Report

Paul Scherrer Institut The European XFEL Intra Bunch Train Feedback Boris Keil For the PSI E-XFEL Team Paul Scherrer Institut Boris Keil, PSI DEELS Workshop 2014 DEELS Workshop 2014 13.5.14 E-XFEL IBFB Overview 1 IBFB Daisy-Chain 2 of BPM Units IBFB Upstream BPM Pickups LINAC IBFB Kicker Magnets (Horizont. & Vertical) H1 V1 H2 V2 IBFB Downstream BPM Pickups SASE 2 e-beam Daisy-Chain 1 of BPM Units - - - - - - - - - - - Analog Signals (Coax Cables) - - - - - - - - - - SASE 1 IBFB Electronics Digital Signals (Duplex Fiber Optic Cables) • Low-latency (~1μs) beam position correction upstream of beam distribution. • Can kick each bunch individually, using feedback + feed-forward algorithm. • Uses undulator BPM data (latency 5-10μs) for fine-tuning of undulator orbit (to correct kicks between IBFB and undulators: Vibrations, distribution kicker, ...). Boris Keil, PSI DEELS Workshop 2014 13.5.14 Transverse Perturbations 2 • IBFB kickers should provide enough kick to correct perturbations, plus reserve. • IBFB removes perturbations, but also adds noise to the beam (dominated by BPMs): Noise should not have negative impact on FEL performance → Low-noise BPMs (goal: <1μm RMS). Pickups: 3.3GHz cavity, same as TL. • Feedback loop latency <1.5μs expected to be sufficient. Spurious dispersion and 3% chirp Nonlinear dispersion and 3% chirp Spurious dispersion and 1e-4 energy jitter Nonlinear dispersion and 1e-4 energy jitter Wake fields Kicker Drift Kicker Jitter Quad Motion Power Supply Jitter Dispersion jitter Total kick xeff [μm] Max. Freq. Plane 15 15 0.5 0.15 25 0 1 28 12.6 2.5 1 kHz 1 kHz 1 kHz 1 kHz 5 MHz 1 kHz 5 MHz 10 Hz 10 Hz 10 Hz x/y x x/y x x/y x x x/y x/y x/y Perturbation Type repetitive repetitive random random repetitive repetitive random random random random X Kick [μrad] Y Kick [μrad] ±0.5 ±0.5 ±0.01 ±0.003 ±0.8 ±0 ±0 ±1.0 ±0.4 ±0.1 ±3.3 ±0.5 ±0 ±0.01 ±0 ±0.8 ±0 ±0.03 ±1.0 ±0.4 ±0.1 ±2.8 *Worst-case estimate (DESY), 30m beta function at kicker & BPM, adding of peak values. Boris Keil, PSI DEELS Workshop 2014 13.5.14 IBFB Kicker Magnet 3 • 50 Ohms stripline kicker (picture shows cut / only half). • Kicker design by PSI (based on CTF3/Daphne design by F. Marcellini et al., INFN Frascati), supported by DESY (wakefield simulations, M. Dohlus). • Tapered 2m long strips. Flexible RF • Wakefield simulations: Kicker vessel needs no taper. feedthrough • Prototype built by company COMEB, RF test successful. • DESY uses modified version (aperture, ...) for dump kickers. Ceramic spacers & RF feedthroughs allow thermal expansion of strip relative to vessel (bakeout, tolerances, ...) DESY standard steel flanges Boris Keil, PSI Aluminum vessel and strips (low weight, easy to fabricate) DEELS Workshop 2014 13.5.14 IBFB Kicker: S-Parameters Boris Keil, PSI DEELS Workshop 2014 4 13.5.14 IBFB Kicker: Diff. Impedance Boris Keil, PSI DEELS Workshop 2014 5 13.5.14 Kicker Positions & Beam Optics Baseline: 4 Kickers of 2m length for IBFB. 6 Reserved space for upgrade: Double number of kickers and max. kick Dump kickers Boris Keil, PSI DEELS Workshop 2014 13.5.14 IBFB Kickers: RF Power Amps 7 • Commercial amplifiers from Company TOMCO (class AB solid state). • Improved at request of PSI: Redundant power supply & amp modules to maximize MTBF. • Two amplifiers purchased & tested extensively: Meet PSI specifications. • Kick: > ±4μrad baseline (4 kickers), > ±8μrad upgrade (8 kickers). Two Amplifiers Boris Keil, PSI DEELS Workshop 2014 13.5.14 IBFB Kickers: RF Power Amps 8 TOMCO guarantees 3kW pulse power, but amp reached 6kW! Prototype test at PSI: IBFB will most likely use 18MHz amplitude-modulated sine or square wave. Boris Keil, PSI DEELS Workshop 2014 13.5.14 IBFB Kickers: RF Power Amps 9 Droop of kick voltage over bunch train (thermal effects in MOSFETs, ...): IBFB digital electronics will compensate droop Boris Keil, PSI DEELS Workshop 2014 13.5.14 IBFB: Electronics Topology BPM6r BPM6y BPM6x BPM5r BPM5y BPM5x undulator BPMs BPM4r BPM4y BPM3r BPM3y BPM3x BPM2r BPM2y BPM2x BPM1r BPM1y BPM1x BPM4x downstream BPMs upstream BPMs 10 RFFE1 RFFE2 RFFE3 RFFE4 RFFE5 RFFE6 6xADC 16-bit 6xADC 16-bit 6xADC 16-bit 6xADC 16-bit 6xADC 16-bit 6xADC 16-bit FPGA1 FPGA2 FPGA3 FPGA4 FPGA5 FPGA6 P0 Boris Keil, PSI PDC 32GFLOPS DSP 6xADC 16-bit 4xDAC 14-bit K1x K2x K2y K2y FPGA8 K1x Pfor K1y Pfor K2x Pfor K2y Pfor FPGA7 Ebeam GPAC3 P0 P0 Feedback/Feedforward algorithm: Same FPGA board as BPMs, but with 0.5-1GSPS DAC mezzanine to generate kicker waveforms RIO P0 GPAC2 RIO RIO RIO RIO RIO GPAC1 to kicker amplifiers DEELS Workshop 2014 13.5.14 kicker signal from y plane kicker signal from x plane ... Data Acquisition from/to control system RIO Link Timing Control x1 x’2 Position & Angle Calculation BPM4r BPM4x BPM3r BPM3x BPM4r BPM4x BPM3r BPM3x BPM2r BPM2x Position & Angle Calculation x-y Plane Decoupling Position & Angle Calculation x2 x3 kicker signal + DDR2 SDRAM BPM1r Kicker Linearization ADC, position, angle, control signals, etc. QDR2 SRAM BPM1x K1x Pfor K2x Pfor 11 K2x K1x IBFB: Algorithm Control & Status Registers Processor Local Bus (PLB) Adaptive Feed Forward Table Ebeam Feedback Kicker Control x’4 x4 Lattice Transfer Matrices Adaptive Feed Forward Algorithm x5 x’5 x6 • Ultra-fast feedback removes random perturbations, e.g. beam offset of whole bunch train due to mechanical vibrations etc. • Adaptive feed-forward corrects reproducible perturbations that are the same for each bunch train (or change very slowly). • IBFB can use same FPGA carrier board as BPMs. Present version (Xilinx Virtex-5 FPGA, PowerPC) sufficient, new version (Artix-7/Kintex-7 FPGAs + DSP) under development, will simplify development of more complex algorithms for future operating modes. Boris Keil, PSI DEELS Workshop 2014 13.5.14 IBFB: Cavity BPM Pickups Prototype at SwissFEL Injector Test Facility Transfer Line Cavity BPM • 3.3GHz, 40.5mm aperture. • Used for: Transverse intra-train feedback, energy measurements, launch jitter control & correction (energy, BAM, linac entry, …), optics measurements, … 12 255mm Similar to undulator type, slightly less resolution (~20%). Main differences: ~16x more angle signal (→ align 16x better), cavity spacing (→ crosstalk). Frequency (both resonators) 3.3GHz Loaded Q (both resonators, desired mode) ~70 Q (uncoupled modes) typ. 200-300 Sensitivity 2.5V/(nC*mm) Thermal noise (lossless cables & electronics, …) 65nm @ 20pC Angle signal (90° to position signal. Cause: Misalignment) ~16mm * dx/dz Boris Keil, PSI DEELS Workshop 2014 D. Lipka DESY 13.5.14 IBFB: Cavity BPM Electronics New: 63dB range, 0.5dB steps RFFE MBU Crate: Removable fan tray, redundant main power supply, ... Differential coax cabling from RFFE to ADCs • I/Q downconversion to baseband. • Active temperature stabilization (several sensors + heaters). • Works with or without external trigger & ref. clock. Boris Keil, PSI DEELS Workshop 2014 13 DOOCS & Timing Interface (SFP/Optical, PCIe/Ethernet /..., up to 6.5Gbps) 13.5.14 ADC Sample Clock Phase Feedback 14 Digital ADC sampling clock phase alignment loop • Eliminates phase drift effects • Retains maximal S/N ratio • Monitors possible reference signal malfunctions & beam arrival time changes Present algorithm: Uses just one ADC sample at top to calculate beam position. Boris Keil, PSI DEELS Workshop 2014 13.5.14 RFFE: Nominal vs. Measured Gain Boris Keil, PSI DEELS Workshop 2014 15 13.5.14 Gain Dependence of Phase Delay Boris Keil, PSI DEELS Workshop 2014 16 13.5.14 Cavity BPM ElectronicsTemp. Drift 17 Temperature drift scales with beam offset. Active temperature stabilization active: <100nm/°C drift at 1mm offset (0.01%/°C) Boris Keil, PSI DEELS Workshop 2014 13.5.14 GUI For Automated Lab Calibration 18 • Presently using commercial RF generator (pulsed) for automated lab calibration (gain & phase delay for each attenuator setting; IQ imbalance, ...). • Developing low-cost test/calibration system (external "customers", ...). Boris Keil, PSI DEELS Workshop 2014 13.5.14 Position Calculation in BPM FPGA Boris Keil, PSI DEELS Workshop 2014 19 13.5.14 SwissFEL BPM Test Area 20 Correlation of 3 E-XFEL Undulator Cavity BPMs Sampled RFFE IQ Signals Only top sample used (so far ...), plus baseline subtraction Histogram (X1+X3)/2 – X2 See IBIC’12, TUPA27, M. Stadler et al. Boris Keil, PSI DEELS Workshop 2014 13.5.14 IBFB: Cavity BPM Performance 21 Position Noise (RMS, 1 Bunch) • Undulator cavity (Ø=10mm): ~11μm @ 2pC (±5mm range) <0.5μm @ 100-1000pC (±1mm range) • Transfer line cavity (Ø=40.5mm): ~1μm @ 100-1000pC (±1mm range) 2x improvement feasible by digital removal of angle signal (15x bigger than for undulator BPMs) – work in progress ... 20mm offset at 1nC: 50V signal! RFFE may need input protection via attenuator (4x worse low-charge resolution), or extra protection circuit (to be developed for IBFB) Charge Measurement RMS Noise (1 Bunch) • Undulator cavity (Ø=10mm): <0.06% @ 100-1000pC <60fC @ 100pC <10fC @ 2pC Boris Keil, PSI DEELS Workshop 2014 13.5.14 IBFB Status 22 IBFB BPMs (Will Dominate IBFB Performance ...) • Using standard E-XFEL cavity BPM electronics (maybe with external RFFE input protection circuit (1nC & big beam offsets ...), necessity being investigated). IBFB (Non-BPM) Electronics Hardware • Can use BPM FPGA carrier board also for IBFB signal processing. • DAC mezzanine to driver kicker amps under development. IBFB Firmware/Software • Feedback/Feed-forward algorithm & feedback network via multi-gigabit fiber optic links to be implemented (re-using building blocks from BPM firmware/software). Boris Keil, PSI DEELS Workshop 2014 13.5.14 Team & Acknowledgements 23 PSI: • M. Stadler (Cavity BPM RF front-end) • M. Roggli, M. Gloor (ADC/DAC Mezzanine) • R. Baldinger, D. Engeler (FPGA carrier board HW) • G. Marinkovic, W. Koprek (Firmware & software) • C. Beard, F. Marcellini, M. Rohrer, D. Treyer, (IBFB kicker magnet & RF power amps) DESY: • S. Vilcins, D. Lipka, D. Nölle (Cavity BPM pickup) • M. Dohlus (Kicker wakefield simulations) • N. Golubeva, W. Balandin, W. Decking (Magnet lattice & beam optics) ... and all other supporters at PSI & DESY/E-XFEL Boris Keil, PSI DEELS Workshop 2014 13.5.14 Paul Scherrer Institut Thank you for your attention! Boris Keil, PSI DEELS Workshop 2014 13.5.14