I. The Problem

Telescopes on small, light, orbiting observatories are susceptible to pointing errors caused by reaction wheel rumble.

The Jupiter Magnetosphere Explorer (JMEX) is a proposed earth-orbiting satellite with a 0.5 m telescope which will observe Jupiter in the UV with 0.25 arcsec (") resolution. If uncorrected, vibrations from the reaction wheels, even though well-damped, would cause random pointing errors with an amplitude up to 5" at frequencies around 1Hz.

II. Proposed Solution

The JMEX approach to mitigating pointing errors is not to eliminate them completely, but to measure the motion of the spacecraft line-of-sight and correct the data with this pointing knowledge. The main science camera, called the Ultraviolet Imager (UVI), is a photon-counting MCP detector which produces data as a time-stamped photon list with 0.08" spatial resolution and ~1 ms temporal resolution. Simultaneously, a 0.5" pixel video camera, called the Image Motion Sensor (IMS), is fed by a pickoff mirror in the main beam and captures visible images of Jupiter's disk at 30Hz. With onboard processing, the centroid of the planet is determined, frame-by-frame, with a resolution <0.02" (1/25 pixel). With inter-frame interpolation, each photon from the UVI camera is position-corrected in ground post-processing to an accuracy of 0.02".

III. Hardware Demonstrator

To rigorously test this 2-camera post-detection scheme, we have constructed a hardware mock-up consisting of a tip-tilt mirror, a beam-splitter, and two video cameras with controlled noise characteristics. The tip-tilt mirror simulates pointing drift of the spacecraft by producing controlled image motion over a range of amplitudes and frequencies.

Acknowledgments

This work was supported by NASA's Small Explorer Program and the Laboratory for Atmospheric & Space Physics at the University of Colorado at Boulder. We are grateful to Ball Aerospace for technical support and for the loan of the tip-tilt mirror. We also thank Theologos Sosonis of Unibrain Inc for expert technical assistance with the camera control software.

Target motion tracked by the 2-camera system. Distance measured in IMS camera pixels. The UVI camera data has been scaled and rotated with the known magnification and tilt to match the pixel scale of the IMS camera.

Measured centroid residuals. Points shown are the differences between UVI reference camera positions and the interpolated IMS camera positions. The horizontal and vertical bars show the FWHM centroid error of about 0.031 pixels = 0.016" .

V. Results

(Top) IMS image of Jupiter, digitally noised to n = 2000 counts/pixel and readout noise = 74 counts. The disk is 80 pixels in diameter. Despite the noise, the disk centroid can be determined with an uncertainty of about 0.02 pixels.

(Left) Clean UVI reference image of Jupiter, to same scale as IMS image.

(Right) Time-averaged UVI reference camera images of Jupiter, showing large amplitude image motion.

IV Images

Michael Dubson, Nicholas Schneider, and Steve OstermanUniversity of Colorado at Boulder

2n ron1 1X ,

n n2 2

VI. Where's the noise floor?

An absolute lower limit on the uncertainty the centroid (center-of-light) of the image of a bright disk is set by photon-counting statistics and readout noise. Assuming an image bright enough to allow clean background subtraction, the 1 uncertainty of the centroid position is given by

where n is the count per pixel and ro is the readout noise per pixel. Notice that the centroid uncertainty is independent of disk radius.

VII. Conclusions

Ughh! We did really good, but still not funded.

Photon-by-photon post-processing correction of pointing errors in an orbiting satellite