OWG meeting 31-Jan-2008

From GlueXWiki
(Redirected from OWG meeting Jan-2008)
Jump to: navigation, search

Agenda


Time/Location

Note change from usual date/time: 10am Thurs 31-Jan-2008 CC F326
(Fri 1-Feb-2008 is tracking review all day)


To connect by telephone:

1) dial:
+1-800-377-8846 : US
+1-888-276-7715 : Canada
+1-302-709-8424 : International

2) enter participant code: 39527048# (remember the "#")


Announcements

Next Meeting

Fri 15-Feb-2008 2pm CC F326


New Action Items from this Meeting

  • Use full-scale FDC test as testbed for controls system.


Minutes

Attendees: Dave L, Dave A, Dave D, Elke, Vardan, Chris C, Fernando, Elliott.


Chris Cuevas showed the latest electronics diagrams

  • Chris showed some new drawings, many things moving well along
  • Chris showed the latest crate, FADC, and F1TDC counts
  • Lots of progress defining newer systems such as TS crate
  • ATCA may be an interesting alternative for the global sum crate (more on ATCA at the next meeting...)
  • Chris reviewed milestone status
  • Four more FADC250's coming, VXS testing continues
  • 12-bit is our standard for the FADC250
  • Now have some space in Stan's lab downstairs
  • TI design in progress
  • New hire starting to work on clock and trigger distribution
  • Collecting specs for F1TDC version 2
  • Crate sum design and prototyping going well
  • 12-bit FADC requires front-end redesign to use new chip...current one runs too hot
  • All drawings will follow Hall D official numbering scheme
  • Need network diagrams, location of patch panels, routes, etc.
  • More work needed on cable routes, rack layouts, etc. taking maintainability into account
  • Three crates per rack seems ok


Dave Lawrence reported for Eugene Chudakov on progress in the event size study

  • Low and high energy photon parts now combined in a single MC generator
  • Threshold to about 3 GeV comes from cross section tables, above 3 GeV is Pythia
  • MC includes coherent bremstrahlung spectrum
  • Still needs realistic trigger algorithm
  • Chinese group will arrive end of March and may work on event size study


Vardan Gurjyan told us about AFECS, the JLab DAQ group experiment controls framework

  • AFECS == Agent Framework for Experiment Controls Systems
  • Includes finite state machines (FSM's)
  • Allows interoperation of hardware and software in a control system
  • Allows integration of DAQ, run control, slow controls, etc.
  • Talking to a component involves talking to its agent using a uniform API for all agents
  • Allows scalability since only the agent needs a driver, and so some bottlenecks can be avoided
  • Allows hiearchical design including FSM's at any level
  • Supports many protocols (cMsg, CA, OPC, SNMP, etc)
  • Written in pure Java
  • Web-based monitoring component exists


Elliott presented preliminary choices for controls field busses and controller protocols

  • This is the bottom of a three-layer proposed controls hierarchy
  • Goal is to minimize number of field bus types, use Ethernet whenever possible
  • Allen-Bradley PLC will be used for solenoid, target, gas systems, experiment cooling systems, perhaps other systems
  • Operator gui's will have uniform look-and-feel, so Allen-Bradly HMI (gui builder) will only be used for expert screens
  • Elliott proposes extensive use of OPC (world standard for talking to controllers), but also SNMP and others
  • Will have CAN-bus and I2C
  • See Field Bus and Controller Specifications for full list of bus/controller choices


Elliott proposed Linux-only EPICS for the slow controls layer in the Hall D control system

  • Elliott attended two ICALEPCS conferences since 2002 and learned about possibilities there
  • Viable alternatives are: EPICS, PVSSII, DOOCS/TINE, and Tango
  • None appear to be compelling alternatives to EPICS at this time
  • Elliott proposed world-standard Linux-only EPICS for the middle control layer
  • EPICS seems to be rapidly developing, has embraced Java, and old utilities are being replaced with modern versions
  • EPICS community seems strong and growing, and is interested in modernization
  • Eclipse-based Control System Studio seems like a good alternative to MEDM, DM, etc, and allows for much improved functionality
  • Concerns about whether promising new development will pan out
  • See Slow Controls Framework Choice for more details