THDAU —  Embedded+Real Time 2   (13-Oct-11   16:00—17:40)
Chair: K. Furukawa, KEK, Ibaraki, Japan
Paper Title Page
THDAULT01 Modern System Architectures in Embedded Systems 1260
 
  • T. Korhonen
    PSI, Villigen, Switzerland
 
  Several new technologies are making their way also in embedded systems. In addition to FPGA technology which has become commonplace, multicore CPUs and I/O virtualization (among others) are being introduced to the embedded systems. In our paper we present our ideas and studies about how to take advantage of these features in control systems. Some application examples involving things like CPU partitioning, virtualized I/O and so an are discussed, along with some benchmarks.  
slides icon Slides THDAULT01 [1.426 MB]  
 
THDAUST02 An Erlang-Based Front End Framework for Accelerator Controls 1264
 
  • D.J. Nicklaus, C.I. Briegel, J.D. Firebaugh, CA. King, R. Neswold, R. Rechenmacher, J. You
    Fermilab, Batavia, USA
 
  We have developed a new front-end framework for the ACNET control system in Erlang. Erlang is a functional programming language developed for real-time telecommunications applications. The primary task of the front-end software is to connect the control system with drivers collecting data from individual field bus devices. Erlang's concurrency and message passing support have proven well-suited for managing large numbers of independent ACNET client requests for front-end data. Other Erlang features which make it particularly well-suited for a front-end framework include fault-tolerance with process monitoring and restarting, real-time response,and the ability to change code in running systems. Erlang's interactive shell and dynamic typing make writing and running unit tests an easy part of the development process. Erlang includes mechanisms for distributing applications which we will use for deploying our framework to multiple front-ends, along with a configured set of device drivers. We've developed Erlang code to use Fermilab's TCLK event distribution clock and Erlang's interface to C/C++ allows hardware-specific driver access.  
slides icon Slides THDAUST02 [1.439 MB]  
 
THDAUST03 The FERMI@Elettra Distributed Real-time Framework 1267
 
  • L. Pivetta, G. Gaio, R. Passuello, G. Scalamera
    ELETTRA, Basovizza, Italy
 
  Funding: The work was supported in part by the Italian Ministry of University and Research under grants FIRB-RBAP045JF2 and FIRB-RBAP06AWK3
FERMI@Elettra is a Free Electron Laser (FEL) based on a 1.5 GeV linac. The pulsed operation of the accelerator and the necessity to characterize and control each electron bunch requires synchronous acquisition of the beam diagnostics together with the ability to drive actuators in real-time at the linac repetition rate. The Adeos/Xenomai real-time extensions have been adopted in order to add real-time capabilities to the Linux based control system computers running the Tango software. A software communication protocol based on gigabit Ethernet and known as Network Reflective Memory (NRM) has been developed to implement a shared memory across the whole control system, allowing computers to communicate in real-time. The NRM architecture, the real-time performance and the integration in the control system are described.
 
slides icon Slides THDAUST03 [0.490 MB]  
 
THDAULT04 Embedded Linux on FPGA Instruments for Control Interface and Remote Management 1271
 
  • B.K. Huang, R.M. Myers, R.M. Sharples
    Durham University, Durham, United Kingdom
  • G. Cunningham, G.A. Naylor
    CCFE, Abingdon, Oxon, United Kingdom
  • O. Goudard
    ESRF, Grenoble, France
  • J.J. Harrison
    Merton College, Oxford, United Kingdom
  • R.G.L. Vann
    York University, Heslington, York, United Kingdom
 
  Funding: This work was part-funded by the RCUK Energy Programme under grant EP/I501045 and the European Communities under the contract of Association between EURATOM and CCFE.
FPGAs are now large enough that they can easily accommodate an embedded 32-bit processor which can be used to great advantage. Running embedded Linux gives the user many more options for interfacing to their FPGA-based instrument, and in some cases this enables removal of the middle-person PC. It is now possible to manage the instrument directly by widely used control systems such EPICS or TANGO. As an example, on MAST (the Mega Amp Spherical Tokamak) at Culham Centre for Fusion Energy, a new vertical feedback system is under development in which waveform coefficients can be changed between plasma discharges to define the plasma position behaviour. Additionally it is possible to use the embedded processor to facilitate remote updating of firmware which, in combination with a watchdog and network booting ensures that full remote management over Ethernet is possible. We also discuss UDP data streaming using embedded Linux and a web based control interface running on the embedded processor to interface to the FPGA board.
 
slides icon Slides THDAULT04 [2.267 MB]  
 
THDAULT05 Embedded LLRF Controller with Channel Access on MicroTCA Backplane Interconnect 1274
 
  • K. Furukawa, K. Akai, T. Kobayashi, S. Michizono, T. Miura, K. Nakanishi, J.-I. Odagiri
    KEK, Ibaraki, Japan
  • H. Deguchi, K. Hayashi, M. Ryoshi
    Mitsubishi Electric TOKKI Systems, Amagasaki, Hyogo, Japan
 
  A low-level RF controller has been developed for the accelerator controls for SuperKEKB, Super-conducting RF Test facility (STF) and Compact-ERL (cERL) at KEK. The feedback mechanism will be performed on Vertex-V FPGA with 16-bit ADCs and DACs. The card was designed as an advanced mezzanine card (AMC) for a MicroTCA shelf. An embedded EPICS IOC on the PowerPC core in FPGA will provide the global controls through channel access (CA) protocol on the backplane interconnect of the shelf. No other mechanisms are required for the external linkages. CA is exclusively employed in order to communicate with central controls and with an embedded IOC on a Linux-based PLC for slow controls.  
slides icon Slides THDAULT05 [1.780 MB]  
 
THDAULT06 MARTe Framework: a Middleware for Real-time Applications Development 1277
 
  • A. Neto, D. Alves, B. Carvalho, P.J. Carvalho, H. Fernandes, D.F. Valcárcel
    IPFN, Lisbon, Portugal
  • A. Barbalace, G. Manduchi
    Consorzio RFX, Associazione Euratom-ENEA sulla Fusione, Padova, Italy
  • L. Boncagni
    ENEA C.R. Frascati, Frascati (Roma), Italy
  • G. De Tommasi
    CREATE, Napoli, Italy
  • P. McCullen, A.V. Stephen
    CCFE, Abingdon, Oxon, United Kingdom
  • F. Sartori
    F4E, Barcelona, Spain
  • R. Vitelli
    Università di Roma II Tor Vergata, Roma, Italy
  • L. Zabeo
    ITER Organization, St. Paul lez Durance, France
 
  Funding: This work was supported by the European Communities under the contract of Association between EURATOM/IST and was carried out within the framework of the European Fusion Development Agreement
The Multi-threaded Application Real-Time executor (MARTe) is a C++ framework that provides a development environment for the design and deployment of real-time applications, e.g. control systems. The kernel of MARTe comprises a set of data-driven independent blocks, connected using a shared bus. This modular design enforces a clear boundary between algorithms, hardware interaction and system configuration. The architecture, being multi-platform, facilitates the test and commissioning of new systems, enabling the execution of plant models in offline environments and with the hardware-in-the-loop, whilst also providing a set of non-intrusive introspection and logging facilities. Furthermore, applications can be developed in non real-time environments and deployed in a real-time operating system, using exactly the same code and configuration data. The framework is already being used in several fusion experiments, with control cycles ranging from 50 microseconds to 10 milliseconds exhibiting jitters of less than 2%, using VxWorks, RTAI or Linux. Codes can also be developed and executed in Microsoft Windows, Solaris and Mac OS X. This paper discusses the main design concepts of MARTe, in particular the architectural choices which enabled the combination of real-time accuracy, performance and robustness with complex and modular data driven applications.
 
slides icon Slides THDAULT06 [1.535 MB]