Difference between revisions of "Controls-status-8-aug-2008"
From GlueXWiki
m |
m |
||
| Line 6: | Line 6: | ||
This layer controls the hardware and contains all control loops. Main components are an Allen-Bradley PLC and the hardware it controls, and manufacturer-supplied ethernet-accessible devices (e.g. CAEN mainframe). | This layer controls the hardware and contains all control loops. Main components are an Allen-Bradley PLC and the hardware it controls, and manufacturer-supplied ethernet-accessible devices (e.g. CAEN mainframe). | ||
| − | * | + | * A number of us met with Steve Lassiter, controls engineer. Discussed general aspects of A-B controllers, implementation strategy, hardware availability and costs, and related concerns. |
| − | * | + | * He has an unused small PLC from another manufacturer (not Siemans) we can use if we want. |
* General agreement that simple analog sensors/actuators (0-10V) and I/O modules are the best choice. | * General agreement that simple analog sensors/actuators (0-10V) and I/O modules are the best choice. | ||
* Initiated discussion with Accelerator on fast-feedback from active collimator to steering magnets. Meeting planned for later in Aug. | * Initiated discussion with Accelerator on fast-feedback from active collimator to steering magnets. Meeting planned for later in Aug. | ||
* Discussed possibility of purchasing A-B equipment early to test with various magnet power supplies. | * Discussed possibility of purchasing A-B equipment early to test with various magnet power supplies. | ||
| + | * Current plan is to run PLC and satellite racks on UPS power from the counting house. Exception may be Tagger area (due to long cable run may install rack-mount UPS there). | ||
| + | * Investigating possibility of running some beamline components on UPS power. | ||
| Line 17: | Line 19: | ||
This layer displays, archives, and alarms on data from the DCS layer. EPICS is our current choice for this layer, but as Elke has noted, we may not need EPICS as the functionality of the DCS and ECS layers grow towards each other. | This layer displays, archives, and alarms on data from the DCS layer. EPICS is our current choice for this layer, but as Elke has noted, we may not need EPICS as the functionality of the DCS and ECS layers grow towards each other. | ||
| − | * Nothing new to report. | + | * Nothing new to report. Continuing to periodically monitor progress in EPICS world. |
| Line 24: | Line 26: | ||
This layer integrates run control and all slow controls into a coherent system. AFECS from the JLab DAQ group is our current choice. This layer implements high-level state machines using the AFECS scripting language COOL. | This layer integrates run control and all slow controls into a coherent system. AFECS from the JLab DAQ group is our current choice. This layer implements high-level state machines using the AFECS scripting language COOL. | ||
| − | * Vardan has been testing and improving AFECS. | + | * Vardan has been testing and improving AFECS and COOL. |
| − | * Run control worked fine with about 1000 components | + | * Carl is optimizing cMsg networking strategy for scalability and to deal with firewalls. Should be ready soon. |
| − | * Recent successful test involved DAQ setup and a Wiener (?) mainframe communicating via SNMP. Disruption of the operation of the mainframe was detected by Run Control. | + | * Run control worked fine with about 1000 components using multiple servers, but some cMsg improvements were needed for scalability. |
| + | * Recent successful test involved a DAQ setup and a Wiener (?) mainframe communicating via SNMP. Disruption of the operation of the mainframe was detected by Run Control and it responded properly. COOL scripts implemented control logic. | ||
Revision as of 15:52, 7 August 2008
Recall I presented a three-tier architecture at the System review:
Device Control System - DCS
This layer controls the hardware and contains all control loops. Main components are an Allen-Bradley PLC and the hardware it controls, and manufacturer-supplied ethernet-accessible devices (e.g. CAEN mainframe).
- A number of us met with Steve Lassiter, controls engineer. Discussed general aspects of A-B controllers, implementation strategy, hardware availability and costs, and related concerns.
- He has an unused small PLC from another manufacturer (not Siemans) we can use if we want.
- General agreement that simple analog sensors/actuators (0-10V) and I/O modules are the best choice.
- Initiated discussion with Accelerator on fast-feedback from active collimator to steering magnets. Meeting planned for later in Aug.
- Discussed possibility of purchasing A-B equipment early to test with various magnet power supplies.
- Current plan is to run PLC and satellite racks on UPS power from the counting house. Exception may be Tagger area (due to long cable run may install rack-mount UPS there).
- Investigating possibility of running some beamline components on UPS power.
Supervisory Control and Data Acquisition - SCADA
This layer displays, archives, and alarms on data from the DCS layer. EPICS is our current choice for this layer, but as Elke has noted, we may not need EPICS as the functionality of the DCS and ECS layers grow towards each other.
- Nothing new to report. Continuing to periodically monitor progress in EPICS world.
Experiment Control System - ECS
This layer integrates run control and all slow controls into a coherent system. AFECS from the JLab DAQ group is our current choice. This layer implements high-level state machines using the AFECS scripting language COOL.
- Vardan has been testing and improving AFECS and COOL.
- Carl is optimizing cMsg networking strategy for scalability and to deal with firewalls. Should be ready soon.
- Run control worked fine with about 1000 components using multiple servers, but some cMsg improvements were needed for scalability.
- Recent successful test involved a DAQ setup and a Wiener (?) mainframe communicating via SNMP. Disruption of the operation of the mainframe was detected by Run Control and it responded properly. COOL scripts implemented control logic.
