Controls Meeting 2-Jun-2011

From GlueXWiki
Jump to: navigation, search



9:30 AM Thurs 1-Jun-2011 CC F326


Next Meeting

9:30 AM Thurs 15-Jun-2011 CC F326


Attendees: Elliott W, Beni Z, David L, Hovanes E.

Coil Test

Elliott briefly summarized results from the recent coil 3 test that were relevant to controls. Basically, the control system worked quite well.

One minor problem was an apparent network interruption that required cycling power to a number of PLC chassis as well as a network hub. This led to a discussion on the relative merits of ControlNet vs EtherNet/IP to connect chassis together. The former requires a special daisy-chained cable between chassis, the latter just needs an Ethernet connection. ControlNet cables are copper-based (a special version of RG-58), and we are trying to avoid copper connections between the hall and counting house, although this can be avoided if additional PLC processors are used. Ethernet systems are vulnerable to network switch problems.

Currently we are planning an Ethernet/IP system, but we might consider ControlNet for certain applications.

Target Control System

Hovanes presented draft P&I diagrams from Chris Keith. His design is very simple: all valves are manually operated, and it mostly consists of one vacuum pump, a large Hydrogen reservoir and a cold plate connected to a self-contained refridgerator. It has very few moving parts. The system is passive...once you clean fill it with H2 gas, lowering the cold plate temperature fills the target, and raising it empties the target. Changing the cold plate temperature is about all the control system has to do, and this is done via RS232 or via Ethernet (not sure).

FDC and CDC Gas System Controls

Hovanes described an EPICS IOC and gave a demo of an operator GUI that control the Brooks 0254 4-channel gas mixing system, part of FDC/CDC gas system. The system operates autonomously, all you need to do is set some parameters via RS232. Hovanes suggests we mount a PC104 next to the controls box to perform the RS232 control. A basic PC104 costs around $250.

Hovanes used CSS (DESY, ORNL) and BOY (ORNL) to develop the control GUI. These use the Java/Eclipse development framework, and Javascript is used for customization. The GUI is specified in XML, which is interpreted at runtime by CSS/BOY. You do not need to know Java to develop GUI's with this system. One concern is that CSS/BOY uses a lot of memory, but this should not be a problem for the computers we purchase.

The custom IOC (see his slides) running in the PC104 implements a "continuously scanning buffer layer" to avoid problems with synchronous control of the Brooks controller. Here EPICS reads/writes values into a local buffer, and the buffer contents are read/sent to the controller asynchronously. This model has been used successfully in Hall B, and may be a good model for our FCAL CAN control system.

P6 Tasks

Elliott noted that all the "Plan Monitoring" tasks started 1-Jun.