Presentation on theme: "Andrew Hutton Director of Operations Jefferson Lab"— Presentation transcript:
1 Andrew Hutton Director of Operations Jefferson Lab THE SCIENCE BEHIND THE CEBAF ACCELERATORAndrew HuttonDirector of OperationsJefferson Lab
2 Outline Review of the basic accelerator layout Injector Linacs making electronscreating electron bunchesLinacsHow does RF acceleration workSetting up the RFSecrets of the ArcsEmittance and Optics notionsSpecial features
3 Summary Accelerator Description Beam performance objectives:≥ 5.7 GeV, 200 µA, CW3 simultaneous beams, independent energy and current adjustmentBeam polarization≥ 75%Parity qualityDesign concept: recirculating, superconducting CW LinacCW for high beam qualityRF superconductivity for efficiency5-pass beam re-circulation gives multi-energy operationIsochronous, achromatic beam recirculation arcs
6 Making Electrons Photo-electric Gun Shine laser light on a semiconductor waferPhoto-electric effect kicks out an electronsame effect as solar cells which make electricity from the sunBecause the laser can be pulsed, electron beam is also pulsedUse three separate lasers, one for each Hallprovides greater flexibilityCathode material is strained Gallium ArsenideLaser light of 780nm gives weakly polarized beams ~35%but high Quantum Efficiency (QE~1%)Laser light of 840 nm gives strongly polarized beam ~75%but low QE~0.2%
7 Polarized photoinjector 2 identical horizontal guns installed in 1998Gun 2Oct 2000toJan 2001Gun 3Feb 2001toMar 2002Gun service required once per year.Both guns provide high polarization (>70%).
8 Source perspective A 100 keV beam from either gun is deflected 15° by a magnet to acommon pre-accelerator beamline.Two laser tables straddle thebeamline and provide a directoptical path to the cathode.
9 Diode Laser Diode lasers – like in your CD player Requires additional amplifier to obtain ~100 mW of lightHighly reliablePulse length δ ~20 psButAmplifier does not fully switch off between pulsesCreates “bleed-through”Have tried to pulse amplifier with some success – still a problemWe always probably use diode laser for Hall BPresently using diode laser for Hall A
10 Ti-Sapphire Laser High power without additional amplifier Switches off cleanly – unmeasurable bleed-throughEarly laser had noise on the beamsLed to trips of the safety systemA Ti-Sapphire laser from Time–Bandwidth Inc. in Switzerland (Tiger) was ordered for the G0 experiment at 31.2 MHzUses a different, proprietary method for locking to RF frequencyHas operated flawlesslyCan create pulses with δ from 20ps to 70psWe have ordered another Tiger laser for 499MHz operationExpect good performanceWill be used for future, high current Hall A experiments
11 Dynamic laser configuration Re-re-re-re-configuration. . .3 end-stations makes for a dynamic physics program which requires that the laser table be configurable for beam qualities:Intensity (power)Polarization (wavelength)RF (1497, 499, )Parity (Independent)G0FutureHAPPEx2
12 Making BunchesBeam from 100 keV gun sent through 499 MHz chopper cavity (one third of accelerating frequency)Rotates beam in circle of ~1.5 cm radiusSlits at 240°, 0° and 120° degrees allow bunches of electrons to passSlits are individually controlled to regulate currents for Halls A, B & CThe three beams are recombined by another 499 MHz chopper cavityCreates 110 picosecond bunchHall A Hall B
13 Bleed-ThroughTo reduce “bleed-through”, Hall B uses about 20 µA beam current from the cathode and rejects all but a few nA at the slitsImproves rejection of leakage from Hall A and Hall C beamsButMakes Hall B current very sensitive to position movement at the slitHave installed a fast feedback to stabilize positionBPMs read position of the average currents for Halls A, B & CDisplaces low current beam if all the beams are not exactly coincidentCannot stabilize against laser spot movements
14 Making Bunches (cont) Beam then goes through buncher section Decelerates front of bunch, accelerates back of bunchAfter a drift space, bunch is compressed (shorter) – 5 picosecondbut energy spread in the bunch is increasedBeam then goes through capture which continues processFinal bunch length – 1 picosecond (3/10 of a millimeter)Phases of Injector Rf must be stable to within ~1 psCan achieve this for short periods of timeBest in summer and winterInjector very unstable in spring and fall – sun-up and sun-down
15 Measuring Bunch Length Measure the time of arrival of the bunch at a high frequency cavity (5.988 GHz)downstream of the buncherMove the phases of the chopper cavities and re-measure the time of arrivalPlot the time of arrival as a function of the phaseEach point is the center of gravity of the beamBy moving the center of gravity, the result mimics the behavior of particles distributed along the bunchThis measurement is accurate to about 15 femtoseconds
18 How does RF acceleration work? Imagine a surfer riding a waveGet on the wave at right time and right speed - accelerationGet on the wave at wrong time or wrong speed, deceleration, wipe-outA good surfer will speed up to catch the wave and will then start to move across the wave to avoid overtaking the wave and wiping outMatching the phase velocity and the group velocityNow imagine an electron traveling in a radio-frequency (RF) waveIf it arrives at exactly the right time it will gain energyIf it arrives at the wrong time it will gain less energyBut If it is traveling near the speed of light, it will not speed up or slow down – it will gain or lose mass as it gains energyNo wipe -out, but not the right energy at the end
19 How does RF acceleration work? cont. RF acceleration comes from multiple, independent klystrons, each feeding one 5-cell cavityThe trick is to ensure that as a bunch arrives at each cavity, the fields are correct to continue accelerationThe waves (fields) in each cavity have to be in phasePhases have to be set up correctlyPhases have to stay correctBoth of these are a challengeEach klystron is fed from the same phase reference, butmust be accurate to a few picosecondsklystrons are up to half a mile away from the sourceThermal problems, stability problems in repeaters
20 Superconducting Cavities Use superconducting niobium cavities to create the RF fields for accelerationVery low losses on the cavity wallscavity resonance very sharp (Qext ~ 106)sensitive to vibration (microphonics)Cooled by superfluid liquid 2 K so heat transfer is easyno bubbles to vibrate the cavityHelium temperature (pressure) changes the tuningkeeping all of the cavities in tune requires automated softwareCEBAF Acceleration System330 cavities in 41 1/4 cryomodules, installed and functionalGradient limit and Q twice as good as specification
21 CEBAF SRF CavitiesCEBAF 5-cell cavities operate at 1497 MHz with an active length of 50 cm eachThere are eight cavities per cryomodule
22 Linac CryomodulesCEBAF has 42¼ cryomodules with a total active length of 169 meters
23 Superconducting Cavity Treatment Gradient specification of CEBAF cavities was 5 MV/mAverage gradient of superconducting cavities as installed was 7.3 MV/mTwo approaches to improve performance of cavities,Helium processing and waveguide vacuum processingCarried out in situAim to reduce field emissionAverage gradient of superconducting cavities is now 7.76 MV/mLimit is established with CW beam under standard operating conditions
25 FSD TripsFast Shut Down (FSD) trips are triggered by RF arcs and protect the SRF cavities from arc damage¿Caused by charging up of ceramic widows?¿Caused by “three dimensional“ gas discharges?Strongly dependent on accelerating gradientWeakly dependent on total linac beam currentPushing energy to:5.7 GeV 10% hit in availability (just acceptable)6 GeV 20% hit in availability (unacceptable)
26 Improving RF arc trip rate Installed >50 stub tuners to improve matchReduces klystron power for same beam powerHelium processed all cryomodules (some twice)Reduces electron emission in cavitiesImproved algorithm for calculating set pointsOptimize for gradient, current, cryogenic load, etcProblem is inherent, getting close to limitThis year, will try to reduce time to reset trip
27 FSD Trip Rate Versus Energy October 99 – June 01
28 Ponderomotive ForcePonderomotive force of RF fields tends to lower cavity resonant frequencyRF fields have energy, tends to push outwards, makes cavity biggerCavity tuner changes to maintain correct operating frequencyIn a few cavities (so far)If cavity trips off at high gradient, resulting frequency shift is so largeCavity goes out of resonanceCannot be switched back onThis is the most serious problem for RF control of the new cryomodules for 12 GeV
29 Linac RF OperationOperation of the linacs requires sophisticated high-level softwareAutomated Cavity TuningMake it resonate at exactly the right frequencyAutomated Cavity phasingMake all the cavities resonate in phaseLinac Energy ManagementSet up exactly the right energyKRESTSet up exactly the right phase for each cavityMOMODMaintain the right phase for each linac
30 Automated Cavity Tuning Sweep Mode (Exact tuning)The cavity frequency is modulated over a range of ± 200 Hz in steps of 5 Hz.Find the typical response of a resonant system to harmonic excitationThe measured de-tuning angle as a function of modulating frequency iscompared to the predicted curve to determine the resonance frequencyas well as the phase offset Autotrack (locks cavity on frequency)Uses the phase offset determined above to keep the cavity tuned to the operating frequency
31 Linac Auto-PhasingPurpose: Precise cavity phasing is achieved by maximizing the Linac energy.Arc is used to measure energy changesThe higher the energy, the heavier it is, the harder it is to bend so it travels on the outside of the bendPhase of an individual cavity is changed by ±30°.Initial beam position and beam position changes are recorded.Crest phase is found from0: initial phase setting, y0: beam position at 0, y±: beam position at = 0 ± ∆Reproducibility better than 1°, Phasing time 2 min/cavity
32 Linac Energy Management (LEM) For a given energy at the end of North or South Linac and available cavities:Optimizes the cavity gradient distribution using individual cavity characteristicsCalculates energy profile along the linacsCalculates Linac quad values consistent with calculated energy profileDownloads and sets RF, Quads (including hysteresis) and skew quadsRequired inputMaximum gradient permitted for each cavityList of available cavitiesIntegrated into the maintenance logMost useful number – energy overhead available (usually 6 – 10%)
33 Global Procedures Krest (intrusive) Momod (parasitic) Modulate the phase of each cavityObserve the change in energy at a BPM in the Arc where there is dispersion (particles with different energies have different orbits)Move the cavity phase until energy does not change with modulationRepeated for each cavityEach cryomodule is done by hand at start-upMomod (parasitic)Modulate the phase of the RF phase reference for each Linac using two different frequencies ~800 Hz during accelerator operationLinac is now on crest
34 What is Emittance?The beam size in an accelerator can be separated into two parts:The emittance or phase space which is a property of the beamThe beta functions or R matrix elements which describe the focusingIn an accelerator, the emittance is defined by the gun characteristics and cannot be improved later – but it can be degradedFor CEBAF, emittance is the extent of the beam in six dimensions:X, X’, Y, Y’, δE, δLAs the beam is accelerated, the transverse emittance should be adiabatically damped (reduced proportionally to the energy)As the beam is accelerated the transverse energy of the beam stays constant but the longitudinal energy increases. It is like pulling on a rubber band – it gets thinner
35 Basic Accelerator Optics DIPOLESmagnets that bend the beam (usually horizontally, sometimes vertically and sometimes at an angle) They define the shape of the machineHigher energy electrons will be bent less (like a prism in optics)The edges of the beam have fields that are not uniform, so they focus as well (like aberrations due to non-flat prism faces)QUADRUPOLESmagnets that focus the beam (like a lens in optics)main difference with an optical lens is that if the lens focuses in the horizontal, it will defocus in the vertical and vice versa“Strong focusing” if the quadrupoles are sufficiently strong (and close together) overall they will focus in horizontal and verticalthe result is rather like a periscope in optics and is stable
36 The R MatrixThe R Matrix relates the output angles and positions from a beamline to the input angles and positions.Xout R11 R12 R13 R14 XinX'out = R21 R22 R23 R24 X'inYout R31 R32 R33 R34 YinY'out R41 R42 R43 R44 Y'inThe matrix for an uncoupled beamline would have the cross-terms zeroIf the beam is kicked horizontally, it will oscillate in the horizontal plane butshould never show a vertical oscillation
37 CEBAF Optics LINACS ARCS A Linac (LINear ACcelerator) is straight – no bendsIn CEBAF, electrons of very different energies travel togetherThere is a regular array of equi-spaced quadrupolesThey focus strongly for the lowest energy pass, more weakly for the highest pass – but still “strong” focusingARCSAt the end of the Linacs the beams of different energies are separated by the “spreaders”There is then a regular circular Arc section (actually five of them)At the end of the Arcs the beams of different energies are brought together by the “recombiner”
38 Achromatic & Isochronous Bunches must be transported around the arc and all of the electrons must arrive at the other end at the same place and timeAchromatic – electrons of different energies arrive at the same placeFrom the Greek “a” without, “khroma” – colorIsochronous – electrons of different energies take the same timeFrom the Greek “isos” – equal, “chronos” – timeThe Arc was designed to be both achromatic and Isochronous
39 Principle of Isochronicity RecombinerLinacSpreaderArcArcSpreaderLinacRecombinerIsochronous if path length difference in Arcs equalspath length difference in spreaders and recombiners
40 Types of Emittance Increase FilamentationIn the longitudinal plane, RF focusing in circular machines causes particles of different energies to rotate around the RF bucket at different speeds. This can cause an effect like an egg beater resulting in real emittance dilution. This effect is not relevant for CEBAFLongitudinal variation of EmittanceIf the front of the beam is not in the same position as the back of the beam, the projection of the beam on a screen is apparently enlarged (think of the beam as a banana – the projection is thicker than the cross-section of the banana). This effect is usually not a problem at CEBAF.Emittance ProjectionIf the beam is strongly X-Y coupled, the projections of the phase space onto the visible axes X, X’, Y, Y’ can all be increased – this has been a problem at CEBAF.
41 Coupled BeamsBeam coupling refers to a coupling between horizontal and vertical oscillations in the beamUnder normal (uncoupled) conditions, the horizontal and vertical motions of the beam are independentNote that a quadrupole has the effect of focusing in one plane and defocusing in the other plane so there is a correspondence between the two planesWhen the beams are coupled, an oscillation in one plane couples into the other plane, like two coupled oscillatorsThis can lead to an apparent increase in the beam emittanceNote that Louiville’s theorem says that phase space (emittance) is conserved, so the increase is only apparent – but very real!
42 Uncoupled Emittances X’ X In the absence of coupling, the product of the projectionsof the phase space area on the X and X’ axes is a constant
43 Coupled Emittances X’ X In the presence of coupling, the product of the projectionsof the phase space area on the X and X’ axes is a never aconstant and is usually much larger than when uncoupled
44 Causes of Coupling Point coupling in the Injector Most likely candidates are the counter-wound solenoids, the cryounit and the cryomodulesUniform Coupling in the LinacsProduced by skew fields in the SRF cavities due to asymmetry in the HOM couplersCorrected by the skew quadrupolesPoint Coupling in the ArcsProduced by mis-steered beams going through fringe fields in dipolesMost serious in the spreaders and recombinersGoal is to reduce coupling as much as possible
45 Summary Keeping the accelerator tuned up is a demanding, delicate task The operators usually have a Batchelor degree in Physics (some are studying for their Master’s)They are trained on the job for six months to become an OperatorIt takes another 2-3 years to become Crew ChiefThey try incredibly hard to deliver quality beam to UsersThere are new capabilities all of the timeWe are all learning as we goPlease – give the Operators a break and complain to me if you are unhappyThat’s my job!