An Overview of the Chicago Adaptive Optics System

M. F. Smutko, M. Chun, F. Shi, V. Scor, W. Wild, J. Larkin, & E. Kibblewhite

(This is a "reprint" of a poster presented at the San Antontio meeting of the American Astronomical Society, January 1996.)

Table of Contents:


The Chicago Adaptive Optics System (ChAOS) was begun in 1989 to develop an affordable high-order laser guide star adaptive optics system for large telescopes. Department of Defense systems have been too large, complex, and expensive to make them very useful to most astronomers. By concentrating on designing and constructing many of our components in-house, we were able to greatly reduce the size, complexity, and cost of an adaptive optics system. Some of the components that we have created include several deformable mirrors, the wavefront sensing and reconstruction electronics, and the operating and control software of the system. This paper outlines the construction of these components and summarizes the complete operational version of ChAOS.


ChAOS was designed to meet the need for an adaptive optics system that delivers excellent near-infrared image correction while still remaining affordable. Our goal was to achieve a Strehl ratio (the ratio of the on-axis intensity of the actual image to the on-axis intensity of an image produced by an ideal system limited only by diffraction effects) of 0.6 at a wavelength of 2.3 microns in one arcsecond seeing at the Apache Point 3.5m telescope.


ChAOS consists of several major components: a deformable mirror capable of bending on command to compensate for atmospheric distortion, a wavefront sensor to measure atmospheric distortion, and a high-speed reconstructor that receives data from the wavefront sensor and converts it into correction commands to drive the deformable mirror. To observe faint objects, we will employ a sodium laser beacon to create artificial guide stars. Additionally, we have written all of the software to drive our system. These components are described in the following sections.

Optical Layout Here is the optical layout of ChAOS, showing its major components. The interferometer is used to aid in system alignment and to monitor the deformable mirror during operation. It is not part of the control loop.

Deformable Mirrors

In-house construction of deformable mirrors began when early project funding left us without money to purchase a commercial deformable mirror. Early results with our "homemade" mirrors containing 7 and 37 actuators were very encouraging and we continued designing more complex mirrors. These experiments have led to our current state-of-the-art mirror which is a 201 actuator continuous facesheet mirror with a facesheet diameter of five inches.

Our deformable mirrors consist of three components: actuators made from the piezoelectric ceramic material lead zirconate titantate (PZT), an Invar baseplate, and a thin quartz mirror faceplate. The actuators are constructed with one inch long tubes of PZT-5H material, which provides a stroke of 1 micron for each 100 volts applied to the actuator. The quartz faceplates are 1 mm thick, prepolished to lambda/2 peak-to-valley over their surfaces, and coated with protected silver to maximize their reflectivity in the near-infrared. Using prepolished faceplates has two advantages: the facesheet can be made very thin using dual-lap polishing techniques (which reduces axial stresses on the actuators during operation because the force required to bend the facesheet is proportional to the cube of the facesheet's thickness) and superior mirror coatings can be achieved since the faceplate is coated ahead of time with high vacuum/high temperature procedures (and the entire deformable mirror need not be placed in a vacuum chamber).

201 Actuators This is the largest ChAOS mirror to date. It has 201 actuators arranged in a 16 x 16 square pattern (with the corners cut off). The facesheet is five inches in diameter. Note the penny for scale.
Three Deformable Mirrors These are three of the deformable mirrors we have constructed. From left to right they are: 59 actuators in a circular pattern, 87 actuators in a "quasi-hexagonal" pattern, and 97 actuators in a square pattern. Visible on the edges of each mirror are the PZT actuators.
Deformable Mirror and Driving Electronics this picture shows an 87 actuator mirror attached to its driving electronics. The driving cards receive commands from the Reconstructor Box and supply the high voltage signals to each actuator. To drive more actuators, one only needs to add more cards in the empty slots. This rack can drive up to 250 actuators.

Wavefront Sensor

ChAOS uses a Shack-Hartmann wavefront sensor to measure atmospheric distortion. This design uses an array of 16 x 16 small lenses to focus the light from a reference object (either a natural star or a sodium laser beacon) onto a CCD chip creating an array of 16 x 16 spots on the chip. As the atmosphere changes, the position of each focused spot of light produced by the lenslet array also changes. The adaptive optics problem effectively reduces to continually adjusting the deformable mirror such that each spot remains focused on the same position on the CCD, regardless of the atmosphere's motion.

Wavefront Sensor Camera This camera records the positions of the spots created by the Shack-Hartmann wavefront sensor. It was built for our group by J. Beletic and features a 64 x 64 pixel Lincoln Labs CCD chip with a readout rate of 2000 frames per second and a read noise of 13 electrons.

High Speed Reconstructor

Data from the wavefront sensor CCD passes to the wavefront reconstructor package which computes the instantaneous atmospheric wavefront and generates the correction signals that are applied to the deformable mirror. The reconstructor uses 16 digital signal processing (DSP) chips on a total of 4 "DSP cards" to compute the wavefront slopes from 256 subapertures, each having 4 x 4 pixels on the wavefront sensor CCD. 16 multiply-accumulate (MAC) chips on a total of 4 "MAC cards" then generate error signals and feed them back to the DSPs which generate control signals for the deformable mirror. The reconstructor currently takes 2 milliseconds to compute the control signals for a 250 actuator system. A feature of this reconstructor is that the boards operate in parallel and to improve performance, one only needs to add more cards.

Control signals are converted to the high voltages necessary to drive the deformable mirror on custom-designed amplifier cards. Each of these cards can drive up to sixteen actuators using a multiplexed driving technique: one Apex PA-88 high voltage amplifier applies an update voltage to each of sixteen actuators in turn by means of a series of high voltage switches. The time required to update all sixteen actuators is 250 microseconds. By using many of these cards running in parallel, any number of actuators can be updated in less than 300 microseconds. This technique has the benefit of minimizing the heat dissipated by the driving electronics since only a small number of amplifiers is required to drive a mirror. In fact, our amplifier rack is so small it is placed inside ChAOS. Typical power dissipation of the driving electronics is less than 100 mW per actuator.

Wavefront Reconstruction Electronics The black electronics racks house all of the wavefront reconstruction electronics. The racks take information from the wavefront sensor camera and convert it into drive signals for the deformable mirror. The Macintosh visible at the top left controls the operation of the entire system including optical alignment and calibration, selection of reconstruction matricies, and operation of the correction loop servo system. The screen at the top right displays information from the wavefront sensor camera.
Mirror Driving Board One amplifier drives up to 16 actuators.

Sodium Laser Beacon

In order for the wavefront sensor to provide meaningful information to the control electronics, it must have a sufficiently bright source to use as a reference beacon. For ChAOS, this corresponds to approximately a mv=9 star. When ChAOS corrects on a "guide star", anything else lying nearby in the same field (i.e. an galaxy or other faint astronomical object) also gets corrected, because turbulence is roughly identical over small "isoplanatic patches" a few arcseconds in diameter. Unfortunately, most astronomical objects are not lucky enough to be located next to a bright guide star in the sky; for these objects, ChAOS must create its own guide star. Our artificial guide star is created using a laser tuned the sodium D2 line (0.589 Ám) which is focused on the atmosphere's sodium layer--a layer of naturally occurring sodium (thought to be meteoric residue) at an average of altitude of 90 km. When the laser reaches this layer, it excites the layer's sodium atoms which then produce an artificial "star" bright enough to be seen by the wavefront sensor. [See the poster by Shi et al for more details on this laser system.]

Laser Beacon The laser beacon firing near the star Regulus. For more laser photos, click here.


All of the ChAOS control software has been written as Macintosh C code. One of the more unique pieces of software our group has written is the A+ package which allows the user to create reconstruction matrices (the matrices which take wavefront sensor information and "reconstruct" what the atmosphere is doing) quickly and easily. This package can be used to generate matrices for virtually any deformable mirror/wavefront sensor combination.


Construction of ChAOS began in 1990. The correction loop was first closed (i.e. the system corrected for turbulence in real time) in the laboratory on March 8, 1994. First correction of starlight on the Apache Point 3.5m telescope was achieved on May 13, 1995 and is described in the accompanying poster by Chun et al. Closing the loop on the artificial laser guide star is the final milestone for ChAOS and is expected in 1996. This will allow us to use our system on virtually any object in the sky.


Support for this work is provided by NSF Cooperative Agreement AST 89-21756.

Return to ChAOS Homepage