KEITH PEACOCK and KNOX S.

LONG

ASTRONOMICAL TELESCOPES: A NEW GENERATION

After a hiatus of more than a decade, astronomers are poised to build a new generation of large ground-
based telescopes. With their larger areas, rigorous imaging specifications, and modern control systems,
they will allow astronomers to extend traditional astronomical observations to fainter, more distant ob-
jects, and will permit fundamentally novel investigations from the ground. New technologies, such as
segmented mirrors, active figure control, "spin-cast" mirrors, and multiple mirrors, are being applied
to extend the size of today's largest single-mirror telescope (6 m in diameter) and to combine the light
from smaller individual mirrors to create effective apertures up to 16 m in diameter. The Johns Hopkins
University, in collaboration with the Carnegie Institution of Washington and the University of Arizona,
is designing an 8-m telescope using a spin-cast mirror.

INTRODUCTION
A Dutch spectacle-maker, Hans Lippershey, is credit- of London. All modern large telescopes can trace their
ed with the invention of the telescope in 1608. 1 Galileo ancestry to Newton's discovery. Reflecting telescopes
subsequently heard of Lippershey's achievement, made have several advantages when compared with refractors;
several improved telescopes of his own, and used them unlike refractors, they do not suffer from chromatic
to discover the spots on the Sun, the lunar mountains, aberration, and the mirrors are much easier to support
the phases of Venus, and the moons of Jupiter. In 1611, than lenses because the entire back surface is accessible.
at a banquet in his honor near Rome, his instrument Reflecting telescopes increased in diameter. In 1789,
was christened "telescope." Although his discoveries ini- Sir William Herschel built a 122-cm telescope, and by
tially promoted Galileo's career, the last two provided 1845, Lord Rosse had made one 183 cm in diameter.
support for the Copernican theory and eventually led All the early reflectors had mirrors of speculum, an alloy
to Galileo's trial for heresy. of copper and tin. Speculum, however, is an imperfect
Progress on telescopes was slow during the next hun- material for a telescope because, like most metals, it is
dred years. The presence of chromatic aberrations (im- harder to figure and polish than glass and has a high
perfections in image quality caused by variation with thermal expansion coefficient, which distorts the mirror
wavelength in the index of refraction of glass) limited once installed in the telescope. It was not until 1856 that
the usefulness of the instruments. Telescopes made of a chemical method for silvering glass was first developed.
simple lenses were built with diameters up to 20 cm, but This technological advancement, a prerequisite for the
they could be as long as 46 m. big glass reflectors of the first half of the twentieth cen-
Chromatic aberration can be greatly reduced by the tury, culminated in 1948 with the dedication of the 5-m
achromat, which combines two glasses with different dis- Hale Telescope at Mt. Palomar in California.
persions into a lens to give focusing without dispersion. The Hale Telescope represented a watershed in tele-
The two-element achromat was invented by Chester Hall scope construction. It weighs 450,000 kg and floats on
in 1733 (and reinvented and patented by John Dollond oil bearings. The f number (focal ratio), the ratio be-
in 1758)2 and led to the development of the modern tween the focal length of the primary mirror and its di-
refractor. The best telescopes of the nineteenth century ameter, is 3.3, smaller than for earlier reflectors. 3 The
were refractors, but lens diameters eventually were limit- focal ratio establishes the length of the telescope and the
ed to 1 m by the difficulty of preventing the glass from size of the dome housing it. The Hale dome is 41 min
distorting under its own weight, by constraints on the diameter and 40 m high.
uniformity of the glass, and by developments in the tech- Construction of the telescope, begun in 1927, was
nology of reflecting telescopes. lengthy, expensive, and technically difficult. The tele-
Defeated by chromatic aberration, Newton gave up scope was not used regularly until 1949 (demonstrating
all hope of perfecting Galileo's form of telescope and that our frustration over the time needed to build instru-
turned his attention to making concave mirrors that ments like the Hubble Space Telescope is not without
reflected light to a focus without dispersing it. Newton's precedent). The principal problem was the primary mir-
first reflecting telescope, invented 60 years after the ror: $600,000 was spent in a futile attempt to make it
refractor, was 15 cm long, with a mirror 2.5 cm in di- from fused quartz. Even after switching to borosilicate
ameter. So successful was the performance of this pygmy glass (Pyrex), the first attempt to cast the mirror failed.
that he made a larger one, now held by the Royal Society The cost was $6.5 million ($6 million came from the

fohns Hopkin s APL Technical Digest, Volume 10, Number 1 (1989) 29

Peacock, Long - Astronomical Telescopes: A New Generation

Rockefeller Foundation), a considerable amount in the the instruments up to 5 m. The cost followed a power
1940s. Because of these difficulties, the 5-m telescope law, S = 0.37D 2.58 , where D is the diameter in meters
held the size record for almost three decades, until a 6-m and S is the inflation-adjusted cost in millions of dol-
telescope was constructed in Russia in 1974. That tele- lars.5 Larger sizes could be shown to be prohibitively
scope, which also uses borosilicate glass, is affected by expensive; an 8-m telescope would cost about $80 million
diurnal temperature variations; as the temperature falls and a 16-m telescope would cost $473 million. Few dis-
at night, the mirror cools, contracts, and distorts. With sented from this conclusion.
such a massive piece of glass, the image quality can be The Multiple-Mirror Telescope (MMT), begun in the
degraded for the entire night. 4 early 1970s and dedicated in 1979, was an attempt to
The large telescopes of the 1970s and 1980s were built supply a large collecting area at a relatively low cost.
with diameters of about 4 m (see Table 1). The focal Resulting from a collaboration between the University
ratio of the primary mirror is significant for comparison of Arizona and the Smithsonian Astrophysical Observa-
with the new generation of large telescopes. The values tory, the MMT has six 1.8-m-diameter primaries mount-
range from 2.5 to 5.0 (average, 3.25), which, as we will ed on the same structure in a hexagonal array. A central,
see, is considerably higher than the focal ratios planned smaller mirror is used for guidance. Computer control
for the new telescopes. combines the images from all six mirrors into a single
The principal difference between the new telescopes focus. The aperture is equivalent to a single 4.5-m-diam-
and the 5- and 6-m telescopes is the use of low-expansion eter mirror.
glasses and vitreous ceramics (e.g., Cervit or Zerodur) In the 1990s, a new generation of telescopes will be
for the main mirror. Low-expansion glass makes larger constructed. Single-mirror telescopes with diameters of
temperature gradients tolerable in the polishing process 8 to 10 m and MMTs with diameters up to 16 m will
and during use. The surface of a good telescope has a be built by various consortia, including one in which The
figure accurate to 50/0 of a wavelength (25 nm), a preci- Johns Hopkins University is cooperating. The designs
sion very difficult to maintain if temperature gradients for these telescopes will be very different from the Hale
exist within the glass. Telescope design. The equatorial mount has already been
Although sizes did not increase, there were significant replaced by the altitude-azimuth drive. The monolithic
developments in telescope guidance and control, detec- primary mirror, with a thickness of 15% of its diameter,
tors and instruments, and site selection. Charge-coupled is being replaced by multiple mirrors, segmented primar-
devices have enabled astronomers to use more efficiently ies, thin deformable mirrors, and light-weighted spin-
the photons delivered to the focal plane. These devices, cast mirrors. Focal ratios as fast as JI1.2 are contem-
with quantum efficiencies of 50%, have largely replaced plated to keep the telescopes light and the domes small,
photographic film, which has an efficiency of 2% at resulting in lower cost. Several new technologies are be-
most. Thus, the effective diameter of the Hale Telescope ing used to construct the telescopes that will serve as-
is about five times what it was in 1950. tronomers well into the next century. We will review
The cost of telescopes was believed to be predictable, these technical developments, and the major telescope
on the basis of an extrapolation of adjusted dollars for projects they are making possible, in this article.

Table 1-The world's largest optical telescopes.

Diameter oj
Altitude primary
Location Ownership (m) (m) Year

Mt. Pastoukhow, Caucasus, USSR USSR 2050 6.05 1974

Mt. Palomar, Calif., USA USA 1800 5.08 1949
Mt. Hopkins, Ariz., USA USA 2600 4.60* 1979
La Palma, Canary Islands, Spain UK 2400 4.20 1986
Kitt Peak, Ariz., USA USA 2100 4.01 1973
Cerro Tololo, Chile USA 2500 4.01 1974
Siding Spring, New South Wales, Australia UK-Australia 1200 3.88 1974
Mauna Kea, Hawaii, USA UK 4200 3.80 1979
Mauna Kea, Hawaii, USA Canada-France-USA 4200 3.60 1979
La Silla, Chile Europe 2450 3.57 1976
Calar Alto, Spain FRO-Spain 2160 3.50 1979
Mt. Hamilton, Calif., USA USA 1300' 3.05 1959
Mauna Kea, Hawaii, USA USA 4200 3.00 1979

*Multiple-mirror-telescope equivalent light-collecting area.

30 fohns Hopkins APL Technical Digest, Volume 10, Number 1 (1989)

 Peacock, Long - Astronomical Telescopes: A New Generation

SCIENTIFIC RATIONALE Optical Design

Several factors determine the usefulness of any tele- The basics of large optical telescopes are straightfor-
scope. The main reason to build a larger telescope is to ward. Figure 1 shows a schematic of the simplest concept
increase the amount of light delivered to the detector. on which all ground-based and most space telescopes are
An 8-m telescope has 4 times the light -gathering power founded. The large primary mirror collects light from
of a 4-m telescope and 11 times that of the 2.4-m Hubble an object at infinity and images it at the prime focus
Space Telescope. For ground-based telescopes alone, the at a distance/, one-half the mirror's radius of curvature,
increased area of an 8-m telescope compared with a 4-m from the mirror. The ratio fl Dp (where Dp is the di-
telescope implies that one can observe four times as many ameter of the primary) is the focal ratio or f number.
objects on the same number of nights or see objects two The radius of curvature r is an important design param-
to four times as faint as one could before. An 8-m tele- eter of the telescope, because it determines the sag of
scope will deliver about 0.7 photons· cm - 2 • S - 1 • A - I the mirror DZ or the depth of the bowl, given approxi-
from a 22nd-magnitude star. Consequently, medium- mately by DZ = D pl(l6 ·f) number. For an fl1.4 8-m
resolution (AI LlA = 1(00) spectroscopy of 23rd-magni- mirror, the sag will be 0.36 m, which requires the grind-
tude objects should be possible, as should low-resolution ing away of a large amount of glass if one begins with
(AI LlA = 100) spectroscopy of 25th-magnitude objects. a flat blank. The sag of the Hale Telescope was only
(Stellar brightness is measured in magnitudes on a loga- one-quarter this value, yet it required the removal of
rithmic scale in which a difference of one magnitude rep- 4500 kg of glass to get the rough blank. To simplify tele-
sents a brightness difference of .lfIOO or 2.512. Stars scope construction, all modern techniques for manufac-
visible to the naked eye range in magnitude from - 1 turing mirrors involve the generation of a blank with a
to 5 or 6.) shape as close as possible to the fmal figure of the mirror.
Another element in determining the usefulness of a The shape of the primary mirror is parabolic, a shape
telescope is the quality of the images it delivers. The sky yielding a perfect axial image at the prime focus. The
is bright compared with the faint objects one would study departure from sphericity is a further complication in
with an 8-m telescope. The sky background at a site un- manufacturing the mirror. The separation between the
contaminated by man-made light is equivalent to one rims of an fl1.4 8-m spherical mirror and a parabolic
21.25th-magnitude star per square arc second. Therefore, mirror of identical vertex radii is only 0.29 mm, a small
for observations of stars, an 8-m telescope regularly value, but sufficient to produce polishing difficulties. For
delivering images of 0.5" to the focal plane is equivalent comparison, the 5-m Hale mirror would have a separa-
to a 16-m telescope delivering images of 1". tion of only 0.014 mm. Polishing spherical surfaces is
relatively easy, because when two rigid bodies are rubbed
TELESCOPE DESIGN AND CONSTRUCTION against each other with an abrasive between them, a
The telescope builder faces assorted interconnected op- spherical boundary tends to form. For a weak parabola,
tical, mechanical, and environmental problems. The such as the Hale mirror, where the departure from a
Johns Hopkins University is involved in the design of sphere is only a few tens of wavelengths, localized work
anfll.4 8-m telescope (the Magellan Project), which we with small tools puts the aspheric component on an ini-
will use to illustrate the techniques being applied to over- tial sphere. With a strong aspheric, however, the radii
come those design difficulties. along the mirror radius and along a direction perpendic-

Primary
mirror
Cassegrain
Secondary Figure 1-The basic optical tele-
focus
scope has two mirrors, a large con-
cave primary and a smaller convex
secondary, which produce an image
at the Gassegrain focus. A multilens
focal-plane corrector is used to cor-
rect the curvature of the focal plane
and reduce other aberrations. It may
also include thin prisms to correct for
atmospheric color dispersion. Addi-
tional mirrors are used to reflect the
beam to other foci, such as the Nas-
myth (to the side of the telescope) or
the Goude (below the telescope). A
key feature of the mirror is the sag,
~ _ _ To Coude oz, which is the depth of the mirror
focus dish.
Nasmyth
focus

Johns Hopkins APL Technical Digest, Volume 10, Number 1 (1989) 31

Peacock, Long - Astronomical Telescopes: A New Generation

ular to it differ by too much to make this approach prac-

tical. Also, the difference is variable along the radius.
Although it is easier to polish mirrors with high focal
ratios, the overall size of the telescope, the dome enclos-
ing it, and therefore the cost, are also set by this primary
/ number, so there is a strong motivation toward small
focal ratios.
Occasionally, the prime focus is used as the location
for scientific instruments. For example, the 5-m Hale
Telescope has an observing cage at this focus, and the
Hopkins Ultraviolet Telescope locates its spectrometer
here. Because the off-axis aberrations are large and the
location is not easily accessible, this configuration is not
usually preferred. The most common configuration is
to use a convex secondary mirror, which changes the /
number of the beam of light and produces a magnified
image at the more convenient Cassegrain focus behind
the primary mirror. A secondary mirror with a hyper-
boloidal figure produces an abberation-free axial image.
The placement and radius of curvature of the secondary
mirror determine the exact position of the focus and the
focal-plane scale (i.e., the ratio of the angular separation
of two objects in the sky to their linear separation in
the focal plane of the telescope). For wide-field applica-
tions, one desires as fast a final / number as possible,
since even at/16 a 40' field subtends 0.6 m, making it
difficult to image the entire field with modern detectors.
The Magellan Project is planning an /16 optical Cas-
segrain focus, with an/I1.4 8-m primary. The secondary
mirror, with a diameter of 1.9 m on the Magellan tele-
scope, is also aspheric (the Magellan secondary differs
from the best-fit sphere by 0.25 mm); that, together with
the convex shape, makes it challenging to manufacture
and test.
Any instruments using the Cassegrain focus are at-
tached to the telescope and move with it. Two other foci
frequently used are the Nasmyth focus, in which a ter-
tiary mirror reflects the beam out the side of the tele-
scope along the altitude axis, and the Coude focus, which
uses additional reflections to send the beam below the
telescope. Both locations are convenient, since the in-
strument can be fixed and does not change its orientation
as the telescope tracks a star. The image plane will ro-
tate, however. The Nasmyth focus is used on telescopes
with altitude-azimuth mounts, as shown in Fig. 2a,
where the beam is reflected along the altitude axis, which
is always horizontal. The advantage is that the instru-
ment can be built on a platform whose orientation with
respect to gravity does not change as the telescope moves.
Thus, flexure of a massive instrument is not a serious
problem. Also, there are two Nasmyth foci on opposite
sides of the telescope so that one can be dedicated to
a p~rmanently mounted instrument (e.g., a high-resolu-
tion spectrograph), while instruments are switched regu-
larly at the other focus. The basic disadvantages of the
Nasmyth focus are that it requires an additional reflec- Figure 2-Modern telescopes now use the altitude-azimuth
tion and that the focal ratio must be relatively large, typi- mount (a) in preference to the older equatorial mount (b). In ex-
change for greater drive complexity (two drive axes with non-
cally greater than /112.
uniform rates), the altitude-azimuth mount offers a simpler
A telescope using a parabola-hyperbola combination structural support for the telescope, two Nasmyth foci , and fixed
is free of spherical aberration, and the mirrors have no axes-one horizontal and one vertical-for the bearings.

32 Johns Hopkins APL Technical Digest, Volume 10, Number 1 (/989)

 Peacock, Long - Astronomical Telescopes: A New Generation
chromatic aberration. Off-axis, however, it suffers from Another advantage is the provision of two Nasmyth foci
coma, astigmatism, and field curvature. These are re- at the sides of the telescope, where instruments can be
duced by changing the primary mirror to a hyperbola, mounted in an environment in which gravity is constant.
resulting in a configuration called a Ritchey-Chretien, Supporting the primary mirror has always been a
and by using a focal-plane corrector, which is an optical problem. The primary can be made so thick and stiff
system comprising three or more lenses located some dis- that it cannot distort under its own weight, since its orien-
tance ahead of the telescope's focal plane. Correctors tation changes from the horizontal to close to the verti-
have been designed for use with Ritchey-Chretien tele- cal, or it can be supported so that the forces on its rear
scopes, and they yield good imagery up to 20' off axis. surface vary with orientation. In the past, a combination
of these techniques has been used. With the new, lighter,
Mechanical Design
thin mirrors, however, the figure of the mirror may have
The mechanical and structural design of a large tele-
to be variable, as is planned for some of the new tele-
scope is subject to extremely tight tolerances and severe
scopes.
requirements. The structure must hold the relative posi-
The dome housing the telescope has two conflicting
tions of the primary and secondary mirrors to within
requirements: it must protect the telescope from the en-
a fraction of a millimeter to avoid excessive aberrations.
vironment, particularly wind, but it must allow the tele-
Typically, for an 8-m telescope, the separation must be
scope to reach thermal equilibrium with the outside tem-
controlled to within 0.05 mm, and the angle must not
perature. The traditional approach has been to bury the
change by more than 0.2'. These tolerances are main-
telescope inside a dome with a slit to allow it to view
tained by a combination of basic structural stiffness and
the sky. This produces "dome seeing" from the internal
active control. In an 8-m telescope, the mirrors are sepa-
temperature variations, creating air movements that de-
rated by 8 to 9 m and are supported in a structure that
grade image quality. Forced-air ventilation can equalize
moves, is subjected to thermal changes, and may not
the internal and external temperatures, but not as effec-
be protected from wind loading. Various materials have
tively as an open dome. Protecting the telescope requires
been used for the telescope structure, but steel, despite
a large dome, with its consequent effect on cost. An al-
its significant thermal deformations, is normally used.
ternative is the use of more open domes. The dome of
Dynamically, the telescope must be driven so that it
the MMT leaves the telescope well exposed when it is
can point to a location in the sky with an accuracy of
open. With other telescopes, the dome may be eliminated
1" and then track that point with an accuracy an order
entirely when the telescope is in use. Exposed telescopes
of magnitude greater. The mass that must be moved with
are subject to greater wind loading and therefore require
such precision weighs up to several hundred metric tons.
stiffer structures.
All large telescopes of the early twentieth century used
the equatorial mount (as shown in Fig. 2b), which is rela-
ATMOSPHERIC PROBLEMS
tively simple because uniform rotation about one axis
(the polar axis aligned along the rotation axis of Earth) The major telescopes of the world are concentrated
allows the telescope to follow an object as it moves across in a few areas. Within the United States, California and
the sky. The other axis allows the telescope to move in the Southwest have been the locations of choice for ob-
declination (latitude). servatories. Hawaii is now a more popular location. The
One difficulty with the equatorial mount is that mountaintops in Chile, along the western coast of South
changes in the direction of the gravity vector are com- America, are the sites for observatories of many nations
plex. In the last decade, most large telescopes (e.g., the and will be the site for the Magellan Project telescope.
Herschel 4-m telescope just commissioned at La Palma The single parameter having the greatest influence on
in the Canary Islands) have been built with altitude-az- site selection is the size of the image of a star at the tele-
imuth mounts in which one axis, the altitude axis, points scope's focal plane. The angular diameter of a star is
toward the local vertical, while the other rotates horizon- always too small to be resolved by any telescope, since
tally with the telescope. To track an object, the telescope its diameter is always well below the diffraction disk of
must be driven along both the altitude and azimuth axes the telescope. For a large telescope, this disk is much
at rates that are not uniform. Improvements in servo- smaller than the image diameter caused by atmospheric
mechanisms, including the use of autoguiders and the "seeing" (i.e., a broadening of the stellar image caused
availability of inexpensive and powerful microprocessors, by atmospheric turbulence).
have now made that possible. Another disadvantage of Seeing results from changes in the refractive index of
the altitude-azimuth mount is that the image of an ex- the air, produced by atmospheric turbulence. The two
tended object such as a galaxy rotates in the focal plane main causes of the turbulence are convection (the heat-
as the telescope tracks, necessitating some compensating ing of air by the warmer surface, followed by the rising
rotation. The altitude-azimuth mount has one great ad- and expansion of the warm air and its replacement by
vantage, however: the gravity vector is confined to a cold air) and high winds that cause turbulence by wind
plane with respect to the telescope. Thus, because there shear. At the best sites, the stellar image diameter is as
is no change in the orientation of the axes of the drive small as 0.5" for 50070 of the nights. This image diameter
bearings, and the telescope tube and mirrors rotate about decreases into the infrared, so that the diffraction limit
a horizontal axis only, the structural design and the sup- of the telescope optics, rather than the atmosphere,
port of the mirror in its cell are considerably simplified. dominates above a wavelength of a few microns. The

fohns Hopkins APL Technical Digest, Volume 10, Number 1 (1989) 33

Peacock, Long - Astronomical Telescopes: A New Generation

seeing pattern is continuously changing, with a turbu- as 1909 (see the boxed insert). In the Mirror Laboratory,
lence-coherence time of 40 to 50 ms . pieces of borosilicate glass are loaded into a mushroom-
Considerable effort has been applied to finding sites shaped furnace perched on a large rotating bearing, as
where the seeing is good. The best sites have permanent shown in Fig. 3. The floor of the furnace is studded with
temperature inversions for most of the year, which pre- a matrix of refractory material occupying the space that
vent the convective mixing, and laminar airflow in the will shape the ribbed structure on the rear surface. The
prevailing wind. They also must have a dark sky and temperature is raised to almost 1200°C to melt the glass,
little precipitable water above. California and Chile meet and spinning begins. The combination of gravitational
those requirements. The prevailing wind is from the west, and centrifugal forces shapes the surface of the molten
so no mountains intervene to perturb the laminar flow mass of glass into a paraboloid. The revolution rate de-
after its passage across the Pacific. The cold coastal termines the focal length of the blank; for anf/1.4 8-m
waters also maintain the low humidity and the temper- primary, the oven is spun at 6.3 revolutions per minute,
ature inversion. Chile also has a sky that is not polluted similar to the rate of a carousel. Once the glass has melt-
by light from human activities. Consequently, the very ed completely, the temperature of the furnace is lowered
best telescopes in the world today, at optimum sites and until the glass solidifies, and spinning can stop. After
under optimum weather conditions, can deliver images the mirror has cooled (a process taking about one month),
to the focal plane with stellar image diameters of 0.5" . the refractory material, which has the consistency of
cardboard, is removed with a high-pressure stream of
NEW TELESCOPES, water. The resulting surface should be within 1 mm of
NEW TECHNOLOGIES the desired parabola. Grinding will reduce this to a few
Several technical approaches are being considered in micrometers, followed by polishing to complete the fig-
the race to produce telescopes of larger diameter. Each uring process. The finished mirror will weigh 13,600 kg.
approach has advantages and disadvantages, advocates For comparison, the Hale 5-m mirror required 10 months
and detractors, and it is not apparent what the future for cooling, and 4500 kg of glass had to be ground away
holds. In the following discussion, we outline the three to produce the rough dish. 7
basic approaches being actively followed today, and pre- Technological demonstrations of spin casting culmi-
sent some potential advantages and technical challenges nated in 1985 with the casting of two 1.8-m mirrors in
of each. Briefly, the three approaches are as follows: a smaller furnace. One mirror is for the Vatican observa-
tory to be built on Mt. Graham in Arizona. More re-
1. To use a new type of casting process to build large, cently, an oven large enough to cast 8-m-diameter mir-
light-weighted borosilicate glass mirrors, with conven- rors was constructed in the Mirror Laboratory. In April
tional diameter-to-thickness ratios. 1988, the firstf/1.8 3.5-m mirror blank was cast in the
2. To assemble a large mirror from smaller segments Mirror Laboratory. Another 3.5-m mirror was cast in
and actively control the position of the segments. the fall of 1988. This will be followed by anf/1 6.5-m
3. To cast a thin meniscus mirror and actively con- mirror that will replace the six mirrors in the MMT (at
trol the surface figure on the telescope. the same time doubling its collecting area), and then the
Spin-Cast Telescopes first 8-m mirror by the end of the decade.
Conventional primary mirrors on large telescopes are Thermal Control of Spin-Cast Mirrors. Low-ex-
solid blocks of glass with diameter-to-thickness ratios of pansion vitreous ceramics, such as Cervit and Zerodur,
7 to 10. Such mirrors, once built, are sturdy and rela- cannot be used for spin casting, because of the way they
tively easy to support because they are stiff. They must must be annealed. Therefore, the mirrors being con-
be made of expensive ultralow-expansion glass because structed in the Mirror Laboratory use borosilicate glass
they have extremely long thermal time constants, and (Pyrex), which was used successfully on the 5-m Hale,
thermal distortions would prevent them from delivering the 3-m Lick, and the 2.2-m Kitt Peak telescopes (and
0.25" images to the focal plane. As this type of tech- with less success on the 6-m Russian telescope). The sig-
nology is applied to larger and larger mirrors, it becomes nificant difference between the American and the Rus-
increasingly difficult to manufacture the blank, and the sian telescopes is that the former were constructed as
moving weight scales as D3. Using light weighting, a ribbed plates with no section thicker than 10 cm, and
matrix of cavities on the mirror's rear surface reduces the latter is a 60-cm-thick disk with huge thermal iner-
the weight but retains the rigidity without the expense tia. 8 Borosilicate glass has the advantages of low melt-
and challenge of an active figure control system. Light ing point and low cost, but its thermal expansion coeffi-
weighting has been used on several telescopes, includ- cient is 6 times that of fused silica and 30 times that of
ing the 5-m Hale mirror, which has 8- to 13-cm-thick ultralow-expansion fused silica and the vitreous ceramics.
ribs,3 and the Hubble Space Telescope. Light weight- (Borosilicate glass was abandoned for large-telescope use
ing with 1.5- to 2.5-cm-thick ribs was the motivation be- years ago because of this problem.) It is desirable to keep
hind the development of the spin casting of blanks, an temperature variations in the blank during polishing be-
activity centered at the University of Arizona's Mirror low 1°C and to maintain the operating temperature of
Laboratory under the leadership of Roger Angel. the mirror's reflecting surface to within 0.2°C of the am-
The physics behind a spin-cast mirror blank is straight- bient air temperature. The thin ribs and faceplates of
forward. Robert W. Wood 6 used the principle as early the University of Arizona mirrors will reduce their time

34 fohn s Hopkin s APL Technical Digest, Volume 10, Number 1 (1989)

 Peacock, Long - Astronomical Telescopes: A New Generation

THE FIRST SPINNING-MIRROR TELESCOPE

The accompanying illustration shows the observatory and
telescope from which was made the first astronomical ob-
servation using a mirror whose figure was formed by a spin-
ning liquid. The work was performed by a famous physicist \ \
from The Johns Hopkins University, Robert W. Wood, as
a diversion while spending the summer of 1908 on his Long
Island farm. The idea that a liquid rotating under the ac-
tion of gravity would form a paraboloid surface was well
I
known, but it had not been used to make a telescope. A
telescope needs a perfect surface, and, as Wood found, con-
siderable effort and ingenuity are required if the rotating
liquid is to be ripple-free. (His research was instigated by
his search for an effective method of driving a ruling en-
gine for making diffraction gratings, which also requires
a perfectly smooth drive.) Wood's mirror consisted of a .. ",::::':.:: :...
51-cm-diameter dish of mercury rotated by a motor from
which it was isolated to minimize vibrations. He went to
considerable lengths to prevent jarring from the drive mech- Telescope
anism, irregularities in the bearing, and rotational velocity pit
variations. The spinning mirror was placed at the bottom
of an old brick well 76 cm in diameter and 4.3 m deep,
and a separate shaft was made for access. The "dome" was
the cow barn above the well, with a hole knocked in the
roof to give a view of the sky. The rotational speed was

·:.::.i(il~fi~~~jt
chosen to give a focal point at the location of the plate
holder above ground level. The focus could be held con-
stant to 1 mm. The total cost of the telescope was $200.
Some astronomical observations were made with the de-
vice, and Wood was able to separate the components of
a double star 5" apart. Unfortunately, the telescope CQuid
not be pointed and could observe only a small strip of sky
The "observatory" and telescope of Robert W. Wood of The
as Earth's rotation swept it through the small field of view. Johns Hopkins University that used a spinning dish of mer·
Although it was never developed for astronomy, the work cury to produce a paraboloidal mirror for astronomical ob-
Wood did in devising a vibration-free drive system was later servations.
very significant in his development of ruling engines for the
manufacture of diffraction gratings. The idea has recently
been revived. * Some Canadians have built and spun mer- son why large mirrors cannot be made in this way at very
cury mirrors up to 1.65 m in diameter. They show architec- small expense. At all events the method seems well worth
tural plans for an observatory using a 6-m liquid mirror trying." Almost 80 years later, Roger Angel of the Univer-
and argue that it is scientifically useful and technically feasi- sity of Arizona's Mirror Laboratory is bringing Wood's
ble to build mirrors over 15 m in diameter. prediction to life, and the future of American astronomy
Robert Wood 6 predicted that " ... we may be able to may well depend on the spin casting of telescope mirrors.
discover some substance which can be fused, rotated, and
*E. F. Borra, R. Beauchemin, R. Arsenault, and R. Lalande, "Op-
allowed to solidify while revolving with a constant veloci-
tical Shop Testing of Liquid Mirrors," in Proc. IA U, Very Large
ty. Refiguring of the surface would probably be neces- Telescopes, Their Instrumentation and Programs, Colloquium 79,
sary .... If a suitable material can be found, ... I see no rea- p. 147 (1984).

constant to only 20 minutes. 9 Also, temperature stabil- face of the mirror with a rotating polishing tool, known
ity may be achieved by blowing isothermal air, controlled as a lap, that must have almost the same shape as the
to be at precisely the ambient air temperature, into the section of the mirror being polished. With a parabola,
honeycomb structure. The final verification of this ap- however, the radius of curvature of a surface changes
proach under observing conditions awaits the use of a depending on the distance from the vertex of the mirror,
spin-cast telescope on an operating telescope. and it is different in the radial and azimuthal directions.
Active Lap Polishing. Another challenge is to polish This mismatch between the two directions limits the size
the mirror and give it the correct curvature. In the final of the lap used. The smaller the lap, the longer the time
stages of manufacture, material is removed from the sur- it takes to polish the primary. (Difficulties in polishing

Johns Hopkins APL Technical Digest, Volume 10, Number 1 (1989) 35

Peacock, Long - Astronomical Telescopes: A New Generation

(3) (1)

(9)

(4)

(5)

(6) (7) (8)

Figure 3- This dramatic sequence of photographs, taken in April 1988, shows the stages in the manufacture of a spin-cast 3.5-m
mirror. The furnace (1) in which the mirror is made resembles a carousel. In the sequence, we see the chunks of borosilicate
glass being loaded into the furnace (2) and the glass before closing the lid (3). Note the capacity of the furnace to make mirrors
with diameters up to 8 m. As the furnace temperature increases from 829°e (4) through 90re (5) to 96Qoe (6), the glass melts
into a liquid mass. Spinning begins at 1170 e (7), and the glass forms the paraboloidal shape. Note the increased height of the
0

central shape as the dish forms and the rib structure visible through the mirror surface. After cooling has allowed the glass to
harden (8), the furnace lid is cracked (9) to show the still red-hot mirror. (Photographs courtesy of Roger Angel , University of Arizo-
na Mirror Laboratory.)

36 Johns Hopkins APL Technical Digest, Volume 10, Number 1 (1989)

 Peacock, Long - Astronomical Telescopes: A New Generation

are not specifically associated with spin casting, but rath-

er result from the high curvatures of the new, fast pri-
mary mirrors.)
One approach proposed by Roger AngellO to solve
this problem is to construct a computer-controlled polish-
er that continuously changes the lap surface to conform
to the desired surface curvature. The lap itself consists
of a plate that is deformed by wires tensioned around
the back of the lap, 24 tensioning cables, and the con-
trolling computer that updates the force pattern every
millisecond. For an 8-m mirror, the lap will have a 2-m
diameter. The first lap of this type is now being con-
structed and will be used to polish one of the 1.8-m
blanks produced in the Mirror Laboratory's smaller
oven, an fll primary for the Vatican telescope.
Telescopes Based on Spin Casting. A number of
telescopes are being planned in anticipation of the suc-
cessful generation of spin-cast mirrors. Much of the in-
centive for mirror development in the United States came Figure 4- The Magellan Project 8-m telescope. The Johns Hop-
from the ambitious National New Technology Telescope kins University is a key participant in the Magellan Project and
Project, a square array of four 8-m mirrors on a com- is planning the construction of an 8-m-diameter single-mirror
mon mount. Unfortunately, this project is not funded, telescope. One possible mount is an altitude-azimuth rocking
chair design, which rotates about a vertical azimuth axis, while
so the future of a full-size array is uncertain. We present "wheels" rotate about the altitude axis. The primary mirror is
below some continuing projects. the circle in the center, and the secondary mirror is supported
The Magellan Project is a collaboration between the on the open structure. The axial cylinder in the center is a light
Carnegie Institution of Washington, the University of baffle to prevent direct light from reaching the focal plane.
Arizona, and The Johns Hopkins University and is in
the definition I design phase. This will be a single-mirror
telescope, 8-m in diameter, using one of the light- alent diameter of a single mirror is 11.3 m. This project
weighted spin-cast primaries from the Mirror Labora- also is in the definitionl design phase. The telescope will
tory. The telescope will be situated at Las Campanas be located at Mt. Graham, near Tucson, Ariz., a site
approved by act of Congress on the last day of the 1988
Observatory, about 350 km north of Santiago, Chile,
session. A variety of foci will be instrumented, includ-
near La Serena. The primary will be anfll.4 parabola.
ing the individual Cassegrain foci of the two primaries.
Observations will take place primarily at the Cassegrain
focus, although provisions also are being made to in- A major impetus behind the Columbus design is the de-
sire to use the telescopes for optical interferometry, thus
strument the Nasmyth foci. This is a fairly convention-
making it possible to bring light from the two mirrors
al telescope, which traces its origins back to the earlier
to a combined focus between the two primaries. With
4- and 5-m telescopes. Two basic configurations are
a baseline of 23 m, it should be possible in principle to
planned: a wide-field f16.45 optical focus and an fl15
resolve objects 0.005" apart with Columbus. With the
infrared-optimized focus. AtfI6.45, the secondary mir-
two primaries on the same structure, the optical path
ror will be a 1.8-m hyperboloid. A large fused-silica cor-
lengths remain the same as a celestial source is tracked
rector with an atmospheric dispersion corrector will be
across the sky, simplifying the challenge of interferomet-
needed to deliver 0.25" images over a 40' -diameter fo-
ric observations. The rocking chair design for the Magel-
cal plane, the current design goal.
lan Project may be required for the Columbus Project.
Two basic designs for the telescope structure have been
The telescope will be within range of remote control from
developed: a standard yoke mount supported on a large
any of the locations of the four cooperating organiza-
ring by hydrostatic bearings and a "rocking chair" de-
sign, shown in Fig. 4, wherein the telescope is supported tions.
on two large wheels. The weight of the rocking chair Segmented Mirrors
design is estimated to be about 22,700 kg less than the The construction of larger-diameter telescopes using
standard yoke. The rocking chair design also has some- monolithic mirrors presents major problems: there are
what higher modal frequencies than the yoke mount. The no existing facilities for casting such large blanks, they
choice of design has not been finalized. are fragile and expensive to repair, and there are logistical
The Columbus "Two-Shooter" Project (named after problems in transporting them. An alternative approach
the Italian navigator) is a collaboration between the Uni- is to build up a single mirror out of a number of smaller
versity of Arizona, Ohio State University, the University segments.
of Chicago, and the Arcetri Astrophysical Observatory There are two principal difficulties associated with the
in Italy. The two-shooter consists of two 8-m mirrors development of a large segmented mirror. First, each seg-
mounted side by side on the same structure in a config- ment is an off-axis portion of a parabola, which has no
uration resembling a giant pair of binoculars. The equiv- axis of symmetry, making conventional polishing tech-

Johns Hopkins APL Technical Digest, Volume 10, Number 1 (1989) 37

Peacock, Long - Astronomical Telescopes: A New Generation

niques very expensive or impossible (for the same reasons

that it is impossible for very fast monolithic mirrors).
Second, if the segments are to operate as a single dif-
fraction-limited mirror, they must be aligned relative to
one another to the surface accuracy of a single mirror,
that is, to within a fraction of the wavelength of light.
Efforts to solve these problems, led by Jerry Nelson at
the University of California, have been underway for
over a decade. Since the University of California and
the California Institute of Technology are constructing
a 10-m segmented mirror to be installed in the Keck Tele-
scope, we will discuss segmented technology in that con-
text.
With an aperture diameter of 10 m, the Keck Tele-
scope is the largest single-mirror telescope under con-
struction or, indeed, planned. Also, of the new genera-
tion of large telescopes, it is the first to be fully funded
and the closest to completion. Design began in 1977, and
construction began in 1985 upon receipt of a $70 million
gift from the W. M. Keck Foundation to the California
Institute of Technology, which is working with the Uni-
versity of California on the construction of the telescope.
"First light" is expected in 1990. Operating funds for 0. 9m~
the telescope will be provided by the University of Cali-
f-oI
.l-----------10.0 m ---------l~~1
fornia. The telescope is atop Mauna Kea, Hawaii, at an
altitude of 4145 m. Figure 5- The segmentation geometry of the Keck 10-m pri-
The primary mirror comprises 36 hexagonal segments mary mirror. The mirror will consist of 36 individually mounted
thin mirrors held in precise optical alignment to provide a sin-
(see Fig. 5) that fit together to produce the 10-m-diameter gle effective aperture of approximately 10 m in diameter. The
primary with anfnumber of 1.75. The center segment center segment is absent to allow passage of the beam . The
is missing to allow passage of the beam from the sec- entire array is curved into a paraboloid with a focal length of
ondary mirror. II The telescope has been designed to al- 17.5 m. (From Ref. 11 . Reproduced by permission, Keck Obser-
vatory.)
Iowa variety of fl15 foci, including Nasmyth, Casse-
grain, and bent Cassegrain positions. The secondary mir-
ror is monolithic. Infrared observations will be possible move (for aluminizing) and replace. The use of a small
at fl25 at a forward Cassegrain focus, located roughly coating chamber is a significant advantage of segment-
at the same position as the Nasmyth tertiary mirror. ed mirrors, since aluminization costs for 8-m monolithic
The completed dome of the Keck Telescope is shown mirrors can constitute up to 20% of the telescope's to-
in Fig. 6. Its structure is a welded steel space frame sup- tal cost.
porting the telescope inside an altitude-azimuth mount- To overcome the problem of polishing the off-axis
ing. The total moving weight of the telescope, only parabolic mirror segments, a new polishing technique
245,000 kg, is supported on an azimuthal journal 17 m called stressed-mirror polishing was developed. It is com-
in diameter on four hydrostatic oil bearings. The lowest plementary to the stressed-lap polishing technique used
resonance frequency for the structure will be about 5 Hz. by the Mirror Laboratory. A full-scale demonstration
The size of the dome, an important cost consideration, of stressed-mirror polishing was carried out by the Kitt
is less than the sizes of the domes of several telescopes Peak National Observatory in the early 1980s.
in the 3- to 4-m range. The generation of each mirror segment begins with
Each mirror segment of the Keck primary has a max- the casting of a Zerodur blank having a 2-m diameter
imum dimension of 1.8 m and a thickness of only and a 10-cm thickness. The initially flat blank is ma-
7.5 cm, a compromise between the weight and the com- chined so that it has about the right meniscus shape,
plexity of supporting the surface so that no distortions reducing its thickness to 7.6 cm. Twenty-four supports
can occur. Thus the segments are quite thin, having a are cemented around the periphery of the blank (which
diameter-to-thickness ratio of about 24 and an overall is still circular), the mirror segment is mounted in a fix-
ratio for the 10-m aperture of about 133. Thin mirrors ture as shown in Fig. 7, and external loads are applied
are required for the figuring process and have the ad- by placing weights in the buckets below the radial arms.
vantage that the resulting mirror is very light, since each By calculation and experimentation, the weights have
segment weighs only 400 kg. The segments are being been determined that will deform the surface by an
fabricated of ultralow-expansion glass (Zerodur), which amount corresponding to the difference between the
avoids the thermal expansion problems associated with desired shape and a sphere. A spherical surface is then
Pyrex. The segments and their support structure are polished onto the blank, the forces are removed, and
designed so that the individual segments are easy to re- the mirror relaxes into the desired asymmetric shape. The

38 Johns Hopkins APL Technical Digest, Volume 10, Number J (1989)

 Peacock, Long - Astronomical Telescopes: A New Generation

Figure 6-The dome of the Keck

Telescope as it looked in July 1988.
It is located atop Mauna Kea, Hawaii.
The inset is a photograph of a scale
model of the telescope. The enor-
mous size is demonstrated by the
size of the figures included with the
model. (Photographs courtesy of
John Gustafson, Calif. Assoc. Res.
Astron.)

Bending Polishing The passive mechanical support system reduces the

levers lap gravitational deformation across the 10-m mirror to less
n \ than 1 mm, and an active segment control system aligns
the surfaces to optical tolerances (i.e., to better than
-----------------:=J 50 nm or 0.1 wavelength). The active control system is
shown in Fig. 8. A total of 168 displacement sensors are
mounted on the rear surface along the boundaries be-
tween segments. These capacitative displacement sensors
consist of a sensor drive paddle on one segment that fits
within the sensor body of the adjacent segment. The sen-
sor operates at a 10-nm resolution over a range of 1 mm.
Optical polishing A control algorithm running on a multiprocessor com-
table puter system uses the sensor readings to determine the
Figure 7-ln the configuration used for stressed-mirror polish- actuator motions needed to adjust the segment positions
ing, weights are applied to levers bonded to the edge of the to correct for gravitation- and temperature-induced
glass blank to apply forces and moments around the edge. changes. Each segment has three actuators that comprise
These weights , in combination with back supports, warp the
blank in a controlled manner. A spherical surface is polished a motor-driven screw and an optical shaft encoder
onto the blank, the forces are removed , and the blank relaxes mounted to the whiffletree supporting the mirror. Sig-
into the desired off-axis parabolic shape. (From Ref. 11. nificantly, the purpose of the active control system is
Reproduced by permission, Keck Observatory.) not to provide the initial alignment of the segments, but
to maintain the alignment, which is determined using a
bright star.
final step is to cut the mirror into a hexagonal shape
with a diamond saw. Monolithic, Thin Meniscus Telescope Mirrors
Completed hexagons must be mounted into and The third technique for producing a large mirror is
aligned in the telescope to produce a single mirror of to cast or assemble a monolithic, thin meniscus mirror
optical quality. Each segment is supported by a passive with a shape close to the final figure. The glass with
mechanical support system consisting of a combination greatest potential is fused silica. Casting an '8-m blank
of axial and radial supports. Axial support is provided in a single operation is beyond present technology, so
by three whiffletrees, each having 12 points of contact techniques for welding small hexagonal pieces have been
with the rear of the blank. This uniform support prevents developed. An 8-m blank is merely an extension of ex-
distortions across the surface of the blank. A radial sup- isting technology. A second possible material is Zerodur,
port post mounts into a cavity machined into the rear which is being used in the Keck Telescope. Assembling
of the segment and allows the freedom of movement a blank from smaller pieces, however, is more complex
needed for the active alignment system. The segments with Zerodur than with fused silica, since vitreous ce-
are mounted with spacings of a few millimeters between ramics can lose their ideal thermal properties if subjected
them. to remelting. Borosilicate glass is a less likely alternative;

Johns Hopkins APL Technical Digest, Volume 10, Number 1 (1989) 39

Peacock, Long - Astronomical Telescopes: A New Generation

Secondary
mirror

~Sensor sites

/ Flat mirror

Primary
mirror

Mirror
M = Motor

Figure 9-The light of a star is reflected by the primary mirror,

which is supported by many motorized levers that determine
its shape. The light is then reflected by the secondary mirror,
the position of which can be adjusted by motors, and finally
brought to a focus by a third flat mirror. Some of the light is
analyzed by a wavefront sensor coupled to a computer that de-
termines the corrections to be made for optimal imaging. In-
structions are sent to the appropriate motors to change the
shape of the primary mirror and the position of the secondary
mirror. (From VL T, The ESO 16-m Optical Telescope , European
Southern Observatory, Feb 1986.)

Sensor body
Figure 8-For the 36 segments of the Keck Telescope to oper-
ate as a single mirror, they must be aligned to within a fraction
of the wavelength of light. The relative position of adjacent seg-
ments is sensed using two capacitative displacement sensors
along each edge. Each sensor consists of a paddle on one seg-
ment and a body on the adjacent segment. The position of the
Exit pupil/'
paddle can be sensed to an accuracy of about 10 nm over a
of telescope
range of 1 mm. (Reproduced by permission, Calif. Assoc . Res.
Astron. From J. R. Gustafson and W. Sargent, "The Keck Ob-
Figure 10-A diagram of an image analyzer or wavefront sen-
servatory: 36 Mirrors are Better than One," Mercury, Mar-Apr
sor. The focal plane of the telescope falls at (1). A lens (2) colli-
1988.)
mates the beam and images the exit pupil of the telescope onto
a raster of weak lenses (3). Because th is is an image of the tele-
scope exit pupil , each portion of the image plane represents
there is concern that its relatively high coefficient of ther- a particular area of the primary mirror. Each lens in the array
mal expansion may cause problems. (3) images the star in a plane at (4). An optical transfer system,
To minimize weight, the mirrors must be less than lens (4), and objective lens (5) reduce the image plane to the
30 cm thick, so thin that the mirrors will deform under size of a charge-coupled-device array (6). This image consists
of an array of points, each being an image of the star as
their own weight. To maintain good image quality under produced by a different portion of the primary aperture. The lo-
conditions of gravitational and thermal changes, optical cations of these points are compared with the locations of
misalignment, and atmospheric variations, a telescope points in a reference array introduced by the beam splitter (7)
with such a mirror requires a system for correcting the and the source (8). (From 2nd Workshop on ESO 's Very Large
Telescope, European Southern Observatory, p. 421 , Sep-Oct
alignment and the figure of the primary mirror over a 1986.)
low-frequency range of 0 to 0.033 Hz. This is active op-
tics. An upper limit of about 30 s is needed to average
the variations of atmospheric seeing. a reference image, and a computer calculates the cor-
An active optics system requires four components: an rections that must be applied to the optics to cancel any
optical train and image detector, a wavefront sensor, a errors. Error correction is accomplished by moving the
servocontrol system, and a phase-shifting optical ele- secondary mirror to correct focus and coma caused by
ment. The system concept is shown in Fig. 9. After the decentering, and by using a set of push-pull actuators
light has passed through the three telescope mirrors, part to change the figure of the primary mirror, which be-
of the light (either from the target star or from an off- comes the phase-shifting element.
set guide star) is picked off and directed into a wavefront The goals of the European Southern Observatory
sensor (Fig. 10). The sensor compares the image with (ESO) system (see below) are to place 80070 of the light

40 fohns Hopkins APL Technical Digest, Volume 10, Number 1 (1989)

 Peacock, Long - Astronomical Telescopes: A New Generation

from a star into an image 0.15" in diameter at a wave- focus to give the effective aperture of 16 m. During con-
length of 500 nm and to be diffraction-limited at 5 p.m. struction, the fIrst telescope completed can operate while
The limiting magnitude of the guide star is 14.5. The work on the others proceeds. The presence of several
wavefront sensor will operate at 5 Hz, and mirror cor- telescopes provides redundancy, so some observations
rection will be at 2 Hz. Between 250 and 350 actuators will always be possible. By operating the telescopes in
will be needed for each mirror. pairs and by adding the beams coherently (assuming this
proves to be possible), long baseline interferometry with
THE ESO VERY LARGE TELESCOPE different spacings is planned as a major scientific en-
The most complex and technologically sophisticated deavor of the array. The disadvantages of the linear ar-
telescope project under consideration is the ESO Very ray are that it requires a longer site than a square array
Large Telescope (ESO-VLT). The objective is to pro- of four telescopes, the optical and mechanical tolerances
duce an aperture with an effective diameter of 16 m by required to combine the beams are very severe, and the
combining the beams from a linear array of four 8-m telescope field of view at the combined Coude focus is
telescopes. Also, the project will include some major only 10".
technical innovations, such as thin mirrors with active The ESO is taking an incremental approach to devel-
figure control, interferometry with combined beams oping the active optics. It was first demonstrated success-
from pairs of telescopes, automatic guidance accurate fully on a I-m mirror (the actuators are shown in Fig. 12)
to 0.05", and remote operation from Europe. The and has been incorporated on the ESO's 3.5-m New
preliminary study for the project began in 1978, a de- Technology Telescope. This telescope has a Zerodur mir-
sign group was organized in 1983, and unanimous agree- ror 24 cm thick, whose figure can be changed by 75 ac-
ment by the eight ESO member nations to proceed with tuators. Its moving weight is only 109,000 kg, consider-
construction was reached in December 1987. The first ably less than a conventional telescope; for example, the
unit is expected to be completed in 1993, and the total 3.5-m German-Spanish telescope constructed at Calar
array will be finished at a cost of over $180 million by Alto, Spain, uses a conventional mirror and has a mov-
the year 2000. (This is 100;0 of the cost of the Hubble ing weight of 209,000 kg.
Space Telescope for 50 times the collecting area, mak- The VLT has no dome in the conventional sense. Each
ing the Hubble Space Telescope 500 times more costly telescope is exposed to the environment during observa-
for each meter of mirror surface.) The proposed site of tions, which has the operational advantage of rapidly
the VLT is La Silla, Chile, but other locations are also bringing the telescope into thermal equilibrium with the
being considered. environment. When the telescopes are not in use they
The VLT four-telescope linear array shown in Fig. 11 will be covered by inexpensive ro11-on/ro11-off shelters
has some important advantages. It provides great flexi- or by the inflatable covers shown in Fig. 11. The open
bility in that the telescopes can point independently at dome leaves the telescopes exposed to the wind. The
different targets, they can all point at the same target selected location in Chile has a significant advantage,
and use different instruments, or their beams can be however, in that the wind is predominantly from one
combined incoherently into a single beam at the Coude direction, the north. Thus, the use of an active wind

Figure 12-A scaled-down version of the support system of

Figure 11-A model of the complete array of four 8-m tele- ESO's New Technology Telescope has been used to support
scopes making up the ESO-VLT. When in use, the telescopes a 1-m-diameter mirror with a thickness of only 20 mm. The lab-
are completely exposed; they are protected only by the wind oratory tests at ESO have successfully confirmed the feasibility
screen behind them. When not in use, the telescopes are cov- of the concept. This photograph shows the transducers that
ered by inflatable shelters. The building below the telescopes apply forces to the rear surface of the mirror to change its shape.
is the interferometric laboratory. (From VLT, The ESO 16-m Op- (From VL T, The ESO 16-m Optical Telescope, European South-
tical Telescope, European Southern Observatory, Feb 1986.) ern Observatory, Feb 1986.)

Johns Hopkins APL Technical Digest, Volume 10, Number 1 (1989) 41

Peacock, Long - Astronomical Telescopes: A Ne w Generation

shield, both to protect the telescopes from wind loading tinuously variable, two auxiliary telescopes with 1.5-m
and to produce a smooth airflow over them, is being apertures will be used external to the main array. These
studied. The telescopes will be able to operate in winds telescopes will run along two tracks, one parallel to the
up to 9 m/ s. main array (thus extending the baseline) and the second
The telescopes can operate in either of two modes: in a perpendicular direction, giving resolution in a sec-
(1) the beams can be reflected to the individual Nasmyth ond coordinate. The telescopes can operate independent
foci along the altitude axes, or (2) they can be relayed of the main array and can be dedicated to interferometry,
to the Coude focus in the optical experimental facility which will be performed in a separate laboratory at-
below the telescopes. Figure 13 shows the path of light tached to the telescope facility. Other major new tele-
to the optical facility from two of the four telescopes. scopes with multiple apertures are being designed to the
It is anticipated that five to seven reflections will be used tolerances necessary to permit interferometry, so that this
to reach the combined Coude focus. Because this will technique is expected to become an important branch
result in a serious reduction in optical efficiency, three of astronomy in the next century.
sets of mirrors will be selectable, each set coated to give
maximum efficiency in one of three spectral regions: 300 CONCLUSION
to 470, 380 to 700, and above 700 nm. A built-in laser For 25 years the 5-m telescope was the largest optical
alignment system will monitor the optical system. The telescope in the world, and for the past 15 years it has
mirrors will be mounted in helium-filled tubes to main- remained in second place. By the end of the century,
tain the quality of the optics. The beams can be com- it will have fallen farther down the list of largest tele-
bined into a single focal plane with a 10" field of view scopes. The change in the list reflects the revolution that
using figured mirrors, or a fiber-optics system can be is starting to take place in astronomy. Several of the new
used to combine the beams or distribute the signal be- large telescopes are funded; the Keck Telescope is near-
tween instruments. ing completion, and the ESO-VLT received the go-ahead
One exciting aspect of the VLT is its potential for in- from member nations in 1987. Other projects such as
tererometry. Interferometry on this scale is unproven to the National New Technology Telescope will not proceed
date, however. Although the technique has been used because of funding problems. During the 1980s, several
with very long baselines for radio astronomy, the preci- other proposals and design studies have been performed
sion required at optical wavelengths means that only re- for large telescopes . Some have made little progress, such
cently has there been an interest in performing interfer- as the United Kingdom's two-mirror telescope,12 the
ometry with pairs of telescopes. Initial attempts will be spectrographical MMT of a French team, and the Uni-
restricted to the infrared at wavelengths longer than versity of Texas large telescope. 13 Other projects are
3 JLm, with possible future developments to shorter still active; for example, the Japanese National Large
wavelengths. The VLT provides a baseline between the Telescope 14 is a single-mirror telescope in the 8-m class.
extreme telescopes of 104m, which should yield a reso- Pennsylvania State University 15 plans to build a point-
lution of 0.045" at 20 JLm and 0.00075" in the blue part able (but not steerable) telescope by assembling a spher-
of the spectrum. To extend this range and make it con- ical mirror from 73 spherically figured segments, each

Mirror 2

Mir o~
Mirror 1
Mirror 4
o
~
<> . .Nasmyth focus
Figure 13-A long optical path
makes it possible to combine the
beams from the four telescopes of

Mirror 6
y 0 t he ESO-VLT (two are shown here).
Up to seven mirrors can be used to
J1Mirror 5 J1 ' Combined
Coude
bring the beams from the four tele-
scopes to a combined Coude focus
'or tJ focus or to a fiber-optics system that will
distribute the light between instru-
o ments. (From Proc. IAU, Very Large
Telescopes, Their Instrumentation
Beam and Programs, Colloquium 79, p. 776,
combiner 1984.)

Interferometric
table (movable)

42 fohn s Hopkin s APL Technical Digest, Volume 10, Number 1 (1989)

 Peacock, Long - Astronomical Telescopes: A New Generation

with a diameter of 0.9-m and a 26-m radius of curvature. Within the next 20 years, some facilities described in
The effective diameter will exceed 7 m and will be used this article will have a decade of operational experience,
for spectroscopy only. and, possibly, the astronomers will be dreaming of the
The size and expense of these telescopes mean that next new generation of large ground-based optical tele-
no single organization will be able to build its own fa- scopes.
cility. The Magellan Project is a consortium of three or-
ganizations' the Columbus Project has four, and the REFERENCES
ESO has eight member nations. Large groups, each with 1D. Boorstin, The Discoverers, Random House, New York, p. 314 (1983).
an interest in a particular telescope, will become the com- 2E. Hecht, Optics, Addison-Wesley, Reading, Mass., p. 236 (1987).
mon approach, thus placing a burden on the communi- 31. S. Bowen, "The 200-Inch Hale Telescope," in Telescopes, G. P. Kuiper
and B. M. MiddJehurst, eds., University of Chicago Press (1960).
cation channels and raising problems in satisfying many 4M. M. Waldrop, "The Mirror Maker," Discover, 78-86 (Dec 1987).
diverse scientific objectives. 5c. v. Humphries, V. C. Reddish, and D. 1. Walshaw, "Cost Scaling Laws
The Hubble Space Telescope will receive most of the and Their Origin: Design Strategy for an Optical Array Telescope," in Proc.
IA U, Very Large Telescopes, Their Instrumentation and Programs, Colloqui-
public interest in the next decade, and, within certain um 79, p. 379 (1984).
scientific areas, it will reign supreme. It will not be 6R. W. Wood, "The Mercury Paraboloid as a Reflecting Telescope," As-
trophys. J. 29 (1909).
equaled in image quality or ultraviolet access by any fore- 7H. C. King, The History of the Telescope, Charles Griffm & Co. Ltd., Lon-
seeable instrument. The real astronomical revolution will don, p. 405 (1955).
occur on the ground, however. Many important scientific 8A. Y. S. Cheng and 1. R. P. Angel, "Thermal Stabilization of Honeycomb
Mirrors," preprints of the Steward Observatory, No. 812 (Mar 1988); also in
problems can be tackled from ground-based facilities just Proc. European Southern Observatory Con! on Very Large Telescopes and
as well, if not better, than from space. The number of Their Instrumentation (in press, 1988).
investigations and questions to which we seek answers 91. R. P . Angel and J. M. Hill, "Steps Toward 8 m Honeycomb Mirror Blanks:
III. 1.8. m Honeycomb Sandwich Blanks Cast from Borosilicate Glass," in
means that all available telescopes will be in full-time use. Proc. Soc. Photo-Opt. Instrum. Eng., Advanced Technology Optical Telescopes
Although the new telescopes are possible because of II 444, p. 184 (1983).
IOJ. R. P. Angel, "Steps Towards 8m Honeycomb Mirrors-A Method for
lower cost-manufacturing techniques, their expense is Polishing Aspheres as Fast as F/ l," in Proc. IAU, Very Large Telescopes,
still formidable, and they include more high technology Their Instrumentation and Programs, Colloquium 79, p. 11 (1984).
in their operation. Maximum exploitation of these tele- llJ. E. Nelson, T. S. Mast, and S. M. Faber, eds., The Design of the Keck Ob-
servatory and Telescope (Ten Meter Telescope), Keck Observatory Report 90
scopes will require high reliability. At times, such as with (1985).
the 10-m Keck Telescope, failure of the mirror-alignment 12R. G. Bingham, "The Two-Mirror Telescope (2MT)," in Proc. IAU, Very
mechanism can make it valueless. For the VLT, however, Large Telescopes, Their Instrumentation and Programs, Colloquium 79, p. 347
(1984).
such a failure will cause only a reduction in performance. 13R. G. Tull, "The University of Texas 7.6-m Telescope," in Proc. IAU, Very
The involvement of many groups with each system re- Large Telescopes, Their Instrumentation and Programs, Colloquium 79, p. 789
(1984) .
quires remote viewing and diagnostics, sophisticated data 14S. Isobe, "Japanese National Large Telescope (JNLT)," in Proc. /AU, Very
acquisition and analysis, and on-line data reduction. Large Telescopes, Their Instrumentation and Programs, Colloquium 79, p. 845
Clear and organized observational programs and inter- (1984).
15L. W. Ramsey and D. W. Weedman, "The Penn State Spectroscopic Survey
facility cooperation will be needed to enhance the scien- Telescope," in Proc. IA U, Very Large Telescopes, Their Instrumentation and
tific return. Programs, Colloquium 79, p. 851 (1984).

Johns Hopkins APL Technical Digest, Volume 10, Number 1 (1989) 43

Peacock, Long - Astronomical Telescopes: A New Generation

THE AUTHORS KNOX S. LONG is a research sci-

entist in the Department of Phys-
ics and Astronomy at The Johns
Hopkins University. He attended
Harvard College, where he was
KEITH PEACOCK was born in awarded an A.B. in 1971, and the
Hull, England, in 1940. He received California Institute of Technology,
a B.Sc. in physics from Durham where he received a Ph.D. in phys-
University in 1%1, an M .Sc. in ra- ics in 1976. From 1975 to 1981, he
dioactivity from Birmingham in was a postdoctoral fellow and as-
1962, a Ph.D. in astronomy from sistant professor of physics at Co-
Manchester University in 1967, and lumbia University. Since his arrival
an M.S. in technical management at Johns Hopkins, he has served as
from The Johns Hopkins Univer- a member of the senior research
sity in 1984. Before coming to APL staff, working primarily on space-
in 1979, Dr. Peacock worked on a based astronomical instrumentation.
variety of space and remote-sensing He is the project scientist and a co-
programs at the Bendix Aerospace investigator on the Hopkins Ultraviolet Telescope, a major shuttle in-
Systems Division in Ann Arbor, strument designed to study astronomical objects in far and extreme
Mich., and on laser fusion at the ultraviolet, which was built jointly by The Center for Astrophysical
Los Alamos National Laboratory. Sciences and APL.
For his flrst eight years at APL, he Dr. Long has conducted research at X-ray, ultraviolet, visible, and
worked on the Submarine Security Program in the Submarine Securi- radio wavelengths on a variety of astronomical objects. He is current-
ty Department. He joined the Space Sciences Instrumentation Group ly interested in the properties of evolved stellar objects-stars in mass-
in the Space Department in 1988. Before joining the Space Depart- exchanging binary systems, white dwarf stars, and the remnants of su-
ment, he was the recipient of the Lawrence Hafstad Fellowship, which pernova explosions. While at Columbia using the Einstein Observato-
he spent at the Center for Astrophysical Sciences working on optical ry, he carried out the first detailed X-ray survey of the Large Magellanic
problems connected with the Magellan Project telescope described in Cloud, as well as the flrst large-scale survey of the X-ray properties
this article. His primary field of interest is optics. Dr. Peacock is now of normal galaxies .
doing optical design for various Space Department programs and teaches Dr. Long is a member of the Science Working Group for the Magel-
classes in applied and modern optics at the G .W.c. Whiting School lan Project. He is also a Visiting Adjunct Associate of the Carnegie
of Engineering, The Johns Hopkins University. Institution of Washington. As such, he has observed frequently at Las
Campanas in recent years, having just completed a successful search
for supernova remnants in nearby spiral galaxies in the southern hem-
isphere.
Dr. Long is a member of the American Astronomical Society and
the International Astronomers Union. He has been a Visiting Fellow
of the Australian National University and a National Defense Educa-
tion Act Fellow. He is also a Fellow of the Center for Astrophysical
Sciences at Johns Hopkins.

44 fohn s Hopkins APL Technical Digest, Volume 10, Number 1 (1989)