CLIC Kickers

1 Introduction

Challenging kicker systems are required for both the Main Beam and Drive Beam of CLIC: these kickers systems will be briefly discussed in the following sections.

2 Main Beam: Damping Ring and Pre-Damping Ring

The CLIC design relies on the presence of Pre-Damping Rings (PDR) and Damping Rings (DR) to achieve the very low emittance, through synchrotron radiation damping, needed for the luminosity requirements of CLIC. In order to limit the beam emittance blow-up due to oscillations at extraction the combined flat top ripple and droop of the field pulse, for the DR extraction kickers, must be less than ±0.02%. In addition, the total allowable beam coupling impedance, in each ring, is also very low: 1 Ω/n longitudinally and 10 MΩ/m transversally. This section discusses initial ideas for achieving the demanding requirements for the PDR and DR kickers.

The design of the injectors for CLIC is based on a central complex, housing all the subsystems, to prepare the main beams. The main beams are subsequently transported, via two long transfer lines, to the starting point of each main linac at the extremities of the collider facility. To achieve high luminosity at the Interaction Point (IP), it is crucial that the beams have very low transverse emittance: the Pre-Damping Ring (PDR) and Damping Ring (DR) damp the beam to an extremely low emittance in all three dimensions. The PDR is required to decouple the wide aperture requirements for the incoming beams from the final emittance requirements of the main linac. The design parameters of the PDR and DR are dictated by target performance of the collider (e.g. luminosity), the injected beam characteristics or compatibility with the downstream system parameters: the emittances of the beams in the damping rings must be reduced by several orders of magnitude [1].

Table 1: PDR & DR Kicker Specifications

Parameter PDR DR

Beam Energy (GeV) 2.86 2.86

Deflection Angle (mrad) 2 1.5

Aperture (mm) 40 20

Field rise and fall time (ns) 700 1000

Pulse flat top duration (ns) ~160 ~160

Flat top reproducibility 1x10-4 1x10-4

Injection stability (per system) ±2x10-2 ±2x10-3

Extraction stability (per system) ±2x10-3 ±2x10-4

Injection field homogeneity (%) ±0.1A ±0.1A

Extraction field homogeneity (%) ±0.1A ±0.01B

Repetition rate (Hz) 50 50

Available length (m) ~3.4 ~1.7 AOver 3.5 mm radius. BOver 1 mm radius.

Kickers are required to inject beam into and extract beam from the PDRs and DRs. Jitter in the magnitude of the DR extraction kicker waveform translates into beam jitter at the IP [1]. Thus the PDR & DR kickers, in particularly the DR extraction kicker, must have a very small magnitude of jitter. Table 1 shows the specifications for the PDR and DR kickers [2]: the specified stabilities include all sources of contributions such as ripple and droop. The values in Table 1 will be refined as the optics design progresses.

3 Beam Coupling Impedance

The allowable broad band impedances, in the CLIC PDR and DR, are 1 Ω/n for longitudinal beam coupling impedance [4] and 10 MΩ/m in the transverse plane: these values would result in beam stability against single bunch effects [3]. Since the allowable impedances are for the complete PDR and DR, which are composed of many systems including both injection and extraction kicker systems, the permissible beam coupling impedances, per kicker system, are assumed to be 5% of the longitudinal impedance allowance, i.e. 0.05 Ω/n, and 2% of the transverse impedance allowance, i.e. 200 kΩ/m [4].

1.E-02

1.E-01

1.E+00

1.E+01

1.E+02

1.E+03

1.E+04

0 100 200 300 400 500 600 700 800 900 1000

Fer

rite

load

ed:

Rea

l im

peda

nce

per

unit

le

ngth

(Ω

/m).

Stri

plin

es:

Tot

al im

peda

nce

(Ω).

Frequency (MHz)

MKE: L-type (Hap=147.7mm, Vap=35mm), no serigraphy, measuredMKE: L-type (Hap=147.7mm, Vap=35mm), fully serigraphed, measuredMKI: (Hap=54mm, Vap=54mm), unshielded, analyticalMKI: (Hap=54mm, Vap=54mm), 15 screen conductors, measuredDAΦNE striplines: TOTAL Impedance0.05Ω/n: PDR (diameter=398m)

Fig. 1: Longitudinal beam coupling impedance, both with and without beam impedance reduction techniques, for an MKE magnet (real impedance), an MKI magnet (real impedance) and DAΦNE striplines (total impedance, 0.94 m overall length). A fraction (5%) of the allowable longitudinal

beam coupling impedance (Z/n), for the PDR, is also shown.

Fig. 1 shows the real part of the longitudinal beam coupling impedance for an SPS extraction (MKE) magnet [5] and an LHC injection (MKI) magnet [6] in Ω/m length of ferrite: if an equivalent ferrite loaded kicker magnet with a length of 3 m is used for a CLIC kicker system, the real part of the longitudinal beam coupling impedance will be three times that shown in Fig. 1. The total longitudinal beam coupling impedance of DAΦNE striplines [7], predicted by numerical simulations, is also shown in Fig. 1: these striplines have an overall length of approximately 0.94 m. With or without serigraphy, to reduce the beam coupling impedance, the MKE magnets real longitudinal beam coupling impedance exceeds 0.05 Ω/n at all frequencies. The MKI magnets employ a more effective beam impedance reduction technique [6], and the real part of the longitudinal beam coupling impedance is within specification over the complete frequency range considered above. The DAΦNE striplines have a real part of the longitudinal impedance of less than 1 Ω for frequencies above 420 MHz: there is, however,

a peak in the impedance spectrum of ~14 Ω at ~100 MHz. The low-frequency impedance peak of the striplines is a fundamental characteristic of the striplines: ref. [8] shows the following equation for the

longitudinal impedance ( 0Z ) of untapered strip-line Beam Position Monitors (BPMs):

2

200

22 2sin sin

2c

L LZ Z i

c c

(1)

Where:

cZ is the characteristic impedance of the striplines with the beam pipe;

0 is the angle each stripline subtends to the pipe axis (coverage angle);

L is the length of each stripline.

Much research has been carried out, for ILC & DAΦNE, into tapered, elliptical cross-section, striplines and wide-band feedthroughs. An elliptical cross section of stripline minimises the variation of the vertical dimension of the beam pipe between the injection region and the adjacent dipole region and increases the deflection efficiency [7].

By tapering the transition between the stripline structure and the adjacent beam pipe it is possible to [7]:

• reduce the non-uniformity of transverse deflection as a function of the transverse position;

• reduce the beam coupling impedance of the striplines;

• reduce the reflection coefficient at high frequency.

Eq. (1) is valid for untapered striplines which are terminated with impedance cZ at their

upstream end [8]. Eq. (1) can be re-written with a sinl l

c c

term to allow for the effect

of tapers [9].

0123456789101112131415

0 50 100

150

200

250

300

350

400

450

500

550

600

650

700

Longitudinal Im

pedance (O

hms)

Frequency (MHz)

Numerical Analysis

Untapered

25% taper

30% taper

33% taper

50% taper

Fig. 2: Longitudinal beam coupling impedance, for Daphne striplines (~30% taper), as determined by a numerical analysis [10], and Eq. (1).

Fig. 2 shows the longitudinal beam coupling impedance for the Daphne striplines (taper, at each end, of ~30% [10] of the 0.94 m overall stripline length [“Numerical Analysis”]) together with the results of Eq. (1) modified for the effect of tapers [9]. After the first low-frequency impedance peak the analytical solution for a 30% taper is very similar to the numerical analysis of the Daphne striplines, providing confidence in Eq. (1) modified to include tapers.

As pointed out in [11] the uniformity of the deflecting field, as a function of the transverse coordinates, for a given transverse section of the kicker, depends on the coverage angle of the striplines. The optimum case, for circular electrodes, is for a coverage angle of 80 degrees; even so the inhomogeneity of the field can be a few tens of percent inside the good field region [11]. However, for CLIC, although the field uniformity specifications are extremely demanding (Table 1) the required good field uniformity region is relatively small. Simulations have commenced to study the field uniformity issues.

Eq. (1) shows that for a coverage angle of 2 radians per stripline, a peak of longitudinal

beam coupling impedance of 4cZ Ohms occurs at a frequency close to 4c L : hence the

frequency of the peak is dependent upon the length of the stripline. Thus, from Eq. (1), for a 50 Ω characteristic impedance and a stripline length of 0.94 m, a longitudinal impedance peak of 12.5 Ω is expected at 80 MHz. The frequency of this impedance peak is moved upwards by tapers and is approximately 100 MHz with 25% tapers, which is reasonably consistent with the DAPHNE stripline simulation results shown in Fig. 1.

From Fig. 1, striplines have a significantly lower longitudinal beam coupling impedance, than a screened MKI magnet, at frequencies above 400 MHz: in addition, the imaginary component of the longitudinal beam impedance (not shown in Fig. 1) is known to be significant, for the MKI magnet, above 600 MHz. The use of striplines rather than a screened ferrite loaded magnet is supported by experience at KEK/ATF, where metallized ceramic tubes were used to reduce the beam coupling impedance of ferrite loaded kickers: the thickness of the metallization was difficult to accurately control and, as a result, two kickers had very different pulsed magnetic characteristics [12].

0

2

4

6

8

10

12

14

16

0 25 50 75

100

125

150

175

200

225

250

275

300

325

350

375

400

Lon

gitu

dina

l Im

pedn

ace

(Ohm

s)

Frequency (MHz)

1.3m striplines, no taper

1.3m striplines, 10% taper

1.3m striplines, 20% taper

1.3m striplines, 30% taper

0.05Ω/n: DR (diameter=421m)

Fig. 3: Longitudinal beam coupling impedance, for striplines of 1.3 m overall length (length suitable for DR), with a taper at each end (expressed as a percentage of overall length) calculated from Eq. (1) modified for

the effect of tapers.

Fig. 3 shows longitudinal beam coupling impedance, for striplines of 1.3 m overall length (length suitable for DR), with a taper at each end whose length is expressed as a percentage of overall length; the longitudinal beam coupling impedance is calculated from Eq. (1) modified for the effect of

tapers. The permissible longitudinal beam coupling impedance, of 0.05 Ω/n per kicker system [4], is also shown on Fig. 3: short tapers are required, for 1.3m long striplines, to ensure that the longitudinal beam coupling impedances in the range from 150 MHz to 180 MHz, are less than 0.05 Ω/n. Depending upon the degree of tapering, the calculated longitudinal impedance at 55 MHz is a factor of 2.5 to 3.3 greater than 0.05 Ω/n.

0

2

4

6

8

10

12

14

16

0 25 50 75

100

125

150

175

200

225

250

275

300

325

350

375

400

Lon

gitu

dina

l Im

pedn

ace

(Ohm

s)

Frequency (MHz)

3m striplines, no taper

3m striplines, 10% taper

3m striplines, 20% taper

3m striplines, 30% taper

0.05Ω/n: PDR (diameter=398m)

Fig. 4: Longitudinal beam coupling impedance, for striplines of 3 m overall length (length suitable for PDR), with a taper at each end (expressed as a percentage of overall length) calculated from Eq. (1) modified for

the effect of tapers.

Fig. 4 shows longitudinal beam coupling impedance, for striplines of 3 m overall length (length suitable for PDR), with a taper at each end whose length is expressed as a percentage of overall length; the longitudinal beam coupling impedance is calculated from Eq. (1) modified for the effect of tapers. The permissible longitudinal beam coupling impedance, of 0.05 Ω/n per kicker system [4], is also shown on Fig. 4: tapers of ~25% are required to ensure that the longitudinal beam coupling impedances in the range from 60 MHz to 100 MHz, are less than 0.05 Ω/n. Depending upon the degree of tapering, the calculated longitudinal impedance at 30 MHz is a factor of 5.5 to 6.1 greater than 0.05 Ω/n. However, if the low frequency longitudinal beam coupling impedance is an issue, the 3 m overall length of striplines could be sub-divided into two series 1.5 m lengths: the shorter striplines have a lower ratio of the maximum value of calculated longitudinal impedance to 0.05 Ω/n, at the low frequency peak of longitudinal beam coupling impedance.

Further beam stability simulations are required, for both the PDR and DR, to assess the effect of the longitudinal beam coupling impedance exceeding 0.05 Ω/n at low frequencies [4].

Ref. [8] shows the following equation for the transverse impedance ( 0Z ) of strip-line BPMs:

2

20 00 2

0

4sin

2pair

Z cZ

b

(2)

Where:

b is the radius of the (round) beam pipe;

0Z is calculated from either Eq. (1) or of Eq. (1) modified to include tapers [9].

Fig. 5 shows transverse beam coupling impedance, for striplines with an overall length of 1.3 m

(length suitable for DR), with a taper at each end whose length is expressed as a percentage of overall stripline length: the transverse impedances are calculated from Eq. (2), assuming a beam-pipe radius of 0.024 m. A transverse broadband impedance of less than 10 MΩ/m, for the DR, would result in beam stability against single bunch effects [3]. The allowable transverse impedance per kicker system is assumed to be 2% of the beam stability criteria [4], i.e. 200 kΩ/m: the 1.3 m striplines are below this limit.

0.00E+00

2.50E+04

5.00E+04

7.50E+04

1.00E+05

1.25E+05

1.50E+05

1.75E+05

2.00E+05

0 50

100

150

200

250

300

350

400

Tra

nsv

erse

Im

ped

ance

(Ω

/m)

Frequency (MHz)

1.3m striplines, no taper

1.3m striplines, 10% taper

1.3m striplines, 20% taper

1.3m striplines, 30% taper

Transverse impedance limit, per kicker system

Fig. 5: Transverse beam coupling impedance for striplines of 1.3 m overall length (length suitable for DR), with a taper at each end (expressed as a percentage of overall length) calculated from using Eq. (2) [8]. For

calculation purposes, the radius of beam-pipe for the striplines is assumed to be 0.024 m.

Fig. 6 shows transverse beam coupling impedance, for striplines with an overall length of 3 m (length suitable for PDR), with a taper at each end whose length is expressed as a percentage of overall length: the transverse impedances are calculated using Eq. (2), assuming a beam-pipe radius of 0.03 m. A transverse broad band impedance of less than 10 MΩ/m, for the PDR, would result in beam stability against single bunch effects [3]. The allowable transverse impedance per kicker system is assumed to be 2% of the beam stability criteria [4], i.e. 200 kΩ/m: the 3 m striplines are below this limit (Fig. 6).

0.00E+00

2.50E+04

5.00E+04

7.50E+04

1.00E+05

1.25E+05

1.50E+05

1.75E+05

2.00E+05

0 50 100

150

200

250

300

350

400

Tra

nsv

erse

Im

ped

ance

(Ω

/m)

Frequency (MHz)

3m striplines, no taper

3m striplines, 10% taper

3m striplines, 20% taper

3m striplines, 30% taper

Transverse impedance limit, per kicker system

Fig. 6: Transverse beam coupling impedance for striplines of 3 m overall length (length suitable for PDR), with a taper at each end whose length is expressed as a percentage of overall length, calculated from

using Eq. (2) [8]. For calculation purposes, the radius of beam-pipe for the striplines is assumed to be 0.03 m.

3.1 Conclusions re Beam Coupling Impedance

Striplines will be used for the kicker systems in the PDR and DR: their transverse beam coupling impedance is within the allowable 10 MΩ/m per kicker system. In order to limit the magnitude of the longitudinal beam coupling impedance at frequencies after the first, low frequency, peak, tapers of at least 10% and 25% are required for the DR and PDR striplines, respectively.

To achieve the required longitudinal beam coupling impedance in the PDR, instead of using striplines with a length of 3 m, two sets of striplines with an individual length of 1.5 m could be used. The transverse impedance of each set of striplines would be similar, but slightly greater than, shown in Fig. 5: the transverse impedance of two sets of these striplines would hence be slightly greater than twice that shown in Fig. 5. Thus each PDR kicker system would meet the transverse impedance specification.

As mentioned above simulations have commenced to study the field homogeneity of the integrated field. To maximize the efficiency of the stripline kickers, the length of the taper should be minimised, however sufficient taper is required to reduce the HF longitudinal beam coupling impedance to an acceptable level [13]. In addition, tapers affect the non-uniformity of transverse deflection as a function of the transverse position [7, 14], so the length of taper needs to be optimized for the field homogeneity too. Once the electromagnetic simulations are completed, striplines will be prototyped under the Spanish Program “Industry for Science”.

3.2 System Stability

Fig. 7 shows a simplified schematic of a stripline kicker system. The two striplines are driven to an equal magnitude of voltage but of opposite polarity. A High Voltage DC (HVDC) power supply charges a Pulse Forming Network (PFN) or a Pulse Forming Line (PFL). The fast switch is then closed to launch a pulse towards the striplines (note: for simplicity, Fig. 7 only shows one of the two HVDC supplies, PFL/PFN and fast switches). The pulse propagates through the striplines and is then deposited in a terminating resistor. The characteristic impedance of the PFL/PFN, transmission lines, striplines and terminating resistors is matched to avoid reflections, which could cause ripple on the flat top of the deflection waveform. The PFL and fast switch, shown in Fig. 7, could be replaced with an Inductive Adder: this is discussed later in this report.

Fig. 7: Simplified schematic of stripline kicker system

Possible sources of ripple, droop and irreproducibility of the deflection waveform include:

• PFN: a PFL or inductive adder will likely give lower ripple – thus a PFN will not be considered further;

• HVDC supplies (reproducibility is expected to be acceptable for slow charging of PFL);

• Attenuation in the PFL and transmission lines;

• Switch (dynamic characteristic, and both short term and long term temperature effects);

• Feedthroughs;

• Terminating resistor (frequency dependence of value, long-term stability and temperature will affect ripple and reproducibility of the waveform);

• Non-ideal impedance matching of the system.

Several of these items will be discussed in the following sub-sections.

The demanding specifications for droop and ripple can be relaxed for an individual kicker if a double kicker system is employed: in this case the overall response of the double kicker system must meet the specifications.

3.2.1 Double Kicker System

3.2.1.1 Experience at KET/ATF

Extraction from the DR with a single kicker system requires a very uniform and stable field pulse with ultra-low ripple (Table 1). A double kicker system (Fig. 8), consisting of two identical ferrite loaded kicker magnets and a single power supply, has been developed at KEK [12, 15-16]. The first kicker extracts the beam from a damping ring and the second kicker, displaced from the first kicker by a suitable Betatron phase, results in anti-phase ripple to that of the first kicker (Fig. 9).

Fig. 8: 1st and 2nd kickers separated by a betatron phase of 2nπ: for a betatron phase of (2n−1)π

the 2nd kick would be in the other direction.

Fig. 9: Exactly the same flat top ripple from both kickers ideally results in ripple cancellation

Theoretically, using a double kicker system, the effect of ripple in the two kickers and small variations in the output of the HVDC supplies can completely cancel. In addition, theoretically, the double kicker can compensate for field inhomogeneity in each stripline kicker; this therefore permits the field uniformity requirement of an individual stripline to be relaxed, while still achieving excellent overall effective deflection uniformity. However this places demanding requirements on the beam optics.

Measurements have been carried out at the KEK/ATF with a double kicker: the two kickers were nominally identical, ferrite loaded, transmission line magnets. However, to reduce beam coupling impedance, the inside of the ceramic tube in the magnet apertures was coated with 1 µm thick Titanium Nitride (TiN): the thickness of the TiN coating is probably not uniform. This is evidenced by

the fact that the second kicker had an apparent deflection of only 83% of the first kicker (for approximately the same current flow [12]): this deflection was determined from measurements of the beam orbit, shot by shot, in the extraction line [14]. The two KEK kickers each had 25 cells and were manufactured to high mechanical accuracy. In addition the ferrite used (TDK-PE14) was from a single manufactured batch. Thus magnetically the two kickers should perform in a very similar manner.

The phase advance of the two kickers, although nominally π, was experimentally determined to minimise the jitter angle in two kicker mode. Subsequently, to measure the equivalent jitter in single kicker mode, the 2nd kicker was replaced with a dipole [16]: the strength of the dipole was adjusted to give the same average beam trajectory as for the double kicker. The dipole had a small aperture and a high stability power supply thus its field uniformity and field stability are believed to be very good. All measurements were made with single bunches, i.e. at a particular time on the kick field waveform.

BPMs were used to determine the ripple of the double kicker system. The ATF Damping Ring had button BPMs with a positional resolution of 20 µm to 30 µm. The extraction line used both Cavity BPMs (2 µm resolution) and Stripline BPMs (20 µm to 30 µm resolution). The KEK double kicker achieved a factor of ~3.3 reduction in kick jitter angle, with respect to a single kicker [16]. The resolution of the BPMs limited the optimization of the phase advance and thus the jitter angle reduction. The resolution of the Cavity BPMs was limited, at the time, by the readout electronics. Improvement in the electronics now allows a resolution of approximately 100 nm [17].

Research is planned at CERN, into double kicker systems, to try and achieve a greater improvement in jitter reduction: two identical stripline kickers and high precision BPMs will be used.

3.2.1.2 Double Kicker for CLIC

For CLIC, assuming a 10 m separation between the 1st and 2nd kickers, the time of flight is ~33.3 ns for beam and ~50 ns for the kicker current pulse. In order that the beam bunches and kicker field are synchronized in time at the 2nd kicker system either:

• the two kicker systems must be in parallel (Fig. 10), or;

• for a series connection a ~16.7 ns delay loop is required for the beam.

Feeding two parallel kickers, from a common HVDC supply and single switch (Fig. 10) is the best option as it avoids the need for an additional delay loop for the main beam. Fig. 10 shows that, to synchronize in time the beam bunches and kicker field at the 2nd kicker system, there is an additional transmission cable between the switch and second stripline kicker whose temporal delay is the same as the time of flight of the beam. However this transmission cable has the undesirable effect of introducing extra attenuation and dispersion of the electrical pulse arriving at the second stripline kicker in comparison with the pulse arriving at the first stripline kicker.

Fig 10: Double kicker system supplied in parallel

3.2.2 Impedance Matching Considerations

Initial considerations of impedance matching of striplines, for electrical pulses, indicate that to achieve a pulse ripple of not more than 0.02%, impedances should be matched to better than 2.8%. However given the relatively long allowance for field rise and fall times (Table 1), and provided that transmission line lengths are minimised, there may be time for ripple, introduced due to impedance mismatches, to be damped to a suitably low level before the end of the specified rise or fall time. However making use of the relatively long allowance for rise and fall times would have the disadvantage of increasing power dissipation, and hence heating, in the switches and terminating resistors.

Feedthrough connectors can be a significant source of impedance mismatch. KEK/ATF and INFN/DAPHNE both carried out research and development concerning feedthrough connectors to achieve good impedance matching over a wide range of frequencies. The KEK/ATF feedthrough connector, developed using HFSS, had a predicted S11 reflection coefficient below 0.02 up to 300 MHz [16]: this corresponds to impedance matching to 4%. A measurement with a step waveform showed an impedance of up to 58 Ω [16]: not as good as predicted but considerably better than the “old” feedthrough connector (Fig. 11). The new feedthrough connector is shown in Fig. 12.

In order to minimise thermal effects the rise and fall times of the electrical pulse must be as short as feasible. Thus research and development of HV feedthrough connectors is required for the CLIC DR kicker systems in order to be able to achieve the required voltage hold-off together with adequate impedance matching over a wide frequency range.

In addition to the feedthrough connectors, the transmission lines, striplines and terminating resistors are all possible sources of impedance mismatches: the usual datasheet value for the real impedance of suitable HV transmission line, at 1 MHz, is 50 Ω with a tolerance of ±1%. In addition, the impedance matching of the PFL and semiconductor switches is important. For the terminating resistor, due to the presence of parasitic components, the impedance value will be frequency dependent. Temperature and long-term stability will affect ripple and reproducibility of the waveform. All of these affects require further studies.

Fig 11: Measured step-response of old (left) and new (right) KEK/ATF feedthrough connector [16]

Fig. 12: New feedthough connector developed at KEK/ATF

In addition to the above the semiconductor switches used will have an on-state resistance which will be temperature dependent; thus the on-state resistance will depend upon pulse duration. Hence it is desirable to minimise the duration of the electrical pulse, i.e. by minimising the rise and fall times of the electrical pulses and designing the system to minimise impedance mismatching so that ripple is damped rapidly.

3.2.3 Compensation of Droop

If a PFL is used one of several problems, for deflection stability, is PFL droop. Properly impedance matched PFLs deliver low ripple pulses, but low attenuation and dispersion are essential (especially with longer pulses) to control droop and “cable tail”. However it is possible to make use of the frequency dependent attenuation and dispersion, of the transmission line, to compensate for PFL droop, but increased cable tail is a potential problem (Fig. 13).

Fig. 13: PSpice predictions for load current, for a 100 m long PFL of RG220U coax with a

transmission line that is (a) lossless, (b) 60 m of RG220U.

Fig. 13 shows PSpice predictions for load current, for a 100 m long PFL of RG220U coax, with a transmission line that is firstly modelled as lossless then subsequently is modelled as 60 m of RG220U coax. In both cases an ideal 50 Ω terminating resistor is modelled. Over a 160 ns period the lossless transmission line gives a predicted droop of 0.14%; when the 60 m of RG220U is modelled, the load current is flat to within ±0.01% over a 240 ns period: however ±0.01% is significant in comparison with the required ±0.02% stability for the DR extraction kicker system.

3.2.4 Inductive Adder

An Inductive Adder [18-24], instead of a PFL or PFN, is a promising means of compensating for ripple as well as attenuation and dispersion in transmission lines. The inductive adder (Fig. 14) consists of: a multi-cell primary circuit, a single secondary winding, and a fast pulse transformer with adequate voltage isolation. Each primary circuit has a fast switch. The switches can be turned on and off independently, via trigger circuits, to provide some pulse shaping: for the CLIC DR and PDR kickers the control of the primary switches could be digital [18] (on-off), analogue [24] (gain control of semiconductor switch) or a combination of these two schemes. The combined digital and analogue control scheme could involve many primary cells, each primary capacitor charged to a high voltage, controlled in a digital sense and several cells, possibly at lower voltage, controlled in an analogue manner. In order to cancel ripple, by suitable control of primary cells, it is expected that high bandwidth switches will be required.

The inductive adder concept is also good for machine protection and reliability as it inherently contains redundant primary switches. Thus if one or just a few of many primary switches fail, a substantial portion of the required deflection is delivered to the beam.

The inductive adder concept is being studied further and it is planned to build a prototype 20 kV version for evaluation and testing. In order to minimise ripple generated by the inductive adder considerable effort will be made to ensure that, as far as reasonably possible, the characteristic impedance of the adder is matched to that of the system.

Fig. 14: Multi-cell inductive adder

Fig. 15 shows an 18 kV Inductive Adder designed and built at LLNL.

Fig. 15: Multi-cell, 18 kV, inductive adder supplied by LLNL to KEK/ATF

3.3 Main-Beam: Interaction Point Kickers

The design luminosity of CLIC requires transverse beam sizes at the nanometre level at the interaction point (IP), as well as stabilisation of the beams positions at the sub-nanometre level. Different imperfections, for example ground motion, can generate relative vertical offsets of the two colliding beams at the IP which significantly degrade the luminosity. In principle, a beam-based intra-train FeedBack (FB) system in the interaction region can correct the relative beam-beam offset and steer the beams back into collision. In addition, this feedback system may help to considerably relax the required tight stability tolerances of the final doublet magnets. Since the IP-FB system has to operate in an environment with high background radiation, the choice of the position of the IP-FB components is a compromise between the reduction of the latency time and the minimisation of the background/backsplash effects on the FB electronic components.

A fast FB system (Fig. 16) has been developed at Oxford University [31]. The principle of the Feedback On Nanosecond Timescale (FONT) scheme [31] is:

Measure the deflected bunches with a BPM and kick the other beam to eliminate vertical offsets at the IP;

The feedback loop assesses intra-bunch performance and the RF Amplifier dynamically modulates the correction signal to the kicker, assuming that the vertical offset of the next bunch is the same as the offset just measured;

There is a need to minimise the latency of feedback to optimise the collisions in the 156 ns bunch train;

The distance of components from the IP has to be minimised to reduce latency: hence the kicker is to be located within the detector, about 3 m from the IP. Thus radiation hard components are required for the kicker.

A 1.5 TeV e+/e- beam, to be deflected by up to ±3 nrad, requires that the stripline voltages are ±4.5 kV.

A latency of ~37 ns has been demonstrated for such a system [31].

Kicker Gain

Round-Trip Delay

BPM Processor

Reset

Fig. 17: A possible IP Feedback System for the Future Linear Collider [32]

3.4 Drive Beam Kickers

The Drive Beam requires kickers for:

Extraction from the Combiner Rings (CR). Four kicker systems are needed, one for extracting from each of CR1 and CR2;

Turn-Around Kickers. A Turn-Around kicker extracts drive beam towards each decelerator: a total of 48 Turn-Around kickers are required;

Loop Phase Compensation Kickers: 192 Loop Phase Compensation Kickers are required for the CLIC complex.

3.4.1 Combiner Ring and Turn-Around Kickers

Table 2 shows the specifications for the existing CTF3 extraction kicker as well as for the CLIC CR extraction kickers and the Turn-Around kickers. The most challenging requirement for the CR extraction kickers is the burst-rate of up to 688 kHz for 140 µs, each 20 ms: continuous operation of up to 3 MHz has been demonstrated with kickers, although this has been for either a capacitive load (i.e. not a 50 Ω load) [25-27] or for short duration pulses [28]. The Turn-Around kickers have similar specifications to the CR kickers except that the Turn-Around kickers do not have a burst-rate requirement and their rise/fall time requirements are relaxed. Thus the Turn-Around kickers are not expected to be overly challenging.

Table 2: Drive Beam: CR Extraction and Turn-Around Kicker Specifications

CTF3 CR Extraction

CLIC CR1 Extraction

CLIC CR2 Extraction

Turn-Around

Beam energy 300 2380 2380 2380 MeV

Total kick deflection angle 7 2.5 2.5 2.5 mrad

Stripline plate separation 40 20 20 20 mm

Maximum stripline length 1.7 3 3 3 m

Rise/fall time (0.25% to 99.75%) ≤0.07/≤0.07 ≤0.15/≤0.15 ≤0.15/≤0.15 ~5/20,000 µs

Pulse duration 200 Up to 450 Up to 450 Up to 450 ns

Flat-top reproducibility ±0.1 ±0.1 ±0.1 ±0.1 %

Flat-top stability (including droop) ±0.25 ±0.25 ±0.25 ±0.25 %

Repetition rate

Initial 5 Hz

Nominal 50 50 50 50 Hz

Burst mode none 688 for 140µs

172 for 140µs

None kHz

Average 4800 1200 Hz

Pulse voltage across 50 Ω load 12.6 10 10 10 kV

Pulse current (into 50 Ω load) 252 200 200 200 A

3.4.2 Loop Phase Compensation Kickers

Loop phase compensation kickers will be used to synchronize the phase between the drive beams and the main beam.

Fig. 16: Overview of the CLIC drive beam accelerator [29-30]. Bunch Compression chicanes (BC1 and BC2) are placed in front of each decelerating section, i.e. 24 times per drive beam.

The phase Feed-Forward (FF) system consists of two phase measurements, one in front of Bunch Compression chicane BC1 and one behind it, used to determine phase errors and energy jitter. In addition the FF system includes kickers for phase correction. The FF (loop phase compensation) kicker system must be capable of changing the path length by up to ±100 µm (±375 µrad): very low latency is required (modulation is required to correct the phase over 10 ns to 20 ns). The loop phase compensation kickers are probably an ideal application of the RF amplifier technology developed for the IP kickers [31].

4 Conclusion

There are several very challenging kicker systems required for both the drive beam and main beam of CLIC. The PDR and DR kickers are particularly challenging due to their requirements of excellent field homogeneity, low beam coupling impedance and ultra-low ripple/droop. Striplines with suitable tapers can achieve the required beam transverse and longitudinal coupling impedances except for the low frequency longitudinal impedance. Studies of field homogeneity have commenced: a set of prototype striplines will be prototyped under the Spanish Program “Industry for Science”. Striplines will likely be used for all the CLIC kicker systems.

An inductive adder is being investigated for the pulse generators for the PDRs and DRs. The inductive adder will designed to permit pulse shaping and is expected to be inherently highly reliable and provide good machine protection.

The other challenging main beam kicker system is the IP feedback kicker, which must have very low latency and work in an environment with high background radiation. The technology developed has demonstrated a latency of only 37 ns: the RF amplifier technology developed will also be applied to the drive beam loop phase compensation kickers.

The combiner ring extraction kickers are also demanding because of the requirement for a high burst-rate. The turn-around kickers are similar to those required for the combiner ring extraction kickers except that the turn-around kickers do not have a burst-rate requirement.

References [1] Y. Papaphilippou, “CLIC Damping ring beam transfer systems”, March 10, 2010,

http://indico.cern.ch/getFile.py/access?contribId=2&resId=1&materialId=slides&confId=87738

[2] Y. Papaphilippou, “Parameter Specification, EDMS# 989080, Kickers for the CLIC Damping and Pre-damping Rings”, PBS reference: 1.2._.10.

[3] E. Koukovini-Platia et al., “Impedance of the CLIC-DRs: What we know and what else we need to study….”, CLIC DR Impedance Meeting, Oct. 5, 2010.

[4] G. Rumolo, Private communication.

[5] E. Gaxiola et al., “The Fast Extraction Kicker System in SPS LSS6”, LHC Project Report 913, Geneva (2006).

[6] M.J. Barnes et al., “An Improved Beam Screen for the LHC Injection Kickers”, Proc. PAC'07, Albuquerque, USA, June 25-29, 2007, p. 1574.

[7] D. Alesini, “Fast RF Kicker Design”, ICFA Mini-Workshop on Deflecting/Crabbing Cavity Applications in Accelerators, Shanghai, April 23-25, 2008.

[8] A. Chao and M. Tigner (Editors), “Handbook of Accelerator Physics and Engineering”, 1998.

[9] S. Smith, Private communications.

[10] D. Alesini, S. Guiducci, F. Marcellini, P. Raimondi, “Design and Tests of New Fast Kickers for the DAPHNE Collider and the ILC Damping Rings”, Proc. EPAC’06, Edinburgh, Scotland, June 26-30, 2006, pp1502-1504.

[11] D. Alesini, S. Guiducci, F. Marcellini, P. Raimondi, “Fast Injection Kickers for DAΦNE and ILC Damping Rings”, DAΦNE Technical Note, INFN - LNF, Accelerator Division, Frascati, June 6, 2006, Note: I-17.

[12] T. Imai et al, “Double Kicker system in ATF”, XX International Linac Conference, Monterey, California, LINAC2000, pp77-79.

[13] I. Rodríguez et al., “Design, Manufacturing and Testing of the CTF3 Tail Clipper Kicker”, proc. of IPAC 10, Kyoto, Japan, pp3299-3301.

[14] I. Rodríguez, “Calculation methodology and fabrication procedures for particle accelerator stripline kickers...”, PhD Thesis, Universidad Politécnica de Madrid, 2009.

[15] T. Imai, “Highly Stable Beam Extraction by Double Kicker System”, KEK Preprint 2002-16, KEK Accession No. 200227016, May 2002.

[16] T. Imai, “Development of Double Kicker System for Stable Beam Extraction”, Dept. of Physics, Tokyo University of Science, Sept. 2002.

[17] N. Terunuma, “Instrumentation for the ATF2 Facility”, proc. of IPAC 10, Kyoto, Japan, pp2397-2401.

[18] J.A. Watson et al., “A Solid-State Modulator for High Speed Kickers”, Proc. PAC’01, Chicago, USA, June 18-22 2001, pp3738-3740.

[19] E.G. Cook, et al, “Design and Testing of a Fast, 50kV Solid-State Pulser”, 2002 International Power Modulator Conference, Hollywood, USA, June 30-July 3, 2002.

[20] E.G. Cook, et al, “Solid-State Modulator R&D at LLNL”, RPIA Conference, Tsukuba, Japan. October 29-31, 2002.

[21] R.L. Cassel, et al, “NLC Hybrid Solid State Induction Modulator”, LINAC 2004, XXII International Linear Accelerator Conference, August 6-20, 2004.

[22] E. Cook et al., “Solid-State Modulators for RF and Fast Kickers", Proc. PAC’05, Knoxville, USA, May 16-20 2005.

[23] C. Burkhart, “Inductive Adder Modulators for ILC DR Kickers”, Cornell ILC DR Workshop, September 26, 2006.

[24] Gower et al., “Analog Amplitude Modulation of a High Voltage Solid State Inductive Adder Pulser Generator Using MOSFETs”, Power Modulator Symposium, 2002 and 2002 High-Voltage Workshop. Conference Record of the Twenty-Fifth International, 30 June-3 July 2002, pp321-324.

[25] M.J. Barnes et al., “A High Frequency MOSFET Driver for the TITAN Facility at TRIUMF”, proc. of 15th IEEE International Pulsed Power Conf., June 13-17, 2005, Monterey, USA, pp178-181.

[26] M.J. Barnes, G.D. Wait, “Design for a FET Based 1 MHz, 10 kV Pulse Generator”, proc. of 10th International Pulsed Power Conf., Albuquerque, USA, July 10-13, 1995, pp1335-1340.

[27] G.D. Wait, M.J. Barnes, “Measurements on a FET Based 1 MHz, 10 kV Pulse Generator”, proc. of 10th International Pulsed Power Conf., Albuquerque, USA, July 10-13, 1995, pp1341-1346.

[28] T. Naito et al., “Multi-Bunch Beam Extraction Using Strip-Line Kicker at KEK-ATF”, IPAC 10, Kyoto, Japan, pp2386-2388.

[29] F. Stulle, A. Adelmann, M. Pedrozzi, “Conceptual Design of Bunch Compressors and Turn Around Loops for a Multi-TeV Linear Collider, Report on the Deliverables of the BCDES Task within EUROTeV WP6”, EUROTeV-Report-2008-086.

[30] F. Stulle, A. Adelmann, M. Pedrozzi, “Conceptual Design of Bunch Compressors and Turn Around Loops for a Multi-TeV Linear Collider, Final Report on PSI's Activities within the EUROTeV Collaboration”, EUROTeV-Report-2008-025.

[31] J. Resta-Lopez, P.N. Burrows, “The Beam-Based Intra-Train Feedback System of CLIC”, proc. of 1st International Particle Accelerator Conference (IPAC’10), Kyoto, Japan, pp2791-2793.

[32] P.N. Burrows, “Feedback On Nanosecond Timescales (Font): Results from First Beam Tests at the NLCTA at SLAC”, http://www-project.slac.stanford.edu/lc/local/nanobpm/documents/ROPC011.pdf