New Hubble trouble: spectrograph awry.

Add another malfunction to the spate of problems plaguing the Hubble Space Telescope. A faulty power supply has forced NASA to halt indefinitely all research conducted with a key spectrograph aboard the Earth-orbiting craft.

The ailing spectrograph, which uses diffraction gratings to separate ultraviolet light into its component wavelengths, has significantly higher spectral resolution than Hubble's only other spectrograph. Unless engineers can replace or compensate for its defective power supply, astronomers will have to forgo several types of Hubble observations that require resolution of minute differences between spectral lines, says Blair D. Savage of the University of Wisconsin-Madison, who helped design the spectrograph. These would include studies of the chemical composition and abundance of interstellar matter, he notes.

Scientists got their first inkling of Hubble's latest trouble on July 24, when the Goddard High-Resolution Spectrograph failed to relay some of its ultraviolet data. But the problem abated within minutes, and researchers didn't become alarmed until it recurred during observations on Aug. 5.

In the past month, researchers from NASA's Goddard Space Flight Center in Greenbelt, Md., and the Ball Corp. in Broomfield, colo.-which built the Goddard-designed spectrograph 10 years ago-have identified what they consider the likely cause of the communications failure. The suspect: a defective solder joint on a low-voltage power supply that provides the electricity for one of the two adjacent electronic sections of the spectograph. Each section powers roughly half of the instrument.

The balky voltage source can't supply enough power to operate the photon detector and other components on the section known as side 1, says Goddard's Jean Lane, experiment manager for the spectograph.

Side 2, powered by a separate voltage source, still functions normally, she says. But for scientists eager to continue observations, there's a catch. In order to transmit information from the spectrograph to the ground or to an onboard computer or tape recorder, a device called a formatter must first package the raw data. And the formatter now connected to side 2 gets its voltage from the defective power supply. As a result, Lane says, side 2 transmits data only intermittently.

Hubble officials have concluded that continued use of the crippled spectrograph would waste precious observing time. Two weeks ago, they temporarily suspended all science operations with the spectrograph, says Peter Stockman, deputy director of the Baltimore-based Space Telescope Science Institute, which coordinates Hubble observations.

Preston Burch, NASA's deputy project manager for Hubble operations and ground systems, says engineers will begin tests next month to gather more information on the equipment failure. They hope to determine whether changes in such factors as the spectrograph's temperature can boost the power supply's performance enough to allow more frequent and predictable data transmission by side 2. The tests could take up to a month to complete, Burch says.

If testing does not suggest a way to improve the power supply, NASA still has two key options for salvaging the spectrograph, says Goddard astrophysicist Sara R. Heap, a member of the spectrograph research team. Using software relayed from the groun to the telescope, researchers could attempt to connect side 2 to a second onboard formatter, which does not use the defective voltage source. The formatter change could take two to three months. although side 1 would remain inactive, a successful switchover should fully revive operations with side 2, Heap says.

But the undertaking poses some formidable risks, she adds. Hubble's four other science instruments -- which now function normally -- would also have to switch to the second formatter, requiring ground-based technicians to temporarily turn off their power, place the instruments in an active "safe mode" for several days, and then power them back up. Power surges inherent in this process might damage equipment, notes Burch. Moreover, says Heap, the second formatter has not been tested in space. Some scientists worry that if this formatter does not operate properly, the engineers may have difficulty switching the instruments back to the original device.

"It's like a chess game," says Burch. "You don't want to make a move you don't know how to deal with, or you might get checkmated." The formatter change remains a possible option, he says, but if it caused a working instrument to fail, "we'd really look like a bunch of fools."

A second option would involve a fix in space, attaching a working power supply to the spectrograph and leaving the faulty supply intact but disconnected. NASA would piggyback this effort on a Hubble service mission already planned for 1993. The maneuver -- similar to the 1984 repairs on the Solar Maximum Mission - would require an extra day on the three-day service mission, says John Campbell of Goddard, the deputy associate director for Hubble flight projects. After all, he notes, astronauts on the service mission will already have their hands full fixing Hubble's blurry optics, failed gyroscopes and loose solar panels.