Resonant multi-turn extraction: principle and experiments · Massimo Giovannozzi RAL - July 5th...
Transcript of Resonant multi-turn extraction: principle and experiments · Massimo Giovannozzi RAL - July 5th...
-
Massimo Giovannozzi RAL - July 5th 2013 1
Manipulation of transverse beam distribution in circular accelerators: beam splitting by particles’ trapping
into resonance islands
M. Giovannozzi
Summary:
Introduction
Present multi-turn extraction
New multi-turn extraction (MTE)
Other applications
Measurement results
Conclusions
-
Massimo Giovannozzi RAL - July 5th 2013 2
Introduction - I
Three approaches are normally used to extract beam from a circular machine:
Fast Extraction (one turn)
A kicker (fast dipole) displaces the beam from nominal closed orbit.
A septum magnet deflects displaced beam towards transfer line.
This extraction can be used both to transfer beam towards a ring or an experimental area.
Slow Extraction (millions turns)
The separatix of the third-order resonance increases particles’ amplitude until they jump beyond the septum.
The tune is changed to shrink the stable region, thus pushing the particles towards larger amplitudes.
This extraction is used to transfer beam towards an experimental area. What is in between?
Multi-tu
rn!
-
Massimo Giovannozzi RAL - July 5th 2013 3
Introduction - II
Multi-turn extraction
The beam has to be “manipulated” to increase the effective length beyond the machine circumference.
This extraction mode is used to transfer beam between circular machines.
AT CERN this mode is used to transfer the proton beam between PS and SPS. In the SPS the beam is used for
Fixed Target physics (broad sense)
Neutrino experiments (until 1998)
CERN Neutrino to Gran Sasso (CNGS) (from 2006)
These beams are high-intensity (about 3×1013 p in the PS).
-
Massimo Giovannozzi RAL - July 5th 2013 4
Present multi-turn extraction – I
First PS batch Second PS batch Gap for kicker
CSPS = 11 CPS
PS PS
SPS circumference
Beam current transformer in the PS/SPS transfer line
1 2 3 4 5 (total spill duration 0.010 ms)
-
Massimo Giovannozzi RAL - July 5th 2013 5
Electrostatic septum blade
Present multi-turn extraction – II
Length
Kicker
stre
ngth
Four turns
Fifth turn
X
X’
1 3 5
2
4 Slow bump
Slow bump
Electrostatic septum (beam shaving)
Extraction septum
Kicker magnets used to generate a closed orbit bump around electrostatic septum
Extraction line
Efield=0
Efield≠0
-
Massimo Giovannozzi RAL - July 5th 2013 6
Present multi-turn extraction –III
The main drawbacks of the present scheme are: Losses (about 15% of total intensity) are unavoidable due to the presence of the electrostatic septum used to slice the beam.
The electrostatic septum is irradiated. This poses problems for hands-on maintenance.
The phase space matching is not optimal (the various slices have “fancy shapes”), thus inducing betatronic mismatch in the receiving machine, i.e. emittance blow-up.
The slices have different emittances and optical parameters.
-
Massimo Giovannozzi RAL - July 5th 2013 7
Novel multi-turn extraction – I
The main ingredients of the novel extraction:
The beam splitting is not performed using a mechanical device, thus avoiding losses. Indeed, the beam is separated in the transverse phase space using
Nonlinear magnetic elements (sextupoles ad octupoles) to create stable islands. Slow (adiabatic) tune-variation to cross an appropriate resonance.
This approach has the following beneficial effects: Losses are reduced (virtually to zero). The phase space matching is improved with respect to the present situation. The beamlets have the same emittance and optical parameters.
-
Massimo Giovannozzi RAL - July 5th 2013 8
Model used in numerical simulations
Standard approach: nonlinear elements represented as a single kick at the same location in the ring (Hénon-like polynomial maps).
Vertical motion neglected.
Normalised (adimensional co-ordinates).
3
2
2
2
3 x
K
K
2 3
1
ˆ ˆ
ˆ ˆ ˆ ˆ' 'n n
X X
X X X X
R
Quadrupoles Sextupole Octupole
The linear tune is time-dependent
-
Massimo Giovannozzi RAL - July 5th 2013 9
Novel multi-turn extraction – II
Right: intermediate phase space topology. Islands are created near the centre.
Bottom: final phase space topology. Islands are separated to allow extraction.
Left: initial phase space topology. No islands.
-
Massimo Giovannozzi RAL - July 5th 2013 10
Novel multi-turn extraction - III
Tune variation
Phase space portrait
Simulation parameters:
Hénon-like map (i.e. 2D polynomial – degree 3 - mapping) representing a FODO cell with sextupole and octupole
-
Massimo Giovannozzi RAL - July 5th 2013 11
Novel multi-turn extraction – IV
Final stage after 20000 turns (about 42 ms for CERN PS)
About 6 cm in physical space
Slow (few thousand turns) bump first (closed distortion of the periodic orbit)
Fast (less than one turn) bump afterwards (closed distortion of periodic orbit)
Bfield ≠ 0 Bfield = 0 At the septum location
-
Massimo Giovannozzi Oxford - July 4th 2013 12
Novel multi-turn extraction - V
The original goal of this study was to find a replacement to the present Continuous Transfer used at CERN for the high-intensity proton beams. However the novel technique proved to be useful also in different context, e.g.
The same approach can be applied for multi-turn injection (time-reversal property of the physics involved). multi-turn extraction over a different number of turns can be designed, provided the appropriate resonance is used. Multiple multi-turn extractions could be considered, e.g. to extract the beam remaining in the central part of phase space.
-
Massimo Giovannozzi RAL - July 5th 2013 13
Novel multi-turn extraction with other resonances - I
Tune variation
Phase space portrait
Simulation parameters:
Hénon-like map with sextupole and octupole
The third-order resonance is used, thus giving a three-turn extraction
-
Massimo Giovannozzi RAL - July 5th 2013 14
Novel multi-turn extraction with other resonances - II
The fifth-order resonance is used, thus giving a six-turn extraction
The second-order resonance is used, thus giving a two-turn extraction
-
Massimo Giovannozzi RAL - July 5th 2013 15
Non-linear elements can be used to make the fourth-order resonance unstable by canceling the amplitude detuning.
Possible applications: Four-turn extraction.
Deplete the core region to
achieve a better intensity
sharing.
Novel multi-turn extraction: 4th order unstable resonance - new application!
Master thesis Diego Quatraro Master thesis Diego Quatraro
-
Massimo Giovannozzi RAL - July 5th 2013 16
Novel multi-turn injection: new application!
Simulation parameters:
Third-order polynomial map representing a FODO cell with sextupole and octupole
The fourth-order resonance is used for a four-turn injection
Tune variation
Phase space portrait
Efficient method to generate hollow beams! Efficient method to generate hollow beams!
-
Massimo Giovannozzi RAL - July 5th 2013 17
Novel multi-turn injection: new application!
Possible applications: transverse shaping of beam distribution.
M. Giovannozzi, J. Morel, PRST-AB, 10, 034001 (2007) M. Giovannozzi, J. Morel, PRST-AB, 10, 034001 (2007)
“Standard” hollow beam distribution
“Flat” beam distribution obtained by injecting a fifth turn in the centre.
-
The capture process - I
Quantitative analysis needed: To control sharing between core and beamlets.
To control the emittance sharing.
Adiabatic theory is the key concept: Probability trapping is proportional to the speed of variation of the island’s surface.
It is possible to account for the loss of adiabaticity close to the separatrix.
2D case seems under control.
The 4D case requires the usual conditions: weak coupling between the two degrees of freedom
no low order resonances in the vertical plane
Massimo Giovannozzi RAL - July 5th 2013 18
-
Massimo Giovannozzi RAL - July 5th 2013 19
The capture process - II Initial gaussian distribution Initial gaussian distribution Particles in island 2 Particles in island 2 Particles in island 4 Particles in island 4
Particles in island 6 Particles in island 6 Particles in island 8 Particles in island 8 Particles in beam core Particles in beam core
-
The capture process - III
Under these conditions: Scaling law for the size of the adiabatic region is well-reproduced.
Trapping probability is well-described.
Massimo Giovannozzi RAL - July 5th 2013 20
Hénon-like model Hénon-like model
Size of non-adiabatic region Size of non-adiabatic region
Pendulum-like model Pendulum-like model
-
Massimo Giovannozzi RAL - July 5th 2013 21
Analytical computation of island’s parameters
Using perturbative theory (normal forms) it is possible to derive analytical estimate of island’s parameters.
2 0,32 u
Distance of fixed points from origin of phase space
2
0,3
2
4 u
Distance between separatices
Normalised phase space
2
0,3
2
16 u
Island’s surface
2
sec 2 0,34 u
Distance from resonance
Secondary frequency
-
Massimo Giovannozzi RAL - July 5th 2013 22
Experimental results
The pencil beam is kicked into the islands producing a strong coherent signal (filamentation is suppressed). Initial wiggles represent beam oscillations around the islands’ centre.
Normalized Coordinates
-15 -5 5 15-15
15
X
-5
5
H Signal PU1: blue - PU2: red
50 150 250 350 450 550 650 750 850 950 1050 1150 1250 1350 1450 1550 1650 1750 1850 1950
-20-10
01020
Turn
y = -0.9211 x + 0.0398
y = -0.7566 x + 0.0408
0.020
0.025
0.030
0.035
0.040
0.045
0.000 0.002 0.004 0.006 0.008 0.010 0.012
Linear invariant
Sec
on
dary
fre
qu
ency
Measurement results
Numerical simulations
Fit Pesato
Linear (Numericalsimulations)
Measured detuning inside an island compared to numerical simulations.
Measured detuning inside an island compared to numerical simulations.
-
Massimo Giovannozzi RAL - July 5th 2013 23
Data analysis and beam parameters
The wire scanner is the key instrument for these studies.
Raw data are stored for off-line analysis.
Five Gaussians are fitted to the measured profiles to estimate beam parameters of five beamlets.
Beam parameters
Intensity *H(s)/*V(s)
Low-intensity pencil beam
5×1011 2.3/ 1.3
Low-intensity large H emittance
5×1011 6.2/ 1.6
High intensity beam 6×1012 9.4/ 6.4
-
Massimo Giovannozzi RAL - July 5th 2013 24
TimeQ
h
t
Qh=6.25
Profile measurements
180 ms Reversibility
Before
After Fast crossing: t 5 ms.
Trapped particles
Trapped particles
Slow crossing: t 90 ms.
Before
After
Large Gaussian tails
No difference observed if t > 20-40 ms. No difference observed if t > 20-40 ms.
-
Massimo Giovannozzi RAL - July 5th 2013 25
Influence of octupole strength Octupole action
Island size.
Detuning with amplitude.
Problems with the fit
-
Massimo Giovannozzi RAL - July 5th 2013 26
Operational implementation - I
Sextupoles and octupoles are used to
Generate stable islands
Control size/position of islands
Control linear chromaticity
Control non-linear coupling (using an additional set of octupoles, normally used to combat instabilities)
yxy
yxx
JhJhQ
JhJhQ
2,01,1
1,10,2
h2,0 -> detuning with amplitude (H-plane) -> x2 K3
h1,1 -> non-linear coupling -> xy K3
h0,2 -> detuning with amplitude (V-plane) -> y2 K3
-
Massimo Giovannozzi RAL - July 5th 2013 27
Operational implementation - II
-600
-400
-200
0
200
400
600
0.00.10.20.30.40.50.60.70.80.91.0
0.0 0.2 0.4 0.6 0.8 1.0 1.2
Cu
rren
ts (A
)
B-f
ield
(T)
Cycle time (s)
B-field Sextupole 39Sextupole 55 Octupole 39Octupole 55 Octupoles
-600
-400
-200
0
200
400
600
0.00.10.20.30.40.50.60.70.80.91.0
0.72 0.74 0.76 0.78 0.80 0.82
Cu
rren
ts (A
)
B-f
ield
(T)
Cycle time (s)
B-field Sextupole 39Sextupole 55 Octupole 39Octupole 55 Octupoles
Injection: 0.170 s
Extraction: 0.835 s
-
Transverse dynamics - III
Massimo Giovannozzi RAL - July 5th 2013 28
Phase space at septum
0.72 0.74 0.76 0.78 0.80 0.82
-3000
-2000
-1000
0
1000
2000
3000
0.00
0.05
0.10
0.15
0.20
0.25
0.72 0.74 0.76 0.78 0.80 0.82
Seco
nd
ord
er
chro
mat
icit
y,
de
tun
ing,
no
n-l
ine
ar c
ou
plin
g
Frac
tio
nal
tun
e an
d c
hro
mat
icit
y
Cycle time (s)
Qx Qx'/Qx
Qx'' h2,0
h1,1
-
Transverse dynamics - IV
Massimo Giovannozzi RAL - July 5th 2013 29
To achieve good sharing of beam intensity between islands and core various parameters optimised (h1,1 is rather crucial).
12.0
12.5
13.0
13.5
14.0
14.5
15.0
15.5
16.0
16.5
17.0
0 100 200 300 400 500 600
Fra
cti
on
of
tra
pp
ed
pa
rti
cle
[%
]
Octupole current [A]
h11=0
h11=100
h11=200
h11=300
h11=500
Last bit achieved by exciting the beam using the transverse damper.
Current in octupoles 39/55 (A)
-
30 RAL - July 5th 2013 Massimo Giovannozzi
y = -0.0185x + 0.896R² = 0.0026
y = -0.1852x + 4.3783R² = 0.2631
0.0
0.5
1.0
1.5
2.0
2.5
3.0
3.5
12 14 16 18 20
Rat
io o
f co
re e
mit
tan
ce (
afte
r an
d b
efo
re s
plit
tin
g)
Fraction of particles trapped in islands (%)
Improving capture: beam excitation
Effect of transverse excitation
revprev TtNtfAtx 25.016sin'
Noise component
-
Massimo Giovannozzi RAL - July 5th 2013 31
Evolution of beam distribution
Horizontal beam profiles in section 54 have been taken during the capture process (total intensity ~2.1×1013).
-
Trapping performance
32 RAL - July 5th 2013 Massimo Giovannozzi
0 50 100 150 200 250 300
37 nsbin
0
2.0x1010
6.0x1010
1.0x1011
1.4x1011
Inte
nsity
0 50 100 150 200 250 300
37 nsbin
0
2.0x1010
6.0x1010
1.0x1011
1.4x1011
Inte
nsity
12 14 16 18 20
200
400
600
800
1000
Trapping (%)
Fre
quency
5000 10000 15000 20000 25000
Time elapsed from May 1st 15:00 (s)�
12
14
16
18
20
Tra
ppin
g (
%)
20% trapping 18% trapping
5 PS turns
-
Extraction efficiency
33 RAL - July 5th 2013 Massimo Giovannozzi
Regular fluctuations in the extraction efficiency are also observed and seem well correlated to spill fluctuations.
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
0
100
200
300
400
500
600
700
90
.0
90
.6
91
.2
91
.8
92
.4
93
.0
93
.6
94
.2
94
.8
95
.4
96
.0
96
.6
97
.2
97
.8
98
.4
99
.0
99
.6
100.
2
100.
8
Fre
qu
en
cy
Extraction efficiency (%)
Frequency Cumulative %
90
91
92
93
94
95
96
97
98
99
100
14:24 15:36 16:48 18:00 19:12 20:24 21:36 22:48
Ext
ract
ion
eff
icie
ncy
(%
)
Time on May 1st
Distribution of extraction efficiency is peaked at about 98% (NB: the beam is debunched at extraction! Unavoidable beam losses are estimated at about 1-2%)
CT extraction efficiency is about 93%
-
CT vs. MTE: extraction losses
For the same extracted intensity, the CT features more losses, about the double, compared to MTE. The CT losses are spread around the ring whereas for MTE the losses are more concentrated on the extraction septum as anticipated in the MTE Design Report.
CT MTE
34 RAL - July 5th 2013 Massimo Giovannozzi
-
First observations of intensity-dependent effects - I
Usual process for trapping the beam into stable islands.
Varying parameter: total beam intensity.
Massimo Giovannozzi RAL - July 5th 2013 35
Beamlets are moving with intensity!
-
First observations of intensity-dependent effects - II
The observed effect can be explained with an intensity-dependent shift of the single-particle tune.
Ntotal is expressed in units of 1010 p.
Possible sources: Image effects.
Direct effects (interaction between beamlets).
Massimo Giovannozzi RAL - July 5th 2013 36
Consider projection effect
No change in beamlets size
-
Massimo Giovannozzi RAL - July 5th 2013 37
Summary and Outlook - I
A novel multi-turn extraction is studied since a few years: it allows manipulating transverse emittances in a synchrotron! First MTE beam delivered to the SPS by mid-September 2009 (about 1.5×1013 p/extraction). Equal sharing for islands/core achieved by the end of 2009. In February 2010 the commissioning was resumed. High intensity beam was extracted (about 2.1×1013 p/extraction) with record intensity 2.6×1013 p/extraction. 2010 physics run at SPS was started using MTE beam. Open issues:
Variation of the fraction of particles trapped in islands. Activation of extraction septum due to the longitudinal structure of the beam (de-bunched as needed by the SPS).
New optimised extraction scheme to avoid losses on extraction septum -> beam commissioning in 2014.
-
Massimo Giovannozzi RAL - July 5th 2013 38
Summary and Outlook - II
The same principle can be used for injection. Transverse shaping is possible, i.e. generation of hollow bunches. Lines of research:
Complete study of splitting process in 2D and extend to 4D. Study trapping in 4D, i.e., manipulations in x-y plane. Study intensity-dependent effects. Study coupling with longitudinal dynamics (strong non-linear elements generate non-linear terms in the momentum compaction)
And many more!
-
Massimo Giovannozzi RAL - July 5th 2013 39
Selected references
S. Gilardoni, M. Giovannozzi, C. Hernalsteens (2013). “First observations of intensity-dependent effects for transversely split beams during multiturn extraction studies at the CERN Proton Synchrotron”, Phys. Rev. ST Accel. Beams . M. Giovannozzi, D. Quatraro, G. Turchetti, (2009). “Generating unstable resonances for extraction schemes based on transverse splitting”, Phys. Rev. ST Accel. Beams 12 024003. A. Franchi, S. Gilardoni, M. Giovannozzi, (2009). “Progresses in the studies of adiabatic splitting of charged particle beams by crossing nonlinear resonances”, Phys. Rev. ST Accel. Beams 12 014001. M. Giovannozzi and J. Morel, (2007). “Principle and analysis of multiturn injection using stable islands of transverse phase space”, Phys. Rev. ST Accel. Beams 10 034001. S. Gilardoni, M. Giovannozzi, M. Martini, E. Métral, P. Scaramuzzi, R. Steerenberg, A.-S. Müller, (2006). “Experimental evidence of adiabatic splitting of charged particle beams using stable islands of transverse phase space”, Phys. Rev. ST Accel. Beams 9 104001. R. Cappi, M. Giovannozzi, (2003). “Multi-turn Extraction and Injection by Means of Adiabatic Capture in Stable Islands of Phase Space”, Phys. Rev. ST Accel. Beams 7 024001. R. Cappi, M. Giovannozzi, (2002). “Novel Method for Multi-Turn Extraction: Trapping Charged Particles in Islands of Phase Space”, Phys. Rev. Lett. 88 104801. The same principle can be used for injection.