High efficiency – high perveance Klystron (X-band)
EnEfficiency RF sources Workshop
Cockcroft Institute
Franck Peauger CEA/IRFU/SACM/LISAH
June 3, 2014 www.cea.fr www.cea.fr OUTLINE
. Context and objectives . New klystron concept . Simulation codes . Klystron development plan
2 . Context and objectives . New klystron concept . Simulation codes . Klystron development plan
3 WORK DONE UNDER TASK 12.3 OF EUCARD2 WP12 Innovative RF Technologies 2013 - 2017 « In this sub-task, CEA will develop and search for innovative concepts of X band RF power sources and components. The objective is to propose affordable and reliable solutions for future testing capabilities for the CLIC accelerating structures. The task includes the design and the fabrication of prototype RF devices to demonstrate the feasibility of the new concepts proposed. »
Juliette PLOUIN: CEA project leader for this sub-task
Budget available to build a (small) part of the RF power source or component
Fully oriented towards an R&D activity 4 A STRONG MOTIVATION WITH A MORE GENERAL INTEREST • Some important projects (XFEL, ESS, CLIC-drive beam, FCC…) search for “green technologies” and so require ultra-high efficiency RF sources • Modulators are now close to 90% efficiency. Microwave tubes are one of the major weak point in the energy consumption chain
• A road map to reach the 90% efficiency horizon have been established by Igor at different frequencies
• Our Eucard2 task is well in agreement with this road map
• This project is also an occasion to acquire knowledge and experience on RF sources “in-lab design” and why not “in-lab fabrication” in Europe (like what is done at SLAC or on HV modulator at CERN for example) 5 LIST OF POSSIBLE DEVICES
Microwave vacuum tubes for X band and high peak power
O type Gyro M type Gridded tubes devices tubes tubes
Travelling Locked Carcinotron Klystron Gyroklystron IOT wave tube Magnetron
Single beam “affordable and reliable” klystron = Multibeam reduced surface fields klystron (DC and RF)
Sheet beam “high efficiency” klystron = Optimal beam / RF interaction Hollow beam 6 klystron X-BAND 50/75 MW KLYSTRON (USA)
• Development started in 80’s for linear collider studies at SLAC
• XC series: 100 MW design, 65 to 72 MW measured • XL series for NLCTA: 50 MW, 410 kV – 305 A, reduced perveance of 1.2µA/V1.5, 40% efficiency,
Jongewaard
• XL4 : 11.424 GHz produced in 15 units
• PPM focusing investigated and combined with high efficiency (60%) but with limited success
E.SLAC, of
• In 2008 – 2011: XL5 : scaling of XL4 at 11.9942 GHz for CLIC, PSI and Trieste
- 5 tubes built courtesy
• Since 2011: XL5 transferred to industry : CPI VKX8311A Image: - first tube tested at SLAC in 2013
7 N. Catalan, CLIC workshop 2014 X-BAND 75 MW KLYSTRON (JAPAN)
Same parallel development at KEK/TOSHIBA for NLC/JLC
Spec (target): 75 MW, µPerv. 0.8, 60% efficiency More results with the PPM focusing version 10 tubes tested
S. Fukuda, KEK, 2005 Image: courtesy of TOSHIBA
8 X-BAND 150 MW MULTIBEAM KLYSTRON (JAPAN)
Annular cavities working on a HOM “Realistic design” to reduce the beam voltage
A. Larionov et al., KEK
9 9.3 GHZ – MEDIUM POWER « COMMERCIAL » KLYSTRON
10 SHEET BEAM KLYSTRON
Explored in W-band, S-band and at 1.3 GHz by SLAC PPM focused X-band tube under development at CPI to lower the voltage Also investigated in China at IECAS
A. Jensen, SLAC, IVEC2013
T. Johns et al., CPI D. Zhao et al., IECAS11 ANNULAR BEAM KLYSTRON
Developed by CCR at 1.3 GHz – 10 MW Very high perveance (3.3 µA/V^1.5) and high efficiency (65%) Zero compression gun and high current density cathod Try to avoid diocotron instabilities
M. Read et al., CCR
12 XBOX3 TEST STAND AT CERN
XBOX3 will use four Medium Power X-band klystrons recombined and compressed to produce a 50 MW power level This smart concept is well adapted to the reliability and affordability we are looking for and it will constitute the basis of our study
I. Syratchev, G. McMonagle, N. Catalan Lasheras
• 4 turn-key 6 MW, 11.9942 GHz, 400Hz power stations (klystron/modulator) have been ordered from industry. Medium power x-band RF sources have• alsoThe f iarst wider unit is interestscheduled in to FELarrive and at C ERN in October medical applications 2014. The full delivery will be completed before13 Ju ly 2015. XBOX 3 AT CERN
CERN is ordering commercial products for XBOX3 (4 units of each) No particular R&D is required and delivery time is short
TOSHIBA klystrons SCANDINOVA parameters Modulator parameters Peak power: 6 MW Peak RF power: 8.0 MW Beam Voltage: 150 kV Pulsed voltage: 175 kV Beam current: 90 A Pulse current: 115 A Average power: 12.4 kW Average power: 50 kW Efficiency: 47.5% Pulse length (flat): 5 sec Rep. rate: 400 Hz
14 A VERY CHALLENGING OBJECTIVE
Our proposal is to study and design a 12 GHz single beam klystron able to deliver a peak power of 12 MW with a pulse duration of 4.5 s to double the testing capability of XBOX3
It should be compatible with the Scandinova modulator ordered for XBOX3 in order to do fully qualify the klystron at CERN. A possible operating point could be: - Vk = 170 kV - Ik = 100 A -> µP = 1.4 µA/V1.5 - efficiency > 70 % -> Pout = 12 MW
A new klystron concept is required
P. Guidee, Thales, 2001
A classical solenoid based focusing system would be considered for this first step.
The klystron should be designed to operate at a pulse repetition rate of 400 Hz minimum and - why not -, up to 1 kHz, a highly interesting repetition frequency for future FEL projects 15 . Context and objectives . New klystron concept . Simulation codes . Klystron development plan
16 THE ADIABATIC CONCEPT
. A system is called adiabatic when the external forces vary more slowly than the interaction forces in the system. The dynamics of the system is then a succession of equilibrium states and the entropy does not increase.
Example in thermodynamic: adiabatic if no heat exchange with the external medium: the piston velocity must be smaller than the speed of sound in the gas
. In our case, the external forces are the beam induced bunching forces and the interaction forces are the space charge forces.
17 A PARALLEL WITH RFQS FOR PROTON LINAC
An RFQ cavity is used in proton linac injector to bunch, focus and accelerate a continuous beam from few tens of keV to few MeV. It allows a beam transport in space charge regime (high intensity beams is possible) with very low beam losses. TE210 mode
longitudinal modulation on the electrodes creates a longitudinal component in the TE mode that bunch and accelerate the beam
bl
In the IPHI RFQ (3 MeV, 6m, 352 MHz), around 350 cells are used to bunch the 18 beam PHASE SPACE DISTRIBUTION IN AN RFQ
Along the 6m RFQ:
At the RFQ exit:
(MeV)
dW
dispersion
Energy Phase (deg)
The bunching process in an RFQ is adiabatic (« gentle » buncher)
The RFQ preserve the beam quality, has a high capture (~90%) compare to a discrete bunching model (50%)
19 INCREASING THE NUMBER OF CAVITIES IN A KLYSTRON
10 cavities 20 cavities Efficiency 67.2 % Efficiency 78 % Length 197 mm Length 285 mm
. The idea is also to maintain a classical perveance value of 1.4 µA/V1.5 (170 kV – 100 A) : high current leads to shorter plasma wavelengths (and so shorter structures) . In the design proposed, the cavities are weakly coupled to the beam (low R/Q) and largely detuned to 20 avoid strong bunching ANIMATION FROM AJDISK RESULTS (CHIARA MARRELLI)
5 cavities – 1 GHz optimized by C. Marrelli 20 cavities – 12 GHz optimized by F. Peauger 휇Perveance = 0.21 휇Perveance = 1.4 Pout ≈ 2.3 MW Pout ≈ 12.5 MW Efficiency 78. % Efficiency 78 % EXPECTED PERFORMANCES WITH 4 DIFFERENT DESIGNS
Name Numbers R/Q TTF Efficiency Overall of cavities [Ohm] length [mm] AK10-2 10 20 to 50 0.683 67.2 % 197 AK14-1 14 20 0.683 68.5 % 221 AK20-3 20 27.4 0.688 74 % 285 AK20 20 12 0.72 78 % 285
70% efficiency
. These design have been manually optimized and could be certainly fully improved by an automatic procedure. 22 EXPECTED DETAILED PERFORMANCES 14 CAVITIES STRUCTURE
Output power for different cathod voltages Phase variation for 1% of cathod voltage change 14 -50 ∆∅ 175 kV – 104 A =9.4 °/kV
12 -55 ∆푉푘 171.7 kV – 101.5 A
-60 10 170 kV – 100 A -65 -70 8 170 kV – 100 A 150 kV – 82.8 A -75 6 -80 -85
4 Outputphase (deg) -90 168.3 kV – 98.5 A Outputpower (MW) 125 kV – 63 A -95 2 -100 0 0 5 10 15 0 10 20 30 40 Input power (W) Input power (W)
Efficiency at saturation for different cathod voltage Bandwidth Small signal 80 74 (Pin = 0.5 W) 170 kV – 100 A 70 70
60 66
50 62 Saturation (Pin = 10 W)
40 (dB) Gain 58
30 54 Efficiency atsaturation (%) 20 50 120 130 140 150 160 170 180 11.97 11.98 11.99 12 12.01 12.02 12.03 Cathod voltage (kV) Drive frequency (GHz) 23 BUNCHING CAVITY DESIGN
2 j z E (z)dz b C z Ez (z)e dz R M Q 2U L / 2 E (z)dz z L / 2 Eigenmode
F = 11.9108 GHz, Q = 2039 with copper Utot = 2.5735E-19 J R/Q = 23.9 ohm M = 0.6127
XY Plot 1 cav3 0.20 Curve Info ComplexMag_E Setup1 : LastAdaptive gap2='1mm' Phase='0deg' r1='9mm' ComplexMag_E_1 0.18 Setup1 : LastAdaptive gap2='1mm' Phase='0deg' r1='9mm'
0.15
0.13
0.10 Y1 [V_per_meter]Y1
0.08
0.05
0.03
0.00 0.00 5.00 10.00 15.00 20.00 25.00 Distance [mm]
24 INPUT / OUTPUT CAVITY DESIGN
Input cavity Output cavity: many possibilities: Qx = 460 - Same as input cavities but with Qx = 20 (change of the iris dimensions) - Multicell output cavity: SW or TW - Other ideas ? (backward wave, recirculation, …)
SLAC XL5 output cavity (4 cells, p mode)
25 . Context and objectives . New klystron concept . Simulation codes . Klystron development plan
26 BEAM / RF INTERACTION SIMULATION TOOLS
AJDISK ? MAGIC TRACEWIN (CEA) (SLAC) Frequency domain PIC time domain Tracking 1D 2D 2D/3D 2D/3D free Cluster CST PIC time domain Preliminary Design 3D consolidation design Emittance WARP growth, (LBNL) PIC time domain matching, halo 2D/3D and error studies Free, open source (new use for Cluster klystron!)
Full verification of the performances and oscillation study 27 THE WARP CODE
Warp*: PIC modeling of beams, accelerators, plasmas Lawrence Berkeley National Laboratory, CA, USA
28 SIMULATING A MONOTRON CASE WITH WARP
. Monotron oscillator = most simple microwave tube configuration . A well know phenomena (most of the time not desired) in the microwave tube community
DC energy source + resonant system = monotron
. Objective: validate WARP on its ability to simulate the beam/RF interaction where Barroso publication: 20 kV – 10 A, annular beam, two-steps cavity 4.1 GHz . 1D analytical treatment is possible and 2D PIC results is compared 29
MONOTRON CASE
Barroso publication WARP simulation
• Without focusing magnetic field Bz 10 ns
630 ns
> 750 ns Simulations performed by Antoine Chancé - CEA
30 MONOTRON CASE
Barroso publication WARP simulation • With focusing magnetic field Bz 10 ns
542 ns
> 750 ns
31 MONOTRON CASE
Ez field at saturation
WARP simulation 4.1 GHz
Barroso publication
Ez field at saturation at r = 7 mm
The WARP code is valid for simulating all kind of vacuum 32 microwave tubes TRACKING CODE “TRACEWIN”
a 3D code widely used in the accelerator community PARTRAN routine to simulate the space charge effects The idea is to inject the gap voltages calculated by AJDISK Allows to simulate a large number of particles (useful for Halo studies for example) and has automatic optimization procedures
33 . Context and objectives . New klystron concept . Simulation codes . Klystron development plan
34 THE 12 GHZ - 12 MW KLYSTRON PROTOTYPE PLANNING
2014 2015 2016 2017
Preliminary design Commissioning preparation (advanced simulations)
Detailed design and drawings Convergence on simulation codes
Fabrication
Choice of the Design number of cavities Review
Tests
Superconducting solenoid
PhD student
35 TOWARDS HIGHER EFFICIENCY WITH THE ADIABATIC APPROACH Tentative Road map
Advanced Focusing system: superconducting solenoid / Medium Power klystron High power klystron Permanent magnet / 170 kV – 12 MW 170 kV – 50 MW RF focusing?... single beam 4 beams
Output RF structure: SW, TW, … Proof of principle Low power klystron 60 kV (gun in air) 1 MW single beam or 4 to 6 MW multibeam X-band frequency FABRICATION: collaboration with industry is mandatory:
Possible strategy: • Functional drawings by lab (CEA or CERN) Lower frequency klystron • Call for tender for the fabrication 1 GHz, 704 MHz or lower Higher frequency klystron of one prototype > 12 GHz NEW PROCEDURE for klystrons (but usual for Gyrotron in the Directly applicable to To check the limit of the concept fusion community) accelerator projects (ESS, Need to find some applications CLIC, FCC…)
36 CONCLUSION
o Some clear and highly motivating objectives to develop high efficiency RF sources with high reliability have been presented o The new adiabatic approach to optimize the bunching process of the beam have been proposed and declined in few designs
Kl-adi(adiabatic)-stron = « KLADISTRON »
o We are working on new numerical tools for the klystron study o We have also to think of extending the Eucard2 project to a larger collaboration 37 CEA TEAM
CEA permanent staff Post-doctoral position
24 months from April 2014 Modelisation of the beam/wave interaction Franck Juliette Barbara Funding EUCARD2/CEA PEAUGER PLOUIN DALENA
Internship with help of: 6 months from April 2014 Design of electron guns
Lyes Antoine Funding CEA CHANCÉ BOUDJAOUI Beam dynamics, user of Tracewin and Warp + 1 PhD position with funding demand in progress http://www-instn.cea.fr/spip.php?page=Publication_Sujet&idSujet=8343&lang=fr&langue=fr&id_rubrique=70 38
EXTERNAL CONTACTS FOR POSSIBLE COLLABORATIONS
. CERN/CI: Igor SYRATCHEV, Chiara MARELLI, …
. THALES ELECTRON DEVICES: Rodolphe MARCHESIN, Armel BEUNAS
. PSI: Jean Yves RAGUIN, Micha DELHER
New collaborators are WELCOME!
39 EXTRA SLIDES
40 WHAT IS A KLYSTRON?
It is a vacuum microwave electron tube amplifier where:
The input cavity prebunch slightly a DC beam provided by an electron
gun Devices
The intermediate cavities develop an RF voltage induced by the beam loading (image charges). These induced voltage intensify the bunching process.
of Thales Electron Thales of
The beam is strongly decelerated in the output cavity and a high RF power is created
The decelerated beam is collected in a collector courtesy
The beam is focused by an axial magnetic field (solenoid) Image:
Image charges
RF out
RF in
electron
velocity Axial 41 S. Berger - Thales, XB2008 workshop, Cockcroft Institute, UK THE CLASSICAL WAY TO DESIGN A KLYSTRON
One or two cavities around central frequency to ensure the bandwidth Two or three bunching cavities at higher frequency, located at approximately a
quarter of the reduced plasma wavelength lq 푉0.75 ∙ 푏 휆푝 = 0.036 with b beam radius 퐼0.5 휆푞 = 휆푝/퐹 Exemple: Vk=170 kV, Ik = 85 A b = 1.8 mm, a = 3.5 mm lp = 59 mm
ge .b = 0.61 rad F = 0.35 lq = 168.5, lq/4 = 42 mm
At Ik = 1.5 x 85 A = 136 A, lq/4 is decreased to ~ 33 mm
Beck, « Space Charge Wave », 1958, p.106
Maximize the modulated current I1/I0 Sometimes a 2nd harmonic is added to reduce the length of the interaction structure 42
A CLASSICAL 6 CAVITIES KLYSTRON
Quick design with AJDISK (SLAC 1D code)
20 mm 46 mm Drift lengths 55 mm ~ lq/4
Large energy dispersion
Limit of the method: induce large velocity dispersion and so limit the interaction efficiency (except for very low perveance and/or advanced optimization method – cf Baikov, Guzolov et al.) 43 GLOBAL COMPARISON WITH OTHER METHOD (IGOR SYRATCHEV)
44