Difference between revisions of "Field Bus and Controller Specifications"
m |
m |
||
Line 25: | Line 25: | ||
'''Hall D Subsystem Slow Controls Specifications''' | '''Hall D Subsystem Slow Controls Specifications''' | ||
− | Notes: OPC | + | Notes: OPC/xxx means OPC server connecting to the device via the xxx field bus instead of Ethernet. TCP/IP and SNMP run over Ethernet. CAN and I2C each have their own connectors and physical layer. |
Line 37: | Line 37: | ||
|gas systems||Allen-Bradley PLC | |gas systems||Allen-Bradley PLC | ||
|- | |- | ||
− | |high voltage||OPC | + | |high voltage||OPC |
|- | |- | ||
− | |low voltage||OPC | + | |low voltage||OPC or SNMP |
|- | |- | ||
− | |VXS crates||OPC | + | |VXS crates||OPC or SNMP |
|- | |- | ||
|Cockroft-Walton bases||CAN | |Cockroft-Walton bases||CAN | ||
|- | |- | ||
− | |goniomenter||OPC | + | |goniomenter||OPC or Allen-Bradley PLC |
|- | |- | ||
|network devices||SNMP | |network devices||SNMP |
Revision as of 14:34, 29 January 2008
Introduction
There are innumerable possible choices for sensors, actuators, etc. for use in Hall D, independent of the slow controls framework chosen (EPICS, PVSS, DOOCS/TINE, etc). These devices communicate using one of large number of control network systems (CAN, I2C, DEVICENET, ProfiBus, WorldFIP, TCP/IP, etc). For practical reasons we need to minimize the number of network types we deploy in the hall.
Thus in the following I propose a minimal set of control networks for use in Hall D. If an application needs a special sensor or controller that is not compatible with the set below, please contact me.
Note on PLC usage in Hall D
PLC's (Programmable Logic Controllers) are highly reliable process controllers that are in widespread use in industry, and are routinely used in critical applications. They are not general purpose computers, and instead have a single purpose in life, to run a control loop at regular intervals. Although other labs use them routinely, JLab does not, and I'm not sure why (speculation: JLab is too EPICS-happy, and early EPICS systems could not communicate with PLC's). PLC's are typically programmed by trained controls engineers, not by scientists, as special languages and techniques are used (similar to how FPGA programming is done by specially trained electrical engineers).
JLab typically uses PLC's in cryo-control systems, and we plan to use one for our solenoid (Allen-Bradley, model TBD). We expect that the solenoid cryo-control system will use up less than one third of the PLC capacity, so we can use the PLC for other purposes. Note that PLC control loops can be strictly prioritized, so that e.g. solenoid control will not be compromised by a lower priority control loop.
The PLC is ideal to control the following systems:
- solenoid
- cryo-target
- all gas systems
An open question is what other systems might be controlled by the PLC. As programming the PLC takes some time and requires a trained engineer, it may not be appropriate for many systems developed by physicists, particularly simple ones with benign failure modes. Further, we have to understand the cost of setting up a PLC-compatible system at a remote institution (Tim and I are working on this).
I believe Allen-Bradley PLC's talk to their connected devices via DEVICENET, which is a layer on top of CAN-bus. I'm not sure this is the only possibility...
Hall D Subsystem Slow Controls Specifications
Notes: OPC/xxx means OPC server connecting to the device via the xxx field bus instead of Ethernet. TCP/IP and SNMP run over Ethernet. CAN and I2C each have their own connectors and physical layer.
Subsystem | Control System |
solenoid | Allen-Bradley PLC |
cryo-target | Allen-Bradley PLC |
gas systems | Allen-Bradley PLC |
high voltage | OPC |
low voltage | OPC or SNMP |
VXS crates | OPC or SNMP |
Cockroft-Walton bases | CAN |
goniomenter | OPC or Allen-Bradley PLC |
network devices | SNMP |
misc temp, etc. | CAN and I2C |