Approximately two and a half years ago we started to design a comprehensive computer controlled system to handle data and control telescopes at the Hale Observatories. Our previous experience with hard-wired digital data systems had adequately demonstrated that digital systems can increase the astronomical observer’s efficiency by at least a factor of two, that is, we had found that our observers, when using these digital data systems were able to observe twice as many stars per night as they had previously. We became painfully aware of the fact that the hard-wired systems were extremely difficult to modify for new observing requirements. It therefore appeared logical to develop a data system using a small computer which had the facility of permitting the addition of a large number of new devices without modifying the hardware which had been previously constructed. This concept has the advantage that new observing instruments can be added by constructing new hardwire peripheral devices and developing new software. During the development time of a new device it is always possible to insure that the instrument can be returned to its previous configuration by loading the computer with its previous program. This permits an observer to use the same equipment configuration for a period of years even though the capability of the basic system has been greatly enlarged.

As early as 1968, increases in the complexity of photoelectric observing programs and the variety of sophisticated photometric instruments in use at the Cerro Tololo Inter-American Observatory necessitated the acquisition of a digital data-recording and instrumentation-control system. The diversity of projects to be done with such a system and the obvious need of provisions for extensive future modification and expansion made apparatus oriented around hard-wired logic clumsy and uneconomical. Clearly, the heart of the system was to be a digital computer.

We hope to have a well-planned program tied to the astronomical specifications demanded. We also hope to save time and money, very important considerations in high cost, big science.

I have become really aware of the sink of dollars and man-hours that technology involves, especially so when it is not well defined and well planned. We hope to avoid a near bottomless sink if we can. Total systems design is very important, not just a careful design of each subsystem.

In adopting automation, I want to insist that we don’t lock out certain experimental possibilities that we had before, or lock out certain type people (good ones) who are unwilling or unable to work with the highly sophisticated equipment we’ll be providing. Excellent astronomy has been done with poor equipment used by good astronomers; even the best equipment doesn’t do the best astronomy without the full involvement of good scientists. What I’m saying is, perhaps, that the system has to be our slave, not the reverse, and that we should not be limited by it.

What we want is an improvement in astronomical research, not merely in the gadgetry involved. The key to success may well be to strive for the best possible astronomer-technologist interface.

The aims of the automatic control system are to position the mechanical axis of the telescope to specified right ascension and declination coordinates, and to track the right ascension axis smoothly at sidereal rate, atmospheric refraction variations being taken care of by an auto-guider. The design problems involved fall naturally into two groups, the mechanical ones of measuring the rotations of the axes and of applying drives to these, and the electronic ones of providing supervisory control and calculation facilities.

Solutions to these problems depend to a large extent on whether automatic control is being added to an existing telescope with existing drive systems, or a new telescope design is being carried out with automatic control in mind, and on whether an on-line process control computer is to be incorporated or not.

The art of computing was early recognized by astronomers as being the link between observation and theory. Computing became more and more important with the appearance of electronic computers.

Today we can regard the application of electronic computers in three different aspects.

• 1. Theoretical investigations.

• 2. Data handling, that means collecting of astronomical observations and reduction to observational results.

• 3. Control of telescopes and other astronomical instruments.

The present paper deals with item No. 3 only. It should, however, be emphasized that the combination of items 2 and 3 will lead to fully automatic astronomical observation in the future.

In 1969 a new astronomical institution was founded in the Federal Republicof Germany. This “Max-Planck-Institut für Astronomie” will consist of a central institute at Heidelberg and two observatories, one in the South-East of Spain and one in the Southern hemisphere (possibly in South-West Africa). Each observatory will be equipped with a 2.2-metre Ritchey-Chrétien telescope; these instruments are now under construction at Carl Zeiss, Oberkochen, Western Germany. In addition, a 3.5-metre telescope will be built. A smaller Ritchey-Chrétien telescope of 1.2 m will go into operation next year in Spain. The mirror disks of the 3.5-metre and 2.2-metre telescopes have been ordered from Schott & Gen. at Mainz; they are made of glass ceramic “Zerodur”, a zero-expansion material similar to CerVit. In the operation of the 2.2-metre telescopes and their peripherals, small, general purpose digital computers will be employed.

The Vacuum Solar Telescope has been described elsewhere by Dunn (1964, 1969). A brief summary of its characteristics is included here as background for discussing the computer control of this instrument.

This telescope is altazimuth in design. Image rotation is accomplished by rotating the inner tube structure together with all the auxiliary instruments. Azimuth and elevation torque motors drive the two mirrors at the top of the tower to track the sun automatically. The input to the servo is derived from a photoelectric guider. When clouds intervene, the servos are switched from the photoelectric guider to an electromechanical coordinate converter that also generates the signal for the rotation of the table. An elaborate 25: 1 synchro system connects the mirror servos to the coordinate converter.

An on-line computer has been in use at Jodrell Bank since 1964, both controlling the Mk II radio telescope and performing data processing functions in connection with it. However this system is now being superseded, and the following describes the present state of development, including some facilities that are only now being constructed.

An Argus 400 computer is associated with each of the two main telescopes. Each has 12 000 words of core store, the word length being 24-bits, and the time for obeying most functions is 12 μs. The 24-bit word length is well suited to the application since more than 16-bit accuracy is required in the coordinate transformation. This fact, and the very versatile organization of the Argus computer, makes the effective speed relatively higher when compared with faster machines of smaller word size. In addition a drum store of about 50,000 words capacity will be fitted to each machine, and CRT and graphical output devices are available. Each machine operates a permanent program, which performs a number of functions, including in the case of the Mk II telescope the control of the instrument itself. In addition, up to three “user” programs may be operated simultaneously; these may be on-line data collection, with links to the telescope control sections, or completely off-line background jobs. A wide selection of library routine, available to all programs, is permanently in the core store, and this makes the writing of programs a relatively simple matter.

The automatic star follower or autoguider described in this paper was designed for the Edinburgh 40/60 cm Schmidt camera. The stellar image in the focal plane of the 15 cm f/18 guide telescope is split optically to produce four stationary images which are sampled by a single image dissector tube, whose sequential output signals are proportional to the intensity of illumination in each quadrant. These signals are compared to produce error signals which are a measure of the radial displacement of the star. Guiding corrections in both right ascension and declination are applied to the fine motions of the Schmidt drive mechanism.

Measurements made on a typical plate given an exposure time of 30 min, and using guide stars down to magnitude 8.5 indicate a guiding accuracy of better than 0.14 arcsec or 2 μm at the focus of the guide telescope.

Dr. B. K. Ioannisiani was unable to give his scheduled paper on this subject, but Dr. N. N. Mihel’son kindly agreed to answer questions in its place.

Recently an automatic UBVRI photometer has been under construction at the University Observatory, St Andrews, for use on the 1.02-metre telescope at the Royal Observatory, Cape Town. Two problems have been encountered which would seem to be of general relevance. Firstly a Cassegrain focus acquisition head, specific to the photometer, was required, and secondly a large rack of electronics must be shipped to South Africa. Clearly these problems have occurred to other guest observers, and it is suggested that they would be greatly reduced by the adoption of international standards, preferably through the IAU, relating to mechanical and electronic interfaces.

In this paper we describe the 40-channel magnetograph primarily as an example of the automated data system at the McMath Solar Telescope. The principles of the magnetograph itself have been recently given in paper I (Livingston and Harvey, 1970), while the reduction and interpretation of data is covered in paper II (Harvey and Livingston, 1970).

A simultaneous four channel photoelectric photometer has been used successfully by astronomers at the Dominion Astrophysical Observatory in the past three years to measure the colours of some 2000 stars. The optics consist of a small grating spectrograph with quartz lenses. Windows in the focal plane of the camera isolate four spectral regions and the light is fed to separate EMI photomultipliers. The photometer is not designed for very faint star observations, consequently D.C. techniques are used. Although the photomultipliers are not refrigerated, a 2 per cent precision is possible in the measurement of the colours of a star of V = 12 in a 1 min integration with a 16-inch telescope. Provided stars are bright relative to the sky background, measured colours are insensitive (±001) to transparency variations of up to three magnitudes caused by normal light cloud. The outputs of the photomultipliers are integrated on capacitors and timing is controlled by a crystal oscillator. An integrator and associated control circuitry are shown schematically in Figure 1. The integrators are read automatically by a digital voltmeter and the results together with time and other data are punched out on paper tape. A block diagram of the data acquisition system is given in Figure 2. After conversion of the tape to a card deck the data is processed in the computer. The system has proved to be very stable and mean errors in single observations of each colour are between 0004 and 0007. The most time consuming part of the process is in the careful editing and sieving of the results for erroneous data generated by line voltage spikes, mispunches, and observer mistakes.

I wish to present some aspects of a computer controlled pulse counting system for general photoelectric photometry.

The pulse counting technique applied in astronomical photoelectric photometry has yielded significant improvements as to telescope efficiency, especially in connection with the measurement of very faint light sources. This is achieved through the advantages inherent in the pulse counting technique over the conventional charge technique, that is:

After having introduced the electronic computer at the W. Struve Tartu Astrophysical Observatory, an opportunity to put into practice some measuring and data handhng systems presented itself. The simplest measuring of variable star brightnesses by the on-line computer was chosen for the beginning, and for today the first stage has passed with certain success. In the present paper we give a brief report about the work done.

Photoelectric spectrophotometry has been pursued with increasing success in recent years. In studies of single line profiles, restricted spectral regions, or stellar continua at low resolution, the photomultiplier offers an increase in speed of up to 20 times over the photographic emulsion and an output directly proportional to light intensity over a wide range of intensity. Difficulties in slow scanners have been caused chiefly by fluctuations in light due to seeing, guiding and extinction changes, necessitating either wide slits and photometric weather or ratio circuits. The technique of rapid scanning and exact superposition of digital data from rapid scans offers a solution to these problems. The basic requirement for rapid scanning is therefore a multichannel analyser and a suitable triggering mechanism. I shall describe briefly the way this is being developed at the 48-inch telescope coudé of the Dominion Astrophysical Observatory, and other uses this instrument is being put to.