ATST Software Conceptual Design
ATST Conceptual Design Review
26 Aug 2003
Presentation Structure
• Introduction– Approach– Things to watch for
• Requirements• Functional design overview• Technical design overview• Virtual Instrument Model
Design Architectures
Requirements Behavior
Implementation
Functional Design
TechnicalDesign
Special points
• Configurations– How observations are modeled in system
• Virtual Instrument Model– Provides flexibility in laboratory-style operations
• Device Model– Uniform implementation and control of devices
• Container/Component Model– Flexibility in a distributed environment
Key Science Requirements
• Combine multiple post-focus instruments– Operate simultaneously– Coordinated observing with remote sites
• Match flexibility and adaptability achieved by DST– Support ‘laboratory-style’ operation (modular instruments)– Support visitor instruments– < 30 min switching time between of active instrument set
• >40 year lifetime• Massive data rates• Track on/off solar disk (up to 2sr off)
Software Requirements
ScienceRequirements
RealitySoftwareRequirements
SoftwareDesign
CommonSense
What types are there?
• Functional – what must the system do?• Performance – how well must the system run?• Interface – how does the system talk with the outside?• Operational – how is the system to be used?• Documentation – how is the system to be described?• Security – who/what can do what/when?• Safety – what can’t go wrong?
Functional Design
• Purpose– Focus on behavior and structure (what and why)– Measure against requirements
• Use cases/Overall design• Information flow• Principal Systems
– Observatory Control System (OCS)– Data Handling System (DHS)– Telescope Control System (TCS)– Instruments Control System (ICS)
Overall Design Approach
• Want to adapt a conventional modern observatory software architecture to the special needs of the ATST– Avoid re-invention, but…– Concentrate on multiple instruments operating simultaneously in
a laboratory environment– Flexibility is a key requirement of the functional design
• Overall functional design derived from ALMA, Gemini, and SOLIS, with consideration from other projects as well.– All share a common overall structure (with wildly different
implementations)– All highly distributed with strong communications infrastructure
Overall Design
OCSUIs, Coordination, Planning,
Services
DHSScience data collection
Quick look
TCSEnclosure/Thermal, Optics
Mount, GIS
ICSVirtual Instruments
User Interfaces
Core Software
Information flow
Experiment
Configurations
Observations
Data
Virtual Instrument
Component
Component
Component
Device
Device
DeviceDriver
DeviceDriver
DeviceDriver
Hardware
Hardware
Hardware
Engineering ArchivesQuick Look
Operating Characteristics
• Distributed architecture using a communications bus– Components can be placed anywhere (and moved as needed)– Devices may be constrained by device driver constraints– Components locate each other by name, not location– Language and other environment choices are independent of behavior– Behavior separated from control by a control surface
• Communications bus– Multiple channels– Provides inter-language peer-to-peer communication– Locates components and provides connection handles– Monitors connectivity (detects communication failures)– High-speed, robust
Observatory Control System (OCS)
• Roles:– Construct sequence of configurations for each observation– Coordinate operation of TCS, ICS, and DHS– Provide user interfaces for operations– Provide services for applications– Provide ATST Common Software for all systems
Data Handling System
• Roles– Accumulate science data (including header information)
• handle data rates
• handle data volumes
– Analyze data for system performance (quick look)– Provide archival, retrieval and distribution services
Telescope Control System
Requirements• Coordination and control of telescope components
– Interface to the Observatory Control System
– Configuration management
– Safety interlock handling
• Ephemeris, pointing, and tracking calculations.– Time base control and distribution
– Pointing models
– Target trajectory distribution
• Image quality– Active and adaptive optics management
– Thermal management
Telescope Control System
• Subsystems– M1 Control
– M2 Control
– Feed Optics Control
– Adaptive Optics Control
– Mount Control
– Enclosure Control
• General Systems– Time base
– Global Interlock
– Thermal Management
Global Interlock
Thermal Management
Image Quality
Acquisition, Track, and Guidance
M1 M2Feed
OpticsAdaptive
OpticsMount Enclosure
TCS
Global Interlocks
Interlock
Telescope Control System
• High-Level Data Flows– OCS configurations
– ICS configurations
– TCS events and archiving
• Low-Level Data Flows– Subsystem configurations
– Trajectories
– Image quality data
– Interlock events
OCS
TCSICS
cfgs
cfgs events
FOCS ECS
M1CS
MCS
AOCS M2CS
cfgs
Trajectory
Image quality data
Telescope Control System
• Virtual Telescope Model– Tip of the hat to Pat Wallace (DRAL).
– Several points of view (Instruments, AO WFS, aO WFS).
• Pointing and Tracking– Off-axis telescope should be irrelevant.
– Tracking will be at solar rate.
– Coordinate systems include ecliptic and heliocentric.
– Limited references to build pointing map (1 object/6 months).
– Open-loop tracking for coronal and non-AO work.
– Closed loop tracking uses AO as guider.
• Thermal Management– Daily thermal profile.
– Monitor heat loads and dome flushing.
Mount Control System
Pointing and Tracking• All ephemeris and position calculations are done by the TCS.
• The MCS follows a 20 Hz trajectory stream provided by the TCS.
• This stream consists of (time, position) values that the MCS must follow.
• Current position, demand position, torques, and rates are output at 10 Hz.
Thermal Management• The MCS must keep the mount structure at ambient temperature.
• May be provided by a separate controller (TBD).
Interlocks• Interlock conditions cause a power shutdown, brakes on, cover closed.
• Caused by: GIS, over-speed, over-torque, mechanical obstruction (locking pin, manual drive crank, liftoff failure).
• Provided by a separate controller (PLC-based) that is always operational.
Mount Control System
Servo Requirements• Pos = 3 arcsec
• Vslew = TBD °/sec
• Vtrk = TBD °/sec
• Aslew = TBD °/sec2
• Atrk = TBD °/sec2
• Jitter = TBD °/sec
Azimuth
Elevation
Coudé
Trajectory
Trajectory20 Hz
Trajectory summing
and smoothing.
Bias
M2 Bias0.01 Hz
M1 Mirror Control System
• Axial Support: blending and averaging AO information, applying force map, correcting servo feedback.
• Mirror Position: detecting translation and rotation errors, (feedback to actuators?).
• Thermal management: controlling temperature, applying thermal profile estimates.
• Controller: interfacing to TCS & GIS, simulator.
TCS
GIS
M1CSAOCS
AxialSupport
MirrorPosition
ThermalManagement
M1Controller
Actuators,Force Sensors
Force Map
Translation,RotationSensors
ApertureStop
Blowers,Coolers,
Exchangers
ThermalProfile
Interlocks
Simulator
Interfaces
M2 Control System
Tip-Tilt-Focus• Base configuration set
by TCS.
• Corrections from AOCS in 10 Hz stream.
• Blending data
• Conversion into off-axis translation and rotation.
Thermal Management• Secondary Mirror
• Heat Stop
• Lyot Stop
M2CS
TCS
AOCS MCS
Blending
Conversion
X Y Z
Rotation
X Y Z
Translation
ThermalManagement
Base Configuration
aO & AO TTF10 Hz Tip-Tilt Offset
0.01 Hz
Feed Optics Control System
Small Mirrors• M3: Gregorian feed.
• M7: Coudé feed.
Other Optics• aO WFS beamsplitter
• Polarizers
• Filters?
Thermal Management• Mirror Cooling
• Tube Ventilation
• Coudé entrance
Adaptive Optics Control System
M1
DHS
M2
OCS
Command Channel
Event Channel
Data Channel
0.1 Hz 10 Hz
On Demand On Demand
aO WFS
AO WFS
AOCS
TTM6
DM5
2 KHz
On Demand
Mount
0.01 Hz
Active optics system
Adaptive Optics Control System
M2
M1M5
M6
WFS
WFS
Mount
aO/AOCS
TTF bias offload~10 Hz
Low-order figure offload
~0.1 Hz
Tip-tilt bias offload
~0.01 Hz
Tip-Tilt Mirror~2 KHz
Deformable Mirror
~2 KHz
Active Optics10 Hz
Adaptive Optics~2 KHz
Adaptive Optics/Active Optics
Control System~2 KHz
Enclosure Control System
• Azimuth: drives, brakes, encoders, sensors.
• Shutter: drive, brakes, encoders, sensors, sun shade
• Auxiliary: vent gates, thermal controller, cranes
• Controller: interface, interlock, simulator
TCS
GIS
ECS MCS
Azimuth Shutter AuxiliaryM1
Controller
Drives,Brakes
Encoders,Sensors
Drives,Brakes
ThermalManagement
VentGates
Interlocks
Simulator
Interfaces
Sun Shade
Encoders,Sensors
Cranes
Instrument Control System
Requirements• Laboratory/Experiment Style Observing
– Flexible instrument configuration
– Use of multiple components
• Instruments– Facility instruments follow all ATST interfaces
– Visitor instruments obey a minimal set of ATST interfaces
– Multiple instruments must work together.
• Telescope– Control the beam position
– Control modulators, AO, and other image modifiers
• Development– Several development locations with numerous partners.
Available Components
Instrument Control System
Management• Controls the lifecycle of
virtual instruments
• Allocates components
Interface• Controls configurations
from the OCS
• Provides user interfaces
• Presents TCS and DHS as resources.
Development• Provides standard
instrument as template
OCS
ICS DHSTCS
ViSP VisTF VI VINIrSP VI
Component
Component
Component
Component
Component
Component
Component
Component
Component
Component Component
Instrument Control System
Management Lifecycle1. Select from list of components
2. Build a VI or retrieve an existing VI
3. Register VI with ICS
4. Submit configuration to OCS
5. OCS schedules configuration
6. ICS enables your VI
7. VI takes control of components
8. Interact with your VI [Engineering Mode]
Available Components
My VI
ICS
OCS
1
4
22
5
3 6
7
8
Technical Design Overview
• Purpose– Focus on implementation issues (how)– Must allow implementation of functional design– Identify options, make choices
• Tiered hierarchy– Isolates technology layers– Allows technology replacement
ATST Technical Architecture
ComponentSupport
ConfigDB
ErrorSystem
LogSystem
TimeSystem
DataChannels
Admin Apps Applications
AppFramework
APIs &Libraries
ScriptingSupport
UIFLibraries
ContainerSupport
AlarmSystem
ArchivingSystem
DevelopmentTools
CommunicationsMiddleware
High-levelAPIs/Tools
Services
Core Support
Base Tools
DataHandlingSupport
AstroLibraries
DeviceDrivers
IntegratedAPIs/Tools
Communications
• Communications Bus• Notification Service• Synchronous Communications• Asynchronous Communications
Services
• Logging• Events• Alarms• Connection• Persistent Stores
Interfaces
• Logically separated into three classes:– Lifecycle (startup/reset/shutdown of components)– Functional (command/action/response - behavior)– Service access (connection, log, event, alarm, stores)
• Functional interfaces define accessible behavior– Physically extend the lifecycle interface– Narrow interfaces (few commands used)– All devices use a common interface
• Formally specified and enforced using communications middleware
Container/Component Model
• Industry standard approach to distributed operations (.NET, EJB, CORBA CCM, etc).
• Components implement functionality• Containers provide services to components and
manage component lifecycleComponent
Container
Functional interface
Lifecycle interface
Service interface
Containers
• ATST supports two types of Containers– Porous – Components provide direct access to their functional
interfaces– Tight – Containers wrap component functional interfaces
InterfaceWrapper
Components
• Hierarchically named• Can (currently) be either Java or C/C++• Three lifecycle models for components
– Eternal: created on system start, run to system stop– Long-lived: created on demand, run until told to stop– Transient: exist only long enough to satisfy request
• Exist only inside containers– Common base interface allows manipulation by container
• Critical attributes maintained in separate persistent store
Observations - 1
• Program of configurations• Configurations flow through system: status is updated as
they are operated on by system• Components and devices respond to configurations• Configurations implemented as sets of attributes
(name/value pairs)• Configurations are uniquely named and permanently
archived (as are observations)
Observations - 2
• Observations constructed using Observing Tool• Choice of OT is TBD (ALMA, Gemini, STScI, others)• External representation is XML• Configurations may be composed from other
configurations and components may decompose a configuration into a set of smaller configurations
Observations - 3
• Observation managers track sets of observations as they are operated on by the system
• Components tag header information to identify the component associated with the generation of that header information
Real-time Systems
• Laboratory-style operations– Rapid setup and reconfiguration
– Engineering observations
• Inexpensive– Existing software implementation or design
– Rapid deployment
– Code reuse
• Contractual design and development– Easily partitioned work packages
– Common infrastructure and tools
– Simulators
Device Model
• The device model is used by all real-time components– Other models are available for OCS services (log, notification, etc.).
• It provides a common interface to:– High-level objects (OCS and DHS).
– Other devices.
– Low-level hardware drivers.
– Global services (log, event, database, etc.).
• It can be inherited by more complex devices.• It forces all systems to obey the command/action model.• It operates in a peer-to-peer environment.
Device Model
• Devices have common properties:– Attributes: state, health, debug, and initialized.
– Command Interface: offline, online, start, stop, pause, resume, get, set.
– Services Interfaces: event, database.
• Devices have common operations:– Initialize, power-up, check parameters, change state, execute actions,
handle errors, respond to queries, receive and generate asynchronous events.
• Devices have common communications:– Name/connection registration.
– Event listeners and posters.
– Databases, log, and alarm services.
Device Command Interface
• Devices have a simple command interface:– Offline, online, start, stop,
pause, resume, set, and get.
• Each command moves the device state machine to another state:– States are: OFF, IDLE, BUSY,
PAUSED, FAULT
OFF
IDLE
PAUSEDFAULT
BUSY
onlineoffline
offline
offline
offline
stop/done
start
resume(fault) pause
(fault)
Configurations
• All devices use configurations to transport information.– A group of attributes and corresponding value
– i.e., a filter wheel might act upon: {position=red; rate=10; starttime=10:38:18}.
– Values may be any native data type, arrays, and lists.
• Configurations are followed throughout the system.– Each device action is traceable to a configuration.
– Header information is reconstructed from configuration events.
• Configurations have states:– A configuration may be in multiple states in multiple devices.
– States do not iterate.
– Final state has an associated completion code.
Created Queued Running Done
Inherited Devices
• The Device class is an abstract class; it needs to be inherited by another class.– These classes specify information and operations unique to a particular
device. They may create configuration templates, run background tasks, handle hardware control, and generate specific information.
– For example, the MotorDevice inherits the Device class to operate servo motors and define positions, limits, power, and brake operations.
• An extended class may itself be extended.– The DiscreteMotorDevice inherits the MotorDevice class to provide
defined positions and name-to-position conversion.
High-Level Devices
• Some devices do not operate hardware, they operate other devices or connect to high-level, non-device objects (OCS/DHS).– The SequenceDevice runs other devices in a defined order and phase.
– The ControllerDevice executes scripts.
– The MultiAxisDevice coordinates simultaneous actions.
• High-level devices are an aggregation pattern.– They do not override or hide the low-level devices.
– They allow the low-level devices to operate independently.
Command/Action Model
Commands cause external actions to occur.
• A command will return immediately. The action begins in a separate thread.
• Multiple commands can be given while actions are on-going. This allows us to “stop” an action, or queue up the next “start”.
Actions return asynchronous state information.
• A device will transition to the BUSY state, then either back to the IDLE state, or to the ERROR state.
CommandProcess
ActionProcess
SharedData
Config Response
Actions Actions Actions
Completion
Synchronization Mechanism
Peer-to-Peer Communications
• Devices must have flexible connections.– Depending upon the requested operation, a device may need to
communicate with several different devices.
– Devices must not exclusively control another device.
• Peer-to-peer communications allows a loose federation of devices.– No single point failures, outside of the communications system and the
naming service.
– Multiple federations can exist simultaneously in the system.