FIRST CONSIDERATIONS CONCERNING AN OPTIMIZED CAVITYDESIGN FOR THE MAIN LINAC OF BERLinPro

Bernard Riemann, Thomas Weis, TU Dortmund University, Dortmund, Germany∗

Axel Neumann, Wolfgang Anders, Jens Knobloch, HZB, Berlin,GermanyHans-Walter Glock, Carsten Potratz, Ursula van Rienen, Rostock University, Rostock, Germany†

Frank Marhauser‡ , JLAB, Newport News, Virginia, USA

Abstract

The Berlin Energy Recovery Linac Project (BERLinPro)is designed to develop and demonstrate CWLINAC tech-nology and expertise required to drive next-generation En-ergy Recovery Linacs (ERLs). Strongly higher order mode(HOM) damped multicell 1.3 GHz cavities are required forthe main linac.

The optimization of the cavities presented here is primar-ily based on a comparison of the JLab 1.5 GHz 5-cell high-current cavity design with a 7-cell Cornell design. Whilethe JLab cavity was scaled to 1.3 GHz and extended to 7cells, we integrated JLabHOM waveguide couplers into theCornell structure. Modifications to the end group designhave also been pursued, including the substitution of onewaveguide by a HZB-modified TTF-III power coupler.

INTRODUCTION

HZB is building a demonstrator ERL facility that aimsat low emittance and high current operation at 100 mA [1].Preliminary results of the studies for the BERLinPro mainlinac are discussed. For ps long bunchesHOM power in theorder of 100 W/m can be generated in a frequency band upto ≈ 100GHz.

Optimizing cavity shape and cells per cavity is an im-portant method to reduceHOM power beforehand. Sincesuperconducting resonators are used, it is desirable to min-imize Esurf/Eacc by iris radius manipulation of base cellswhile maintaining sufficient intercell coupling to preventHOMs being trapped inside the structure.

To maximize the average acceleration gradient of themain linac module, considerations are based on two differ-ent, existing cavities that were extended to 7 cells, whichis a compromise betweenHOM extraction and acc. gradient(see Fig. 1). The remainingHOM power will be extractedwith specialized waveguide couplers (see Fig. 2) that arehoped to provide efficient coupling while minimizing beamdisturbance. The waveguide couplers should be effectiveup to high frequencies and require minimal space in lon-gitudinal direction. Corresponding models of 7-cell cavit-ies with waveguides, including a modified TTF-III powercoupler [2], were evaluated in search for unwanted high-QHOM resonances using wakefield and eigenmode solvers.

∗This work is partly funded by BMBF contract no. 05K10PEA† This work is partly funded by BMBF contract no. 05K10HRC‡ Now at Muons, Inc.

10JLAB single-cell cavity, 1.15-scaled, 1301.7919 MHz

60

Cornell single-cell cavity, 1300.1897 MHz

10

60 s [cm]s [cm]

r [cm] r [cm]

Figure 1: Comparison of Cornell (left) [3] and scaled JLab(right) [4] base cells withπ-mode electric field lines [5].

In recirculatingBBU theory for ERLs and for an un-coupled lattice, the current treshold for a single cavityroughly depends onR⊥/Qloss · Qtot of the considered de-flecting mode [6]. Unlike in storage rings with one ei-genfrequency for each direction (betatron and synchrotrontunes), theHOM frequencies of all cavities form sidebandsalong the filling pattern, and the alternating stable and un-stable frequency bands have a width of only a half circula-tion frequency (far-field wake). Therefore, everyHOM hasthe high probability of1/2 to generate positive feedback,and the external quality factors of all TM-like dipole higherorder modes need to be minimized to a sufficient extent. Inaddition, conventional linacBBU can occur at harmonics ofthe bunch repetition rate and must also be taken into ac-count for all modes.

STRUCTURE DESIGN

Esurf/Eacc Optimization of Base Cells

Base cell studies with longitudinalπ-periodic boundaryconditions for the fundamental mode have been performedusing the 2D codeSUPERFISH[5] (see Fig. 3). The irisradius was gradually decreased from the (scaled) originalradius (JLab 40.2 mm, Cornell≈ 36mm) to 30 mm. Thedesign goal of 2.0 could not be reached by this decrease forthe JLab standard cell.

Module Simulation Setup

For a first estimation of relevantHOM resonances, bothmulticell designs, consisting of the two different base celldesigns with corresponding endcells, 5 waveguide couplersand one modified TTF-III power coupler, were simulatedusing a wakefield solver [7] to wake lengths of≈ 500m

Proceedings of IPAC2011, San Sebastián, Spain MOPC080

07 Accelerator Technology

T07 Superconducting RF 259 Cop

yrig

htc ○

2011

byIP

AC

’11/

EPS

-AG

—cc

Cre

ativ

eC

omm

onsA

ttri

butio

n3.

0(C

CB

Y3.

0)

Figure 2: left: preliminary sketch of the 3-fold symmet-ric waveguide. right: asymmetric waveguides and powercoupler in the Cornell-type model.

3 3.1 3.2 3.3 3.4 3.5 3.6 3.7 3.8 3.9 41.9

2

2.1

2.2

2.3

2.4

2.5

Iris radius [cm]

Esurf

/ E

acc

Figure 3: Surface-to-acc. gradient ratio for different irisradii of the 1.15-scaled JLab (blue) and Cornell cell(green). The maximum iris radius is the standard radius.

1

2

3

4

5

0 1 2 3 4 5 6f [GHz]

Wa

ke

Im

pe

da

nce

[a

.u.]

Figure 4: Comparison of longitudinal wake impedances ofCornell- (red) and Jlab-based (blue) cavities with modifiedTTF-III couplers and waveguides.

1.4 1.6 1.8 2 2.2 2.4 2.6 2.810

0

101

102

103

104

105

106

107

108

109

Frequency [GHz]

Qext

7 cell scaled

5 cell monopole

5 cell dipole

5 cell dipole

5 cell quadrupole

Figure 5: Results of eigenmode calculations forQext. The7-cell 1.15-scaled JLab-type cavity is compared toQext

data from the original 5-cell JLab-type cavity [4, 8].

0 1 2 3 4 5 61

2

3

4

5

6

f [GHz]

Wake Im

ped

an

ce [a

.u.]

Figure 6: Wake calculation Results (blue: longitudinal,red/green: transversal) mixed with shunt impedances (max-imum of longitudinal shunt impedances sampled on axisand two parallel off-axis paths) of eigenmode solvers (gray)for the Cornell-based model with modified TTF-III couplerand asymmetric waveguide configuration.

(∆ν ≈ 600 kHz, νmax = 6GHz, see Fig. 4) with a bunchpath parallel to the design axis. The endcells were takenfrom the original geometries, e.g. JLab endcells are 4,2%shorter than base cells.

This approach allows to observe dipolar and highermodes and to separate most spectral lines. Additionalfield probes were included for localization of modes (seeFig. 8). Subsequently, small frequency bands with con-siderable wake impedances are investigated using Jacobi-Davidson and Krylov-subspace eigenmode solvers [7], re-trieving Qext (see Fig. 5) andR/Q of the relevant modes(until now up to 4.3 GHz, see Fig. 6).

MOPC080 Proceedings of IPAC2011, San Sebastián, Spain

260Cop

yrig

htc ○

2011

byIP

AC

’11/

EPS

-AG

—cc

Cre

ativ

eC

omm

onsA

ttri

butio

n3.

0(C

CB

Y3.

0)

07 Accelerator Technology

T07 Superconducting RF

Figure 7: Overview and detail of the modified 7-cell JLabcavity with one modified TTF-III coupler. Note that fortechnical reasons the waveguide couplers are bent out-wards, while they will be bent inwards inside the cryomod-ule.

Figure 8: Field intensity in cell probes of JLab-type modelwith modified TTF-III coupler (colors) and correspondingoutgoing port power (black).

Modified TTF-III Coupler and Waveguide Modi-fications

The HZB-modified TTF-III power coupler works withhigh input power up to 10 kW and allows adjustment ofQext in operation [2]. Thus it was included in both cavities,substituting one of the six waveguides, and the remainingtwo waveguides were adjusted to create a right angle (seeFig. 7).

Studies of introducing more asymmetric features to thewaveguide angles are in progress to further account for thesubstitution of one waveguide (see Fig. 2).

To find the right antenna position, the surface distance ofthe antenna from the beam axis was manually optimized toyieldQext ≈ 4 · 107. The corresponding surface distance is58mm.

Choice of Tuner

It was considered to move the waveguides away fromthe endcell for installation of a Saclay-type tuner. For theJLab-type cavity, this would result into a decrease ofHOM

coupling by four orders of magnitude because manyHOMsare evanescent for the given beam tube radius. Therefore,

blade tuners will be used which do not require any addi-tional space between endcells and waveguides.

OUTLOOK

The results of this study will be used forBBU calcula-tions with different lattice configurations. Based on the out-come of this calculations, a decision will be made which ofboth cavity designs will be further developed. This cav-ity will be used for construction of a modular, normal-conducting brass model consisting of lateral half cell brasselements (cut in half along an axis parallel to beam path)and different endgroups. The model will be characterizedusing a bead-pull measurement stand at TU Dortmund Uni-versity.

SUMMARY

We have started design considerations for theBERLinPro main linac on basis of two known linaccavity designs that were changed for high acc. gradient,flexible input power coupling and higher order modedamping with waveguides.

The first results obtained by electromagnetic simulations(base cell optimization, external quality factors ofHOMs,shunt impedances, power coupling) have been presentedand act as input to further numerical studies.

REFERENCES

[1] A. Jankowiak et al., “BERLinPro - a compact demon-strator ERL for high current and low emittance beams”,Proc. LINAC 10, Tsukuba, Japan, 2010.

[2] O. Kugeler et al., “Warm test of a modified TTF-IIIinput coupler up to 10 kW CW RF-power”, Proc. SRF09, Berlin, Germany, 2009.

[3] N. Valles and M. Liepe, “Seven-cell cavity optimiza-tion for Cornell’s energy recovery linac”, Proc. SRF09, Berlin, Germany, 2009.

[4] F. Marhauser, “JLab high current cryomodule develop-ment”, Proc. ERL 09, Ithaca, NY, 2009.

[5] K. Halbach and R.F. Holsinger, “SUPERFISH - a com-puter program for evaluation of RF cavities with cyl-indrical symmetry”,Part. Acc., 7:213–222, 1976.

[6] G.H. Hoffstaetter and I.V. Bazarov, “Beam-breakup in-stability theory for energy recovery linacs”,Phys. Rev.ST Accel. Beams, 7(5):054401, May 2004.

[7] CST AG, “CST Studio Suite”, Darmstadt, Germany,2011.

[8] R.A. Rimmer et al., “Recent progress on high-currentSRF cavities at JLab”, Proc. IPAC 10, Kyoto, Japan,2010.

Proceedings of IPAC2011, San Sebastián, Spain MOPC080

07 Accelerator Technology

T07 Superconducting RF 261 Cop

yrig

htc ○

2011

byIP

AC

’11/

EPS

-AG

—cc

Cre

ativ

eC

omm

onsA

ttri

butio

n3.

0(C

CB

Y3.

0)