Accelerator Division ESS AD Technical Note ESS/AD/0013 Rihua Zeng LLRF Overview and Development in Other Labs 29 March 2011
""#$%&'(&)'*+,-.'&'/012',3),%34'("+56 Rihua Zeng March 29, 2011 Outline "#$%&'(&)'*$+,-$.'&'/012',3$),$%34'($+567777777777777777777777777777777777777777777777777777777777777777777777777777777777777777778$ 9,3(0-:;3)0,777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777<$.'&'/012',3$),$%34'($+56 77777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777=$ >+5/'$8?$>4'$2+),$1+(+2'3'(6$+,-$20-:/'6$)21/'2',3'-$+3$#@ABCADACEF@"G777777777H$ >+5/'$<?$>4'$20-:/'6$)21/'2',3'-$),$#'(2)/+5CGI"DCJDCKL""B@77777777777777777777777777M$ A02'$"#$"'N:)('2',3$;0,6)-'(+3)0,6 77777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777O$ 966:'6$),$A:5$K0-:/'6 7777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777O$ IP1'()',;'6$+,-$'660,67777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777Q$ A02'$IP1'()',;'$),$9BIF 777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777R$ A:22+(S777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777R$ @;T,0*/'-U'2',36777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777 8V$ "'W'(',;'77777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777 8V$ #@AB7777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777 8V$ ADA7777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777 88$ EF@"G77777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777 88$ #'(2)/+5 77777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777 8<$ GI"D 777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777 8=$ JD7777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777 8=$ KL""B@77777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777777 8=$ 1
7,3(0-893)0, Generally, LLRF system can be broken into sub-modules by functionality as shown in the figure below. The LLRF System can run well only if all sub-modules have the good performances. First of all, one should identify the tasks and challenges in each module. By the experience of other labs, for example, there are some: phase noise in Frequency Generation, drift in Phase Reference Distribution, noise and drift in Frequency Conversion, Lorentz detuning and microphonics in Cavity Field and Resonance Control, nonlinearity and noise in RF Control Hardware, and automated operation of large scale LLRF stations. ESS will suffer most of them, and probably some more: spoke cavity, longer pulse, higher current and larger scale. Therefore, it would be helpful to firstly investigate the solutions that have been taken in other labs and learn from them. Below I will introduce the development of LLRF in other labs. 2
.'&'/012',3),%34'("+56 LLRF R&D in FLASH is evolved from the development in TTF (Tesla Test Facility) RF digital control system started about 15 years ago, and will continue to be developed for the new European XFEL in DESY. Their achievements and expertise in pulsed SRF control are well known in the LLRF community, and more important they are well documented in TESLA Reports. http://flash.desy.de/reports_publications/tesla_reports/. SNS LLRF system has been going through the accelerator commissioning and operation over 5 year, and therefore the experience and lessons achieved will be good reference for us. There are a lot in common between SNS and ESS. JPARC Linac LLRF System s experience during their 181MeV linac development (RFQ, Bunch, Rebunch Cavities, DTL, SDTL) and 400MeV linac upgrade (adding ACS) is good reference for our normal-conducting cavity s LLRF R&D. These three lab s LLRF parameters and sub modules implemented are listed in table1. The rests are the labs whose LLRF systems are not very similar with ESS (CERN), or under development (Fermilab, LBNL), or at the beginning (MYRRHA). The sub modules realized in these labs are given in table 2. CERN LHC has 8 superconducting cavities (signal cell, CW mode, ring) per ring. Although different with ESS (pulse, multi-cell, linac), it is still valuable to look through their RF feed back, Klystron Polar loop and the Tuner loop. Moreover, there is Open Hardware Repository at CERN, a place for collaborating on open hardware designs, much valuable information. http://www.ohwr.org/. And there is also other digital LLRF developing at CERN, such as LEIR, Linac 4 LLRF [4~7, CERN] In Fermilab, there are going on different kinds of the LLRF system for ILC, Project X, HITS, NML [13, Fermilab], based on the hardware platform MFC, newly developed in Fermilab (33 channels). And the spoke cavities might be taken in Project X (already having the test stand for spoke cavities). LBNL has developed the FPGA programming for SNS (many new ideas in algorithms), and developed the 4 th generation evaluation LLRF board to test concepts for ILC with low noise (Open verilog codes and hardware designs http://recycle.lbl.gov/~ldoolitt/llrf/). Then the next generation board design and fabrication will be carried out. MYRRHA has a similar linac with ESS (same frequencies, normal conducting, spoke cavity, low beta, high beat), and desires more extreme reliability, with some LLRF prototypes under development in Orsay. And more interest is their proposal of fault tolerance LLRF. The other more information about the LLRF developments worldwide can be found in the LLRF Workshop held every two years since 2005: LLRF Workshop 2005, http://indico.cern.ch/conferencetimetable.py?confid=a050#20051010 LLRF Workshop 2007, http://neutrons.ornl.gov/workshops/llrf2007/presentations.shtml LLRF Workshop 2009, http://www-conf.kek.jp/llrf09/llrf-intro.html LLRF Workshop 2011, it is coming in October. 3
:+5/';<:4'2+),1+(+2'3'(6+,-20-8/'6)21/'2',3'-+3$"=>?@>A>@BC=#D FLASH SNS JPARC Main Linac Length ~200m[1,2] 311m [1] 248m[1] LLRF Stations 6[3] 96 24 Beam Pulse 800 us [4] 1000us 500us RF Pulse 1300us 1300us[1~3] 650us Beam Current (average dc) 9 ma 26mA 50mA Repetition Rate 5Hz~10Hz 60Hz 50Hz Bunch Frequency 1MHz~40kHz Beam Power N/A 1.4MW(on target) 1.0MW[1,2] Beam Energy 1.2Gev (electron) 1.0Gev 400MeV[3] Energy Spread <510-5 <±0.15%, 0.033%(rms) <±0.1% Arrival Time Jitter Required RF Amplitude/Phase Stability <30fs 0.01%, 0.1 [6,7] (achieved with FB only) ±0.5 ±0.5 ±1%, ±1 [1, 4, 5] Klystron Work Point K2~20% of 5MW, K3~62% of 5MW, K4~53% of 10MW, K5~60% of 5MW. [8][4] RFQ, DTL: 2.5MW CCL: 5MW SRF linac: 0.55MW (Klystron peak power) RFQ: 0.53MW, saturation 3.0MW DTL: 5.7 MW for 3 tanks SDTL: 23.6 MW (total) ACS: 43.8 MW (total) 324MHz, 24 in total, 3.0/2.5MW 972MHz, 26 in total, 3.0/2.5 MW Gradient 25MV/m [9] DTL: 1.58~2.777(ave. EoT) CCL: 1.98~2.143 Med beta: 13.4~16.4 (E0) High beta: 17.9~24.4 DTL: 2.5-2.9 MV/m SDTL: 2.5-3.7 ACS: 4.2-4.3MV/m Cavity Beta 1 0.61/0.81(Med/High) Load Q Cavity Half Bandwidth Expected Microphonics 3e6 DTL: 23554~30863 CCL: 12309~13975 Med beta: 7.3e05 High beta: 7.0e05 (variation 20%) DTL, SDTL: ~20000 ACS: ~8000 [6,7] 220Hz Med beta, High beta: 550, 575Hz 8.1kHz(DTL, SDTL), 61kHz(ACS) 3-7Hz [10] ±100Hz(six sigma, amplitude limit) Not important Lorentz Detuning 1 Hz/(MV/m) 2 Med beta, 2300Hz; High, 1000Hz <470Hz (with piezo, max.) (Without piezo, klystron power limit) Not important Dominant Mechanical modes 280(20) Hz 160/230/2000 Hz [10,7] N/A R/Q 1036[11] Med, High beta: 220-440, 170-570; Cavities Per Klystron RF Frequency 8, 16, 1(RF gun) 1 2[1,8,9] 1.3GHz, 3.9GHz[12] (3rd harmonic cavities) 402.5MHz, 805MHz 324, 976MHz[10,11] IF Frequency 250kHz, 54MHz 50MHz 12MHz Sample Frequency 1MHz and 81MHz[13] both IQ and non-iq 40MHz (non-iq) 48MHz (IQ) 4
Frequency Conversion Downconvet and Vector Modulator New: low-noise 9/54MHz IF Multichannel Downconverter [14] Down/ Up convert in one board [Drift problem later] New: Temperature controlled board [5,6] Downconvert, Vector Modulator [10~14] Hardware VME Crates (ATCA based system will be taken for future XFEL): 1. DSP, ADC, DAC as a VME card 2. Latest generation (SimconDSP) [Virtex II Pro Xilinx FPGA+DSP, 10channel 14bit 105MS/s ADCs] 3. SPARC VME CPUs module running Solaris OS 3. A function generator with VME interface [6,7,15] VXI Crates: [2,3,4] 1. Field Control module (FPGA Xilinx XC2S150) 2. Slot-zero controller (IOC) running vxworks system 3. High Power RF protection module 4. Timing Module 5. Utility Module Compact PCI [10~14] 1. DSP and FPGA (14bit ADC, DAC) 2. General purpose Pentium CPU module with Widows OS system, ACU-128-1, PentiumM 1.6GHz 3. RF and Clock Generator Module 4. Mixer and I/Q modulator module 5. Control I/O FPGA: Vertex-E XCV 600E. One for the real time FB, another for tuner control Cavity Field Control 1. IQ Sampling; Calibrations; Vector-sum; Kalman filter; P gain control; adding adaptive feed forward; FB P gain ~100 2. Klystron linearization; Work near the performance limit 3. Upgrade: New learning feedforward algorithm. Beam load compensation, new feedback controller; [6,7,13, 3~5] 1. SISO real-time PI control; [7~9] 2. 1KByte for set point curve Feedback gain 50~100 for SRF cavity, ~10 for normal cavity. [13] 3. CORDIC-based phasor rotator; AFF for cavity filling (8-Kbyte buffer), the other AFF for beam loading compensation (8K) 4. PID type AFF for proton [20] [10~14] 1. IQ sampling, Calibration, LPF, PI, Feedforward (behind FB switch), IQ output control (offset, limit). Block RAM (after LPF, to fill the I/Q read table) communicates with DSP 2. FB gain ~5, instability [11,15] 3. Feedforward for cavity filling [5] 4. Beam chopped compensation, macro beam compensation [16,10] Cavity Resonance Control 1. Two piezo, one for control, the other for sensor 2. Feed forward and find proper excitation signal for piezo tuner. 3. Cavity tuning is determined from forward, reflected and probe signal 4. Reduce detuning to less than a tenth of bandwidth 5. Microphonics control [16~23,13] The Piezo actuator is designed as part of the cavity mechanical tuning system, and becomes the weakest link of the mechanical system [10,11] 1. Normal conducting cavity pine tuner 2. 3 DTLs and 15 SDTLs controlled by feedback through a cavity tuner. 3. The I/Q components of klystron (after circulator) are measured, stored in DSP. 4. Once the tuning error (cavity to klystron) is larger than a set value, the DSP starts to control the cavity tuner [14] 5. F0 setting. Improved way to adjust tuning position [18] Diagnostics 1.Measurement and Calibration of reflect/forward power; 2. Calibration of vector sum; Beam based cavity gradient calibration; 3. Decay curve measurement; parameters calculation [6,7,13] 1. Registers map: 2 bytes 512 12(PRO I&Q, FW I&Q, REF I&Q, Decay I&Q, DAC I&Q, DIAG I&Q) 2. From firmware to EPICS directly, not via Labview 3. Interlocks. [12~13] 1. Set tables update every 1us during pulse. Measured data stored in FPGA Memory. DSP for data processing and communication (FW/REF/PRO) [10~12] 2. Detuning ("#) detected from phase decay curve, then "# used for FF filling. Variable FB gains [5] 3. Set Values of exponential pattern [13] 4. Q value calculation 5. Interlock in other module [17] Control system Server DOOCS (C++) 1. Sun SPARC platforms and Linux 2. Operation is automated by DOOCS finite state machine Server 3. The data accessed by DOOCS API (Matlab, Labview) 4. DSP state and Start/Reconnect; Load and Start program; Firmware control 5. Feed forward table upgrade according to beam current varies [6,7,13,24] EPICS; 1. Automated operation of each and all FCMs 2. One button starts up LLRF system: auto-run sequencer ramps up cavity field in open loop. Auto-run sequencer maintains cavity frequency. Auto-close loop [12~14] EPICS; PLC [17], touch Panel 1. Field control, high power protection, analog and status monitor and klystron 2. PLC is used as the main system controller. PLC will be locally operated by a touch panel on PLC LAN and remotely by an EPICS Operator Interface on EPICS LAN 3. PLC control through the control I/O board 4. Practical operation, the half number of gain margin (Kp=4~5), is considered to be stable [11] Frequency Generation 1. MO system. By phasing locking 81MHz VCXO to 9MHz OCXO to get the best phase stability. Time jitter: 78fs (integrating measurement bandwidth from 10Hz to 10MHz) 2. All other frequencies come from 81 MHz signal by division and multiplication [25] 1. MO provides 6 outputs at 2.5, 10, 352.5, 402.5, 755 and 805 MHz (directly synthesized from a single 10MHz low noise SC-cut quartz crystal oscillator), by Wenzel Associates in Austin, TX, USA 2. Output signal power +20dBm 3. The MO is located in the klystron gallery in between the DTL (402.5MHz) and CCL (805MHz) sections [15] 1. 12MHz master oscillator at the central room and distributed this frequency to each LLRF station; But later is changed to 312MHz due to temperature drift 2. Other frequencies are generated in LLRF Local station y VCXO with PLL synchronizing 312MHz [18,4] Phase Reference Distribution 1. Short term (phase noise): 1ms 0.1ps 0.05 100ms 0.3ps 0.15 2. Long term (drift): 1s 1ps 0.467 1hour 2ps 0.936 1day 10ps 4.679 3. A temperature-stabilized coaxial line is employed for RF power distribution. Stabilization of ±0.5 C was applied to improve long-term phase stability 4. Optic fiber used for monitoring drifts along coaxial line [25] 1. Reference RF transport ±0.3 C 2. RF reference Lines are 3-1/8"rigid copper coaxial lines with temperature and pressure regulated 3. Temperature zones are maintained at 100±0.1 F 4. RF line phase stability is ±0.1 C between adjacent cavities and ±2.0 C the whole linac [15,13] 1. Goal of Phase stability of reference: ±0.3 at 972MHz 2. PSOF (phase stabilized optical fiber, 0.4ppm/ C, 0.2 /300m/ C@972MHz) 3. Temperature controlled by cooling water (Temp. stability ±0.1 C) to control temperature of the fiber to be ±0.5 C 4. O/E in master oscillator; E/O in each LLRF station; jitter and temperature dependency from O/E to E/O should be considered [18,4] 5
:+5/'E<:4'20-8/'6)21/'2',3'-),$'(2)/+5@DF#A@"GA"@HI##?= Cavities/klystron Fermilab CERN LBNL MYRRHA 8(In one crymodule) 1(spoke cavity) 1[1~3] Evaluation board testing new concepts [1,2] CW mode [1~2] RF Frequency 1300, 325, 201.25MHz [1~3] 400MHz(CW) 1.3GHz 352, 704MHz IF Frequency 13MHz 20MHz 55.981MHz 10MHz[3,4] Sample Freq. 54.17, 56.33MHz(non-IQ)[4,5] 80MHz(IQ) 77.751MHz(non-IQ) 40MHz(IQ) Frequency Conversion 8 channel down conversion upconvert (IF IQ drive). [5] New: 96 channels receive, 12 boards; 8 channel/board, built on aluminum plate with heaters and sensors regulated by temperature controller. [6] FB gain upto 1000, test? [7,8] Downconvert and IQ modulator Analogue modulator and demodulator Downconvert and upconvert integrated into the FPGA board; Compact but some problems in LO distribution; [3,4] Drift, channel isolation? Downconvert, IQ vector modulator Hardware VXI crates: [8,9] 1.MFC board (Altera Cyclone II FPGA. 4 8-channel 12 bit 65MHz ADC, 2 dual channel 14bit 260Hz DAC; another channel with 14bit 105MHz ADC used for fast Klystron feedback) 2. Slot0 CPU Host Two VME creates: (Cavity control) 1. High gain RF feedback 2.A tuner loop module 3. A klystron Ripple loop 4.Conditioning system USB based (no crates): All clock distribution, up/down convert, ADC/DAC/FPGA are in one board, four RF/IF input, LTC2249 ADC, XC3S 1000 FPGA. [3~5] NI Labview PXI crates customer fabricated FPGA board Cavity Field Control 1.Serial data to parallel data. 2.Digital downconvert, Vector sum 3.Klystron linearization 4. Feedback, feedforward, fast klystron loop 5. Reference signals for beam and cavity phase 6.Digital upconvert to IF 7. Drift calibration [3,5,7~9] 1. Digital feedback paralleled with analogue feedback for transient (low delay, 10ns) 2. Feedback gain must be changed to keep impedance constant 3. Feedforward for bunch; 4. The klystron polar loop, klystron limiter Non IQ, Calibration, Cordic, Filters, DC reject, Rotation, Integrator, Bit-serial arithmetic; hard core or soft core CPU on chip.[5~8] Fault tolerance control in case of a RF cavity failure. 1. Fast detect RF fault and trigger beam shut down. 2. New correcting field and phase set points update; 3. Failed cavity quickly detuned, 4. Beam reinjection when reached steady state, recovery at max 1s. [3~5] Cavity Resonance Control 1. Slow and piezo tuner for 1.3GHz SRF CCII cavity [4] 2. Identify the mechanical resonance (165Hz) 3. Optimize the voltage and delay of excite signal to piezo 4. Compensation reduce detuning 275Hz to 20Hz [10] 5. Microphonics control for 3.9GHz cavity [11] 6.Spoke cavity freq. track [5] 1.Step motor control, based phase difference between cavity drive and probe 2.Digital demodulator (IQ sample), CIC filter 3.It can track minute fluctuations; with 25Hz resolution (0.11deg. @400MHz) N/A Fast cold tuning system for microphonics and Lorentz detuning, simulation.[5~7] Quick detuned the failed cavity by 1kHz and more slow but bigger detune with stepper motor. And the adjacent cavity is retuned. (20deg. 50Hz, 20% in amp.)[3] Diagnostics 1. Cosine/sine table; Gain table; Set point table; FIFO; diagnostics buffer used by DSP / CPU; 2. Data transferred to the SDRAM. Both FPGA and DSP have SDRAM for waveform, data storage; [7~9] 3.Vacc, Q0, Ql measurement. [5] 1.Automatic measurements for cavity tune and transient response and so on... in FPGA and DSP 2.Tune command sweeps the tuner through its range, measuring and storing values in buffer. System on chip, CPU communicates with SDRAM, use network to communicate with external devices [5~7] 1. Fast diagnostics to detect the failure 2. Fast update the setpoints in fault recovery 3. Calibration cavity field and phase, etc. 4. Other dedicated diagnostics, interlocks [4] Control System Server EPICS, DOOCS; 1.Drives and software build in CERN's Front End Software Architecture (FESA) 2. The memory map is used to build the driver and application 3. Remote diagnostics 4. Labview/Matlab application USB to a PC; Network communicate based afterwards, UDP protocol developing; [5~7] LABVIEW windows (prototype testing) Frequency Generation 1.10 MHz crystal oscillator from Wenzel Associates used as phase reference, 1300 MHz dielectric resonator oscillator (DRO) from Poseidon Scientific Instruments locked on crystal reference via PLL 2. All other frequencies are divided by 1.3GHz[12] 1. Direct digital synthesizer generates a 400MHz reference 2. A master 400MHz is generated by VCXO, sent to the cavity controller 3. VCXO is loop locked on DDS output 4. LO 380MHz; All clocks are phase locked Only LO distributed over coax; Clock distribution on the board [1] 704.4, 694.4, 40MHz [3] Phase Reference Distribution N/A The two Master 400MHz signals are sent to cavity controller by fiber optic links; A fs level sync. system for LCLS. Optical interferometer to measure length change, active feedback [9,10] N/A 6
>02'""#$#'J8)('2',390,6)-'(+3)0,6 1. Cavity field phase and amplitude stability. 2. Cavity frequency control Automated cavity tuning (normal conducting); piezo (superconducting) 3. Reliable (minimize LLRF induced downtime, maximize availability), exception handling (recovery from quench ) Redundant subsystem; modularity design 4. Support Automated operation Exp. Automated start, run, and close loop; Fault detected and recovery. 5. Build in Diagnostics (also providing remote access for diagnostics and control) 6. Minimize the RF power needed for control (work near the saturation of klystron, piezo tuner, overshoot control ) 7. LLRF modeling (cavity, klystron, detuning ) 8. Interlock 7668'6),>85H0-8/'6 1. Cavity Control Perturbations: Beam loading, phase noise from MO, Klystron ripple, Thermal drift, Lorentz detuning, microphonics, etc (know to what extent they can be controlled, identify those out of the feedback close loop control, drift in LO, downconvert, high frequency noise etc ) Feedback: Loop gain limitation, loop bandwidth, loop phase, delay and Instability; Beam based feedbacks? Feedforward: Cavity filling; Beam loading compensation; Piezo tuning; Klystron Linearization algorithms; 2. Frequency Generation Low phase noise stable reference master oscillator; Phase locked to low frequency (10MHz?) 3. Reference Phase Distribution (and time system) Drift caused by temperature changed; Fiber or coaxial cable? How frequencies generated? Distribute all or major frequencies? Where the MO located? Temperature stabilization. Power distribute to each LLRF station; Phase stability monitoring and correction? 4. Frequency conversion Low noise, Drift control for mixers, amplifier, and splitters 5. Diagnostics Forward/reflected power calibration, and calculation for tuning control (redundant). Drift calibration; IQ to phase & amplitude transfer; Loop phase rotation matrix; field calibration rotation matrix (based on rf, beam based transients, spectrometer?); Cavity detuning (average during pulse, detuning curve during pulse, by measuring probe or reflected power decay curve); load Q calculation; Beam phase measurement; 6. Cavity Frequency control Slow tuner (step motor)(maintain average resonance frequency, pre-detuning, maximize tuner lifetime); Faster tuner (piezo), dynamic Lorentz detuning compensation; microphonics control; Minimize RF power; System identification (beam phase and current, loaded Q, incident phase); 7. Control System server Automated fault recovery; Automated operation for all LLRF stations (remote, local?); Finite state machine; Interface to control system; RF system Database (calibration coefficients, subsystem characteristics). 8. Hardware Low noise; Fast high resolution ADC/DAC; Power convert switching noise; Nonlinearity, Signal integrity. 7
FK1'()',9'6+,-"'660,6 FLASH: Diagnostics for beam energy at many points along the linac are always a great help for setting up the RF, Slow drifts of resonance frequency was an issue High closed loop bandwidth (30kHz - 60KHz), G>150. [Brian Chase, Lessons Learned from the 9 ma Test ] SNS: The piezo tuner has not used for some failures. To compensate, about 15% more RF power is consumed to achieve the designed phase & amp stability [11]. The high voltage converter modulator system is below the design, so that beam pulse length only reaches 0.8 ms in a 60Hz production. [11,17,18] Acceleration gradients of several cavities are limited by HOM couplers;[11,17,18] Temperature dependences in downconvert; [6] Klystron circulator mismatch [16]; Modulator droop (causing phase margin degrade) and 20kHz ripple [16,19]; JPARC: Temperature dependency of the PLL-VCXO was larger than the expectation by multiplying 12 MHz reference frequency to get RF, LO. Instead, in new design the 312-MHz LO is directly distributed as reference [5]. There are also HVPS ripples induced klystron amplitude droop and phase modulation, but they were controlled down to the required stability accuracy [8]. 8
>02'FK1'()',9'),7?FC Good hardware platform is essential to make design easy. DC/DC power convert 120kHz switching frequency noise and harmonics; 120 khz and harmonics DAC output clock jitter dependence on frequency; Downconvert and upconvert on the same board with low isolation; Strong interference between RF and low IF Loop bandwidth (try to lower delay, try simple algorithms); >822+(L By dividing the LLRF system into sub-modules, it would be a bit clearer to identify the tasks and challenges in LLRF system. It is valuable to learn that how the other labs dealt with these problems and then find out the proper solutions for ESS. But there are still some challenges: 1.Work near the saturation of klystron to improve efficiency (80% or higher? about 60% in existing accelerators) 2. Automation of Operation for larger scale LLRF stations (at ESS, ~200 stations) 3. Field and resonance control (feedback, feedforward, tuner, etc. optimization for perturbations suppression) 4. Availability and reliability. 5. Spoke cavity. 6. Longer pulse, higher current, higher power 9
=9M,0*/'-N'2',36 I would like to acknowledge the help and suggestions received from J. Anders and P. Steve during writing this notes. #'O'(',9' $"=>? 1. LLRF System Components Development. TESLA Report 2008-03. Editor: R.Romaniuk. 2.http://flash.desy.de/sites2009/site_vuvfel/content/e66400/infoboxContent66401/flash-layout.pdf 3. LLRF Operation Experience at FLASH. Valeri Ayvazyan, DESY. LLRF workshop 2009; 4. LLRF Development at DESY. Simrock S. LLRF workshop 2007. http://neutrons.ornl.gov/workshops/llrf2007/presentations/006%20- %20LLRF%20Development%20at%20DESY%20-%20Simrock.pdf. 5. DESY LLRF. Markus Hoffmann. LLRF workshop 2009. 6. DESIGN OF THE DIGITAL RF CONTROL SYSTEM FOR THE TESLA TEST FACILITY. S. Simrock, 1996. 7.V. Ayvazyan et al., Digital RF Control System for The DESY FLASH Linear Accelerator, in EUROCON, 2007. The International Conference on" Computer as a Tool", 2007, 1178 1185. 8. LLRF Control: Operational Issues, Wojciech Cichalewski et.al. http://flash.desy.de/sites/site_vuvfel/content/e870/e2343/infoboxcontent2344/llrf_control.pdf. 9. T. Schilcher. Vector Sum Control of Pulsed Accelerating Fields in Lorentz Force Detuned Superconducting Cavities. Ph. D. Thesis of DESY, 1998 10. S. N. Simrock, Achieving phase and amplitude stability in pulsed superconducting cavities, in Proceedings of the 2001 Particle Accelerator Conference, Chicago. 11. M. Pekeler, Experience with superconducting cavity operation in the TESLA Test Facility, in Particle Accelerator Conference, 1999. Proceedings of the 1999, vol. 1, 1999, 245 249. 12. RECENT LLRF MEASUREMENTS OF THE 3RD HARMONIC SYSTEM FOR FLASH. C. Schmidt et al. IPAC 10. 13. LLRF CONTROL SYSTEM UPGRADE AT FLASH. V. Ayvazyan et al. PCaPAC 2010. 14. M. Hoffmann. Development of A Multichannel RF Field Detector for the Low-Level RF Control of the Free-Electron Laser at Hamburg. Ph.D. Thesis of DESY, 2008 15. Evaluation of an ATCA Based LLRF System at FLASH. Simrock S., Grzegorz Jabáoski, et al. 16th International Conference Mixed Design of Integrated Circuits and Systems, 2009 16. LORENTZ FORCE DETUNING COMPENSATION SYSTEM FOR ACCELERATING FIELD GRADIENTS UP TO 35 MV/M FOR SUPERCONDUCTING XFEL AND TESLA NINE-CELL CAVITIES. P. SEKALSKI et al. 17. S.N.Simrock, Lorentz Force Compensation of Pulsed SRF Cavities, Proceedings of LINAC 2002, Gyeongju, Korea 18. M. Liepe, W.D.-Moeller, S.N. Simrock, Dynamic Lorentz Force Compensation with a Fast Piezoelectric Tuner, Proceedings of the 2001 Particle Accelerator Conference, Chicago 19. L. Lilje, S. Simrock, D. Kostin, M. Fouaidy, Characteristics of a fast Piezo-Tuning Mechanism for Superconducting Cavities, Proceedings of EPAC 2002, Paris, France. 20. Field Stabilization in a Superconducting Cavity Powered in Pulsed Mode. Alban Mosnier. 21. PIEZOELECTRIC TUNER COMPENSATION OF LORENTZ DETUNING IN SUPERCONDUCTING CAVITIES. G. K. Davis, J. R. Delayen. PAC 03. 22. FIRST DEMONSTRATION OF MICROPHONIC CONTROL OF A SUPERCONDUCTING CAVITY WITH A FAST PIEZOELECTRIC TUNER. Simrock. S.N. PAC 03. 23. MEASUREMENT AND COMPENSATION OF MICROPHONICS IN CW-OPERATED TESLA- TYPE CAVITIES. Oliver Kugeler et al. Proceedings of ERL07. 10
23. ADAPTIVE FEEDFORWARD CANCELLATION OF SINUSOIDAL DISTURBANCES IN SUPERCONDUCTING RF CAVITIES. T. Kandil. Linac 04. 24.K. Rehlich, Status of the FLASH Free Electron Laser Control System, DESY, Hamburg, Germany. ICALEPCS 07. 25. PERFORMANCE OF THE NEW MASTER OSCILLATOR AND PHASE REFERENCE SYSTEM AT FLASH. Simrock S. PAC07. >A> 1. sns_parameters_list_june05.pdf, http://neutrons.ornl.gov/media/pubs/pdf/sns_parameters_list_june05.pdf. 2. M. Champion et al., Overview of the Spallation Neutron Source Linac Low-Level RF Control System, in Particle Accelerator Conference, 2005. PAC 2005. Proceedings of the, 3396 3398. 3. M. Champion et al., The Spallation Neutron Source Accelerator Low Level RF Control System, in Particle Accelerator Conference, 2003. PAC 2003, 2003. 4. H. Ma et al., Digital Control of Cavity Fields in the Spallation Neutron Source Superconducting Linac, in Linac 2006, 2006. 5. A. Regan et al., Newly designed field control module for the SNS, in Particle Accelerator Conference, 2003. PAC 2003. Proceedings of the, vol. 5, 3371 3373. 6. M. Crofford et al., SPALLATION NEUTRON SOURCE LLRF TEMPERATURE DEPENDENCE AND SOLUTION, in Linac 2010, 2010. 7. Hengjie Ma et al., Low-level rf control of Spallation Neutron Source: System and characterization, Physical Review Special Topics - Accelerators and Beams 9, no. 3 (3, 2006), http://link.aps.org/doi/10.1103/physrevstab.9.032001. 8. H. Ma et al., SNS low-level RF control system: Design and performance, in Particle Accelerator Conference, 2005. PAC 2005. Proceedings of the, 2005, 3479 3481. 9. K. Kasemir et al., Adaptive Feed Forward Beam-Loading Compensation Experience at the Spallation Neutron Source Linac, in Particle Accelerator Conference, 2005. PAC 2005. Proceedings of the, 2005, 1467 1469. 10. J. R. Delayen and G. K. Davis, Microphonics and Lorentz transfer function measurements on the SNS cryomodules, in The Proceedings of the 11th Workshop on RF Superconductivity, Travemunde, Germany, 2003. 11. Y. Zhang, Experience and Lessons with the SNS Superconducting Linac, Procs. of IPAC01, Kyoto, Japan (2010): 26. 12. M. Crofford et al., Operational Experience with the Spallation Neutron Source High Power Protection Module, in Particle Accelerator Conference, 2005. PAC 2005. Proceedings of the, 2005, 3411 3413. 13. M. Champion, SNS RF System Performance and Operation. PAC 07 talk. 14. Hengjie Ma, Progress in RF Controls at SNS, LLRF workship 2007. 15. M. Piller et al., The Spallation Neutron Source RF Reference System, in Particle Accelerator Conference, 2005. PAC 2005. Proceedings of the, 2005, 3573 3575. 16. M. McCarthy, M. Crofford, and S. K ORNL, SPALLATION NEUTRON SOURCE SUPERCONDUCTING LINAC KLYSTRON TO CAVITY MISMATCH EFFECTS AND COMPENSATION. Linac 08. 17. Sang-ho Kim, Operational Aspects and Experiences of the SNS LLRF System, LLRF Workshop 2009. 18. S. H. Kim et al., SNS SUPERCONDUCTING LINAC POWER RAMP-UP STATUS AND PLAN, PAC2009, Vancouver, CANADA (2009): 4 8. 19. D.E. Anderson et al., Operational Performance of the Spallation Neutron Source High Voltage Converter Modulator and System Enhancements, in Power Modulator Symposium, 2006. Conference Record of the 2006 Twenty-Seventh International, 2006, 427-430. 20. S. Kwon et al., SNS Superconducting Cavity Modeling-Iterative Learning Control, Arxiv preprint physics/0008161 (2000). BC=#D 1. JPARC Main Linac Parameters.pdf. 2. Y. Yamazaki and M. Kinsho, Discussion on RCS versus AR on the Basis of J-PARC Beam Commissioning for Pulsed Spallation Neutron Source, in Proceedings of the 11th European Particle Accelerator Conference (EPAC08), 3557 3559. 11
3. H. Ao, STATUS OF J-PARC LINAC ENERGY UPGRADE, in LINAC 2010, 2010. 4. T. Kobayashi, S. Anami, and S. Yamaguchi, PERFORMANCE OF RF REFERENCE DISTRIBUTION SYSTEM FOR THE J-PARC LINAC, in LINAC 2006, 2006. 5. T. Kobayashi et al., AUTOMATIC FREQUENCY MATCHING FOR CAVITY WARMING-UP IN J-PARC LINAC DIGITAL LLRF CONTRO, in PAC2009, Vancouver, CANADA, 2009. 6. T. Ito et al., RF Characteristics of the SDTL for the J-PARC, in LINAC 2006, 2006. 7. H. Ao et al., Improvement in the ACS Cavity Design for the J-PARC Linac Energy Upgrade, in Proc. of the 2008 Linac Conf, 2008, 915 917. 8. T. Kobayashi et al., Performance of J-PARC linac RF system, in Particle Accelerator Conference, 2007. PAC. IEEE, 2007, 2128 2130. 9. S. Yamaguchi et al., An Overview of the Rf System for the JAERI/KEK High Intensity Proton Linac, in Proc. of Linac2002, Pohan, Korea, 2002. 10. Z. Fang et al., LLRF CONTROLLER UPGRADE FOR THE J-PARC 400 MeV LINAC, in IPAC 10, 2010. 11. S. Michizono et al., Digital RF Control System for 400-MeV Proton Linac of JAERI/KEK Joint Project, in Proceedings of the 2002 Linac Conference, Gyeongju, Korea. 12. S. Michizono et al., Digital Feedback Control for 972-MHz RF System of J-PARC Linac, in Proceedings of the 2009 Particle Accelerator Conference (PAC 2009), Canada, 2009. 13. S. Michizono et al., Digital feedback system for JPARC linac RF source, in LINAC 2004, 2004. 14. S. Michizono et al., Performance of Digital LLRF Field Control System for the J-PARC Linac, LINAC06, Knoxville (2006): 574. 15. T. Kobayashi, STABILITY EVALUATION FOR LONG FB LOOP DELAY IN THE ACS CAVITY FIELD CONTROL FOR THE J-PARC LINAC 400-MEV UPGRADE, in LINAC 2010, 2010. 16. T. Kobayashi et al., PULSE-BY-PULSE SWIT ING OF BEAM LOADING COMPENSATION IN J-PARC LINAC RF CONTROL, in LINAC 08, 2008. 17. S. Anami et al., Control of the Low Level RF System for the Jparc Linac, in LINAC 2004, 2004. 18. T. Kobayashi et al., RF REFERENCE DISTRIBUTION SYSTEM FOR THE J-PARC LINAC, in LINAC 2004, 2004. $'(2)/+5 1. J. Branlard, S. Simrock, and S. Michizono, Survey of LLRF development for the ILC, in Particle Accelerator Conference, 2007. PAC. IEEE, 2007, 3810 3812. 2. R. C. Webber et al., First high gradient test results of a dressed 325 MHz superconducting single spoke resonator at Fermilab (Fermi National Accelerator Laboratory (FNAL), Batavia, IL, 2010). 3. T. A Butler et al., New LLRF System for Fermilab 201.25 MHz Linac, in Proc. LINAC 08, vol. 8, 2008. 4. J. Branlard et al., Capture cavity II results at FNAL, in Particle Accelerator Conference, 2007. PAC. IEEE, 2245 2247. 5. J. Branlard et al., LLRF Design for the HINS-SRF Test Facility at Fermilab, in this conference, LINAC10, Tsukuba, 2010. 6. U. Mavric et al., A 96 channel receiver for the ILCTA LLRF system at Fermilab, in Particle Accelerator Conference, 2007. PAC. IEEE, 2007, 2271 2273. 7. U. Mavric et al., DESIGN AND EVALUATION OF THE LOW-LEVEL RF ELECTRONICS FOR THE ILC MAIN LINAC, in LINAC 08, 2008. 8. U. Mavric, B. Chase, and M. Vidmar, Design and evaluation of a low-level RF control system analog/digital receiver for the ILC main LINACs, Nuclear Instruments and Methods in Physics Research Section A: Accelerators, Spectrometers, Detectors and Associated Equipment 594, no. 1 (2008): 90 96. 9. P. Varghese et al., Multichannel vector field control module for LLRF control of superconducting cavities, in Particle Accelerator Conference, 2007. PAC. IEEE, 2008, 2298 2300. 10. R. Carcagno et al., FIRST FERMILAB RESULTS OF SRF CAVITY LORENTZ FORCE DETUNING COMPENSATION USING A PIEZO TUNER (2007). 12
11. R. Carcagno et al., Microphonics detuning compensation in 3.9 GHZ superconducting RF cavities, in Proceedings of the 11th Workshop on RF Superconductivity SRF2003, 2003. 12. J. Branlard and B. Chase, Master Oscillator for Fermilab ILC Test Accelerator, in LINAC 2006, 2006. 13. B. Chase, LLRF Challenges for ILC, ProjectX, HTS, NML, Achievements, Plans, May 18, 2009. DF#A ;P P. Baudrenghien et al., The LHC Low Level RF, in EPAC 2006, 2006. 2. J. Molendijk et al., Digital Design of the LHC Low Level RF: The Tuning System for the Superconducting cavities, in Proc. 10th European Particle Accelerator Conf., Edinburgh, United Kingdom, 2006. 3. J. Tückmantel, Adaptive RF Transient Reduction for High Intensity Beams with Gaps, in EPAC 2006, 2006. 4. M. E. Angoletta et al., CERN s LEIR Digital LLRF: System Overview and Operational Experience, in paper TUPEA057, this conference. 5. M. Vretenar, Status of Linac4 construction at CERN, in Linac conference, 2010. 6. 6-CERN_lab_Talk_LLRF09_v2.ppt. LLRF workshop 2009. 7. P. Baudrenghien, Low Level RF III. CERN CAS Courses: RF for Accelerators. 2010. "GA" 1. L. Doolittle, H. Ma, and M. S Champion, Digital low-level RF control using non-iq sampling, in Proc. LINAC, 2006, 568 570. 2. http://recycle.lbl.gov/llrf4/ 3. LLRF4 Board, http://recycle.lbl.gov/llrf4/stat-20070521.pdf. 4. LLRF Evaluation Board, http://www.dimtel.com/_media/support/manuals/packet.pdf. 5. L. Doolittle, Embedded Networked Front Ends Beyond the Crate, in ICALEPCS 2003, 2003. 6. L. Doolittle, Low-Level RF Control System Design and Architecture, in Asian Particle Accelerator Conference, 2007. 7. L. Doolittle, Applications of General-Purpose Reconfigurable LLRF Processing Architectures, in PAC 2009, 2009. 8. L. Doolittle, Digital Feedback for Cavity Field Control using Near-IQ Sampling, 2006. http://recycle.lbl.gov/~ldoolitt/llrf/neariq.pdf 9. J. M. Byrd et al., Femtosecond synchronization of laser systems for the LCLS, in This conference. 10. J. W. Staples, R. Wilcox, and J. M. Byrd, Demonstration of femtosecond-phase stabilization in 2 km optical fiber, in Particle Accelerator Conference, 2007. PAC. IEEE, 2007, 494 496. HI##?= 1. J. L Biarrotte et al., Accelerator reference design for the MYRRHA European ADS demonstrator, in LINAC 2010, 2010. 2. J. L Biarrotte, T. Junquera, and A. C. Mueller, A European Advanced Technology Programme for ADS Accelerator Development, SRF 2005 (2005), http://hal.in2p3.fr/in2p3-00123373/en/. 3. F. Bouly and J.L. Biarrotte, LOW LEVEL RADIO-FREQUENCY DEVELOPMENTS TOWARD A FAULT-TOLERANT LINAC SCHEME FOR AN ACCELERATOR DRIVEN, in LINAC 2010, 2010. 4. O. Piquet et al, Task 1.3.4 final report, EUROTRANS deliverable, D1.66, December 2008. 5. F. Bouly et al., DEVELOPMENTS AND TESTS OF A 700 MHz PROTOTYPICAL CRYO- MODULE FOR THE MYRRHA ADS PROTON LINEAR ACCELERATOR, in LINAC 2010, 2010. 6. F. Bouly et al., CONSTRUCTION OF A 700 MHz PROTOTYPICAL CRYOMODULE FOR THE EUROTRANS ADS PROTON LINEAR ACCELERATOR. SRF 2009. 7. M. Luong and O. Piquet, RF Control System Modeling, EUROTRANS deliverable, D1.40, 2007. 13
14