rf design of the power coupler for the spiral2...
Post on 02-Feb-2020
8 Views
Preview:
TRANSCRIPT
RF DESIGN OF THE POWER COUPLER FOR THE SPIRAL2 SINGLE BUNCH SELECTOR*
F. Consoli, A. Caruso, G. Gallo, D. Rifuggiato, E. Zappalà, INFN-LNS, Catania, Italy M. Di Giacomo#, GANIL-SPIRAL2, Caen, France
Abstract
The single bunch selector of the Spiral2 driver uses high impedance travelling wave electrodes driven by fast pulse generators. The characteristic impedance of 100 Ohm has been chosen to reduce the total power, but this non standard value requires the development of custom feed-through and transitions to connect the pulse generators and the matching load to the electrodes. The paper reviews the design of these devices.
INTRODUCTION The single-bunch selector of the Spiral2 accelerator
reduces the bunch repetition rate onto the experimental target of a factor 100 to 10000. The device is constituted of a static magnetic deflector and of a pulsed RF kicker whose fields are perfectly compensated for the selected bunch. All other bunches are deviated onto a beam stop by the magnetic field. This principle, shown in Fig. 1, inverts the duty cycle required for the RF kicker and uses constant length pulses.
Figure 1: Principle of the inverted duty cycle single bunch selector.
This solution was developed in the framework of the Eurisol Design Study [1] and is justified by the high repetition rate, high voltage, and fast transient time required by high intensity ion drivers. Table 1 reports the values for the Spiral2 application.
Table 1: Single Bunch Selector Requirements
Parameter Value Units
Repetition rate 1 MHz
Rise and fall time 6 ns
Pulse total length 19 ns
Pulse max voltage 2.5 kV
Pulser power 1 kW
Table 2 gives the operating parameters for a plate
length of 546 mm and for the different kinds of ions
(identified by their mass to charge ratio A/q) of the Spiral2 driver. The difference between the beam and the plate voltages is due to the coverage factor of the meander plate, which is ~0.75.
Table 2: Voltage Requirements for the SBS
A/q Beam Voltage Pulse Voltage
1 473 V 630 V
2 792 V 1056 V
3 1182 V 1576 V
6 1773 V 2364 V
High voltage switches being too slow and not enough
powerful, the travelling wave solution was proposed in the inverted duty cycle configuration in order to reduce the involved RF power.
The electromagnetic field of the RF kicker is generated by two pulsers of opposite sign travelling along two meander-shaped microstrips (beta 0.04), each one ended by a matching load.
Characteristic impedance (Zc) of 100 Ohm has been chosen as a compromise among pulser RF power, meander loss, and meander feasibility.
Figure 2: Schematics of the meander feeding chains.
As shown in Fig. 2, the signal of each generator is transmitted via 100 Ohm cables supplied by the pulser manufacturer. From the cable, the pulse travels through the feed-through, where the vacuum window is located, and a short section of transmission line before reaching the transition to the meander strip. A similar path is followed at the output to reach the matching load.
All the feeding chain elements have been designed to minimize the insertion and transition loss to transfer the best possible pulse. CST Microwave Studio was used for RF simulation in both time and frequency domains and results of the different simulations are reported here. An
+ HV Pulser
Beam
- HV Pulser
Static B- field
z
R
RTravelling E- field
• • • • •
• • • • •
• • • • •
• • • • • • • • • •
• • • • •• • • • •
• • • • •• • • • •
1 kW – 100 Ωdummy load
100 Ω coaxial line
Connector
VACUUM CHAMBER100 Ωvacuum feedthrough
Transition between the 100 Ω line and the meander electrode
Pulse generator (~1 kW)
100 Ω coaxial line
Connector
100 Ωvacuum feedthrough
Transition between the 100 Ω line and the meander electrode
Meander line
___________________________________________
*Work supported by the European Community FP7 – Capacities –SPIRAL2 Preparatory Phase n° 212692. #digiacomo@ganil.fr
TUP277 Proceedings of 2011 Particle Accelerator Conference, New York, NY, USA
1346Cop
yrig
htc
2011
byPA
C’1
1O
C/I
EE
E—
ccC
reat
ive
Com
mon
sAtt
ribu
tion
3.0
(CC
BY
3.0)
Accelerator Technology
Tech 28: Subsystems and Technology (Other)
analysis bandthe pulse spe
Feed-ThroThe feed-t
cable are strtwo custom disk (b) vacuHN 50 Ohm
Figure 3: Simgeometry a)
Solution abandoned bwith the diskdetails. The iand the feedmachining. TMV/m, lowconsidered ointroduced bsparking riskthat the Tefldone in HV is granted byside, the diaminner conducthis value it the resulting electric fielddielectric stre
To simulaapproximatiosince it hasnfrom the datadielectric, anbe worse duAlumina, usgeometry incwhich has chand the sectiline on the vof each sectio
Figure 4: HN(left size) and
Comparativedeformation
dwidth of 150ectrum.
POWER
ugh and Cothrough and tictly related. 100 Ohm de
uum windowsstandard cabl
mulated 100 Ois almost twic
(a) has beebecause it reqk break (b), winner conduct
d-through is toThe maximum
wer than the of 3 MV/m. Cbetween the k. In fact, theon can be supstandard conny tight mechmeter ratio chctor diameter
is more diffiouter diamete
d is ~2 MVength value. ate the set oons in the mn’t been possia sheets. Only
nd probably thue to the higsed for the vcludes the secharacteristic imon before the
vacuum side, on are shown
N simulated d ending trans
e results for thare shown in
0 MHz has b
R COUPLE
onnector the connectionThree cases hvices, using c
s as shown in le and feed-thr
Ohm feed-throce smaller than
en studied quires more rwhich has beeor connection
o be brazed tom electric field
air dielectriConsequently,two conductoe disc Aluminperposed to thnectors. The anical contachanges to Zc is chosen of 1icult to braze er is 6.4 mm. V/m, well be
of standard Hmodelling had
ble to have py Teflon has b
he feed-througgher dielectriacuum windo
ction where thmpedance higshort transitiowhere Zc is ~in Fig. 4.
structure withsition to 100 O
he S parametern Fig. 5, wher
een considere
ER
n to the 100 have been stucylindrical (aFig. 3 and a srough connect
ughs. Scale on that of case
first but raroom than theen studied mon between the o avoid compd at this point ic strength v, a Teflon shors to reducena is T-shapehe ceramic, asground conne
ct. On the vac= 100 Ohm.
1.2 mm, as bethe Alumina
The corresponelow the vac
HN devices, to be introd
precise informbeen consider
gh behaviour wic constant oow. The simuhe cable is inseher than 100 Oon to the 100 ~60 Ohm. Le
h 100 Ohm cOhm (right siz
rs and for the re (c) indicate
ed for
Ohm udied: ) and set of tors.
of b).
apidly e one ore in cable licate is ~2
value, hell is e the ed so s it is ection cuum The
eyond a, and nding cuum
some duced, mation red as would of the ulated erted, Ohm, Ohm ngths
cable ze).
pulse es the
50 Othe f
FiguthrouFig.
100Be
100 equippumThe trans
Thwhilthe t
CouTo
sectithe c
F
Ohm geometryfunction:
ure 5: S paramughs of Fig. 34. Time doma
0 Ohm Coaxetween the feeOhm coaxial
pped with lmping inside th
slots do nosmission loss,
Figure 6: Slo
he ground coe the inner coip ends onto a
upler Mechao avoid the pion of line is coupler has to
Figure 7: Feed
y. The input p
meters for the and for the H
ain results for
xial Line ed-through anl line is inseongitudinal she volume coot change sias shown in F
otted coaxial l
onnection is onductor is paa pad of the m
anics presence of inused as showface the trans
d-through and
pulse is appro
(a) and (b) 10HN 50 Ohm st
the (b) and (c
nd the meanderted. The outslots for prooncerned by tignificantly iFig. 6.
ine and S para
granted by tart of the feed
meander strip.
nternal spacewn in Fig. 7. sition collar.
d 100 Ohm slo
oximated with
(1)
00 Ohm feed-tructure (c) ofc) cases.
er transition, ater coaxial isoper vacuumthe RF fields.insertion and
ameters.
tight contact,d-through and
ers, a straightConsequently
otted line.
h
-f
a s
m .
d
, d
t y
Proceedings of 2011 Particle Accelerator Conference, New York, NY, USA TUP277
Accelerator Technology
Tech 28: Subsystems and Technology (Other) 1347 Cop
yrig
htc
2011
byPA
C’1
1O
C/I
EE
E—
ccC
reat
ive
Com
mon
sAtt
ribu
tion
3.0
(CC
BY
3.0)
TRANSITBoth long
been investigline and the similar behashape.
Figure 8: geometries apulse respon
As a 1 kWthe contact talso grant thAlumina. Bwarming the 6 A of DC cu
All the prdifferent eleresults to thethe perpendic78-periods m
Different coupler of ththe transitionis the feed-thcoaxial linedeformation.through baseresults, as theof nanosecoprinciple beconclude thnegligible wi
TION TO Tgitudinal and gated for the 100 Ohm mic
aviour, with
Longitudinal and comparisses.
W pulse has to to the strip hahe cooling of
Brazing on thplate up to so
urrent. revious simulements of the pulse deformcular transitio
meander shown
CONCsolutions hav
he single buncn to the microhrough, the me having ve The coupler
ed on a disk e transient timonds only.
e achieved wat the effectith respect to t
THE MEA perpendicultransition be
crostrip. Resunegligible ef
(a) and pson of their
be transferredas to be brazef the coaxial he microstripome 45°C by d
lations show he coupler. Tmation due to ton has been usn in Fig. 9.
CLUSIONSve been studich selector meostrip. The mo
meander transitry low effedesigned withvacuum wind
mes are slowedAcceptable
with the HN t of the desthe pulse defo
NDER STRar solutions etween the coults of Fig. 8 ffect on the
perpendicularS parameters
d, we consideed. In this wa
line wire vip is possibldriving it with
the effect oTo compare the electrode ised with the w
S ied for the peander line anost critical eletion and the slect on the h the custom dow gives thed down of fewresults coulset too. We
signed couplormation due t
RIP have
oaxial show pulse
r (b) s and
er that ay we ia the e by h 5 or
of the these itself, whole
power nd for ement lotted pulse feed-
e best w tens d in
e can ler is to the
lengtof thhavechamcoupquansolut
Figumean
Thwho equareali
[1]
[2]
th, the attenuahe meander line been studiedmber, two waplers is being dntify maximution.
ure 9: Pulsnder line fed b
ACKhe authors are
reviewed thation (1) for stic simulation
G. Le Dem,Selector for tion beam", USA, 25-29 Jhttp://www.JAP. BalleyguieFremont, PImprovementSelector", LSeptember 20
ation and the ne. Manufactud too and a proater cooled edesigned at INum perform
se deformatioby the perpend
KNOWLEDe grateful to Phe design wothe input signs.
REFEREM. Di Giacthe next low PAC'07, AlbJune 2007, MACoW.org. er, M. Di GP. Bertrandt in the SLINAC2010, 010.
frequency depuring and assemototype includelectrode holdNFN-LNS. Po
mances of th
on due to thdicular transit
DGEMENTP. Balleyguier ork and sugggnal, in order
NCES como: "A Sβ continuous
buquerque, NMOPAN008, p
iacomo, M. d: "ElectroSPIRAL2 Si
Tsukuba, J
pendent delaymbling issues
ding a vacuumders and fourower tests willhe proposed
he 78-periodtions.
T (CEA/DAM)
gested to user to get more
Single Bunchs wave heavyNew Mexico,p. 158 (2007);
Michel, G.ode Designingle BunchJapan, 12-17
y s
m r l d
d
) e e
h y , ;
. n h 7
TUP277 Proceedings of 2011 Particle Accelerator Conference, New York, NY, USA
1348Cop
yrig
htc
2011
byPA
C’1
1O
C/I
EE
E—
ccC
reat
ive
Com
mon
sAtt
ribu
tion
3.0
(CC
BY
3.0)
Accelerator Technology
Tech 28: Subsystems and Technology (Other)
top related