Adaptive Optics

By Mike Luciuk

Atmospheric turbulence has bedeviled astronomers since the invention of the telescope. Degradation caused by the unstable atmosphere severely limits the utility of telescopic imaging. The invention of digital computers now permits real-time compensation for atmospheric seeing. Astronomers first proposed this in the 1950s.

The Department of Defense started developing techniques to overcome the effects of atmospheric seeing in the 1970s when they began to use surveillance satellites. The declassification of military accomplishments in the 1990s has combined with significant private sector and educational institutional developments to allow observatories to begin installing adaptive optics (AO) in many locations. Many pioneer researchers such as Babcock, Freid, and Roddier have contributed a great deal to the technology of AO.

An adaptive optics system is made up of three basic components: a wave front corrector (deformable mirror), a wave front sensor, and a control system (real-time computer). The illustration below is a schematic of the relationship between these three elements in a typical AO system.



The purpose of the wave front sensor is to estimate the distortion of the incoming wave front. The control system uses the wave sensor's output to update the deformable mirror. As the mirror tries to compensate for the incoming distorted wave front, part of the light is sent back to the wave front sensor. This real time closed loop system operates in the millisecond range.

Sensing the wave front requires a bright source, a laser beam, a bright star, a planet, or an asteroid. The laser beam stimulates Sodium atoms at an altitude of about 90 kilometers, creating a star-like point of light that can be positioned near the area of interest. Since its light passes along nearly the same path as the object of interest, it suffers almost the same distortion. A laser source is useful since bright stellar objects may not be available near every faint object of interest.

One well-known wave front sensor is the Shack-Hartman device. This is made up of many sub-pupils that focus onto a CCD. The slope of the incoming wave front can be determined from the projected positions from the sub-pupils.

The heart of the control system is a very fast real-time computer. It calculates the corrections based on input from the wave front sensor and it commands the actuators on the deformable mirror. These calculations must be executed in the sub-millisecond range to keep up with changing atmospheric conditions. For a 250-actuator system, computing power is on the order of 200 million instructions per second.

The diameter of the deformable mirror is usually in the range of 8 to 20 centimeters, but work is being done on larger secondary mirrors. The number of piezoelectric actuators on the mirror depends on the degree of correction required, the wavelength of interest, and the size of the telescope objective. Near perfect corrections on an 8 meter telescope in the visual range would require about 6400 actuators, an unrealistic number. As a result, perfect correction in the visual range is not possible at this time for the largest telescopes. For infrared, only about 250 actuators are required.

There are several concepts that deserve some mention in quantifying AO performance. The Freid parameter, ro, can be viewed as the maximum diameter telescope which can support diffraction limited images with a particular atmosphere. Typical ranges for ro range from 1 to 20 centimeters and these values can change rapidly with time. In effect, the Freid parameter is a measure of atmospheric turbulence, with ro decreasing as turbulence increases. In addition, . In other words, for the same atmospheric condition, the size of the Freid parameter depends on the wavelength. This is why corrections are easier to make in the infrared than in the visual spectrum. If ro = 10 centimeters at a wavelength of 550 nanometers, it would be at 3.4 micrometers in the infrared.

Another important AO concept is the Strehl ratio, R. This is a traditional criterion for image quality and is the ratio of maximum intensity of the point spread to that of a theoretically perfect point source image (Airy disk). A perfect (no distortion) diffraction limited image would have R = 1.0. Experience has shown that images with Strehl ratios greater than 0.37 look good, while images with lower R's look poor. All things being equal, . The Strehl ratio (image quality) improves as the Freid parameter increases. An AO configuration that produces a Strehl ratio of say, 0.9 in infrared, might yield only an R of 0.1 to 0.2 in visual wavelengths.

The error variance from basic statistics, of the incoming wavefront, , can be used to define the Strehl ratio. . The Strehl ratio, , where D is the diameter of the telescope objective and ro is the Freid parameter. From these relationships, it's clear that image quality from an AO system is sensitive to the incoming wave front error as well as the ratio of the Freid parameter (which is dependant on wavelength) to the size of the telescope.

Techniques for improving image resolution in amateur equipment have become available through the SBIG AO-7 Adaptive Optics system. This arrangement is more accurately an active rather than an adaptive optics system. It uses basic tip-tilt adjustments but there is no deformable mirror. It does have promise, however, in reducing wind turbulence effects and periodic drive errors.

It's clear that many challenges remain in the field of adaptive optics. This is an area that's finding utility in solar as well as nighttime astronomy. The Big Bear Solar Observatory in California is in the process of installing an AO system for better resolution at very narrow bandwidths. In the future, with the advent of faster real-time computers, sensitive and low-noise detectors, mirrors with thousands of actuators, and large deformable secondary mirrors, we can expect high Strehl ratios in visual wavelengths.



See "Adaptive Optics Comes of Age" by Govert Schilling on page 30 of the October 2001 issue of Sky and Telescope.

Return to Tutorials

Maintained by Ray Shapp
Page last updated 04/16/2004