RF re-commissioning LHC season II

LHC RF re-commissioning for run 2

with planned operation at 6.5 TeV/c and 0.55 A DC current

P. Baudrenghien BE-RFwith contributions from J.E. Muller, E. Shaposhnikova and H. TimkoSeveral slides are copied from previous presentations:on the way we intended to commission the LHC (run 1) from a presentation at the LHCCWG June 2007on the first commissioning from a presentation in Sept 2008, after the original start-up week

LBOC July 29th, 2014

Period 1.1: Single bunch (or few bunches) capturedPeriod 1.2: ramping single bunch(es) to 6.5 TeVPeriod 2: 25 ns, 450 GeVPeriod 3: 50 ns, 6.5 TeVPeriod 4: 5-20 ns, 450 GeVPeriod 5: 25 ns, 6.5 TeV

What was done between Sept 2008 and Dec 2009

Period 1.1: Single bunch (or few bunches) capturedA1First Turn PilotJune 5, 2007LHCCWG meeting5GoalInject pilot and centre first turnRF sub-goal: Label buckets (numerology and cogging) Adjust front end gains to see PU signals (APW and BPM)

We do not anticipate changes but it must be checkedJune 5, 2007LHCCWG meeting6Strategy (1)Generate the LHC injection kick (RF).Observe kick+beam. (BT). They adjust their delay (or ours?) to kick the beam.Get the beam to make a few turns (OP)Adjust the gain of the RF front end of the Beam Position and Beam Phase module + time alignment and Frev marker (RF). Dedicated RF 8 hours. To do that: Set the Observation memory to trigger on the Beam In timing.Observe the PU signal (APW in Beam Phase module and BPM in Beam Position module).Adjust gain/attenuationAlign the signals from the 2 inputs (OK for APW and D,S from BPM. More difficult for cavity sum as the beam induced voltage will be very small with pilot. Coarse adjustment must be done without beam)Adjust the Frev marker (offset in memory addressing) so that marker points to bucket 1

Sector testsNo change in hdw -> 4 hours should be OKJune 5, 2007LHCCWG meeting7Strategy (2)Adjust delay in the Frev sent to the beam dump. Observe our signal at the beam dump location. Compare to bunch position. Adjust. (BT). The abort gap ends just before the passage of bucket 1 (proposed convention)Adjust delay in the 40 MHz/Frev clock received by the experiments (EXP)Should be quick but must be done for MP. Was done during sector test in 2008Should be quick. Done during sector test 2008A2 Capture and Circulating Pilot June 5, 2007LHCCWG meeting9GoalCapture and centre the closed orbit. Get pilots to collide at the right point (cogging)RF sub-goal:Commission phase loop and synchro loop.CaptureAdjust relative positions of the 2 rings for collisions in IPs (cogging)

Still relevantSept 17, 2008LLRF commissioning10Adjust Inj Freq (2008) Sept 11, 14:00 Synchro module not working anymore.17:30 Synchro modules replaced with lab versionTest Synchro module Step Response -> OK21:00 Beam on ring 2, 100 turnsKeep Synchro loop On, Phase loop OFF, Radial loop OFF, RF OFFMeasure the Beam/RF phase slip turn after turn, using Beam Phase Module with RF OFFRe-adjust inj freq from 400.788933 MHz to 400.788 963 MHz

Beam Phase module DAC outFiltered PU signalBunch by bunch phasePhase avgShould be quick. Magnetic centre should not have changed muchSept 17, 2008LLRF commissioning11Capture (2008)Sept 12, 00:30 Cavities are now on same ref as low-level. Switch phase loop ON at injection. Transient lasts for 10 rev period as expected.Trajectory on MR very goodPhase loop can be ON before inj as we have a threshold on PU signal

MR of one of the very first capture beams. T. Bohl and U. Wehrle

Bunch avg phase at inj (phase loop on)Took 3 h from RF ON to captured beam in 2008. Should not be much longer in 2015June 5, 2007LHCCWG meeting12Adjust Synchro Loop dynamicsWith both loops ON, measure the synchro loop step responseAdjust synchro loop gain and phase advance to fine tune the responseAs this depends on the RF voltage via synchrotron freq we want to do it at different voltages. But it is not urgent. Optimization of the LLRF loops dynamics was never done. This is now more important if we ever want to manipulate the RF loops for blow-up in physics, as this likely requires the opening of the phase loop with beam.

June 5, 2007LHCCWG meeting13Check phasing of cavity sumNow try to capture the beam with one cavity at the timeObserve synchro loop phase discri output Dfsync after transientIf non-zero, fine-tune the delay in Cavity Sum for that cavity8 hours RF

Must be done as we have replaced Mod1B2June 5, 2007LHCCWG meeting14Align the two ringsThis stage has been advanced compared to the original OP scenarioGet 1 circulating pilot in bucket 1 of each ringMeasure the collision point. This requires an acquisition in a PU that sees both rings (first BI, then EXP).Adjust one ring with respect to the second to get collision in IP1 (?), using Frev prog 2 B# of the second ring. See diagram on page 124 hours OP,BI and RFDone in 2009. Should be quick in 2015. But must be done early in the start-up

what was done in 2010 @ 3.5 TeV

Period 1.2: Ramping. Longitudinal stability. Blow-upJuly 29, 2014LBOC meeting16

May 15th, 2010. First attempt to ramp single bunch nominal. UNSTABLE

May 17th, 2010. We inject with phase loop open and 45 degrees phase error -> large blow-up during capture (1.4 1.6 ns). First successful ramp. STABLE

May 28th, 2010. Blow-up in the SPS (1.6 ns). STABLE

June 15th, 2010. Blow-up in the LHC ramp. STABLE16Longitudinal blow-upis essential!Several types of blow-up were proposed during run1Promising alternative methods were tried at start-up 2012, but after initial problems, were discarded due to lack of time for optimizingTheoretical studies (diffusion models) and simulations (PyHEADTAIL tracking code) have been developed during LS1 to include the effect of controlled RF noiseWe wish to optimize the process in 2015We also want to set-up longitudinal profile flattening with RF modulation

LBOC meeting17July 29, 201417Loss of Landau damping and impedanceExtrapolating from results in 4 TeV, the min. long. emittance is 1.32 eVs @ 6.5 TeV with 1.15E11 pp bunch (Evian 2014)We want to check it by ramping a few bunches, nominal intensity but different long. emittances, or different intensities and similar emittances. (CERN-ATS-Note-2013-001 MD)

That will set the minimum voltage required during physicsLBOC meeting18July 29, 2014

18Bunch lengthening at 6.5 /eVEmittance growth is caused by IBS and RF noiseDamping comes from synchrotron radiationThe net effect was predicted to be bunch shortening at 6.5 TeV (20% reduction in length over 6 hours,Evian2014)

If so, we may need to apply longitudinal emittance blow-up or bunch flattening, on regular intervals at flat topTo be identified asap

July 29, 2014LBOC meeting19

Reproduced from J. Tuckmantel, LHC Project Report 819, Synchrotron Radiation Damping in LHC and Longitudinal Bunch Shape, 2005 19

25 ns scrubbing beam @ 450 GeV/c

Period 2

Potential problemsCapture losses will increase sharply with bunch intensity. The SPS can produce bunches with nominal longitudinal parameters (0.5 eVs, 1.5 ns) up to 1.15E11 p but the bunch length increases quickly above that intensity (1.65 ns with 1.35E11 p). That will create large capture losses.

Heating of parasitic resonators (HOMs) must be monitored carefully. The beam spectrum will have lines spaced by 40 MHz (used to be 20 MHz). For narrow-band resonator the power scales as the square of the total beam current -> large increaseLBOC meeting21July 29, 2014Sunglasses?21

Potential problems (contd)The ACS HOMs have indicated less heating than expected, so far. We have monitoring of both temperature and power. Do we need interlocks?CBI: we do not anticipate stability problems from the cavity impedance at the fundamental. How about the HOMs?LBOC meeting22July 29, 2014Scrubbing May 19th, 2011, 1092b, 1.3E14 total

22

50 ns ramp-up @ 6.5 TeV/c

Period 350 ns intensity ramp-upMonitoring of HOM powerCheck stability. If we have a few ramps to 4 TeV/c, we can check that the stability is comparable to 2012 (no large modification to the machine impedance)LBOC meeting24July 29, 201424

5-20 ns scrubbing beam @ 450 GeV/c

Period 45-20 ns scrubbingThese beams have not been accelerated in the SPS yet. What longitudinal parameters can we expect at injection into the LHC?We anticipate much problems with capture losses. 1.6E11 p per pair!On the hardware side the beam phase loop must be optimized for 5-20nsDiagnostics tools are being developed to monitor the bunch-by-bunch stable phase (energy loss caused by e-cloud)LBOC meeting26July 29, 2014Sunglasses?26

25 ns intensity ramp-up @ 6.5 TeV/c

Period 525 ns intensity ramp-upWe will reach record beam current at high energy (0.55 ADC @ 6.5 TeV vs. 0.35 A DC @ 4 TeV in 2012)-> potential Coupled-Bunch instability (CBI)We want to measure the CBI stability margin by a few test ramps with reduced target bunch lengthIf blow-up does not work for a fill, the high intensity/energy beam will be unstable. That happened in 2012, and the beam was dumped by an interlock on heating. Do we need a specific interlock on bunch length? Do we need dedicated hardware?

LBOC meeting28July 29, 20142825 ns intensity ramp-up (contd)What is the optimal voltage in physics? In 2012 we have observed a saturation effect in bunch lengthening suggesting limitation of momentum aperture. Lower voltage may be better. To be tested with a few physics fills at different voltagesBunch shaping may have to be applied at regular intervals in physics as the shape will return to Gaussian. Delicate as we must limit the risk of debunching

LBOC meeting29July 29, 2014

29A word on ControlsThe RF controls became robust after the implementation of basic macro-operations in the sequencer (begin 2010)All RF CPUs (RIO3) are being replaced by MEN A20 running Linux. FESA classes have been re-compiled but remain in FESA 2.10Very limited upgrades -> We do not anticipate big problemsMany features will be re-commissioned during dry runs and sector testsLBOC meeting30July 29, 201430

Conclusions

Except for the replacement of one full cryomodule (four cavities) the RF upgrade has been limited and we do not anticipate hardware problems with the re-commissioningEarly in the restart we wish to measure the minimum longitudinal emittance that preserves Landau damping at 6.5 TeVWe also want to optimize the longitudinal blow-up in the ramp using single bunch or a few bunches per beam. This manipulation has a big effect on the bunch spectrum and should best be optimized (and understood) before moving to multi-bunchWith multi-bunch the challenges are the total beam current (heating and CBI) and the capture losses due to the large SPS bunch length (25 ns and 5-20 ns operation). For the later sunglasses at injection will make life easier. Heating must be monitored closely. Test ramps with reduced target bunch length would identify the CBI limitLBOC meeting32July 29, 2014Thank you for your attention32