[4.0] The Reinvention Of The Telescope

v3.0.0 / chapter 4 of 5 / 01 nov 16 / greg goebel

* Following World War II, the 200-inch Hale telescope went into operation at Mount Palomar in California. It was effectively the ultimate classical telescope design, in more ways than one: not only was it by far the biggest telescope to be built to that time, but it pushed the available technology to its limits, with some telescope makers believing that the laws of physics prevented the construction of larger telescopes that were actually worth the trouble. Several decades would pass before new thinking, new materials, and computer processing would open the door to new telescope designs. By the 1990s, new telescopes were in operation that greatly surpassed the Hale telescope, with projects in the works that would dwarf it.

Keck telescope



* George Ellery Hale had worked himself to exhaustion in building his giant telescopes. He tended to be obsessive, becoming so stressed in his work on the Hooker telescope that he ended up in a sanatorium for a time, holding conversations with a little green elf that only he could see. Hale finally had to retire in 1923. However, given his personality, he simply couldn't leave well enough alone, and began to lobby for a new record-breaking reflecting telescope. He originally set his sights on a 7.62-meter telescope, but wisely decided that was too ambitious, and settled for a 5.08-meter -- 200-inch -- design.

Hale began to accumulate grants for the super telescope and put his plan into motion. He originally wanted to put it on Mount Wilson, but Los Angeles was growing rapidly at the time and the "light pollution" from the city, as well as the air pollution from increasing numbers of cars, had undermined the value of the site. A bit of searching suggested that Mount Palomar, 145 kilometers to the southeast of Mount Wilson, would be a good site for the new instrument.

Site selection wasn't the biggest of Hale's worries, however. Not only was fabricating such a huge mirror a challenge in itself -- the mirror for the Hooker telescope had been enough of a pain, and it was relatively small -- but the size of the mirror would create problems of its own. Weight was one issue, since the mirror might easily deform as the telescope was shifted around, but thermal expansion was a worse one. Such a large mirror would expand and contract with temperature changes, ruining the curvature of its precision surface. The larger the mirror, the longer it would take to stabilize.

Fortunately, the Corning company had recently developed a new type of glass tradenamed "Pyrex", which had a coefficient of thermal expansion only a third as great as that of normal glass, meaning that it would only expand a third as much for a specific increase in temperature. It was also relatively cheap and easy to work with. To reduce weight, the blank was to be cast over an array of solid "cores", giving it a honeycombed appearance from the rear. Reducing the mass of the mirror also meant that it would stabilize more quickly with changes in temperature.

The first blank was poured at the Corning works in New York state on 24 March 1934. Unfortunately, some of the cores came loose and remained embedded in the blank, distorting and ruining it. Structural reinforcement and a core cooling system were added to ensure that the cores wouldn't pop loose, and the second blank was poured on 2 December 1934. The blank was cooled for ten months to make sure air bubbles weren't formed, and then hauled by train across the country to Pasadena, California. The train had to move slowly, and the route had been carefully planned to ensure that the wide blank in its huge packing crate wasn't blocked by narrow tunnels or bridges. In some places it had to be eased through, with a clearance of only centimeters on each side.

After the blank arrived in Pasadena, the long and laborious process of grinding began. 4.5 tonnes of Pyrex were polished off, with the final mirror having a weight of 14.5 tonnes. The central hole in the mirror was 1.01 meters across, as big as the objective lens of the biggest refractor in the world. The mirror was silvered with a new "vapor deposition" technique that had been invented by an American astronomer named John Strong, in which the mirror was placed in a vacuum chamber in which aluminum wires had been strung; the wires were vaporized with a strong electric current, with the aluminum vapor deposited on the mirror in a very thin layer that didn't need polishing.

The mirror was mounted on a plate with 36 support studs penetrating into holes in the bottom of the mirror, with the support studs featuring an internal system of levers and counterweights to ensure the weight of the mirror was evenly distributed. A cover was fitted around the top edge of the mirror, with metal petals that folded in to the center to keep out dust and debris. Imagery could be obtained through the prime focus, with a viewing cage big enough to accommodate an astronomer; a Cassegrain focus; or a coude focus. The telescope assembly was placed on an equatorial mount with a complicated electrical tracking drive assembly. The mount pivoted on oil bearings to ensure smooth movement. Total weight of the entire telescope was 482 tonnes, and it was accommodated in a neat dome with a height of 41.1 meters.

Hale telescope

World War II slowed down work on the giant telescope, but it was finally dedicated on 3 June 1948 -- though initial observations showed the mirror had to be slightly reground, with the instrument back on line by the end of 1949. It was named the "Hale telescope" after its godfather, George Ellery Hale, who had died in 1938. The telescope had been completed under the direction of Max Mason (1877:1961), a professor of math and physics, and the director of the Rockefeller Foundation.



* The Hale telescope was a wonder of the world when it went online, and remained so for decades. It was, in a sense, the end of the road for classical telescope design; there was not much that could be done to seriously improve on it as it stood. There were three limiting factors on building better telescopes:

The Soviets built a 6.1-meter -- 240-inch -- Richey-Chretien telescope, much bigger than the Hale, in the Crimea, with the telescope going into service in 1977. The "Bolshoi Teleskop Azimutal'nt (BTA / Large Azimuthal Telescope)" featured a computer-controlled alt-azimuth mount that was much lighter than a classic equatorial mount and would become standard later, but it was otherwise seen as a disappointment. The initial mirror suffered from cracks; it was quickly replaced by a better mirror, but the big mirror would not stabilize overnight if the change in temperature from day to night was more than 2 degrees Celsius.

The site, though at high altitude and isolated from ground lighting, also suffered from variable weather and high winds. The mirror had degraded due to use of solvents to strip off aluminum for refinishing; the original blank was reground to strip off the defects. It would have been preferable to obtain a new mirror made of materials with less thermal sensitivity, but the money wasn't available.

* The arrival of diminishing returns didn't mean that astronomy had come to a dead end by any means. A single telescope can of course only observe one bit of sky at a time, and so building more big telescopes helped improve the ability of astronomers to search the sky and keep an eye on transient or repetitive events. The Hale telescope, the world's biggest telescope at the time, was soon joined by what was then the world's biggest Schmidt camera, with a 1.22-meter -- 48-inch -- lens and a 1.83-meter mirror. The correcting lens for a Schmidt is thinner than that for a refractor, and can be made bigger. The instrument was known as the "Big Schmidt", in contrast to the "Little Schmidt", which had a 45.7-centimeter correcting lens, a 66-centimeter mirror, and had been set up at the site in 1936.

The Palomar Big Schmidt complemented the Hale telescope, providing wide-area photographic sky surveys with unprecedented detail. In 1949, a team of researchers backed by Caltech and the US National Geographic Society used the two Palomar Schmidts to conduct a survey of the northern skies, with the survey taking five years to complete and producing 1,870 plates.

A bigger Schmidt camera, with a 1.34-meter lens and a 2-meter mirror was later set up at Jena, then in Communist East Germany, but for whatever reasons it was still overshadowed by the Palomar Big Schmidt. The Big Schmidt was renamed the "Samuel Oschin Telescope" in 1987, in honor of a benefactor.

Oschin Schmidt telescope

* Supporting technologies were undergoing improvements as well. New photographic emulsions were introduced in the 1970s that provided much greater color selectivity, finer grains, and (with special treatments) faster exposure times. Photographic printing processes were also improved, allowing prints to be made in which faint details weren't drowned out by bright details; such techniques allowed more information to be extracted from old archival plates as well as new imagery.

By that time, plates could also be digitized by scanning them using a "photomultiplier tube (PMT)". The PMT was a type of vacuum tube that featured a "photocathode" at front that would emit electrons when hit by light. The electrons would then strike the first in a chain of cathodes plates with an increasing positive charge, with this cathode emitting more electrons; these electrons would fall on the second cathode, emitting more electrons; and so on down the chain of cathodes, greatly increasing the electron flow, which was then finally measured and converted to a digital value. A PMT could also be used to directly scan images from a telescope, where its extreme sensitivity would allow it to pick up very faint objects.

In the 1960s, technology conceptually similar to an array of PMTs was developed to create the "night vision scopes" used in Vietnam, which led later to modern "night vision goggles". Such "image intensifier" technology not only provided greater sensitivity than traditional photographic schemes, but it also provided a lever to help deal with the problems of atmospheric "seeing".

In 1969, a French optical physicist named Antoine Labeyrie (born 1943), then working in at the State University of New York at Stony Brook, came up with a scheme called "speckling" that leveraged off the fast response time of an image intensifier. Labeyrie reasoned that at any one instant, atmospheric distortion would break up the point image of a star into a scatter of multiple points of light, which he called "speckles". An image intensifier was fast enough to pick up the pattern of speckles in less than a 30th of a second; a sequential set of such "specklegrams" could be digitized and then "crunched" in a computer with a statistical algorithm known as "vector autocorrelation" that sorted out the true image.

As mentioned, Labeyrie was an optical physicist, not an astronomer, and when he got in touch with the management of the Palomar observatory to tell them of his idea, they didn't exactly know what to make of him. They contacted the astronomy department at Stony Brook, and Professor Stephen Strom went over with two grad students, Robert Stachnik and Daniel Gezari, to chat with Labeyrie. They decided he knew what he was talking about, so in 1970 Stachnik and Gezari arranged for time on the Hale telescope to check out the idea. Sure enough, it worked.

At about the same time, Bell Telephone Laboratories -- "Bell Labs" -- was working on a solid-state silicon memory scheme known as a "charge coupled device (CCD)". It featured rows of memory elements, in which charges could be stored to represent a digital "1" or "0" or "bit". The bits in a row couldn't be accessed directly; they had to be shifted around on a regular cycle.

The CCD seemed to offer improved memory densities for computer systems, but conventional solid-state memory systems quickly outpaced it. However, the CCD could also be designed so that light falling on the elements would set up electric charges in them, which could be then shifted out and read. Bell Labs demonstrated a grayscale (black and white) CCD camera in 1971 and a color CCD camera in 1972. Researchers at the US National Aeronautics & Space Administration's Jet Propulsion Laboratory (NASA JPL) in Pasadena, California, thought the CCD would have considerable potential for lightweight cameras for planetary probes, and assigned an engineer named James Janesick to see what he could do with the idea.

Janesick investigated CCD technology and decided they were a big step ahead of any astronomical imaging system known to that time. The CCD was sensitive, could pick up a wide range of wavelengths, and had low noise levels (particularly when it was cooled). In 1975, Janesick, colleagues from JPL, and researchers from the University of Arizona demonstrated a CCD camera with a 1.55-meter telescope at Mount Lemmon near Tucson. They obtained an electronic image of Uranus that was far clearer than any obtained to that time.

That got the CCD ball rolling. By 1976, the Hale telescope was using a CCD with a resolution of 800 x 800 picture elements or "pixels" known as the "Prime Focus Universal Extragalactic Instrument", a somewhat tortured name that converted into the acronym "PFUEI". (For non-native English speakers, "phooey" is a somewhat antiquated exclamation of disgust or contempt.) PFUEI was built by an astronomer with a bent for tinkering named James Gunn (born 1938), and the name appears to have reflected some annoyance with the low pixel resolution of the instrument -- even a cheap modern digital pocket camera would put it to shame. Gunn came up with an idea for ganging four CCD chips into a single imaging system, and the "Four-Shooter" was put online with the Hale in 1982.

CCDs are now standard kit for any serious telescope. The Oschin Schmidt at Mount Palomar, for example, is fitted with an array of an array of 112 CCD imagers, giving it a field of view of about 4 x 4 degrees and allowing it to cover 500 square degrees of sky per night. The camera system, built by Yale and Indiana University and named after the "Quasar Equatorial Survey Team (QUEST)", produces a flood of data every month.

The QUEST camera was not the first digital camera fitted to the Oschin Telescope; a much less powerful CCD system, the "Near Earth Asteroid Tracker (NEAT)", was installed on the Oschin in 2001. It is likely that QUEST will be followed by an even more capable system.

Incidentally, up to the introduction of CCD imaging, the Oschin Schmidt had accumulated 19,000 photographic plates, which were archived at Caltech. They were all packed up and hauled off to the Palomar Observatory in 2002. Presumably they have, or will be, digitized, the astronomy community having a desire to get all the huge number of astronomical photo plates accumulated up to the digital age, to ensure preservation and permit computer analysis.



* The new electronic imaging systems and computer processing were a big step forward, but making more progress with telescopes meant coming up with new schemes for telescope design.

The first major attempt to rethink the telescope came in 1979 with the entry into service of the "Multiple Mirror Telescope (MMT)" at Mount Hopkins Observatory in Arizona. The MMT consisted of six 1.8-meter mirrors on a single alt-azimuth mount and feeding a common focus, giving an equivalent aperture of 4.5 meters. This was a respectable aperture, making the MMT the world's third-biggest "eye" when it was completed, but not a massive step ahead in capability compared to contemporary technology; the MMT was partly intended as a prototype for a new generation of telescopes.

The basic concept of a multiple-mirror telescope had been published by an Irish scientist named Edward Synge in 1930. Frank Low (1933:2009) of the University of Arizona reinvented the idea in the late 1960s, later finding out about Synge's work during literature search. Low and his colleagues obtained funding from the University of Arizona and the Smithsonian. They managed to obtain the mirror blanks from the US Air Force, which had built them for a military space station named the "Manned Orbital Laboratory (MOL)" that had been canceled.

Using six relatively small mirrors instead of one big mirror addressed the problems of mirror fabrication and thermal stability. The alt-azimuth mount also reduced cost compared to an equatorial mount, but it made pointing and tracking the telescope to keep up with the rotation of the Earth more troublesome; in fact, the alt-azimuth mount had been effectively abandoned for large telescopes after the introduction of astronomical photography, because it was only possible to keep a long exposure on track using an equatorial mount. The answer was to use a computer to direct the telescope, with computer control completely reversing the logic that made the equatorial mount standard for large telescopes. An equatorial mount is relatively heavy and expensive; steering is a simple task for a computer, and can be performed with minimal computer hardware. Equatorial mounts also demanded carefully alignment of the mounting system, while tweaks to a computer-controlled alt-azimuth mount could be made, somewhat less painfully, in software.

Computer control meant that for future large telescopes, the alt-azimuth scheme would be the norm instead of the equatorial mount. A computer control system was also used to ensure that the six mirrors stayed on a common focus. In addition, the MMT used electronic detectors from the outset, and it has been referred to as the first "all-electronic" telescope.

Mount Hopkins Multiple Mirror Telescope

* It took over a decade longer for next-generation telescopes to truly arrive, in the form of the 9.82-meter Keck Telescope, which went online at the Mauna Kea Observatory on the island of Hawaii in 1991. The Keck took a radical approach to building a large telescope, constructing the mirror out of 36 hexagonal segments, each 1.9 meters across. It was built by the University of California and Caltech using a grant from Keck Foundation, set up by William Keck (1880:1964), a wealthy oilman and Caltech alumnus.

The concept of the segmented mirror was not entirely new. Lord Rosse had tinkered with the idea a bit, and in the 1950s an Italian astronomer named Guido Horn-d'Arturo built a segmented-mirror telescope with an effective aperture of 1.8 meters using 61 hexagonal tiles, each with a spherical curvature but arranged on a parabolic base. The tiles were manually adjusted using a system of screws to keep the telescope on focus. Horn-d'Arturo's telescope was regarded as little more than a curiosity, and in fact the concept wasn't really practical until the development of computerized optical control. Keck chief scientist Jerry Nelson and his team picked up the idea and realized the technology was available to put it to good use.

The segmented mirror represented a successful attempt to trade "smarts" for brute force, allowing the construction of a telescope whose scale left the Hale Telescope in the dust. The individual segments were relatively easy to handle and transport and generally avoided the thermal-stability problem. However, it wasn't an easy approach by any means and there were difficulties getting it to work.

A total of 43 blanks made of "ZeroDur", a modern low-expansion ceramic, were obtained, with the additional seven as spares and insurance. Grinding each segment took one to two months, and it was a very tricky operation, with a carefully planned sequence of steps:

Once assembled into the final telescope, the shape of the segments was continuously modified by a computer-controlled feedback system. There were two sensors mounted on every inter-segment edge, for a total of 168 sensors, and there were three actuators for each segment, for a total of 108 actuators. The control system read the sensors, computed the proper shape for the mirror, and then modified the actuators to change the segment shapes accordingly. The sensors measured the height difference between the two segments using capacitive sensing. A paddle attached to the back of one segment fit between a pair of plates attached to the other, and the change in capacitance indicated the mirror displacement.

Like the mirror itself, the sensors were built out of ZeroDur to ensure they remain stable with temperature. The actuators were attached to the wiffletrees and the telescope structure. They were screw drives, allowing precise travel over a range of 1 millimeter. The control system scanned the mirrors and adjusted them twice a second, giving the illusion of a mirror that was "absolutely stiff".

Keck twin telescopes

A second Keck telescope was completed on Mauna Kea in 1996. Each was, at the time, the biggest telescope in the world by far. They were equipped with the latest instruments and digital imaging systems. The "seeing" from the Mauna Kea site is unsurpassed, doing much to reduce the problems of atmospheric interference.

* The Mount Hopkins MMT itself was replaced in 1999 by a new telescope in the same building that was more or less a classical single-rigid-mirror reflector design. Telescope maker Roger Angel, an expatriate Briton at the University of Arizona, came up with an idea for fabricating a large honeycomb mirror made of low-expansion borosilicate glass using "spincasting". Angel wanted to defy the belief that the limits had been reached with the Hale on building bigger one-piece mirrors. Spinning the mirror blank while it cooled gives the surface a parabolic curvature, reducing the need for grinding. The mirror blank has a honeycomb core build of alumina silica fibers that are flushed out after the mirror is cast. This results in a very big mirror that is very light for its size, and so does not have a great tendency to distort under its own weight. Spincasting can be used to fabricate mirrors up to 8.4 meters in diameter.

Magnum Mirror Telescope

Using this technology, a 6.5-meter mirror was fabricated for the new Mount Hopkins telescope, which was dedicated in 2000 and was the biggest single-mirror telescope in the world at that time. The mirror was said to have been ground to such precision that if it were scaled up to the size of the United States, its curvature would deviate from perfection by only 2.5 centimeters at most. The new telescope kept the "MMT" acronym in honor of its famous predecessor -- though it was redefined as the "Magnum Mirror Telescope".



* Another approach to building a very large mirror was to make it relatively thin for its size, only about 20 centimeters thick. That in itself would result in a very "wobbly" mirror, but it could be kept rigid by a framework of sensors and actuators conceptually similar to that used on the Keck Telescopes to keep all the segments aligned. Such "smart" telescope mirrors not only permit the construction of workable, precise unitary or segmented telescope mirrors, they also help address the problem of atmospheric interference through a scheme known as "adaptive optics".

The idea is conceptually simple: instead of using the sensor-actuator system to maintain a constant surface, it modifies the surface on an ongoing basis to compensate for the variations in the atmosphere. Focus is maintained by a feedback system that focuses on a bright star in the field of view, or if there's no bright star available, a laser shining into the sky above. Adaptive optics can only really provide a tight focus at the center of the field of view, making it well-suited for close-ups but not for wide-field surveys. The technology was developed for the US "Star Wars" missile defense program during the 1980s, and declassified in 1991.

* Another telescopic technology that has come of age roughly in parallel with adaptive optics is optical interferometry. As discussed earlier, optical interferometry involves using optical interference techniques to perform precise measurements of the difference in phase of light sensed by two separated telescopes, allowing them to perform observations with a resolution equivalent to that of a single telescope with a mirror the diameter of the "baseline" separating the two small telescopes; however, the light-gathering power remains the simple sum of the areas of the two telescopes. Michelson and Pease had performed interesting experiments with optical interferometry on the Mount Wilson telescope back in the 1920s, but ran into technical obstacles in scaling up the concept.

At the long wavelengths used by radio astronomy, interferometry requires much less precision, and it has been used in radio astronomy almost since the beginning of the field. In 1975, a physics graduate student at the Massachusetts Institute of Technology named Michael Shao, now with the US National Aeronautics & Space Administration's Jet Propulsion Laboratory (NASA JPL) in California, finally devised a scheme to make optical interferometry practical. Shao's invention was the "mechanical delay line", which consisted of a set of movable mirrors mounted on rails and controlled by a feedback positioning system that tracked the brightest fringe in the optical interference pattern.

The positioning system tracked shifts in the fringes and adjusted the position of the mirrors every hundredth of a second by as little as a few tenths of a micron (millionth of a meter), damping out the noise. The mechanical delay line also compensated for the effect of the Earth's rotation, which otherwise caused the interference fringe pattern to "drift", by moving the mirrors a millimeter a second.

Shao developed a series of optical interferometers based on his concepts, resulting in the "Mark III", which went into operation at the Mount Wilson Observatory site in 1986, not far from where Michelson and Pease performed their experiments. The Mark III consists of two moveable light collectors that can be spaced as far as 31 meters apart. At maximum extension, it can perform measurements to a resolution of 0.0005 arc-second.

In the meantime, other researchers at the University of Sydney in Australia and the Observatoire de la Cote d'Azur in France had also developed workable optical interferometers with even longer baselines, using computers to compensate for noise computationally, instead of delay lines.

* These early optical interferometers were capable of making precise measurements of the positions and diameters of stars, but they could not be used to construct images. That required using computing power to soak up phase comparison observations and generate an image out of them.

The first imaging interferometer systems were put into operation in the early 1990s. These included the University of Cambridge's "Cambridge Optical Aperture Synthesis Telescope (COAST)", and the US Naval Research Laboratory (NRL) and US Naval Observatory's (USNO) "Naval Prototype Optical Interferometer (NPOI)", also known as the "Big Optical Array (BOA)", set up on Anderson Mesa in northern Arizona.

COAST consists of a set of four 40-centimeter telescopes mounted on steel trusses in a 6-meter-wide "Y" pattern. Light from the telescopes is reflected through buried pipes with mechanical delay lines to a central combiner system that uses a computer to interpret the interference data and produce and image.

A few experimental infrared interferometry systems preceded COAST, but it became the first operational imaging optical interferometry system in September 1995, when it obtained images of the Capella binary star system. NPOI, which uses six telescopes, followed it into operation a few years later. A more recent array, built by the Georgia State University's Center for High Angular Resolution Astronomy (CHARA) at Mount Wilson, features six 1-meter telescopes, ringing the Hooker telescope, the array featuring a baseline of 330 meters.

Other optical interferometer systems have been, or are being, built. Although radio interferometry can be performed with telescopes on different continents, optical interferometry's demand for vastly greater precision means that an optical interferometer array can only really span a single site. The maximum baseline now under consideration -- for an optical interferometer connecting some of the big telescopes on Mauna Loa with fiber-optic links -- is 800 meters.