Realtime Control Tools in the PSI Accelerator Control System

The distributed control system for the PSI accelerators provides optimized input output functions for closed loop control. These basic functions are then used to build generic applications for global online beam control. Some of these applications, including the basic mechanism used, are explained. Questions about testing and the quality of these closed loop controls are discussed. Resulting performance, timing and real-time response in the real system and under normal load are presented. 1 General structure of control system The PSI Accelerator Control System [1] implements a distributed architecture. Frontend computers (FEC) provide logical access to all accelerator devices. Workstations serve as operator interfaces and minicomputers provide functions like central data storage and accelerator data logging. The whole communication is handled by a 10MBit Ethernet. 1.1 System performance The following diagram (fig.1) shows maximum achiev able values for input output (i/o) and communication bandwidth throughout the different levels of the control system. The estimation is based on the following distribution of i/o into different lists. 10% random i/o (5 devices per list), 40% repetitive i/o for average applications (10 devices per list) and 50% read i/o for display and monitoring tasks (50 devices per list in compressed format). Operation of the accelerator now loads the system to a level of about 10 to 15% of the theoretical Ethernet maximum load. 2 Control requirements We have a need for closed loop control in many different cases. To assist the operator in choosing optimal parameters, good feedback for process identification, optimization and troubleshooting is needed. The closed loops have to observe numerous boundary conditions on the loop parameters as well as other external parameters (interlock conditions). All parameters can be located anywhere in the control system. The described control applications for closed loop stabilization are generic applications that have evolved out of older specific programs. Based on the distibuted control system, they are independent of the underlying structure. Any combination of parameters can be used irrespectiveof the concerned Frontend. This application Workstation Workstation Workstation Ethernet 10 Mhz ( 90 % Load ) to next FEC to next WS ≤ 10 % ≤ 10 % ≤ 10 % FEC HPrt 743/64 Mhz FEC HPrt 743/64 Mhz FEC HPrt 743/64 Mhz 24 % 24 % 24 % 33k IO/s 7,8k IO/s 7,8k IO/s CAMAC Loop bit-serial 5MHz CAMAC Crate CAMAC Crate CAN ROAD-C Asynch. Interface Asynch. Interface 10 % 10 % 780 IO/s 780 IO/s 3,7k IO/s to Control Unit’s to Control Unit’s 25k IO/s ROAD-C ( 25 kHz | 24 bit parallel ) CAN, 1MHz to next CAMAC Crate to next CAMAC Crate Maximum Ethernet Performance : for one FEC System total 50 % 50 Devices/List (repetitive-compressed) 120 List è ⇒ 55 kByte 60900 Devices/s 40 % 10 Devices/List (repetitive) è 150 List ⇒ 45 kByte 14900 Devices/s 10 % 5 Devices/List (random I/O) è 30 List ⇒ 12 kByte 1500 Devices/s 33k IO/s 33k IO/s