Hao Zha 2013Oct09 Outlines 1 Chokemode structure design 2 Optimization on RF parameters 3 Experiments 4 Future plan Overview of CDS design Name Description Parameters Manufactory amp ID: 185380
Download Presentation The PPT/PDF document "CLIC choke-mode structure" is the property of its rightful owner. Permission is granted to download and print the materials on this web site for personal, non-commercial use only, and to display it on your personal computer provided you do not modify the materials and that you retain all copyright notices contained in the materials. By downloading content from our website, you accept the terms of this agreement.
Slide1
CLIC choke-mode structure
Hao Zha2013-Oct-09Slide2
Outlines
1. Choke-mode structure design;2. Optimization on RF parameters;3. Experiments.4. Future planSlide3
Overview of CDS
design
Name
Description
Parameters
Manufactory &
Measurement
CDS-AOriginal design;regular choke;Irises: CLIC-G(0.15m): 20~30 V/pC/m/mmPrototype cells made in CERNCDS-BHOM tuned choke;Wakefield suppression Improve;Irises: CLIC-G(0.15m): 15~20 V/pC/m/mmNoCDS-CNew choke design, Optimized on wakefield suppression;Irises: CLIC-G(0.15m): ~5 V/pC/m/mmη: 24.2%Prototype cells (aluminium) made in Tsinghua.Wakefield tested in AWA.CDS-DNew choke design;Optimized on RF parameters.Irises: a/λ=0.125(0.15m): ~4 V/pC/m/mmη: 26.0%No
NameDescriptionParametersManufactory &MeasurementCDS-AOriginal design;regular choke;Prototype cells made in CERNCDS-BHOM tuned choke;Wakefield suppression Improve;NoCDS-CNew choke design, Optimized on wakefield suppression;Prototype cells (aluminium) made in Tsinghua.Wakefield tested in AWA.CDS-DNew choke design;Optimized on RF parameters.NoSlide4
Equivalent circuit model
For accelerating mode,
ϕ
3
=
π
/2, so z=∞. Then It is fully reflected by choke.
Some HOMs could be also fully (or mostly) reflected.Slide5
Thin-neck choke design
Zc
a
Short load
L
a
section a
Plane CZcb =K*ZcaLbsection bzazbz3Plane D
Plane APlane APlane DPlane C
S
12
K
1
2
3
4
f2
36GHz
49GHz
60GHz
69GHz
T
he joint planes (Plane A and Plane D) are equivalent as impedance transformers
Purpose of thin-neck design:
Increase the 2
nd
fully reflected frequency
Reduce reactance of HOMsSlide6
Matching step
Purpose of matching step design:
Reduce
reactance of
some HOMs.
Compensate residual reflection from the load,
p
ossible to design loads with lower price.Slide7
Dipole detuning
(
~16GHz)
(
~18GHz)
c1
c2
(mm) (GHz) Purpose of dipole detuning: Detune lower band HOMs (one cell).
Provide weak detuning between cells (like DDS).Slide8
Wakefield simulation results
W
⊥
(s=0.15m)
WDS
5
CDS-A
20~30+Thin-neck10~17+Matching step7~10+Dipole detuning (CDS-C)5~6CDSWDSFc11Frms45Fworst2520Slide9
RF parameters of CDS-C
HFSS simulation results:
CDS-C
CLIC-G
Iris aperture(
mm)
3.15, 2.35
3.15, 2.35Q(Cu)4895, 53855538, 5738Shunt impedance(MΩ/m)59, 8381, 103Group velocity(%c)1.38, 0.731.65, 0.83Max E-field(MV/m)246235Sc (MW/mm^2)5.725.43Pulse heating(K)23.247.9Peak input power(MW)67.4
60.5Filling time(ns)72.462.2RF-beam efficiency24.2%28.1%Slide10
Optimization on RF parameters
Optimization target : RF-beam efficiency (η), Max surface field (Es, Sc
, Hs).Geometry sizes of irises are the focus points.
Input:
Geometry sizes of irises
Output:
η
, Es, Sc, HsOptimizationSlide11
Genetic algorithm
Lots of geometry variables => Huge searching space.Genetic algorithm
=> find optimum solutions more quickly (actually much more quickly).
Use C++ to implement it.
Fitness
η
, Es, Sc
Individuals Gene: GeometryBorn children (with gene mutation)Higher fitness, more childrenCalculationDiscardedJoin inlowHighRandom initialization
Here we use a simplified way to calculated RF parameters.Slide12
Optimization results (1)
Max population: 400~800Birth rate: 4/iterationIteration: 800
Searching time: 2.5×106
;
CDS-C
CDS-D
CLIC-G
Iris aperture(mm)3.15, 2.353.49, 2.503.15, 2.35Shunt impedance(MΩ/m)59, 8353, 7481, 103Group velocity(%c)1.38, 0.732.00, 0.751.65, 0.83
Bunch population (109)3.724.503.72Max E-field(MV/m)246223235Sc (MW/mm^2)5.725.175.45
Pulse heating(K)
23.2
22.0
47.9
Peak input power
(MW)
67.4
79.3
60.2
Filling time
(ns)
72.4
60.8
62.2
RF-beam efficiency
24.2%
26.0%
28.1%Slide13
Optimization results (2)
Wakefield suppression: 4V/pC/m/mmF
c=1.01; Frms
=3.8;
F
worst
=3
5CDS-CCDS-DSlide14
Radial line experiments
Target: (1) To verify HOMs absorption (S12) in choke-structures; (2) To test the RF load;Reflection S11 can be tested on VNA after calibration (“multi-short load calibration”)
To VNASlide15
Radial line experimentsSlide16
Reflection of RF load
Results are promising;Slide17
HOMs absorption in chokeSlide18
Layout of AWA
G1: Photo cathode + Faraday cup. To study the phonemes of laser trigged RF break down.
G2: Witness beam line, Mg cathode + gun + 1*linac, 15MeV, max 100nC. G3: Drive beam line, Semiconductor cathode + gun + 6*linac, 75MeV; >=100nC, can produce bunch train.Slide19
Set up experiments (1)
GUN
LINAC
Chamber
ICT
ICT
Spectrometer
20GHz oscilloscope, 0~20GHz mixer
1#
2#
3#
4#
oscilloscope
With load
Without loadSlide20
Set up experiments (2)
Chamber
YAG
ICT
Bellows
YAG
Bellows
Be windows:50um thickness;For 14MeV electron beam, ~30% (calculated by Fluka) of the beam can pass it without any scattering.After optimized on beamdynamics:1nC Beam size: ~0.7m,5nC Beam size: ~1.6m,Beam aperture: 5.5mmSlide21
Set up experiments
(3)Slide22
Wakefield measurement results (1)
L
0
10GHz
CABLE
Attenuation
1# 20 dB 2# 20 dB 3# 0 dB 4#0 dB Slide23
Spectrum of signal in cable 4#
Unidentified dipole mode? It may be the 36GHz – 2*10GHz = 16GHz
Fundamental
mode 12GHz
Dipole mode
~16GHz
Unidentified dipole mode?
Dipole mode ~19GHzDipole mode ~23GHzDipole mode ~28GHzSlide24
Wakefield measurement results (2)
L
0
No Mixer
CABLE
Attenuation
1# 30 dB 2# 20 dB 3# 10 dB 4#0 dB Slide25
Spectrum of signal in cable 3#
Dipole mode
at 16GHz and 19GHz
Looks
like the dipole mode
at 50-22=
28GHz
Fundamental mode UnidentifiedQ (@12GHz) = 160~400Q (@16GHz) = 12.5Q (@19GHz) = 15.3Slide26
Future plan
Manufacturing & High power test;Start manufacturing choke-mode cells, 6cells/24cells(Tsinghua
); Single choke-cell for high power test designed (Tsinghua & KEK);
Start to design the compact coupler for choke-mode (
Tsinghua
).
Wakefield test with two beams;
- Two beam line facility (AWA)Further optimization (geometry, load design);Slide27
Thank you!!Slide28
Multi-short calibration
At least 3 different load is needed in the standard calibrationSolve the Inconsistent equations to get more precise network parameters (Least square method).
- y: reflection detected by network analyzer ;
- x
: reflection of radial line at plane A
……Slide29
Genetic representation
There are 3 selections for
genetic representation.The second way
has the best practical results.
For the first and third selections, it is very difficult to put iris parameters of every cell into chromosomes. We just selected some cells (e.g. first, middle and last cell) as chromosome
.
linear fitPolynomial co-efficiencypolynomial fitSlide30
Fitness function
Early convergence is a big problem of GA:
Better solution has more children, and soon it and its posterity will monopolize the whole world.
Fitness = Performance – CompetitionSlide31
Reproduction
Evolution will be aroused in the reproduction by two ways: Mutation
; Crossover (not very suitable in our case).About mutation, we find this way of mutation has a better effect
:
Sequential
mutation (rather than randomly mutation) + Mutation step becomes gradually smaller (rather than constant)