ILC Damping Ring Kickers Presenter: Josef Frisch Dec 7, 2004.

ILC Damping Ring Kickers

Presenter: Josef Frisch

Dec 7, 2004

Basic kicker types

• Conventional pulser (Remainder of this presentation).– High voltage pulse driving (probably) strip line kicker– Simple, minimal impedance problems (screen electrodes)– May require exotic pulser.

• RF kicker– Pulsed RF source driving low Q deflection structure. – Makes use of available high power, broadband RF sources– Possible impedance issues

• Resonant deflection system– Uses multiple resonant cavities driven with a set of frequencies

to select bunches– Multiple designs – too varied to discuss here– Quasi-CW RF eliminates ringing, provides good stability– Impedance, RF kicks within bunches, etc need to be understood.

Approximate requirements for pulsed kicker

• Deflection angle 0.6 mrad (0.01 T-M) for TESLA ring design

• If we allow 10 Meter total kicker length– Need 50 Amp kicker drive

• Length of each kicker << bunch spacing (assuming speed of light kicker).– For 3 nanosecond, need ~20 Kickers

• Stripline kicker impedance probably ~100 Ohms (assuming speed of light kicker)

• Pulsers: 5Kv, 50Amp, 20 units, 2nanosecond rise and fall time .– Just for scale: Actual specifications depend on detailed ring

design• Un-kicked bunches must not be disturbed by more than

7x10-4 of kicked bunch.

Basic Extraction Scheme

Comments on Basic Scheme

• Shortest ring for given current– No unused bunches, or gaps (except ion clearing).

• Tight requirements on kicker stability and fall time– Need to not disturb neighboring bunches by more

than 7x10-4 of kicked bunch

• Falling edge of a pulse typically more difficult to control than rising edge. – Ringing from impedance mismatches, stray

inductance etc.

Buffer pulse scheme

Comments on buffer pulse scheme

• “Buffer” pulses are not needed for luminosity– Can probably kick by 10-2 of main kick– Allows longer settling time to 7x10-4.

• Need to replace buffer pulses– May be tricky with positrons if bunches are generated

by main electron beam: Might need to waste a machine cycle.

• Slightly longer ring for same average current.

Kicker Gap Scheme

Comments on Kicker Gap Scheme

• Gap allows extraction kicker with fast rise, but unrestricted settling for next pulse, settle to 7x10-4 after gap

• Injection kicker (larger pulse) only needs 1% interference with preceding bunch, and 1% after gap.

• Scheme does not work if positrons generated by luminosity generating electron beam– Works if you have a pre-damping ring.

• Scheme requires that ring empty, then re-fill in 2 milliseconds – might cause ring stability problems.

Kicker Gap Extraction & Injection

Injection and extraction with fast rise slow fall if DR size is not determined by kicker rise time

Comments on Kicker Gap Scheme

• Average ring current constant

• Allows use of slow fall time kicker

• Requires 2X ring length for same bunch spacing.

• Beam current harmonic content changes– DR physics question

Kicker Driver Requirements

• 20 Units• ~5KV, 50Amps

– Depends on damping ring design, kicker length, etc– 250KW peak power– 2.5KW, average power, 1 millisecond– 25 Watt long term average

• Few nanosecond rise and fall, with settling to <7x10-4 for preceding and following pulses.

Snap / Step Diodes

• <50 picoseconds to 20V. • sub-nanosecond to 300V, 6 Amps.

(1800W peak). • High power devices from Institute of

Electrophysics– 600 picosecond to 1000 Amps (?? Voltage)– 5 nanosecond to 400KV. – Repetition rates to few KHz.

• Power dissipation probably limits rep rate.

Step recovery diode pulses

Very fast high voltage pulsesRepetition rate limited to ~KHz(for these devices).

Institute of Electrophysics

Avalanche Transistors• Avalanche Transistors

– <200 picoseconds to 200V, ~50 Amps (10KW)

– Arrays (tapered transmission lines) demonstrated to 40KV, 800A, 200ps. (Kentech)

– Recovery time too long except in liquid nitrogen (50nsec reported)

– Average power limited to ~1W / device.

– Combining may lead to ringing.

• Low Repetition Rate

40 KV in 200ps rise time (Kentech)

MOSFETs• Individual devices to ~1Kv, ~50Amps, ~3ns rise and fall

times. • Variety of combining schemes to high power, medium

fast rise. – DARHT-2 kicker: 20KV, 10ns rise / fall, 1.6MHz burst ( 4 pulses)– Belkhe / TESLA: 7.5Kv, 72A, 5.3ns, 1MHz (200 pulses). (fall

time slower)– Belkhe – datasheet 3Kv, 80A, 2ns, 1 MHz (MAX) burst. (10

pulses). – Kentech: 10Kv, 2ns – could operate at high rate.

• Relatively low impedance – (~10 Ohms), stray inductance ringing can be a problem– Gate drive is very low impedance <<1 Ohm.

• Probably OK for ~10 nsec rise / fall times. (maybe faster)• May be used as driver for additional stage / compressor

MOSFET Pulses

4MHz pulses, but with 18ns risetime

More MOSFET pulses (Kentech)

5 channels

Lower voltage2.5MHz pulser

MOSFET pulser comments

• Very likely to use MOSFET technology in pulser – maybe with shock line for compression.

• Some designs (Belkhe) are very fast but have limited repetition rate. Problem is not thermal – but design in proprietary.

Shock Lines - Ferromagnetic• Nonlinear transmission line: Wave velocity increases with pulse voltage• “Sharpens” front end of pulse• Ferromagnetic (most common): (used for SLAC Kicker (Cassel)• 95KV, 380 Picoseconds rise (Seddon et al, 1987), Ferroxcube B2 ferrite

Shock Lines – Ferromagnetic (SLAC)

Ferromagnetic Shock Lines, Falling edge

Shock Lines - others

• Ferroelectric– 20KV, 400 Picosecond (Oxford – Web report)

• Diode loaded line– Monolithic (Allen, 1994 thesis), 4V at <700

Femtoseconds! (expect <170 fsec in future)

• Vacuum Magnetron line: – At high voltages, magnetic field insulates line– Probably only applicable at higher power than

we require.

Shock Lines - comments

• Most work in shock lines has been to obtain very fast, very high power pulses – well beyond our requirements

• Typically operated at low repetition rates.• Need to eliminate (typical) slow tail from release of

energy stored in non-linear material. • For ILC high repetition rate may lead to heating

problems (need low loss nonlinear material). • Ferromagnetic and ferroelectric materials tend to also

have magnetostrictive / piezoelectric effect– For millisecond pulse burst could lead to stability problems.

• Many non-linear materials have strong temperature sensitivity – may lead to stability problems.

Hard Tube Switches

• Pulser based on Eimac Y-690 tube used for Pockels Cell drive at SLAC (M. Browne, D. Brown). – 6KV, 30 Amps, <1.5ns rise time.– Driven by avalanche transistors – not appropriate for

high repetition rate– Would need to parallel 2 tubes for ILC kicker (easy)

• Nonlinearity of tubes helps with settling time. • Average power “not unreasonable” but would

need to check. (grid dissipation)

Hard Tube Pulser

Note, tail on pulse believed to be due to output Transformer (not needed for ILC kicker)

Custom Tube Pulser

Custom tube Comments

• Single beam switched between multiple (~20) Anodes.

• Tube parameters comparable to other big power tubes (klystrons).

• Something of this sort would very likely work, but would require a large development effort.

• Only consider if conventional pulsers will not work.

Kicker Magnet

• Probably need speed of light kicker• Kicker fill time + pulse rise time -> effective rise

time– Need short (< 1 Meter) kickers.

• Need to avoid reflections / ringing– Must be designed as a RF component– Full E+M simulation / optimization

• Possibly shield beamline with thin screen (to block beam wakefields (~10GHz), but transmit kicker fields (~300 MHz).

• Want optimized design to minimize kicker power

Stability / Settling time issues

• Multiplicity of kickers helps with random noise

• Reproducible and small settling time problems can be fixed with additional kicker driven by AWG and power amplifier.

• Probably want feed forward from beam position / angle out of ring to kicker in main beam line– Assumes turn-around after damping ring

Correction Scheme

Correction Scheme - variant

Correction scheme / layout issues.

• The 7x10-4 stability specification is Heroic!– Difficult to measure without a beam line

(ATF?)

• The kicker driver will likely have pulse – pulse feedback to flatten the waveforms

• Would like a beamline arrangement which allows feed-forward from output beam

Ongoing Work

• ATF Damping Ring in Japan: Proposal to build a single pulse extraction system– Good test bed for kickers, and stabilization– Provide ILC – like test beam (~200 bunches)– Requires higher Kicker drive power than ILC

• Working on kicker / optics design to reduce

• DHART FET pulser -> shock line– Test high repetition rate shock lines

• DESY working on paralleling Belkhe pulsers to increase repetition rate.

• SLAC to obtain Belkhe pulser for testing shock lines.

Overall Comments

• Can probably build a kicker to meet any likely damping ring requirements, for a small fraction of the damping ring cost– Optimize the ring design, see what is needed.

• Best guess: MOSFETs driving Ferromagnetic shock line. – Hard tubes an option.

• Custom tube can probably solve the problem, but expensive to develop – leave as a backup plan.

• Want technology demonstration prototypes soon, to allow selection of technology, and system development.

• Kicker parameters (voltage, current, etc) depend on details of ring design.