The United Kingdom Infra-Red Telescope (UKIRT) is currently undergoing its first major upgrade since its construction. The upgrades program consists of an adaptive tip-tilt and focus system closed with a CCD system at rates of up to a few hundred hertz, an active primary support system, extensive dome thermal work, and other miscellaneous improvements. This paper outlines how we propose to control the new systems, and how these systems are integrated into the existing telescope control system.

A number of improvements have been made to the servo control systems of the 4.2 m William Herschel Telescope (WHT) at the Isaac Newton Group of telescopes on La Palma in the Canary Islands. The successful upgrading of both the Cassegrain and prime focus rotators to meet more stringent science and engineering requirements is described. Simulation (using Matlab^R and Simulink^R) of a model reference adaptive controller to improve azimuth tracking in the presence of torque disturbances is presented together with some preliminary results and a discussion of the way forward. Further enhancements to the WHT's subsystems are also discussed. The smaller 2.5 m Isaac Newton Telescope (INT) and the 1 m Jacobus Kapteyn Telescope (JKT) are also being considered for major improvements to their drives and encoders. Studies are being carried out to determine the requirements and appropriate goals of such improvements and whether modern control approaches can offer cost-effective solutions with minimal re-engineering work. The current performance, generally pointing and tracking, of these telescopes is presented and the subsystems which limit performance are examined; these may be drives, encoders, mirror supports, and structural components. A range of solutions is considered and the technical proposals developed so far are discussed.

The William Herschel Telescope is a 4.2 m altazimuth design which has been in operation since 1987. The architecture of its control system is briefly outlined, and measurements of pointing, tracking, and field rotation are presented. Recent developments, including accurate algorithms for offset guiding, correction of errors due to refracting optics, and automation of the acquisition procedure, are described.

We control the Keck ten meter telescope with a point-and-click interface called `SKY.' We choose a position either by entering coordinates directly, by reading a star list, or by searching one of two on-line catalogs. Targets can be previewed using current or future sidereal time. SKY provides instrument specific rotator control, usually by finding a guide star within the available annulus and then directing the rotator to the appropriate position angle. All this is done by pointing and clicking, without retyping coordinates and angles. A preview popup displays time-to-limit information. SKY features a stand alone mode for planning and training at the observer's home institution.

Remote observing has many definitions, ranging from unattended batch-mode use through simple remote logins to fully faithful off-site observing centers indistinguishable from the on- site telescope control room. There are problems with each of these ideas: batch mode operation, for example, precludes remote interactive target acquisition and remote access to targets of opportunity. Simple remote login suffers from network problems such as full-duplex character latency; shipping screens instead of the underlying data can cause bandwidth problems and interferes with analyzing or archiving data. Brute-force reproduction of the control room requires expensive fiber or satellite connections. The WIYN Telescope control system was designed to be inexpensive to build and inexpensive to maintain. We emphasized the use of standard tools, portable implementations, and network friendliness. These techniques and features are precisely those that underlie a powerful remote observing capability. The WIYN Telescope control system therefore supports remote observing from the very lowest levels, and does so effectively and inexpensively using a carefully planned architecture, standard software and network tools, and innovative methods to ship large digital images over low bandwidth connections such as phone lines. Even before the construction was complete, these techniques proved their value by allowing remote access for the purposes of eavesdropping, troubleshooting, and servo tuning. This paper presents a block diagram and detailed descriptions of the WIYN Telescope control system architecture. Each aspect of the control system is discussed with respect to its contribution to the overall goal of remote observing, including multi-user access, bandwidth conservation, interoperability, and portability.

The control system for the 45 m radio telescope and the millimeter array at the Nobeyama Radio Observatory, which has been used since 1982, has a centralized structure based on an IBM compatible mainframe. However, due to developments of new instruments for these telescopes, the upgrading of the telescope control system has been started. To provide better availability and flexibility, the new control software is completely distributed, being based on LANs interconnecting workstations and personal computers. The new system is functionally divided into three levels: console (first layer), followed by supervisor and data archiver (second layer), and by local controllers for each device (third layer).

The 100 m radiotelescope in Effelsberg (Germany) has been in continuous operation for over 24 years. Consequently, its control system has evolved from a centralized Ferranti Argus 500 based architecture via one using Modcomp Classics coupled with `dumb' CAMAC crates to our current multi-level system, where two clustered VAX 3000 (VMS) machines coordinate several `intelligent' CAMAC crates. Receiver timing, ULO control, and also the control of the backends happens on the CAMAC level using dedicated crates equipped with DC-J11 (PDP11) auxiliary crate controllers. A similar approach is used for the positioning of the telescope, where the astronomical calculations are made by the VAX but the PID control loops for the telescope axes are calculated by the auxiliary crate controllers. A specialized 16-bit parallel interface developed by the MPIfR's digital group enables the observational data from the continuum, spectroscopic, and pulsar backends to flow directly to the VAXes, where it is buffered and then transferred to DATapes and via network connections to UNIX workstations for further processing.

For more than 16 years the Radio Astronomy Group of the ETH Zurich has used solar radio telescopes, which were designed to run automatically in an unmanned observing station. The antennas are azimuthally mounted and the position of each axis is measured with absolute shaft encoders. One antenna is driven by stepping motors while the two others use analog motors. The latest version of the antenna steering uses a velocity control of both axes. The antenna follows the Sun continuously and the velocity is adjusted due to the difference of nominal and actual position of the antenna. The Sun's position is calculated in real time using date and time as input parameters. Pointing to galactic objects is also possible and is used for system calibration (Cassiopeia A). The clock of the antenna control is synchronized each minute with a radio controlled time receiver. In addition, the antennas can be operated manually or remotely. A status signal inhibits data recording, if the antenna is not positioned.

The near infrared camera and multi-object spectrometer (NICMOS) and the space telescope imaging spectrograph (STIS) are next-generation orbital replacement instruments for the Hubble Space Telescope (HST). Ball Aerospace Systems Group is currently designing and building these instruments for the National Aeronautics and Space Administration. Substantial components of the flight software for these two instruments are being designed in common in order to save development costs and to make in-orbit commanding and maintenance of the flight software easier. The majority of the flight software (FSW) for both NICMOS and STIS is embedded within the electronics of the instruments. This paper begins with an overview of the HST-environment in which the embedded FSW for both STIS and NICMOS executes. The paper then describes: (1) the common electronics environment in which the FSW executes, (2) the FSW development environment, (3) the FSW real-time environment, (4) the method for in-orbit commanding of instrument operations via macros, (5) the method for gathering analog hardware engineering data, (6) the method for commanding the STIS CCD detectors and the NICMOS detectors, (7) the structure and management of the science data buffer, (8) fault management, (9) the collection of diagnostic data, (10) the capabilities for uplinking new FSW code and data once the instruments are in orbit, and (11) instrument-specific software executing in the NASA Standard Spacecraft Computer (NSSC-1), external to STIS and NICMOS.

The Gemini 8-m Telescopes Project presents a unique managerial challenge in that not only the detailed design and fabrication of the mechanical subassemblies will be contracted out to the international community but also the software and control systems. The project has developed a model for how to do distributed design and how to manage this effort. The author presents lessons learned in the endeavor and also what he would do differently the next time.

The design of the VLT Telescope control software is driven by requirements of different nature and origin: technical/astronomical requirements, operational requirements, test and maintenance requirements, and general system engineering requirements. A driving factor is also the multitude of devices to be controlled and coordinated, which in itself leads to complications due to the large number of cases to be handled. The VLT Telescope control software will control and coordinate a large number of devices for each of the four 8 m telescopes, each with four individual foci, and additionally two combined foci. It is a requirement to support a large variety of observing instruments and procedures, some developed in-house, some by outside institutes and companies, and some still to be defined. The VLT control system is a highly distributed system; the control of devices is implemented in one VME unit for each device. All VMEs requiring accurate timing are also connected to the central time distribution system, which is of course a vital component in the tracking system. They are also connected to multiple LANs, to cope with different performance requirements. The use of a real-time database is an important design concept. A commercially available product is used (HP/Rtap), on which ESO has developed an additional layer of software that is used for all VLT application software. ESO has also developed a similar standard package for the VMEs that is compatible with the equivalent workstation layer. The paper describes different requirements and involved equipment, and it presents the main parts and characteristics of the software that should meet these requirements. The first part of the paper is an introduction to the telescope control part of the total VLT control system. Then there is a description of the environment and infrastructure on which the telescope control system is based, and finally a presentation of the telescope control software itself.

Many diverse software systems are in use on the Keck I telescope. This is mostly because software standards were low-level (e.g., choice of programming language, computer or operating system) and did not specify use of a particular software environment. Selection of directory structures, messaging systems, tasking environments and support packages was largely left up to individual development groups, although there were some successful instances of group collaborations. For the Keck II telescope, a common set of standards and tools has been agreed on, and the provision and maintenance of these tools is regarded as a group effort. These standards and tools are known as the Keck II Software Infrastructure and include EPICS (Experimental Physics and Industrial Control System), the successful Keck I concept of making all system control available via keyword/value pairs, the Tcl command language, standard logging and error reporting, and common programing standards. This paper discusses some of the successes and failures of the Keck I approach and describes how the Keck II system is evolving from the Keck I system. Some examples of the use of EPICS for telescope control are given, and EPICS as a vehicle for future collaboration is considered.

The Instituto de Astrofisica de Canarias (IAC) is taking preliminary steps towards the building of an 8-m class optical-infrared telescope to be placed at the Observatorio del Roque de Los Muchachos (ORM) on the island of La Palma in the Canary Islands, Spain. This paper presents a brief description of the preliminary conceptual design of the control system for the telescope and its instrumentation in all planned modes of observation and configurations. An outline of the devices that are to be controlled and the functions that are to be performed is presented, together with a description of the system architecture. The system will be distributed and highly modular, and will allow for an easy and efficient exchange of instruments, focal stations and observing modes. It will also be designed so that technological upgrades can be implemented at low cost and minimum down-time.

TNG first light is foreseen for the latter half of 1996. Due to its main characteristics (alt- azimuthal mounting with tight pointing and tracking requirements, deformable primary mirror with active control, actively controlled secondary and tertiary mirrors), the computer support is essential, together with the need to command instrumentation and the possibility to operate remotely. Most of the control system has been developed and is now under test phase. The main characteristics of the computing environment are presented here. The implications of the desired performances on pointing and tracking are discussed and solutions to define the time base and synchronization are proposed.

The generation and distribution of an accurate time signal has been the cause of some technical difficulties in the past. Modern telescopes with their distributed computer architecture have increased this problem even more. Since the availability of the Global Positioning System (GPS), the generation of an accurate time signal of isolated telescope sites is not a problem any more. However, distribution of a time signal with a high accuracy to all computers is still an issue. The IRIG-B code, which is commonly used, is not adequate if accuracy requirements are in the 10 microsecond range. Also, synchronization of software processes in one computer and in different computers must be solved. This paper presents the design that is adopted for the ESO Very Large Telescope (VLT). An overview of the requirements of the time reference system (TRS) is given. These requirements have been defined on the basis of experiences with the timing system of the ESO NTT telescope. The hardware components are described with the emphasis on the distribution and the decoding module in the computers. As timing is vital in a `real time' software environment, the operating system and application software requirements have driven the TRS requirements from the beginning of the definition. This has led to a TRS system that supports the software needs on time synchronization between processes, even in different computers. A description is given about the software implementation.

The use of `scenarios' that describe desired telescope operating behavior can serve as an effective tool for ensuring a telescope design provides the functionality that is desired by the user community. This report examines the role of scenarios and the descriptions of how the system responds to these scenarios (`walk throughs') in the Gemini 8 m Telescopes Control System.

The Gemini servo simulation is a 6 degree of freedom (6-DOF) time domain simulation which is meant to represent the interaction between the servo controls and the telescope structure. The performance measure is root mean square image motion, based on the translations and rotations of the optical elements. The telescope structural model is based upon a sophisticated finite element analysis (FEA) to include the expected bending and torsional modes. Active control of the secondary mirror is modeled in a realistic way, including delays and noise, in order to show the expected improvement to image smear. Among the error sources for the altitude and azimuth control loops are angular encoder quantization, nonlinear bearing friction, motor D/A quantization, motor torque cogging, drive eccentricity, and tachometer ripple. Other relevant nonlinearities include drive amplifier voltage and current limits. Some other smaller simulations are also included: a simplified model for the tip-tilt response to telescope windshake, a model for azimuth drive slip-suppression, and a simple image-quality simulation.

A new controller has been developed for the JCMT chopping secondary mirror. The controller is based on the `proximate time-optimal servo' (PTOS) design. The PTOS controller incorporates both linear and nonlinear controller designs. The nonlinear portion offers minimal time response while the linear portion allows traditional compensator design techniques for higher order flexible modes in the plant. The PTOS controller was implemented with a floating point digital signal processor (DSP). Advantages of the PTOS design as described here include minimal time response, robustness, stability and the ability to accept `random' reference inputs.

Very precise models are required to simulate drive system performance in terms of tracking accuracy. In any case it becomes very hard to identify the correct system transfer function because of the near zero speed nonlinear characteristics of several system subsections. On the other hand, very precise models are unnecessary because of the relative implication with the mechanical structure performance often represented by simplified models due to the complexity of the mechanical structure itself. For these reasons a simplified drive system model can be sufficient to define the macro capabilities of the same system (i.e., required torque, ripple, encoder resolution) while system tracking accuracy can be foreseen by means of a detailed study of any subsystem performance. On the other hand, a model identification can be performed when a precise simulation of the system behavior is necessary but it is possible only if the system has been already done. As a matter of fact a system model is not only necessary to study and to forecast the system performance but it is also necessary to allow a very precise simulation of system capabilities necessary for future improvements. The present paper describes an approach that allows us to identify a real and reliable model of the system starting from the same system real behaviors.

As part of a general upgrade of the 20 m radio telescope at Onsala Space Observatory a new digital control loop has been implemented using discrete time state space control theory. The model of the telescope is reduced to second order. The controller has been designed for two different modes, corresponding to small (tracking) and large (slewing) reference position changes. For the tracking mode the control includes position error integration and reference value feedforward. For the slewing mode the velocity is controlled with regard to maximum allowed acceleration and control voltage. Incremental encoders on both axes are the only inputs to the system, consequently the velocity is reconstructed via a current estimator. The main objective was to create a robust controller, given the uncertainties in the parameters and non-linearities of the telescope system. The control loop runs on a standard VMEbus CPU- board in parallel with astronomical coordinate transformation routines, display, and watchdog programs. The result is a stand-alone telescope unit which is controlled through high level commands using ethernet or serial communication lines.

A model for simulations of the dynamical behavior of the Large Earth-based Solar Telescope (LEST) is described. The model includes the various subsystems of the telescope optics and mechanics, as well as its live optics system. Also, various disturbances, including wind gusts and atmospheric turbulence, are included. The simulation results indicate the fundamental importance of the live optics system of the telescope.

The VLT must provide high quality images in a wide range of operating conditions and for a number of different modes. Performances of tracking and guiding impacts heavily on the final image quality. Control of the tracking axis is implemented in VME controllers, one per axis. They are accurately synchronized with a central time distribution system. This allows each axis to operate independently from the others, deriving its own position from the star coordinates and the time. This coordinate transformation includes basic corrections like refraction and pointing model compensations. To improve tracking performances, guiding devices (guiding cameras located in the adapter, or even instrument detector itself) can be used to measure position errors and derive small corrections for tracking. These corrections can be applied either to the secondary mirror, or to the tracking axis. To coordinate all these activities, the VMEs are connected through several LANs together with supervising workstations. While the basic functionality is controlled by applications implemented in the VMEs running a real time commercial operating system: VxWorks, the presentation layer and the non time critical operations can be implemented on workstations running standard Unix. The distribution of control software on VMEs or workstations is then dictated by real time constraints and/or availability of proved solutions for one of the operating system. This paper presents the choices in the hardware and software architectures, as well as the design concepts, made to support the different operational modes and fulfill the performance requirements.

The Very Large Telescope (VLT) utilizes high accuracy absolute encoders for the altitude and azimuth axes. The size of the telescope makes it impossible to use encoders of the glass disc type, like have been used on the ESO NTT telescope. The solution was found in using the laser Doppler displacement measurement (LDDM) technique. Although this technique is primarily developed for linear displacements, it can also be used for rotational motions. The arguments that have led to the specifications of the VLT encoders are described. The principle of operation of the LDDM, together with the application for rotational motion, is explained. The application of the LDDM for large scale, high accuracy rotational motions is unique. Therefore, a prototype of the encoder was made. The results of the prototype tests, which confirmed the theoretical analysis, are presented. The mounting of the LDDM on the available diameters of the VLT required a detailed analysis of the mechanical tolerances and thermal effects. These aspects are evaluated. A description of the system integration and calibration is presented.

Encoding the angular position of large telescopes is typically achieved through the use of friction driven rotary encoders, tape style encoders mounted on the circumference of each telescope axis or co-axially mounted high precision rotary encoders. These forms of encoding have very stringent mounting requirements, are expensive, adversely affected by contaminating particles and often difficult to retrofit to existing telescopes. The advent of long CCDs presents the opportunity to develop accurate position encoding for telescope control using digital image metrology. In this paper we present the design of a high precision non- contact encoding system which uses the detection of multiple redundant visual edge features to develop sub-pixel edge position measurements to a precision of 1/50th pixel. The method is described in detail and is validated with both simulation trials and experimental results from a testbed setup.

In this paper we describe the active optics system of the WIYN 3.5 meter telescope put into operation in the spring of 1995 on Kitt Peak, Arizona. The active optics system provides real time collimation of the telescope and optical figure control of the primary mirror. The individual subsystems are first described. These include the wavefront curvature sensing technique, the support and articulation of the secondary mirror, the control of the primary mirror figure and rigid body motion, and the mechanics and electronics used for controlling and monitoring the optics. Algorithms for the complete loop are then discussed. This involves mapping coma terms used to actively correct the collimation, while residual phase errors are corrected by active control of the forces supporting the primary mirror. In the next section we compare two operational modes: open loop using mapped collimation and optical figure corrections, and closed loop using feedback from the wavefront sensor directly. Finally, preliminary stellar images obtained with the actively controlled telescope are presented.

During 1988 - 1993, Itek Optical Systems developed software for a demonstration program of several critical technologies for adaptive optical systems. The software implements a variety of data acquisition and control functions for an adaptive telescope system with hundreds of degrees of freedom. The completed software contains approximately 250,000 lines of real- time Ada code and was developed using a tailored DoD-STD-2167A methodology. Testing of the optical system was successfully conducted in December 1993. The system-level requirements for this software, the design of the software, the development methodology, and some lessons learned are discussed.

The 4 m telescope at Cerro Tololo Inter-American Observatory has recently been fitted with an active optics system. As with the NTT and other active-optics telescopes, coma is corrected by tilting the secondary mirror while the other low-order aberrations are removed by deforming the primary mirror. The modifications to the telescope include a new computer- controlled collimation/focus unit for the secondary mirror, and the conversion to active control of the 33 axial supports for the primary mirror. The system works quite well, in spite of having been built at low cost and having to deal with an `old technology' thick primary mirror. We describe here the hardware and software control approach, along with the first results following installation.

The Space Telescope Science Institute (STScI) has developed a new tool for the preparation of an observing program with NASA's Hubble Space Telescope (HST). This tool, called the Remote Proposal System Generation 2 (RPS2), provides the astronomical user with immediate feedback on important aspects of the observing program such as fundamental observing constraints and the detailed structure of the observations, allowing the user to detect errors and make informed trade-offs in observing plans. RPS2 was developed by `gluing' together several existing systems that are used in the day-to-day operations of the HST, and adding new components where necessary.

As institutions and observatories are required to handle more tasks with fewer resources, the need to automate processing becomes crucial. One of the easiest tasks to partially automate is the front-end process of requesting to use a service, e.g. a telescope. Automation can include making the proposal forms available, and allowing them to be submitted, electronically. By providing a standard proposal form, the necessary information contained in the proposal can be extracted and partially processed by software including tracking and low-level error detection. The need to have people in the loop is still necessary, but may of the mundane, repetitive tasks can be given to software, while the mentally difficult tasks can be handled by people. This paper discuses tools available for automating a proposal submission process. It will also discuss a system which was implemented in two months to partially automate Phase I submissions for using the Hubble Space Telescope. This system was used for the most current call for proposals and will continue to be used for upcoming cycles.

The control system for the Subaru telescope is designed to consist of distributed workstations, local processors, and data acquisition computers, which are interconnected by control LANs and data LANs. The control software achieves its functionality with message-based communication. Two key processes, a scheduler and a status logger cooperating with any other processes, are designed to perform efficiency and security in observation. Control flow of observation scheduling and functionality of sub-processes which constitutes the scheduler and the status logger are described. For data acquisition from instruments, Subaru control system provides a variety of data highway which enable instruments to transfer data by up to 20 Mbytes/second. Functionality and characteristics of other subsystems which compose the Subaru control system are described.

The performance of a telescope position control servosystem depends on its ability to minimize changes in position due to wind and other disturbances. Modern, lightweight telescopes such as WIYN rely on feedback to achieve good disturbance rejection. While the simplicity of a direct, friction-drive system is attractive, the small drive ratio greatly diminishes the effectiveness of motor velocity control in reducing disturbance sensitivity. The WIYN servosystems use position feedback and motor torque control to achieve tracking accuracy and disturbance rejection. This type of control allows the use of larger position control bandwidth than is possible with motor velocity control. The factors affecting disturbance rejection are discussed. Motor velocity controlled and motor torque controlled servosystems are described and compared. The WIYN telescope is analyzed to show the expected performance of each type of system, and measured performance is presented.

The Berkeley Illinois Maryland Association (BIMA) array consists of 6 antennas, each 6 meters in diameter, operating at a wavelength of 3 mm. The telescope control is fully automated, allowing round-the-clock observing with little or no supervision. The array can also be controlled from a remote site. One of the major challenges of automated operation is the ability to tune multiple receivers to the desired operating frequency in a reliable manner. The large tuning range required at millimeter wavelengths (85 to 115 GHz), the nonlinear response of the microwave cavities and oscillator phase lock problems have been stumbling blocks for remote receiver tuning. At BIMA, we have developed an automated system capable of tuning all the receivers to any observing frequency within a few minutes. The system uses a Sun workstation in conjunction with dedicated hardware to control the receivers. Large disparities between receiver characteristics are handled in an efficient manner through the use of lookup tables. We describe the system design in detail, and discuss the problems encountered along with appropriate solutions. The generation of lookup tables in the laboratory is also presented.

The RGO is involved in a number of mirror support programs, ranging from new controllers for its existing Isaac Newton Group (ING) telescopes to new primary mirror supports for the UK Infra-red Telescope (UKIRT) and design proposals for the active support of the Gemini 8 m meniscus mirrors. This work has led to the identification or development of critical components such as load cells and control valves which have high precision and stability. Even so it is still necessary to develop servo controllers capable of minimizing the effects of non- linearity and maintaining stability, particularly in regard to the highly non-linear behavior of pneumatic supports. In order to predict the performance of mirror supports and compare differing control strategies, components and systems are modelled using Matlab^R and Simulink^R. These models are presented, together with parameters derived experimentally, and results from recent laboratory tests are discussed. Specific applications are described and current status of the work at the time of submission is presented.

The Canada-France-Hawaii-Telescope has placed in operation a servo controlled roller-screw support system for its primary mirror. This paper will address the goals of CFHT in upgrading from a fixed `bendix' style mirror supporting defining pads. The main design criteria was to have a system that would define the mirror like the original pads, but have adjustability to remove coma dependent on telescope position.

The Canada-France-Hawaii-Telescope is in the process of upgrading the primary mirror cell for cooling of the mirror during non-operating hours. This paper addresses the goals of CFHT in insulating and actively cooling the primary mirror cell environment. The main design criteria is to have a system that would introduce chilled dry air to a sealed and insulated primary mirror cell during daylight hours. This could remove the deleterious effects of having the primary mirror warmer than the ambient air. A system is planned to protect the mirror in case of sudden moisture exposure and also decrease the frequency of recoating due to contamination.

In this paper the statistical analysis of mirror segments' spatial fluctuations depending on noises of their own sensors is given. The research is executed for round and hexagonal mirror elements. The received results permit us to predict the main limitation on stabilization accuracy for various control systems of segmented mirrors.

This report is devoted to the method of optimal regulator synthesis for the control of segmented mirror surface form. The segmented mirror operating frequency range spreads from several hertz up to hundreds hertz. Positioning accuracy is several nanometers. That is why it is important to develop a method of optimal synthesis of the control system, meeting the `basic' functional criterion as well as the given dynamic characteristics of the segmented mirror. Minimum root-mean deviation (RMD) of the current surface form from the desired form is used as the basic functional criterion. This criterion is defined as a quadratic form of the segment displacement at the points of the actuator location. The desired dynamic characteristics of the closed-loop system are given in a form of the standard differential operator. The construction method for the quality function is proposed, which minimizes the difference between the closed-loop object dynamic characteristics and the standard operator.

The Instituto de Astrofisica de Canarias (IAC) has presented the proposal for the Observatorio Roque de los Muchachos (ORM) 8 m telescope. As a part of this project, we have proposed a distributed control system using the Controller Area Network (CAN) protocol for the control of hundreds of force actuators for the support of the 8 m diameter thin meniscus mirror, allowing active optics applications. Recently, our efforts in this area have focused on the development of a low cost, compact, force actuator `node' with a view towards large scale production. We present hardware and software details of the node prototype and our plans for the incorporation and evaluation of 18 such nodes in the IAC 1 m mirror support test rig. We also provide details of an extension in actuator closed loop bandwidth to around 10 Hz.

The complexity of modern astronomical instrumentation generates a large amount of status information for the operators to assess. In practice there is far more information than an individual can assimilate. In order to make optimum use of complex systems such as the WIYN active primary mirror system, it is essential that any system events which may impact on the quality of data be reported in a timely and comprehensible manner. Using toady's powerful workstations we can support the use of real-time visualization of telescope functions and system status. We have developed an interface to the commanding and status layers which allows us to take advantage of the anticipated potential for 3D, photorealistic, and virtual display technologies.

The processing of high-level system commands within an experimental physics and industrial control system (EPICS) database application shares many issues with the design of EPICS motor control records in that the traditional synchronous/asynchronous record processing model is not adhered to. This paper presents the design of the Gemini CAD (command action directive) and CAR (command action response) EPICS database records and illustrates their use in a telescope mount control context dealing with the handling of repeated offset and slew commands.

The Gemini 8-m Telescopes Project has been charged with extremely challenging performance requirements. The author discuses these goals, the design that has been adopted to achieve them, and the current status of this design and its predicted performance.

The fine error sensor (FES) system is a star tracker like device that is used within the Lyman far ultraviolet spectroscopic explorer (FUSE) instrument. The FES extends the performance achievable from the conventional star trackers by providing very low noise guidance reference signals. In addition the FES is used for scene and target identification for the UV system operation. This is accomplished by the FES looking through the Lyman FUSE telescope system by means of a pick-off mirror at an intermediate image. This paper describes the proposed readout electronics for the FES CCD and the CCD clock driving system. The clock driving system is unique in that software can reconfigure the hardware required to generate the various CCD clocks and the timing signals for the readout electronics. An adaptable method used to define the location of the six sub-rasters, used to identify the selected guide stars, is described. A variable clocking rate is utilized to minimize readout time while minimizing read noise. The benefits of using RAM based field programmable gate array (FPGA) technology for flight programs also is highlighted.

The DRAMA software environment has been developed at the Anglo-Australian Observatory to facilitate the development of instrument and telescope control systems. It is designed to run on a distributed system consisting of machines running a variety of operating systems including UNIX, VAX/VMS, and VxWorks. DRAMA builds on ideas from the ADAM system in use at a number of observatories. It is based on the concept of a `task' which is a software object which responds to messages requesting it to perform actions. The message system provides network transport using TCP/IP as well as optimized local transport for each machine. Messages are encoded using a self-defining hierarchical data system (SDS) which allows complex data structures while transparently handling differences in machine architecture. Tasks are coded using a standard event driven structure which can be used for applications ranging from low level real-time systems to user interfaces. The latter are developed using the Tcl/Tk package incorporated into DRAMA tasks.

Modern telescopes have demonstrated the possibility of obtaining very high quality images (as good as 0.3 arc sec) from ground-based telescopes. To maintain this quality during regular use places stringent requirements on many other aspects like alignment (traditionally zero coma) and pointing and tracking, especially if active optics is used. We argue that for these stringent conditions to be met, for correcting coma, tilt and decentering have to be measured and corrected separately, both at the initial set-up of the telescope, and during normal operations. Otherwise: (1) a complex pattern of astigmatism can be introduced during the initial alignment phase; (2) pointing changes are introduced during observations, which though possible to correct, can limit the optical quality.

This report describes a method for using precision tilt-sensors to measure the position of an equatorial telescope relative to the local horizontal plane. Unlike conventional systems which measure the telescope position using position encoders coupled to the telescope axes, this method avoids many sources of non-repeatable error, such as hysteresis in the telescope structure due to inelastic flexure of the fork or yoke, or random slippage in the couplings between the position encoders and the telescope axes. In this respect, it shares many of the advantages of optical gyros, but achieves these at much lower cost. We present a design for a compact and relatively inexpensive dual-axis tilt-table whose frame is rigidly attached to the telescope's primary mirror cell. The table contains two precision tilt-sensors, aligned orthogonally with the tilt axes of the table. The sensors are used as nulling devices to close a servo loop which keeps the table level at all times. This provides a precise and stable reference against which the telescope position is measured. A high resolution incremental encoder is directly coupled to each tilt-table axis and measures the angle by which that axis rotates to keep the table level. A mathematical transform converts these two encoder readings into local hour angle and declination. Preliminary tests of the tilt sensors and of a single-axis prototype tilt-table are reported, and future plans described. The use of tilt-tables for measuring the positions of non-equatorial telescopes is also briefly examined.

The antenna of the EISCAT Svalbard Radar is designed for the frequency range 470 - 530 MHz. The servo system has thyristor controlled dc-motors with digital cascade controllers for tachometer and position loops. The motors are rated on the basis of wide load considerations. The motors are connected via gears and are electrically preloaded using a special scheme. Special velocity profiles are used during acceleration. A non-linear model has been used for simulation of servo performance under wind load. A linear model was formulated to study stability conditions. A special active damping algorithm is applied to damp vibrations in drive motors coupled in parallel. Frequency response measurements on the antenna were carried out using the control computer. Good agreement was found between measurements and model results.

The new generation telescopes need very high quality and reliable control systems in order to guarantee the quality and continuity of the service requested by the high cost of scientific observations. Telescopes require many years to be developed and sometimes they appear to be obsolete at the first light. For this reason many efforts are to be made in terms of technical management allowing telescopes to reach the first light in the shortest possible time, rejecting any technical obsolescence phenomenon. This paper is a general overview on the technical management approach, and contemporarily a description of the solutions adopted to improve the TNG digital drive system capabilities and reliability.

The author reviews the makeup of the recent SPIE Conference on Telescope Control Systems held in Orlando, Florida during April 1995. Also included is the results of a questionnaire circulated to attendees at the conference during the final session.