ADT Tune Measurement F. Dubouchet, W. Hofle, D. Valuch Acknowledgement: R. Calaga, F. Roncarolo, E. Bravin, shift crews New developments and tests on August 1 st and 12 th 2012 W
ADT Tune Measurement Reminder (Chamonix): Plan: implement gain modulation to selectively reduce ADT gain for witness batch (Chamonix: 12 bunches Now: 6 bunches) done, but no improvement for BBQ situation: BBQ gating difficult, BI working on it, tests upcoming long term ADT plan: tune from residual ADT signal massive data treatment, limitations of VME technology not operational before LS1 as new hardware required Plan B: kick beam (witness batch) selectively by ADT and record free oscillation make an operational tool by combining with already commissioned batch by batch gain modulation use new word long dedicated study buffer on FPGA of ADT DSPU VME card method gives also the ADT damping time along the cycle monitoring of performance W
Timeline of tests on August 1 st, : 6 turns kick (16x) 2: 3 turns kick (21x) 3: 5 turns kick (28x) 4: 5 turns kick (7x) 5: 5 turns kick (500x) 6: 5 turns kick (52x) total: 624x 5: frequency trim 6: tune trim 6 6 bunches, nominal intensity 1.5x10 11 /bunch beam measurements with 1 beam only, V-plane; both beams injected (6 bunches) using existing firmware / infrastructure (abort gap cleaning), kicking at 0.31 in tune 450 GeV, but damper gain adjusted to model situation at flat-top (damping time of 100 turns) W
Lifetime during excitation of 6 bunches start of continuous burst excitation [1 burst every 2.34 s] BLM lifetime drops from 200 hours to below 100 hours W
Lifetime cont. end of burst mode of 500 excitations BLM lifetime recovers after stop of excitation excitations re-started W
Blow-up of the 6 bunches as a result of kicks initial test (manual triggers) then 500 bursts (each 5 turns) from 13:45 to 14:05 (19.5 min) 2.34 seconds per burst d = 100 turns max amplitude 100 m at Q7V.B1 of damper 1 st period: 45 minutes, 3.3 nm / min m for 72 excitations + normal blow-up 2 nd period: 19.5 minutes, 43.3 nm /min 43.3 nm /min m in 19.5 min m for 500 excitations + normal blow-up deduce blow-up w/o excitation (assumed to be a constant rate per unit of time): 0.65 nm/min, i.e nm during 1 st period nm during 2 nd period blow-up per trigger (excitation) 1 st period: 72 triggers: 1.66 nm per excitation 2 nd period: 500 triggers: 1.66 nm per excitation excitation bursts (approx) F. Roncarolo, E. Bravin W
Tune trim (data set ‘6’) 1e-3 up and down tune time (hours, decimal) W
RF frequency trim VCXO word 40 Hz up, i.e. from -20 Hz to +20 Hz with respect to central frequency of ’837 MHz FGC frequency (logging on change) time [1 min / tick] 14:03:37 W
RF frequency trim (data set ‘5’) 40 Hz up, i.e. from -20 Hz to +20 Hz with respect to central frequency ’837 MHz Q = -4.9e-4 p/p = - 3.2e-4 Q’ = single bunch (!), single pick-up Q7.VB1, 2730 turns, 500 times kicked tune time (decimal) 14:03:3x W
2 nd peak at -20 Hz also visible for FFTs of individual bunches Real modulation of beam tune or artifact ? Data Set 5, 500 measurements, FFT average over 6 bunches
W Example of a single bunch, single measurement FFT (V.B1) 50 Hz 20 Hz 2730 turns resolution too limited to clearly resolve if there is a systematic peak at 20 Hz in the low frequency range
3.5 TeV, 8 MV f s =22.9 Hz compatible with Q s and 2Q s, f s =22.9 Hz 50 Hz Clean spectrum between 50 Hz and tune tune 2010 data 17 May 2012 damper off low bunch intensity FFT of 4 bunches (65536 turns) frequency in Hz FFT (B2.V Q7L) 4.5 Hz broad Hz 2010 data W
Data Set 5, 500 measurements x 6 bunches
W Data Set 5, 500 measurements x 6 bunches
W average: 1.8x10 -3 in tune 20.2 Hz cut: only measurements are displayed in which upper and lower tune peaks are the highest peaks for at least one bunch
2 nd test (12 th August 2012) dedicated firmware deployed (Daniel), new dedicated fesa classes (Frederic) for excitation similar to AGC, but “skimmed down” excitation scales automatically with normalised damper gain dedicated parameters and resources: window (gating on part of the beam) DDS to generate excitation with programmable excitation frequency number of kicks (in turns) dedicated memory on FPGA (16384 bunch x turns) max 2730 turns for each of the 6 bunches kicked beam test (12 th August 2012) injected 2x6 bunches per beam with standard, i.e. minimal possible spacing symmetric pattern with collisions in IP1 and IP5 only (bucket 1 and 651) use first 6 as witness bunches, kicked by ADT, 4 planes x beams kicked every beam and plane once per 4 seconds (rotating kicks every second between beam and plane) 4 turns kicked at 0.28 (H) and 0.31 (V), amplitude 0.8 of max real life W no changes to operational Parameters and functions
Damping times Fill 2956 collisions flat top and squeeze Beam modes: 19:33 prep4ramp 19:37 ramp 19:50 F-top 20:15 squeeze 20:39 adjust 21:26 dump W injection Daniel’s tool ramp
Horizontal Beam 1 Fill 2956 collisions with / modes tune change at start of squeeze ramp setting-up then refill Beam modes: 19:33 prep4ramp 19:37 ramp 19:50 F-top 20:15 squeeze 20:39 adjust 21:26 dump flat bottom high ADT gain need lower ADT gain for the first 6 W FFTs from 6 bunches averaged
Horizontal Beam 2 Fill 2956 collisions spurious at n x f rev +/ Hz messes up spectrum need to kick harder ramp: tune took a dive due to tune FB locking on wrong line switched off and corrected at FT flat bottom high ADT gain need lower ADT gain for the first 6 setting-up then refill Beam modes: 19:33 prep4ramp 19:37 ramp 19:50 F-top 20:15 squeeze 20:39 adjust 21:26 dump spurious W FFTs from 6 bunches averaged
Vertical beam 1 Fill 2956 collisions flat top spurious messes up spectrum need to kick harder Beam modes: 19:33 prep4ramp 19:37 ramp 19:50 F-top 20:15 squeeze 20:39 adjust 21:26 dump W FFTs from 6 bunches averaged
Vertical Beam 2 Fill 2956 collisions flat top with spurious messes up spectrum need to kick harder Beam modes: 19:33 prep4ramp 19:37 ramp 19:50 F-top 20:15 squeeze 20:39 adjust 21:26 dump W FFTs from 6 bunches averaged
Individual bunches W Hz line moves with beam tune !
Individual bunches W Hz line moves with beam tune !
W F. Roncarolo, E. Bravin Emittance blow-up
W F. Roncarolo, E. Bravin Emittance blow-up
W F. Roncarolo, E. Bravin Emittance blow-up
W F. Roncarolo, E. Bravin Emittance blow-up
Towards making an operational tool – next steps implement control of parameters in sequencer, plus start / stop functionality (t.b.d.) implement FFTs in front-end (0.45 ms/FFT, 12 FFT 5.4 ms (6 bunches, 2 PU) average FFTs in front-end publish averaged FFT and tune (peak), fitting (?) make operational with 1374 bunches once sequencer work completed, and acquisition working explore parameter space to get desired resolution understand and decide how to deal with the spurious 20 Hz line next to the tune (!) needs: gain modulation programmed to tailor damping time of the 6 bunches kicking harder at flat top when spurious lines perturb (kicking to larger amplitudes by increasing the number of turns kicked or by increasing normalised gain + additional gain modulation to maintain same gain for main beam) suggestion: use more turns of excitation when amplitude is insufficient and flat-bottom: gain modulation 16/128 for the 6, 127/128 for rest ramp / flat-top / squeeze: gain modulation 64/128 for the 6, 127/128 for rest W