Feedbacks & Stabilization Getting them going

1. Feedbacks & Stabilization Getting them going J. Wenninger AB/OP • Machine parameters to stabilize. • How far can one get without FBs ? Thanks to A. Burns & R. Jones Chamonix 03 / Presentation 7.5 / J. Wenninger

2. Not discussed here • The following (really) fast feedback systems : • Feedback loops within the RF system. • Transverse damper system : • This system is required at an early stage to damp injection oscillations and instabilities. • Instrumentation details : • BPMs, Q-meters, PLLs, AC-dipoles, reference magnets… • Please refer to the various presentations of the instrumentation and magnet groups. Chamonix 03 / Presentation 7.5 / J. Wenninger

3. Main parameters to stabilize • Orbit ~ 50 - 250 mm • Energy @ injection 10-4 • Tunes < 0.003 • Chromaticity ~ 1-2 units • Coupling < 0.005 • Luminosity in physics  Mike’s presentation(s). Growing complexity • Injection and snapback (ramp)… • affected by large dynamic effects ! • are the first steps we have to master ! • Concentrate the discussion on those phases for the nominal cycle. Chamonix 03 / Presentation 7.5 / J. Wenninger

4. Anticipated variations during decay & snapback • Energy : few 10-4 b1 decay, Earth tides… • Orbit : ~ 1-4 mm rms. • Tunes : not so tough but … snapback • Chromaticity : ~ 80 units ! • Coupling : feed-down… snapback Q’ is the really tough one ! Chamonix 03 / Presentation 7.5 / J. Wenninger

5. Feedbacks… • Open loop feedback : • The reference magnet / multipole factory is our main OL system. • Such systems are very sensitive to model errors. • Closed loop feedback : Preferred choice ! • Feed-forward : • Measurements at time t  anticipate changes at t+ Dt • (injection/filling  snapback, one fill to the next, …) • Relies on the machine reproducibility  magnet cycling …. Chamonix 03 / Presentation 7.5 / J. Wenninger

6. Setting up for injection… • Magnets on pre-injection plateau • Multipole factory & reference magnets : evaluate corrections for b2, b3…. incorporated into the PC settings. • With experience : • include ‘empirical’ corrections based on beam measurements. • Magnets to injection settings • The persistent curr. decay changes orbit, energy, Q, Q’… during the next ~ 15 minutes. Changes snapback at the start of the ramp. • Get some beam – perform rough corrections… • If we have them, beam based FBs take full control with beam ! •  need beam all the time to track the parameters. • When do we need the beam-based feedbacks ? Chamonix 03 / Presentation 7.5 / J. Wenninger

7. Orbit feedback • Measurement and correction are well established, • details to be sorted out / finalized  yesterday’s talk ! • At injection one can survive with a simple LEP-like measure-correct orbit ‘auto-pilot’ feedback : • - relaxed tolerances • - drifts are slow enough • The orbit FB should be implemented at an early stage with moderate gain, in any case before we start ramping. Chamonix 03 / Presentation 7.5 / J. Wenninger

8. Energy feedback on injected beam • First beam(s) : match energy scale of SPS & LHC ! • Regular operation : • Follow energy changes by adjusting hor. orbit correctors in the LHC. • ! Sort out interferences with orbit FB – both use the same elements. • Measurements : • Comparison trajectory of injected beam – closed orbit. • Phase of injected beam… • This feedback is not very critical with first beams • – mainly required for larger intensity to keep the amount of • un-captured beam at an acceptable level. Chamonix 03 / Presentation 7.5 / J. Wenninger

9. Q, Q’ & coupling : single kick measurements • Good for Q, Q’ (head-tail monitor) and coupling. Detection : • Dedicated high resolution monitors (mandatory for head-tail). • Orbit system (multi-turn). data readout : problem for the orbit FB due to front-end load ? Main side effect : emittance blow up  limit on kick amplitude and (rate) number ! Collimation : damage to collimators, coll. efficiency, BLM dump triggers  kick amplitude limits : • s/4 to s/2 “default” • up to ~ 1 s for low intensity (1 nominal bunch ?) Simple “measure and correct” system, once every few seconds ! Not suited for real-time feedback. Chamonix 03 / Presentation 7.5 / J. Wenninger

10. Q kicking • Emittance increase due to single kicks (A. Burns) : • 0.4 mm kicks ~ s/2 • resolution 20 mm • 50 turns damping time  reasonable Q’.. • dQ < 10-3 Snapback Injection Will not work as well at higher energy because s decreases… We have to accept some e blow-up if we kick all the way up to 7 TeV… Chamonix 03 / Presentation 7.5 / J. Wenninger

11. Q, Q’ & coupling : AC dipole measurements • Good for Q and (best for ?) coupling. • Detection : • Orbit system (multi-turn).  data readout : problem for the orbit FB due to front-end load ? ‘No’ emittance blow upif Q is ~ known  excitation outside Q spectrum. Collimation : same amplitude limitations than for single kicks. Could be operated continuously (!) for a real-time correction. • Comments on single kick/AC dipole : • They are potential collimator-busters ! • Careful amplitude interlocking and energy tracking is mandatory ! • Difficult to operate at high energy  very small amplitude ! Chamonix 03 / Presentation 7.5 / J. Wenninger

12. Q, Q’ & coupling : resonant BPM • Excellent for Q measurements. • Q’ obtained from RF frequency (slow) or phase modulation (fast). Detection of ~ mm excitations: • Requires a 40 MHz structure on the beam  not suited for single bunches, TOTEM beams !  test ramps must be performed with 12 cons. bunches or so ! Not limited by e blow-up or collimation. Input signal for the Q PLL system  real-time Q FB. Delicate device  not available immediately. Q’ measurement remains delicate even with Q PLL. Chamonix 03 / Presentation 7.5 / J. Wenninger

13. Operation without RT FB Reference magnets / multipole factory : • Track multipoles (n ≥ 2) @ few Hz during injection, snapback… Single kick Q,Q’… measurements : • Adjust / fine tune beam parameters. • Improve predictions of ref. Magnet / M-factory. During filling : • Track parameter changes. • Limit number of kicks or use sufficient number of ‘test’ shots. At end of filling,before launching the ramp : • Collect the integrated changes observed during injection. • Anticipate snapback effects by feed-forward to the PC functions.  should take care of the worst dynamic effects.  need a very strict book-keeping of trims ! Chamonix 03 / Presentation 7.5 / J. Wenninger

14. Tolerances without RT FB dQ dQ’ ±0.01~ ± 15 units  ± 0.001 ~ ± 1-2 units  ± 0.003 ~ ± 6 units Assume 20% error on the predictions/ corrections from ref. magnets & M-factory : Beam measurements during filling : Feed-forward for snapback (prediction) : Measure & correct every few seconds in snapback… With experience we might achieve for the snapback : Q ~ OK 0 < Q’ < 10 or better – unlikely to reach ultimate tolerances. How far can we go like that ? With what kind of beams ? Chamonix 03 / Presentation 7.5 / J. Wenninger

15. Remarks on operation without FB : • All effects come together – makes life a bit more difficult ! • We must rely on a good understanding of the dynamic effect and on a reproducible machine – with a machine that just started up !! • The emittances will surely be large(r) ! • There will be interferences with machine protection system that is also in the ‘commissioning’ phase. • Turn-around time is long, over 1 hour even in the best case. •  learning by trial and error costs a lot of time… • We need a good control system (applications !). • … • To minimize frustrations and keep a reasonable efficiency • (LEPI best was ~ 55 %), we should aim to have operational feedbacks… • ASARP !! Chamonix 03 / Presentation 7.5 / J. Wenninger

16. When should we get FBs into operation ? Orbit : • We should start using it at an early stage, before we ramp. Energy : • Required at latest when RF capture losses become too high. Q,Q’ : • We could survive without FB … but life will be difficult ! It is worth investigating this point in detail (LHC-OP WG ?). • We should aim for an operational Q PLL system ASARP ! This will make life  easier ! • TOTEM beam , single bunches : Must rely on AC-dipole and beam kicking + feed-forward. Beam-kicking incompatible with small TOTEM e ? Chamonix 03 / Presentation 7.5 / J. Wenninger

17. This is not LEP ! • LEP was commissioned 1989 and … • The tune feedback was operational in 1997, • A simple orbit feedback in 1995 (physics only). • …for operation of the LHC we cannot wait such a long time ! Chamonix 03 / Presentation 7.5 / J. Wenninger