A Detector Array for Direct Control of a Deformable Mirror
Robert Winsor, Margaret Frazier, Michael Krueger, Tim Myers
 
Discussion
    This detector array can be thought of as an integrated wavefront sensor that outputs the shape of the wavefront.  It allows for high refresh rates due to the signal processing electronics residing within the array.  The detectors are expected to have significantly lower noise than existing wavefront sensors due to the pixels being active and the signals do not need to travel down long wires.  Although the fill factor seems low, it is actually comparable to existing wavefront sensors that map single actuators to clusters of pixels, typically 4 x 4 pixels or larger (where each pixel has a 6% to 10% fill factor of the actuator map).  There is no loss of sensitivity as a result.
    This array can easily be expanded to a 63 x 63 matrix (and even larger), allowing it to drive deformable mirrors in excess of 3100 actuators, with no reduction in refresh rate.  This is particularly attractive to the Military and Astronomical communities.
    A prototype of this detector is being fabricated using a CMOS 0.5mm process.  Testing of this prototype will reveal the actual refresh rates that can be attained as well as the sensitivity (flux) levels that are needed.  There are some potential problems that might arise when using this detector in a scientific instrument.  One concern is the use of the metal 3 layer to block light incident to the areas of the array that are processing the signals.  Light incident on this layer is reflected back to the instrument, possibly causing an unacceptable rise in the background signal level.  If this proves to be a problem, methods will be explored at a later time to solve the problem, however, some potential solutions can be presented.  A chip without a window is a good intial step toward solving this issue, as this would substantially increase the path length of the return signal.  If this is not adequate, an investigation into extra lithography steps to mask the metal is the next step.  Other methods can be explored using optical techniques of tilting the array slightly to get the reflected path to walk off into a beam dump.
    It is useful to comment on refresh rates.  For applications involving low light (such as Astronomy), refresh rates present a trade-off in performance.  Higher refresh rates require higher photon fluxes, which are not always (in fact seldom) available.  However, they also result in better overall system performance due to the ability to correct for distortions occurring at a faster rate.  For this detector array, it is expected that peak fluxes as low as 1x108 are needed per pixel, or for the entire array, 4x1010.  For astronomy, this flux rate is too high for all but the brightest stars (unless laser guide stars are used).  However, this technique is still worth exploring as it will likely improve with fabrication processes outside of those available in the MOSIS tinychip program.  For now, this is a good test of the technique.  Also, there are many applications that do not have a limitation on flux, and this method of wavefront reconstruction and control of a deformable mirror should be very attractive.
 
Abstract
Methods
Results
Discussion
Ackowledgements

Send correspondence to: winsor@stsci.edu

07 December, 2001