Troubleshooting with the space shuttle Atlantis fully fueled has narrowed the source of the problem that kept it stuck on the pad this month to a connector that passes through the walls of the big external fuel tank, raising hopes that International Space Station (ISS) assembly can resume as early as January.

Wayne Hale, NASA's shuttle program manager, said Dec. 18 a tanking test earlier in the day reproduced the problem with an open circuit in the engine cutoff (ECO) sensor system, and sophisticated Time Domain Reflectometry (TDR) gear traced the fault to the three-part pass-through connector at the tank wall.

Hale said the TDR gear - which sends an electrical pulse down a wire and measures the time before impedance in the circuit is reflected back - lacked the resolution necessary to determine if the fault is at the outside of the tank, inside the tank where wires leading to the five ECO sensors emerge from the wall, or in the metal and glass assembly that passes signals from ambient to cryogenic temperatures.

One theory about the intermittent problem suggests the circuits may somehow be pulled open by the partial vacuum that results in that central portion of the pass-through when the liquid hydrogen reaches it, Hale said. Bench tests are underway at Kennedy Space Center and Marshall Space Flight Center to determine the exact source of the problem, and the pass-through connector on the Atlantis tank will be examined in detail.

Engineers and managers are scheduled to meet formally Dec. 19 to begin mapping a plan to recover the tank and make another launch attempt. Atlantis is tentatively scheduled to launch Jan. 10, and some managers expressed the hope that a January launch date could be achieved.

But Hale - who earlier faced criticism from the astronaut office at Johnson Space Center when it appeared he was willing to fly without fully understanding the cause of problem if a reasonable flight rationale could be found - said the decision would not be made until the problem is better understood.

"I would like to fly as soon as practical, but more than that, we need to fix this problem, so we're not being driven by any particular launch date," he said.

Technicians equipped with TDR gear went to launch pad 39A at KSC after ECO sensor No. 1 "failed wet" during the tanking test, indicating the open circuit. Two other sensors experienced intermittent failures during the test, and all returned to normal after the tank was drained, Hale said.