Cautiously optimistic NASA engineers said yesterday they will attempt a launch Tuesday of the space shuttle Discovery after days of exhaustive troubleshooting turned up possible causes of the sensor malfunction that scrubbed the launch last week.

Bill Parsons, the shuttle project manager, said the best way to test their analysis will be to load the external fuel tank with liquid oxygen and hydrogen during countdown in order to re-create as closely as possible the conditions that existed when the sensor malfunctioned less than three hours before liftoff.

Parsons, speaking from the Kennedy Space Center during a televised news conference, said the team was prepared to launch with one bad sensor -- the shuttle has four -- but only if engineers understand why it is malfunctioning. "If it doesn't go the way we think it will, we'll scrub and think about what to do next," he said.

Michael Wetmore, the shuttle processing director, said the three-day countdown will start at midday Saturday and, barring further mechanical troubles or bad weather, will end Tuesday at 10:39 a.m. Eastern time, when Discovery will lift off Launch Pad 39B at the space center for a 13-day mission to the international space station.

If the launch is scrubbed for a minor reason, Wetmore said the team could try again Wednesday or July 29 or 31. Parsons said planners may try to steal a couple of extra days at the beginning of August, but lighting conditions will make it impossible to launch after Aug. 4.

Discovery's flight would be the first shuttle launch in the 21/2 years since Columbia disintegrated on reentry. NASA views the mission as a test of its procedural changes and of the extensive safety modifications made since the disaster.

Discovery's sensor problems began in April during a "tanking test" in which the orbiter's external fuel tank was loaded with 1.6 million pounds of liquid oxygen and liquid hydrogen cooled to several hundred degrees below zero.

During the test, two sensors designed to trigger a main engine shutdown in case of a sudden loss of fuel began to behave erratically, toggling back and forth between "wet" and "dry" settings for reasons that remain unclear.

Engineers replaced the system -- sensors, wiring and the "black box" controller -- and went ahead with the launch last Wednesday, only to scrub it after one of the sensors locked in the "wet" position during the final stages of the countdown.

Since then, 12 teams of troubleshooters from around the country have analyzed the system. They suspect that the sensors may be affected by faulty grounding or electromagnetic interference from devices or equipment installed during the post-Columbia redesign.

"Analysis of the circuitry suggested grounding would be a problem," said John Muratore, the shuttle systems engineering and integration manager in charge of the trouble-shooting effort. "We're going to replicate the electromagnetic environment at launch, and this is best accomplished during countdown."

Muratore said engineers will exchange the black box connectors of two of the sensors, which should enable the team to pinpoint the cause of a malfunction, if the team's conclusions are accurate.

"If Sensor Four fails, we'll say that we understand the problem," Parsons said. "But if it was a Sensor Two problem, we would say we don't understand it and we'll give ourselves some time to think about it."

Launch rules require all four of the hydrogen sensors to function properly before liftoff can take place, but Parsons said the team was prepared to fly with three good sensors by "taking an exception" to the rule, but only "if we get a [fault] signature we understand."

A technician tests wiring in the shuttle Discovery yesterday at the Kennedy Space Center.