IS IT POSSIBLE TO INCREASE THE p INTENSITY FOR CNGS BY A FACTOR 2 OR 3 IS IT POSSIBLE TO INCREASE THE p INTENSITY FOR CNGS BY A FACTOR 2 OR 3 ? R. CAPPI / SL Seminar, 21.03.2002 Introduction Main limitations (some of) acceptances & emittances space-charge double batch injection bunch flattening 5 turn Continuous Transfer new 5t CT List of various schemes Conclusion
speculations => studies & experiments Introduction The talk is a ‘simplified’ summary of the paper: CERN/PS 2001-041 (AE) or CERN/SL 2001-032 speculations => studies & experiments all results are PRELIMINARY and generally OPTIMISTIC the talk will be mainly devoted to PSB-PS issues I will not talk about collective effects ( except sp. ch.), longit. beam dynamics issues , transition crossing, etc.
Introduction: basic limitations NB: The present scheme is “consistent” i.e. LINAC, PSB, PS and SPS are all close to their limits, i.e. there is not a single weak point Linac2 Close to its max Ip PSB Space charge ~limited Ek,max limited (1.4GeV) PS Acceptance ~limited 5t Continous Transfer …. SPS Acceptance limited Common: T & L collective effects, losses, transition, PRF , etc. recent results
Acceptance & emittance issues PS acceptance: Ax=60mm, Ay=20mm ex2 < 22mm, ey2 < 9mm LHC ~ 5 5 Ex2 Experiments Ax limit Ey2 Ay limit Courtesy of R.Steerenberg
Present scenario & associated problems 50 MeV PSB PS SPS Nt = 3.3 Limit ex< 22 ey< 9 1.4 GeV, h<0.9 DQ x,y~ 0.13 , 0.23 ex= 25 ey= 12 Nt = 3 14 GeV/c; 5t CT ; h=0.8 NB: in all transparencies: ex= 4sx2/bx in mm intensities Nt are in 10^13 p 3) h is the transfer efficiency 4) yp is the p flux on target in 10^13p/s X Limit ex= 4.2/3 = 1.4 ey= 2.5 ex< 3 ey< 2 Nt = 4.8 G.Arduini filling time = 1.2s yp = 4.8/6 = 0.8 G = 1
Space charge (at low energy in the PS) Self field tune shift: In the PS, to be safe : If : T=1.4 GeV, ex = 22mm, ey = 9mm Nt < 4.8 E13 p/p (Kb=8) to reach it WE NEED A DOUBLE BATCH INJECTION NB: the SPS filling time will increase by 1.2 s (or 0.6 s if PSB can pulse 2x faster* ) PS LIMIT *) M.Benedict et al. , undergoing study
Double batch injection into PS: forecast 50 MeV Limit PSB PS SPS Nt = 2 x 2.4 ex< 22 ey< 9 1.4 GeV; h=1 ex= 21 ey= 9.2 DQ x,y~ 0.21 ; 0.35 Nt = 4.8 => Intensity limit for a PS @ 1.4 GeV 14 GeV/c; old 5t CT; h=0.8 Limit X ex= 3.4/3 = 1.13 ey= 1.4 ex< 3 ey< 2 Nt = 7.7 yp = 7.7/7.2 = 1.07 G = 1.34 yp = 7.7/6.6 = 1.17 if PSB@.6s, G = 1.46
Recent results of high intensity double batch injection studies Experiments PS transformer Beam intensity ( E10 p/p) 1st batch 2nd batch Time (ms) Courtesy E. Metral
Comparing with LHC “ultimate beam” DQ = 0.20, 0.26 PS transformer Beam intensity ( E10 p/p) Time (ms) Courtesy G.Metral,E. Metral
Can we improve space charge limits? Increase injection energy (e.g. with SPL) Reduce Ip by ‘bunch flattening’ techniques: (gain <1.5) time
A new bunch flattening technique (*) (*) C.Carli /CERN-PS-2001-073-AE and EPAC2002
Bunch flattening in PSB: recent results Final bunch Initial bunch Experiments DQ reduction of ~28% Courtesy C.Carli
5 turn Continuous Transfer It is the way the PS uses to fill the SPS (at 14 GeV/c) CSPS = 11 x CPS PS PS SPS Present system: + it works - it is lossy (~20%) x’ 2 Qx = 6.25 3 1 5 x Extracted beam 4 . TT2 transfo 1 2 3 4 5 ES blade time, 2ms / div
Proposal for a new 5t CT (*) The principle: the beam is adiabatically captured into 4 islands of a 4th order resonance properly adjusted with sextupoles and octupoles, Initial state Simulation results Final state Simulation results ES 2) then the beam is extracted similarly to the present scheme. (*) M.Giovannozzi, R.Cappi ; Phys. Rev. Lett., V.88, i.10
n 5t CT: pro / con + it should be less lossy (~5%) + the five beamlets will match the phase space topology better => less betatron mismatch at injection in the SPS=> lower transv. emittance beam to SPS => lower losses => higher intensity - it has to be tested experimentally
n5tCT: (x, x’ ) topology qx Courtesy M.Giovannozzi time ~ 30 ms
n5tCT: x-x’ measurement results Courtesy M.E.Angoletta, A-S.Muller, M.Martini,…)
MAD simulations Courtesy A-S.Muller
MAD simulations (suite) Courtesy A-S.Muller
Expected results from: double batch+ n5tCT 50 MeV PSB PS SPS Nt = 2 x 2.4 ex< 22 ey< 9 1.4 GeV, h=0.9 ex= 21 ey= 9.2 Nt = 4.8 14 GeV/c; new5t CT; h=0.9 ex= 3.4/5 = 0.68 ey= 1.4 ex< 3 ey< 2 RMKS: 10% improvement => h=0.9 =>lower transfer losses, better matching, etc. Nt = 8.6 filling time = 2.4s yp = 8.6/7.2 = 1.19 G = 1.49 yp = 8.6/6.6 = 1.30 if PSB@.6s G = 1.63
What about the SPS ? Single bunch coll. effects: Beam loading: 8.6E13ppp => 2 E10 p/b [LHC~10 E10; e-cloud > 4 E10 (5ns?)] Transverse impedance strongly reduced since 2002 => ~OK Beam loading: 8.6E13ppp => 0.4 E13/ms [ LHC~0.5 E13p/ms] ~OK better if p=26GeV/c Transv. & long. Feedbacks HW modifications? 20=>100 MHz? octupoles :YES (some e x,y b.u. accepted) ~OK ? Transition: now 5% losses, Etc. K.Cornelis, T.Linnecar, E.Schaposnikova,…
The various schemes
we need a.s.a.p. clear priorities to continue at efficient speed. Conclusion first studies show encouraging results not only for CNGS but for LHC itself and for cleaning up the machines by improving reliability a gain in p flux of ~1.5 seems feasible though difficult (cost ~0-2MCHF) a gain of ~2 is maybe possible but will be more expensive (~50MCHF) a gain of 3 will be VERY expensive ( ~300MCHF) and probably technically unrealistic we need a.s.a.p. clear priorities to continue at efficient speed.