1 Consequences of RF system failures during LHC beam commissioning T.Linnecar AB-RF

January 2005Chamonix XIV2 Acknowledgements This talk gives preliminary ideas from: Olivier Brunner Edmond Ciapala Elena Shaposhnikova Joachim Tuckmantel

January 2005Chamonix XIV3 Outline The RF system – main parameters Possible failures Starting conditions Hardware protection Beam issues Summary

January 2005Chamonix XIV4 Main Parameters of RF system Each beam: 8 SC cavities in 2 cryogenic modules Klystron power source for each cavity 300kW nominal Each cavity can supply 0-2MV(3MV at low beam intensity) Beam is captured, accelerated and stored by this 400 MHz system. (The 200 MHz capture system will be installed later for high intensities) Each cavity has a controllable tuner and coupling loop There is a strong short delay RF feedback and a 1-turn feedback around each cavity to reduce it’s impedance f 0 = Hz :tuning range ~100 kHz R/Q = 44  Q ext : ~15,000 to ~200,000 V a : 0 to 3MV / cavity (depending on beam current, energy)

January 2005Chamonix XIV5 Possible Failures (1/2) Increasing seriousness: Loss of control of a single cavity Loop electronics problem, RF power system failure (but not HT) Loss of several cavities Multiple failures of above type, HT problem – four cavities in one module lost Removal of one module Beam-vacuum leak, insulation vacuum leak, power or HOM coupler failure. Can be assimilated to loss of 4 cavities Etc.

January 2005Chamonix XIV6 Possible Failures (2/2) For the present talk we consider the problem of the loss of control of up to four cavities – what are the consequences? Two questions to try and answer for commissioning: Can we continue running if failure with beam present? Can we inject with this hardware failure? Two types of problems: Hardware protection (don’t increase seriousness) Effect on the beam (will it survive?)

January 2005Chamonix XIV7 Starting Conditions – RF system Multi-variable situation: analyse optimum situation and then look at changes.  Minimise transient peak power and mean power at high beam loading. Fully compensate phase modulation at injection to minimise capture losses. Optimise beam lifetime in store. 1/2 detuning, f det = ¼ f 0 R/Q I RF / V a Q ext = 2 V a / (I RF R/Q) V a = 1MV/cavity at injection, 2 MV/cavity at 7 TeV (At injection / top energy we can go to higher voltages if we have lower beam currents)

January 2005Chamonix XIV8 Starting Conditions –beams considered BeamMN b I rf A f det kHz Q ext I rf A f det kHz Q ext Injection7 Tev 25ns Nominal , ,000 ½ Nominal , , ns Nominal , ,000 ½ Nominal , ,000 Initial Comm. 156 bunches , ,515, bunches ,800, ,545,000 I rf = 2 I DC F b. Inj: F b =0.75, V a =1MV 7TeV: F b =0.87, V a =2MV Time

January 2005Chamonix XIV9 Hardware protection - Limits Beam induced over-voltage in cavity – arcing, surface damage We condition to 8 MV/m, V a  3 MV – take this as limit Beam induced power goes via circulator to load Rating 300 kW – take this as limit. Conservative, but non-perfect matching can lead to locally high field values.

January 2005Chamonix XIV10 Hardware protection - Transient effects If we lose drive to a cavity the stored energy in the cavity decays at the resonant frequency while the beam induced voltage builds up at f rf = hf rev 25 ns nominal, Q ext =120,000, f det =2kHz, R/Q=44, I rf =1.01A, V a =2MV The transient is fast - peak voltage decaying to steady-state value. Both values decrease as f det increases and Q ext decreases. Also true for the power in the load. Envelope of cavity voltage after trip

January 2005Chamonix XIV11 Hardware Protection – V cav and P load - nominal f det kHz P load kW V a kV time to 90% s Peak V a kV ns nominal beam Q ext =120,000 R/Q=44 I rf =1.01A V b =2MV Energy 7 TeV N.B. Opt. f det = 2.23 kHz  With 2.23kHZ we have ~ 1MW taken from beam.  Kills the load.  Also must be provided by the remaining 7 cavities, 143kW/cav  They will probably trip. (Cascade effect)  Serious danger with nominal beam! Dump the beam!

January 2005Chamonix XIV12 Hardware Protection – V cav and P load I RF A F det kHz Q ext P load kW V a kV (SS) V a Pk kV 25ns ½ nominal 450 GeV , TeV , , ns nominal 450 GeV , TeV , bunches450 GeV , TeV ,  Cavity protected up to half-nominal by slight change in de-tuning and Q ext, (possible because power is available).  Coupler and tuner must remain under control!  Beam dump interlock should come from V cav and P load  The power is supplied by the remaining cavities Commissioning beams:

January 2005Chamonix XIV13 Beam issues – Transient effects Cavities trip with beam: Transient on voltage (increased power from other cavities) Phase loop takes care of dipole oscillation Quadrupole oscillation: Quadrupole/Hereward damping could help if necessary. 25ns ½ nominal, could survive 2(3) tripped cavities at injection. 25ns ½ nominal beam InjectionTop VrfExtra power /cavity Vrf leftExtra power /cavity MVkWMVkW Start off off off off At low beam currents V rf decreases proportionally to the number of cavities off

January 2005Chamonix XIV14 Beam issues - Instability  For instability it is the difference in impedance at the nf rev + mf s sidebands that matters.  Cavity exactly on tune – no instability.  Off-tune the impedance increases. Difference impedance for one cavity with detuning Q ext =20000 R/Q=44 f det in steps of 2kHz Vert. -1 to 1 M Horiz. n=-5 to n=5

January 2005Chamonix XIV15 Beam issues – instability thresholds Estimation of thresholds: E. Shaposhnikova, Longitudinal beam parameters during acceleration in the LHC, LHC Project Note 242. R thr   3/2 V ¼ h 9/4 / (E 3/4 I 0 ) Instability frequency is ~ 400 MHz Assume blow-up to 2.5 eVs during acceleration Threshold is weakly dependent on V  For nominal beam R thr ~ 0.8 M ( 1/I 0 ) With 200 MHz system R thr is a factor 6 lower. If we stay at 1eVs, the threshold impedance at 7TeV drops factor 4 Assume negligible coherent freq. shift from Im(Z/n)

January 2005Chamonix XIV16 Beam issues – Instability with cavity off Injection7 TeV Mode N f det kHz 1 off 2 off 3 off4 off f det kHz 1 off2 off3 off4 off 25 ns nominal ns ½ nominal  Nominal, R thr ~ 0.8 M, ½ nominal R thr ~ 1.6 M  With standard detuning: May survive with nominal beam with 2 cavs off. ½ nominal beam stable with one module off. (Also true for 1eVs at 7 TeV)

January 2005Chamonix XIV17 Beam issues – max. detuning for instability Impedance k f det kHz 1 off2 off3 off4 off For half-nominal beam the threshold is ~ 1.6 M. The impedance with detuning is:  If the detuning is kept <3 kHz we can have 1 module (4 cavities) off.  When the cavity is off, we rely on the tuning setting – this should be ~  0.75   5 mbar.  The instability is always n=1, mode 1 damper? Note that the impedance contribution from active cavities is negligible due to the RF feedback.

January 2005Chamonix XIV18 Beam issues – acceleration and 7 TeV  With V total = 4 MV,  s rises to ~ 8 at ~3.5 TeV, bucket area is ~ 2.1 eVs.  Can accelerate with >4MV. V total MV  b eVs  b ns Bucket eVs  With lower voltage at 7 TeV  The bucket is more full  possibly faster blow-up if RF noise  The bunch is longer  reduced luminosity.  Could push voltage on remaining cavities at low beam current  For new store could have smaller emittance if intra-beam scattering allows and no instability, R thr   3/2 V ¼ At 7 Tev

January 2005Chamonix XIV19 Summary 1/3  Trip of a cavity or cavities with beam present: Can probably survive with 2(3) cavities tripping at ½ nominal 25ns or 75ns nominal (power required from other cavities main concern). At half these beam currents we can survive a trip of one module. To do this we should work at or near half de-tuning and slightly below optimum Q ext. When control of a cavity is lost with beam-in, the tuner and coupling should be blocked – RF interlock.

January 2005Chamonix XIV20 Summary 2/3  A new injection We can inject and work with up to 4 cavities off, i.e. one module, up to half nominal intensity if we retain control of the tuner and coupling. For this, the tuning is set to ~ half de-tuning and Q ext is made as small as possible on the dead cavities. Can work up to ½ nominal with reduced performance in store with only one module in the ring The frequency stability of the passive cavity must be  0.75 kHz.

January 2005Chamonix XIV21 Summary 3/3  General The beam-dump interlock to protect the RF hardware should be derived from the cavity voltage and the circulator load power. The beam dump should not be driven from an RF trip The beam-dump (~3-turn reaction time) will not prevent arcing at high beam-currents (nominal) but will limit the duration of arc, similarly will limit energy dissipated in the load. The rest of the machine will be protected (instabilities – beam loss) by non-RF measurements. A mode n=1, m=1 damper may be interesting These results are preliminary – it is possible that other combinations of parameters may be better. RF interlock policy is the area we will now attack. e.g slow interlocks where drifts in parameters can be used to adjust parameters of cavities to help prevent switch-off or minimise effects of switch off.