2. T. Argyropoulos, LIU 2013 12/04/2013 Progress in the SPS: RF studies and beam quality T. Argyropoulos, T. Bohl, J. E. Muller,, H. Timko, E. Shaposhnikova With condributions from: H. Bartosik, H. Damerau, S. Hancock, Y. Papaphilippou LIU 2013 12 April 2013 Many thanks to all the SPS, PS and PSB OP and RF/FB section for LLRF

3. T. Argyropoulos, LIU 2013 12/04/2013 Outline  Beam studies in transverse plane see talk of H. Bartosik  Beams studies in longitudinal plane  PS – SPS transfer (Q26)  LHC 25 ns beam quality (Q20)  Instability thresholds in a single RF  Multi-bunch  Single bunch  Longitudinal impedance  Long injected bunches (τ 4σ ~ 25 ns) with RF off  Quadrupole synchrotron frequency shift at injection  Summary

4. PS – SPS transfer studies  A larger longitudinal emittance is desirable for the LHC beams in both PS & SPS  For beam stability  For beam quality at SPS extraction  improvement seen in operation  Limitation due to losses  PS-to-SPS transfer studies with 50 ns beam with single batch (36 bunches) 40 % larger ε l can be achieved with the same transmission using optimised rotation and an additional 40 MHz cavity Bunch length at SPS injection is not the best indicator of transmission Studies with 25 ns LHC beam and 4 batches are planned after LS1 (with 2 nd 40 MHz cavity)

5. LHC 25 ns beam quality (Q20)  Stable beam up to ~1.35x10 11 p/b at 450 GeV: in double RF + BUP (after the LLRF set up, necessary for N p >1.2x10 11 p/b)  Ramp was starting just after 4 th inj. => 300 ms moved from ft to fb  time for the 4 th batch to filament  SPS controlled emittance BUP is now more efficient  Remaining issues at higher intensities (>1.4x10 11 p/b)  “U-shape” of the bunch length distribution at ft  CEBUP+Beam loading  Bunch length decrease along fb  1 st batch smaller ε l (phase loop) Not trivial going for higher intensities! Longer fb and N p ~1.35x10 11 p/bShorter fb and N p ~1.2x10 11 p/b

6. LHC 25 ns beam quality (Q20)  Transmission Bunch intensity at ft versus bunch intensity at injection for 3 MDs  High losses: 15-20 %  Slow losses at flat bottom (large transverse beam size, orbit, …?)  Fast losses at injection at the end of 3 rd and 4 th batch (e-cloud, coupled-bunch instability, transverse damper, …?)  50 ns beam: losses ~ 5-6 % for N p ~1.7x10 11 p/b at ft

7. LHC 25 ns beam quality (Q20)  Strong dependence of beam stability on the injected parameters (bunch length, emittance)  Instability observed after the BUP  BUP cannot be optimized when the spread in the injected bunch lengths is large  especially for smaller injected emittances Different potential well distortion from bunch to bunch  large variation of the incoherent synchrotron frequency shift (Δf s ~Δτ/τ 4 ) Before BUP Larger ε l from PS  more stable beam in the PS  less spread in bunch lengths at extraction  more stable in the SPS

8. Instability thresholds – single RF (Q26)  Multi-bunch (one batch 50 ns beam of ε l ~0.35 eVs):  Threshold ~ 2x10 10 p/b !! (with RF FB, FF and LD):  Clear dependence on energy E and long. emittance ε l  Does not depend on the number of batches  short range wake  compatible with the main impedance of the TWC 200 MHz (Q = 150) and TWC800 (Q = 300)  Comparison between 25 and 50 ns beam  threshold scales with current, roughly as 1/E th ~ N b /T b 50 ns: E th ~ 160 GeV with N b = 1.6x10 11 25 ns: E th ~ 110 GeV with N b = 1.2x10 11  Same scaling in Q20 only more stable (factor ~ η)

9. Instability thresholds – single RF (Q26)  Single bunch:  Measurements with phase loop ON and OFF (compare with simulations )  Threshold estimation for nominal emittances (0.35 eVs): Flat bottom: N th ~ 1.3x10 11 (2 MV) Flat top: N th ~ 1.1x10 11  Theory predicts that threshold at flat bottom should be much higher than flat top  Measurements show continuous oscillations at flat bottom (~11 s)  Loss of Landau damping Large voltage mismatch (2-3 MV, as in nominal operation)  strong quadruple oscillations (loss of Landau damping)  instability growth on flat bottom Small voltage mismatch (1 MV)  beam is more stable Can not be used in operation due to beam loading in the RF systems  beam loss Not an issue in Q20! Possible explanation: initial filamentation + bunch distribution coming from the PS after bunch rotation (studies on-going) Examples N p ~ 1.1x10 11

10. Longitudinal impedance sources  Motivation  Instability of the LHC 50 ns beam in Q26 at flat bottom  High frequency modulation of the bunch profile  Long injected bunches (τ 4σ ~ 25 ns) with RF off : method to identify resonant impedances used in the past (Measuring the Resonance Structure of Accelerator Impedance with Single Bunches, T. Bohl, T. Linnecar, and E. Shaposhnikova, PRL, 1998)  Measurements in Q26 (reported on LIU review day in August): − Strong resonant peak at 1.4 GHz above N p ~ 8x10 10 − Already observed in 2001 and 2007 2007 N p = 8x10 10 2012 N p = 1x10 10

11. Longitudinal impedance sources  Measurements in Q20 in 2013  to exclude coupling due to transverse impedance  similar results as in Q26 It should be a longitudinal resonant impedance with high R sh /Q that causes this instability

12. Longitudinal impedance sources  Simulations to identify the parameter space of this impedance  Using the present SPS longitudinal impedance model − RF systems (200 MHz + HOM (629 MHz), 800 MHz) − All kickers (from C. Zannini) − Resonance at 1.4 GHz (varying parameters, f r, Q, R sh )  Initial distribution: Tomography in the PS  ESME simulations until extraction to SPS  Simulations with 2 codes (code in Matlab and HEADTAIL) Parameters derived from simulations f r (GHz)QR sh (kΩ) 1.35 – 1.455 – 10300 – 400 SimulationsMeasurements Example – Q26 - N p = 1x10 11

13. Longitudinal impedance sources  Possible candidates for this impedance source:  HOM in the 200 MHz and 800 MHz RF systems: new measurements (lab, tunnel) and simulations are planning (F. Caspers, R. Calaga, B. Salvant, J. E. Varela)  Enamel flanges at each side of each BPM in the SPS At least 436 flanges  R sh ~300 kΩ! Low frequency capacitive part! Can possibly explain the results of the quadrupole frequency shift measurements (next slides) Survey of similar devices is needed during LS1. Talk of B. Salvant, LIU-SPS BD WG 12/05/2011

14. Longitudinal impedance model  Quadrupole synchrotron frequency shift as a function of intensity  Many measurements have been done in the past (E. Shaposhnikova et al. PAC 2009)  Complicated dependence on bunch lengths  Need of same experimental conditions to compare with old results  very small injected longitudinal emittances (~0.1-0.2 eVs at injection)  Beam prepared in the PSB (S. Hancock, H. Damerau) − ε l was adjusted by longitudinal shaving during acceleration − Intensity was later selected by adjusting the CEBUP − Very reproducible results!! Injected bunches in the SPS for different emittances and intensities

15. Longitudinal impedance model  Single proton bunch injected in the SPS (26 GeV) in single RF (V200 = 900 kV)  Bunch profile acquisitions (first 1400 turns)  Gaussian fit (4σ bunch length)  fit the quadrupole oscillations to get 2f s  Results from 2 MDs with different emittances The slopes (b) give us information about the inductive part of the impedance Compare with measurements in 2008 Compare with simulations as a benchmark to the SPS Impedance model Example of ε l = 0.15 eVs and N p = 10 10

16. Longitudinal impedance model  Comparison with measurement of 2008  Comparison with simulations: 2013 Smaller slopes in 2013  Impedance reduction  effect of serigraphy on the MKEs! Simulations with the present SPS impedance model does not give good agreement Something is overestimated or the capacitive part of the enamel flanges is missing (?)

17. Summary  PS to SPS transfer studies 40 % larger ε l can be achieved with the same transmission using optimised rotation and an additional 40 MHz cavity  LHC 25 ns beam quality with Q20  Stable beam, acceptable for transfer to LHC up to N p ~ 1.35 10 11 p/b (τ 4σ ~ 1.55 ns - ε l ~0.45 eVs at ft)  Not trivial going for higher intensities!  Large Losses: 15 – 20 %  Strong dependence on the parameters of the injected beam  Instability thresholds in single RF with Q26  Single bunch: similar threshold at fb and ft probably due to mismatch at injection  Multi-bunch: Impedance of TWC 200 MHz, 800 MHz  Q20 thresholds are increased as expected (~ η)  Longitudinal impedance  1.4 GHz: enamel flanges at each side of each BPM in the SPS seem to be good candidate  Quadrupole frequency shift: impedance reduction since 2008  serigraphy of MKEs  Simulations and measurements are ongoing to verify the present SPS impedance model and identify new sources based on all of these measurements  crucial for future studies!

18. Back up slides

19. T. Argyropoulos, LIU 2013 12/04/2013 LHC 50 ns beam quality  The beam quality was degraded in the mid of June  BQM rejections for LHC  Comparing two consecutive fills with bad and good beam quality and similar intensities (N p ~ 1.5x10 11 p/b at flat top): All cycles of these LHC Fills Bad LHC FillGood LHC Fill Difference: probably PSB adjustments  At these intensities SPS is very sensitive to beam quality coming from PS (at the limit of stability) T. Bohl, LIU-SPS BD WG 21/06/2012

20. T. Argyropoulos, LIU 2013 12/04/2013 LHC 50 ns beam quality  Detailed measurements (MD Week 25 )  Bunch intensity 1.6 x 10 11, higher than for LHC filling at that time  Batches are unstable on flat bottom  beam blows up: big bunch tails (difference between Gaussian fit and scaled FWHM)  Different optimum settings for filing with 2 and 4 batches. Why? Bunch length evolution along the cycle Measurements on 18/06/2012 Typical bunch profile -Time(cycle)~3s

21. T. Argyropoulos, LIU 2013 12/04/2013 LHC 50 ns beam quality  Bunch intensity ~1.5 x 10 11 p/b at flat top  Similar observations as on 18 th June: Unstable bunches at flat bottom Batch 4 unstable at the end of acceleration ramp or flat top  Large spread in the injected bunch lengths with the nominal emittance blow up in the PS (3x3.5 kV)  correlated with instabilities in the SPS More measurements in the dedicated MD (Week 26) at 25/06/2012

22. T. Argyropoulos, LIU 2013 12/04/2013 LHC 50 ns beam quality  4 th batch unstable at the end of ramp  less (no) time spend on flat bottom  smaller emittances  SPS controlled BU is not optimum and less efficient Solution 1: take out the 4 th dip in the 200 MHz RF Voltage program (const at 3 MV at flat bottom)  blow-up the beam with more mismatched voltage Cost: slightly worse transmission (typically 1%) OPERATIONAL Solution 2: increase injected emittance: controlled emittance BU in the PS from ε L ~0.35 eVs (3x3.5 kV) to ε L ~0.38 eVs (3x4.5 kV) Improves also the uniformity of the injected bunch lengths  more stable bunches coming from PS NOT YET OPERATIONAL  Modifications of the 800 MHz RF Voltage didn’t help the beam quality

23. T. Argyropoulos, LIU 2013 12/04/2013 Longitudinal impedance identification The impedance identification measurements in W28 and W30 were initiated by:  The increase of the average bunch length of the batches along the flat bottom  Observations of high frequency pattern on the nominal single bunches at flat bottom after loss of Landau damping (see later)

24. T. Argyropoulos, LIU 2013 12/04/2013 Longitudinal impedance identification 23 Same method used in the past (Measuring the Resonance Structure of Accelerator Impedance with Single Bunches, T. Bohl, T. Linnecar, and E. Shaposhnikova, PRL, 1998) Experimental conditions:  Long single proton bunches (ε L (90%) ~ (0.23 – 0.26) eVs – τ inj ~ (25-30) ns)  Small momentum spread (to be more unstable and debunch slowly)  Intensity scan from 0.5x10 11 to 2.0x10 11 p  SPS RF off Method: Acquisitions of beam profiles for a period of ~90 ms after injection each 10 turns. followed by a Fourier analysis  The presence of resonant impedances with high R/Q and low Q leads to line density modulation at the resonant frequencies Video of bunch profile 1.4 GHz Example with N p =1.58x10 11 and τ inj (4σ) = 28.5 ns

25. T. Argyropoulos, LIU 2013 12/04/2013 Longitudinal impedance identification 24  Averaged for all (15) acquisitions with intensity ~1x10 11 p  The mode amplitude at 1.4 GHz peak is developing faster than the main at 200 MHz Mode amplitude versus Time  Strong peak above N p ~ 8x10 10 p Already observed in measurements in 2001 and 2007 – this impedance was always there! These measurements are used now for impedance evaluation through simulations. Very preliminary: R/Q ~ 30 kOhm, Q~7 Mode amplitude ratio of 1.4/0.2 versus Intensity

26. T. Argyropoulos, LIU 2013 12/04/2013 Instability thresholds in a single RF – Single bunch 25  Measurements in W17: Plan: Find the instability thresholds at FT for single bunch in single RF for Q26 and Q20 optics Aim: Use these results to compare with simulations  verify the current SPS longitudinal impedance model Due to many problems (POPS, RF TX were tripping) limited data Results for ε L ~ 0.3 - 0.35 eVs: Q20 : intensity threshold at flat top > 1.5x10 11 Q26 : intensity threshold at flat top ~ 1.0x10 11 scope saturation Δτ : bunch length amplitude oscillations during the ramp The horizontal line in the plots indicates the noise level

27. T. Argyropoulos, LIU 2013 12/04/2013 Instability thresholds in a single RF – Single bunch 26  Measurements in W34 (last week, preliminary results):  Only in Q26  Phase Loop ON and OFF (for better comparison with simulations) Difficult to measure accurately due to strong dependence on bunch parameters (emittance and distribution) Examples of bunch length evolution along the cycle for different 200 MHz RF voltage configurations  Bunch intensity ~1.1 x 10 11 p at injection  Phase loop ON  Large voltage mismatch (2-3 MV, as n nominal operation)  strong quadruple oscillations due to loss of Landau damping  instability growth on flat bottom  Small voltage mismatch (1 MV)  larger synchrotron frequency spread  beam is more stable Can not be used in operation due to beam loading in the 200 MHz RF system  beam loss

28. T. Argyropoulos, LIU 2013 12/04/2013 27 Results at flat bottomResults at flat top  Bunches injected at 1 MV are stable at flat bottom (ε L ~0.28 – 0.35 eVs)  Become unstable at the end of ramp – flat top with N p >1x10 11 p  Bunches injected in 2-3 MV blown up at flat bottom  stable at flat top (these int.) Instability thresholds in a single RF – Single bunch

29. T. Argyropoulos, LIU 2013 12/04/2013 28  Measurements in Week 26 – Flat Bottom: Intensity ~ 1.47x10 11 p/b at injection Intensity ~ 1.8x10 11 p/b at injection Unstable 1.5x10 11 < N th < 1.8x10 11 p/b  Issues when LHC was asking these intensities – instability threshold in Q26!  Threshold is close (above) the single bunch threshold for similar (higher) longitudinal emittance  is this a single bunch effect? Stable Instability thresholds in a single RF – Single batch

30. T. Argyropoulos, LIU 2013 12/04/2013 29  Measurements in Week 26 – Flat Top: Beam is unstable at the end of ramp even below 0.4x10 11 p/b !  At flat top multi-bunch threshold is much lower than for single bunches  most probably due to effect of 800 MHz beam loading (stronger for short bunches)  V bl comparable to the 800 MHz voltage (when in operation) but only in completely wrong phase between the two RF systems (~π/2)  Test planned next (last) MD with re-programmed phase with the 800 MHz on but very difficult to take into account intensity and bunch shape  FB and FF are absolutely necessary – should be operational after LS1  time for commissioning!  Effect present also in a double RF operation (corrected by V rf at 800 MHz)  Controlled longitudinal emittance blow-up is always necessary (even in Q20), Vind depends strongly on bunch length (> factor 10 between FB and FT) Instability thresholds in a single RF – Single batch

31. T. Argyropoulos, LIU 2013 12/04/2013 30  Measurements in floating MDs W30 and 32 and on 11/08 when injectd to LHC (intensities ~ 1.6x10 11 p/b at flat top):  Stable beam on FB for these intensities (factor 3 gain is expected)  Without controlled emittance blow-up : some dipole or quadrupole oscillations at the end of ramp or flat top: τ FT ~ 1.5 -1.6 ns and ε FT ~ 0.4 - 0.46 eVs Beam was sent to the LHC!  With controlled emittance blow-up: stable beam along the cycle τ FT ~ 1.55 – 1.75 ns and ε FT ~ 0.42 – 0.53 eVs Increase of bunch lengths along the batches at flat top – not understood LHC 50 ns beam in SPS Q20

32. T. Argyropoulos, LIU 2013 12/04/2013 Summary of MD results (2012)  LHC beam quality in Q26:  Unstable bunches at flat bottom for N > 1.5 10 11  Batch 4 spends no time at flat bottom  unstable at the end of ramp or flat top  Enhanced instabilities due to increased spread of incoming bunches  Solutions:  Take out the dips in the 200 MHz RF voltage program (done)  Introduce more time at flat bottom after the 4 th injection (longer FB)  Increase emittance blow-up in the PS (3x4.5 kV)  more stable in PS  less spread of bunch parameters at injection in the SPS  Cost: more losses - about 1 %  Q20 (still better with longer FB)  Longitudinal impedance identification using long injected bunches (~25 ns) with RF off :  Measurements as in the past (1997,2001,2007…) of the beam spectrum show a high mode amplitude at 1.4 GHz  low Q impedance to be identified

33. T. Argyropoulos, LIU 2013 12/04/2013 Summary of MD results (2012)  Longitudinal instability thresholds measured at flat bottom and flat top in single RF:  Single bunch (ε L ~0.28 – 0.32 eVs): Flat bottom: Loss of Landau damping due to injection V mismatch – stable with low (matched) capture voltage > 1.4x10 11 Flat top: - N th ~1x10 11 p  Single50 ns LHC batch: Flat bottom: 1.5x10 11 < N th < 1.8x10 11 p/b  similar to single bunch Flat top: N th < 0.4x10 11  stronger effect of beam loading at 800 MHz RF (?)  50 ns LHC beam Q20  N p ~(1.5 - 1.6)x10 11 p/b on flat top  Stable beam on FB  Acceptable beam parameters at extraction (bunch length) – injected to LHC  Scaling of bunch length/stability between Q20 and Q26 as expected  More possibilities with upgraded 200 MHz RF (LS2)

34. T. Argyropoulos, LIU 2013 12/04/2013 Plans for the rest of 2012 (2013?) MDs:  High intensity 50&25 ns beam in Q20  operational. And Q26? (losses increased in the first test)  Study (thresholds) of single and multi-bunch instability at injection for Q26 and during ramp for Q20 and Q26 in a single and double RF systems  Reference impedance measurements from quadrupole frequency shift at injection and stable phase shift (? – more difficult) Simulations:  Identify the impedance from comparison with measurements done with long (RF off) and short (RF on) bunches  Reproduce the measured thresholds for single bunches (PL on/off)

35. T. Argyropoulos, LIU 2013 12/04/2013 Questions to be answered  Requirements for the 800 MHz system:  2 nd cavity operational (idle at the moment) – more voltage and better control  FB and FF for phase and voltage control under strong beam loading for the two cavities (work in progress)  Phase calibration (now based on bunch shape/stability)  What else do we need to know to estimate performance after LIU:  Impedance source of longitudinal instabilities  measurements of HOMs in the 200 MHz and 800 MHz TW RF systems in labs (spares, not fully equipped) and in situ (tunnel) during LS1 (F. Caspers, E. Montesinos + new fellow/PhD student)  1.4 GHz  Complete impedance model