Feedback Requirements for SASE FELS Henrik Loos, SLAC, Kyoto, Japan 1 1 Henrik Loos
Outline Stability requirements for SASE FELs Diagnostics for beam parameters Transverse: Beam position monitors Longitudinal: Bunch length/compression/arrival monitors, synchrotron radiation monitors Feedback implementations LCLS transverse feedback XFEL orbit IBFB LCLS longitudinal feedback FLASH longitudinal IBFBs Summary 2 2 Henrik Loos
SASE FEL Feedback Ensure electron beam quality for lasing Provide stable photon beam for users 100 MeV E, φ E, φ E, φ ~ GeV λ, Δt, Δx Injector Acc BC Acc BC Acc Undulator Users Energy (GeV) Wave length Und. length Bunch Charge Peak Current Gain length Beam size Rate (Hz) 13.6 1.5 Å 100m 0.25-1nC 3kA 3.5 m 8 1 Å 100m 0.3nC 2.5kA ~10 m 17.5 1 Å 130m 0.1-1nC 5kA 3.7 m 30 μm 120 35 μm 60 45 μm 10/ 5E6 3 3 Henrik Loos
SASE FEL Feedback Requirements Transverse requirements Undulator orbit x' < λ L G for efficient SASE L G ~ 3 10 m, λ ~ 1 Å x < 5 μrad over several L G Beam position x < σ/10 for stable photon beam β ~ 30 m, ε n ~ 1 μm x < 5 μm 4 LCLS example: Transverse jitter in undulator from leaked dispersion FEL Pulse Energy (mj) 3 2 1 Lasing rms width at 6.7 GeV σ = 90 μm 0-60 -40-20 0 20 40 60 80 100 Undulator Beam Position X (μm) 4 4 Henrik Loos
Feedback Requirements cont d Longitudinal requirements SASE process: ρ parameter ~ 10-4 Photon BW ~ ρ energy stability 10-4 Bunch compressor R 56 ~ 4 cm timing jitter Δt ~ R 56 ρ/c ~ 10s of fs Energy measurement R 16 ~ 10 cm R 16 ρ ~ 10 μm Energy in BC from position measurement in BC or from TOF measurement with beam arrival monitors Bandwidth requirements NC accelerator ~100 Hz rate Feedback stabilizes slow drifts SC accelerator bunch train MHz rate Intra Bunch FB required 5 5 Henrik Loos
LCLS Strip Line BPM Performance Strip line BPMs Continuous calibration with test pulse between beam triggers Beam synchronous data acquisition system at 120 Hz Noise level measurement Measure beam orbits at ~150 BPMs for 500 shots in main linac through 4 undulator 2 Average value for strip-line 3.5 μm, for RF cavity 250 nm at 250 pc 0 Noise rms (μm) 6 250 pc Strip Line BPM σ = 3 μm Stripline BPM RF BPM RF BPM σ = 300 nm E. Medvedko et al., BIW 2008, TUPTPF037 6 Noise rms (μm) 30 20 Strip Line BPM σ = 25 μm 10 20 pc RF BPM σ = 2 μm 0 0 500 1000 1500 Position (m) 6 Henrik Loos
LCLS Undulator RF Cavity BPMs Few micron beam orbit straightness in undulator required for FEL operation Sub-micron resolution met with RF cavity BPM design 11.4 GHz dipole cavity Reference cavity for normalization Calibration with beam signals Move supporting girder of undulator Induce known orbit oscillation upstream of undulator 7 7 Henrik Loos
RF BPMs at XFEL/Spring-8 Dipole mode cavity at 4.76 GHz + monopole cavity Shifted from main RF frequency to avoid dark current Measurements at SCSS test accelerator Position resolution < 200 nm Timing resolution from TM 010 cavity < 25 fs Position resolution Timing resolution H. Maesaka et al., DIPAC09, MOPD07 See also H. Maesaka et al., MOPE003 S. Matsubara et al., MOPE004 8 8 Henrik Loos
RF BPMs for X-FEL Test stand at FLASH 40 mm RF BPM for IBFB Based on Spring-8 design Frequency 3.3 GHz D. Noelle, BIW10, WECNB01 Low Q to resolve bunch train at 5 MHz 10 mm high precision version for undulator 40 mm version for IBFB Designed for 1 μm resolution See also B. Keil et al., MOPE064 9 9 Henrik Loos
LCLS Transverse Feedback Launch FB for each linac section Loops for transport line and undulator FB are independent of each other Decoupling by use of different time scales FB response matrix from online model GUN GUN Laser V 0 δ 0 σ z1 Steering Loop σ z2 BPMs CER detectors L0 L0 δ 1 δ 2 δ 3 ϕ 1 V 1 ϕ 2 V 2 V 3 L1 X L2 DL1 BC1 BC2 DL2 J. Wu et al., PAC 2009, WE5RFP046 L3 10 10 Henrik Loos
Undulator Feedback Performance Upstream LTU FB runs at 10 Hz Undulator FB slower with 1 Hz Horizontal jitter 13 μm / 2 μrad 30 40% larger than vertical due to 40 dispersion leakage 0-40 Residual jitter ~ 25% -80 of beam size Position (μm) Angle (μrad) 10 5 0-5 Beam at Undulator Entrance, 6 GeV σ x = 13 μm σ x' = 2 μrad Actual Correction 0 50 100 150 200 Time (s) 11 11 Henrik Loos
XFEL/PSI Intra-Bunch Orbit Feedback Use downstream BPMs for feedback loop Latency ~ 1 μs bunch spacing FPGA for feedback calculation Fast strip-line kicker for orbit correction Use upstream BPMs for calibration BPMs in undulator for slow feedback B. Keil et al., EPAC08, THPC123 12 12 Henrik Loos
LCLS Bunch Length Monitor Edge radiation from last dipole of each BC Integrated measurement sensitive from mm to 20 μm Block NIR radiation from bunching instability with filters 3% rms noise from correlation with bunch length dependent wake field energy loss in undulator Mesh Filter Pyro Detector Paraboloid Beam Splitter Beam BC2 Peak Current (A) 1700 1600 1500 1400 <I> = 1507 A σ I = 50 A 1300 Edge Radiation 0 5 10 15 20 25 30 Wake Energy Loss (MeV) 13 13 Henrik Loos
LCLS BLM Calibration BPM provides only signal related to bunch length Calibration with absolute measurement from transverse deflecting cavity Detector Signal (10 5 cts) 8 6 4 2 e σ z 2.44 m V(t) S-band β d RF streak Δψ 90 β s σ y 0 10 20 30 40 50 Bunch Length (μm) Empirical fit of signal to (σ z ) - 4/3 Use fit to calculate peak current 14 14 Henrik Loos
LCLS Longitudinal Feedback Cascaded FB at 5 Hz (Matlab implementation) Fixed energy gain in L2 & L3 klystrons Change global L2 phase Adjust L2 & L3 energy with several klystrons at opposite phases Feedback uses orthogonal actuators to separate energy gain and chirp of L2 GUN GUN Laser L0 L0 V 0 δ 0 ϕ 1 V 1 L1 X σ z1 δ 1 ϕ 2 V 2 L2 Steering Loop DL1 BC1 BC2 DL2 σ z2 δ 2 V 3 L3 BPMs CER detectors δ 3 15 15 Henrik Loos
Longitudinal Feedback Performance I Peak (A) δe/e (10-4 ) 40 0-40 1750 1500 1250 E Φ (mj) Beam energy/peak current, 6 GeV 4 2 σ δ = 11 x10-4 σ I = 98 A σ E = 0.25 mj 0 0 10 20 30 40 Time (s) See also F.-J. Decker et al., TUPE071 16 7% peak current jitter 6% X-ray pulse energy jitter (best 3%) Stability achieved over hrs Feedback controls enable bunch length & energy changes (few %) in 10s of seconds Operation soon at 120 Hz Fast orbit and energy/phase feedback in development Time-slot aware control for different 60 Hz phases 16 Henrik Loos
LCLS Phase Cavities Jitter between two cavities 15 fs Not used for e-beam FB Signal used for offline analysis Phase Cavity 2805 MHz Adjustable Attenuator Trigger Mixer X6 Multiplier 51 MHz 2856 MHz 16 Bit Digitizer ¼ Divider 119 MHz Phase Measurement Software Synchronize laser of user experiment to electron beam J. Byrd et al., MOOCRA03 476 MHz Reference J. Frisch et al., TUPE066 See also J. Byrd et al., TUPEA033 T. Ohshima et al., TUPEA030 17 17 Henrik Loos
FLASH Bunch Compression Monitor C. Behrens et al., MOPD090 Coherent diffraction radiation detector Radiator is metal screen with slit Optical radiation transport with GHz to THz bandwidth Signal from pyroelectric detector Fast detection resolves bunch train 18 18 Henrik Loos
FLASH Beam Arrival Monitor Laser clock via length stabilized fiber with 6 fs stability Beam signal from 4 button pick-ups Electro-optic modulator encodes beam signal on laser amplitude Fast sampling with 108 MHz ADC Operate at zero-crossing of amplitude modulation Delivers arrival time of each bunch in bunch train with < 10 fs resolution F. Loehl, TESLA-FEL2009-08 See also M. Bock et al., WEOCMH02 M. Bock et al., FEL09, WEPC66 19 19 Henrik Loos
Longitudinal Intra-bunch Feedback FPGA based controller board PID controller for amplitude correction from BAM signal Phase control from BCM signal Rapid change at head of bunch train from beam loading Latency of 30 μs due to SC RF Phase feedback F. Loehl et al., FEL08, THBAU02 Amplitude feedback F. Loehl et al., EPAC08, THPC158 20 20 Henrik Loos
FLASH Synchrotron Radiation Monitor A. Wilhelm et al., DIPAC09, TUPD43 SRM signal resolution Energy measurement with < 10-4 resolution ICCD for energy spread of single bunches Fast centroid readout with multi-anode PMT 14-bit ADC at 1 MHz for bunch train resolution C. Gerth et al., DIPAC09, TUPD22 21 21 Henrik Loos
FLASH Energy Feedback using SRM Correct stochastic and deterministic disturbances with a learning FF algorithm Effect of beam loading at head of bunch train minimized after a few iterations of the FF algorithm C. Gerth et al., DIPAC09, TUPD22 22 22 Henrik Loos
Summary Diagnostics available to meet resolution requirements for SASE FELs SASE FEL feedback systems achieve beam stability to do user experiments over many hours Optical synchronization schemes enable < 10 fs timing measurements and synchronization of user experiments Energy stability of ~ 10-3 still exceeds photon beam bandwidth 23 23 Henrik Loos
Acknowledgements Thanks to all the people working on X-ray laser facilities worldwide and to my colleagues from the LCLS commissioning team to make stable X-ray beams a reality 24 24 Henrik Loos