Scientific satellites currently performing important missions in orbit could be called collections of electronic parts because the role of LSI (Large-Scale Integration) circuitry has become very large in such satellites. Japan currently imports most of its high-performance, space-application LSIs from overseas, however. The number of LSI used by Japanese scientific satellites, usually launched at a rate of one a year, is six orders of magnitude less than that used in personal computers. Therefore, it is very difficult for Japan alone to build and maintain the framework to develop and manufacture space-application LSIs. In the meantime, there is concern about high-performance commercial LSIs because of their resilience to the severe radiation environment in space. To mount commercial LSIs "as is" onto satellites is like proceeding with space science by leaving it to fate. The use of commercial LSIs in space is not appropriate because high-level space missions require high-reliability. Is there a third strategy?
One answer may lie in common-use space and commercial applications. The biggest issue in developing space-application LSIs is radiation hardness. The same problem has arisen recently even in commercial LSI development because the micro-fabrication of commercial LSIs has advanced to an extreme level. Therefore, common use between both applications is now feasible.
In 1996, in addition to the radiation problem caused by impure materials within the packages, a new neutron radiation problem attributable to cosmic rays hitting the ground became evident for commercial LSIs, which were becoming increasingly micro-fabricated. IBM reported its concern that neutron radiation could cause a Single Event Upset (SEU), a major failure caused by radiation. A Single Event Upset can rewrite data stored in LSIs, leading to malfunctions. The automobiles, aircraft and construction-equipment are exposed to severe mechanical and temperature environments and their safety also has lives at stake. So, high-reliability and especially high radiation hardness is demanded in these industries in particular. Thus, we came up with a strategy of common use, where these industries and the space community can cooperate to develop LSIs and maintain the manufacturing line.
Target specification of space-application LSIs
The partner we selected for common-use LSI development was Nagoya Guidance and Propulsion Systems Works, Mitsubishi Heavy Industries, Ltd. (MHI). The company is engaged in a variety of electronics development including space, automobile, and construction-machinery applications. Joint development started in 1999. Our first task was to clarify target specification required for the space-science community. We selected the most appropriate design, circuit and manufacturing-process technologies in order to fulfill the target specification of our LSI.
The most critical point in developing the space-application LSIs is to consider trade-off of processing speed, power consumption, and radiation hardness to set up target specification. The enhancement of radiation hardness would result in a drop in processing speed and increase of power consumption as well as increase of chip size in proportion to hardness level. These were just some of the difficulties we faced that are not seen in the development of commercial semiconductor integrated circuits. Fig. 1 shows the history of processing speed of one LSI, the highest-performance microprocessor (MPU). It shows that the processing speed of space-application MPUs is low compared to commercial MPUs at the same age due to the trade-off mentioned above.
To implement the trade-off, we first needed to determine mission requirements. The requirements for LSIs onboard scientific satellites/planet explorers include the followings: autonomous satellite operation function including three-axis attitude control; processing speed capable of processing huge observation data and forwarding them to ground station within the period flying above the station; power consumption equivalent to the low power generated by a small spacecraft; and radiation hardness higher than that required for the Space Station circling at low earth orbit. We carefully traded off the requirements and aimed at more than simply maximizing processing speed and radiation hardness. As a result, the development target of the MPU was defined as follows: processing speed 100MIPS; power consumption 1W or below; and world standard radiation hardness.