BNL-41950
• ...
8th International Symposium on High Energy Spin Physics, Minneapolis, MN,
9/12-17/88.
OPERATION OF THE AGS POLARIZED BEAM*
BNL
41950
Lelf A. Ahrens
DE89 005
*43
Rrookhaven National Laboratory, Upton, NY 11973
INTRODUCTION
A polarized proton physics run took place during January,
1988,
at the Brookhaven
AGS*
It is the purpose of this paper to
review the tune-up period preceding chat run. This was the third
such run at the AGS, the others occurred in June of 1984 and Febru-
ary of 1986. Some comparisons will be drawn among these. A
thorough review of the history and hardware associated with the
acceleration of polarized protons at the AGS can be found in the
proceedings of the last meeting of this group at Protvino
1
and will
not be repeated here.
THE 1988 RUN
The primary objective for the 1988 run was to obtain a beam
suitable for polarized proton physics as rapidly as possible. To
maintain polarization 30 imperfection and 5 intrinsic depolarizing
resonances had to be dealt with. After about 2-1/2 weeks of round-
the-clock tuning effort a beam with average polarization of 43%,
intensity 1-2 x 10
10
protons per 2.8 second cycle, at the extrac-
tion momentum of 18*5 GeV/c was achieved. Tuning necessarily con-
tinued during the three week physics run which followed. Over this
period, the polarization fluctuated from just below 403 to just
above 50%.
The conditions for the 1988 run differed from earlier runs
primarily in that the acceleration rate was lower by about a factor
of 0.6 due to problems with the main ring magnet power supply.
This reduction in crossing speed enhanced all the depolarizing
resonances.
On the other hand, it resulted in more time between
resonances to adjust magnetic corrections. Prior to the '88 run a
vertical realignment of the main ring magnets was carried out using
a more sophisticated procedure for analyzing the survey data, and
taking more sets of measurements than for similar surveys before
the earlier
runs.
In addition, and unlike previous runs a
hori-
zontal realignment was accomplished. Particular attention was paid
to centering the ferrite quadrupoles used in jumping the intrinsic
resonances on the actual beam equilibrium orbit in both the verti-
cal and horizontal planes*
Before discussing the resonances and their correction, a brief
summary will be given of the performance of the other polarized
proton acceleration systems. The polarized proton source, which
*Work performed under the auspices of the U.S. Department of
Energy.
wsra
I'.:,
(_•":.-.'.'.
•
obtains polarized H through charge exchange between a neutral
cesium beam and a neutral polarized proton beam routinely provided
a 500 Ps pulse of 25 uk current. Acceleration to 750 keV was ac-
complished using an RFQ with about a 75% transmission efficiency.
By the time the beam arrived at the 200 MeV end of the Linac, the
intensity was about 10 uA. At this point, the polarization was
periodically monitored by scattering from a carbon target. The
polarization was extremely stable at 78 ± 2Z. Of the 3 x 10
l0
protons injected into the synchrotron, about 50Z survived to full
energy. Capture and acceleration require more sensitive electronic
pickups than for normal (10
13
protons per pulse) running but that
is a solved problem at this intensity level.
A polarimeter located in the main synchrotron ring provides
the primary signal for resonance tuning. This "internal" polarirae-
ter carried a 6 mil diameter (0.15 mm) nylon fish line this year (4
mil in the
past).
The line both spools (1 meter/sec) and flips
into the circulating beam when a measurement is desired. The
polarization measurement is derived from the asymmetry measured in
two symmetric counter telescopes monitoring the scatter from the
string. This yields a relative measure of polarisation. The 200
MeV measure and a measure after extraction from the synchrotron
(18.5 GeV/c) provide two points where the absolute polarization is
known.
THE RESONANCES
The bulk of the effort and the rest of this report will deal
with the correcting of the depolarizing resonances* For resonance
tuning, the polarization is measured using the internal polarimeter
on a fixed field porch set slightly above the resonance in ques-
tion. When all resonances below the porch momentum have been cor-
rected, the porch is moved up in a large enough momentum step to
allow a few resonances to be corrected, but not such a large step
that the initial (untuned) polarization on the new porch is negli-
gible.
In practice we moved in ten steps from 0*7 GeV/c to 18.5
GeV/c in tuning the 35 resonances, taking larger steps in regions
known to contain only weak resonances. The resulting plot of
polarization at measuring porch against time is a somewhat distort-
ed saw tooth curve (Figure 1). The maximum polarization gradually
declined with increasing momentum, but without clear indication of
the structure seen in the past in the region of GY • 36 - v and
GY - 27.
The depolarizing resonance condition for the intrinsic reson-
ances is given by GY • nP - v where Y " ^n^
B
o» ** *
8 t
*
ie
P
roton
anomalous magnetic moment
(1«7§),
P is the periodicity of the
machine (12 for AGS), and n is any integer. The horizontal field
driving the spin is just that field responsible for containing the
beam vertically in the beam pipe and hence the resonance name.
Five such resonant tunes cross the machine tune in accelerating to
POLARIZATION
vs
TIME
ICO
15
IS 17 18 19 20 21 22 23 24 29 30 31 1
Figure
1.
Polarization
at the
measurement porch momentum
as a
function
of
time during
the
tune
up
period.
18.5 GeV/c.
As an
example,
we
will concentrate
on the
strongest,
at
n • 3, GY * 36 - v or v » 36 - GY.
Figure
2
shows this reson-
ance crossing,
and
trie effect
of
jumping
the
tune
on the
rate
of
H
9.1
9.0
B.9
8.8
8.7
8.6
8.5
8.4
8.3
8.2
"NORMAL
TUNE
-
-
1
1 1
y
/
VERTICAL BETATRON TUNE
AT "36-i>"
3ESONANT TUNE
(-
36-Gr)
(y
- 60)
\
^-
MACHINE TUNE
JUMP
.
N
s
\
( I i I Vi I
-2
2
4 6
TIME
(ms)
10
Figure
2.
Vertical betatron tune (machine
and
resonance)
vs
time.
crossing. The system which produces the tune jump consists of 10
ferrite core quadrupoles and associated power supplies. What is
"tuned" for an intrinsic resonance is the timing of the jump. The
clock used
("Gauss"
clock.) ticks at a rate proportional to the rate
of change of magnetic field in the main ring magnets. The optimum
timing (or jump field) depends on the machine tune and the particle
Y (which is uniquely associated with the magnetic field provided
the orbit circumference or radius is held
fixed).
The timing of
the tune jump demands more accuracy and stability of the Gauss
clock than is required for any other AGS running mode. For the '88
run the optimum setting for the jumps occurred within about 10
Gauss of the predicted clock valve* In addition, there was no
evidence for a loss of polarization at any intrinsic once the
resonance had been tuned. Since the plateau in the curve of final
polarization vs jump time is 10 to 20 Gauss wide, these observa-
tions imply that the calibration and stability of the Gauss clock
over the run was better than one part in one thousand.
The lower acceleration rate for this run mentioned above af-
fected the intrinsics but only in an indirect way. The actual
crossing speed is dominate by the tune jump and so only the distant
tails of the resonance are aware of the lower crossing rate. How-
ever, the curve of v with energy (Figure 2) has a smaller (nega-
tive) slope for lower acceleration rate. The wave form for
v
niachine
vs tirae
(the fast quad jump and recovery) is cast in
concrete (or at least in non-trivial capacitors and resistors) and
so could be not changed for the run. The separation between the
machine tune and the resonance tune just after the jump was less
than for the normal acceleration rate. This pzcblem turned out to
be important only for GY = 24 + v where it showed up as a narrowing
of the corrected plateau.
A second problem associated with the intrinsic corrections
which has plagued us in the past is associated with the very es-
sential non-adiabatic nature of the jump. Although an eoittance
growth of 10Z per jump is expected from the non-adiabatic shifting
of the machine betatron function, in the past ('86) we typically
saw increases averaging 50% per jump and responded by moving the
machine tunes around to empirically find the point of minimum grow-
th.
As mentioned above, in preparation for the 1988 run, the
transverse position of the fast quads was adjusted to center the
quads on the orbit. Further, and perhaps more importantly, we paid
particular attention to holding the machine orbit at the centered
radius during acceleration. The result of all this was a satisfy-
ing reduction in emittance growth (Figure 3) which somewhat reduced
the strength of later intrinsic resonances and improved extraction
efficiency. Freed of the emittance growth problem the slow adjust-
ment of the machine tune prior to a jump could be used as original-
ly designed, namely to increase the jump headroom. By suppressing
the nominal 8.75 vertical tune to 8.55 (at GY - 36 - v)
a
jump of
0.1 units to 8.85 was possible; without the slow adjustment a jump
of even 9.2 units caused beam loss from the v
y
= 9 stopbands. Slow
tune adjustment proved unnecessary at the other intrinsics.
o
c
7
1—
k_
i—
«—.
z
v i.l Ik
<
II.M.
>
30
50
4.0
20
:
0
0.
1
L
i
I
0
1
RON
o
—}
z
i
•
>
o
+
o
o
A
/
o
so©©
o
1
5.0
1
1
I——^~—~^-^^^
QQOQOOQO O*)j
1
1
10.
MOMENTUM
i
>
o
1
+
j
12.
1
0
(GeV/c)
i
>
•O
.
i
/
/
_/
o
o
<
i
15.
>'
o
1
+
Ij
•^
1
I
u
1
j 1988
0
V
1986
MM
20.0
Figure 3. Vertical normalized emittance vs time in the
acceleration cycle for '86 and '88 runs.
finally, the imperfection resonances are discussed. Because
there are so many (~ 30 below 18.5 GeV/c) and all are strong enough
to require tuning, imperfection resonance correction is the primary
activity of the tune-up period. The depolarizing condition is
simply GY * n. The resonance is driven by a horizontal magnetic
field of periodicity n, B(6)
-
B
s
sin(n 0)+ B
c
cos(n 9). Tuning
involves adding by aeans of 95 correction dipoles a magnetic har-
monic to cancel the harmonic initially in the machine.
The
magnets are energized as the machine approaches resonance energy
with the pulse timing and width set using the Gauss clock calibra-
tion to guarantee the resonance is covered.
As in the past imperfection resonances with values of n « CY
near GY « 36
- v
(i.e., 27, 28, and 29), were more effectively
corrected using "beat" harmonics (9, 8, and 7 respectively) than
with the field harmonic directly.
2
These "beat" magnetic harmonics
which are integers near 8.75 result in relatively large vertical
equilibrium orbit distortions of the machine which carry the
