Recent High Efficiency RF Source Developments at SLAC National Accelerator LaboratoryPresented by Jeff Neilson on behalf of Electrodynamics Dept members:Mark Kemp, Aaron Jensen, Erik Jongewaard and Sami Tantawi, Chief Scientist for RFARED division

Work supported by the Department of Energy

2

Overview

• RF and Accelerator Research and Engineering Division at SLAC

• Energy Recovery for Pulsed RF Sources

• Scalable High Efficiency Klystron

HPRF capability at SLAC is Highly Vertically Integrated

• World’s only integrated capability to conceive, design, build test

− Very high peak power sources (up to 150MW) and components (up to 500MW)

− Associated modulators for sources− High gradient (170 MeV/m) normal conducting rf accelerator

structures

• This capability all under one roof allows prototype and test in a tight, rapid development cycle

• Unique capability to provide multivendor source of RF vacuum devices through licensing to industry

SLAC High Power RF Research & Engineering Has Spanned 0.3 -100 GHz and up to 150 MW Peak Power

5045 KlystronS-Band

•2.856 GHz, 65 MW

•> 800 produced since 1983 •MTBF > 90,000

hrs

B Factory Klystron476 MHz

1.2 MW CW

XL4 & XL5X-Band

11.4 GHz, 50 MW@SLACCERN

Sinc. TriestePSIBNLLLNL

XPX-Band

11.4 GHz, 50 MWPPM focused

W-Band Sheet Beam Klystron

95 GHz

Span wide range• 0.3 - 100 GHz• 1.2 MW CW – 150 MW Peak

Unique infrastructure geared toward fabrication and testing of specialized high power vacuum RF electron beam devices

B Factory klystron (MW-class cw) completing bake-out

• Precision in-process machining• 8 Ultra-high vacuum bake-out

stations• 12 Hydrogen Braze/Retort

Furnaces and 3 vacuum furnaces• 5 vacuum cathode processing

stations• Sputter and evaporative coating

chambers

Large Test Capability

• 13 instrumented 150 MW pulse power modulators

• Two MW-class CW test stands

• Two shielded test bunkers

Infrastructure is unique. Although originally sized for a higher klystron production rate, now utilized to build a much larger variety of high power RF devices and structures

5045 klystron in final test

High Power RF, Accelerator, and Pulsed Power Electronics Capability “Under One Roof” Enables Integrated System Design

Klystron modulators• SLAC inductive adder topology is

generally replacing line-type topologies

• SLAC Marx topologies coming into use for long pulse SCRF applications

Ultra-fast beam kicker drivers

• Solid state nanosecond-switching pulse generators for transmission line beam deflectors Inductive adder modulator next to a SLAC

6575 standard modulator. 3x volume reduction.

RF Source Development Renaissance at SLAC

After nearly 10 years of continuous attrition of personnel devoted to RF source technology, recent changes underway:• Upper management decision to maintain as one of SLAC core

competencies• All departments reorganized with new management• New funding being allocated for rf source R&D• Encouraged to actively seek outside funding sources • Two new hires in Electrodynamics department• New THz initiative – seeking to fill new staff scientist position to

lead program

Seeking to apply HPRF source and accelerator expertise to a broader set of problems, beyond DOE Office of Science

Spent Beam Energy Recovery for Pulsed RF Sources

10

Motivation for Energy Recovery in Pulsed RF Sources

• Growing attention placed upon energy usage at laboratories• Potential for providing a time-phased way to improve an

existing, aging facility• Future, high-rep rate (>kHz) applications place very tough

demands on modulator

11

Depressed Collector Technology

• Used to improve the effective efficiency of vacuum tubes• CW depressed collector technology is mature

12

CW Spent Beam Energy Recovery Methods not Suitable for Pulsed Systems

Typical CW biasing methodology

Ringing on cathode from parasitic elements results in phase jitter – Need better solution

Parasitic elements

13

The SLAC Pulsed Depressed Collector

• The collector stages self-bias as electrons impact the stage surfaces

• The time-varying potentials of the stages are determined by the spent beam characteristics and the collector electrical impedance

• Recovers energy for the next pulseClaims:

1)This is the first demonstration of a pulsed depressed collector (using a single power supply) in a high power vacuum device

2)This is the first method to recover energy from the spent beam during the rise and fall time of the pulse

14

Self-Biasing Concept

15

Multiple Benefits of SLAC Self-Biasing Approach

• Recovers energy during rise, fall, and flat-top

• Modulators would have a relaxed requirement on rise and fall times

• Existing systems can be retrofit

• No extra power supplies. Cost no longer proportional to number of stages• Concept is not limited to a particular klystron or modulator topology

• Bias tuning can be accomplished through adjustment of the storage capacitance. Can be done “automatically” if desired

16

Case for Economic Justification : A 5045 depressed collector

5045

Klystron RF Efficiency 45%System Efficiency (no recovery) 25%Pac,no recovery 107 kWSystem Efficiency (with recovery) 33%Pac, with recovery 82 kW

Average power consumption can potentially be reduced by over 25%

Implementation expense recover in ~10 years

>$800k/year electricity savings if implemented on 80 LCLS stations.

17

Pulsed Depressed Collector Experiment

Purpose

• Show that self-biasing concept works

• Confirm models

• Answer major concerns brought up in internal discussions

Approach

• Get results as quickly as possible

• Not about optimizing for best efficiency

18

Subbooster Klystron Depressed Collector Fabrication

ExistingSubbooster

Klystron

AfterBakeout

stage 1stage 2

location ofold collector

spentbeam

19

SLAC Klystron Test Lab Experimental Setup

SubboosterKlystron

DepressedCollector

Energyrecovered

Energy dumpedbetween pulses

20

Example Result of Self-Biasing Collector Stage

21

Klystron Test Lab Experimental Setup

• Five separate circuit element parameters are available to tune

N:1

Zc,load

ZL,leakage

ZL,magZc,primaryCollector

Stage

22

Comparison to PIC Simulations

•The stage potentials vary over time and depend on the biasing impedances and the collected currents•This problem is solved iteratively with a PIC and SPICE code

2D Magic PIC Simulation

SPICE Circuit Simulation

Collected Stage Currents

ResultingStage Potentials

23

Comparison to PIC Simulations

Biasing Impedance/Stage Voltage Maximum

22 nF/15kV

44nF/11kV

66nF/9kV

Measured Collector Efficiency

18% 19% 17%

Simulated Collector Efficiency

20% 19% 18%

•To the left, simulated results are compared for two different circuit nodes

•A good match is obtained over a fairly large range of different biasing conditions

24

Application to Long Pulse Systems

• Modulator rise time less of an issue for long pulse (ms and longer) but potential interest for retrofit of existing systems for energy recovery during flat top

• Unfortunately transformer approach does not scale well as loss and cost go up with pulse length

Solution - An “Inverse” Marx Energy Recovery Modulator

25

An “Inverse” Marx Energy Recovery Modulator

•Capacitors charge in series, and discharge in parallel

•A transformerless, solid-state topology

•Using resonant recovery, can passively recover energy back to the modulator

In-between pulses

During pulse

26

Repeat Test Using Inverse Marx Approach

27

Comparison of Simulation to Experiment

• Good match between PIC/SPICE model and experiment• Additionally, we can pre-charge biasing capacitors to produce a more-square pulse

28

Next-Generation Modulator Systems Both Provide and Recover Energy

Next-generation modulator development completes holistic approach to RF system design

Traditional Modulator

(DC to Pulse Converter)

AC Power RF Power SourceRF Out

SLAC Energy Recovery Modulator

(Pulse to DC Converter)

SpentBeam

Energy

Recovered RF Power

29

Summary and Next Steps

Summary• Basic concept has been experimentally demonstrated• Good matches with circuit and PIC simulations

Next step

• Implement in a challenging application:- Scale up to the SLAC workhorse klystron, the 5045

(three orders of magnitude greater peak power than “subbooster” klystron)

- Apply to ultra-short pulse, high repetition rate klystrons

Scalable High Efficiency Klystron

31

Unique Modular MBK Combining Scheme As Method to Produce a Scalable Design

Goal is to generate a design which can be scaled to higher power levels using modular design• Each module to use

low perveance beam for high efficiency

• Reduced NRE costs for new designs

• Economy of scale• Graceful power

degradationProposed scalable design where power

scales as (2N)2

32

X- Band Multi-Beam KlystronDesign Specifications

Parameter Design GoalBeam Voltage (kV) 60Frequency (GHz) 11.424Output Power (MW) 5Beamlets 16Beam Focusing Periodic Permanent Magnet (PPM)Efficiency (%) 60+Cathode Loading (A/cm2) < 10

X-Band Multi-Beam Klystron

33

Gun Design

X-Band Multi-Beam Klystron

Semi-automated gun design using R. Vaughan’s gun synthesis approach

34

RF Cavity and PPM Design

X-Band Multi-Beam Klystron

RF Cavity PPM Period

Magnet IronCavity

Drift Tube

To minimize size and the number of beamlets, the PPM iron will be plated with copper to act as a cavity wall.

RF Cavity Goes Here

35

65%+ Efficiency Predicted

X-Band Multi-Beam Klystron

One-Dimensional AJDISK RF Design

36

Rapid 2D Transport Simulation

X-Band Multi-Beam Klystron

Pseudo Port Model (Using MAGIC2D)1D Simulation voltages are set at the port boundary. Simulation Time: ~1minute

I1/I0 (consistent with 1D simulation)

37

Initial PPM Field Profile Based on Charge Density

X-Band Multi-Beam Klystron

38

2D PPM Transport98%+ Transmission

X-Band Multi-Beam Klystron

Normalized Axial PPM Field

Beam Transport Using Port Model

39

Port Model Agrees with Cavity Model

X-Band Multi-Beam Klystron

Port Model: Black ParticlesCavity Model: Red Particles

40

Two Energy Product PPM Stack

X-Band Multi-Beam Klystron

Simple Script for Quickly Building the PPM Stack

New PPM stack with optimized pole piece geometry to reduce the number of energy products required

41

Beam Transport Including Collector

X-Band Multi-Beam Klystron

MAGIC Port Model

Collector

PPM Stack

42

Magnet Modification for the Input and Output Cavities

X-Band Multi-Beam Klystron

Magnet Geometry for the Input and Output Cavities

43

3D PPM Stack

Iron Magnet

Cutout for input and output waveguide

X-Band Multi-Beam Klystron

44

Magnet Cutouts Have Minimal Effect on Beam Transport

X-Band Multi-Beam Klystron

3D DC Transport

X-Band MBK 3D Layout

Collector

Gun

Input and Output Waveguide

The cathode to collector tip distance is ~30 cm(the maximum diameter is ~6.5 cm)

PPM StackThe outer diameter of the iron pole piece is used to shunt the field to match the current density as the beam bunches.

X-Band Multi-Beam Klystron 45

46

Summary

• Simulations confirm the scalable MBK module will meet desired specifications

• Mechanical design and drafting are underway

• Modeling of the combining scheme is being finalized