Beam Diagnostics, Low Level RF and Feedback for Room Temperature FELs Josef Frisch Pohang, March 14, 2011
Room Temperature / Superconducting Very different pulse structures RT: single bunch or short bursts SC: Long bursts or CW RT: ~100 bunches / second SC: ~10,000 bunches / second Data collection and processing RT machines usually accumulate data for an entire train before processing SC machines process within bunch train SC accelerator looks a lot like a ring from a diagnostics, LLRF, and feedback perspective Diagnostics and LLRF sensors, actuators and analog systems can be similar but Digital systems and software are very different! 8ms Room Temperature Single bunch Multi bunch Train <400ns total Superconducting 100ms Multi bunch train, 10ms 3MHz bunch rate CW operation, MHz bunch rate Slide 2
Types of Beam Diagnostics Diagnostics for feedback. Continuous operation High Reliability Beam Position Monitors, Bunch Length Monitors Diagnostics for tuning Profile Monitors, Wire Scanners, Transverse Cavities Diagnostics for machine physics Transverse cavities, X-ray spectrometers, COTR spectrometers, etc. Focus of this talk Slide 3
Beam Position Monitors: Striplines Well understood technology 5um RMS resolution at 250pC Use conventional designs unless there are special requirements Requires good RF engineering! Very different from storage ring BPMS Single shot broadband, not narrowband Low average data rate 3 out of 4 channels shown Variable Attenuator Variable Attenuator Digitizer cross coupling Band pass Amplifier Amplifier Variable Attenuator Variable Attenuator Digitizer Calibrate Signal Variable Attenuator Variable Attenuator Digitizer Slide 4
Beam Position Monitors: Cavities Better resolution than striplines Theoretical resolution 1nm at 1nC Practical limitations (mechanical mounting, dynamic range etc) limit typical resolution to 100nm. 10nm demonstrated at ATF@ More complex and expensive than striplines Axis Cavity Variable Attenuator Mixer Digitizer Local Oscillator Vector division in software to get position Reference Cavity Variable Attenuator Mixer Digitizer Slide 5
Bunch Length Monitors Absolute bunch length measurement requires transverse cavities Resolution limited to ~10fs Spectral bunch length monitors can provide approximate bunch lengths Measurements are invasive, can t be used for feedback Relative bunch length monitors based on total radiated power in some band. Can use diffraction, synchrotron, or edge radiation. Pyroelectric detectors have sufficient sensitivity, single pulse response Bunch length monitor using edge radiation and pyro detectors Mm-wave filters used to reduce sensitivity to beam horns Bunch length monitor using high frequency RF diodes (BC1) Slide 6
Low Level RF Systems Control and measure the high power RF systems Waveform shaping for SLED and other applications Correct drifts and nonlinearities in high power systems Used for feedback from both RF and Beam information Different for RT and SC Accelerators RT: typically no feedback within pulse Simple digital control electronics and software Low average data rate High RF bandwidth ~10MHz SC: Need feedback within pulse to lock to high Q SC cavities Typically use programmable logic (Xilinx) for real-time feedback High average data throughput during pulse. Modest RF bandwidth ~100 KHz SC type systems may not be appropriate for room temperature RF. Slide 7
Generic LLRF System RF Reference I / Q Modulator KW Driver Amplifier High power RF Fast waveform generator Low power pickoff Mixer Beam Data Software: Feedback, RF and Beam data Software: decode phase and amplitude Transient Digitizer Slide 8
LLRF: Medium Power RF control In some accelerators (including SLAC) phase and amplitude control is done at ~KW power Cost tradeoff needs careful evaluation LLRF Drive 10KW Driver Amplifier KW power drive line Other stations Fast control of a group of stations Slow control of differences ΔΦ Mechanical ΔA LLRF Readout High power RF Slide 9
Feedback Functions Reduce Noise and Drift Will always amplify high frequency noise Design depends on noise spectrum Often low bandwidth is OK At LCLS / SLAC not much noise above 1Hz that isn t random shot to shot. Isolate parts of the accelerator Upstream changes corrected by feedbacks Improves tuning Reduces effects of klystron cycling etc. Provides orthogonal controls Can change beam compression without changing energy Valuable for understanding performance vs. single parameters Slide 10
LCLS Longitudinal Feedbacks Energy and bunch length feedbacks Act as 6 independent loops In principal possible to combine into a single loop, but the matricies depend on the operating conditions (compression) gun V 0 d 0 s z1 s z2 1 V 1 d 1 d 2 2 V 2 d 3 V 3 Linac X Linac Linac DL1 BC1 BC2 DL2 Slide 11
LCLS Transverse Feedbacks Multiple loops, each with multiple BPMS, 2 X/Y corrector pairs Most low rate / low gain, LTU runs at high rate Undulator uses cavity BPMS to stabilize orbit 1 to 10 Beam Position Monitors (BPMs) gun 2 or 4 corrector magnets example Injector L2 L3 LTU Undulator LTU Feedback : stabilizes beam for jitter frequencies < 10Hz @ 120Hz rep-rate Slide 12
120Hz Operation and timeslots US power line frequency is 60Hz LCLS operates at 120Hz Modulator voltages, ambient magnetic fields, etc. have 2 states. Use main and difference feedbacks Feedback Sum Fast Actuator BPM Data Switch Average Difference Feedback Data Switch Slide 13
Feedback Software LCLS_I Basic algorithms operate as originally designed Software layer to convert LLRF amplitude and phase -> in-phase and out-of-phase components Feedback uses in-phase to control energy, out-of-phase to control compression Most effort on feedback development was for handing errors, beam loss and other special conditions Feedbacks prototyped in Matlab: allows rapid modification of code Now being converted to real-time system for faster performance Feedbacks operate independently Ideally would combing into a single more complex feedback. Difficult not clear if this is worth the effort. Slide 14
LCLS_I Automated Operation LINAC Energy Management: Adjust quad strengths when klystron complement changes Matching Adjust quads based on measured beam parameters Energy Change Adjust klystron complement and magnets to change beam energy Works well for small energy changes Large changes often require additional tuning. Charge changes 250pc, 150pc, 80pc, 40pc, 20pc Need configuration changes Require different source laser aperture, bunch length monitor Automation reduces chance of operators missing something Photon energy change by users being tested Slide 15
Comments Diagnostics, LLRF and Feedbacks form a single integrated system Superconducting and Room Temperature systems have very different requirements Some technology may transfer, some will not. LCLS Feedback system worked nearly as designed Lots of technical issues. General concept probably a good starting point for the design of other RT FELs. Slide 16