Jan. 7, 1999

How a down-to-the-wire computer fix at Cornell enabled a troubled spacecraft to take images of an asteroid

asteroid 433 Eros
Scott Murchiel/Applied Physics Lab
A composite photograph of asteroid 433 Eros, taken on the NEAR flyby on Dec. 23, 1998.

Some 240 million miles from Earth, a spacecraft hurtled through the black void of space, off its intended course. But thanks to the creation of a last-minute fix by Cornell University mission engineers during a tense 24 hours just before Christmas, the $150 million mission now has hundreds of new images of a distant asteroid.

The Near Earth Asteroid Rendezvous (NEAR) had almost been given up for dead after a signal failure Dec. 20, but after contact was renewed scientists quickly had to formulate a new mission plan. Thanks to the considerable ingenuity of researchers at Cornell, NASA's Jet Propulsion Laboratory (JPL) in Pasadena, Calif., and Johns Hopkins University's Applied Physics Laboratory (APL) in Baltimore, signals were sent enabling the spacecraft to capture images as it approached the asteroid 433 Eros. But it was touch-and-go.

Cornell astronomy professor Joseph Veverka leads the mission's science team in charge of the visual light camera and near-infrared spectrometer, two of the five science instruments carried by NEAR. His operations team here designs the spacecraft command sequences that point and operate the instruments. NEAR's primary mission is to remain in orbit around Eros for one year collecting high-resolution images and other science data.

The drama began Sunday, Dec. 20, when a signal was sent from APL to implement a main engine burn to reduce the spacecraft's speed from 2,180 to 700 miles per hour on its approach to Eros. This was to be followed by three additional burns to further reduce the speed to 11 miles per hour relative to Eros and enable the spacecraft to go into orbit around the asteroid Jan. 10. But the first engine burn went awry.

When the first main engine burn began, the spacecraft's built-in safety devices detected a problem and shut down the burn after one second. The spacecraft began to tumble violently, and onboard communications systems also were shut down. Contact between scientists and NEAR was lost for 36 hours. Instead of closing in on Eros at a relatively slow pace, the spacecraft continued to speed along at more than 2,000 miles an hour.

"Afterward, it was Black Monday. We thought we had lost the spacecraft. We thought that we had lost everything we had worked toward for the past four years," says Ann P. Harch, a research support specialist in Cornell's astronomy department, who designs command sequences to acquire camera and spectrometer data.

The failure of the burn meant the postponement of NEAR's orbit of Eros until next year. But mission scientists calculated that within about 60 hours NEAR would pass Eros, about two weeks ahead of schedule.

Cornell engineers and astronomers with help from colleagues at APL cobbled together a plan to salvage at least something from the flyby by commanding NEAR to take photographs and infrared spectrographic images of Eros. But no computer programs were available to send such commands.

"It was nerve-racking. We really weren't sure about the spacecraft. We didn't know if this was perhaps our last and only chance to see Eros," says Harch. "We really weren't sure if the main engine would burn or not."

For the impromptu Eros flyby, beginning at about 10:30 a.m. Eastern time Dec. 23, Harch, Maureen Bell, also a member of Veverka's Cornell team, and Scott Murchie of APL were asked at noon Dec. 22 to write a new, error-free, computer command sequence for the spacecraft. The sequence would have to slue the spacecraft to keep the instruments pointed continuously in the vicinity of Eros throughout the flyby, while simultaneously keeping the spacecraft's photovoltaic cells pointed toward the sun. Since Eros' location was not known precisely, it would require taking a mosaic of four images to guarantee capture of each view of Eros.

Harch had written the command programs for NEAR's flyby of asteroid Mathilde in June 1997, a task that took her six months. But the new sequence would have to be written in just 12 hours in order to reach the spacecraft in time to make the complex onboard revisions.

Using electronic mail and constant phone conferencing, Harch and Bell at Cornell kept in contact with Murchie in Baltimore, and with true grit, and much sweat, the trio wrote the sequence. It was sent electronically to APL in Baltimore close to midnight.

From midnight and into the morning hours, sleep-deprived APL scientists tested the sequence for errors that could permanently incapacitate NEAR. Then, the sequence was sent by microwave transmission to the craft during the early morning hours Dec. 23. The "enable command" for the sequence was sent at about 10 a.m. Eastern time and received by the craft at about 10:22 a.m., eight minutes before the first scheduled event in the imaging sequence. Had the enabling command arrived eight minutes later, the information would have been too late.

Then, more nail-biting. Scientists at Cornell, APL and JPL waited anxiously to see if the program worked. If it didn't, the scientists would have to wait until the rescheduled February 2000 rendezvous with Eros.

Hours later, Harch received an e-mail from Mark Robinson of Northwestern University, a member of imaging science team: "We have an asteroid! ... Here's a cut-out (partial frame) of M0089838063-F0-s showing 'Christmas Dinner!' "