IMPLEMENTATION OF SYNCHRONISED PS-SPS TRANSFER WITH BARRIER BUCKET M. Vadai H. Damerau M. Giovannozzi A. Huschauer A. Lasheen CERN Geneva Switzerland

2025-04-24 0 0 1.25MB 7 页 10玖币

侵权投诉

IMPLEMENTATION OF SYNCHRONISED PS-SPS TRANSFER WITH

BARRIER BUCKET

M. Vadai∗, H. Damerau, M. Giovannozzi, A. Huschauer, A. Lasheen, CERN, Geneva, Switzerland

Abstract

For the future intensity increase of the ﬁxed-target beams

in the CERN accelerator complex, a barrier-bucket scheme

has been developed to reduce the beam loss during the 5-

turn extraction from the PS towards the SPS, the so-called

Multi-Turn Extraction. The low-level RF system must syn-

chronise the barrier phase with the PS extraction and SPS

injection kickers to minimise the number of particles lost

during the rise times of their ﬁelds. As the RF voltage of the

wide-band cavity generating the barrier bucket would be too

low for a conventional synchronisation, a combination of a

feedforward cogging manipulation and the real-time control

of the barrier phase has been developed and tested. A de-

terministic frequency bump has been added to compensate

for the imperfect circumference ratio between PS and SPS.

This contribution presents the concept and implementation

of the synchronised barrier-bucket transfer. Measurements

with high-intensity beam demonstrate the feasibility of the

proposed transfer scheme.

INTRODUCTION

The Multi-Turn Extraction (MTE) scheme [1] replaced the

previous Continuous Transfer (CT) extraction method [2

–

in the CERN Proton Synchrotron (PS) to deliver the high-

intensity proton beams for the ﬁxed-target physics at the

CERN Super Proton Synchrotron (SPS), and the operational

implementation of MTE allowed to signiﬁcantly reduce ex-

traction losses in the PS ring (see, e.g. [5–7]).

The circumference diﬀerence between the PS and the

SPS, the latter eleven times longer than the ﬁrst, suggests

extracting the beam from the PS over ﬁve turns to maximise

the duty factor for ﬁxed-target experiments. This means two

transfers from the PS which ﬁlls 10/11th of the SPS and

leaves time for a gap for the injection kickers. MTE has

been designed to split the beam into ﬁve beamlets in the

horizontal plane by crossing adiabatically the fourth-order

resonance [1, 8]. Only losses due to the longitudinal beam

structure and the rise time of the extraction kickers remain.

The intense studies on the theory behind transverse beam

splitting [8] had ensured that the PS performance is close to

an optimum. Studies [9, 10] conﬁrmed that no showstopper

is in sight when increasing intensity for ﬁxed-target physics

at the SPS.

The remaining extraction beam losses are a consequence

of the SPS requirements for the beam transfer from the PS,

as a debunching is performed after transverse beam split-

ting and prior to extraction. While this is necessary for the

SPS, it is certainly a drawback for the PS, given that the

non-negligible rise time of the extraction kickers induces

∗mihaly.vadai@cern.ch

beam losses during beam extraction. A much more promis-

ing approach appears to be the implementation of a barrier

bucket [11

–

16] which has the potential to practically remove

the extraction losses in PS. The advantage of this approach

is that no new hardware is required, as a wideband RF cavity

is already present in the PS ring. It is loaded with Finemet

material, which makes it usable in the frequency range from

400

kHz to well above

MHz. This device was installed

in 2014 [17, 18] as a part of the longitudinal coupled-bunch

feedback system.

Initial beam tests pursued at the PS gave extremely encour-

aging results [19

–

21]. It has been possible to successfully

implement barrier-bucket manipulations in the PS, and even

combine it with the transverse beam splitting yielding a

substantial reduction of beam losses at PS extraction (see,

e.g. [22] for a review of this beam manipulation).

SYNCHRONISATION CONCEPT

Figure 1 shows the magnetic ﬂux density during the accel-

eration cycle with and without the synchronisation process.

After injection, the beam is accelerated in

ℎ=8

buckets,

then longitudinally blown up, split into ℎ=16 buckets and

accelerated to the ﬂat-top momentum of 14 GeV/

𝑐

. The

longitudinal blow-up is needed to stabilise the beam at tran-

sition crossing at high intensities. The bunches are then

transversely split and debunched at the end of the cycle prior

to extraction to the SPS.

Figure 1: A comparison of the magnetic ﬁeld function of

the operational SFTPRO cycle without the synchronization

and the modiﬁed cycle with the longer ﬂat-top needed for

synchronisation.

The transfer of the coasting beam from the PS to the SPS

is triggered by the SPS, which means that the extraction

kickers of the PS are synchronous with the injection kickers

of the SPS. In order to mitigate the losses caused by the

rise time of the PS extraction kickers, a longitudinal gap is

arXiv:2210.05416v2 [physics.acc-ph] 26 Oct 2022

made with the wide-band RF system of the PS. This new

requirement means that RF systems in PS and SPS that

were previously operated stand-alone must be synchronised.

The beam energy has to be temporarily changed in the PS

to overlap the position of the longitudinal gap with the rise

time of the PS extraction kickers. This can only be done with

enough voltage available in the main RF system, especially

if the manipulation is to be performed on the order of ten

milliseconds.

Due to the low RF voltage requirements of the transverse

beam splitting,

𝑉RF

must be lowered well before the end

of the cycle. The Finemet system can only generate a low

RF voltage, too small to synchronise the beam. Hence the

synchronisation is proposed to take place at the beginning of

the magnetic ﬂat-top, where the RF voltage is still suﬃcient

to re-phase the beam quickly. Due to the periodicity it only

requires at maximum a half of a

ℎ=16

bucket phase change

in either direction. This is also required because the barrier-

bucket transfer from

ℎ=16

is non-adiabatic, which means

that the barrier is most eﬀective once the voltage is already

raised between two existing

ℎ=16

buckets, as shown in

Fig. 8.

The beam phase measurements for

ℎ=16

and then

ℎ=1

have to take place at the same frequency as the SPS. At a

ﬁxed bending ﬁeld in the PS this means that the mean radial

position at the time of the measurements must be the same as

the one at extraction. This poses a problem for the transverse

splitting process as its ideal radial position is centred. Hence,

in open loop, a constant frequency excursion is programmed

to centre the beam for the transverse splitting and then after

the process steer it to the extraction orbit with the RF systems.

This only introduces a ﬁxed, repeatable phase oﬀset that does

not aﬀect the synchronisation.

Since the bunches after the cogging are

ℎ=16

syn-

chronous with the SPS, the ﬁnal

ℎ=1

synchronisation is

only a bucket selection, which can be pre-calculated. There-

fore, no change in beam energy is needed for the

ℎ=1

part.

The steps of the proposed synchronisation sequence are

summarised in Table 1.

Table 1: Synchronisation sequence in PS

Cycle time [ms] Action

590-600 Loops oﬀ, ℎ=16 phase measurement

600 ℎ=1phase measurement

600-620 Cogging

620-650 Radial steering to central orbit

650-770 Constant oﬀset for MTE

770-800 Radial steering to extraction orbit

800-815 Hand-over ℎ=16 to barrier bucket

800-815 Debunching

835 Extraction

Cogging

The principle of phase correction at a ﬁxed magnetic

ﬁeld,

𝐵

is to change the frequency of the beam such that the

integral of the frequency change equals the desired phase

diﬀerence. This requires a phase measurement at the refer-

ence frequency, a phase correction, and then returning to

the original frequency, thus implying that the beam has to

be accelerated and then decelerated.

Since the bending ﬁeld is constant, the beam is also radi-

ally oﬀset according to the relation [24,25]:

𝑑𝑓

𝑓

𝛾2

tr −𝛾2

𝑑𝑅

𝑅,(1)

where

𝑓

is the beam revolution frequency,

𝑅

the PS radius,

𝛾

the Lorentz factor, and

𝛾tr

the gamma at transition. The

frequency oﬀset needs to be small enough due to aperture

limitation. Longitudinal macro-particle tracking simulations

have been performed to validate the phase curve [26]

𝜙(𝜙set, 𝑇, 𝑡)=

𝜙set

𝑇𝑡−𝑇

2𝜋sin 2𝜋

𝑇𝑡 (2)

𝑡∈ [0, 𝑇];𝜙set ∈ [−2𝜋, 2𝜋],(3)

which corresponds to the programmed frequency curve of

𝑓(𝜙set, 𝑇, 𝑡)=

𝜙set

2𝜋𝑇 1−cos 2𝜋

𝑇𝑡 .(4)

It deﬁnes the set of frequency curves for the

ℎ=16

correc-

tion, which can be seen in Fig. 7 (middle). Note that the

phase range is set conservatively for a whole bucket move-

ment in either direction, which is useful for testing - half of

this phase change is suﬃcient in operation.

IMPLEMENTATION

Synchronisation can be tried with the current PS beam

control system if one acts on the master direct digital synthe-

sizer (DDS) frequency as depicted in Fig. 2. Phase slips and

frequency steering can be implemented for all RF systems in-

volved in the synchronisation simultaneously since the mas-

ter clock frequency drives the clock signal for the low-level

RF (LLRF) cavity controllers. Since this is an open-loop ma-

nipulation, the handover from closed-loop (see long latency

loops in Fig. 2) to open-loop has to be implemented. In

order to signiﬁcantly reduce phase drift during ﬂat-top, the

precision of the frequency word has to be increased from 23

bits to 32 bits in open loop.

To reduce development time and cost, a rapid proto-

typing solution was chosen, also taking into account the

challenge of electronic component availability. A custom

board was designed to interface an ARM

Cortex

M7 con-

troller (STM32H723ZG) on a development board with the

current beam control (see Fig. 3). The controller’s digital

interface is compatible with the LVTTL/TTL signals of the

frequency distribution [28]. Hence, adding only a thin layer

of power management and IO interface was suﬃcient. The

文档加载中……请稍候！
如果长时间未打开，您也可以点击刷新试试。

下载文档到电脑，查找使用更方便

10 玖币 0人已下载

立即下载

摘要：

IMPLEMENTATIONOFSYNCHRONISEDPS-SPSTRANSFERWITHBARRIERBUCKETM.Vadai,H.Damerau,M.Giovannozzi,A.Huschauer,A.Lasheen,CERN,Geneva,SwitzerlandAbstractForthefutureintensityincreaseofthefixed-targetbeamsintheCERNacceleratorcomplex,abarrier-bucketschemehasbeendevelopedtoreducethebeamlossduringthe5-turnextra...

展开>> 收起<<

IMPLEMENTATION OF SYNCHRONISED PS-SPS TRANSFER WITH BARRIER BUCKET M. Vadai H. Damerau M. Giovannozzi A. Huschauer A. Lasheen CERN Geneva Switzerland.pdf

共7页,预览2页

还剩页未读，继续阅读

声明：本站为文档C2C交易模式，即用户上传的文档直接被用户下载，本站只是中间服务平台，本站所有文档下载所得的收益归上传人(含作者)所有。玖贝云文库仅提供信息存储空间，仅对用户上传内容的表现方式做保护处理，对上载内容本身不做任何修改或编辑。若文档所含内容侵犯了您的版权或隐私，请立即通知玖贝云文库，我们立即给予删除！

IMPLEMENTATION OF SYNCHRONISED PS-SPS TRANSFER WITH BARRIER BUCKET M. Vadai H. Damerau M. Giovannozzi A. Huschauer A. Lasheen CERN Geneva Switzerland

相关推荐

开通VIP享超值会员特权

作者详情

相关内容

热门标签

举报选择: