Delivering full longitudinal polarized beam to multiple halls

Delivering full longitudinal polarized beam to multiple halls
Full longitudinal polarization for
multiple halls
or
A Wien for each beam
Jay Benesch
1
Outline
•
•
•
•
•
•
•
•
•
Present injector
Idea and scheduling issue
Present Wien
New Wien model
New Wien with focusing solenoids
Triple Wien
OPERA constraints
Future work?
Conclusion
2
What does the injector contain?
(in order)
1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.
12.
130 keV gun (200 keV 2015)
vertical Wien filter (90 degrees precession)
two spin flipper solenoids for left/right flip with vertical Wien
prebuncher to reduce longitudinal spread in bunch due to space charge effects
aperture 1 of two, defining transverse bunch extent
horizontal Wien, setting final precession angle for all 3 halls, one optimally or a compromise
except at two-hall “magic” energies. All focusing solenoids thereafter are counter-wound for
zero net precession.
aperture 2 of two, finishing transverse bunch definition.
chopper, allowing different currents for three halls – the only place in the injector where the
three beams are separated, at 120o intervals on a 30mm diameter circle. Holes 24o by 6mm
Buncher cavity, again reducing bunch length
Capture cavity, accelerating to 500 keV kinetic energy (incorporate in (12) in 2015)
apertures 3 and 4, transverse beam cleanup, usually no intercepted current
two-cavity cryomodule, accelerating to 6.3 MeV/c momentum (change in 2015)
3
Idea
• The injector group within Accelerator Operations is
working on a new front end with 200 keV gun and a new
cryo-unit incorporating the capture.
• Reza Kazimi and I were chatting about it, discussing the
problems caused by the pre-buncher between the two
Wiens. Reza asked if one could build a minature Wien and
install three in the chopper region. I decided to
investigate.
• Current density required appeared too high, ~2000 A/cm2,
for conventional conductor – until the YR melted. That
runs at 3700 A/cm2 at 6 GeV, so it’s possible.
4
Scheduling Issue Example
• Single horizontal Wien can be set to maximize polarization
to one (parity) experiment or maximize the sum of P or P2.
• For example, Qweak energy was reduced 2 MeV in
October 2011 to improve the polarization for hall A on
passes 2-5. With C at P2=1 Hall B has P2=1 on passes 2
and 4 and P2=0 on passes 3 and 5. HD-Ice wanted
polarization on pass 3 as well, but ….
Hall A Pass
P2 549 MeV/linac
P2 548 MeV/linac
2
.870
.884
3
.869
.902
4
.909
.957
5
.969
1.00
5
Wien equations
precession=
Wien condition
6
Present Wien
B(y)
hole for HV
E(x)
Steel box with 5.375” inside width,
3.125” inside height. Nickel disk at
each end to reduce field Z extent.
7
Present Wien Internals
8
Present Wien Top View
5mm square array of 200 keV particles enter at left. BdL 3600
G-cm, electrode voltage +- 18.2kV. Note horizontal focusing.
9
Present Wien: Side view at exit
Note modest spread in vertical direction
of the five rays which started co-planar.
X-Y coupling in this Wien is small.
Again, 5mm square array starts at Z=-25
10
Field fall-off at end of present Wien
0
-20
-19
-18
-17
-16
-15
-14
-13
-12
-11
-10
-2000
Field (V/cm or 160*G)
-4000
-6000
160*By
EX_5chamfers
-8000
-10000
-12000
-14000
Z (cm)
11
electrodes with 5
chamfers follow
mechanical drwg as
well as possible
Electrode taper in model
12
Field integral comments
•
•
•
•
•
B is determined by precession desired. 90o here.
E is determined by need for straight orbit on axis.
100 keV: BdL= 2097 G-cm, E=-11.2 kV/cm
200 keV: BdL= 3600 G-cm, E=-24.3 kV/cm =-2.43 MV/m
electrodes are 1.5cm apart, so latter equals +-18.2 kV
13
Things to check for present Wiens
• Power supplies +-15 kV now. 20 kV needed for 200 keV.
• Cables and connectors 20 kV capable?
• 20A magnet supply needed? Double up wires down to
tunnel?
14
Chopping system
• Symmetric pair of transverse 499 MHz cavities and
solenoids put the three bunches at 120o intervals on a 30
mm diameter circle.
• Moving slits may then independently reduce the current for
each hall.
• Master slit has 24o apertures, +-3mm radially about 30mm
circle
• 24o of 30mm circle is 6.3mm = width of beam between the
slits and the de-chopping half of the system.
• Anything within the chopping system must have at least
this aperture and have “unity” transfer matrix. -1
preferred and currently implemented.
• RF power adequate for 200 keV per R. Kazimi
15
Aperture choice for new Wien
• I assume in what follows that a parallel beam fills the A
and C chopping apertures and fills half the B angle extent.
Full 6mm radial extent in three beams.
• Less than 1% of the 180 A Qweak beam at 130 keV is
intercepted by the “master slit”.
• If the chopper slits were 20o by +-2.5mm radially the Wien
design task would be much easier. The Operations
Injector Group would have a more difficult design task.
16
Basics of new Wiens
• Three osculating 26mm circles have their centers on a
30mm circle.
• There’s a stock metric steel tube size 26mm OD, 1mm wall
– perfect for shielding one Wien from the next.
• E-field limit suggested 4 MV/m
• 200 keV beam, 90 degree precession capability needed
• For those values and sharp edged fields 30 cm long,
BdL=3600 G-cm and E=2.5 MV/m.
• Since E field and B field don’t fall off the same way in Z,
the electrode length has to be adjusted to minimize
steering.
17
single Wien model
18
End fields
0
13
13.5
14
14.5
15
15.5
16
16.5
17
Electrode length was
adjusted to match
magnetic field fall-off.
-5000
Field strength (V/cm or G)
-10000
Magnetic shield length
was varied in a failed
attempt to adjust the B
field end profile.
-15000
-20000
Final electrode 14.8 cm
half-length.
-25000
Final steel tube 15cm
half-length.
-30000
208*By
-35000
-40000
Z (cm)
19
EX
x axis cm from center
y axis V/cm or 208*G
Electrode curvature
I explored electrodes as arcs of circles, keeping 1cm spacing
on mid-plane. 3.05 cm chosen
Electric field uniformity
maxE/minE over 6mm circle
0.99
0.98
0.97
0.96
0.95
0.94
0.93
0.0
1.0
2.0
3.0
4.0
5.0
6.0
7.0
electrode radius of curvature (cm)
20
8.0
9.0
10.0
E field in midplane of Wien
8mm circle
-24912+-760 V/cm
21
E field in midplane of Wien
schematic of A beam
22
Time delay as a function of precession
• J. Grames sees degree phase delays in the present Wien at
130keV at large angles. The new model doesn’t show a
problem at 200keV. I haven’t checked the model of the
old Wien.
Time delay as function of precession
degrees of 499 MHz
0.025
0.02
0.015
0.01
0.005
0
0
20
40
60
precession angle
23
80
100
120
conductor choice
• Some gap between the OD of the conductor and the steel
desired to reduce field distortion. 0.5mm final choice.
• The smaller the angle subtended by the dipole conductor,
the better the field. I tried 15o, 12o and 10o .
• For this round of modeling, I used a square “helical end”
coil built into Tosca, 3mm square, 15cm half-length.
• Hollow conductor may be fabricated in 10o wedge, 7.445
mm2 , or 1/8” diameter refrigeration tubing used,
5.776mm2 of copper. Use tubing – cheaper and less
chance of water leaks into vacuum.
24
field quality from these choices
• Magnetic dipole field +/- 0.6% over 6mm diameter circle
• Electric field +/- 0.9% over 6mm circle
• Power supplies must be stable to better than 0.05% to
avoid beam steering.
• Scaling up the design doesn’t help because the 24o of
chopping circle scales too.
25
5.24 mm square array through Wien
Why 5.24 mm? 20 degrees. One
might need steering magnets
between the chopper slits and the
Wiens, but no focusing at this size.
26
view showing horizontal focusing
27
A full length quad doesn’t help
• A full length quad was added to the model in an attempt to
counteract the horizontal focusing.
• Vertical focusing was so great as to preclude its use.
• Since it complicated the assembly and was no use, delete.
• Next two slides show the beam divergence in the XZ and
YZ planes 10cm beyond the steel tube.
• Conclusion: focusing elements are needed fore and aft to
allow full chopped beam through the filter.
• To date, counter-wound solenoids have been used in the
models. Quadrupoles will likely be required for
engineering reasons.
28
YZ projection 10cm downstream
X-Y coupling is larger than in present
Wien, as seen by bundle spread
29
XZ projection 10cm downstream
30
Single Wien with solenoids
15.7M elements, twice the size of
previous model.
31
5mm bundle with focusing
End of
Wien (on)
focusing solenoids
Exit of Wien at left with vertical divergence due to focusing by counterwound
solenoids upstream of Wien. 5mm bundle is nicely preserved. However …
32
8mm bundle with same focusing
Wien on
Vertical spacing of eight planes is 8/7 mm, closer than 5/4 mm of previous slides. X-Y coupling
at larger excursions. View from +Z at upper left. Entire bundle is rotated and some rays
extend beyond electrodes, likely between the solenoids shown . The rays don’t hit any material.
33
Is the magnetic shield adequate?
• Model was copied twice with offsets to create a larger
model with three Wiens on 30mm diameter chopping
circle.
• E and B fields are identical in the three Wiens. Current in
B dipole is 0.39% higher than A/C to get same BdL.
• Square bundles of 200 keV “particles” were propagated
through the model and its parts to answer the question.
34
Three Wien filters on 30mm circle
35
Three Wiens Z+ end view
Field 6mm
downstream of
solenoids. Fields are
not symmetric about
center of each Wien
so beams don’t
propagate properly.
More shielding is
needed.
36
Caps added to solenoids
Only the solenoids are turned on in
this model. Caps with 0.65cm radius
holes have been added to reduce
cross-talk. These solenoids can’t
easily be built because it’s difficult to
deal with the leads and get pure Bz.
This is a proof-of –principle model.
Wien off in this figure.
steel cap
37
Cross-talk eliminated with steel end caps
38
Full model with Wien on
8mm square arrays propagated
through summed fields of
models. Electrostatic and
dipole fields calculated on
model without solenoid end
caps. Solenoid fields calculated
with end caps. Note that some
of the rays escape the bunch.
39
Chopped beam through the model
Full 6mm radial by 24 degree
theta extent for A and C beams.
Full radial extent by 12 degree
theta for B beam. All the rays
remain within the bunches.
IT’S POSSIBLE!
40
Chopped beam view 2
41
Chopped beam view 3
Space charge will blow up the
beam at the tight focus in the
middle of the Wien. Fields
calculated here must be transferred
to a particle-based simulation code.
Effect of the manipulations on dechopping must be modeled too.
42
Conclusions on 3 wien model
• Magnetic shielding is sufficient on the Wien proper
• Magnetic shielding was insufficient on the focusing
elements fore and aft. Two options will be explored:
– steel caps on tubes, 0.65cm radius holes
• They work; engineering very difficult with return current and leads
– investigate change to quadrupole arrays with steel tubes
like those around the Wiens proper, no end caps
• Exploring the options on focusing elements will take some
months due to computer hardware constraints.
• Unity transfer matrix must be achieved, including nulling
x-y coupling.
43
Modeling constraints in OPERA
1. Entire model must reside in RAM
2. single-threaded with some use of SIMD instructions in Intel CPUs
3. single-Wien model with solenoids occupies 6GB first iteration, 1GB for
each additional, takes ~20 hrs for first and 5 hrs for subsequent
4. triple Wien occupies 16.7GB first iteration, 3GB for each additional, so
three iterations max in my 24GB machine of 2010 vintage. 62 hours for
first, 12 for second/third
5. replacing the entrance and exit solenoid pairs with quad triplets expands
the model 50%, too much for my workstation, so I’ll have to coarsen it.
6. maximum problem size OPERA will run takes ~60GB RAM
7. Jlab has four OPERA 3D network licenses
8. ME has two unused individual OPERA licenses, one of which they will
allow to be converted to network for $5900 (first year) and $1180/year
thereafter by another Division
9. I have about 4TB of solved models, half backed up by sneaker-net.
620GB on the Wiens
44
Possible FY12 Milestones and Resources
• Management decision on further work. If yes:
• Modeling of the 200 keV injector sufficient to determine if
smaller bunch extent (20o by 5mm) is feasible.
• Model focusing and steering arrays
• Begin modeling of proposed chopper region including RF.
Can unity transfer matrix be achieved?
• Machine wooden dowels and form refrigeration tubing into
a dipole for feasibility tests.
• 1 FTE mostly modeling (includes me)
• ~$40K burdened for the fifth license and an OPERA
compute server for use by me and Physics
45
Possible FY13 Milestones and Resources
• Complete 200 keV injector modeling
• Design and fabricate single Wien assembly including
focusing and steering elements as needed
• Test single Wien in EGG test stand
• Decide whether to proceed to triple-wien project based on
EGG tests and models.
• 3 FTE: modeling, engineering, design, assembly, test
• $100K: materials, machining, power supplies
• WAG: Full installation (FY 14-16) might be $300K + 10
FTE-years, labor mostly for radical re-arrangement of
injector. Significant labor savings if combined with
planned 2015 injector upgrade.
46
WAG Discussion
• Capture is 40cm long. In the very optimistic scenario,
smaller bunch at 200 keV allows only the triple Wien to be
installed by adding this 40cm to the chopping region,
displacing the girder downstream of the chopper 40cm.
• In the optimistic scenario, only the triple Wien and
downstream corrector/focusing array will be required in
the chopper region, ~80cm total. Capture plus ~40cm shift
in quarter cryo needed.
• In the pessimistic scenario, about 1.4m needed in the
chopper region for focusing and steering, precession,
diagnostics, focusing and steering. Capture plus 1m shift
in quarter cryo and compress 6 MeV lattice.
•
Existing horizontal Wien is between A1 and A2 and their spacing can’t be compressed,
so there’s no real estate gained by removing it. Leave it in for back-up.
47
Conclusion
The physics and scheduling benefit of the proposed device
merits funding through experimental testing in the Injector
Test Facility. The injector test facility is crucial for
bumpless integration into the injector. After such testing
incorporation in the 2015 injector AIP project or making it
a separate AIP proposal would be debated.
48
Can I coarsen the model?
The discretisation is inadequate to model the coil field.
If you are not interested in the local values of the field
increase the value of #MAXEDGEHDLPTS (current value is 32,
minimum value 8, maximum value 1024, a power of 2).
If you want good local values of fields, improve the mesh at
the following locations ...
... at -2.034969652,0.5932691501,-17.90397636 (cm), Error 2.377E-03%
... at -2.120387067,0.6651094925,-17.92213716 (cm), Error 3.288E-03%
... at -2.091743085,0.6425010534,-17.92364535 (cm), Error 2.49E-03 %
... at -2.061793882,0.6186492849,-17.91912528 (cm), Error 3.618E-03%
... at -1.827145311,-0.274035773,-17.7994915 (cm), Error 3.422E-03%
... at -1.781289417,0.0324909175,-17.66482969 (cm), Error 3.814E-03%
... at -1.962044227,0.5151815544,-17.87561671 (cm), Error 1.551E-03%
... at -1.902698577,0.4300843549,-17.85722873 (cm), Error 2.399E-03%
... at -1.884628211,0.4008031438,-17.85226522 (cm), Error 3.58E-03 %
... at -1.869043299,0.3699945223,-17.84801607 (cm), Error 3.007E-03%
... at -1.782625648,0.0655654358,-17.58918861 (cm), Error 2.3E-03 %
This continues for ~450 pages. The model is already coarser than VF
defaults suggest. I’m using 1mm maximum extent elements in sources
and the volume occupied by the beam. 0.5mm around the dipole coil
ends.
49
Was this manual useful for you? yes no
Thank you for your participation!

* Your assessment is very important for improving the work of artificial intelligence, which forms the content of this project

Download PDF

advertisement