,,
Io Introduc'l;ion., -----
CE.RN~·PS/K3 30 lfov, . ·_. ·: · J.9~i6
INJECTION PHILOSOPHY FOR SYNCHROTRONS .R)
The injec:ton system is a very important part of any acoelers.tor~ and the beam
intensity one can expect to get out of the accelerator depends very much on having
the ic"ight deaigi~ for the injectoro By injector is in this respect meant everything up
to the synchrot~: ono In most large synchrotron projects om plans to use a linear
accelerator ae :injectoro Designs for such linear accelerators exiet, both for electrons
and for protons., However, such ma.chines are constructed for nucleE.r physics experiments9
for medical purposes or even for industrial applications, and the requirements put on
these machines for such purposes may be quite different from the requirements one would
like to impose on an injector.
The purpose of this lecture is to show the principles one sho'+ld follow in
analysing the injection problems to find the special features one would like an injection
ayetem to haveo I do not know the new German 6 GeV project well enough to be able to say
vhat would be the result of such an analysis ·carried out on that project, but I hope
that the method~ to be described will be useful, and we shall also during the lecture
be able to draw a few conclusions relevant to that projecto
IIo LioUTille 0s Theoremo
The equations of 1110tion o! the particles in an accelerator can~ lfi th the normal
approximations rmd assumptions, be derived from a Hamiltoniano This makes it possible to
apply some of th~ general theorems of Hamiltonian dynamics in anal.ysing the beam dynamics l
of particle accularatorso
Probably the most useful of these theorems, &nd the only one I shall use in this
lecture~ is the ao-oalled Liouville~s theoremo This states that if the motion of the
particles can ba derived from a Hamiltonian~ tm density of pointe in phase space -====~"'======;==-==-=======----="'!"""1"====================~=====================
~) Lecture given at the Seminar on PQrticle accelerators at the University of Bonn 3rd-6th Octobar, 19560
K !l~I~ ~ DE f-'A1:\f El"i f.: tff T 0~6190D21.DEC. 5 6
I
- 2 - C"ERN-PS/KJ 30
representing a group of particles does not change with time. Others prefer to state
the theorem in the following way: particles move in phase space like an incompressible
fluido
The motion of a particle is in general a three.dimensional problem in ordinary
space, and the phase space would therefore be a six-dimensional space, and it would
be very cumbersome if we had to consider the motion in this space in order to get the
information wanted. There are cases when that would be necessary, but in general one
can separate the motion in the various directions into independent one-dimensional
motions. For example, the longitudinal motion is usually independent of the transverse
motion, or the so-oalled betatron oscillations. Further, when the analysis is not too
detailed, one _hormally al.so assumes that there is negligible coupling between
horizontal and vertical transverse oscillationo
Although it should be kept in mind that these are approximations, they are
justifiable approximations for moat purposes, and we shall therefore introduce them
here., This simplifies. the problems. Instead of a six-dimensional phase space, we now
have to operate with only three independenttwo-dimensional phase spaces (phase planes).
Further, as the vertical and the horizontal betatron oscillations a.re of the same
nature, we only have to consider tbs two motions, the phase oscillation and s~ the
horizontal. betatron oscillations, and Liouville 0 s theorem now states that in each of
the corresponding phase planes tbs particle density round a given particle does not
change with time.
We could now continue to discuss in such general terms the problems involved in
determining optimum injection conditions that no reference would be needed to which
type of motion we are considering~ However, in practice, the problems are so different
that I prefer to split the problem al.ready here and first consider the betatron
oscillations and afterwards the longitudinal motiono This has also the advantage that
we have something more concrete to ~ncentrate on in the analysis.
IIIo Free Transverse Oscillations (Betatron Oscillations).
We ahall first consider the free transverse oscillations, the so-called betatron
oscillations, confining ourselves to the horizontal oscillations only. As already
mentioned, this means neglecting coupling between horizontal and vertical oscillations
and in that case the analysis applies equally to the vertical betatron oscillationso '
C:,i iN-PS/KJ 30
1) Defini1ion of acceptance.
·rbe optical system under consideration, in our case the linear
accelerator or the synchrotron, is characterizttd by a few quanti tiee of special
intere~t in this discuesion8 th& most important one of these being the
sok~calJ .ad acceptance o
!.et us assume that the half-width of the available horizontal arerture
is ac Thia is in general not the width of the vacuum chamber but the width .
a:f"ter E<pace for phase oscillations, misalignments» etc~ e h~ been subtracted,,
Le·t us further, for the momentp assume that the particle motion is sinuso.idalo
This iE1 nearly correct for a fixed gradient machine, and ir:J a ro\lgh approxima
tion f,,r an alternating-gradient machine o We shall later consider the Ao Go
maclU.llEi· in more detail o
With this assumption a particle will in the phase-plane move in an
ellipSf1, as indicated in figure lo
-a
,,"' I I
' '-
--
-b Figo l
a
l!Urther9 all the ellipses described by the various partieles in the beain
'Will have the same shape, ioeo the axis ratio ~./! ic-. t~ same far all
.e:W.pse:Jo Here 1 and ~x are the amplitudes of the transverse displacement and
transvorse momentum of a particle reapectivelyo
The largest of these ellipses one can draw without goirig outside the
availa"ole ,aperture.± a is drawn in full on Figolo It is then readily noticed
tbat it we inject a particle in such a way that its point in thia phase plane
is out'.lide this ·fully drawn ellipse, the particle will sooner or later hit the
wall a::id get losto If, on the other hand, the particle is injected inside thif~
fully ..irawn ellipsep it will au.rviveo Consequently, tbs focusing system can
only a1Jcept tMt part of the phase plam that is inside t:ki! ellipeeo One
therefore cells the area of this limiting ellipse the accxrnta.nst~ of the system ..
If one ooneiders the horizontal betatron oscillations one talks about the
= 4 -
hc-rizorita:t acceptan:e, and oorrespondingly the vertical acceptance for the
'1erttci1t oacillatio:'l.Bo Generally cne triel3 to make the two acceptances equal
and talks ju.st· a.bou!; the acceptan.cs of t:OO synchrotron"
As pointed o-it a.bove 9 this limiting cu:rve is always an ellipse 9 and a.a it
ie~ aMkl1a.:rd always t ·:> keep a 'It in the expressions, it has become conventional
to d.t:ftne the acceptance as the araa divided by 1t ? or if N stands for
acceptence
N ~ ab (1)
A more convenient w.g,y, in some cases, to write the acceptance is
(2)
The ratio~ fJ/x is given by the focusing properties only, and for a given
focusing strength the acceptance :1.s seen to be proPQrtion~ to the square of
tbe aperture. For sinusoidal osoillations1 as we have dealt with so far, the
acceptance can be written in ternw of the wavelength ha of tha oscillations
(3)
Here Pt is the longitudinal momentum of' the particle. For convenience one
often 1U1asures the momenta in terms of m c 9 in which case N has the 0
d:7..mension of length. o
Qw.te often one also finda that tha acceptance is defined in terms of
ang1.tlar spread x 0 instead of p , If we call the acceptance defined in this ' ' J:
lll\Y A ve get
and in:::tead of (3) 11e then get
2 A~ 2~ a /'A')
(4) ,
(5)
- 5 - C~HN-PS/KJ 30
The relation between A and N is given by
(6)
It should be noted that ( x 0 ~) are not canonically conjugate
variables in a system vith acceleration, where PL changes, and that one there
fore hae ,no area preservation in this planeo It is therefore preferable to
work in the ( x PP:z:) - plane and use N if one .c.o').eiders accelaratore. But as
the relation between N and A is very simple, it does not make much difference
which definition one uses, as long as one keeps in mind the fundamental
difference of the variables in the two syatemso
Let us now consider what happens in an AoG. synchrotron. Siguraeirsson
(1952) was the first J:erson to show that one could use ooneiderations like this
also with AoG• focusing, only would one get ellipses of different sha~s
according to what part of the structure one considerso Let us consider that
part of the structure that gets the largest transversal excursiono That is the
mid-point of a focusing eectiOno Observing in this point, we again find that
the particle describes en ellipse in phase space, only does it so by discrete
point plottingo Again we therefore get a limiting ellipse, which, for this
symmetry point, has the same orientation as indicated in Figolo A particle
being outside this ellipse at injection will sooner or later be lost and a
particle being inside will surviveo We therefore use the same definition of
acceptance as before : the area of this limiting ellipse divided by 1to •
The symmetry point of a focusing section has here been chosen as the
reference point for oonv<tnienoe, as it is easy to find the limiting ellipse
in this caseo In any other point there will be a limiting ellipse of' the same
area but of a different ahapeo If om ia inte~~ted in finding any of these
other limiting ellipses, it is easier to do this by a suitable matrix
tnnaformation or the ellipse in the symmatr.r point chosen, mald.D« use of the
transfer matrix from the symmetry pn.\llt to th~ point in queationo
The axis ratio and the area o'! the limiting c:'lllil>Se 1n the symnatry
point can be found from the transfer matrix over one period of the structure ,,
en 6 Kt• ( ~ '; '' ,,; t 'S/KJ 30 . L;.:.L·.
I shall not go thxough the calculation hero but only give thrt result:
where 'l'jk are the matrix elements defined by
(9)
vhsre t~ indices 1~2 indicate the symmetry point in question and the oi:JS
being one period away"
2) D&fini tion of &ii ttance., Simple beam charu(.Jlg eystemso
Lat us then consider what the beam looks like in this same planep how
it changes with time and how we can change 1 ts shape by simple devices.
Especially tbs ls.st problem. is of .importanoe in solving injection problemao
If we take the beam at a certain point, say just after t,ne linear
accelerator, the beam will occupy a certain area in the phase pla.nep tor
instance o.s indicated in .Fige2o
.Fi.go 2
"': 7 -
If we now follow the particlesw represented by this area~ as time goes
on~ it follows from Liouville 0 ei tbeore11. that the area is an invariant. ThiP
invariant ls an important property of the beam and is called "emi ttancs11" R
In order tei get confonnity between the definition of emittance 81ld the definition
of acceptance P one does not in practice use this definition of end ttance
direotly, but defines instead the emittanee as this area divided by 1to
I:f' the beam has gone through a rather loDg optical system, like for
instance a linear accelerator where it bas been aperture 11.mi ted, it will
look considerably more regular than indicated in Figo2a in fact it shOul.d
be clear from the previous section that it is marly ellipse ah.ai:ed, aa
indicated in Figo3o
/
/ //
/
/ ""
/ /
( ,,/'"
' ---=- - --
----- ....--....... ,.,- ' / / / /
/ /
/
/ /
Figo '
lC
Fig., 4
Altl'oough the emittanctt of the beam. is invariant, its shape can be
modified considerably by rather simple beam shaping deviceso Thi111 is indeed
important, as we Shall see later that it ia very desirable lo be able to makB
such changeeo First we shal.1 9 as examples, consider two of the very simplest
beam shaping devices, a d1"1.tt space and a point lenso
a) Drift .... space o
It is self evident what. is meant by a drift~spaceo It is a space where
there are no.forces present that can change pxo Such drift-spaces are
unavoidable, whether they are use:f'Ul. or not, but often they can be made usef'ulo
Mathematically the effect of a pui-e drift s~ce can be wr1 tten:
• The notation is not quite universally established in this field yet and should be taken with some resenation.
)
- 8 11911'1 OE.Hlki: '3/KJ }0
(10)
Here '( is directly proportions.al to the drift ... length» but depends also
on many other factors, such as how close one is to velocity of light and,
of coursep on the system of variables one has choseno
The dotted curve in Fi.go} shows ·what happens to the beam wh~n trans
versing such a drift space .. Since the equation (10) gives a linear relation
between p and x at the two ends of the drift space, the shape of the ·beam
is still an ellipse 9 the area is unchanged and the maximum p is unchanged,
and 1 t is easy to construct the new ellipse o One only needs to know the d-ift
of one point, say the point (O~p)o
b) Po:ffit ~o
A point lens is by definition a system that changes only p . and mt x o
The effect of a linear point lens, which is the only one we shall consider9
can be written
C:) = (:: : ) (~) (11)
where 6 is the ,.strength" of the lens, closely related to its .focal length
in an ordinary optical lanso
.,, 9 -
The effect of such~ lens is shown in Figo4o.As a starting ellipse :tn
I have chosen tho result of the operation/Figc3~ and 6 ia so chosen that
the final ellipse agai::i ha.a its axes coinciding with tha co~ordinate axea"
What thia axample shove in particular is that by these two simple operations
we .have been able to transfar an ellipse with x~,p orientation into another
ellipse with the same orientation but with quite a diffe~nt axis ratio 11 the
area of all ellipses being t~ same. Thia is an important ,r·etru.lt that we are
going to make u5e of latero
J.t is further noticed that by o~ratione like this it is possible to
transfer any ellipse into an;y other ellipse of the same area but with
arbitr~1.:f a~ia ratios and orientationso
}) Beam Matchingo
With the two previous sections as a basis, it is now easy to pass over
to beam matching~ Let us first take the 'particular example of the emittanca
of the injected beam being equal to the acceptance of the synchrotronp that
the shape of the beam in the phase plane is an ellipse, but that the shape
and orientation of this emittance ellipse is different from the acceptance
ellipse, aS indicated .in Figo5o
emittance
aeoeptance
Figo 5
Ii ie then 1)bvious that we shave off a portion of the hoam equ&.l to
the }X:lrt of the eud:ttanca that is outside the accepte.nc.'"'·
fa \f.ewr, it :i.s now poesiblei, for inBtanc-e by uning eLnple bean; shaping
devfoe5" a.e indicated in Fi.go~ and 4~ to gt\te ths emittance r:}J.llpse the sama
aha.Jj(l 6tlc •Jrientation as the acceptance, and. thus one just a.ccepts all tlie
})3.rlic:le" into the aynchrotrono Thie is what we usually mean. by beam matchingo
One impo1 tant fact to be noticed is that i.t is no aim in itself to get tm
cross sectton (in configuration spac~) as small as possibJ.A., In Fig,, 5 for
insttm.ce" the ti'allSverse dimension of the beam is already too small o Al tholJ8'h
this :ls f. ·very obvious conclusion. from considerations like this~ it is often
diffJ.cul i to get the idea round to the people making for instance Van der Graaff
gene:rat1n·s or linear accelerators9 who seem to thi.nk that their machines are
better~ ths smaller spot size they can make .,
If the beein emittance is smaller than. the acceptance~ one still wants
a good nwtch if possible, i~e. the two ellipses having the same shape and
orientatjon at the injection point, although being diffGrent in sizeo This is
b&eauae or.e wants to occupy as little aa poasible of the vacuum chamber, as
one uu:.y in the end have less aperture than expected, or the barun emittance may
be larger than expected o
It' tha emi ttanoe i• larger than the acceptance, the complete matching
is not so important as long as one matches ao well that one gets full overlap
of tha al Upses. However, perfect matching is still advantageous~ as the density
in the phase plane is in £'Emeral highest mar th& centre and decreases outward.so
Ir. p:r.actice the beam emittance may not be quite an ellipse~ but 19rhaps
more as :indicated in Figo2o In that case one chooses the ellipse that fits
best, and matches that one o
Tl!e emi ttanco and shape of the beam injected into the synchrotron is
mainly determined by the optics of the injr:ictor (linear accelerator), and what
we hfave c;alled a beam matching is in fact as much a matching between two
machl.:neao Realizing this we also understand that the requirelllf~nts one wants to
put on the linac are very much determined by the synchrotron acceptance and by
the possibilities one has to put in matching devicesp si.ich as drift-spaces and
lenses~
.,, 11 -
I :n practics, the group in which I tmrk only dealt with problems like
these in proton w.chineso In principle$ the problems are the same for proton
and electron ma.chines~ but the practical. Sf>lutions may differ considerably.
I shall therefore not go into detail of th•:l solutions chosen by CERN o
Bowever~ a few nUJrerical figures for possible eleotron machines may be
of intersat. I then use as a basis some figures that are available for the
Cambridga Project in the UoS.;A. (Livingston.a, 1956)P as that project is similar
to ti/.e 110w Germain project in many respects •
.Jlss~:og an available aperture iltf .± 12 mm one gets
H ~ Oo45 x l0-3m for 20 MeV injection
~ Oo9 x 10-3m for 40 MeV injection
lt has been. difficult to find any figures f~r existing electron
linear e.ceelerators, but reasonable -ctalues seem to be a spot size of 10 mm
diameter and divergence of .± t x 10-~ radians at about i5 ... 20 Me Vo This
would give an emittance of
If one could increase the beam current from a linao by pemitting it to give a
larger emitt,:nce, one would do so, preferably by asking for a larger spot Size 9 L.
again an example that 011S should discourage the ~r of the injecticr to
produce the smallest possible spoto
Al though the beam shaping devices nentioned earlier a.re simple P one
should not forget that matching is needed in two planes, which complicates
the problem considerablyo One should thel"Elfore plan the whole injection system
such that a minimum of beam shaping is needed. This is not easy in proton
maohinsa, where the f oousing of the linear accelerator is a very tricky problem
in iteeJ.f ~ In electron machines, however, the focusing problem is considerably
easier, and one should therefore keep in mind the possibility of getting the . proper beam shape by suitable design of the lina.c foous~ng systemo
Pf" Plm;l' and Eno:· Q' Oscillations (.Synchrob:·on Oscillations '., "'-"'~·---o". ,..,,.____ ,..________________ ~
~ ·o Eihall n1 w- pass over ti) th~ lcmgi tu<.li.nal motion m1r: cons:>.der the
'oacHlai:'.ons nbou·: the phase s·httonar-j' poi.n:t:o These a.re th:: SO·=~alled phase
ose:.illa1:·.ons, or Gynchrotron oscillations" It should then firat of' all be
recalla( ~;hat in 1m electron synchrotron o:ne would always bJect above the
trausitHm energ:r:. and in general far abov'3Q Thia and the f~"ct that the
injecte~ olectronu are far up in th9 rolat ivi.stic regioni mtkes the phase
trappin£: ;?roblem fook different in an Ao~o synclrrotron for e>•lectrons as from
that of p:rotonsa
fiince injef,tion is ah~ve trans! tion~ the phase stationary point is on
the :fulling side of too radio-frequency wave, at !l distance qi8
from th.a
zero 1:~rossing" As a variable we use A ip ~· which is tm deviation in phase
C·f th1 particl1::i under consideration fro:n <p8
o The corraaponding can1Jm.cally
conj11ttate' variuble is 4p/m0
cf., where ~p is the devie.tion of the loDgi"~
tucli..J!Hl momenti:.1m of the particle from too longitudinal momentum of the
phal*· rJtationary particle. Since the pa:rticle velocity iE: close to that of 2
ligbf~, this conjugate variable can also be written AE/D' c ~ wherE• IE is 0
·the e:nergy dif:f erence between the particle in question ar:.d the phase
statjona.ry particieo
hJ before na car: in such a pha~~ pl'lnfJ draw the limiting curve for the
acceptance of J,articles. This looks somewhat like indicated in Figo6a.
1r,p/m c 0
----:----+-~·~
A'P
Figo 6b
.~ radia1 oscillation is also associated with the energy oscillation~
lf tl:·f;;ae radial oscillations are so lart~ that the trapping ia limited by
the r~i.dial aperture rather than by the -phase stability limits, a curve
f'urth~·r in than the one given in Figo6a is limiting the tra.pping efficiency~
for example the curve indicated in Figoobo In a proton synchrotron with
AaGa focusingt one has usually phase U~nitation, -whereas it looks as if, one
may easily get aperture limitation in a:o. electron synchrotrong
The curve in Figo6a is not an ellipse because the forces acting on the
phase oscillations are rather non-linearo Particle trajectories insi~e this
limiting curve will therefore not be ellipses either, but as these tra~
jectories depart from the limiting ourvH they approach more and more
ellipses, and for most purposes the linear approximation to. the phase
oscillations is sufficient" ~f we have :rather strong aperture limitation
the curve in Figo 6b will therefore be ~lose to an ellipse~
The beam from the linear ~ccelerator will come in bunches as indicated
AP/m c 0
cp
If. the frequency of the linear acceJ.erator is high comps.red with that
of the synchrotron, there will be many of these bunches within the trapping
range in Figo6, and we can treat the benm more or less like a continuous
beamo This will always be tbs case in proton ayn~hrotronso
In electron synchrotrons there is 9 however, the possibility of having
a harmonic relationship between the linnc and the synchrotron, say by
running tbs linac on the second harmonic} of the synchrotron, or Etven o:n the
fundamentalo The injection problems are somewhat differ~nt in these two
caaes, and we shall treat them sepa.rateJ.yo
. .
' ' 'I' · '. [ :
> 1 this c.~ :11 the beam ie conaiderec1 to be a contini.ic\1s beam as
' I
i.rp · ;!il c. !)
Fig .. 8b
'Ibo si tuatl.on after about a quarter of a synchrotron oscillation is tl! .
ehown ~.n FigoSb o The bending ove-.r/t'be baam is due to the non--linaarityo Th:Ll
l!Ollc-·] inearity ulso leads to an increase of the effective area of the beam in
i: he I l>ase planf'· ~ as the beam Winds round. and round in phase space effecti vel~
takir.6 all the ap&co inside the li.mi.ting c'ilrve. This may seem like a violation
c1f Lfouville 0a theor.emp but~ of course~ it is not. It is sormwhat analogous
t.o wil:,ding a tl1read into a ballo There will be plenty of air making the
'roJ.tuu: of the ball larger than the vol111oe of the thread itself, whereas the
three.a has not ~hanged its volumeo 1-lowe·rer, the analogy breaks down on one
imp01-t:an-t poini ~ The ball can usually b13 unwound. 9 also in practice. Tm beam
can .1n princip:ler- but not in practice. 'rhe process is in practice unro
c·oinwD.bla, alth1ugh it may be ~versibl1l in principleo Hareward (1954}P who
:firn1. describe':' such a process in an ac~lerator, cal.lad it filamantatioiiv
F:LgoBa now indicates wha·t requiremei:lts to put on the phase trapping
region and the parameter.a dete:r.mining i !:~ a.'1.d on th1:1 baron its"'lf.,
Erstly 0 or;e wants the width in pha3e to be as large as poasibleo That
u·eanf; that one should work with a large RF. voltage compared with the
1.i.ecam:~:i.ry acco; 0:ration~ .or in other worisv the working poj.nt on the RF wa.w
abouJ d. be rathe:J' clcse to the zero oros3:Lng, In proton machines one hae to
1"trH.:i:1 a comp:r(:nLise ~ usually auoh that the trapping width is about 'It ( 50 o/o;
In electron mad1.i.nes, however, the acce laration needed later to compensate
i;ha n~diation · NJses ia so large that 0:1e can: easily get a trapping width SD!.Plitude
:naa:i:- 2n~ or .1 00 o/oo One shall in fact ha-ve to/nvdu.1at~ the RF in order
not 'i;e> come to.: cloae to tha zero croestng" 'r.:b:R.t is for other reasons 11 but
orw Ei'!;ill g9ta a fairly large tr~:p.iiittt;; i'rldth~ ~rtainly aomewhare between
50 o/n ~ JOO '' "'"
thE1 next requireraent ia on the width in energy of the ti·apping region,
or c.n i;h.E~ depth of the potential we.11~ as we r.Uao may call it~wbich shoulO. be
ama:11 enough not to gjve apei-turo limitation. · Here ie ot'lft of the
between electron maohi.:ngs and proton machinEHJo Xn AoG,
:9roton .:iynchrot!'Ona the depth of the pntr:>rrf;ial well is usuall:y rather smal 1
and the associated radial displacement ie a small fraction of tm vacuum
oharnbe:r,. To get such a situation is vei"'Y difficult with electron machims,
One liaa to work on a vecy high ham.onfop and ae ah"'eady mentionedv have
wnp~_j .. tude modulation of the RF to give a relatively small RF voltage at
injection compared with at ejectiono In the Dambridge project~ for instance~
it :i.s planned to work on the 350th harmonic and to amplitude l'JJOdulatep and
ati!l they plan to have a horizontal apextl.ire a little more than 3 times ·ctie
vertical aperture, which they certainly reed .. Personally I feel one might
cor1aider to go up to the 700th hamonic, i,,eo about 1000 Mc/S o
let us then consider the requirements om would like to impose on tm beaw.~. From fi~;n 8a it is seen that we want its energy spread to be an.all
com:r:ered with the depth of the pot.ential well, say Y4 of this deptho
'1'.'here isl' however, another effect and that is the betatron oscillation
tha"c is associated wi.th an energy spread,. A pa.rt:i.cle with an energy
deviation of &J~ from the correct sneJ.·gy will have its ~losed orbit moved
by fir~ given in the extreme relativistic approxi.mation by !Jr/rm= ex IJE/E~
whex"t:"l et is the momentum compaction fa~tor. A 2,5 o/o er.iergy spread gives
for example in -Che Cambridge machine e. fh: of about 3 cm L' If this parti1;:1.e
:.1.s ::.m.rnohed on i;o the same orbit as the particle with f.'orroct energy~ tbiB
diapJ.a.wment hall to be taken up by a betatron oscilliition Oif ~ cm amplit11d~h
Thi1:. is serious ; and if nothing is don.:1 about it~ it w:i.ll either increase
the requirement on the vacuum chamber width or lead to particle losso
Ona can roduce this betatron oscillation amplitude by using the prope:r
double bending cm the in:flector to pJ.a1:.-a each of the injected partial.ea on
its closed orbit. J or at least near to :.\.t" Thia is a complication, and a.
rather unnecessnry one for protonso For electron mac-.hines, however, it should
bo looked into :Jeriouslyo Probably ow UJB.Y get the energy spread of the
li:oa~ below the figl.U"a indicated above~ but perhaps not so very f'ar below~
~Jne should notice that one has to wake any wante:;d :L1.fl'ovement of the
{-m~::·,;y opread in the .linea:x- acceJ.erator :i.tself9 as thCJ:\3 is nothing one can
do to the energy spread between the linear accelerato1· and the synehrotrono
Thi 8 may seem .surprising if we r~member the close analogy there is be"b..i~en
the phase oscillations and the betatron oscillationsw and for the betatron
oscUlations lie discussed earlier very simple beam shaping deviceao
Such beam Bha:pi:o.g devices do indeed also exist for the phaoo oscil.lationsp
and· i;.he so-caUed debuncherr which has been described in a C.'ERN-~port
{Jobnsen~ 1955), wai'I invented just for the vecy purpose of reducing the
ene:cgy spread of the beam going into a synchrotrono The only draw-back~
however~ is that although it probably uorks beautifully for proton machines,
it wHl require completely unrealistic drift-spaces when the p;i.rticles are
as htgbly relativistic as electrons of 20 - 50 M3Vo
If we run the linear accelerator o:n the same frequency as the
synchrotron or on say its second ha.rmo~1ic, we get a qui-te different situation
from the one just described~ This ie e possibility that has been touched upon
sava1•al ti.Joos also for proton synchrot:r·ons, but seems to be unreal:i.stic for
such machinesc. One should thsn either use t..h.e fundamental or the· second
harmonic (which already throws away half of ·the particles)" Any higher
hanQanic: 9 al though doing no harm11 does not seem to give improvement over
the case of no harmon..'lc relationo
F'or simplfoi ty we here consider the case of dr.l.iri:ag the linac a.t the
fundamen.tal cf the frequency of the synchr.:itron$ althciugh the arguments
apply equally well t:o the case of the If:econd. harmonic"
I.n tact~ the case is now completely analogous to the casa of matching
the te"&atron oe;cillations (section lIL))o ~~e bunches have a certain
shape and a certa:i.n siz&, and one has to match"' if possV:ile* by giving the
trak!ping reg.ton (Figa6) the same shape e.s the buncheso The non,·linearities
make this not quite possible for very large buncheeo However~ in practice 9
one would then match by giving the elli.pses riear the origin the same shape o
The bunches wi.11 probably not be larger than the linear :region of t.he phase
trapp:ing area, If one runs on ths funde::nentaJ.~ one wocld in this way P in
,·.
...... \ ... ? "'
p:x1.nciple~ be a'ule w trap 100 o/o of the p<lrticles,. on the second. harmonic
50 o/oo A further advantage would be that the synchrotron oscillations would
bo smaller, as the area. near the edges of tm ·phase trapping region will
not be filla~L
I have not had time to investigate in detail if it is possible to do
this in practice~ but if it is not possible to get a complete match~ it
stilJ. looks as if much can be gained by this method even if one has to be
satisfied with an incomplete matcho
One difficulty one has with this matcn over th~ matchi.ng of the
betatron oscillations, is the difficuHy mentioned earHer when H ~1as shown
that a debuncher does not wom in the case of a'l electron synchrotron., One
cannot rely upon any bunch-shaping device between the linear accelerator
e.nd. the synchrotron. All matching of the S'.r'l!lchrotron oscillations must be
done by adjusting the parameters of the synchrotron and possibly the early
part of the linear accelerator (the bu.uching section)u
There does not seem to be any difficulty in making a linac run on the
second harmonic of an electron synchrotron if this is run at about 500 N.c/s,,
However~ if. one wants to run on the fundamental j one would probably prefer
not to run the linac much lower than at about 1000 Mc/a and one would trere~
fore have to increase the synchrotron frequency to about this value. This ia outside .
rather / what one is used to~ but there does oot seem to be ~ fundamental
reason why this should not be doneu In my opinion, it would have advantages
that makes it worth oonaideringo
4) )i'reguency tolerancesu
At last we shall briefly consider frequency tclnrariceso A frequency erro1·
shifts the phase stationary point outwards or inwards. This shift must
therefore be small compared with the radial width of the phase trapping
regiono Rather careful considerations of timing requirements and possible
frequency errors are therefore needed for proton synchrotrons (Hereward,
Johnsen, Lapostolle, 1956L whereas the problem is simpler for electron
machinea, and we shall consider those here.
CEfff: .. } ':~;/KJ 30
If LLe frequency ie sh;.fted by IA f 0 arid we c:alJ the corresponding radial
abifi: xf r wa have the relation (Courant~ Livingston and Seyderr, 1952)
/;,f/f:; (J.2)
and jf we insert for instance the Cembrtdge parameters~ we get
-4 -1 3 x 10 om
Since this is a fixed frequency machir.e, the difficulty in keeping
the :frequency error better than 10-4 is small~ and effects of frequency
exrors can therefore be considered negligible.
Vo Conclusionso
I have here mainly tried to give the philoephy one should follow in
tryirit: to a nalyss the problems one gets involved in when wanting to optimize
in,jection conditions for a eynchrotrono For proton synchrotrons the problems
ha·re been studied in consider.ably more detail than I have found it preferable
to ropoJ~t here., I can refer to CERN reports by Lapoatolle (1955), Hereward
(1954f l955) and myself' (Johnsen, 1952v 1953 and 1955L. and a summing up of the
main reaul ts will be found in the Proceedings from the Conference on Accelerators!
held J.a::1t June in Geru\ve (Hareward, Johnsen and Le.postolle~ 1956)0
The matching of the betatron oscillations are not dissimilar in the case
of electrons from that· of protons, but the problems involved should be kept in
mind l'l'hen the requirements for an injector is going to be specified~ The question
of space is important~ as beam shaping devices require more space the more
difference there is between the two systems to be matched (longer drift-spaces)o
(; ' 3/KJ 30
Fo~ the syr.(jhrotron oaciUat;.ons thE< condi tfons are no dlf:ta:r-ent that
a few r;~·11.ierical e:ir.am.p1ea may throw some e1.:t .:ra. JJ.ght: on the problemsp al th<>l1Bh
theee mrueric:Eil examples should ba taken \l.d. th ~ll.l z-eservatj.ons,,
~l?J~~l.o Assuurlng the main parameters RV "in 1;he C8lllbrid~ projeot
(Livi1~ston, 1956)P but assuming no amplitude tlOdvl.ation ,~j.' the RF., we fiud
that ·the radial half-width of the trappir.&£~ region 1.s 10-30 cm depending on
the DC bi.as chosen and on the injection einergy. This is i.Illpoasible, and RF
modllle.tion must be incorporated to get t.h:l.s &Dlnll enough~ although there·
are waya of pressing it down a littleo
~~-!!,_G_,,. Same ae above but with RF amplitude being smaller by a fao1;or of
about lC at injection, gives about .t 3 cm width., If one :further incNaaea the
h.arn.oru.c number to say 700 it should be ~ossible to come down to say .t 2 cmo
DotiJ. these figures are reascmableg but they put rather strict energy.,apread.
i-equiratrenta on the linear accelerator~ of the order of.± l o/op preferably
lesso
There is or,.e question I have not rai ead yotp that ie whether there may
b& al tematives t-0 the linear accelerator a.a an· injector. ~L'he only other
possibility seeme. to be the microtron~ With es~cially tha nice feature of
low ene:rgy spread. On the other hand, the low energy spread is obt.5linad :tn
the mic:rotron becP.luse of the aruall stable region,, 'l'he same csn perhaps also
be obtaiLed in a linear accelerator if oDa for instance is pl'i'tpa.Ted to reduce I
the pha.s~ trappir:,g region :ln the bunching section of the H.nac and consequently
lt.ei tra:pping .affi~iency.,
H1Jwewr~ I 'fael that in the analysts of cll the injeetiori probbms for
an ale¢tron synchrotron all possibla aolutiona should be included till they
can be dlsrEtga:rded on safe teclmical grounds~ fM both tha lineEU" accelerator
and the microtron a.re still in the pictureo
Ka Johnsen
/jdw
•
References:
OOURANT~ EoDo~ LJYINGSTON1, MoS01 and SNYDERr H~~io r 1952
HEBE\iARD9 H~ Go I• ~rQJINSEN 9 Kc 9 8Jld LAPOSTOLLE, Pnp J.9560
HEREWARD, Ho Go p J.9550
SIGURGEIRSsoN~ To~ 19520
Physo Reva .§§_p 11900
"Problems of Injection"., CERN Symposium on High Energy Acceleratorr, and Pion Physics~ June 1956~ Proceedings Volol. 0 ppo 179-1910
"Energy Sproad and Phaee-fodusing in Particle Accelerators"? CERN/PS/HGB/lo
"A note on injection timihg and tolerances for an AoGo Synchrotron"; CERN/PS/HGH/5o
0 Bla:ee Oscillations" 11 CERN/PS/i:J/lla
"Frequency and Momentum tolerances at Injection", CERN/PS/KJ/l8o
"'l'he "Debuncher'' : A device for reduoing the energy spread in a linac11
11 r:ERN/PS/U/29
"Etude graphique des trajectOiresa Rep~~qtation da.ns 1 aespace des phases", CERN/PS/Pl. 0 2"
"Injection dans le synchrotron,, R6glage de l'acceptanc& du synchrotron."o CERN/PS/PLo3n
"The Cambri~ Electron Accelerator"o CERN Symposium on High Energy Accelerators and Pion Physics, June 19560 Proceedings~ Voloi9 ppn 439-4460
"Betatron Oscillations in the Strong Focusing Synchrotron"~ CERN/T/TS-2"