Upload
zuzana
View
36
Download
2
Tags:
Embed Size (px)
DESCRIPTION
CLS Control System Overview. E. Matias Canadian Light Source University of Saskatchewan. Introduction ( § 1.0) CID Staff Goals Background Comparisons Major Projects Functionality ( § 2.0) Architecture ( § 3.0) OPI Software ( § 7.0) IOC Software ( § 8.0) Device Level ( § 9 & 10). - PowerPoint PPT Presentation
Citation preview
E. MatiasCanadian Light Source
University of Saskatchewan
CLS Control System Overview
Agenda
• Introduction (§1.0)– CID Staff– Goals– Background– Comparisons– Major Projects
• Functionality (§2.0)• Architecture (§3.0)• OPI Software (§7.0)• IOC Software (§8.0)• Device Level (§9 & 10)
• Other Interfaces (§11)• Timing System (§12)• Secondary Sys. (§13)• Tools (§14, 5, 6)• Servers (§15)• Networking (§16)• System Software (§17)• Anticipated Changes• Syn-apps• Conclusions
CID Department
Students – Intern (3)Students – Summer (1)
Controls & Instrumentation Development Department Manager
Elder Matias
Controls Systems Engineer/LeadRobby Tanner
RF TechnologistJonathan Stampe
Staff Scientist (Instrumentation)Johannes M. Vogt
Electronics TechnologistHeinz Buchmann
Control System Engineer/LeadMike Mckibben
Electronics TechnologistDenis Beauregard
Computer TechnologistCarl Finlay
Electrical Technologist/ForemanNeil Hovdestad
Systems Analyst/Control LeadsRuss Berg
Tony WilsonGlen Wright
Dioni Medrano (Term)
Control System Engineer/LeadHao Zhang
David Beauregard (Term)
Electrical Technologist Wayne McWillie
Doug Starns
RF TechnologistGlen PeggFuture (1)
Instrumentation Technologist Carl Jansen
Carmen Britton
Software Specialist (EFD)Ru Igarashi
Electrical Engineers/Controls Lead (ETS)Neil JohnsonWade Dolton
Goals
• Review our existing design.• Identify areas that we can improve.• Identify functionality that is needed.• Identify options that we should explore.• The architecture should be:
– Maintainable– Flexible– Reliable– Support our functional requirements– Low cost to maintain and support (labour)
• Once a system is deployed and commissioned we provide on-call support during scheduled beam time.
Background
• SAL (1964 to 1999)– Linear accelerator, EROS ring, Experimental Areas 1, 2
and 3– Control system based on CAMAC, Digital PDP, VAX and
Sun with some NeXT and Amiga computers.– LUCID data acquisition program
• CLS (1999 ….)– Highly distributed system (extensive use of RTEMS, single
board computers, and optical links to VME crates)– Based on EPICS– Diagnostics beamlines (2)– phase I beamlines (7)– phase II beamlines (7)– phase III beamlines (TBD)
Making Comparisons
• Scope– Initial build project $5 M (equip)
+ labour– In operations approximately $2 M /year
+ special projects
• Comparisons– Larger Staffing Level than MAXLab– Smaller Staffing Level than APS
• Good Comparisons– SPEAR 3? - similar parameters – BESSY? – similar size– other mid. size regional/national machines?
• SAC, and others compare us to APS; is this reasonable? ….
SafetyInterlocks
(17)
BeamlineControls
(11)
Controls(26)
RF(20)
PowerSystems
(13)
Diagnostics(17)
BuildingServices
(?)
Conv.Facilities
(?)
APS = 110CLS = 20
AccelSoftware
(??)
Leveraging Our Size
• Perhaps APS and CLS are not the best comparison.
• We need to take advantage of our size– Innovative, decisive, cross-disciplinary– Leverage technology and design choices across systems
• Technically evaluate our design choices• Adopt best-practice, from both within and outside of
the EPICS and synchrotron community• Move more towards a “software engineering”
paradigm– Provincial Legislation for “certification” of system analysts -
software engineers became law in 2005, as a first step towards licensing
– The “software engineering” methods are intended to reduce development costs and improve quality
Major Projects
• Phase I and Accelerator Operational Support• Phase II beamlines• CANARIE Remote Access• Infrastructure
– VME Monitor Program Upgrade (Russ, Neil J. + Intern)– CS-Studio Tree Explorer (Glen W. + Intern)– Data Acquisition Program Upgrade (Glen W, & Ru)– Next Generation Motor Control (Mike M. & Tony W.)
• Upgrades– Diagnostics Kicker, X-ray BPM and time resolve (Johannes)– Linac Gun &RF (Neil J. & Hao)– Linac ACIS System (Robby)– Replace remaining pre-1980 controls (Hao + summer student)
Agenda
• Introduction (§1.0)– CID Staff– Goals– Background– Comparisons– Major Projects
• Functionality (§2.0)• Architecture (§3.0)• OPI Software (§7.0)• IOC Software (§8.0)• Device Level (§9 & 10)
• Other Interfaces (§11)• Timing System (§12)• Secondary Sys. (§13)• Tools (§14, 5, 6)• Servers (§15)• Networking (§16)• System Software (§17)• Anticipated Changes• Syn-apps• Conclusions
VLANs for: each beamline, machine control, development, office, visitors
VME Crate
(Reflective Memory)
MicroStep
EROCIOC
RTEMS
FieldDev. RS-232
Devices
OPI
Linux
IOCStep Controller
RTEMS
Motors
MicroStep
OPI
Linux
OPI
Linux
Touch PanelOPI Linux
NetworkServer
(bootp, dhcp,auto restore)
Linux
DataArchiveServer
Linux
AlarmServer
MS-Win
MS-SQLServer
MS-Win
PowerEdgeIOC
Linux
PS BoardsIOC
RTEMS
PowerSupplies
EROCIOC
RTEMS
FieldDev.
EthernetDevices
PLC & GPIB
FieldDev.
MagnetsMotors
1GigBridge
IOC
Linux
FieldDev.
ProfibusPLC
System Architecture (3.0)
Agenda
Turning to the design document….• Introduction (§1.0)
– CID Staff– Goals– Background– Comparisons– Major Projects
• Functionality (§2.0)• Architecture (§3.0)• OPI Software (§7.0)• IOC Software (§8.0)• Device Level (§9 & 10)
• Other Interfaces (§11)• Timing System (§12)• Secondary Sys. (§13)• Tools (§14, 5, 6)• Servers (§15)• Networking (§16)• System Software (§17)• Anticipated Changes• Syn-apps• Conclusions
Tools
• PV Crawler
• Cables Database
• Sequencer
• MKS Source Integrity
• CS Studio
• Rational Software
Network
Office
Gateway
MS-SQL
50
BAOPI
AlarmPrinter
OPI
IOC
IOC
IOC
PLC
OPI
IOC
IOC
Server Subnet
Office Subnet
Beamline BBeamline A
Main Control Subnet
Syn-apps
• Considered in 2001– reviewed as part of EPICS fact finding – part of the EPICS versus Altersys ECR
• Considered (partially) in 2004– an option for Mono-control
• Considered in Feb. 2005– see memo from E. Matias.
• Considered again in Nov. 2005– see ECR
The Next Step…• Deploy MKS Problem Tracking and the New MKS
– This should provide more responsive and predictable operational support– More advanced and configurable access control
• Participate in Building and Deploying CS Studio– This tool will provide the ability to permit users and beamline scientist more direct
access to configure PV variables – Provide automation for the generation of configuration files
(db files, alarm handler, gateway, data archiver, auto-save-restore, edm screens)– Better support for re-factoring
• Exploit Lessons Learned from CANARIE Project– First major project based strongly based on UML– First major project to make extensive use of CASE tools– First major project to use Java and Web Services paradigm
• Enhance VME Monitor Program– Improved performance– Improved data collection
• Improve Data Acquisition Program
• Move to Next Generation Motor Control (perhaps based on ESRF design?)
• Improve capability for time resolve and diagnostics
Development Process
E. Matias
Inception Elaboration Construction Transition
Requirements Analysis Design Implementation Test
Unified Process
Core Workflows In Each Phase
- Establish Feasibility- Establish Business/Scientific Case- Capture Essential Requirements- Identify Critical Risks- Establish initial budget & schedule
Deliverables (as required):- Project Plan- Risk Assessment- Initial Requirements (10-20%)
Requirements: - Refine System ScopeAnalysis: - Establish what to buildDesign:- Create an ArchitectureImplementation- Build an architecture baseline- Build any prototypesTest- Test the architecture baseline- Test any prototypes
Deliverables (as required):- PID Drawings- Wiring Diagrams- Updated Requirements Document- System breakdown
Requirements: - Uncover missing requirementsAnalysis: - Finish the analyis Design:- Finish detailed designImplementation- Build and install the systemTest:- Test and ring-out the system
Deliverables (as required):- Running System
- Correct any defected- Provide support for commissioning - Prepare final documentation
Deliverables (as required):- Final documentation- Working system
CLSI System (Beamline) Engineering Process
ProposalConceptual
DesignPreliminary
DesignDetailedDesign
Build Commissioning
Some recommendations?
• Should we get rid of requirement gathering?– If anything the reverse is true, we need to improve our requirements capture
process…• Should we get rid of P&ID drawings?
– They have become critical to our design process. We are moving to add more not less information to P&IDs….
• Should we get rid of Wiring Diagrams?– Without them (or something equivalent) we can’t maintain a system this size with
the staff we have. • Should we get rid of design standards?
– They have been critical to being able to maintain a system this size with the staffing levels we have.
• Should we get rid of ECRs & ECOs?– As part of our license renewal applications the CNSC was concerned we did not
give them sufficient detail and asked for more.• Should we remove access controls?
– For the most part, the beamline administrator accounts provide a great deal of control.
– Difficult to justify after trying to undo changes in the middle of the night.
Winter at the CLS
Moving Forward
• Current CLS Control System– Built on a common design (circa 2000 and 2002)– Homogonous – Common structure and design across the facility– Built on 6 years of EPICS experience – Built on SAL (30 years) of accelerator and nuclear physics science
• Critical Questions– Does it represent best industry practice?– How can we improve quality (scientific capability and the user experience)?– How can we improve efficiency?– Is it safe and ethical?
• How Do We Answer These Questions?– Collectively (cross-disciplinary)– Open to new ideas and methods– Invite people from inside/outside of EPICS to help us….– Invite people from inside/outside synchrotrons to help us…– Built on our in-house experts…– Decisions to change are technically driven with backup (ECR)– Exploit automation to reduce costs and increase reliability
• Shift in approach, from just-in-time make-it-workbuilding well designed, structured reusable applicationssupport highly configurable applications configured just-in-time