._
_I
.
Submitted
to:
workshop
on Iilcreasing the
AGS
Polarization
’
UniverSity
of
Michigan, Nov.
6
-
9,
2002
BNL-71027-2003-CP
Thoughts
And
66Factss’ From The
AGS
Polarized Proton
Runs
During The
1980’s*
Leif
A,hrens
Brookhaveit
National
Lab
Abstract.
This workshop’s focus is on considering ways
for
improving the proton beam
polarization that the AGS delivers to the
WC.
This
talk attempts to review the
first
decade
of
AGS polarization
-
the
1980’s;
to briefly describe some aspects
of
the machine situation, the
depolarization avoidance strategies employed and the success achieved in AGS from the
perspective
of
one of those involved.
THE
GENERAL CORRECTING SCHEMES
OF
THE
SO’S
First a very brief description of the
1980’s
polarized proton setup will be given.
Reference
1
goes through this in detail. Differences with the situation in
2003
will be
mentioned as we
go.
The intensity of the polarized beam delivered to the AGS directly
from the
200
MeV Linac was at most 2~10’~ protons per AGS cycle.
(In
2003
we will
have more than
10
times this intensity, now coming in from the Booster at about
1.5
GeV.) The initial polarization was
75%;
we will have
80%
this year. The initial
transverse emittances
of the beam, in AGS,, were about 10n
mmmr
normalized
95%.
The need for “polarization-based” measurements will be indicated occasionally in the
following. Machine setup based on polarization measurements are much more time
consuming and difficult than either dead reckoned setups or beam-based setups that
only need beam properties such as intensity, betatron tune, and transverse emittance,
which get more difficult in that order.
Intrinsic Resonances
Intrinsic resonances were handled by pulsing very fast (rise time less than the time
for the protons to make one turn around the AGS) ferrite quadrupoles, located
symmetrically in each of the twelve AGS superperiods at positions where the vertical
betatron function is a maximum (22m), and the horizontal a minimum (1Om). Only ten
quads were actually used, for reasons of cast. Hybrid pulsing systems were built for
the resonances occurring at
0+,
12+, 36-, 24,+
and
48-,
where the code being used here
is e.g.
“O+”
means the resonance when the spin tune
(Gy)
is equal to the integer
0
“plus” the vertical betatron tune (which is close to
8.75
in AGS). The resonance at
24-
was too weak to require a pulse. The resonance at
36+
(Gy
=
44.75)
was judged too
strong for jumping and strong enough to rely on spin flipping. The strengths of these
*
Work
performed under the auspices
of
the
U.S.
Department
of
Eniergy
resonances were well predicted by Courant and Ruth. The most aggressive pulsing
was at 36-, with the fast (1.6ysec rise time) components requiring 12kV, and the slow
part (20 psec) requiring 2 kV. The timing of these pulses during the acceleration ramp
was derived from measurements of the field in the main magnets using the AGS
“Gauss Clock”. Although the timing setups could be “dead reckoned”, the clock was
neither accurate nor stable enough to avoid using polarization measurements
to
both
carry out the initial setup and to occasionally confirm that things were still ok later in a
run. The situation has improved since the 80’s. We have a new clock, a new orbit
measuring system, and claim (not tested) an accuracy that would allow such
a
setup to
be marginally possible without any polarization checks.
A second system required for the intrinsic tune jumping involved the normal AGS
slow quadrupoles. These were used to shift the tunes around each jump to allow more
tune headroom for the jump. The timing requirements were mild.
Imperfection Resonances
The imperfection resonances were corrected by making the machine equilibrium
orbit perfect for the relevant driving harmonic. There are more than 30 resonances
below 18.5 GeV. Each must be set up. The timing requirements
are
loose enough to be
learned from the Gauss clock. Learning the two strength parameters
-
i.e.
the
amplitude and the phase of the correction
-
is completely polarization based. The
existing system of vertical correction dipoles, eight per superperiod, was connected to
stronger pulsing power supplies using a new control system. The system allowed the
harmonic corrections to change with time in order to maximize the current available to
the one relevant for the spin survival at that moment in the acceleration cycle. This
system was pushed to its limits in order to cope with imperfections encountered below
Gy
of 42. Some of the stronger of these imperfections were ultimately corrected by
flipping rather than by correcting since the resulting machine was more stable and the
tuning was simple. At highest energies the strength of the correction system was
marginal to correct and too weak to flip. Aside from remembering the complexity of
the system and the associated setup, this discussion is no longer particularly relevant.
The solenoidal Partial Siberian Snake replaced this entire system in the go’s, flipping
all
the imperfections but also introducing its own interesting problems associated with
relatively strong and uncorrectable betatron coupling which then strengthens other
higher order resonances.
SPECIFIC DESCRIPTION
OF
THE
MAJOR
1980’s
RUNS
The 1980’s polarized proton accelerator activity at the AGS can be described by
three running periods, a pre-commissioning run in June of 1984, a commissioning run
in February of 1986, and a production run in January of 1988. During the 1984 run
beam was accelerated up to
16.5
GeV/c or Gy=31.5 where
40%
polarization was
achieved. The higher energy pulsing systems not yet available, the lower energy
systems were “being commissioned”.
The
1986
Pollarized
Run
The 1986 run produced the highest energy polarized beam,
45%
at
21.7
GeVk or
w41.5
of the 80’s. (This
run
is described
in
reference
1.)
The acceleration rate
available in the AGS was the nominal
2
Tesldsec from the Siemens motor generator
set (which will not be true for the 1988 nm). During the course of the ‘86 run the
correction systems described above were pushed to their limits. Their reliability was
better understood, and diagnostic systems to help “Operations”
-
which for this run
was mostly physicists
-
respond to problems were being developed.
Maintaining the beam intensity and
if
possible the emittance through the intrinsic
jumping was a major effort. The fast quads produce an inherently nonadiabatic change
to the orbit betatron motion. Though some concern was expressed over the
implications of this for emittance growth clue to the implied changes to the machine
beta functions, we did not worry about putting the beam on the axis of the quads
during the jumps, in particular in the horizointal plane.
As
a result, each pulsing excited
significant oscillations in both the vertical, and the horizontal planes. The shifting of
the tunes to provide maximum headroom. for the fast tune shift complicated this
situation because the horizontal and vertical tunes would sometimes cross slowly
-
adiabatically
-
before and after each tune jump. As a result the beam transverse
emittances were systematically swapped (slow quad setup), increased (fast quad
kicks), and then swapped again (slow quad recovery) around the intrinsic jump. This
was not understood at the time though the fascinating transverse size evolution as seen
by the ionization profile monitor
(IPM)
was known. Figure
1
shows the trajectories in
tune space of the vertical betatron tune.
SLOW
SHlFf
-----.-
FAST
SFtfFT
-
72
t
QV
QV
0
4
Qv
QV
r
8.3
t
FIGURE
1.
AGS
Horizontal Betatron tunes as set
up
during the
1986
run.
Figure
2
gives one snapshot of the vertical emittance evolution. These figures are
from a talk given in
1987
(ref
2).
There is no discussion of the behavior of the
horizontal tunes, which must also jump, though only half as far as the vertical and of
course are involved in the emittance swap.
1
€
E
-
I
*
#
0,
#
u?-
-
+
r.
W
r‘
0
TOO
ZOC)
300
4-00
500
TIME
?N
ACCEZLERATION
CYCLE
(mserc]
FIGURE
2.
Typical vertical normalized emittance (95%) during the 1986 run.
The solution given in these figures was arrived at empii-ically, to first of all allow
beam survival, and then to minimize emittance growth. We well understood that
emittance growth was a bad thing for the intrinsics. Note that for the
O+
jump, the
vertical tune crossed the half-integer line at
8.5,
actually crossed it twice, once very
fast going down, and then more slowly on the recovery side of the fast pulse. That this
line can be crossed is consistent with earlier
AGS
experience, at least at injection.
I-
+
FIGURE
3.
Polarization (figure
40
from ref 1) measured using extracted
beam
for
the
1986 run.
Figure 3, again from reference
1,
gives measurements of the polarization using a
polarimeter in an extracted beam line for the
1986
run. The measurement at
13.3
GeV/c corresponds to extraction at Gy=25.5, and at 16.5GeV/c to v31.5. The
drop
between these points is associated with the region near the strong
36-
resonance. Aside
from this and despite or because of the large emittance growth, there is no measurable
polarization
loss
later in the cycle.
The
1988
Polarized
Run
The 1988 run was explicitly to be a prolduction run. We had 2.5 weeks to tune up
the machine, and then
3
weeks for physics. Extraction at 18.5 GeV/c (Gy35.5)
avoided the higher energy trouble with the imperfection corrections and the
48-
intiinsic. Interpretation of the results of the run are complicated by a failure with the
Siemens motor-generator set, which forced the run to occur with the lower
acceleration rate associated with the backup power supply, the Westinghouse.
(If
you
think
I
have slipped a decade and am describing the 2002 run, you
are
wrong, but your
confusion is understandable.
So
the resonances are all stronger because the
acceleration rate is cut in half. In some ways this made the machine setup easier.
Tolerances and adjusting room for timing tihe imperfection resonance corrections and
the slow quadrupole tune shifts were relaxed by this factor of two. We had learned
enough about the cause for the emittance growth in
'86
to fix it. Both the vertical
positions (beam based quad repositioning) and the horizontal position (care for the
radial position) were corrected.
As
a result the emittance growth essentially
disappeared. Figure
4,
from reference 3, compares the emittance growth in the two
runs.
80
n
W
&
60
E
U
w
40
a
E
ZE
W
J
20
<
0
1-
E
w
=-0
>
0
-1-
01
>
W
'1
I
OmQaQg
~
0.0
5.0
10.0
15.0
20.0
MOMENTLJ
M
(GeV/c)
FIGURE
4.
Comparison
of
the vertical emittance
growth
(normalized, 95%) between 1986 and 1988.