RTG 2128 AccelencE

LLRF Control Systems and Beam Diagnostics

The realization of an ERL operation mode includes challenging requirements on the rf-system and requires new concepts in beam diagnostics. A LLRF system topology design needs to be based on the physical requirements and the available signals, considering phase and amplitude accuracy, the reduction of microphonic effects and the handling of beam loading including beam current variation and imbalances. High beam power and the 180° phase shift between accelerating and decelerating bunches resulting in a change of bunch spacings between ERL and conventional mode, which is even worse in case of non-isochronous operation since the phase shift is not exactly 180° anymore, complicates the use of conventional beam diagnostics like e.g. rf-monitor devices. Project Area C of AccelencE addresses the challenges of developing and adapting both LLRF control systems and beam diagnostics for ERL operation.

  • In a multi-turn ERL operation there are at least two beams of the same energy. In case of S DALINAC and MESA the beams of the same energy are transported through the same beamline. It is a big challenge to optimize a lattice for the first accelerated beam and bring a decelerated beam through the same section with only tiny options to tune it. It is also a challenge to have a diagnostic system measuring sensitively both beams, ideally non-destructive tools should be used.

    The biggest difference between the two beams is a phase shift of 180°, resulting in a 6 GHz repetition rate of bunches for interleaved accelerated and decelerated bunches. This project will investigate the applicability of three different systems: (a) 6-GHz BPM cavities, (b) 3 GHz BPM cavities and a pulsed beam, and (c) determination of the beam position by the usage of a wire scanner along with the measurement of the secondary particles.

    Applications of these diagnostic systems will include measurements on the stability of the ERL operation concerning beam break up, sensitivity to beam dynamics etc. These will be compared to simulations.

    Supervisor: Prof. Dr. Dr. h.c. mult. Norbert Pietralla

  • In the extraction beamline of the S DALINAC a high-energy scraper-system was installed. The commissioning was done during the first funding period of this RTG. The measurements proved that this system is able to minimize the momentum spread at the experiments and to reduce the background significantly. Furthermore, it showed its capability as beam diagnostics system.

    An automatized stabilization of the beam energy will be implemented in this project. It is based on the current measurements of the slits and can correct the beam energy by adjusting slightly the energy gain of the main accelerator, accordingly. Operating the scraper in this stabilizer mode will result in desired stable beam intensities at the experiments downstream.

    The scraper system will also be applied as an advanced diagnostic tool for an optimized setting of the non-isochronous recirculation mode of the S-DALINAC. If the S DALINAC’s recirculation arcs are not set correctly (e.g. non-vanishing angular dispersion), the beam may fluctuate in position at the scraper system without fluctuating in energy. This situation would result in changing energies passing the energy scraper. Then a position sensitive analysis of the beam hitting a downstream target in a dispersive section makes it possible to measure that change in beam energy. Within this project dedicated simulations will be done to use the scraper system and the downstream target for further optimization of the non-isochronous operation. The position-sensitive target will be set up at the appropriate downstream position.

    Supervisor: Prof. Dr. Dr. h.c. mult. Norbert Pietralla

  • Multi-turn Energy-Recovery Linacs used for long-time high-precision experiments (like the P2 experiment at MESA) rely on a stable beam quality and thus on a precise and reliable adjustment of the radio frequency parameters. Even for short-time experiments, temporary shutdowns due to component failure, replacement or readjustment require to restart the ERL with the same beam quality and properties. Both, long-time stability and restarts, are impaired by various uncertainties.

    Therefore, investigations on reliable RF-control for ERLs in terms of reproducibility and easier restarts are necessary.

    The structure and main goals of this project can be summarized as follows: (a) finding and analyzing possible uncertainties, (b) assessing the influence of the identified uncertainties on RF parameters, (c) formulate requirements in terms of robustness and resilience, and (d) selecting appropriate methods to realize these requirements (these may also include AI or machine learning approaches).

    For future ERLs with beam powers exceeding the installed RF power it is important to estimate the required overhead RF power for ensuring reliable RF operation in the presence of the identified uncertainties and according to the requirements formulated in this project.

    Supervisor: Prof. Dr. Harald Klingbeil

  • This project aims at a substantial increase of the operational efficiency of multi-turn ERLs using methods form computational sciences. Innovative and advanced computational techniques shall be applied to optimize the beam tuning process. An intelligent learning algorithm would improve the beam-tuning process in finding an optimized parameter vector for a given task. It may also speed up the beam tuning in predicting a suitable parameter vector if external effects have slightly changed the lattice or RF properties, for instance, during a temporary shutdown.

    We plan to make use of nature-inspired algorithms, especially Q-learning Neural Networks, to perform the desired optimization tasks. Beside this, the beam stability can also be improved by applying this learning algorithm to the RF control loop.

    Further applications, such as a systematic survey of process data of the whole machinery for uncovering linear and non-linear relationships between machine processes, are conceivable. Knowing them may help to locate malfunctions of the machinery faster and to prevent failures. Both, S DALINAC and MESA, make use of the worldwide established EPICS framework for their control systems. Both, the optimization and data science tools to be established will have interfaces to an EPICS based control system. The S-DALINAC will be used as the first test site for the developments. All the written software are planned to be tested and characterized at both facilities, S-DALINAC and MESA.

    Supervisor: Dr. Jonny Birkhan

  • Within this RTG a laser Compton backscattering (LCB) setup in its simplest version will be installed in the third recirculation beam line at the S-DALINAC. At the interaction point a laser with a photon energy of 1.2 eV will collide with the electron bunches of a maximum energy of 100 MeV to produce backscattered gamma rays with an energy of up to 180 keV exceeding all X-ray energies of stable elements. A challenge is to collide an electron beam with a laser beam, both with a size of 100 µm or smaller. A high precision alignment of the lattice, the laser beam transport and the interaction point and their stabilization are absolutely necessary.

    Concerning the LCB experimental site, simulations are required to determine the accuracy needed for the elements of the laser beam transport or the interaction region. Aligned and accurately controllable targets will be used to precisely guide the laser and the electron beam to the interaction point. Even a stabilization of the interaction region in temperature might be needed.

    To compensate for long-term drifts of both beams a precise stabilization of the interaction will be integrated. A possibility could be to stabilize the laser beam behind the interaction region at the laser beam dump. The electron beam will then be stabilized on the signals of the backscattered photons via a feedback loop to magnetic steering elements. Short-term and long-term stability of the LCB setup will be characterized.

    Supervisor: Dr. Michaela Arnold