Peter L. Wizinowich

Phase shifting interferometry in the presence of vibration: a new algorithm and system

A modification to the usual phase shifting interferometry algorithm permits measurements to be taken fast enough to essentially freeze out vibrations. Only two interferograms are time critical in this 2 + 1 algorithm; the third is null. The implemented system acquires the two time critical interferograms with a 1-millisecond separation on either side of the interline transfer of a standard CCD video camera, resulting in a reduction in sensitivity to vibration of 1-2 orders of magnitude. The required phase shift is achieved via frequency shifting. Laboratory tests comparing this system with a commercial phase shifting package reveal comparable rms errors when vibrations are low; as expected from an analysis of potential phase errors. However, the 2 + 1 system also succeeded when vibrations were large enough to wash out video rate fringes.

Millisecond phase acquisition at video rates

We can measure the phase of a static wave front with great accuracy and resolution by using standard phase-shifting interferometry techniques. Since wave-front changes during or between the acquisition of the individual interferograms cause measurement error, the measurement of transient phenomena still requires an essentially static wave front during acquisition. The 2 + 1 phase-shifting algorithm, originally developed to freeze out vibration when testing large optics, permits fast acquisition by requiring only two time-critical interferograms. The third interferogram provides a dc level or null and is only required once. The current 2 + 1 implementation is a low-cost interferometric system that is capable of phase acquisition in <2 ms at video rates. The repetition rate is limited by the readout time of the detector. The 2 + 1 has produced map movies of turbulent air acquired at the 30-Hz frame rate of a CCD camera.

Neural network adaptive optics for the Multiple Mirror Telescope

The MMT consists of six comounted 1.8 m telescopes from which the light is brought to a combined coherent focus. Atmospheric turbulence spoils the MMT diffraction-limited beam profile, which would otherwise have a central peak of 0.06 arcsec FWHM, at 2 microns wavelength. At this wavelength, the adaptive correction of the tilt and path difference of each telescope beam is sufficient to recover diffraction-limited angular resolution. Computer simulations have shown that these tilts and pistons can be derived by an artificial neural network, given only a simultaneous pair of in-focus and out-of-focus images of a reference star formed at the combined focus of all the array elements. We describe such an adaptive optics system for the MMT, as well as some successful tests of neural network wavefront sensing on images, and initial real-time tests of the adaptive system at the telescope; attention is given to a demonstration of the adaptive stabilization of the mean phase errors between two mirrors which resulted in stable fringes with 0.1 arcsec resolution.

System For Phase Shifting Interferometry In The Presence Of Vibration

Very accurate and detailed testing of small and moderate size optical surfaces can be realized by phase shifting interferometry. However, for the very large primaries needed for the next generation of telescopes, it is difficult and expensive to keep vibrations low enough for this method to work. A modification to the usual phase shifting interferometry reduction algorithm permits measurements to be taken fast enough to essentially freeze out vibrations. Only two interferograms are needed with an exact phase relationship; and these can be recorded very rapidly on either side of the interline transfer of a standard CCD video camera, prior to charge transfer readout. The third required interferogram is a null. In the developed implementation, two frequencies, dv/v.10-8, are generated with orthogonal polarizations. A Pockels cell rapidly switches the frequency entering the interferometer, resulting in a phase shift over the long path difference of the interferometer. The two time critical interferograms are acquired with a lms separation resulting in a reduction in sensitivity to vibration of one to two orders of magnitude. Laboratory tests were performed to compare this 2 + 1 system with a commercial phase shifting package. Similar phase determination accuracies were found when vibrations were low. However, the 2 + 1 system also succeeded when vibrations were large enough to wash out video rate fringes.

High resolution imaging at the Multiple Mirror Telescope using adaptive optics

The next generation of 6 to 10 m class telescopes is being planned to include the capability for adaptive wavefront correction. The MMT with its 7-m baseline, provides an ideal testbed for novel techniques of adaptive optics. Using a new instrument based on a six-segment adaptive mirror, a number of wavefront sensing algorithms including an artificial neural network have been implemented to demonstrate the high resolution imaging capability of the telescope. These algorithms rely on a freely available property of starlight, namely, its coherence over large scales, to sense directly atmospheric and instrumental phase errors across large distances. In this paper, we report results obtained so far with resolutions between 0.08 and 0.3 arcsec at 2.2-micron wavelength. We also show data indicating that at the level of 0.1-arcsec imaging in good seeing, the isoplanatic patch at this wavelength is at least 20 arcsec across.

Real-time adaptive wavefront correction by an artificial neural network at the multiple mirror telescope

Summary form only given. The authors reported the first results from an adaptive system operating online at the Multiple Mirror Telescope (MMT), in which the wavefront is sensed by a neural network, implemented on an array of 20 transputers. Star images were formed at a wavelength of 2.2 mu m by two coherently phased apertures of the MMT, and analyzed by the net. Outputs from the net were used to drive piezoelectric actuators on two segments of an adaptive mirror in the optical beam train, controlling five degrees of freedom of the wavefront. With the servo loop in operation, the corrected image shows significant power at the diffraction limit of 0.1 arcseconds (0.5 mu rad) and a much improved peak intensity.<<ETX>>

Atmospheric modeling with the intent of training a neural net wavefront sensor

Modeling atmospheric turbulence which plays a critical role in the training of neural network wavefront sensors is discussed in the framework of an adaptive optics program for the multiple mirror telescope. It is concluded that the accuracy of the wavefront correction possible with a neural network directly depends on the similarity of the training images to those seen in the telescope. The image simulations used in the training of neural network wavefront sensors are based on a random mid-point displacement (RMD) algorithm and sine wave summation algorithms. The RMD algorithm is considered to be an extremely fast method of wavefront generation used for very large arrays and image sequences without time evolution. Multiple turbulent layer atmospheric models based on the sine wave summation algorithm create image sequences with temporal structure functions that closely match real structure function data.

Interferometric testing of large optics in the presence of vibration

A modification to the usual phase-shifting interferometry algorithm permits measurements to be taken fast enough to essentially freeze out vibrations. Only two interferograms are time critical in this 2 + 1 algorithm; the third is a null. An error analysis has been performed for this new algorithm. The implemented system acquires the two time-critical interferograms with a 1-msec separation, on either side of the interline transfer of a standard CCD video camera, resulting in a reduction in sensitivity to vibration of one to two orders of magnitude. The required phase shift is achieved via frequency shifting.

Polarization Device For Improved Spectrograph Efficiency

Grating, and hence spectrograph efficiency, can be a strong function of the polarization of incident light. A polarization device has been designed to divide the light entering a spectrograph into two orthogonal polarizations, one of which is then rotated 90°, so that all of the light incident on the grating is polarized either parallel to or perpendicular to the grating grooves. Spectrograph throughput is thereby im-proved while also becoming independent of input polarization. Tests at the coude spectrograph of the Canada-France-Hawaii Telescope with an 830 f/mm grating revealed a 28% improvement in spectrograph throughput at a wavelength of 500 nm.