1. Accelerator control at iThemba LABS

2. Some background No formal reliability procedures Cost considerations SSC operational 24/7 Shutdown total of 2 months/year Equipment often unavailable during shutdown Want it working NOW Inadequate testing time Reliable delivery of beam

3. Some history Cyclotron control systems originally designed (late 70s) around a few mini-computers (HP 1000s running RTE) Control electronics and instrumentation interfaced via CAMAC Lab-built interactive devices (joysticks, set- point units, etc)

4. Some more history Control system migrated to distributed PC- based system in the early 90s Distributed memory-resident tables of control variables Originator node computers controlling interface electronics maintain their own control variables The console and other nodes requiring access to these variables link to this originator node Adequate diagnostic messages for debugging and more efficient fault-finding Communication over Ethernet LAN Development of in-house interfaces (SABUS)

5. Contributions to control system reliability Control system migrated to distributed PC- based system running OS/2 If a particular node computer fails, other nodes are not affected Minimal preventative maintenance All fans and filters checked, cleaned and/or replaced during annual major shutdown Daily archiving of control system database Communication over Ethernet LAN The original CSMA/CD Ethernet is a “passive” broadcast bus which is resilient to most node failures

6. Contributions to control system reliability Development of in-house “simple” interfaces (SABUS) to control electronics Simple, robust, noise-immune 8-bit parallel differential bus connecting up to 15 SABUS crates per PC controller card Each crate contains up to 13 cards, each card controlling between 2 (e.g., power supplies) and 64 (e.g., relays) pieces of equipment An assortment of I/O cards developed using readily-available components Simple bus design allows for easier maintenance and development, easier migration to new/upgraded operating system and longevity of hardware

7. Contributions to control system reliability Design for graceful degradation Highest level of control at the console nodes Control can devolve down to the instrumentation interface nodes with each having a graphical control screen for the variables originated on that node Enables convenient testing of variables and their associated interface electronics Many hardware interfaces can be switched into local mode and controlled from front panel

8. Improvements to control system infrastructure Network with 1Gb/s fibre-optic backbone to distributed managed switches VLAN with private IP addresses (restricted routing to campus VLAN) Nodes assembled from good quality components (motherboards, CPUs, RAM, disk drives, power supplies, etc) Adequate cooling (particularly of disk drives) On-site stock of spares (PC components, interface modules, etc) Back-up clones of critical disk drives for fast replacement following failures Most nodes and I/O module crates powered via individual UPSs

9. Current control system developments Migrate control system onto EPICS platform Mature stable code Active development in, and support from, a number of similar international labs Many useful utilities available in EPICS (logging, archiving, alarming, etc.) Run old and EPICS-based subsystems in parallel Retain hardware (SABUS) interfaces

10. EPICS migration roadmap Cyclotrons – all new developments on EPICS platform (for example, beam splitter control) Develop gateway between old table-based control variables and EPICS process variables Port old subsystems onto EPICS over time Electrostatic accelerators’ controls at both lab sites (Cape and Gauteng) have successfully migrated almost 100% to EPICS

11. EPICS-based VDG vacuum control

12. Improve resilience of cooling, power and network infrastructure Eskom Transformer 1 Transformer 2 Diesel Generator UPS

13. High-intensity beam project Increase 66MeV proton beam intensity up to 500µA New vertical beam target station to accommodate high intensities Tighter control required to reduce possibilities of equipment damage and other safety issues Flattop systems for SPC1 injector and SSC cyclotrons Non-destructive beam position monitors Continuous beam position monitoring and automatic alignment Beam splitter to deliver beams simultaneously to two radioisotope production targets

14. Control diagnostics for high-intensity beam Target ruptures and other damage

15. Initial analysis of damage Autoradiography of targets

16. Simulation of target charge distribution

17. Search for instabilities

18. Control diagnostics for high-intensity beam Urgent development of control diagnostics to increase protection Beam halo monitors to detect stray beam in high-energy beam line Real-time analysis of beam distribution on production target Monitoring of RF systems for instabilities