ICALEPCS2013 - List of Keywords (real-time)

Paper	Title	Other Keywords	Page
MOCOAAB07	Real Time Control for KAGAR, 3km Cryogenic Gravitational Wave Detector in Japan	controls, network, EPICS, feedback	23
	O. Miyakawa ICRR, Chiba, Japan
	KAGRA is a 3km cryogenic interferometer for gravitational wave detection located underground Kamioka-mine in Japan. The next generation large scale interferometric gravitational wave detectors require very complicated control topologies for the optical path length between mirrors, and very low noise feedback controls in order to detect an extremely tiny motion between mirrors excited by gravitational waves. The interferometer consists of a Michelson interferometer with Fabry-Perot cavities on its arms, and other two mirrors as, so called, a power recycling and a resonant sideband extraction technique. In total, 5 degrees of freedom for length between 7 mirrors should be controlled at a time, and the control must be continuously kept during the observation of gravitational waves. We are currently developing a real time controls system using computers for KAGRA. In this talk, we report how the control system works.
	Slides MOCOAAB07 [8.536 MB]

MOPPC028	High-Density Power Converter Real-Time Control for the MedAustron Synchrotron	controls, timing, operation, FPGA	127
	J. Gutleber, A.B. Brett, M. Hager, J. Junuzovic, M. Junuzovic, M. Marchhart, R. Moser, H. Pavetits, C. Torcato de Matos CERN, Geneva, Switzerland A. Ambrosch, A.B. Brett, P. Fraboulet, M. Hager, J. Junuzovic, M. Junuzovic, M. Marchhart, R. Moser, H. Pavetits, C. Torcato de Matos EBG MedAustron, Wr. Neustadt, Austria J. Dedič, M. Mehle, L. Šepetavc Cosylab, Ljubljana, Slovenia
	The MedAustron accelerator is a synchrotron for light-ion therapy, developed under the guidance of CERN within the MedAustron-CERN collaboration. Procurement of 7 different power converter families and development of the control system were carried out concurrently. Control is optimized for unattended routine clinical operation. Therefore, finding a uniform control solution was paramount to fulfill the ambitious project plan. Another challenge was the need to operate with about 5'000 cycles initially, achieving pipelined operation with pulse-to-pulse re-configuration times smaller than 250 msec. This contribution shows the architecture and design and gives an overview of the system as built and operated. It is based on commercial-off-the-shelf processing hardware at front-end level and on the CERN function generator design at equipment level. The system is self contained, permitting use of parts and the whole is other accelerators. Especially the separation of the power converter from the real-time regulation using CERN's Converter Regulation Board makes this approach an attractive choice for integrating existing power converters in new configurations.
	Poster MOPPC028 [0.892 MB]

MOPPC123	Extending WinCC OA for Use as Accelerator Control System Core	controls, ion, interface, status	395
	M. Marchhart, A.B. Brett, M. Hager, M. Janulis, J. Junuzovic, M. Junuzovic, R. Moser, H. Pavetits, C. Torcato de Matos EBG MedAustron, Wr. Neustadt, Austria A.B. Brett, J. Gutleber, M. Hager, M. Janulis, J. Junuzovic, M. Junuzovic, M. Marchhart, R. Moser, C. Torcato de Matos CERN, Geneva, Switzerland J. Dedič Cosylab, Ljubljana, Slovenia
	The accelerator control system for the MedAustron light-ion medical particle accelerator has been designed under the guidance of CERN in the scope of an EBG MedAustron/CERN collaboration agreement. The core is based on the SIMATIC WinCC OA SCADA tool. Its open API and modular architecture permitted CERN to extend the product with features that go beyond traditional supervisory control and that are vital for directly operating a particle accelerator. Several extensions have been introduced to make WinCC OA fit for accelerator control: (1) Near real-time data visualization, (2) external application launch and monitoring, (3) accelerator settings snapshot and consistent restore, (4) generic panel navigation supporting role based permission handling, (5) native integration with interactive 3D engineering visualization, (6) integration with National Instruments based front-end controllers. The major drawback identified is the lack of support of callbacks from C++ extensions. This prevents asynchronous functions, multithreaded implementations and soft real-time behaviour. We are therefore striving to search for support in the user community to trigger the implementation of this function.
	Poster MOPPC123 [0.656 MB]

MOPPC124	Optimizing EPICS for Multi-Core Architectures	EPICS, controls, Linux, software	399
	R. Lange, F. Di Maio ITER Organization, St. Paul lez Durance, France R. Lange HZB, Berlin, Germany
	Funding: Work supported by German Bundesministerium für Bildung und Forschung and Land Berlin. EPICS is a widely used software framework for real-time controls in large facilities, accelerators and telescopes. Its multithreaded IOC (Input Output Controller) Core software has been developed on traditional single-core CPUs. The ITER project will use modern multi-core CPUs, running the RHEL Linux operating system in its MRG-R real-time variant. An analysis of the thread handling in IOC Core shows different options for improving the performance and real-time behavior, which are discussed and evaluated. The implementation is split between improvements inside EPICS Base, which have been merged back into the main distribution, and a support module that makes full use of these new features. This paper describes design and implementation aspects, and presents results as well as lessons learned.
	Poster MOPPC124 [0.448 MB]

MOPPC128	Real-Time Process Control on Multi-Core Processors	controls, framework, operation, software	407
	M. Ishii, Y. Furukawa, T. Matsumoto JASRI/SPring-8, Hyogo-ken, Japan
	A real-time control is an essential for a low level RF and timing system to have beam stability in the accelerator operation. It is difficult to optimize priority control of multiple processes with real-time class and time-sharing class on a single-core processor. For example, we can’t log into the operating system if a real-time class process occupies the resource of a single-core processor. Recently multi-core processors have been utilized for equipment controls. We studied the process control of multiple processes running on multi-core processors. After several tunings, we confirmed that an operating system could run stably under heavy load on multi-core processors. It would be possible to achieve real-time control required milliseconds order response under the fast control system such as an event synchronized data acquisition system. Additionally we measured the response performance between client and server processes using MADOCA II framework that is the next-generation MADOCA. In this paper we present about the tunings for real-time process control on multi-core processors and performance results of MADOCA II.
	Poster MOPPC128 [0.450 MB]

TUPPC029	Integration, Processing, Analysis Methodologies and Tools for Ensuring High Data Quality and Rapid Data Access in the TIM* Monitoring System	monitoring, database, controls, data-acquisition	615
	A. Suwalska, M. Brightwell, M. Bräger, E. Koufakis, R. Martini, P. Sollander CERN, Geneva, Switzerland
	Processing, storing and analysing large amounts of real-time data is a challenge for every monitoring system. The performance of the system strongly depends on high quality configuration data and the ability of the system to cope with data anomalies. The Technical Infrastructure Monitoring system (TIM) addresses data quality issues by enforcing a workflow of strict procedures to integrate or modify data tag configurations. TIM’s data acquisition layer architecture allows real-time analysis and rejection of irrelevant data. The discarded raw data 90,000,000 transactions/day) are stored in a database, then purged after gathering statistics. The remaining operational data (2,000,000 transactions/day) are transferred to a server running an in-memory database, ensuring its rapid processing. These data are currently stored for 30 days allowing ad hoc historical data analysis. In this paper we describe the methods and tools used to guarantee the quality of configuration data and highlight the advanced architecture that ensures optimal access to operational data as well as the tools used to perform off-line data analysis. * Technical Infrastructure Monitoring system
	Poster TUPPC029 [0.742 MB]

TUPPC057	New Development of EPICS-based Data Acquisition System for Electron Cyclotron Emission Diagnostics in KSTAR Tokamak	controls, electron, EPICS, diagnostics	699
	T.G. Lee, K.D. Lee, S. Lee, W.R. Lee, M.K. Park NFRI, Daejon, Republic of Korea
	Korea Superconducting Tokamak Advanced Research (KSTAR) will be operated in the 6nd campaign in 2013 after achievement of first plasma in 2008. Many diagnostic devices have been installed for measuring the various plasma properties in the KSTAR tokamak during the campaigns. From the first campaign, a data acquisition system of Electron Cyclotron Emission (ECE) Heterodyne Radiometer (HR) has been operated to measure the radial profile of electron temperature. The DAQ system at the beginning was developed with a VME-form factor digitizer in Linux OS platform. However, this configuration had some limitations that it could not acquire over 100,000 samples per second due to its unstable operation during the campaigns. In order to overcome these weak points, a new ECE HR DAQ system is under development with a cPCI-form factor in Linux OS platform and the main control application will be developed based on EPICS framework like other control systems installed in KSTAR. Besides solving the described problems main advantages of the new ECE HR DAQ system are capabilities of calculating plasma electron temperature during plasma shot and displaying it in run-time.
	Poster TUPPC057 [1.286 MB]

TUPPC062	High-Speed Data Acquisition of Sensor Signals for Physical Model Verification at CERN HiRadMat (SHC-DAQ)	data-acquisition, hardware, LabView, software	718
	C. Charrondière, M. Guinchard, S. Marques Dos Santos CERN, Geneva, Switzerland
	A high-speed data acquisition system was successfully developed and put into production in a harsh radiation environment in a couple of months to test new materials impacted by proton beams for future use in beam intercepting devices. A 4 MHz ADC with high impedance and low capacitance was used to digitize the data at a 2 MHz bandwidth. The system requirements were to design a full speed data streaming on a trigger during up to 30 ms and then reconfigure the hardware in less than 500 ms to perform a 100 Hz acquisition for 30 seconds. Experimental data were acquired, using LabVIEW real-time, relying on extensive embedded instrumentation (strain gauges and temperature sensors) and on acquisition boards hosted on a PXI crate. The data acquisition system has a dynamic range and sampling rate that are sufficient to acquire the very fast and intense shock waves generated by the impact. This presentation covers the requirements, the design, development and commissioning of the system. The overall performance, user experience and preliminary results will be reported.
	Poster TUPPC062 [9.444 MB]

TUPPC083	FPGA Implementation of a Digital Constant Fraction for Fast Timing Studies in the Picosecond Range	detector, FPGA, neutron, timing	774
	P. Mutti, J. Ratel, F. Rey, E. Ruiz-Martinez ILL, Grenoble, France
	Thermal or cold neutron capture on different fission systems is an excellent method to produce a variety of very neutron-rich nuclei. Since neutrons at these energies bring in the reaction just enough energy to produce fission, the fragments remain neutron-rich due to the negligible neutron evaporation thus allowing detailed nuclear structure studies. In 2012 and 2013 a combination of EXOGAM, GASP and Lohengrin germanium detectors has been installed at the PF1B cold neutron beam of the Institut Laue-Langevin. The present paper describes the digital acquisition system used to collect information on all gamma rays emitted by the decaying nuclei. Data have been acquired in a trigger-less mode to preserve a maximum of information for further off-line treatment with a total throughput of about 10 MByte/sec. Special emphasis is devoted to the FPGA implementation of an on-line digital constant fraction algorithm allowing fast timing studies in the pico second range.
	Poster TUPPC083 [9.928 MB]

TUPPC133	Graphene: A Java Library for Real-Time Scientific Graphs	controls, operation, interface, background	901
	G. Carcassi, K. Shroff BNL, Upton, Long Island, New York, USA
	While there are a number of open source charting library available in Java, none of them seem to be suitable for real time scientific data, such as the one coming from control systems. Common shortcomings include: inadequate performance, too entangled with other scientific packages, concrete data object (which require copy operations), designed for small datasets, required running UI to produce any graph. Graphene is our effort to produce graphs that are suitable for scientific publishing, can be created without UI (e.g. in a web server), work on data defined through interfaces that allow no copy processing in a real time pipeline and are produced with adequate performance. The graphs are then integrated using pvmanager within Control System Studio.
	Poster TUPPC133 [0.502 MB]

TUPPC134	Pvmanager: A Java Library for Real-Time Data Processing	controls, framework, EPICS, background	903
	G. Carcassi, K. Shroff BNL, Upton, Long Island, New York, USA
	Increasingly becoming the standard connection layer in Control System Studio, pvmanager is a Java library that allows to create well behaved applications that process real time data, such as the one coming from a control system. It takes care of the caching, queuing, rate decoupling and throttling, connection sharing, data aggregation and all the other details needed to make an application robust. Its fluent API allows to specify the detail for each pipeline declaratively in a compact way.
	Poster TUPPC134 [0.518 MB]

WECOBA04	Effective End-to-end Management of Data Acquisition and Analysis for X-ray Photon Correlation Spectroscopy	detector, photon, experiment, status	1004
	F. Khan, J.P. Hammonds, S. Narayanan, A. Sandy, N. Schwarz ANL, Argonne, USA
	Funding: Work supported by U.S. Department of Energy, Office of Science, under Contract No. DE-AC02-06CH11357. Low latency between data acquisition and analysis is of critical importance to any experiment. The combination of a faster parallel algorithm and a data pipeline for connecting disparate components (detectors, clusters, file formats) enabled us to greatly enhance the operational efficiency of the x-ray photon correlation spectroscopy experiment facility at the Advanced Photon Source. The improved workflow starts with raw data (120 MB/s) streaming directly from the detector camera, through an on-the-fly discriminator implemented in firmware to Hadoop’s distributed file system in a structured HDF5 data format. The user then triggers the MapReduce-based parallel analysis. For effective bookkeeping and data management, the provenance information and reduced results are added to the original HDF5 file. Finally, the data pipeline triggers user specific software for visualizing the data. The whole process is completed shortly after data acquisition – a significant improvement of operation over previous setup. The faster turn-around time helps scientists to make near real-time adjustments to the experiments.
	Slides WECOBA04 [9.540 MB]

THPPC014	CMX - A Generic Solution to Expose Monitoring Metrics in C and C++ Applications	monitoring, controls, diagnostics, operation	1118
	F. Ehm, Y. Fischer, G.M. Gorgogianni, S. Jensen, P. Jurcso CERN, Geneva, Switzerland
	CERN’s Accelerator Control System is built upon a large number of C, C++ and Java services that are required for daily operation of the accelerator complex. The knowledge of the internal state of these processes is essential for problem diagnostic as well as for constant monitoring for pre-failure recognition. The CMX library follows similar principles as JMX (Java Management Extensions) and provides similar monitoring capabilities for C and C++ applications. It allows registering and exposing runtime information as simple counters, floating point numbers or character data. This can be subsequently used by external diagnostics tools for checking thresholds, sending alerts or trending. CMX uses shared-memory to ensure non-blocking read/update actions, which is an important requirement for real-time processes. This paper introduces the topic of monitoring C/C++ applications and presents CMX as a building block to achieve this goal.
	Poster THPPC014 [0.795 MB]

THPPC102	Comparison of Synchronization Layers for Design of Timing Systems	timing, interface, network, Ethernet	1296
	A. Aulin Söderqvist, N. Claesson, J. Neves Rodrigues Lund University, Lund, Sweden J. Dedič, R. Štefanič, R. Tavčar Cosylab, Ljubljana, Slovenia
	Two synchronization layers for timing systems in large experimental physics control systems are compared. White Rabbit (WR), which is an emerging standard, is compared against the well-established event based approach. Several typical timing system services have been implemented on an FPGA using WR to explore its concepts and architecture, which is fundamentally different from an event based. Both timing system synchronization layers were evaluated based on typical requirements of current accelerator projects and with regard to other parameters such as scalability. The proposed design methodology demonstrates how WR can be deployed in future accelerator projects.
	Poster THPPC102 [1.796 MB]

THPPC104	A Timing System for Cycle Based Accelerators	timing, software, LabView, hardware	1303
	J. Gutleber CERN, Geneva, Switzerland Z. Croflic, J. Dedič, R. Štefanič Cosylab, Ljubljana, Slovenia
	Synchrotron accelerators with multiple ion sources and beam lines require a high degree of flexibility to define beam cycle timing sequences. We have therefore decided to design a ready-to-use accelerator timing system based on off-the-shelf hardware and software that can fit mid-size accelerators and that is easy to adapt to specific user needs. This Real Time Event Distribution Network (REDNet) has been developed under the guidance of CERN within the MedAustron-CERN collaboration. The system based on the MRF transport layer has been implemented by Cosylab. While we have used hardware on NI PXIe platform, it is straightforward to obtain it for other platforms such as VME. The following characteristics are key to its readiness for use: (1) turn-key system comprising hardware, transport layer, application software and open integration interfaces, (2) performance suitable for a wide range of accelerators, (3) multiple virtual timing systems in one physical box, (4) documentation developed according to V-model. Given the maturity of the development, we have decided to make REDNet available as a product through our industrial partner.
	Poster THPPC104 [0.429 MB]

THPPC115	Fast Orbit Feedback Implementation at Alba Synchrotron	software, FPGA, hardware, target	1328
	X. Serra-Gallifa, S. Blanch-Torné, D. Fernández-Carreiras, A. Gutierrez-Milla, Z. Martí, O. Matilla, J. Moldes, A. Olmos, R. Petrocelli CELLS-ALBA Synchrotron, Cerdanyola del Vallès, Spain
	After the successful accelerator commissioning and with the facility already in operation one of the top short term objectives pointed out by accelerator division was the Fast Orbit Feedback implementation (FOFB). The target of the FOFB system is to hold the electron beam position at submicron range both in vertical and horizontal planes correcting the inestabilities up to 120Hz. This increased beam stability performance is considered a major asset for the beamlines user operation. To achieve this target, the orbit position is acquired from the 88 Libera BPMs at a 10KHz sampling rate, distributed through an independent network and the corrections are calculated and sent to the 176 power supplies that drive the corrector coils. All this correction loop is executed at 10 KHz and the total latency of the system is characterized and minimized optimizing the bandwidth response.
	Poster THPPC115 [0.732 MB]

THPPC125	Evaluation and Implementation of Advanced Process Control with the compactRIO Material of National Instrument	controls, FPGA, LabView, feedback	1355
	G. Maire, A. Kehrli, M. Pezzetti, S. Ravat CERN, Geneva, Switzerland B. Charnier, H. Coppier ESIEE, Amiens, France
	Programmable Logic Controller (PLC) is very commonly used in many industries and research applications for process control. However a very complex process control may require algorithms and performances beyond the capability of PLCs, very high-speed or precision controls may also require other solutions. This paper describes recent research conducted to implement advanced process controls with the cRIO material from National Instruments (decoupling of MIMO process control, steady state feedback, observer, Kalman filter, etc). The cRIO systems consist of an embedded real-time controller for communication and processing, a Reconfigurable Field Programmable Array (FPGA) and hot-swappable I/O modules. The paper presents experimental results and the ability of the cRIO to treat complex process control.
	Poster THPPC125 [1.004 MB]

FRCOAAB05	JOGL Live Rendering Techniques in Data Acquisition Systems	GPU, detector, controls, experiment	1477
	C. Cocho, F. Cecillon, A. Elaazzouzi, Y. Le Goc, J. Locatelli, P. Mutti, H. Ortiz, J. Ratel ILL, Grenoble, France
	One of the major challenges in instrument control is to provide a fast and scientifically correct representation of the data collected by the detector through the data acquisition system. Despite the availability nowadays of a large number of excellent libraries for off-line data plotting, the real-time 2D and 3D data rendering still suffers of performance issues related namely to the amount of information to be displayed. The current paper describes new methods of image generation (rendering) based on JOGL library used for data acquisition at the Institut Laue-Langevin (ILL) on instruments that require either high image resolution or large number of images rendered at the same time. These new methods involve the definition of data buffers and the usage of the GPU memory, technique known as Vertex Buffer Object (VBO). Implementation of different modes of rendering, on-screen and off-screen, will be also detailed.
	Slides FRCOAAB05 [1.422 MB]

FRCOBAB03	The New Multicore Real-time Control System of the RFX-mod Experiment	controls, plasma, Linux, framework	1493
	G. Manduchi, A. Luchetta, C. Taliercio Consorzio RFX, Associazione Euratom-ENEA sulla Fusione, Padova, Italy
	The real-time control system of RFX-mod nuclear fusion experiment has been in operation since 2004 and has been used to control the plasma position and the MagnetoHydroDinamic (MHD) modes. Over time new and more computing demanding control algorithm shave been developed and the system has been pushed to its limits. Therefore a complete re-design has been carried out in 2012. The new system adopts radically different solutions in Hardware, Operating System and Software management. The VME PowerPc CPUs communicating over Ethernet used in the former system have been replaced by a single multicore server. The VxWorks Operating System , previously used in the VME CPUs has now been replaced by Linux MRG, that proved to behave very well in real-time applications. The previous framework for control and communication has been replaced by MARTe, a modern framework for real-time control gaining interest in the fusion community. Thanks to the MARTe organization, a rapid development of the control system has been possible. In particular, its intrinsic simulation ability of the framework gave us the possibility of carrying out most debugging in simulation, without affecting machine operation.
	Slides FRCOBAB03 [1.301 MB]

Paper

Title

Other Keywords

Page

MOCOAAB07

Real Time Control for KAGAR, 3km Cryogenic Gravitational Wave Detector in Japan

controls, network, EPICS, feedback

23

O. Miyakawa
ICRR, Chiba, Japan

KAGRA is a 3km cryogenic interferometer for gravitational wave detection located underground Kamioka-mine in Japan. The next generation large scale interferometric gravitational wave detectors require very complicated control topologies for the optical path length between mirrors, and very low noise feedback controls in order to detect an extremely tiny motion between mirrors excited by gravitational waves. The interferometer consists of a Michelson interferometer with Fabry-Perot cavities on its arms, and other two mirrors as, so called, a power recycling and a resonant sideband extraction technique. In total, 5 degrees of freedom for length between 7 mirrors should be controlled at a time, and the control must be continuously kept during the observation of gravitational waves. We are currently developing a real time controls system using computers for KAGRA. In this talk, we report how the control system works.

Slides MOCOAAB07 [8.536 MB]

MOPPC028

High-Density Power Converter Real-Time Control for the MedAustron Synchrotron

controls, timing, operation, FPGA

127

J. Gutleber, A.B. Brett, M. Hager, J. Junuzovic, M. Junuzovic, M. Marchhart, R. Moser, H. Pavetits, C. Torcato de Matos
CERN, Geneva, Switzerland
A. Ambrosch, A.B. Brett, P. Fraboulet, M. Hager, J. Junuzovic, M. Junuzovic, M. Marchhart, R. Moser, H. Pavetits, C. Torcato de Matos
EBG MedAustron, Wr. Neustadt, Austria
J. Dedič, M. Mehle, L. Šepetavc
Cosylab, Ljubljana, Slovenia

The MedAustron accelerator is a synchrotron for light-ion therapy, developed under the guidance of CERN within the MedAustron-CERN collaboration. Procurement of 7 different power converter families and development of the control system were carried out concurrently. Control is optimized for unattended routine clinical operation. Therefore, finding a uniform control solution was paramount to fulfill the ambitious project plan. Another challenge was the need to operate with about 5'000 cycles initially, achieving pipelined operation with pulse-to-pulse re-configuration times smaller than 250 msec. This contribution shows the architecture and design and gives an overview of the system as built and operated. It is based on commercial-off-the-shelf processing hardware at front-end level and on the CERN function generator design at equipment level. The system is self contained, permitting use of parts and the whole is other accelerators. Especially the separation of the power converter from the real-time regulation using CERN's Converter Regulation Board makes this approach an attractive choice for integrating existing power converters in new configurations.

Poster MOPPC028 [0.892 MB]

MOPPC123

Extending WinCC OA for Use as Accelerator Control System Core

controls, ion, interface, status

395

M. Marchhart, A.B. Brett, M. Hager, M. Janulis, J. Junuzovic, M. Junuzovic, R. Moser, H. Pavetits, C. Torcato de Matos
EBG MedAustron, Wr. Neustadt, Austria
A.B. Brett, J. Gutleber, M. Hager, M. Janulis, J. Junuzovic, M. Junuzovic, M. Marchhart, R. Moser, C. Torcato de Matos
CERN, Geneva, Switzerland
J. Dedič
Cosylab, Ljubljana, Slovenia

The accelerator control system for the MedAustron light-ion medical particle accelerator has been designed under the guidance of CERN in the scope of an EBG MedAustron/CERN collaboration agreement. The core is based on the SIMATIC WinCC OA SCADA tool. Its open API and modular architecture permitted CERN to extend the product with features that go beyond traditional supervisory control and that are vital for directly operating a particle accelerator. Several extensions have been introduced to make WinCC OA fit for accelerator control: (1) Near real-time data visualization, (2) external application launch and monitoring, (3) accelerator settings snapshot and consistent restore, (4) generic panel navigation supporting role based permission handling, (5) native integration with interactive 3D engineering visualization, (6) integration with National Instruments based front-end controllers. The major drawback identified is the lack of support of callbacks from C++ extensions. This prevents asynchronous functions, multithreaded implementations and soft real-time behaviour. We are therefore striving to search for support in the user community to trigger the implementation of this function.

Poster MOPPC123 [0.656 MB]

MOPPC124

Optimizing EPICS for Multi-Core Architectures

EPICS, controls, Linux, software

399

R. Lange, F. Di Maio
ITER Organization, St. Paul lez Durance, France
R. Lange
HZB, Berlin, Germany

Funding: Work supported by German Bundesministerium für Bildung und Forschung and Land Berlin.
EPICS is a widely used software framework for real-time controls in large facilities, accelerators and telescopes. Its multithreaded IOC (Input Output Controller) Core software has been developed on traditional single-core CPUs. The ITER project will use modern multi-core CPUs, running the RHEL Linux operating system in its MRG-R real-time variant. An analysis of the thread handling in IOC Core shows different options for improving the performance and real-time behavior, which are discussed and evaluated. The implementation is split between improvements inside EPICS Base, which have been merged back into the main distribution, and a support module that makes full use of these new features. This paper describes design and implementation aspects, and presents results as well as lessons learned.

Poster MOPPC124 [0.448 MB]

MOPPC128

Real-Time Process Control on Multi-Core Processors

controls, framework, operation, software

407

M. Ishii, Y. Furukawa, T. Matsumoto
JASRI/SPring-8, Hyogo-ken, Japan

A real-time control is an essential for a low level RF and timing system to have beam stability in the accelerator operation. It is difficult to optimize priority control of multiple processes with real-time class and time-sharing class on a single-core processor. For example, we can’t log into the operating system if a real-time class process occupies the resource of a single-core processor. Recently multi-core processors have been utilized for equipment controls. We studied the process control of multiple processes running on multi-core processors. After several tunings, we confirmed that an operating system could run stably under heavy load on multi-core processors. It would be possible to achieve real-time control required milliseconds order response under the fast control system such as an event synchronized data acquisition system. Additionally we measured the response performance between client and server processes using MADOCA II framework that is the next-generation MADOCA. In this paper we present about the tunings for real-time process control on multi-core processors and performance results of MADOCA II.

Poster MOPPC128 [0.450 MB]

TUPPC029

Integration, Processing, Analysis Methodologies and Tools for Ensuring High Data Quality and Rapid Data Access in the TIM* Monitoring System

monitoring, database, controls, data-acquisition

615

A. Suwalska, M. Brightwell, M. Bräger, E. Koufakis, R. Martini, P. Sollander
CERN, Geneva, Switzerland

Processing, storing and analysing large amounts of real-time data is a challenge for every monitoring system. The performance of the system strongly depends on high quality configuration data and the ability of the system to cope with data anomalies. The Technical Infrastructure Monitoring system (TIM) addresses data quality issues by enforcing a workflow of strict procedures to integrate or modify data tag configurations. TIM’s data acquisition layer architecture allows real-time analysis and rejection of irrelevant data. The discarded raw data 90,000,000 transactions/day) are stored in a database, then purged after gathering statistics. The remaining operational data (2,000,000 transactions/day) are transferred to a server running an in-memory database, ensuring its rapid processing. These data are currently stored for 30 days allowing ad hoc historical data analysis. In this paper we describe the methods and tools used to guarantee the quality of configuration data and highlight the advanced architecture that ensures optimal access to operational data as well as the tools used to perform off-line data analysis.
* Technical Infrastructure Monitoring system

Poster TUPPC029 [0.742 MB]

TUPPC057

New Development of EPICS-based Data Acquisition System for Electron Cyclotron Emission Diagnostics in KSTAR Tokamak

controls, electron, EPICS, diagnostics

699

T.G. Lee, K.D. Lee, S. Lee, W.R. Lee, M.K. Park
NFRI, Daejon, Republic of Korea

Korea Superconducting Tokamak Advanced Research (KSTAR) will be operated in the 6nd campaign in 2013 after achievement of first plasma in 2008. Many diagnostic devices have been installed for measuring the various plasma properties in the KSTAR tokamak during the campaigns. From the first campaign, a data acquisition system of Electron Cyclotron Emission (ECE) Heterodyne Radiometer (HR) has been operated to measure the radial profile of electron temperature. The DAQ system at the beginning was developed with a VME-form factor digitizer in Linux OS platform. However, this configuration had some limitations that it could not acquire over 100,000 samples per second due to its unstable operation during the campaigns. In order to overcome these weak points, a new ECE HR DAQ system is under development with a cPCI-form factor in Linux OS platform and the main control application will be developed based on EPICS framework like other control systems installed in KSTAR. Besides solving the described problems main advantages of the new ECE HR DAQ system are capabilities of calculating plasma electron temperature during plasma shot and displaying it in run-time.

Poster TUPPC057 [1.286 MB]

TUPPC062

High-Speed Data Acquisition of Sensor Signals for Physical Model Verification at CERN HiRadMat (SHC-DAQ)

data-acquisition, hardware, LabView, software

718

C. Charrondière, M. Guinchard, S. Marques Dos Santos
CERN, Geneva, Switzerland

A high-speed data acquisition system was successfully developed and put into production in a harsh radiation environment in a couple of months to test new materials impacted by proton beams for future use in beam intercepting devices. A 4 MHz ADC with high impedance and low capacitance was used to digitize the data at a 2 MHz bandwidth. The system requirements were to design a full speed data streaming on a trigger during up to 30 ms and then reconfigure the hardware in less than 500 ms to perform a 100 Hz acquisition for 30 seconds. Experimental data were acquired, using LabVIEW real-time, relying on extensive embedded instrumentation (strain gauges and temperature sensors) and on acquisition boards hosted on a PXI crate. The data acquisition system has a dynamic range and sampling rate that are sufficient to acquire the very fast and intense shock waves generated by the impact. This presentation covers the requirements, the design, development and commissioning of the system. The overall performance, user experience and preliminary results will be reported.

Poster TUPPC062 [9.444 MB]

TUPPC083

FPGA Implementation of a Digital Constant Fraction for Fast Timing Studies in the Picosecond Range

detector, FPGA, neutron, timing

774

P. Mutti, J. Ratel, F. Rey, E. Ruiz-Martinez
ILL, Grenoble, France

Thermal or cold neutron capture on different fission systems is an excellent method to produce a variety of very neutron-rich nuclei. Since neutrons at these energies bring in the reaction just enough energy to produce fission, the fragments remain neutron-rich due to the negligible neutron evaporation thus allowing detailed nuclear structure studies. In 2012 and 2013 a combination of EXOGAM, GASP and Lohengrin germanium detectors has been installed at the PF1B cold neutron beam of the Institut Laue-Langevin. The present paper describes the digital acquisition system used to collect information on all gamma rays emitted by the decaying nuclei. Data have been acquired in a trigger-less mode to preserve a maximum of information for further off-line treatment with a total throughput of about 10 MByte/sec. Special emphasis is devoted to the FPGA implementation of an on-line digital constant fraction algorithm allowing fast timing studies in the pico second range.

Poster TUPPC083 [9.928 MB]

TUPPC133

Graphene: A Java Library for Real-Time Scientific Graphs

controls, operation, interface, background

901

G. Carcassi, K. Shroff
BNL, Upton, Long Island, New York, USA

While there are a number of open source charting library available in Java, none of them seem to be suitable for real time scientific data, such as the one coming from control systems. Common shortcomings include: inadequate performance, too entangled with other scientific packages, concrete data object (which require copy operations), designed for small datasets, required running UI to produce any graph. Graphene is our effort to produce graphs that are suitable for scientific publishing, can be created without UI (e.g. in a web server), work on data defined through interfaces that allow no copy processing in a real time pipeline and are produced with adequate performance. The graphs are then integrated using pvmanager within Control System Studio.

Poster TUPPC133 [0.502 MB]

TUPPC134

Pvmanager: A Java Library for Real-Time Data Processing

controls, framework, EPICS, background

903

G. Carcassi, K. Shroff
BNL, Upton, Long Island, New York, USA

Increasingly becoming the standard connection layer in Control System Studio, pvmanager is a Java library that allows to create well behaved applications that process real time data, such as the one coming from a control system. It takes care of the caching, queuing, rate decoupling and throttling, connection sharing, data aggregation and all the other details needed to make an application robust. Its fluent API allows to specify the detail for each pipeline declaratively in a compact way.

Poster TUPPC134 [0.518 MB]

WECOBA04

Effective End-to-end Management of Data Acquisition and Analysis for X-ray Photon Correlation Spectroscopy

detector, photon, experiment, status

1004

F. Khan, J.P. Hammonds, S. Narayanan, A. Sandy, N. Schwarz
ANL, Argonne, USA

Funding: Work supported by U.S. Department of Energy, Office of Science, under Contract No. DE-AC02-06CH11357.
Low latency between data acquisition and analysis is of critical importance to any experiment. The combination of a faster parallel algorithm and a data pipeline for connecting disparate components (detectors, clusters, file formats) enabled us to greatly enhance the operational efficiency of the x-ray photon correlation spectroscopy experiment facility at the Advanced Photon Source. The improved workflow starts with raw data (120 MB/s) streaming directly from the detector camera, through an on-the-fly discriminator implemented in firmware to Hadoop’s distributed file system in a structured HDF5 data format. The user then triggers the MapReduce-based parallel analysis. For effective bookkeeping and data management, the provenance information and reduced results are added to the original HDF5 file. Finally, the data pipeline triggers user specific software for visualizing the data. The whole process is completed shortly after data acquisition – a significant improvement of operation over previous setup. The faster turn-around time helps scientists to make near real-time adjustments to the experiments.

Slides WECOBA04 [9.540 MB]

THPPC014

CMX - A Generic Solution to Expose Monitoring Metrics in C and C++ Applications

monitoring, controls, diagnostics, operation

1118

F. Ehm, Y. Fischer, G.M. Gorgogianni, S. Jensen, P. Jurcso
CERN, Geneva, Switzerland

CERN’s Accelerator Control System is built upon a large number of C, C++ and Java services that are required for daily operation of the accelerator complex. The knowledge of the internal state of these processes is essential for problem diagnostic as well as for constant monitoring for pre-failure recognition. The CMX library follows similar principles as JMX (Java Management Extensions) and provides similar monitoring capabilities for C and C++ applications. It allows registering and exposing runtime information as simple counters, floating point numbers or character data. This can be subsequently used by external diagnostics tools for checking thresholds, sending alerts or trending. CMX uses shared-memory to ensure non-blocking read/update actions, which is an important requirement for real-time processes. This paper introduces the topic of monitoring C/C++ applications and presents CMX as a building block to achieve this goal.

Poster THPPC014 [0.795 MB]

THPPC102

Comparison of Synchronization Layers for Design of Timing Systems

timing, interface, network, Ethernet

1296

A. Aulin Söderqvist, N. Claesson, J. Neves Rodrigues
Lund University, Lund, Sweden
J. Dedič, R. Štefanič, R. Tavčar
Cosylab, Ljubljana, Slovenia

Two synchronization layers for timing systems in large experimental physics control systems are compared. White Rabbit (WR), which is an emerging standard, is compared against the well-established event based approach. Several typical timing system services have been implemented on an FPGA using WR to explore its concepts and architecture, which is fundamentally different from an event based. Both timing system synchronization layers were evaluated based on typical requirements of current accelerator projects and with regard to other parameters such as scalability. The proposed design methodology demonstrates how WR can be deployed in future accelerator projects.

Poster THPPC102 [1.796 MB]

THPPC104

A Timing System for Cycle Based Accelerators

timing, software, LabView, hardware

1303

J. Gutleber
CERN, Geneva, Switzerland
Z. Croflic, J. Dedič, R. Štefanič
Cosylab, Ljubljana, Slovenia

Synchrotron accelerators with multiple ion sources and beam lines require a high degree of flexibility to define beam cycle timing sequences. We have therefore decided to design a ready-to-use accelerator timing system based on off-the-shelf hardware and software that can fit mid-size accelerators and that is easy to adapt to specific user needs. This Real Time Event Distribution Network (REDNet) has been developed under the guidance of CERN within the MedAustron-CERN collaboration. The system based on the MRF transport layer has been implemented by Cosylab. While we have used hardware on NI PXIe platform, it is straightforward to obtain it for other platforms such as VME. The following characteristics are key to its readiness for use: (1) turn-key system comprising hardware, transport layer, application software and open integration interfaces, (2) performance suitable for a wide range of accelerators, (3) multiple virtual timing systems in one physical box, (4) documentation developed according to V-model. Given the maturity of the development, we have decided to make REDNet available as a product through our industrial partner.

Poster THPPC104 [0.429 MB]

THPPC115

Fast Orbit Feedback Implementation at Alba Synchrotron

software, FPGA, hardware, target

1328

X. Serra-Gallifa, S. Blanch-Torné, D. Fernández-Carreiras, A. Gutierrez-Milla, Z. Martí, O. Matilla, J. Moldes, A. Olmos, R. Petrocelli
CELLS-ALBA Synchrotron, Cerdanyola del Vallès, Spain

After the successful accelerator commissioning and with the facility already in operation one of the top short term objectives pointed out by accelerator division was the Fast Orbit Feedback implementation (FOFB). The target of the FOFB system is to hold the electron beam position at submicron range both in vertical and horizontal planes correcting the inestabilities up to 120Hz. This increased beam stability performance is considered a major asset for the beamlines user operation. To achieve this target, the orbit position is acquired from the 88 Libera BPMs at a 10KHz sampling rate, distributed through an independent network and the corrections are calculated and sent to the 176 power supplies that drive the corrector coils. All this correction loop is executed at 10 KHz and the total latency of the system is characterized and minimized optimizing the bandwidth response.

Poster THPPC115 [0.732 MB]

THPPC125

Evaluation and Implementation of Advanced Process Control with the compactRIO Material of National Instrument

controls, FPGA, LabView, feedback

1355

G. Maire, A. Kehrli, M. Pezzetti, S. Ravat
CERN, Geneva, Switzerland
B. Charnier, H. Coppier
ESIEE, Amiens, France

Programmable Logic Controller (PLC) is very commonly used in many industries and research applications for process control. However a very complex process control may require algorithms and performances beyond the capability of PLCs, very high-speed or precision controls may also require other solutions. This paper describes recent research conducted to implement advanced process controls with the cRIO material from National Instruments (decoupling of MIMO process control, steady state feedback, observer, Kalman filter, etc). The cRIO systems consist of an embedded real-time controller for communication and processing, a Reconfigurable Field Programmable Array (FPGA) and hot-swappable I/O modules. The paper presents experimental results and the ability of the cRIO to treat complex process control.

Poster THPPC125 [1.004 MB]

FRCOAAB05

JOGL Live Rendering Techniques in Data Acquisition Systems

GPU, detector, controls, experiment

1477

C. Cocho, F. Cecillon, A. Elaazzouzi, Y. Le Goc, J. Locatelli, P. Mutti, H. Ortiz, J. Ratel
ILL, Grenoble, France

One of the major challenges in instrument control is to provide a fast and scientifically correct representation of the data collected by the detector through the data acquisition system. Despite the availability nowadays of a large number of excellent libraries for off-line data plotting, the real-time 2D and 3D data rendering still suffers of performance issues related namely to the amount of information to be displayed. The current paper describes new methods of image generation (rendering) based on JOGL library used for data acquisition at the Institut Laue-Langevin (ILL) on instruments that require either high image resolution or large number of images rendered at the same time. These new methods involve the definition of data buffers and the usage of the GPU memory, technique known as Vertex Buffer Object (VBO). Implementation of different modes of rendering, on-screen and off-screen, will be also detailed.

Slides FRCOAAB05 [1.422 MB]

FRCOBAB03

The New Multicore Real-time Control System of the RFX-mod Experiment

controls, plasma, Linux, framework

1493

G. Manduchi, A. Luchetta, C. Taliercio
Consorzio RFX, Associazione Euratom-ENEA sulla Fusione, Padova, Italy

The real-time control system of RFX-mod nuclear fusion experiment has been in operation since 2004 and has been used to control the plasma position and the MagnetoHydroDinamic (MHD) modes. Over time new and more computing demanding control algorithm shave been developed and the system has been pushed to its limits. Therefore a complete re-design has been carried out in 2012. The new system adopts radically different solutions in Hardware, Operating System and Software management. The VME PowerPc CPUs communicating over Ethernet used in the former system have been replaced by a single multicore server. The VxWorks Operating System , previously used in the VME CPUs has now been replaced by Linux MRG, that proved to behave very well in real-time applications. The previous framework for control and communication has been replaced by MARTe, a modern framework for real-time control gaining interest in the fusion community. Thanks to the MARTe organization, a rapid development of the control system has been possible. In particular, its intrinsic simulation ability of the framework gave us the possibility of carrying out most debugging in simulation, without affecting machine operation.

Slides FRCOBAB03 [1.301 MB]