LANL, Los Alamos, New Mexico
Funding: This work has benefited from the use of the LANSCE at LANL. This facility is funded by the US DOE and operated by LANS for NSSA under Contract DE-AC52-06NA25396. LA-UR-08-03585.
Tuning particle accelerators is time consuming and expensive, with a number of inherently non-linear interactions between system components. Conventional control methods have not been successful in this domain, and the result is constant and expensive monitoring of the systems by human operators. This is particularly true for the start-up and conditioning phase after a maintenance period or an unexpected fault. In turn, this often requires a step by step restart of the accelerator. Surprisingly few attempts have been made to apply intelligent accelerator control techniques to help with beam tuning, fault detection, and fault recovery problems. The reason for that might be that accelerator facilities are rare and difficult to understand systems that require detailed expert knowledge about the underlying physics as well as months if not years of experience to understand the relationship between individual components, particularly if they are geographically disjoint. This paper will give an overview about the research effort in the accelerator community that has been dedicated to the use of artificial intelligence methods for accelerator beam line tuning.
JLAB, Newport News, Virginia
Funding: Authored by Jefferson Science Associates, LLC under U.S. DOE Contract No. DE-AC05-06OR23177.
In the last 20 years the requirements for rf control has increased as the target use has broadened from electron/ion accelerators for Nuclear and Particle Physics to light sources such as Free Electron Lasers. The increasing requirement of cavity field control to meet the spectral and jitter performance specifications for light sources has led system designers to a more rigorous approach in designing the rf controls. Design attention must be applied not only to the hardware and control algorithms but also to the overall accelerating system to meet performance and cost requirements. As an example, cavity QL in Energy Recovery Linacs (ERL) must be optimized such that the rf controls can accommodate the lowest possible rf power given the background cavity microphonics. This paper presents the status and future directions of high QL superconducting rf control systems.
BESSY GmbH, Berlin
Funding: Work partially funded by the EU Commission in the sixth framework programme, contract no 011935 EURO-FEL-DS5, BMBF and Land Berlin.
All mechanical tuning systems are subject to hysteresis effects: For coarse tuning with a stepper motor, the exercised forces lead to a visco-elastic deformation of the tuner body. In piezo-based fine tuning, even if the smaller deformations of tuner and cavity can be regarded as fully elastic, the piezo-actuators themselves suffer from remanent polarization effects. The extent of these nonlinearities has been measured in three different tuning systems (Saclay I, Saclay II and Blade Tuner) utilizing high-voltage as well as low-voltage piezo actuators. An estimate of the resulting tuner-resolution and performance degradation with respect to microphonics compensation is given. Experiments were performed in the HoBiCaT facility at BESSY.
JLAB, Newport News, Virginia
ODU, Norfolk, Virginia
Funding: Supported by US DOE Contract No. DE-AC05-06OR23177
In low current applications superconducting cavities have a high susceptibility to microphonics induced by external vibrations and pressure fluctuations. Due to the narrow bandwidth of the cavities, the amount of rf power required to stabilize the phase and amplitude of the cavity field is dictated by the amount of microphonics that need to be compensated. Electronic damping of microphonics is investigated as a method to reduce the level of microphonics and of the amount of rf power required. The current work presents a detailed analysis of electronic damping and of the residual cavity field amplitude and phase errors due to the fluctuations of cavity frequency and beam current.
Fermilab, Batavia
One of the goals of the low energy 60 MeV section of the Fermilab HINS H- linac is to demonstrate that a total of 40 rf cavities can be powered by a single 2.5 MW, 325 MHz klystron. This requires individual vector modulators at the input of each rf cavity to independently adjust the amplitude and phase of the rf input signal during the 3.5 ms rf pulse. Two versions of vector modulators have been developed; a 500 kW device for the RFQ and a 75 kW modulator for the remaining rf cavities. High power test results showing the vector modulator phase and amplitude responses will be presented.
TRIUMF, Vancouver
The ISAC 2 superconducting rf cavities use self-excited, phase-locked mode of operation. As such the microphonics are sensitive to the alignment of the phase control loop. Although initial alignments can minimize the effect of microphonics, long term drifts, particularly in the power amplifiers, can cause the control loop to misalign and an increase in sensitivity to microphonics. The ISAC 2 control system monitors several points in the control loop to determine the phase alignment of the power amplifiers as well as the rf resonant cavities. Online adaptive feedbacks using Self Tuning Regulators are employed to bring the different components back into alignment.
TRIUMF, Vancouver
Most rf feedback control systems today uses the I/Q demodulation and modulation scheme because of its simplicity. Its performance, however, depends on the alignment of the feedback loops. If the loop contains elements that have a high AM-PM conversion such as a class C amplifier, then the misalignment is dynamic and power dependent. In the extreme case the I/Q loops can become unstable and the system settled into a limit-cycle oscillation.
DESY, Hamburg
For the LLRF system of superconducting linacs, precision measurements of the rf phase and amplitude are critical for the achievable field stability. In this paper, a fast ADC (ADS5474) has been evaluated for the measurement of a 1.3 GHz rf signal directly without frequency down conversion. The ADC clock frequency is synchronized with the rf frequency and chosen for non-IQ demodulation. In the laboratory, the Signal to Noise Ratio (SNR) of the ADC was studied for different clock and rf input levels, and the temperature sensitivity of the ADC has been determined. A full bandwidth phase jitter of 0.2 degree (RMS) and amplitude jitter of 0.32% (RMS) was measured. For field control of superconducting cavities with a closed loop bandwidth up to 100 KHz, one can expect to achieve a phase stability close to 0.01 degree. The main limitation will be the jitter of the external clock. We present a measurements at the cavities at FLASH and compare the result with the existing system.
RIKEN/SPring-8, Hyogo
University of electro-communications, Tokyo
RIKEN Spring-8 Harima, Hyogo
Requirement on a Low Level rf (LLRF) system is very tight and allowable jitter is less than several tens femto seconds for the XFEL/SPring-8. To satisfy this requirement, we have developed special components; a low-noise master oscillator, a high precision IQ modulator/demodulator, a high speed DAC/ADC, and a delayed pulse generator with 700 fs jitter to a 5712 MHz reference clock. These components were installed in the SCSS test accelerator and their performance was checked. The standard deviations of the phase and amplitude were less than 0.02 degree and 0.03% for a 238 MHz SHB acceleration cavity. Measured rms jitter of the beam arrival time relative to the reference rf signal was 50 fs, which demonstrated the high performance of the total LLRF system. For the XFEL, the length of reference signal transmission line is long, about 1 km. Therefore an optical system is adopted because of low transmission loss and an ability to keep precise time accuracy using fiber length control, which has 0.2 um/sqrt(Hz) noise floor. Achieved performance of the LLRF and timing system, and development status on the optical transmission system will be presented in this paper.
KEK, Ibaraki
JAEA/J-PARC, Tokai-Mura, Naka-Gun, Ibaraki-Ken
JAEA, Ibaraki-ken
At the J-PARC 181 MeV proton linac, the rf sources consist of 4 solid-state amplifiers and 20 klystrons with operation frequency of 324 MHz. The rf fields of each rf source are controlled by a digital feedback system installed in a compact PCI (cPCI). A very good stability of the accelerating fields has been successfully achieved about ±0.2% in amplitude and ±0.2 degree in phase, much better than the requirements of ±1% in amplitude and ±1 degree in phase. Besides, the tuning of each accelerator cavity including 3 DTL and 15 SDTL is also controlled by this LLRF system through a cavity tuner. We pre-defined the cavity resonance states with the tuner adjusted to obtain a flat phase during the cavity field decay. The cavity auto-tuning is well controlled to keep the phase of rf fields within ±1 degree. Furthermore, from the amplitude waveform during the cavity field decay, the Q-value of each cavity is calculated in real-time and displayed in the PLC TP of the LLRF control system.
KEK, Ibaraki
In a superconducting accelerator, an FPGA/DSP-based low-level rf (LLRF) system with feedback control is adopted to satisfy the requirement of stability in the accelerating field. An rf probe signal picked up from cavity is down-converted to an intermediate frequency and sampled by an analog-to-digital converter (ADC) in the digital LLRF control system. In order to decrease the number of the ADCs required for vector sum feedback operation, a digital LLRF control system using different intermediate frequencies has been developed. At STF (Superconducting RF Test Facility) in KEK, the digital LLRF system with four intermediate frequencies was operated and the rf field stability under the feedback operation was estimated using a superconducting cavity. The result of the performance will be reported.
KEK, Ibaraki
RF operation has started at the STF (Superconducting RF Test Facility) in KEK. The digital feedback system, which consists of one FPGA, ten 16-bit ADCs and two 14-bit DACs, was installed in order to satisfy the rf-field regulation requirements of 0.3% rms and 0.3 deg.rms in phase. The rf field stability under various feedback parameters are presented. Various studies were also carried out such as cavity detuning measurements (microphonics, quench detection, etc.). These results will also be summarized.
KEK, Ibaraki
In the superconducting rf test facility (STF) at KEK, high power tests of the nine-cell superconducting cavity for the international linear collider (ILC) have been performed. Although the cavity was operated in π-mode, the feedback instability due to 8/9π and 7/9π modes was observed in the STF. The intensities of 8/9π and 7/9π modes were measured by changing the feedback loop-delay and stable/unstable region appeared periodically as expected.
KAERI, Daejon
The requirements for the field amplitude and phase stability of the PEFP linac are 1% and 1 degree, respectively. To achieve the requirements, a digital LLRF control system has been developed using a commercial digital board for general purpose(FPGA). The feedback with PI control and feedforward are implemented in the FPGA. The LLRF control systems are currently used for the linac test. In this paper, test results and discussion on the advantage and disadvantage of the LLRF system based on a commercial board are presented.
RAS/INR, Moscow
Stabilization of amplitude and phase in linear accelerator cavities can be realized by means of control systems, operating both in polar (A/P) and rectangular (I/Q) coordinates. In analyzing of linear control systems, as a rule, transfer functions are used, which, in turn, are the symbolic representation of the linear differential equation, connecting the input and output variables. It's well known that generally in A/P coordinate it is impossible to get two separate linear differential equations for amplitude and phase of rf voltage in a cavity except for estimating of the control system stability in the small near steady state values of variables. Nevertheless, there is a possibility of numerical simulation of nonlinear A/P control system using up-to-date programs. Some results of the simulation are presented.
Fermilab, Batavia
The Fermilab Proton Plan, tasked to increase the intensity and reliability of the Proton Source, has identified the Low Level RF (LLRF) system as the critical component to be upgraded in the Linac. The current 201.25 MHz Drift Tube Linac LLRF system was designed and built over 35 years ago and does not meet the higher beam quality requirements under the new Proton Plan. A new VXI based LLRF system has been designed to improve cavity vector regulation and reduce beam losses. The upgrade includes an adaptive feedforward system for beam loading compensation, a new phase feedback system, and a digital phase comparator for cavity tuning. The new LLRF system is phase locked to a temperature stabilized 805 MHz reference line, currently used as frequency standard in the higher energy accelerating section of the Linac. This paper will address the current status of the project, present the advancements in both amplitude and phase stability over the old LLRF system, and discuss commissioning plans.
LBNL, Berkeley, California
Funding: This work was supported by the Office of Science, U. S. Department of Energy, under Contract No. DE-AC02-05CH11231.
A fiber-based frequency and timing distribution system based on the principle of heterodyne interferometry has been in development at LBNL for several years. The temporal fiber drift corrector has evolved from an rf-based to an optical-base system, from mechanical correctors (piezo and optical trombone) to fully electronic, and the electronics from analog to fully digital, all using inexpensive commodity fiber components. Short-term optical phase jitter and long-term phase drift are both in the femtosecond range over distribution paths of 2 km and more. The temperature dependence of group and phase velocity correction is measured and applied. We will discuss the results of field tests, integration into various client subsystems and further applications.