upgrade of the injection kicker system for j-parc main...

UPGRADE OF THE INJECTION KICKER SYSTEM FOR J-PARC MAINRING

T.Sugimoto∗, K.Fan, K.Ishii, H.Matsumoto, KEK, Ibaraki, Japan

Abstract

Four lumped inductance injection kicker magnets for theJ-PARC main ring (MR) produce a kick of 0.1096 T·m witha 1% to 99% rise-time of about 400 nsec. A residual fieldof about 6 % of the flat-top exists at the tail of the pulsedue to an impedance mismatching. The residual field is re-quired to be suppressed less than 1% to reduce injectionlosses. For a higher intensity beam operation, the kickerrise-time of less than 300 nsec is required to inject longerbeam bunches which reduces a space charge effect. Dur-ing the long shutdown in FY2013,140 Ω resistor and 7 nFcapacitor were connected to the thyratron to improve thepost-pulse shape. In addition, an optimization of a capac-itance in the matching circuit was carried out to optimizethe waveform. As the result, the rise-time of 195 nsec andthe residual tail field of 2 % were achieved. However, an-other reflection peak of about 9 % is appeared. We plan tocompensate the effect of the new peak by using a new smallkicker magnet. This paper discusses the detail of the circuitand the beam test results.

INTRODUCTION

J-PARC (Japan Proton Accelerator Complex) consists ofthree accelerators, a 400-MeV Linac, a 3-GeV Rapid CycleSynchrotron (RCS) and a 30-GeV Main Ring (MR). TheMR provides a high intensity proton beam to the long base-line neutrino experiment (T2K) and the hadron experiments.During 2013, 220 kW beam has been provided to the neu-trino experiment [1]. For the higher precision measurementof the neutrino oscillation and the observation of the CP vi-olation of the neutrino, a higher power beam operation isrequired. Two scenarios are being considered; one is to in-crease the beam current, the other is to increase a repetitionrate of the MR. For former, it is necessary to use 2nd har-monic RF cavities to mitigate the instability due to the spacecharge effect [2]. However, there is not enough margin fora rise-time of a injection kicker pulse of the MR to modifythe longitudinal bunch shape. In addition, it is importantto reduce a beam loss which introduces a radioactivation ofthe instruments installed in the accelerator tunnel. In 2013,average injection loss was approximately 200 W. A majorsource of a loss during injection period is a coherent oscil-lation of the circulated beam deflected by a residual fieldof the injection kicker [3]. Therefore, an upgrade of the in-jection kicker is one of the key issues to realize the higherintensity operation.

∗ [email protected]

MR INJECTION KICKER SYSTEMFigure 1 shows the injection section of the MR. Proton

bunches are extracted from RCS, and transported via a beamtransport line (3-50BT). Two types of septum magnets andfour injection kicker magnets are equipped to introduce pro-ton bunches to the circular orbit.

septum-2

septum-1

3-50BT

kickerQD QFQFMR

Figure 1: Schematic diagram of the injection section of theJ-PARC MR.

The kicker magnets consist of four identical windowframe lumped inductance kickers installed in a vacuumchamber. Each kicker is driven by two independent powersupplies to realize a short rise-time. A Ni-Zn ferrite core(CMD10 provided from Ceramic Magnetics Inc.) was cho-sen as a return yoke because of the high saturation flux den-sity and the high Curie temperature.

The equivalent circuit of the kicker system is shown inFig. 2. An inductance of each magnets was designed as600 nH. Including a stray inductance of a copper plate con-nected between the high voltage feed-through and the coil,however, the total inductance of the magnet (L1) was esti-mated as about 1 µH. To match with the transmission cablewith a characteristic impedance of 9.7Ωmeasured by a timedomain reflectometry method, a capacitance (C1) shouldbe10.6 nF which results in a large rise-time (it is estimatedas more than 500 nsec using a SPICE software). Further-more, this mismatching also induces a reflection peak. Theoptimized values for the circuit are indicated in Fig. 2.

Non-inductive ceramic resistors and ceramic capacitorswere installed in individual boxes filled with nitrogen. Theywere mounted on the vacuum chambers and connected tothe coils for the impedance matching. The kicker systemhas been operated since December 2011 [4]. However, se-vere discharge problems of the ceramic resistors were foundon March 2012. We have overcome the problem by apply-ing a brazing technology to attach electrodes to the end ofceramic resistor [5].

A waveform of the excitation current measured by a cur-rent transformer is shown in Fig. 3. Four injection timingsignals (K1-4, 40 msec interval) are synchronized to the ex-traction timing of the RCS. For each timing, an excitationcurrent is transmitted to the magnet to deflect two protonbunches into the circular orbit. Nominal deflection angleis 8.5 mrad for 2.7 kA excitation current. The measured

5th International Particle Accelerator Conference IPAC2014, Dresden, Germany JACoW PublishingISBN: 978-3-95450-132-8 doi:10.18429/JACoW-IPAC2014-MOPME069

MOPME069526

Cont

entf

rom

this

wor

km

aybe

used

unde

rthe

term

soft

heCC

BY3.

0lic

ence

(©20

14).

Any

distr

ibut

ion

ofth

isw

ork

mus

tmai

ntai

nat

tribu

tion

toth

eau

thor

(s),

title

ofth

ew

ork,

publ

isher

,and

DO

I.

07 Accelerator Technology Main SystemsT16 Pulsed Power Technology

ChargingPower Supply

Pulse Forming Line(~140m)

Thyratron(e2V CX1193)

Transmission Line(~150m)

L1=1μH R1=9.3Ω

C1=7.1nF R2=9.8ΩZ0=9.7Ω Z0=9.7Ω

R=9.7Ω

Figure 2: Equivalent circuit of the injection kicker system.

rise-time is approximately 400 nsec (from 1% to 99%). Cir-cumference of the MR is 1567.5 m and the revolutionperiodis 5.38 µs before acceleration. Interval of each bunches is597.7 nsec, and the bunch length is about 150 nsec with us-ing fundamental harmonic RF cavities (h=9). The rise-timeis required to be about 200 nsec because the longitudinalbunch length will be increased up to 400 nsec by adding a2nd harmonic RF cavity to reduce the space charge effect.

As shown in the waveform, there is a tail of about 5 %for 1 µs duration and a reflection peak of about 6 % at 3.5µsec. Due to this residual field, some of the circulatingbunches were kicked additionally (especially No.1 bunch atK3 timing, No.1-3 bunches at K4 timing). This extra-kickbecomes a source of a transverse coherent oscillation whichinduces an emittance blow-up and a beam loss during the in-jection period. To damp the oscillation, two kinds of beamfeed-back system have been equipped [3] [6]. However, theinjection loss is still dominant for the total amount of beamloss for the MR.

-1 0 1 2 3 4 5 6

0

0.2

0.4

0.6

0.8

2 2.5 3 3.5 4 4.5 5 5.5 6-0.01

0

0.01

0.02

0.03

0.04

0.05

0.06

0.07

tail reflection peak

time (μs)

norm

aliz

ed e

xcita

tion

curr

ent

K1 (t=0ms)

K2 (t=40ms)

K3 (t=80ms)

K4 (t=120ms)

1 2

3 45 6

7 8(empty bucket)

5.38μs = revolution period597.7ns

150ns → >300ns

450ns→ <300ns

1400nsec

(injection timing)

1% 99%

Figure 3: A measured waveform of the excitation current ofthe injection kicker and the position of injected bunches foreach injection timings (K1-4).

CIRCUIT MODIFICATION

During a long shutdown in 2013, two additional circuitsshown in Fig. 4 (indicated as dotted rectangles) were at-tached to the thyratron housing. Furthermore, the capaci-tance of C1 was optimized to adjust the rise-time and thereflection peak.

ChargingPower Supply

PFLThyratron

Trans. Line

Speed-upCircuit(SUC)

Tail MatcingCircuit(TMC)

L1=1μH R1=9.3Ω

C1=7.1nF R2=9.8Ω

R=9.7Ω

R3C2

Figure 4: Upgraded circuit.

(1) Tail Matching Circuit (TMC)

According to the SPICE simulation study, the pulse tailwas caused by a impedance mismatching due to a conduc-tion loss of the thyratron and a resistive component of thethyratron housing and the high voltage connectors. To im-prove the mismatching, a resistor (R3) was attached to thecathode of the thyratron. Figure 5 shows a measured wave-form with respect to the various R3. As shown in the figure,the pulse tail was reduced less than 1% successfully withR3<140 Ω. However, the flat-top value of the pulse wasalso decreased. Considering the trade-off between the tailreduction and the flat-top drop, R3=140 Ω was chosen.

time (μs)

norm

aliz

ed e

xcita

tion

curr

ent

1.5 2 2.5 3 3.5 4 4.5-0.02

0

0.02

0.04

0.06

0.08

0.1

1%

(beam bucket)

(tail part)

no resistor

R3 = 300 Ω

R3 = 140 Ω

R3 = 100 Ω

R3 = 75 Ω

R3 = 50 Ω

Figure 5: Measured waveform of kicker excitation currentfor various resistors (tail part of the pulse is expanded.).

(2) Speed-up Circuit (SUC)

Reverse current flows into a capacitor attached to the an-ode of thyratron at the leading and trailing edge of the pulse.This current also flows into the magnet and then makes anovershoot waveform which is effective to improve the rise-time. The upper two figures of Fig. 6 show the waveformsmeasured with a SUC capacitor (C2). The leading edge andthe reflection peak were changed by scanning C1 with fixedC2=7 nF. The lower figure of Fig. 6 shows the rise-time andthe fraction current of the reflection peak as a function of C1.Optimization of C1 was carried out to control the rise-timeand the reflection peak. As the results, the rise-time of 195nsec with the original peak fraction of 6 % were achievedwith C1=6.5 nF. While the requirement of the rise-time wassatisfied with this condition, another reflection peak of 9 %which corresponds to a reflection at C2 was appeared beforethe original reflection peak.



MOPME069527

Cont

entf

rom

this

wor

km

aybe

used

unde

rthe

term

soft

heCC

BY3.

0lic

ence

(©20

14).

Any

distr

ibut

ion

ofth

isw

ork

mus

tmai

ntai

nat

tribu

tion

toth

eau

thor

(s),

title

ofth

ew

ork,

publ

isher

,and

DO

I.

2.8 3 3.2 3.4 3.6 3.8-0.02

0

0.02

0.04

0.06

0.08

0.1

capacitance of C1 (nF)0 2 4 6 8 10

rise

time

(1-9

9%, n

sec)

0

100

200

300

400

500

600

norm

ariz

ed c

urre

nt o

f 2nd

ref

lect

ion

peak

0

0.05

0.1

0.15

0.2

0.25

0.3

0.15 0.2 0.25 0.3 0.35 0.4 0.45 0.50.9

0.92

0.94

0.96

0.98

1

1.02

1.04

1.06

1.08 w/o SUC (C1=3.5nF) w/ SUC (C1=3.5nF)

w/ SUC (C1=4.4nF) w/ SUC (C1=5.7nF)

w/ SUC (C1=6.5nF) w/ SUC (C1=7.0nF) w/ SUC (C1=7.2nF)

w/ SUC (C1=8.0nF) w/ SUC (C1=9.5nF)

w/o SUC

w/ SUC

norm

aliz

ed c

urre

nt

time (μs)time (μs)

norm

aliz

ed c

urre

nt

99%

(leading part) (tail part)

(new peak)(original peak)

C2=7nF (fixed)

(beam bucket)

Figure 6: Measured waveform of kicker excitation current(upper) and measured rise-time of the kicker pulse as a func-tion of capacitance in matching circuit (lower).

BEAM TEST RESULTIn Fig. 7, three kicker waveforms are summarized. Al-

though the tail was successfully reduced by R2, the new re-flection peak which is bigger than the original one was ap-peared. To estimate the effect of the modifications, a beamintensity were measured by a DCCT equipped in the MR.As shown in Fig. 8, a part of the beam was lost at K3 andK4 injection timing due to the injection oscillation. Whilethe beam loss at K4 which corresponds to No.1 bunch de-flected by the tail was improved after attaching R2 (red line),it was increased after attaching C2 because No.2 bunch wasdeflected by the new reflection peak (blue line). The beamloss at K3 timing was almost unchanged after the modifica-tion.

(beam bucket)

1.5 2 2.5 3 3.5 4 4.5-0.02

0

0.02

0.04

0.06

0.08

0.1

time (μs)

norm

aliz

ed e

xcita

tion

curr

ent before upgrade (May 2013, C1=7.1nF)

after upgrade (TMC, C1=3.5nF)

after upgrade (TMC + SUC, C1=6.5nF)

1%

(new peak)

(original peak)

(tail part)

Figure 7: Tail part of kicker waveform.

To suppress the injection oscillation, two beam feedbacksystems (bunch-by-bunch and intra-bunch feedback) havebeen equipped in the MR [3] [6]. In 2014, a performance ofthe intra-bunch feedback system (Intra-B FB) was measuredby using the high intensity proton beam (220 kW equiva-lent). The result listed in Table 1 indicates that the feedbacksystem suppressed the coherent oscillation effectively evenif the additional oscillation was induced by the new reflec-

if the additional oscillation was induced by the new reflec-tion peak. Furthermore, a new kicker magnet is planed tobe installed in the injection section of the MR at about 80m downstream from the injection kicker magnet to compen-sate the extra-kick of the reflection peaks [7].

0 50 100 150time from K1 [msec]

Bea

m In

tens

ity [a

.u.]

K1 K2 K3 K4

beam loss

before upgrade (May 2013)after upgrade (TMC)after upgrade (TMC + SUC)

Injected bunch: #1 and #2Measured Device: DCCTBunch-by-bunch FB: ONIntra-bunch FB: OFF

No.1 bunch kicked by tail

No.2 bunch is kicked by new reflection peak

No.1 bunch is kicked by original reflection peak

Figure 8: Beam intensity measured by DCCT.

Table 1: Relative value of injection beam loss for high in-tensity beam test.

Intra-B FB TMC+SUC no circuitON 0.6 0.5OFF 1.3 1

CONCLUSIONIntroducing two additional circuits and optimizing the ca-

pacitance of the matching capacitor, the pulse waveform forthe injection kicker was improved. While the rise-time of195 nsec was achieved, new reflection peak which inducesthe additional injection beam loss was appeared. A newsmall kicker will be developed to compensate the single kickby the residual field for the higher intensity operation.

REFERENCES[1] T.Koseki et. al., “Present Status of J-PARC - after the

Shutdown due to the Radioactive Material Leak Accident”,these proceedings, THPME061, IPAC’14, Dresden, Germany(2014).

[2] Y.Sato et. al., ”Recent Commissioning of High-intensity Pro-ton Beams in J-PARC Main Ring”, Proc. HB2012, Beijin,China, (2012).

[3] Y.Kurimoto et. al., “The Bunch by Bunch Feedback System inJ-PARC Main Ring”, Proc. DIPAC2011, Humburg, Germany,(2011).

[4] K.Fan et.al., “Design and test of Injection Kickers for J-PARCMain Ring”, Proc. IPAC12, New Orleans, USA, (2012).

[5] T.Sugimoto et.al., “Development of a Non-Inductive ceramicresistor”, Proc. IPAC13, Shanghai, China, (2013).

[6] K.Nakamura et.al., “Transverse Intra-bunch Feedback in theJ-PARC MR”, these proceedings, THOAA03, IPAC’14, Dres-den, Germany (2014).

[7] K.Fan et.al., “New Design of J-PARC Main Ring InjectionSystem For High Beam Power Operation”,these proceedings,WEPRO065, IPAC’14, Dresden, Germany (2014).


MOPME069528

Cont

entf

rom

this

wor

km

aybe

used

unde

rthe

term

soft

heCC

BY3.

0lic

ence

(©20

14).

Any

distr

ibut

ion

ofth

isw

ork

mus

tmai

ntai

nat

tribu

tion

toth

eau

thor

(s),

title

ofth

ew

ork,

publ

isher

,and

DO

I.


upgrade of the injection kicker system for j-parc main...

Documents