High Power S-Band RF Loadfor SLAC Linac

A. Krasnykh, F-J Decker, et al.SLAC 2014

Introduction

• The LCLS Injector showed the phase jitters (PJ) at L0A, L0B, and L1S stations on beginning of injector setup.

• Initial PJ measurement setup employed both RF signals from Input and Output WG coupler of each accelerating section.

• Initial thoughts was that the PJ problem linked with the 5045 klystrons (see movie L1S, July 2008)

• I was asked to jump into the PJ investigationsand two main week components were selected.They are RF Loads and Thyratrons.

*) PJ was acceptable for the Injector commission but it was close to the upper spec value.

Original and Present RF Setups

Originally 20 MW peak klystronswere employed. All vacuum RFcomponents ( waveguide bends,power dividers, RF loads, etc.)are matched with a power levelfrom available RF sources.

7 RF loads per each station: 4 “High” and 3 “Low” power loads

Comparison of RF Power Levels

Klystron Output Power,

MW

Power at

SLED Output,

MW

Power in Waveguide

Feeders after First

Power Divider

Power at the Input Port of Accel.

Section,MW

Power at the Output of Accel. Section

(Power at Load),MW

Original RF Setup 24 n/a 12 6 1.8 Present

RF Setup 65 240 120 60 18.2

*) From our “Blue Book” bible

*)

5

High Power RF Loads in LCLS Injector

~50MW In

~14 MW out

6

Result of RF Load Performance @ Klystron Test Lab

A dozen of original Kanthal RF loads were tested.We found not one good RF load for P>3 MW peak.

It was confirmed thatthe original loads canabsorb RF peakpower no more than3 MW peak withoutbreakdown.

Dilemmas # Actions Result Comments1 To find good RF loads from

existing spares Unsuccessful 14 Loads had been tested,

practically all linac RF loads were inspected, load lifetime is unknown?

2 To find a national vendor(s) and to buy RF loads

Unsuccessful “vacuum dry” spec

3 To find an international vendor(s) and to purchase

a) NIHON KOSHUHA Co., Ltd, Japanb) BINP, N-sk, Russia

a) 5 loads tested, 2 were accepted, $$b) “vacuum dry” spec

4 To find alternative technical solutions for the RF linac setup

There are couple risky proposals

Alternatives were not tested experimentally

5 To find a company capable of design and fabrication

a) Muons Inc.b) INTA Technologiesc) SLAC (a “HY” approach)

a) & b) Under DoE SBIR grants

Result of Linac Inspection of the RF Loads

For example, waveforms of reflection signal @ Sector #21

All waveforms have an unstable (harsh) periods. The harsh period is delayed. The delay is equal toa round trip of RF power from the KLY to the RF Load and back,i.e. ~1.8 usec.

Result To Find National Vendor(s) to Buy the RF Loads

A vacuum dry specification was a main stopper for the National Vendors.

Their technical solutions are similar to the SLAC H2O RF Load concept. Such proposals are not acceptable for the SLAC linac.

Existing SLAC “vacuum dry” RF Load

High Power Loads from Japan and RussiaFive SiC-type loads were tested: accepted two and failed three loads.

• This load does not meet the SLAC requirements to be the “dry load”• It is a resonant load (sensitive to temperature , incident power level, etc.)

HFFS simulation shows that the SiC surface gradient is high(@ 30 MW peak 35 kV/cm @ surface of the SiC pins)

TE020 mode

Possible Temporary Technical Solution

• Each RF window had been tested at 60 MW, 180 pps, 3.5 usec before to be installed in 5045 assembly

• Double overhead power (i.e. four RF windows are employed)

Double RF window approach with a vacuum interlock

Item #4

Item #7

Possible Temporary Technical Solution #2 (cont.)

~ 1usec

W, MeV

Pin, MW

Pout, MW

P_ref@KLY, MW

L0A 57.3 33.5 9 0.18L0B 70.6 50.8 13.7 0.27

• Stable RF Power termination• Good vacuum (no RF window)• Pref @ 5045 is less 0.3 MW (is still rather high)• 5045 tube practically OFF (i.e. no KLY beam)• Multi bunch mode is feasible

Anatoly -- Help Yourself with the RF Load Problem!

Alike Anatoly worker fabricates dirtyabsorber parts for his RF loads

Permanent solution is needed for the SLAC RF Loads.• 1,000 loads are needed (i.e. a cheap cost

load concept) for whole linac *)• Lifetime load should be “forever” under a

continuously mode of the linac operation It was proposed to work with Muons Inc. and INTATechnologies to design and fabricate the RF load for theSLAC linac under the DoE SBIR grants.

Muons Inc. & SLAC INTA Tech. & SLACAn absorber is based on the lossy dielectric.TE01 mode in round WG is a working mode.

An absorber is based on the low conductive layer. TE10 mode in rectangular WG is a working mode.

M. Neubauer is the Muons PI (design and fabrication). Anatoly is responsible for the SLAC sub-grant

Anatoly is a designer and tester. Mr. LeClair is the PI for the load technology and its fabrication.

14

Concept of the Muons RF Load

*) courtesy Muons Inc.

A prototype configuration of the complete dry vacuum load *).

The Muons RF Load design employs a mode converter. A design of the mode converter is the SLAC responsibility.

Cross section view of the lossy ceramic cylinder in the compression assembly *).

Accelerating Section and RF Load

15

Mode Convertors for the Muons RF Load

The S-Band “wrap around” mode converter had been scaled from the Sami Tantawi X-Band converter

Under Muons-SLAC SBIR DoE Grant a new compact mode converter for Muons RF Load had been designed with V. Dolgashev

TE10-TE11-TE01

16

Design of Dielectric Cylinders for the Muons RF Load

Lossy material in the TE01 round waveguide will be mechanically confined in compression (without brazing), in order to eliminate tensile stresses in the lossy material.

A number of lossy porcelain ceramics were fabricated by Muons with varying amounts of SiC powder incorporated.

Dielectric constant and loss tangent had been measured at SLAC.

The case where 52% by dry volume of the SiC powder was used produced a lossy ceramic rings with an average dielectric constant of 7.44-0.37i, yielding a loss tangent of 0.05 as measured by the Agilent probe method.

A heat gun is used to evaluate an RF loss sensitivity vs. temperature

Where Are We with the Muons RF Load Design?• Two prototypes of the mode convertors were proposed at SLAC for the Muons RF Load

Design• The prototypes were manufactured by Muons Inc.• S-parameters of prototypes were measured/studied at SLAC.• A compact version of the mode converter was recommended for the Muons RF load.• A matrix of the porcelain with the SiC powder doping was sintered by Muons Inc. for the

RF lossy dielectric material.• The lossy ceramic cylinders in the compression assembly is currently in the fabrication

stage by Muons Inc.

This development is sponsored by the Phase II DoE SBIR Grant.

Activity at SLAC is conducted under Muons-SLAC CRADA with a total SLAC budget of $450k paid by Muons Inc.

Remaining to do:• Fabricate cylinders, test them individually and after welding them in array to test

at mW-kW levels, vacuum and thermal cycle tests of the pre assembled unit.• Fabricate the high power version of the TE10-TE01 mode converter, low level

RF test, vacuum leak test, bake up cycle tests.• Build the load assembly and perform low-middle-high power tests

Development of New Load Based on High Resistive Absorber (SLAC design)

This high power RF load concept is a conservative approach of further development of the original design.

A Kanthal wire (matrix of components) and a flame spray technology are used to produce the RF absorption layer.

Similar to the Kanthal matrix (but atomized powder instead of the wire) and a plasma spray technology are used to produce an the RF absorption layer

Original New

Kinetic energy of melted drops, their temperature, and sizeare dissimilar. Oxidation process of components is alsodifferent. Interaction particles with a base substrate isdifferent too. Etc.

Formation of RF Absorbing Layer

Development of New Load Based on High Resistive Absorber (cont.)

Original New

In 2012 we (at SLAC) were working with INTA Technologies that was granted aPhase I SBIR to develop the new high power RF load.

The load design was proposed at SLAC.

INTA was a responsible for the technology and manufacturing methodologiesincluding welding of Skarpass RF flanges, which are made from dissimilarmaterials.

Tapering concept No tapering concept

Development of New Load Based on High Resistive Absorber (cont.)

During the Phase I INTA fabricated a pilot version of the RF Load that had been conditioned and tested up to 30 MW peak power at SLAC.

We (SLAC & INTA Technologies) did not get a Phase II SBIR support. INTA Technologies stopped to work on this project in 2013.

30 MW peak, 1 kW average, 30 Hz, cooling by two fans

Development of New Load Based on High Resistive Absorber (cont.)

Main received results: • The plasma deposition of ferritic FeCrAl alloy onto modified rectangular

waveguide with kerfs on H-planes is a promising concept. • The technology achieved a 5.5 dB/m attenuation rate at 30 MW peak power and

1 usec pulse width with no evidence of breakdown activity in the vacuum envelope on the coated surfaces.

• All technological aspects of RF load fabrications had been practically solved.• SLAC Relies Drawings SA-701-267-10 was set for a fabrication in this year.

Remaining for a development of the load prototype :• Demonstrate a reliable RF absorption at 5 kW average RF power• Demonstrate an acceptable lifetime at 20 MW peak and 5 kW average• Conduct optimization of plasma coating to get an uniform RF absorption

along the load length• Cost optimization

Where Are We Now with the SLAC RF Load Design?APS Materials delivered coated aluminum extruded angles for eight load prototypes.Plasma spray quality for three pilot loads was not accepted after their visual inspection. Company will rework. A preparation of rejected pieces for a shipment back to APS is in progress. The accepted parts are in the fabrication process.

Where Are We Now with the SLAC RF Load Design? (cont.)

VNA cold test of the WG parts which were accepted and welded

Next steps are:Welding flanges, leak check, bake up cycles, test @ mW and kW RF power levels after temperature cycles, assembly of cooling system, and in final, to perform an RF test at bldg044, etc.

Pilot S-Band Dry Vacuum Load for SLAC Linac

Skarpass RF Flange made by the Al-SS explosion bond technology

Liquid cold plates: Cu pipes enclosed into Al alloy housings

Extruded Al alloy angles with kerfs, a surface of which are plasma coated by the FeCrAlYmatrix

CF 8 l/s vacuum portwith the Al-SS interface

Estimated cost is less than $5k per each

Al extrusions was from APEL Extrusion, Inc. Plasma coating performed by APS Martials, Inc.Dissimilar Al-SS plates purchased from Spur IndustriesLiquid cold plates fabricated by Wakefield-Vette, Inc. SLAC labor

How RF Load Activity Does Relate to Improvement of the LCLS Phase Stability? **)

A qualified speculation:

RF Power propagates from klystron to the RFload through an accelerating waveguide. Thepresent RF load can not absorb the residualRF power and creates a local vacuum burstduring each pulse. A result of the vacuum burstpropagates into the beam line. A baselinepressure will be higher than for a case of aproper RF power termination. The collision ofthe bunch during its acceleration with theresidual gas is higher accordingly. A result ofthe collision has two folds: (1) the nonrecoverable emittance rise and (2) the pulse-to-pulse jitter due to statistical uncertainty of acollision. Both effects are not needed forLCLS.

**) PJ is acceptable but we are looking the improvements