CLIC RECOMBINATION SCHEME FOR THE LOW ENERGY OPERATION MODE

A. Gerbershagen, D. Schulte, CERN, Geneva, Switzerland P. N. Burrows, Oxford University, Oxford, UK

Abstract The CLIC recombination scheme is a concept of

multiplication of the drive beam frequency in order to generate a 12 GHz RF wave for the main beam acceleration. CLIC is designed to be operated at nominal energy and in low energy modes. The low energy operation modes require the train length to be increased by different factors in order to maintain the same level of luminosity. Also the number of initial trains that are merged to form each final train is changed. The combination scheme must be able to accommodate and recombine both long and short trains for nominal and low energy CLIC operation modes. The recombination hence becomes a non-trivial process and makes the correction of the errors in the drive beam more challenging. The present paper describes in detail the recombination process and its consequences.

INTRODUCTION The Compact Linear Collider is a potential future linear e+e- collider, which is designed to provide acceleration with a gradient of ~100 MV/m. CLIC is supposed to be operated at 12 GHz frequency with 244 ns long RFpulses at nominal energy of 3 TeV [1]. Such pulses cannot be produced by any conventional RF source, but can be extracted from a high-current low energy beam called the drive beam. Drive beam can be accelerated by conventional klystrons at the frequency of 0.5 GHz, later drive beam buckets can be recombined in the so called CLIC drive beam recombination scheme. The buckets of the subsequent trains are repositioned longitudinally between each other, hence increasing the bunch frequency of the beam (see Fig. 1).

Figure 1: CLIC drive beam combination principle.

The frequency is increased by factor two in a delay line and by factors three and four in the two combiner rings, hence giving a drive beam at the final frequency of 12 GHz.

Figure 2: CLIC drive beam accelerator complex and recombination scheme.

DIFFERENT ENERGY MODES Operating at lower energies is necessary to study the

properties of the particles that are expected, or hoped to be discovered, at CLIC. Hence CLIC must allow the possibility of energy scans, which can be performed by reduction of the accelerating gradient. CLIC error tolerances require the bunch charge N to be reduced proportionally to the collision energy E, leading to a significant luminosity drop. In order to compensate for it, it is planned to increase the pulse length and the number of bunches per pulse nb. This pulse length increase can be performed for several stages of energy reduction, which defines the different energy operation modes [2].

The baseline design implies the construction of four different delay lines for different energy modes. The combiner rings, in contrast, will be designed in order to be able to accommodate all possible pulse lengths in the same structure. The length of the combiner rings is 292.8m for CR1 and 439.2m for CR2.

NOMINAL 3 TeV OPERATION MODE At the nominal operation mode CLIC drive beam

pulses with both even and odd buckets are 244 ns long. After the recombination of the buckets in the delay line, the beam consists of 244 ns long 1 GHz pulses and 244 ns long gaps (as shown in Fig. 1). CR1 will be filled with two pulses simultaneously, separated by two gaps (see Fig. 3). When next two pulses arrive, they will be combined with the ones circulating in the ring, up to the combination factor three. Afterwards, the pulses will be extracted from the combiner ring.

TUPPR024 Proceedings of IPAC2012, New Orleans, Louisiana, USA

ISBN 978-3-95450-115-1

1864Cop

yrig

htc ○

2012

byIE

EE

–cc

Cre

ativ

eC

omm

onsA

ttri

butio

n3.

0(C

CB

Y3.

0)—

ccC

reat

ive

Com

mon

sAtt

ribu

tion

3.0

(CC

BY

3.0)

01 Circular and Linear Colliders

A03 Linear Colliders

Fm

Hoci

FPafrt

be

FTwr

Figure 3: Themode – 122 ev

The above However, the of 6×244 ns combiner ringinjected into C

Figure 4: SchPulses arrive fand leave to thfilled with bunrepresents the two or three re

In CR2 pulsseparated by being recombieach cycle (Fig

Figure 5: ComThe factor twwithout recomrecombined pu

e switching paven and 122 od

scenario descrextraction musfor the prop

g CR2. This mCR1 must be ex

heme of first from the left, che right. Each pnches, hence a

combined puespectively.

ses of 3 GHz f5×244 ns longined with the fg. 5).

mplete recombinwo recombine

mbination from ulses are used f

attern of the ndd buckets [3].

ribes the steadst happen in re

per operation means that thextracted before

several circulacirculate clockwpoint represent

doublet or a tlses with com

frequency and g gaps, circulafollowing pulse

nation by DL, d pulse is thCR1, the follo

for main beam

nominal 3 TeV

dy state modeegular intervalof the second

e second pulsit is combined

ations in CR1wise in the ringts pulse buckettriplet of point

mbination facto

244 ns lengthate four timeses at the end o

CR1 and CR2he one ejectedowing factor 24acceleration.

V

e. ls d e

d.

1. g ts ts or

h, s, of

2. d 4

2.25 TeVThe 2.2

244 ns. Hfollowed followed

Figure 6: even and

Figure 7: (top) and

Other MFor 2 T

The switcmode and122 odd afterward

The 1.5the first cpulses insinstead ofcombiner

The swto the 2.25

CPRO

Main BeThe de

deflector Hence, thfrequencybunches idistributedTeV modbuckets be

LOWER E

V Mode 25 TeV mode

Hence, it is planby 81 odd

162 odd bucke

The switchingodd pulses com

Bunch positiofor the 2.25 Te

Modes TeV mode the ching pattern fud implies the a

buckets, theds 62 even and5 TeV mode hacombiner ringstead of two af four [3]. The rings stay the

witching pattern5 TeV mode.

ONSEQUEOPAGATIO

eam Gradienlay line and thfrequency at

hey create dry with some ofinstead of cred bunches. E.

de is not 9 GHeing empty.

ENERGY M

requires the pnned to producbuckets, then

ets and so on (s

g pattern of thme alternating

oning for the neV mode (botto

pulse length mfunctions similaalternation beten 183 even 182 odd.

as the pulse leng accommodatand the secondrecombinationsame.

n for 1.125 TeV

ENCES FORN AND CO

nt Error he combiner rnominal and

rive beam witf the buckets neating a pulseg. the bucket Hz, but 12 G

MODES

pulse length ofce 163 even bun 82 even busee Fig. 6).

he 2.25 TeV min groups [3].

nominal 3 TeV om).

must be 3/2×24arly to the 2.25tween 122 eve

and 61 odd

ngth of 2×244 tes only one od combiner rinn factors provid

V mode is ana

R ERROR ORRECTION

rings have the low energy mth 12 GHz bnot being fillede with equidis

frequency forGHz with 1/4 o

f 4/3 uckets uckets

mode –

mode

44 ns. 5 TeV en and d and

ns, so of the

ng two ded by

alogue

N

same modes. bucket d with stantly r 2.25 of the

Proceedings of IPAC2012, New Orleans, Louisiana, USA TUPPR024

01 Circular and Linear Colliders

A03 Linear Colliders

ISBN 978-3-95450-115-1

1865 Cop

yrig

htc ○

2012

byIE

EE

–cc

Cre

ativ

eC

omm

onsA

ttri

butio

n3.

0(C

CB

Y3.

0)—

ccC

reat

ive

Com

mon

sAtt

ribu

tion

3.0

(CC

BY

3.0)

aeoln

M

bpe

rwtp

if

fcc

fppl

Fd

Fd

e

Because of amplitude wilestimate the reorder of severlimit. Hence necessary.

Main Beam

The drive bbeam acceleratparameters areerror being 0.2

Part of the precombinationway: when tratrains are posiproportion of lshifted to the in the main befilling time in

Important efrequencies, acorresponds tocancel out. Th8: the noise wifrom 10 kHz particular at hipeaks at n×4.1length.

Figure 8: Maidrive beam bun

Figure 9: Maidrive beam RF

In order to senergy case, tstructure has b

that, the resull not be stabesulting main bral per mille, wa further inv

Acceleratorbeam provides tion, hence thee very strict, e.25° at 12 GHz phase error is rn scheme, whicains get recombitioned next tolow frequency higher frequen

eam RF structuthe order of 60exceptions to at which the o the train lene example of sith originally eto 20 MHz isigher frequenc1 MHz, which

in beam RF pnch charge erro

in beam RF pF amplitude err

suppress the rethe length of

been set to 244

ulting main bble. Preliminarbeam energy erwhich is abovvestigation of

r RF Phase Ethe RF powe

e tolerances on g. the toleranc[4]. educed due to ch functions inbined, buncheso each other. noise (up to s

ncies, which arures, since the 0 ns.

this filtering wave length

gth and so thesuch filtering isequal amplitudes significantly cies, with the eh correspond to

phase error caor for 3 TeV m

phase error cauror for 3 TeV m

esonant peaks fthe drive beans. Hence for

beam RF wavry calculation

rror being in thve the tolerancf the errors i

Error er for the mainthe drive beam

ce for the phas

filtering by thn the followings from differenHence, a larg

several MHz) ire later filteredstructure has

are the jitteh of the jittee errors do nos shown in Fige at frequenciefiltered out, in

exception of tho 244 ns puls

aused by 0.1%mode.

used by 0.05%mode.

for the nominaam accelerating

the phase erro

e ns e e is

n m e

e g

nt e is d a

er er ot g. es n e e

%

%

al g

or

caused byresonant p(compare

This filits multiplengths arbe filterestructure the exampns long puwhich are2.0 MHz,

Figure 10drive beam

The peathe peaks at 12 GHzcase. Thiforward particular not supprethe total compared

Operatiuse of diff

The recmain beambe investi

The droptimizedenergy mthe resonConsequeand the feHence, adconsidere

[1] F. Temeeti

[2] D. ScKyoto

[3] D. Scchapt

[4] D. Sc

y the errors ofpeaks at 4.1 MFig. 9 with Fig

ltering can be ples. For differre different, aned out by fittlength to themple of the 2.25 ulses are divide recombined. 3.1 MHz and

0: Main beam m RF amplitud

aks at n×4.1 Mremains unfilt

z compared wiis difference system funct

r because the laessed. E.g. forRMS error d

d with 0.11° for

SUMMARYion of CLIC afferent drive becombination pam linac and thigated. drive beam ad to cancel ou

mode; however nant errors

ently, the phaseed-forward sydditional filterid.

REFecker at al., Cing.web.cern.cchulte et al. “Co, Japan, WEPchulte et al., Cter 8:“CLIC Opchulte et al. for

the drive beamMHz, or its mult

g. 8). set up only forrent lower enend hence all reting the drive

m. Fig. 10 demTeV mode – i

ded into ~325 nHence the res6.1 MHz.

RF phase errode error for 2.2

MHz are filteredtered. The totalith 4.96° at 12is not large,

tions much large peaks at 2r 20 MHz feeddrops only tor the 3 TeV cas

Y AND OUTat low energy eam recombinaatterns lead to he amplitude o

accelerating sut resonant err

it cannot be aof all lowe

se error is highystem cannot reing or correctio

FERENCESCLIC parametch/clic-meetingCLIC Energy SPE022 CLIC Conceptperation at Lowr LINAC2010,

m accelerator Rtiple, are suppr

r one frequencergy modes thesonant peaks ce beam accel

monstrates this in this mode thns and ~163 nssonant modes

or caused by 025 TeV mode.

d out, but the rl RMS error is

2 GHz for the 3however the

less effectivel2.0 and 6.1 MHd-forward bando 1.05° at 12 se.

TLOOK modes require

ation patterns [3gradient error of these errors

structure has ors for the noadapted to filter energy mher for these meduce it sufficion methods mu

S ter table, http:g/clictable2010Scans” for IPA

tual Design Rw Energies” MOP024

RF the ressed

cy and e train cannot lerator using

he 488 s parts are at

0.05%

rest of 5.58° 3 TeV

feed-ly, in Hz are dwidth

GHz

es the 3]. in the

s must

been ominal er out

modes. modes iently.

must be

://clic-0.html AC’10,

Report,

TUPPR024 Proceedings of IPAC2012, New Orleans, Louisiana, USA

ISBN 978-3-95450-115-1

1866Cop

yrig

htc ○

2012

byIE

EE

–cc

Cre

ativ

eC

omm

onsA

ttri

butio

n3.

0(C

CB

Y3.

0)—

ccC

reat

ive

Com

mon

sAtt

ribu

tion

3.0

(CC

BY

3.0)

01 Circular and Linear Colliders

A03 Linear Colliders