Ernest L. Williams Jr. Controls Commissioning October 2007 Controls Commissioning Facility Advisory Committee Meeting Oct 30, 2007 IOC Engineering Team Infrastructure and subsystems Integration Testing Lessons-Learned
Ernest L. Williams Jr. Controls Commissioning October 2007 EPICS Team Ernest Williams Kristi Luchini Stephanie Allison (50%) Till Strauman (50%) Arturo Alarcon Debbie Rogind (on loan from high-level apps) Dayle Kotturi Sheng Peng Doug Murray Stephen Schuh Diane Fairley (on loan from high-level apps) Judy Rock
Ernest L. Williams Jr. Controls Commissioning October 2007 Infrastructure and Subsystems Controls Network was simply a subnet of the Office enterprise network. One big Channel Access Domain Used different CA ports to segregate development and production network All files needed by the control system were on AFS. Easy access to files: development and production were in the same tree.
Ernest L. Williams Jr. Controls Commissioning October 2007 Infrastructure and Subsystems (Cont’d) Special AFS/NFS translator was used to provide NFS access to IOCs Computer Infrastructure: Boot Server based on the AFS/NFS translator. Daemon Server: ran softIOCs and ChannelWatcher Operator Consoles were managed by SCCS and used the AFS system as well. OS: RHEL WS 4 (linux-x86) EPICS: Version R
Ernest L. Williams Jr. Controls Commissioning October 2007 Infrastructure and Subsystems (Cont’d) IOC Infrastructure (EPICS: Version R ) Targets: MVME6100 (VME64-X) – rtems 4.7.x uC5282 (stand-alone) – rtems 4.7.x linux-x86 (soft-IOCs) – linux rhel-ws4 Serial port system to connect to all IOCs: Watch the boot process and/or debug IOC Remote power Cycle for all IOCs mostly in place. IOC Health Monitoring needs work.
Ernest L. Williams Jr. Controls Commissioning October 2007 Infrastructure and Subsystems (Cont’d) Types of IOCs SoftIOCs ran with high-reliability and uptime. Number of softIOCs ~12 ColdFire (uC5282) IOCs: BPMs are using the uC5282 and ran with high reliability especially with respect to Beam Synchronous Acquisition. (Qty ~26) LLRF are also using ColdFire and ran with good reliability. There is work to do to improve the beam synchronization acquisition. (Qty ~28)
Ernest L. Williams Jr. Controls Commissioning October 2007 Infrastructure and Subsystems (Cont’d) Types of IOCs VME-based systems worked fairly well this commissioning period. (Qty ~40) Issues related to timing receiver boards (i.e. are under investigation) Camera system vulnerable to timing glitches Magnet IOCs still not using EVR due to an issue with interrupts.
Ernest L. Williams Jr. Controls Commissioning October 2007 Subsystems
Ernest L. Williams Jr. Controls Commissioning October 2007 Integration Testing A plan was implemented to get more controls test time. Typically, the schedule seems to always squeeze the control systems test time. We instituted a plan to allocate one day every week for controls integration testing and checkout as part of the beam commissioning schedule. Integration testing was performed in close coordination and cooperation from operations and physics.
Ernest L. Williams Jr. Controls Commissioning October 2007 Lessons-Learned Improve IOC Infrastructure and Management Remote power cycle integration into EPICS for all IOCs Standardized Test Plans to improve integration testing and coordination. (Optimize Testing schedule) Move to production network to improve overall reliability and independence from office network. Beam Synchronous Acquisition is very important for LCLS and uses the same infrastructure required for fast feedback. This infrastructure will be completed for LLRF, Toroids, and Bunch Length. The systems that are using ColdFire + VME combination need to use a dedicated network and special driver for deterministic data transfer. Should not use Channel Access for this. Configuration Management via relational database in the works Not enough personnel resources to cover all systems.