space between HV- and E-plate [m] 0.0086 0.0268 0.0115 0.016 0.016Plate area [m²] 0.035 0.242 0.224 0.381 0.321Active plate length (square) [m] 0.188 0.492 0.474 0.618 0.567HV plate length [m] 0.348 0.652 0.634 0.778 0.727HV plate height [m] 0.188 0.492 0.474 0.618 0.567Earth plate length [m] 0.198 0.502 0.484 0.628 0.577Earth plate height [m] 0.188 0.492 0.474 0.618 0.567


Cost Estimates

2012 Estimates 50Ω SC /65kV (SF6) 50Ω/25Ω SC /40kV MKPS+PFL

50 ns injection kicker kCHF FTE kCHF FTE kCHF FTE kCHF FTE

2 vacuum tanks 200 0.6 220 0.66 300 0.9

8 kicker magnets 1600 3.2 1600 3.2 2000 4.0

8 PFLs 400 0.8 600 1 200 0.8 300 1

8 switches 800 2.4 800 2.4 500 2.0 600 2

8 Dump Switches 800 1.2 800 1.2 1000 1.5

8 Terminating Dump Resistors 400 1.2 400 1.2 500 1.5

8 Resonant Charging System 400 0.8 200 0.8 250 0.8 300 0.8

HV transmission line 320 0.8 700 1.0 200 0.8

BA1 extension (100m2, 4kCHF/m2) 400 0.6 600 0.6 400 0.6 100 0.5

Oil system 100 0.6 100 0.6 100 0.6 100 0.6

Gas system 150 0.5

Cabling 400 0.8 500 0.8 400 0.8 100 0.2

Vacuum (incl. sector valves) 100 0.2 100 0.2 100 0.2

Slow controls 200 0.4 200 0.4 200 0.4 150 0.4

Fast controls 200 0.4 200 0.4 200 0.4 150 0.4

Interlocking 100 0.2 100 0.2 100 0.2 75 0.2

Total 6420 14.2 7240 15.2 6450 15.5 1875* 6.1

*Switch between PFN/PFL not consideredDump switch & TDR suppressed!

41

Conclusions• New kicker system in principal feasible with the constraints given

(50ns 10-90%) however not practicable in the global view (blow-up). Thus:

• Redefine rise-time & ripple. • Increase total system length (from 4.5 to ??) to stay with conventional

cables.

• Enlarged aperture (56mm) feasible. • For relaxed rise time a 25Ω SC system is preferred (shorter, no

TMR). • In case a beam screen is needed detailed studies have to be done

to determine the reduction in rise time. • Present MKPS could be upgraded to 100 ns (2-98%).

• 40kV max (~2.6 mrad).• No extra SPS impedance. • No extra radiation dose.

LIU-SPS 50 ns Injection System for Pb Ions Review


ReferencesG. Schröder, Fast Pulsed Magnet Systems, CERN-SL-98-017 BT

M. Barnes, Beam Transfer Devices: Septa & Kickers, proc. CAS 2009

B. Goddard, et. al., A new lead iopn injection system for the CERN SPS with 50ns rise time, proc. IPAC2013

L. Ducimetière, et. al., Upgrading the SPS injection kicker system for LHC requirements.

J. Uythoven, The new SPS injection channel, proc. Chamonix IX, p.120

J. Uythoven, et. al., The future of the SPS injection channel, SL-Note-99-023 SLI

M. Barnes,T. Stadlbauer, Determination of coating thickness for the MedAustron kicker magnets vacuum chambers, ES-111108-a-TST, Geneva, 2011

Documents

MKP-I Kicker System Design and Feasibility