WASHINGTON — Three satellites not much bigger than birthday cakes arrived on orbit this year, arranged themselves into a constellation and made research-quality measurements of the Earth’s magnetosphere before being decommissioned just 90 days later.

Such a short run, unusual for a healthy science satellite, has become par for the course on NASA’s New Millennium technology-demonstration program.

But it has not always been that way.

Christopher Stevens, NASA’s New Millennium program manager at the Jet Propulsion Laboratory in Pasadena, Calif., said the program was redefined in 2000 to focus “very specifically on just what was needed to validate the technology for future use. Any science that’s derived is incidental rather than a requirement driver.”

Created in 1995 by NASA’s space and Earth science directorates, which at the time were separate, the New Millennium program was envisioned as the agency’s primary venue for demonstrating breakthrough spacecraft technologies with fairly broad applications. Developing more specialized technologies is still primarily the responsibility of the programs and missions that need them.

“We are the primary way that the Science Mission Directorate (SMD) validates advanced technology that needs to go to space to be flight validated,” Stevens said. “We focus on technologies that are broadly applicable to the SMD portfolio.”

New Millennium got off to an auspicious start with the 1998 launch of Deep Space 1 on what would become a three-year mission that validated a dozen advanced technologies, including the xenon-ion propulsion system that will power the Dawn spacecraft when it begins its asteroid-hunting mission in June 2007. In the process, Deep Space 1 returned some of the best-ever pictures of a comet.

Earth Observer 1, launched in 2000, also exceeded expectations, demonstrating an advanced land-imaging camera and a host of other technologies. The satellite remains in service today, demonstrating breakthroughs in remote-sensing satellite autonomy with the aid of New Millennium-funded software uploaded after the mission achieved its primary objectives.

But the stunning successes of Deep Space 1 and Earth Observer 1 have been tempered by frustrating setbacks that underscored the danger of heaping science requirements onto missions that are high risk by design.

New Millennium officials also have seen at least two missions canceled for exceeding their budgets, lost one at Mars, and generally have had a hard time finding launch opportunities.

Deep Space 2, consisting of two miniature probes carrying 10 experimental technologies, was lost when the host spacecraft, the 1998 Mars Polar Lander, crashed into the red planet. Earth Observer 3, a $200 million experimental weather satellite NASA was building in partnership with the National Oceanic and Atmospheric Administration and the U.S. Navy, never flew. The satellite, also known as GIFTS-IOMI, lost its reservation aboard a U.S. Air Force-sponsored launch when the Navy said a $2 million funding shortfall would delay its completion. NASA looked into other options for finishing and flying Earth Observer 3, but eventually canceled the project amid continued budget uncertainties.

Even the successful Space Technology 5 mission — the three-satellite constellation that launched this past March and studied the magnetosphere — was delayed three years while NASA scrambled to find a suitable launch opportunity. NASA at one point considered designing a custom payload adapter so that the three satellites could launch as secondary payloads on an Atlas 5 or Delta 4 rocket. Ultimately, NASA flew them as the primary payload on Orbital Sciences Corp.’s air-launched Pegasus.

Since revamping the New Millennium program, NASA generally has tried to alternate between fully integrated flight demonstrations like Deep Space 1 or Space Technology 5, and subsystem-level experiments that often must find their way into space by hook or by crook, Stevens said.

Take, for example, the Inertial Stellar Compass. Selected as part of Space Technology 6 in 2001, the $10 million attitude-determination system built by Draper Laboratory in Cambridge, Mass., had been penciled in to fly on everything from a space shuttle to a commercial solar sail mission. The experiment finally found accommodations as part of the Pentagon’s TacSat-2 mission, which at press time was slated to launch Dec. 11 aboard a Minotaur rocket out of NASA’s Wallops Flight Facility on Virginia’s eastern shore.

“We were actually ready for flight back in 2004-2005, but had to wait for the right launch vehicle,” explained Linda Fuhrman, Draper’s program manager for the Inertial Stellar Compass, or ISC. “The ISC was originally designed to fly as a HitchHiker payload in the space shuttle, and has been manifested on one launch vehicle after another over a period of several years, each of which was canceled for one reason or another. TacSat-2 is the first launch opportunity that has actually made it all the way to the launch pad.”

The low-power, low-mass device, consisting of a star-tracker camera and miniaturized gyroscopes, is on board the TacSat-2 microsatellite and will be turned on about 7 days after launch. Fuhrman said a successful demonstration would point the way to an attitude determination and navigation sensor that offers dramatic weight and power savings over conventional systems.

With the New Millennium program holding the line on science objectives, NASA has seen per-project costs decline. Whereas NASA spent $170 million on Deep Space 1 and nearly $200 million on Earth Observer 1, fully integrated demonstration platforms like the upcoming Space Technology 8 — four technologies bundled into a dedicated carrier spacecraft slated to launch in February 2009 on a Pegasus — are being done for $130 million to $150 million including launch, according to Stevens. Subsystem demonstration missions, meanwhile, are generally costing $10 million to $15 million each, he said.

NASA’s backlog of New Millennium flight projects also includes the Disturbance Reduction System, also known as Space Technology 7, which is being developed for inclusion on the European Space Agency’s Laser Interferometry Space Antenna Pathfinder mission, formerly known as Smart-2. Stevens said the Disturbance Reduction System, slated to launch in October 2009, will demonstrate new methods for very precise position control.

Looking ahead, NASA plans to select its next integrated flight demonstration, Space Technology 9, either before the end of the year or in early 2007. Stevens said the five concepts in contention for the $130 million flight opportunity include a solar sail mission; a prototype of an aerocapture system for planetary missions; a package of large space telescope technologies, a precision formation-flying experiment; and a suborbital pinpoint landing and hazard-avoidance system.

Stevens declined to provide much detail on the five concepts since they are still in competition. He did say that all of the proposals with the exception of the landing and hazard-avoidance system — which he referred to a terrain-relative guidance system — would feature a demonstration in Earth orbit.