Space Station shuns cutting-edge technology for the tried and true

The International Space Station has been called mankind`s single greatest engineering and construction effort — the culmination of generations of dreams and aspirations. But, from an electronics standpoint, it is most assuredly not state-of-the-art. In fact, the 50 primary computers aboard the U.S.-built segment that will run the Space Station are based on the Intel 80386 microprocessor — a chip that is at least five generations into obsolescence compared to what is available for the n

Dec 1st, 1999

By J.R. Wilson

The International Space Station has been called mankind`s single greatest engineering and construction effort — the culmination of generations of dreams and aspirations. But, from an electronics standpoint, it is most assuredly not state-of-the-art. In fact, the 50 primary computers aboard the U.S.-built segment that will run the Space Station are based on the Intel 80386 microprocessor — a chip that is at least five generations into obsolescence compared to what is available for the newest commercial desktop computers.

The Russian and Japanese elements have their own computer systems, tied into the U.S. computers via MIL-STD 1553 databus and slightly more modern than the `386, but still far from current standards. The European Space Agency and the U.S provided the Russian computers.

But to those building the station, the question is not how advanced are the systems, but how proven is their reliability in the harsh environment of space.

"When you contrast what you see in a desktop to what we`re launching, it does seem like what we use every day is much higher capability — but that stuff would not survive in space very long," says Ted Goetz, Space Station chief engineer for prime contractor Boeing in Seattle. "It wouldn`t take very long for a Pentium III to start failing in orbit because of the radiation environment. It would take more weight for shielding to harden it — and even then you couldn`t protect it from cosmic radiation."

Larry McWhorter, deputy manager for avionics integration at NASA`s Johnson Space Center in Houston, says the performance of the planned computer hardware is not a problem. "Even with the `386, we`re only running about 60 percent on the CPU, which is one reason we haven`t pushed an upgrade."

The design challenges that engineers anticipate most revolve around maintenance, upgrades, testing, and integration. "We do have a program in the background to just change out the processor card without having to pull the entire package," McWhorter says. "But we`re talking about integration and test as the real pusher in the computer system because of the number of computers and software providers, U.S. and international."

Space Station electronics represent a layered, distributed architecture. On the top layer is the primary command-and-control computer. Below that are computers for power distribution, payload management, systems, guidance, and control. Below those are a series of multiplexer/demultiplexers (MDMs), which have discrete I/O card capability and interface directly with all of the box-level hardware. The MDMs send out commands to turn specific switches on or off.

Each circuit card is removable. In orbit, the crew will be able to pull a box out of the rack and either replace individual cards or, if necessary, install a complete spare MDM.

The astronauts will interface with the onboard computers through Pentium-based IBM Thinkpad 760 laptop computers, which they can upgrade as necessary. Ada is the primary programming language on all 50 computers in the American segment. Some of the laptops will run C++ and some Solaris, which is the same configuration of laptop hardware and software on the Space Shuttle.

Engineers from Honeywell Inc. of Minneapolis are making all of the computers on the U.S. segment, as well as providing some for the Japanese centrifuge module. The Italians also are using the Honeywell units in their mini-pressurized module, which goes into the shuttle cargo bay to carry payloads to and from the station.

Perhaps one of the most surprising elements aboard the station will be its mass-memory system — the equivalent of a home computer`s hard drive, which in most new computers now averages 10 gigabytes or more. The first elements of the Space Station are launching with 300-megabgyte mechanical drives and three circuit cards from the Raymond Engineering operations of Kaman Aerospace Corp. in Middletown Conn. Later, Space Station crews are to replace those with a new 1-gigabyte solid-state system comprising only three circuit cards and no mechanical drive.

"The question is, what do you really need?" McWhorter says. "The 386 processor and 1-gigabyte storage more than meet the identified needs. But we will continue with a process of upgrades, mainly to deal with obsolescence."

The Space Station`s communications system is standard Ku S-band, the same as the shuttle and most satellites.

The guidance and control systems also involve little new technology. One exception is the global positioning system (GPS) receiver in place of a star-tracker for attitude correction.

"Attitude determination from GPS on the Space Station is the only program I know of actively pursuing this technology to determine a vehicle`s inertial attitude in space," McWhorter says. "We`re getting ready, on an upcoming shuttle flight, to do a demonstration to validate the hardware and software in a space environment. That is called SOAR [Standard Integrated GPS System Onboard Attitude Reference]."

Engineers from the Honeywell Space Systems division in Clearwater, Fla., built the Standard Integrated GPS unit, but elements of it came from Trimble Navigation Ltd. in Sunnyvale, Calif., and some of the software was developed by NASA`s Goddard Space Flight Center in Greenbelt, Md.

Other than the GPS, "there is no real leading-edge technology in the avionics from a hardware point of view," McWhorter says. He cites the primary goal of packaging and integrating proven elements from several different vendors and nations so they will work together with the least amount of conflict or oversight requirement once in orbit.

Boeing`s Goetz says he believes "state-of-the-art" with respect to the Space Station should involve only what has been tried to date in space, not in terms of what is available on the desktop commercially.

Some of the Space Station subsystems process more power than any boxes that have gone into space before — and in turn generate a great deal of heat, Goetz says. The station`s thermal coatings and electronics mountings and, in some cases, some special reflective coatings, had to be designed to accommodate the need to dissipate 500 watts of heat from a very small space. This is not an easy task, given segments of the station will be exposed to the sun`s heat for extended periods, then face the intense cold of clear space at other times.

The thermal cooling system is outside the station, because installing it inside would take too much room and add too much to launch costs, experts say. All of the electronics going aboard the station will have been hardened against cosmic and ionizing radiation. Some parts have been designed to withstand intense radiation, while some of the software and embedded processors have been designed with automatic internal error-correction codes.

The fault-protection capability is a combination of hardware and firmware. The brain behind the remote power controller is a hybrid microcircuit. A quick-reacting system, it responds to an overload condition in a matter of five or six microseconds. The closest switch protection device is the one that trips first, enabling the crew to quickly isolate the problem area.

Harris Corp. of Melbourne, Fla., provides the high-speed, high-power switching sets for the system and most of the boxes have embedded firmware developed and qualified at Boeing`s Conoga Park, Calif., facility.

"The biggest problem we had was coming up with radiation-hardened chips. A lot of the computers are located outside the vehicle, so we`ve had to work those to temperature extremes and other environmental areas," McWhorter says. "When we get into the permanent systems, they will be radiation-hardened and built to special requirements to deal with the thermal and radiation problems. And when we upgrade, our biggest concern will be the software, not the hardware. We need to keep the cost of re-verifying the software down when we go to a new processor."

Onboard batteries proved to be another problem. While nickel-hydrogen batteries are frequently used in orbit, previous systems did not need them in the size or with the life span required by the Space Station. Each of the Space Station`s 24 batteries must produce 81 amp hours — equivalent to about 13 standard automobile batteries in a rack.

"But if you treated that rack the way we`ll be treating these batteries, they would die in about three months compared to our requirement of 6.5 years," says Goetz. "The power system is a large capacity, relatively high voltage [160-volt] system, which presents its own challenges in being able to survive the plasma environment."

Dan Olberding, Boeing`s electrical power system manager for the Space Station, says power components are mature developments, not leading-edge technology. "The challenge has been to qualify a big system of this type in space. We have more than 40 different power configurations to deal with in orbit because of the different providers and our own developments and loading requirements and configurations. It`s a lot more difficult than doing a system for a single satellite."

Even with a higher voltage direct-current distribution system than the Space Shuttle`s 28-volt configuration, meeting the system challenges aboard the Space Station led to some dramatic improvements in size and weight.

"With the remote power controllers, we`ve been able to pack the same amount of power-control capability we have on the shuttle — which took a 65 pound box — into a 25 pound box," notes Larry Moon, manager of NASA`s subsystems integration office at Johnson Space Center. The remote power controllers involve a remotely operated switch that can receive instructions from the ground or from an onboard laptop tied into the station`s basic computer architecture.

The shuttle box is roughly 12-by-9-by-20 inches and the Space Station box is 5-by-9-by-12 inches — a considerable size and volume shrinkage. "It`s handling an equivalent load in terms of the equipment it is shifting power to, but using standard microcircuit technology that is much more advanced than what went into the shuttle 20 years ago," Moon says.

Another primary concern for the Space Station will be water processing, which will mean recycling everything to reduce dependence on shuttle re-supplies.

"In developing that capability, we had to come up with specialized electronic sensors," Moon says. "One of those, for example, senses gases in a liquid stream, which could pose a problem in the reprocessor."

Placing so many critical elements outside the Space Station`s pressurized and radiation-buffered work areas did create one other potential problem: forcing the crew to work outside make repairs or upgrades. To keep astronaut exposure to that harsh environment at a minimum, virtually all orbital replacement units (ORUs) and critical systems can be maintained using a 65-foot robot arm from Spar Aerospace Ltd. of Mississauga, Ontario.

"Everything can be replaced using that arm, so an astronaut in shirtsleeves inside can do repairs," Goetz says. The arm is attached to a cart that runs on rails up and down the main truss of the Space Station. The arm will help maintain all systems within its reach. Astronauts working outside the station will maintain everything else.

Boeing engineers draw on their commercial aircraft manufacturing experience in seeking to reduce the amount of work the astronauts will need to do in space as they assemble parts from all over the world. Central to this is a multi-element integration test (MEIT) performed at Cape Canaveral, Fla., before anything launches into space. As each element arrives, it runs through a range of operations — including how well it works with other systems.

"We also have digital pre-assembly, which was a technique developed for the Boeing 777," Goetz says. "It allows you to take very, very accurate measurements of as-built hardware, then the computer analytically makes sure they will fit on orbit."

"It requires a pretty sophisticated test and verification approach to make sure everything will work once we get it into orbit," says Goetz. "So far, everything we`ve put into orbit is working well — much better than some people thought they would."

More in Power