Joint CLIC Beam Dynamics and LAT Section Meeting
Joint (CLIC) Beam Dynamics and LAT Section Meeting CLIC 380 Ge. V: Optimisation of the 2 nd Bunch Compressor of the Main Beam Base Alignment Correction for CLIC-RTML Xingguang Liu, Chetan Gohil and Andrea Latina xingguang. liu@cern. ch May 7, 2020
Content • CLIC 380 and CLIC-RTML-BC 2 • RTML-BC 2 -RF Lattice Update • RTML BBA tuning process and results • Conclusion 2
CLIC 380 Layout * P 2, CLIC-PIP, 2019 3
CLIC RTML Layout * P 8, CLIC-PIP, 2019 BC 2 lattice RTML End RTML Start 4
Match Section Units BPM Zero length dipole/corrector Q Chicane x 11 RF sections BPM Q Structure(s) CLIC RTML Bunch Compressor 2 (BC 2) 200 Match – 210 220 230 240 250 260 270 280 BC 2 RF – match – BC 2 Chicane 1 – match – BC 2 Chicane 2 – Match Section – Diagnostic – Match to ML *200 -280: Section ID in the scripts 5
Motivation- Introduction Table 1: Normalized emittance budgets in CLIC RTML initial Design with Static Imperfections With Dynamic Imperfections 700 <820 <850 5 <6 <8 <10 * P 10, CLIC-PIP, 2019, also see: Y. Han, etc. Journal of Instrumentation, vol. 12, no. 06, P 06010, 2017 Everything seems perfect except that the aperture of RF cavities for BC 2 RF is assumed to be 3. 6*1. 5=5. 4 mm 6
Why Large Aperture? Wt_mean* [V/mm/m] a/a 0 Wt/Wt_0 CLIC-KB 2. 94 89955. 64 1 1 3. 5 56733. 04 1. 19 0. 63 BBA fail 5. 4374 15899. 18 1. 8495 0. 17 BBA pass a[mm] Power Loaded [MW] Power Loaded in terms of <a> Averaged wake in terms of <a> 120 100 80 60 40 200 MW/structure ! x 2. 5 What do we need? To pass BBA test with reasonable RF power consumption. 3. 5 4. 5 <a>[mm] n_bunches 352 n_particles 5. 2 e 9 bunch_sepration 0. 5/1 e 9 G [MV/m] 70 a 1 0. 145*ak a 2 0. 09*ak n_cell 28 5. 5 7
*See first part of X. Liu https: //indico. cern. ch/event/896561/ BC 2 RF Structure Parameters* • provide the same total accelerating voltage as present use smallest number of Klystrons Name Existing New Units num of sections 5 11 - number of structures/sec 16 2 - Total number of structures 80 22 - structure length 0. 23 0. 9248 m gradient 70. 11521739 63. 4098 MV/m cell_a 5. 4374 1. 5 x 3. 625 4. 4053 (1. 5*2. 9) mm cell_g 8. 5417 3. 5316 mm cell_l 10. 417 (1500) 8. 333 (1200) mm ü Power/meter is greatly reduced: 200/0. 23 -> 155/0. 92 ü Reduced number of klystrons*: 80 -> 22 *Klystron is very expensive (200 K – 400 k CHF*, private communication with Carlo Rossi). 8
Existing Lattice Profile* 150 30 beta_x_i[m] alpha_x_i[m] 100 βx, y[m] 25 beta_y_i[m] 20 alpha_y_i[m] 75 15 50 10 25 5 0 9193. 2 -25 9243. 2 9293. 2 -50 * Produced by PLACET plain tracking 9343. 2 9393. 2 9443. 2 9493. 2 0 9543. 2 -5 -10 Z[m] 9 αx, y[m] 125
New Lattice Profile* beta_x_i[m] 125 100 βx, y[m] 30 beta_y_i[m] 25 alpha_x_i[m] 20 alpha_y_i[m] 75 15 50 10 25 5 0 9193. 2 -25 -50 9243. 2 9293. 2 9343. 2 9393. 2 9443. 2 9493. 2 0 9543. 2 -5 -10 Z[m] 10 αx, y[m] 150
BBA Settings: Static Imperfections Dipole σ_pos[µm] σ_roll[µrad] σres[µm] 30 100 — Strength[%] 0. 1 Quadrupole 30 100 — 0. 01 for CA and TAL, 0. 1% for others Sextupole BPM 30 30 100 — 1 0. 1 — In scripts: Scripts/survey. tcl Main/set_imperfections. tcl 11
Correction Methods* One-to-one (OTO) correction (tune the beam to BPM center) BPM BPM *Also see C. Gohil’s summary at: https: //indico. cern. ch/event/913683/ Or Y. Han’s paper: Y. Han, etc. Journal of Instrumentation, vol. 12, no. 06, P 06010, 2017 12
Dispersion Free Steering (DFS) correction (minimize beam trajectory of different energy) 13
Sextupole Tuning (Moving sextupoles horizontally/vertically to minimize the beam emittance) • 14
Correction Setting Table Scripts/set. Param. tcl OTO DFS o o o o o SR BC 1 Boost Linac ncarray(1, 1) ncarray(1, 2) ncarray 2(1, 1) ncarray 2(1, 2) 40 0. 5 100 0. 5 de 0. 05 nbins beta 0 beta 11 wgt OTOnloop DFSnloop 1 5 3 4 5 1 7 2 30 2 2 Long Turn Central Vertical Transfer Around Arc Transfer Line Loop 1 40 50 50 40 0. 5 20 0. 5 Turn Around Loop 2 40 0. 5 BC 2 30 0. 5 0. 05 0. 1 4 1 3 30 2 2 4 1 5 3 3 1 2 1 30 2 2 4 1 3 30 2 2 30 2 15 2
CLIC-RTML BBA correction/tuning • • Tuning until TAL 2 -end (OTO & DFS & ST) BC 2 • • BC 2 -ST • • OTO & DFS & ST -> x 3, starting from 6 th, five sextupoles BC 2 -ST-extra • • • (OTO & DFS -> x 5) TAL 1 ->TAL 2: ex , ey, TAL 2 ->BC 2: ex, ey, starting from 1 st, five sextupoles Tuning on selected cases: • • BC 2 -ST-extra-select • TAL 1 ->TAL 2: ex , ey • TAL 2 ->BC 2: ex, ey, starting from 1 st, five Sextupoles BC 2 -ST-extra-select-2 • CA->LTL: ex, ey, starting from 1 st, five Sextupoles • TAL 1 ->TAL 2: ex , ey • TAL 2 ->BC 2: ex, ey, starting from 1 st, five Sextupoles 16
Tuning until TAL 2 -end (OTO & DFS & ST) TAL 2 STx 2 TAL STx 3 84 93 93 99 99 99 TAL 2 STx 1 Folders: BBA_2020 -1 -new-TAL 2 -update 17
BC 2 -Correction and Tuning Results RTML-BC 2 -end ex<820 ey<8 BC 2 60 67 BC 2 -ST 60 90 BC 2 -ST-extra 82 97 extra-2 86 98 & 54 80 83 Section-270 -end ex<820 ey<8 & 64 87 80 97 78 86 97 83 86 98 84 * Case 10 improved by setting beta_1=8 in Scripts/Set. Pram. tcl, good at 270 -end but not at BC 2 -end 18
Difficult Cases 19
Summary and further studies • Aperture <a>=4. 4 mm is not good enough to pass BBA test but very close • Possible solutions: • More tuning • Slightly large aperture • Lattice re-design • (Loosen emittance requirement) 20
Acknowledgement This study has been based on many people’s previous work: Yanliang Han, Andrea Latina, Chetan Gohil, Jim Ögren, etc. Special thanks to: Andrea Latina for overall guidance and in-time advice Chetan Gohil for sorting out the BBA lattice and scripts Jim Ögren for usage of PLACET on Condor Xiaowei Wu, Igor Syratchev, Jinchi Cai, Carlo Rossi for RF related topics (I am totally newbie to RF) And many others for jump-in discussions 21
Failed Lattice Profile 900 800 beta_x_i[m] 700 beta_y_i[m] Beta [m] 600 500 400 300 200 100 0 9193. 2 9243. 2 9293. 2 9343. 2 9393. 2 Z[m] 9443. 2 9493. 2 9543. 2 9593. 2 22
Tuning until TAL 2 -end (OTO & DFS & ST) 820 800 8 7 * Merit function: ex/700+ey/5 Folders: BBA_2020 -1 -new-TAL 2 -update 23 First five sextuples in TAL 2 selected
BC 2 (OTO & DFS ->x 5) folder: BBA_2020 -1 -new-TAL 2 -BC 2 24
Present process 1) Plain tracking to CA-start ->beam 4 Main/plain_sr_ca. tcl make_particle_beam 1 beamparams $beamdir/particles. in $wakedir/zero_wake. dat 2) Varying offset_X/Y with 10 x proposed by fminsearch 3) Tracking from CA-start to LTL-start 4) test emittance with 1% uncerntainty(randn) ->2) run. tcl system("placet -s Main/plain_sr_ca. tcl machine $machine"); [cor, cor_min] = fminsearch("correction 1", X, options); system("placet -s Main/save_ca_sextupole_offset. tcl machine $machine"); source $scriptdir/load_mag. tcl Octave { Start = placet_get_name_number_list("rtml", "Marker-SR-start"); End = placet_get_name_number_list("rtml", "Marker-CA-start"); [E, B] = placet_test_no_correction("rtml", "beam 1", "None", 1, Start, End); printf("n. Emittance entering CAn"); printf("Emittance: %10. 2 f, %10. 2 f nmn", E(end, 2)*100, E(end, 6)*100); save -text $beamdir/beam-$machine-4. dat B; } CA-start Scripts/sextupole_tuning_functions. m function [ex, ey] = read_emittance(filename) # reads emittance from plain tracking and adds an error E = load(filename). E; ex = E(end, 2)*(1 + 0. 01*randn()); ey = E(end, 6)*(1 + 0. 01*randn()); end 1% uncerntainty Main/save_ca_sextupole_offset. tcl source $scriptdir/load_mag. tcl Octave { X = load("X_tal 1 -$machine. dat"). X; offset_X = round(10*X(1: 5)); offset_Y = round(10*X(6: 10)); Multipole_all = placet_get_name_number_list("rtml", "Multipole-left*"); N_Multipole = length(X)/2; Multipoles = Multipole_all(1: N_Multipole); placet_element_vary_attribute("rtml", Multipoles, "x", offset_X); placet_element_vary_attribute("rtml", Multipoles, "y", offset_Y); } source $scriptdir/save. tcl Save tuned result function demitt = correction 1(X) global machine; save(['X_ca-' num 2 str(machine) '. dat'], 'X'); success = system(['placet -s Main/sext_ca. tcl machine ' num 2 str(machine)]); if(success == 0) [ex, ey] = read_emittance(['Emitt 0 -CA-rtml-30 -' num 2 str(machine) '. dat']); demitt = ex/7. 0 + ey/0. 05; else # there was an error in tracking demitt = 10000; end Merit function Main/sext_ca. tcl source $scriptdir/load_mag. tcl Octave { X = load("X_ca-$machine. dat"). X; offset_X = round(10*X(1: 5)); offset_Y = round(10*X(6: 10)); Multipole_all = placet_get_name_number_list("rtml", "Multi-CA*"); N_Multipole = length(X)/2; Multipoles = Multipole_all(1: N_Multipole); placet_element_vary_attribute("rtml", Multipoles, "x", offset_X); placet_element_vary_attribute("rtml", Multipoles, "y", offset_Y); } 10* proposed Octave { load "$beamdir/beam-$machine-4. dat"; placet_set_beam("beam 4", B); } Reset to CA-start plain. Tracking CA rtml Marker-CA-start Marker-LTL-start beam 4 Track to LTL-start File output and input takes time -> merge into single script without handling disk 25 files
Comments On The Computation Platform • Single computer with multiple CPU (6 cores/12 threads @ 4. 9 GHz) • Fast for single case test • Slow/impossible for many (e. g. 100) cases • HTCondor (PLACET not capable of parallel computing) • Slow for single case (hours ~ days) • Capable of many cases (a MUST for RTML BBA study) 26
- Slides: 26