You are on page 1of 8



C. P E L L E G R I N I t a n d A. M. S E S S L E R

Lawrence Radiation Laboratory, University of California, Berkeley, California, U.S.A.
Received 6 April 1970

In one m o d e o f operation o f an electron ring accelerator (ERA), the square o f the electron oscillation frequency (A 2) is (a) m u c h
at the e n d o f c o m p r e s s i o n rings are slowly m o v e d t h r o u g h the smaller a n d (b) m u c h larger t h a n the c o n t r i b u t i o n to the square
radialintegral b e t a t r o n resonance Qr = 1. A l t h o u g h the coherent o f the oscillation frequency f r o m the ions (A2). It is s h o w n that
radial oscillation frequency o f the ring as a whole remains below for the E R A , where case (a) applies, the increase in ring m i n o r
unity, t h e oscillation frequencies o f individual electron are dimensions, for given field errors a n d rate o f resonance crossing,
(incoherently) caused to pass t h r o u g h the resonance because o f is less t h a n in case (b) by a factor o f (A/A) 2. Numerical examples
t h e additional focusing f r o m ions t r a p p e d in the ring. In this show that the degradation o f ring quality in case (b) should,
p a p e r the effect o f field errors on ring m a j o r a n d m i n o r radii is with suitable attention to the design a n d construction o f the E R A
e v a l u a t e d - t h e o r e t i c a l l y - f o r the cases in which the spread in apparatus, be acceptably small.

1. Introduction The increase in oscillation amplitude of a single
In the electron ring accelerators (ERA) now being particle crossing an integral resonance at a rate
studied at Dubna, Berkeley, Karlsruhe and Garchingl), r = d~off/dt is given approximately by
an electron ring is compressed in a magnetic field
having field index n - ~ - (r/B)(dB/t?r) such that Xs = eg2 2 , (1)
0 < n < 1. At the end of compression positive ions are
captured in the ring, which is subsequently extracted
from the compressor and brought into an accelerating where R is the beam radius, ~ the revolution frequency,
column having a constant magnetic field and hence and ( A B / B ) the magnetic field perturbation driving the
// == 0.
During the compression process the radial betatron Formula (1) shows, using typical E R A parameters,
frequency c0r = Qf2, where f2 is the revolution fre- that in order to maintain the increase in amplitude
quency and Q is approximately given by ( l - n ) ~, within reasonable limits, the requirements on the
stays below f2 or, equivalently, Q stays below unity. magnetic field are very strong; for instance, assuming
The capture of ions in the electron ring introduces an x, = 0.1 cm, R --- 3 cm, and Q = 104 sec -1, one has
additional focusing force on the electron, which has ( A B / B ) < 10 -5. Various possibilities have been sug-
the effect of increasing Q. During the extraction gested for reducing Q, so as to avoid crossing the
process n goes to zero, so that, in the absence of ions resonance: The use of image forces obtained by sur-
or other additional forces, Q would become equal to rounding the electron ring with a dielectric cylinder 6)
unity. As a result of both effects Q crosses the value or a slotted metallic cylinderT), or keeping Q > 1
Q=l. throughout compression and acceleration of the ring
As is well known, when Q = 1 an integer resonance by using the azimuthal magnetic field generated by a
is excited. This can produce a large displacement of current along the axis of the ringS).
the electron orbits and hence a beam loss. Moreover, The use of image forces seems to provide a practical
even if the beam is not lost it is possible that the way to avoid the resonance crossing when there are
crossing of the resonance could produce a large increase only few ions in the ring, but not when the ring is
in beam dimension and a corresponding decrease in charged with more ions than of the order of 1% of
the electric field that keeps the ions inside the ring. the electrons. The use of an azimuthal magnetic field
As a consequence, the external electric field which is to keep Q always above unity requires currents in the
applied so as to accelerate the ring would have to be conductor on the axis of the order of l0 s A - a n
lowered to an uninterestingly small value. inconvenient, but possible, design requirement.
It has, however, been pointed out by Van der
* This work was s u p p o r t e d by the U . S . A t o m i c Energy C o m -
Meer9), on the basis of qualitative arguments, that the
t P e r m a n e n t address: Laboratori Nazionali di Frascati, Frascati application to the ERA of the formula for the single-
( R o m a ) , Italy. particle increase of amplitude during the resonance

crossing may be incorrect. In this paper we study the also in the case of the storage ring the crossing of the
effect of resonance crossing in detail. In particular we resonance produces only a beam widening, and that
consider the cases when Q would stay below unity in this widening is not too dangerous because of the strong
the absence of ions (i.e., the coherent integral resonance reduction introduced by the factor A2/A 2. This con-
is not crossed), but is shifted above unity by the ion clusion is in agreement with the experimental obser-
focusing force (i.e., the incoherent integral resonance is vations performed on electron storage rings.
crossed). We find that in this case the formula (1) is not
valid and that the behavior of the beam in crossing the 2. Formulation of the problem
incoherent resonance depends on the ratio of the We assume that the electrons move on a circular
spread in the square of the frequency in the electron orbit with a constant angular velocity f2, and that they
ring, A2, to the shift in the square of the frequency, oscillate in a direction orthogonal to this orbit under
A 2, induced by the ions. the action of the focusing forces due to the external
The result described by (1) applies only when the magnetic field and to the ions. The ions are assumed
condition to have zero angular velocity and to oscillate in the
A 2 / A 2 >~ 1, (2) same direction as the electrons under the action of the
focusing force due to the electrostatic field of the
since in this case each electron behaves as a single
electrons. We ignore ion-ion forces, since in practice
electron having a frequency (oo2 +A2) ½, where co is the
the ion density is sufficiently low that these terms are
frequency due to the external magnetic field and image
forces, and A is the shift in frequency caused by the
Let us call Xk, Ok and ~j, ffj the transverse and the
ions. Thus resonance crossing leads to an increase in
azimuthal coordinates of the kth electron and jth ion.
beam minor dimensions, but no change in the beam
The equations of motion can be written as
center of mass.
On the contrary, in the case more often encountered xk (t) + CO2(t) x k (t) + a~ °2 (t) [Xk (t) -- ~ (t, Ok)] +
in the ERA, when
+ A~kO2(t)[xa(t)-- ,2(t, Ok)] = a COS (~0 k + (a),
A 2 / A 2 ,~ t , (3)

there is a (small) change in the local beam center of mass, Ok = ~td-O~k,
but the beam minor dimension increase is smaller, by a
factor of (A2/A2), than that expected on the basis of (1). ~j (l) + M j2..[~j (t) -- .~ (t, I{/j)] = O,
Hence the limit on the tolerable magnetic field imper-
fections, AB/B (which is set by the strong requirement ~bj = const, (4)
of small minor dimensions of the ring), is lowered and
can more easily be satisfied. Thus our detailed analysis where ¢OgXa is the focusing force due to the magnetic
field, the A (e)2 term describes the force of electron on
supports the general conclusions of Van der Meer and
is in qualitative agreement with observation1°). electrons, A(i)R [Xk--~(t, Ok)] and M~ [ ~ i - ~ ( t , ~bj)] are
That the simple formula (1) does not apply to circular the forces between ions and electrons and a cos (~0 k + 4)
is the perturbation in the guide magnetic field. Note
electron beams partially or totally neutralized by ions
that we consider only field bump errors and do not
is of importance, also, for electron storage rings. In
include gradient error terms as they are - in practice -
this case, too, due to the long beam lifetime, a large
negligible4). We consider only the ~ - Fourier component
number of ions are captured by the beam, when
in the magnetic field perturbation, where ~ f ~ - COg.
clearing field electrodes are not used. Once again, the
The electron ion forces are written, in the linear
frequency shift introduced by the ions can cause a
approximation, as proportional to the distance of the
crossing of an integer resonance. Both the conditions
kth particle from the local center of mass of the
that Q remain below the nearest integer during the ion
particles of the other species, .~(t,O) and ~(t, rp). The
loading process and condition (3) are well satisfied in
local center of mass can be defined, with the help of the
storage rings. However, in this paper we have consi-
step function S(O), as
dered only azimuthally uniform beams, while the
electron beam of a storage ring is bunched. Hence, xk(t ) S(O k - O) S(O + d O - Ok)
we cannot directly apply our results to storage rings. ~(t,O) = k
Notwithstanding, we think that, at least to a first y S(Ok--O)S(O+dO--Ok)
approximation, the results of this work indicate that k

~i (t) S ( O j - ¢) S (~ + d f f - ~j) By use of (8), the system of (9) can be reduced to
~(t, ¢) = j (5) A kS( t O -2-~(e)Z+A(i)z__~2~Q2 } - A~i)2 ~-A(e)2 ~ a
k T I1 k = •

The nonlinearities of this force, as well as the non-
linearities in the external focusing force, are taken into The first of (9), together with (8), shows simply that,
account approximately by allowing a dependence of under the action of the external perturbation, the local
0.)2, M 2, A (e)2 and A (i12 o n some of the parameters of ion center of mass undergoes the same displacement
the particles such as oscillation amplitude or energy. as the local electron center of mass.
Newton's third law implies a subsidiary condition This result is also valid for slow changes of
among the A(~) and Mj. We need not invoke this COk,Mk, A(~,), and A(~), so that in general we can reduce
relation, as well be seen below. the equations of (4) to an equation for the electrons
The quantities COR,M), A(~,), and A(~) are functions of only, namely
time, because of the changes in the external magnetic
field and in the number of ions with time. Both these 5Ck+CO2(t)Xk+A2(O [Xk--X ] = a cos (~Ok+¢), (11)
variations are assumed to be very slow compared with
the electron and ion oscillation period. where we have set A] A(e)2±A(i)2 When COkand Ak
We are only interested in studying the closed-orbit are constant in time this clearly reduces to (10).
perturbations due to the magnetic field imperfections,
i.e., the particular solution of the nonhomogenous (4). 3. Normal mode analysis
We will first consider the case in which cok, A(~), Mk We have reduced the problem to solving (11),
and A(~) are constant in time. Since the driving force, which task is accomplished in this and the next two
a cos ( ~ 0 + ¢ ) , is periodic with respect to 0, we look sections. We can limit ourselves to the case in which the
for a solution having the same periodicity. Let us variation in time of cok and Ak is small compared which
assume fig2. It is then possible to perform a power-series
Xk(t ) = A k COS (fi0k+qS), expansion of these quantities, and to consider only
(6) terms up to first order, namely, to write
~j(t) = Bj cos (ff4/j+¢).
O12 = 602 (to) + r ( t - - to),
The local centers of mass are then given by (12)
A~ = A ~ ( t o ) + r ' ( t - t o ) .
~(t, 0) = ~ cos (~0+4~),
We also assume that r and r' are different from zero
~(t, lp) = B cos ( h ~ b + ¢ ) .
only in a time interval t o - t 1 during which the reso-
The amplitudes A, B are given, in the case of a beam nance is crossed, and that the initial and final values
containing Ne electron and N i ions uniformly distri- of co2 and CO2k + A 2k are respectively well below and well
buted along the circumference, and assuming that the above the resonant value ~2122. Notice, also, that we
distribution of the A k, Bk is independent of the have assumed r and r' to be equal for all particles. This
azimuthal position, by is a good approximation when the frequency spreads
for both co and A are small compared with ~g2.
Z Ak,
We can now obtain a solution of (11), assuming Xk
k=l to be of the form
(8) xk(t) = ~ An(t)Ck(n)exp [i(~Ok+¢)], (13)
= ~ Bj.
j=l n=l

where the An(t ) are unknown functions and the C(7,)
Substituting (6) and (7) into (4), we obtain
are a complete orthonormal set of vectors defined as
Bj = A , the eigenvectors of the linear system of equations
A k¢ (Ok2 -+/Jk
-(e)2+A~i)2 j~2 Q2} __ A~e)2~__A(kI)2 B = a . [6o2 (to) + A 2 (to)] C~ n)- A~(to) C (n)

(9) = /"(2.)C(,n) . n = 1. . . . . N . . (14)

where C (") is defined as and

6 = {(l/Ne) Z [xk-ffl2} ~ = {(t/Ne) Z IAn(t)[2-x2} k'
k n
N e k=l (19)
as follows from (5) and (13); and F~.) is an eigenvalue. which are the local center-of-mass amplitude and the
Substituting (13) into (11) and using (12) and (14), root-mean-square (rms) b e a m size. Both ~ and 3 2, as
we get well as (17), depend on the Q") only through the
average values C (").
Z {A'.+2i~2A.+[F~.) -~2~22 +
n=l 4. Determination of the eigenvectors
+ (r + r') (t-- to) ] A,} C(k") -- i (t-- to) Z A , C (") = a. In this section we determine the eigenvectors and
n eigenvalues of (14) in the two cases: (a) A2/A 2 ~ 1,
(15) and (b) A2/A 2 >~ l. We consider case (a) first; case (b)
Using the orthonormality property of the Ck("), we is rather trivial and is discussed at the end of this
obtain section. It is convenient to start by solving (14) for
the case of zero frequency spread. The eigenvectors
A', + 2 i ~ A . + [r(.2) - ~2 ~2 + (r + r') (t - to)] A. - Ck(n)° are given, for A = 0, by

N© C(k")° = (1/X/N) e 2~i"k/u , (20)
-- r ' ( t - - t o ) N e Z A mC (m) C (") = a N eC ("). (16)
C(")° = (1/x/N) 6,,o, (21)
We assume that A , ( t ) is a function varying slowly
with respect to the characteristic oscillation periods, where we have employed N as a notation for Ne. The
so that it is possible to neglect the second derivative corresponding eigenvalues are
of A,(t) in (16) and write it as /~(2n) 0 2
= coo+Ao2 [1-6,,o]. (22)
2 inQA, + [F~,) - ~2 ~2 + (r + r') (t - to)] A, -
Notice that all the F~,) are equal, with the exception
Ne of F~o).
- r'(t-to)N ~ ~ A,,C'(m)C (") = a N ~ C ("). (17) For a small frequency spread, we can use pertur-
bation theory to determine the Ck("). Let us rewrite
The p r o b l e m is now reduced to finding the Ck(") and eq. (14) as
A,(t), i.e., to solving (14) and (17). (H (°) + H (')) C (") = ri 2) Q("), (23)
The solution will depend on the ratio A2/A 2, where
A 2 is the width of the distribution of the frequencies where C (") is a vector of c o m p o n e n t s C ("),
co2, and Ao2 is the average value of A 2. (We assume that
the widths o f the distribution o f COk and A k are small H(~°) = (co2 + A~)),~ - A g / N , (24)
c o m p a r e d with the average values of COkand Ak.)
In the remainder of this p a p e r we will study only /(1)
k~ = (co2 + A 2 2
-coo--A2)~kt - (Ak2 -- a o2) / N , (25)
the two cases
and COo 2 and A g are the average values of COk, 2 A 2. For
(a) A2/A 2 ~ |,
H (1) ---- 0, C (") is equal to C (")° as given by (20), and
F~n) = F(,)o2 as given by (22).
T o apply perturbation theory when H (1) # 0, one
(b) a~/A~ ,> 1, must r e m e m b e r that the unperturbed solution is
degenerate (all eigenfunctions, except C (°)°, belong to
for both of which solutions of (14) and (17) can be
the same eigenvalue), and use instead of the C(")°'s,
for n 4= 0, a linear combination of these vectors such
We also notice that we are interested in the deter-
that H (1) is diagonalized. Calling these new vectors q~("),
mination of the two quantities
one has
= I(I/N~) ~ Xk[ = I~ A,,(t) C(")[ (18)
k n 4~(°) = C (°)° , (26)

and We can now use these results to simplify (17). In
N-1 the case n = 0 the equation contains zero-order terms
~(n) = B~C 0)° for n:#0, (27) and first-order terms in A2/AZo. Neglecting the first-order
terms, one has
2i~Odo+[O~--fi2Q2 +r(t-to)]Ao = a~/N. (37)
1 e2nint/(N_l)
B~ = ( N _ I ) + For n -# 0 (17) contains first- and second-order terms
in A2/A~. Keeping only lowest-order terms, one has
It is easy to verify that
2 i ~ o A n + [o~z + A~ - Kz 0~ + (r + r') ( t - to)] A, -
(q$,(n) H(1)q$(m)) = 0; n, m # 0 and n # m, (29)
- r' ( t - to) ~/(N) C(") Ao = a N C t") • (38)
In case (b) the coupling between particles is negli-
N-1 N-1
gible and the eigenvalues are almost equal to the single
(4,*(")u('4,("b = (1/N) kZ= O t Y= l ~ ~
(~Ok +Ak -- O~o--Ao)~ particle frequencies, i.e.,
= ~ o , + A n + O [ ( A o / A ) ]. (39)
x { e x p 2 7 r i t I ~+nk ] } N - 1 ; n # 0, (30)
The corresponding eigenfunctions are
and that
C~") = 6,, k + ~) [(A~/A 2)], (40)
N--1 N-1
(~*(°)H(')#")) = (i/N) E E (~-~b x and the C"(") are given, to lowest order, by
k=0 t=l
C(")= 1/N. (41)
1 {exp2rdt I ~ - + k]}-
x -( -N- - - -1)~ 1 (31) Eq. (17) now becomes, neglecting the coupling between
The solution of (23) is now given by 2 i~QA, + / m z + A,-2 ~2 oz + (r + r') ( t - to)] A, = a.
C(n) = 4/n) + ~ A~"4,(~), (32) (42)

and, to first order in the perturbation, one has 5. Determination of the amplitude functions
(4,,(m) H (. #n)) In this section we solve (37), (38), (42) for the
A n =
ra (33) functions A,(t).
5.1. CASE (b)
F~2.) = F(,~0 + (q~*(")H(1)qg(")). (34)
We start from (42), which we write in the form
Notice that with our choice ofm 2, A z one has also
A , ( t ) - i g , ( t ) A,(t) = - ia, (43)
(~b*(°) H °) ~b(°)) = 0, where
so that there is no first-order correction to the coherent 1 [m2 + A 2 _ ~2 02 + (r + r') ( t - to)], (44)
g.(t) =
frequency F(~). The quantities C"~n) are now easily
obtained, and, to first order, one has

C (°) = (1/~/N) + first order term, (35) = (45)
~-,t.) = (1/~/N) (d?*C°) H (1) q~(n))/A2o The solution of (43), with the initial condition
N--1 N--1
A(to) = 0, is
= (1/NAzo) Z Z (o~-cog) x
k=0 t=l A.(t)=- i~{expli f[o O.(t')dt'l}x
i Iexp2nitlk+ n ]} (36)
x IN(N- 1)] ~ N- 1 X f[o dt' exp [-i f'i g.(t")dt"]. (46)
114 C. P E L L E G R 1 N I AND A. M. S E S S L E R

Evaluating the integrals, and using the notation for A o, substitute the result in (38), and solve for A,.
The result will be different according to whether the
D. = (092 + A 2 - ii 2 ~22)/2/iI2,
coherent frequency, coo, does or does not cross the
p = (r + r')/2 nf2, (47) resonance. We will consider here only the case in
which coo does not cross the resonance (i.e., the coherent
one has: integral resonance is not crossed), since this is the
situation which usually confronts us in practice. Under

L exp { -- i [D. (t' -- to) + ½ p (t' - to) 2] }dt'

= x/(n/p) exp [iD2/(2 p)] x
this assumption one can neglect the variation in time
of the coherent frequency and of Ao, and obtain
from (37)
x {h ( P ( t - t ° ) + D " ) - h ( D. ~[, (48) A o = cog_(~f2)2. (54)
\ ~/(nP) ,/ \~(~p)/J
where Substituting this in (38) one obtains
h (x) = C ( x ) - iS (x), (49) ~ [ , - i [D + p ( t - to)] A, = - inN C (") [ 1 + q ( t - to)], (55)

and C(x), S(x) are the Fresnel integrals. where
It is usually possible, when p is small and the 2 2 --2
integral extends from well below to well above the D = °)°+A°-n ~2
resonance, to make the approximation 2 ~2
D./~/(Trp) ~ -- 1, p - r+r'
2 iif2
Ep(t- to)+D.]lx/Ozp) >> 1. (50) a

2 ~if2
Since C(___oo) = S(___oo) = _+½, one has in this case

~ o
exp { - i [ D , ( t ' - t o ) + k p ( t ' - to)Z]} dt ' q -e)o2_~zQ2"

(2rr/p) ½ exp [+i(Dz,/Zp)-izr/4]. (51) The solution can again be written, assuming A. (to) = 0,
The value of A, after crossing the resonance is then
given by
A.(t)=-iaNC(")exp{i fto [D+p(t'-to)]dt' I
A. ,,~ - i d ( 2 ~/p) "~exp {i [ D , ( t - to)+½p(t- to)2]} x
× exp [+i(D2./Zp)-i rc/4] x f, dt'[l+q(t'-to) ] x
d to

=- i~(2~z/p) "texp [ L
_ iTr/4}. (52)
xexp (f/
[D+p(t"-to)]dt" . ) (57)

The final amplitude after crossing the resonance is The integrals of (57) can be evaluated by using (48)
therefore and
IA.I = a(2~/p)* = a [Tr/nt2(r+r')]~, (53)
a well-known result. ft ' d t ' ( t ' - to) exp { - i [D(t'-
to) + ½p(t'- to)Z]}

5.2. CASE (a) =2ei(n~/2p,I D Fh(P(t-to)+D~_h( D )]
p / 2~/(~p)L \ ~/(rcp) ,/
In this case the frequency spread A 2 is small com-
pared with the frequency shift Aoz. The situation is
described by (37) and (38), and is clearly more com-
plicated than case (b). The procedure is to solve (37)
..,_, Es,,,\ ~/(2p) /

Assuming the conditions (50) to be satisfied, one 1°'
obtains an amplitude, after crossing the resonance, 62 = { a ( ~ f 2 ( r + r , j ) [I - - X

----~p x (
× l + co~_~2a2/_lj
Ill × ~ ~j

×exp [ L
li[P(t--to)+Ol~ }
i -in/a ,
7(~- ~ (59)
z __afi2 f22 a ( f i Q (2+
COo n ) r') ½
where negligible contributions from the last term of (58)
have been dropped. By use of (56), (59) can be written
as x
[1 r (
n(r+r') 1 "~-CO2__~2~2
A+ )] X

A.(t) ~ i a v/(n) N(~ C")

[iif2(r + r')] +
x 2Re
exp i [co~+A2°-n2f22+(r+r')(t-t°)]2
4/if2 (r + r')

x 1 n(r-+r') 1 + CO2--~2~ 2 X
2 2 ~2
x exp i [ c o ° + A ° - 02+(r+r')(t--t°)]2 If
4nO(r+r') '4) (. n , ./+ 1 (65)
(60) >~
\fit2(r+r )] coo2 - - ~2 0 2 '

6. Evaluation of beam position and size and
We are now in a position to evaluate the local
center-of-mass displacement, ~, and the rms beam n i) + AI I (66)
width, & which were defined in (18) and (19). ~Q(T+r' A-~ ~ ~ -2~,
CO 0 - - n 2 ~

6.1. CASE (a): A2/A 2 < 1 then (63) and (64) become
Using (35), (36) (54), and (60), and introducing the a
~ (67)
quantities 0)2 -- ~2 ~r~2 '
A4 = (l/N) E (mR
2 --m0)
2 2, (61) rt , ) ~ Az
k=0 ×
6 ~ a fif2(r+r) AZ~o
A~ = (I/N) ~ (CO2--co~)2e-2~k/U, (62)

so that A s is the rms spread in the square of the fre- x
[ 1 ~0"7/)
r' ( 1 -F °92o - .ll
~2 (22/1"

quency shift, one obtains
Eq. (67) shows that, when the rate of change of co
~_ a ia ( ~ ,)+x and A and the frequency spread are such as to satisfy
co~ -- (tTf2)z \fif2(r + r ) (65) and (66), the local beam center of mass is essen-
tially not influenced by the resonance crossing (but
only by the proximity of the coherent integral reso-
x l n(r+r') 1 ~ coo2 -- ~2 f22 x nance). However, and under the same conditions, the
crossing of the resonance can lead to an increase of
x exp{i[co2+A~-fi2122+(r+r')(t-t°)]2-i41 A---~4~ beam size, as shown by (68).
4 iiO(r + r') A 4' It is interesting to compare these results with the
increase in amplitude of a single particle crossing the
(63) resonance. For a single particle the amplitude after

crossing is given by Assuming, again, that condition (50) is satisfied, (69)
and (30) become, to a good approximation,
xs = a ( r r ~ ~.
\~f2r} ~ O, (71)

Taking, for the sake of comparison, r' = 0 the increase
in beam size, ~, is seen to be equal to xs multiplied by l z~ , /t (72)
the factor AE/A2, i.e., the ratio of frequency spread to
frequency shift. These last results are equivalent to saying that each
As a numerical example consider the case of an particle behaves as a single particle, so that, because of
ERA with parameters f2 = 10 a° sec -1, R (ring radius) the large frequency difference between particles, their
3 cm, cOo-f2 ~ 2 x 10-2 f2, r = 2O)o(do)o/dt ) ~ 2 x center of mass averages to zero and one gets essen-
(A/A) ~ 10 - 1 , a = - R I 2 2 ( A B / B )
X 1024 sec - 3 , r ' = 0, tially only a beam widening. But the width increase
3 x 102°(AB/B)sec -2, ~ = 1. The quantity r cor- is larger, by a factor of A2/Ag, than that obtained in
responds to a case such that e)o/f2 changes by 0.1 in case (a).
10 psec, a value typical for the ERA. One sees that (65)
and (66) are satisfied for these parameters. From (67)
We are indebted to G. Lambertson, L. Smith, and
and (68) one has
S. Van der Meer for stimulating conversations and,
in particular, to L. Smith who pointed out an error
~ 30(AB/B)cm, in the original manuscript.
6 ~ 37.5 (AB/B) cm,
so that a value of AB/B less than 10 -a should suffice 1 ) V. P. Sarantseev, Proc. Conf. Particle accelerator, Washington,
to keep the effect of the resonance crossing within D.C., IEEE Trans. Nucl. Sci. NS-16, no. 3 (1969) 15;
tolerable limits. D. Keefe, Proc. Conf. Particle accelerator, Washington,
D. C., IEEE Trans. Nucl. Sci. NS-16, no. 3 (1969) 25.
2) A. Garren et al., Nucl. Instr. and Meth. 18-19 (1962) 525.
6.2. CASE (b): A2/A 2 >> 1 3) j. D. Lawson, Syrnp. Electron ring accelerators, UCRL-18103
(Lawrence Radiation Laboratory, Berkeley, California,
From (52), (40), and (41) and from (18) and (19), I968) p. 301.
we have 4) K. R. Symon, Symp. Electron ring accelerators, UCRL-18103
(Lawrence Radiation Laboratory, Berkeley, California, 1968)
p. 304.
~- , 2 exp {i [(Dk+p(t--to))2/2p]} 2) L.J. Laslett, ERA Internal Report (Lawrence Radiation
= a\iif2(r+r)] ~[ k Laboratory, Berkeley, California, 1969) unpublished.
6) A. M. Sessler, Symp. Electron ring accelerators UCRL-
(69) 18103 (Lawrence Radiation Laboratory, Berkeley, California,
and 1968) p. 11.
7) A . G . Bonch-Osmolowskii et al., Preprint JINR-PG-4135
62 = a2zc { 1 ~exp[i(Dk+P(t--to).~2 (Joint Institute for Nuclear Research, Dubna, USSR, 1968).
iiQ-~r+r') 1--NSk,h \ ~/(2p) / 8) See ref. 7, and also L. J. Laslett, Internal Report ERAN-51
(Lawrence Radiation Laboratory, Berkeley, California,
9) S. Van der Meer, CERN Preprint ISR-PO-169-57 (1969).
(70) lo) V. P. Sarantseev et al., in Proc. VII Int. Conf. High energy
\ ~/(2p) / j j accelerators (Yerevan, USSR, 1969) to be published.