Next: The CARMA Control System
Up: Control System
Previous: Using OpenOffice as a Portable Interface to JAVA-Based Applications
Table of Contents -
Subject Index -
Author Index -
Search -
PS reprint -
PDF reprint
Axelrod, T. A. & De La Peña, M. D. 2003, in ASP Conf. Ser., Vol. 314 Astronomical Data
Analysis Software and Systems XIII, eds. F. Ochsenbein, M. Allen, & D. Egret (San Francisco: ASP), 704
An Overview of the Large Binocular Telescope Control System
T.A. Axelrod and M.D. De La Peña
Large Binocular Telescope Observatory,
University of Arizona, Tucson, AZ 85721
Abstract:
The Large Binocular Telescope (LBT) consists of two 8.4-meter mirrors on a
common mount. This configuration provides the light gathering power equivalent
to an 11.8-meter telescope and the resolving power of an 22.8-meter telescope.
Due to the binocular nature of the telescope, there are unique requirements
imposed on the telescope control system (TCS) to ensure the health and
safety of the telescope, while also enabling observations in both independent
and interferometric mode.
This paper presents an overview of the design of the TCS, from the graphical
user interface (GUI) and the plotting and analysis capabilities at the highest
interaction level to the low-level interfaces for the software.
A logical view of the LBT TCS subsystems, represented by rectangles,
and the controllers, represented by rounded-corner rectangles, is shown
in Figure 1.
Subsystems are autonomous entities which control attached telescope
hardware through one or more interfaces. In contrast, controllers are
autonomous entities without attached hardware. The connecting lines in
Figure 1 represent the major communication paths
between components.
From the physical point of view, the TCS
has three support (control, global memory, and telemetry) and an
instrument network, all of 1 Gb/s capacity.
Figure 1:
TCS Logical Overview of Subsystems and Controllers
(Axelrod 2003)
 |
The control network is used for shipping commands through the
system. The global memory network is used to maintain a
globally accessible data structure which contains complete state
information for the system. This global or shared memory is
replicated on each platform, and therefore, is robust to individual
subsystem failures. Telemetry streams can originate in any subsystem
and are stored in the telemetry archive. Certain subsystems can require
high bandwidth (MCS servo loops, AO loops) which cannot be
accommodated on the other networks, raising the need for
a dedicated telemetry network. The instrument network
services the requirements of the instruments and storage of the acquired
data.
This software layer provides the fundamental foundation upon which the
TCS is constructed and functions. This layer is comprised of a series
of services and structures which are accessed by the higher levels of
the system.
- System Data Dictionary and Network Shared Memory
All system commands, variables and their corresponding attributes, and
ancillary information are defined in a system data dictionary which
enforces the concept of a single information source, allowing for a
greater level of system integrity. Telescope
status is organized by subsystem and maintained in a segment of
network shared memory specifically designated for the subsystem.
Collectively, the subsystem segments comprise a ``telescope-wide
shared memory'' where each segment is propagated to all TCS platforms for
a contemporaneous update of all the platforms.
- Event Subsystem
Events are the means of recording actions, as well as errors or abnormal
conditions in the
system. This mechanism provides a means to both generate events and to be
notified via a callback when a particular event has occurred elsewhere in
the system. Notification of events is only sent to subsystems which have
subscribed to receive such events; the subsystem decides on the appropriate
action to handle the event. All events are stored in a central logging
archive for diagnostic purposes.
- Telemetry Streams
Subsystems may register telemetry streams, sets of time-sampled variables,
which are delivered over the telemetry network to an archive.
- Customized RPC Library The library was created to support a
multi-threaded environment. Since communication is accomplished via XML
strings, only strings need to be accommodated by the library.
- Time Service
A GPS synchronized time server will be present on the LBT ethernet
and will offer time services to any client via ntp. Since
the system design is intended to keep the ethernet load light, ntp
should easily provide time accurate to 1 msec.
Each subsystem has a dedicated GUI, created with Qt Designer, which is
comprised of a main window and ancillary support panels and dialogs.
The LBT TCS GUI guidelines are defined by De La Peña (2003).
Each GUI is an independent executable and
- never communicates directly with its corresponding subsystem or
to any hardware,
- never communicates with GUIs for any other subsystem,
- issues its commands as customized RPC calls to the command
sequencer (CSQ),
- receives a handle from the CSQ which it uses to poll the status of
the issued command,
- obtains subsystem state and other critical information from the
network shared memory, and
- registers itself with the event handling system for direct
communication of alarms.
The GUIs have been designed such that they
are instantiated in their ``design-time'' form ( disabled).
Only when status is successfully obtained from networked shared
memory is the GUI enabled for use.
A GUI can be aborted and restarted without any interference to/from the TCS.
Each subsystem GUI has exclusive use of a monitor, giving rise
to a TCS display wall. The display wall provides individuals in the
control room simultaneous state information for all of the telescope
subsystems. Finally, a standalone Matlab GUI application provides
capabilities for plotting telemetry streams.
Command and control of the telescope is done primarily through a series
of GUIs which communicate directly with a centralized controller or command
sequencer (CSQ).
Commanding is accomplished by sending XML strings using a
customized remote procedure call (RPC) library. The customized
library properly supports a multi-threaded environment and
only needs to accommodate the transfer of strings. The controller
which initiates the command maintains a handle from the receiving system
which can be used to determine the status of the command. System status
(e.g., running, stopped) and state (nominal, alarm) are read from network
shared memory.
The primary responsibilities of the CSQ are to validate incoming requests and
to route the requests to the appropriate subsystems. Since a single command
can blossom into a series of necessary actions which need to be assigned
to a variety of the telescope subsystems (e.g., mount control, pointing
control), a petri net technique is employed by the CSQ to handle the
sequencing. Petri nets are well suited to model concurrent and discrete
events, and therefore, are ideal for control system specification.
While subsystems control specific hardware and unique characteristics,
all subsystems share common properties. These properties consist of:
autonomous operation, heartbeat, remote reboot, subsystem network
interfaces, subsystem API, and subsystem template. A few of these
properties are discussed here.
- Autonomous Operation
For testing purposes, each subsystem must be able
to work autonomously. While the entire TCS depends upon the common
software layer, a minimal version of the common software should be
able to support autonomous operation of a subsystem. The minimal
common software has the following capabilities:
manage subsystem configuration
via the data dictionary and configuration files, log events to local
log files, collect telemetry streams to a local archive, and use customized
RPC services to communicate with local or remote clients.
- Heartbeat
As the TCS is a distributed system, a software implemented heartbeat is
used to monitor the health of all the subsystem nodes. Each subsystem
provides a ``heartbeat'' function which returns with minimal latency a
system health code. In the event of an abnormal health code, the
controllers monitoring the heartbeat, GUI and CSQ, will alert the
telescope operator and client (respectively) that the subsystem should
be restarted.
- Subsystem API
The subsystem API consists of two major components: commanding and
getting/setting values. Commands are issued as customized RPC calls,
and a handle can be retrieved from the server such that the commanding
subsystem can monitor the progress of the issued command and continue to
perform other tasks. All subsystems support a minimal set of commands
which accommodate an orderly startup and shutdown of the subsystem.
All values which can be configured and all read-only state values for a
subsystem comprise the keyword interface. The keywords define the
external state of the subsystem, and are stored in the LBT System Data
Dictionary as discussed in Section 2. The configurable values are
accessible via accessor or modifier functions.
References
Axelrod, T. 2003, LBT CAN 481s001
De La Peña, M.D. 2003, LBT CAN 481s010a
© Copyright 2004 Astronomical Society of the Pacific, 390 Ashton Avenue, San Francisco, California 94112, USA
Next: The CARMA Control System
Up: Control System
Previous: Using OpenOffice as a Portable Interface to JAVA-Based Applications
Table of Contents -
Subject Index -
Author Index -
Search -
PS reprint -
PDF reprint