ATF Fast Kicker R&D at LBNL ILCDR06, Cornell University Stefano De Santis Center for Beam Physics ILCDR06, Cornell University September 26th, 2006 S. De Santis ILCDR06 Sept. 26-28, 2006
Summary Motivation of the project and its main requirements Calculation of kicker’s main parameters Computer simulations of stripline electrodes Conclusions S. De Santis ILCDR06 Sept. 26-28, 2006
The ATF as test-bench for ILC DR technologies The challenge: Developing a pulser capable of generating elevated voltages over few nanoseconds, with short rise and fall times and relatively high repetition rate. Developing a kicker structure capable to efficiently transform the voltage pulse into a deflecting field, maintaining the pulse’s time structure and without introducing undesired beam impedance. S. De Santis ILCDR06 Sept. 26-28, 2006
Fast kicker’s main specifications Deflection angle k: 5 mrad Bunch spacing tb: 5.6/2.8 ns Pulse repetition rate frep: 3 MHz Total kicker length LT: 1.4 + 0.8 m Kicker field falls to <0.07% of max value before next bunch S. De Santis ILCDR06 Sept. 26-28, 2006
Pulser and bunch separation V tr td tft=2Lk/c bunch 0 bunch 1 bunch 2 tk tb -tk tk< [tb- max(tr, td)]/2 1.16 GHz 3 MHz 5.6 ns/3 MHz pulse train spectrum FID pulsers: - V up to 10’s of kV - tr down to 50 ps (100-200 typ.) - tft up to 10’s of ns - frep up to 100’s of kHz … and kicker’s electrodes can only make things worse S. De Santis ILCDR06 Sept. 26-28, 2006
Calculation of stripline kicker’s main parameters (tb=5.6 ns) Max. stripline length: ~84 cm Lk= 65 cm (allows 2 modules in the 1.4 m sector) Trying to minimize the impedance, we choose the plates separation h=24 mm ( same as beam pipe diameter), with a traditional angle of 120° (coverage factor g=0.93) The required pulser voltage is Vk/2 ≈ 21 kV MHz kΩ Rs ≈ 250 kΩ and the shunt impedance The peak power is and the average power at 3 MHz is Pavg ≈ 100 W S. De Santis ILCDR06 Sept. 26-28, 2006
2D Computer Simulation Determination of the beam pipe radius required to obtain a 50 characteristic impedance for the striplines. We get an outer pipe radius of 22 mm, with a 120º coverage angle and a 0.5 mm thickness for the striplines. even mode odd mode S. De Santis ILCDR06 Sept. 26-28, 2006
3D electromagnetic modelling Detail of the mesh (with coax for impedance measurements) Kicker module with feedthroughs and tapers as seen by Microwave Studio™ S. De Santis ILCDR06 Sept. 26-28, 2006
S-parameters and stored energy trailing bunch enters module Excitation: tr=150 ps, tft=5.45 ns, td=300 ps S. De Santis ILCDR06 Sept. 26-28, 2006
Deflecting field along bunch orbit Since the kicker is substantially shorter than half the bunch separation, it is possible to prefire the pulser, at the price of a slightly higher power dissipation, to ensure deflecting field uniformity along the bunch path. bunch enters kicker (avg. 695) … at midlength … leaves kicker S. De Santis ILCDR06 Sept. 26-28, 2006
Transverse field uniformity variation < 1% over 5 mm S. De Santis ILCDR06 Sept. 26-28, 2006
Field decay Average and rms values are of the order of 5 10-3. Anyway we have to consider that this is a free oscillating field and things will get better - Attenuation from ohmic losses - Better feedthroughs - Effects of different modules don’t add trailing bunch at kicker’s midlength S. De Santis ILCDR06 Sept. 26-28, 2006
Longitudinal impedance and loss factor We can simulate the classic coaxial wire measurement (Walling’s formula for distributed impedances ) Loss factor of the order of 0.1 V/pC S. De Santis ILCDR06 Sept. 26-28, 2006
Transfer impedance We can simulate this measurement as well in order to evaluate the field levels on the feedthroughs, loads, etc induced by the circulating beam. downstream ... and upstream port S. De Santis ILCDR06 Sept. 26-28, 2006
Shorter kicker electrodes 650 -> 350 mm residual field If power dissipation is not a concern and/or more powerful pulsers are available this could be a viable method for reducing perturbations on the trailing bunch S. De Santis ILCDR06 Sept. 26-28, 2006
Conclusions For 5.6 ns bunch spacing the specifications seem to be attainable. tb = 2.8 ns is much harder, due to (non) commercial availability of high voltage pulsers with appropriate rise and fall times. The frep = 3 MHz is likely the hardest requirement on the pulsers in the 5.6 ns scenario (FID say they are being developed). Shorter modules, if power figures allow, are beneficial, yet more expensive. The effects of modules in different locations on the ring need to be investigated, as well as computer simulations at higher frequencies. S. De Santis ILCDR06 Sept. 26-28, 2006