A device installed to help prevent another Three Mile Island caused a short circuit that shut down the Crystal River nuclear power plant in Florida last week, nuclear regulators learned yesterday.

The device, called a saturation, or subcooling, meter, was installed on all 70 of the nation's operating reactors by order of the Nuclear Regulatory Commission in the wake of Three Mile Island. It was designed to give operators more direct information on the condition of a reactor's radioactive core.

The one at Crystal River was installed within the last two months. The Feb. 26 shutdown occurred when two tiny pins short-circuited on their connection to one of the meter's four circuit boards, according to Ed Blackwood, an NRC nuclear systems engineer in the operations inspection office.

Blackwood said it was his "tentative conclusion" that the problem was the result of an installation error and not a design difficulty, since the circut board appeared to have been slightly out of alignment.

"It's a tight squeeze. There was lots of stuff in this cabinet," Blackwood said.

However, other NRC sources said the possiblity of a design problem, which would be much more serious, has not been ruled out.

The NRC formally warned all reactors yesterday to check these circuit boards whever they occur in the plants. The boards are made by Bailey Meter Co. of Wickliffe, Ohio, a subsidiary of Babcock & Wilcox Co., which built both the Crystal River and Three Mile Island reactors.

The NRC's Atlanta office, which handled the Crystal River incident, completed its formal report yesterday to the NRC's advisory committee on reactor safeguards, which supplies the NRC with technical advice.

The committee had complained formally in January that the time allowed by the NRC staff for checking items that would be required as Three Mile Island fixes was "too short to give reasonable assurance that all changes will be in the direction of greater safety."

Dr. Harold R. Lewis, physics professor at the University of California at Santa Barbara, wrote in additional remarks on the complaint: "The plan consists of an uncritical listing of anything anyone has suggested be done in the aftermath of (not necessarily as a result of) the accident at Three Mile Island."

Lewis asked NRC staff members yesterday if it now appeared that their recommendations had helped or hindered things at Crystal River.

"I think we helped, and the Crystal River operators think we helped," Denwood Ross, chief of the task force on post-Three Mile Island orders, said later. "The training made them more aware of what could go wrong and they responded a lot quicker than they otherwise would have."

Roger Mattson, head of the "lessons learned" group that made recommendations on installation of devices -- including the faulty subcooling meter -- agreed with Ross.

"It's ridiculous to imply that a short-term lesson learned from TMI caused what happened at Crystal River," he said. "We told them to put an instrument on, and they are required to put on a reliable instrument in a reliable fashion. Any instrument that causes such a power failure is not properly designed or installed."

NRC Chairman John Ahearne also defended installation of the meter. "I'm confident it was a good move to make," he said. "But it would certainly concern me if there were not adequate quality control over the manufacture or the installation of any of these devices."

The Crystal River plant remains shut down while necessary repairs and some planned maintenance are performed. Florida Power Corp., which owns the plant, said the sequence of events that led to the shutdown was being studied.