Brian Chase

Multichannel vector field control module for LLRF control of superconducting cavities

The field control of multiple superconducting RF cavities with a single Klystron, such as the proposed RF scheme for the ILC, requires high density (number of RF channels) signal processing hardware so that vector control may be implemented with minimum group delay. The MFC (Multichannel Field Control) module is a 33- channel, FPGA based down-conversion and signal processing board in a single VXI slot, with 4 channels of high speed DAC outputs. A 32-bit, 400MHz floating point DSP provides additional computational and control capability for calibration and implementation of more complex control algorithms. Multiple high speed serial transceivers on the front panel and the backplane bus allow a flexible architecture for inter-module real time data exchanges. An interface CPLD supports the VXI bus protocol for communication to a SlotO CPU, with Ethernet connections for remote in system programming of the FPGA and DSP as well as data acquisition.

Neural Networks for Modeling and Control of Particle Accelerators

Particle accelerators are host to myriad nonlinear and complex physical phenomena. They often involve a multitude of interacting systems, are subject to tight performance demands, and should be able to run for extended periods of time with minimal interruptions. Often times, traditional control techniques cannot fully meet these requirements. One promising avenue is to introduce machine learning and sophisticated control techniques inspired by artificial intelligence, particularly in light of recent theoretical and practical advances in these fields. Within machine learning and artificial intelligence, neural networks are particularly well-suited to modeling, control, and diagnostic analysis of complex, nonlinear, and time-varying systems, as well as systems with large parameter spaces. Consequently, the use of neural network-based modeling and control techniques could be of significant benefit to particle accelerators. For the same reasons, particle accelerators are also ideal test-beds for these techniques. Many early attempts to apply neural networks to particle accelerators yielded mixed results due to the relative immaturity of the technology for such tasks. The purpose of this paper is to re-introduce neural networks to the particle accelerator community and report on some work in neural network control that is being conducted as part of a dedicated collaboration between Fermilab and Colorado State University (CSU). We describe some of the challenges of particle accelerator control, highlight recent advances in neural network techniques, discuss some promising avenues for incorporating neural networks into particle accelerator control systems, and describe a neural network-based control system that is being developed for resonance control of an RF electron gun at the Fermilab Accelerator Science and Technology (FAST) facility, including initial experimental results from a benchmark controller.

Precision vector control of a superconducting RF cavity driven by an injection locked magnetron

The technique presented in this paper enables the regulation of both radio frequency amplitude and phase in narrow band devices such as a Superconducting RF (SRF) cavity driven by constant power output devices i.e. magnetrons. The ability to use low cost high efficiency magnetrons for accelerator RF power systems, with tight vector regulation, presents a substantial cost savings in both construction and operating costs compared to current RF power system technology. An operating CW system at 2.45 GHz has been experimentally developed. Vector control of an injection locked magnetron has been extensively tested and characterized with a SRF cavity as the load. Amplitude dynamic range of 30 dB, amplitude stability of 0.3% r.m.s, and phase stability of 0.26 degrees r.m.s. has been demonstrated.

A 96 channel receiver for the ILCTA LLRF system at fermilab

The present configuration of an ILC main LINAC RF station has 26 nine cell cavities driven from one klystron. With the addition of waveguide power coupler monitors, 96 RF signals will be down-converted and processed. A down-converter chassis is being developed that contains 12 eight-channel analog modules and a single up- converter module. This chassis will first be deployed for testing a cryomodule composed of eight cavities located at New Muon Laboratory (NML) - Fermilab. Critical parts of the design for LLRF applications are identified and a detailed description of the circuit with various characteristic measurements is presented. The board is composed of an input band-pass filter centered at 1.3 GHz, followed by a mixer, which down-converts the cavity probe signal to a proposed 13 MHz intermediate frequency. Cables with 8 channels per connector and good isolation between channels are being used to interconnect each down-converter module with a digital board. As mixers, amplifiers and power splitters are the most sensitive parts for noise, nonlinearities and crosstalk issues, special attention is given to these parts in the design of the LO port multiplication and distribution.

Technique for monitoring fast tuner piezoactuator preload forces for superconducting RF cavities

The technology for mechanically compensating Lorentz Force detuning in superconducting RF cavities has already been developed at DESY. One technique is based on commercial piezoelectric actuators and was successfully demonstrated on TESLA cavities [1]. Piezo actuators for fast tuners can operate in a frequency range up to several kHz; however, it is very important to maintain a constant static force (preload) on the piezo actuator in the range of 10 to 50% of its specified blocking force. Determining the preload force during cool-down, warm-up, or re-tuning of the cavity is difficult without instrumentation, and exceeding the specified range can permanently damage the piezo stack. A technique based on strain gauge technology for superconducting magnets has been applied to fast tuners for monitoring the preload on the piezoelectric assembly. The design and testing of piezo actuator preload sensor technology is discussed. Results from measurements of preload sensors installed on the tuner of the Capture Cavity II (CCII )[2] tested at FNAL are presented. These results include measurements during cool-down, warm- up, and cavity tuning along with dynamic Lorentz force compensation.

BPM System for Electron Cooling in the Fermilab Recycler Ring

We report a VXI based system used to acquire and process BPM data for the electron cooling system in the Fermilab Recycler ring. The BPM system supports acquisition of data from 19 BPM locations in five different sections of the electron cooling apparatus. Beam positions for both electrons and anti‐protons can be detected simultaneously with a resolution of ±50 μm. We calibrate the system independently for each beam type at each BPM location. We describe the system components, signal processing and modes of operation used in support of the electron‐cooling project and present experimental results of system performance for the developmental electron cooling installation at Fermilab.

Capture cavity II results at FNAL

As part of the research and development towards the International Linear Collider (ILC), several test facilities have been developed at Fermilab. This paper presents the latest Low Level RF (LLRF) results obtained with Capture Cavity II (CCII) at the ILC Test Accelerator (ILCTA) test facility. The main focus will be on controls and RF operations using the SIMCON based LLRF system developed in DESY. Details about hardware upgrades and future work will be discussed.

Analog Receiver and Transmitter Design for the ILC Main LINAC LLRF Control System

Particle accelerators of the next generation, like the International Linear Collider (ILC), will need exceptional beam quality in order to achieve the required luminosity and beam energy dictated by the experiments. Performance of a low-level radio frequency (LLRF) control system plays a major role in preserving beam quality in a particle accelerator. LLRF system regulation specifications are usually given in terms of phase and amplitude stability of the detected fields in a cavity and in the case of the ILC these are not the same for all parts of the machine. For instance, bunch compressor and damping rings have more severe specifications in terms of regulation than the main linear accelerator (LINAC). In the introduction, we present the main design issues for the ILC main LINAC LLRF system and focus on the analog receiver and transmitter design. In the first section a short overview of the main design approaches is given. In the following section we present measurements of the main building blocks in an analog receiver and transmitter and apply measured data on a model. We study interactions between the two modules, when integrated in a closed loop LLRF system. At the end, we present results followed by conclusion.

Longitudinal momentum mining of antiprotons at the fermilab recycler: past, present and future

The technique of longitudinal momentum mining (LMM) in the Fermilab Recycler was adopted in early 2005 to extract thirty-six equal intensity and equal 6D- emittance antiproton bunches for proton-antiproton collider operation in the Tevatron. Since that time, several improvements have been made in the Recycler and the mining technique to handle higher intensity beams. Consequently, the Recycler has become a key contributor to the increased luminosity performance observed during Tevatron Run lib. In this paper, we present an overview of the improvements and the current status of the momentum mining technique.

Uniform longitudinal beam profiles in the Fermilab Recycler using adaptive RF correction

The Fermilab recycler ring is a permanent magnet based 8 GeV anti-proton storage ring. A wideband RF system, driven with ARBs (ARBitrary waveform generators), allows the system to produce programmable barrier waveforms. Beam current profile distortion was observed, its origin verified both experimentally and theoretically, and an FPGA-based correction system was designed, tested and implemented to level the bunch profile.