Mul$- bunch accelera$on in FFAG. Takeichiro Yokoi(JAI)

Transcription

1 Mul$- bunch accelera$on in FFAG Takeichiro Yokoi(JAI)

2 Introduc$on For high intensity applica9on such as ADSR, high repe99on opera9on is a requirement to diminish the influence of space charge force For the realiza9on of high repe99on rate opera9on in circular machine, rf is a real obstacle è an approach : high Q ferrite cavity (ex PAMELA ferrite rf cavity : Q~60) The nature of V 2 causes surge of cooling power, running cost and ini9al cost P V 2 V 2 R sh µqf Efficient and reliable rf rf accelera9on system is is the key for the realiza9on of of high intensity proton machine

3 Mul$- bunch accelera$on : what is it? Beams of different energies can be circulated simultaneously in a fixed field accelerator. In broad band rf cavity such as MA cavity ( Q~1 ), rf pattern for multiple bunches can be generated with single cavity. RF buckets must be separated (2 bucket height) one another With PAMELA proton ring (E:30~250MeV, f rev :2~6MHz), more than 20 bunches can be accelerated in parallel. The scheme has advantages of both MA cavity and Ferrite cavity δ 2δ B MA : No bias current, robust against beam loading Ferrite : relatively low rf loss Δf δ B = 2ν s hη Y (φ s) ~ 2ν s h Φ Any fixed field lattice with stable betatron tune can employ the scheme It also improve the injector efficiency (ε = T inj /T bunch )

4 Mul$- bunch accelera$on : average power Single bunch accelera9on Energy Mul9 bunch accelera9on Energy time time Low Q cavity (ex MA) can mix wide range of frequencies (ΣV ) 2 P = R dt (ΣV ) 2 (ΣV i sin[ f i (t)]) 2 = Σ i (V isin[ f i (t)]) 2 + Σ i j (V isin[ f i (t)] V j sin[ f j (t)]) 1 T dt " 0 Average power wise. P N rep2 /N bunch P N rep2 /N bunch Mul$- bunch accelera$on is preferable from the viewpoint of efficiency and upgradeability

5 Mul$- bunch accelera$on : peak power 1bunch The voltage distribution of the multibunch acceleration is completely different from that of single bunch acceleration Through instinct, available rf voltage per bunch is V max /n bunch As the number of wave superposition increases, the high amplitude fraction gets reduced drastically. Ex. for 15bunch case, 99% of bunches, the peak rf voltage is below 0.5(nbunch Vrf) In result, the peak voltage can also be reduced considerably in multi-bunch acceleration(~60% reduction in 15bunch case. è peak power wise, ~85% reduction) 15bunch 1bunch 2bunch 5bunch 10bunch 15bunch Cumula9ve frac9on of normalized maximum voltage in rf bucket

6 Past experiment Mul9- bunch accelera9on has already been demonstrated at KEK PoP FFAG f 4 f sy (PAC 01 proceedings p.588, Koba et al.) In the experiment, two bunches were injected/accelerated, and revolu9on frequencies of two bunches were measured as the indica9on of two bunch accelera9on In the experiment, only two bunch accelera9on was carried out and no other measurements such as beam loss was done

7 KURRI booster FFAG For the beam simula9on, KURRI booster FFAG was employed It is the booster ring of KURRI FFAG complex Now, it is separated from main ring (change of injec9on scheme) Energy(MeV) Laece 2.5/20(inj/ext) Scaling( Radial sector) R(m) 1.27/1.74 F rev (MHz) 2.6/5.3 N cell 8 Field index 2.5

8 Beam accelera$on (single bunch) ΔE(keV) Rf paoern (KURRI booster FFAG:2.5 è 10MeV) F rf (MHz) Φ(radian) δ inv T(msec) To beam understand the dynamics during beam accelera9on, ordinary longitudinal phase space (ΔE,Φ) is not appropriate ç adiaba9c dumping, and change of η,ν s θ inv

9 Mul$- bunch accelera$on (number of bunch dependence) Fixing the accelera9on 9me per bunch, mul9- bunch accelera9on was simulated in various number of bunch (up to 15bunch, change bunch separa9on) Up to 15bunch, no significant beam loss was observed Basically, the beam mo9on follows synchrotron mo9on. As the number of bunch increases ( bunch separa9on 9me decreases), the perturba9on gets more significant δ inv δ inv δ inv 1 bunch θ inv 6 bunch (bunch separa9on: 250 μsec) θ inv 15 bunch (bunch separa9on: 100 μsec) θ inv

10 Mul$- bunch accelera$on(single par$cle mo$on) Single par9cle mo9on is easier to understand the dynamics with rf perturba9on. ( fix : the bunch separa9on and vary: the number of bunch.) The perturba9on is linear and adjacent 2 bunches mainly contribute the perturba9on. Dominant factor is the bunch separa9on. Thus, as long as the bunch separa9on is fixed, the dynamics is basically similar to the case of 3 bunch accelera9on 1 bunch 2 bunch (reference + adjacent former bunch) bs:125μsec 3bunch(reference + adjacent 2bunches) Energy 15bunch t bunch time

11 Bunch separa$on dependence Longitudinal phase space plot suggests the bunch separa9on is the dominant factor of perturba9on. Radial distribu9on of quasi- isochronous par9cle in invariant phase space was adopted as a measure. The results supports that the adjacent bunches are the dominant source of perturba9on Basically, the dynamics of mul9- bunch accelera9on can be explained within the context of synchrotron theory. T bunch : 70µsec Energy t bunch time T bunch : 250µsec

12 Satura$on effect The voltage distribu9on suggests small rf satura9on might be tolerable. By clipping the rf voltage, output satura9on was introduced in the simula9on Up to ~5% of rf error, beam can accelerated to the end without significant beam loss The beam loss/blow- up mechanism is not clear at the moment. Due to the nature of V 2, it dras9cally influence the cost of rf system. 1bunch 2bunch 5bunch 10bunch 15bunch Typical beam blow- up

13 How efficient with an exis$ng rf system? In the case of 10 bunch acceleration, peak rf voltage ~0.5 n bunch V bunch è it is equivalent to double the available rf voltage. Thus, if the peak voltage is the limiting factor, the acceleration rate of a bunch is 1/5 of single bunch acceleration in 10 bunch acceleration In result, the intensity is doubled (1/5 10 = 2 ) On the other hand, the rf power is 2/5 compared to single bunch acceleration. (1/ ) Thus, rf power wise, it is as expected factor of 10 more efficient [2 2 (2/5)] 15bunch 10bunch 5bunch 1bunch 2bunch Cumula9ve frac9on of normalized maximum voltage in rf bucket

14 RF response in MA cavity rf superposition in MA cavity has already been examined using PRISM RF cavity in 2008 The linearity of superposition was found quite good for 2-frequency case 1. Frequency range : 2~6MHz 2. Δf : 60kHz~200kHz (tested up to 20kHz ) 3. Saturated condition Probably it is a good support for the hardware feasibility ç For the multi-bunch acceleration, the closest rf bunch is the major perturbation source But, beam experiment is required to demonstrated the feasibility

15 Beam Experiment at KURRI At the moment, beam experiment using KURRI FFAG was proposed. Test machine is Booster FFAG. In the experiment, the single bunch motion with multi-bunch rf pattern is examined è No need for multiple bunch injection (simpler monitor response, no worry of beam loading, space charge) Beam loss/longitudinal motion are the measured items Due to the limitation of available rf voltage, the experiment will be carried out in the following steps 1. Low acceleration voltage, single bunch 2. Multi-bunch acceleration 3. Saturation effect Energy t bunch time

16 Remaining issues Beam loading ç In multi-bunch acceleration, beam loading compensation might be complicated/difficult (How to feed-back) Beam dynamics of peripheral region. Realistic rf pattern (HOM, amplitude dependence etc.) At the moment, ideal superposition was assumed Saturation rf pattern modeling (At the moment, just clipping) Loss mechanism in saturation operation

17 Toward 1~10mA opera$on For >1mA machine, repetition rate over 1kHz is required for beam acceleration under space charge forces. Assuming 10-bunch parallel acceleration and 1kHz beam cycle, target acceleration rate/bunch is 100Hz. (for PAMELA proton ring, about ~8kV/ turn :30MeV->250MeV) KEK 150MeV FFAG has already realized 100Hz DC operation in similar energy range (rf core loss:90kw) In >1mA operation, ferrite cavity and MA (multi-bunch) has no significant difference (beam power core loss) However, is a proton accelerator with full intensity really required for ADSR? Rf power for 30- >250MeV proton machine (1kHz opera9on) Reference ring Ferrite PAMELA proton ring MA: 1bunch KEK 150MeV FFAG N cavity Q R sh (Ω) N bunch Vmax/ cavity/bunch rf power/ cavity(kw)* Total rf power(kw) * Core loss = rf power (unload rf power) MA: 10bunch KEK 150MeV FFAG

18 On reliability/redundancy The reliability is said to be the key of ADSR, but what does the word actually mean in terms of accelerator system? For real system design, at least 4 parameters need to be specified 1 Tolerable instantaneous power drop è ring redundancy 2 Acceptable operational power drop è operational margin (injector intensity etc.), component redundancy 3 Transient time to temporary operation è component control 4 Full recovery time [beam power] (5 Full recovery time [hardware]) è maintenance scenario power 1 Failure Ex. RF system-wise, pulsed machine might be more robust than dc machine(cyclotron) against failure. ç A pulsed machine with multiple rf cavity in principle might be able to be operated with lower repetition rate even in a failure of cavity. (ex 3-ring, 8rf system case->1 :33%, 2:4% ) Full $me recovery

19 Summary Mul9- bunch accelera9on is an efficient op9on for high intensity pulsed accelerator. It has the advantage of both MA cavity and ferrite cavity Any fixed field laece with stable tune can employ the scheme The beam dynamics can be basically understood with conven9onal synchrotron mo9on theory Satura9on opera9on needs further study Beam experiment with KURRI FFAG is being planned For a realis9c machine design for ADSR, failure/maintenance scenario must be prepared as soon as possible and be included in the very early design stage.

20 PAMELA: overview PAMELA (Par$cle Accelerator for MEdicaL Applica$on) aims to design a par$cle therapy facility using NS- FFAG It aims to provide spot scanning with proton and carbon beam. 2 cascaded rings (For proton, 1st ring is used. For carbon, 1st ring is used as a booster) Flexible change of beam energy and par$cle is required Extracted beam : pulsed beam high repe$$on rate is required for ac$ve beam scanning For high repe$$on opera$on, rf system is the key element Particle p,c Ext Energy:p (MeV) 70~250 (variable) Ext Energy:C(MeV/u) 110~450 (variable) Repetition rate(khz) 0.5~1 Voxel size (mm) 4 4~10 10 Active beam scanning Spot scanning Switching time:p c(s) <1 # of ring 2 (*2nd ring :for C)