Speaker
Dr
Paul Joireman
(Fermilab)
Description
Fermilab’s Main Injector, Recycler and Tevatron synchrotrons Low Level RF systems
perform a wide variety of tasks in support of accelerator goals, colliding beam HEP
and neutrino production. In this paper, we will focus on the architecture and
features of the LLRF software, the development practices we use and the lessons we
have learned in the process. These specific features are presented within the
framework of a standard control model.
The LLRF systems are called upon to perform a variety of different tasks including
data acquisition, beam acceleration, machine to machine beam transfer, feedback
control, data transfer, and user interface. To do this, we implement separate
processes that run with different periods, ranging from a few microseconds to
minutes. These systems must also implement a flexible toolset that allows an
end-user to reconfigure the system, combine the tools in unforeseen and useful ways,
and request new features. The systems must be highly reliable, and both fault
detecting and fault tolerant. In addition, changes to any system must minimally
affect beam operation of all systems.
In order to meet these requirements we use a variety of techniques. We employ a set
of size C VXI modules built in-house from a standard base design, which includes,
SHARC DSP, ALTERA FPGA, and VXI interface. The remainder of the module is
customized
to perform specific functions, such as data acquisition, frequency control, or RF
switching. The common DSP platform enables significant savings in software design
and implementation by allowing us to develop, test and deploy a single solution to a
common system problem across multiple VXI modules. In order to meet system
specific goals, such as acceleration or beam transfer, cooperation between modules
is
required. We implement an object-based methodology which provides a common
interface to data and services provided by a VXI module. System level software
employs these module objects by requesting services to fulfill system tasks and
exposing module specific data to the accelerator control system. The system level
software also implements the interface to support user requests for system
reconfiguration. The interface allows the user to reconfigure a single parameter,
such as feedback loop gains, or the user can specify a specific timed scenario of
operations for the system to perform. Error logging and reporting are tightly
integrated into the system software to enable debugging and fault diagnosis.
Author
Dr
Paul Joireman
(Fermilab)
