Designing from first principles makes space more affordable

Paul Boughton

Designing for space is very different from designing for terrestrial applications. Maarten Meerman – also known as Max (Fig.1) – perhaps knows as much about this as anyone else in Europe, being the director of research for Surrey Satellite Technology Limited (SSTL).

This UK-based company received global publicity on 28 December 2005 for the successful launch of its latest satellite, Giove-A (Fig.2), a preliminary component of the Galileo project, Europe’s own navigation system designed to provide a higher service level than is available from the USA's GPS (global positioning system). 

Giove-A was launched at 05.19 GMT, carried aboard a Russian Soyuz-Fregat launcher. Within ten minutes, the Fregat upper stage had separated, ignited and was pushing Giove-A into orbit approximately 23 000 km above the earth. And that 23000km orbit is exactly where the challenges had started for the satellite’s designers.

An orbit 23000km above the earth’s surface is termed a mid-earth orbit, as opposed to the low-earth orbits used by remote sensing, communications and similar satellites. Maarten Meerman explains: “As the distance from earth increases, the radiation spectrum varies considerably, and there is limited data available for mid-earth orbits (the Americans are reluctant to share the data they have from their GPS satellites). Furthermore, we believe that a lot of the computer model data that is available relating to radiation is inaccurate, as it was gathered by Skylab at a time when radiation levels were unusually high. Over the past 20 years, SSTL has gathered data from its own satellites and demonstrated the computer models to be over-pessimistic. SSTL is therefore now working with both NASA (National Aeronautics and Space Administration) and ESA (European Space Agency) to improve the models.”

Indeed, one of the tasks for Giove-A is to measure the radiation so that the designs for the constellation of 30 Galileo satellites can be confirmed. Plastics, glass and, most importantly, electronics can all suffer if struck by heavy radiation particles, and metallic enclosures do not provide complete protection. For example, if an alloy enclosure for electronic componentry is too thin, radiation particles will pass through. And if the wall of the enclosure is too thick, although the radiation particle will be halted, its energy can be such that it can generate new radiation particles from the inside of the enclosure wall. Hence wall thickness has to be optimised, and the Giove-A has been designed with enclosures having a wall thickness about double that typically used on low-earth-orbit satellites in order to cope with the higher radiation levels at 23000km.

Coping with radiation

But whatever measures are taken with the enclosure design, it has to be accepted that some particles will inevitably interfere with the electronics. Maarten Meerman discusses the example of a particle altering the crystal structure of a chip’s semiconductor material and flipping a memory bit. “We employ a triple redundant architecture on the basis that a particle might pass through one chip and flip a memory bit, but it is extremely unlikely to pass through two and corrupt both of them in exactly the same way. So when the processor reads the memory, it normally only reads the data from two chips. If the result is the same, it is accepted but, if the results are different, the third memory chip is read and the answer from that is accepted (which is effectively a two-out-of-three voting system).”

He continues: “To prevent the level of corruption from building up over an extended period, at times when the processor would otherwise be idle, it reads through all three memories and compares bits, correcting any discrepancies that are found.”

Interestingly, this triple-redundant architecture and memory washing procedure is now starting to be adopted on aircraft. Although radiation was never a significant problem at 30000feet, modern aircraft can cruise at 39000feet (and 41000feet may soon be the norm), at which point radiation issues certainly do need to be addressed.

Aside from radiation, another important factor when designing for space applications is the vacuum. “Oils, greases, paints and plastics all suffer from evaporation in space,” says Maarten Meerman. “Plastic components can become brittle and fall apart; if PVC is used, the chlorine off-gases and attacks other components. Even ball bearings have to be specially designed for space applications, as they have to operate without lubrication – plus they need to be preloaded to avoid the rolling elements rattling and the bearings subsequently failing due to premature wear (with no gravity and no air damping, there is nothing to otherwise restrict the motion of the rolling elements).”

Indeed, the lack of gravitational effects causes major problems in space. Almost no matter how much testing is performed on the ground, a small wire or other non-rigid component can easily move to an unanticipated position and create a problem. The Giove-A satellite’s solar array panels had to be folded for the launch and, when the satellite was in position, deployed to measure 7m across the tips (Fig.3). At each moving joint, the cables have to be carefully tied into position, but with stress-relieving loops. Another point to be aware of with mechanisms in space is that there is no air damping, so they behave differently and components can be subjected to shock loads that they would not experience in a similar terrestrial application.

Thermal management

“Thermal management is also important for satellites,” adds Maarten Meerman, “as the absence of gravity means that convection currents do not form, so heat flows have to rely on thermal paths – which is a very different situation from that on earth. Depending on the orbit, satellites might spend most of the year in sunlight and only short periods in the earth's shadow, or, at the other extreme, they might alternate between sunlight and shadow on a 90-minute cycle. Thermal cycling can be detrimental to electronic components, so critical components are usually located near the centre of the satellite, where the temperature variation can be restricted to a range of around 5°C. Towards the outside, components may experience a temperature range of 10°C. Reflective blankets are used on the satellite’s exterior to minimise heat gain in sunlight and insulate against heat loss when in shadow. But solar panels can cycle between -10 and 40°C over a period of 90 minutes, depending on the orbit, and they might experience extremes of-50 and 80°C during deployment.”

As with special-purpose machinery, space design is usually a matter of designing ‘working prototypes’. However, unlike terrestrial machinery, satellites cannot be physically modified once they have entered service. Although the Space Shuttle has been used to perform repairs, it can only attend satellites that it launched itself – and the cost to do so remains substantial. So satellite designs have to be analysed and tested as much as reasonably practicable, through computer modelling and physical testing.

Maarten Meerman describes the step-by-step evolutionary approach adopted successfully by SSTL: “We typically install a tried-and-tested technology – such as a processor – in the satellite that we can rely on to operate as intended. In addition, a newer design will be installed in parallel, performing the same tasks as the primary technology, and possibly with some extra, non-mission-critical functionality. When this newer technology has been proven in two or three satellites, it will become the primary technology in the next satellite, and another newer technology will be introduced alongside it for evaluation.”

It was this process that led to Giove-A making use of a CAN (controller area network) fieldbus. “CAN was originally developed for automotive applications, which means it is extremely robust, well proven and of high enough integrity to be suitable for safety-critical tasks such as anti-lock braking systems (ABS). SSTL has used CAN, Ethernet and other fieldbus networks on previous satellites and, in the light of the results, selected CAN for Giove-A. One of the advantages of this approach is that any of the CAN nodes on the satellite can be addressed directly from the ground. So any of the processors can be used to control the satellite’s functions, resulting in high availability, even if one or more processors fail.”

Fit for purpose

The use of ‘conventional’ technology for space applications is one of the features that sets SSTL apart from its competitors. Maarten Meerman explains his company’s philosophy: “Whether you want an integrated circuit, resistor, nut, bolt or whatever, you usually have a choice of commercial, industrial, military or space specifications. Space specification is fine for ‘traditional’ space applications such as for a mission to Mars, where the exceptional reliability is essential, the long lead time is not an issue, and the components represent only around two per cent of the overall project cost. But for the satellites we design and build here, which might be for, say, imaging or DMC (disaster monitoring constellation,) there are bigger benefits to be gained from using newer designs of component with considerably improved performance and sufficiently high reliability. For example, consider an iPod: this is incredibly powerful and is robust enough to survive being dropped. A huge investment in its development has been possible because of the high production volumes; such a high degree of development is simply not feasible for space applications where only a very small number of parts is required. Although we do not use iPod components in our satellites, we do use – often in a modified form – conventional parts if they offer the right balance between cost, performance, reliability and availability.”

Most of SSTL’s satellites are built as one-offs, so mechanical components are normally machined from solid using CNC (computer numerically controlled) machine tools. Because of the complexity of these parts, the CNC programming tends to be a lengthy process, leading to a part cost that is substantial compared with the cost of a similar mass-produced component. Nonetheless, on the rare occasions that SSTL builds a series of three to five satellites, the part cost reduces dramatically. For the intricate, tightly toleranced machining that is required, SSTL is fortunate in that its location in the south of England gives convenient access to numerous specialised machine shops that work largely for the Formula One race car industry.

When SSTL was first formed as a spin-off company to commercialise technologies developed at the University of Surrey, most components and systems were designed and manufactured in-house out of necessity. Maarten Meerman describes how SSTL’s design policy evolved: “In the early days of space, little was known about radiation levels in low-earth orbits, so most of our competitors were constructing satellites to the ‘gold’ standard, and that simply continued. However, we realised that the use of ‘space specification’ components was largely unnecessary for the types of projects in which we were involved. This was partly due to the requirements for only having to withstand the lower radiation levels in low-earth orbits, but also because a momentary glitch in something like an imaging satellite is far less serious than a break in the signal for a televised global sporting event.”

Niche market

By taking a fresh look at satellite design, and by virtue of the fact that SSTL was a small, nimble company in comparison with its competitors, SSTL created its own niche market for small, yet technologically advanced, modular microsatellites weighing 50 to 130kg for low-earth orbits and with a life expectancy to suit the relatively short-term applications.

SSTL restricts its use of ‘space specification’ components to those parts where it is absolutely necessary – such as solar cells – while using off-the-shelf industrial-grade or military-grade components elsewhere. The company still builds almost all of its own equipment, including transmitters, receivers, computers and cameras. In some cases, such as Giove-A, the payload (clocks, transmitters and the main antenna) is supplied by the customer.

Because of the emphasis placed on designing components and systems from first principles, few of the staff at SSTL are aerospace engineers; most are physicists, mechanical engineers, electrical engineers and software engineers.

While SSTL certainly has exceptional technological capability, the company wins its business through being cheaper than its competitors. For example, the company’s involvement with the Galileo project has already saved the ESA around E100million compared with what the organisation would have had to pay a ‘conventional’ satellite builder (Fig. 5). Moreover, Giove-A was designed and built in 30months for a cost of E28million and launched in time to secure the frequencies allocated by the ITU (International Telecommunications Union). In contrast, Giove-B is being built by Galileo Industries, a consortium that includes Thales, EADS Astrium and Alcatel, and is not expected to be complete until mid-2006, plus it will cost considerably more than Giove-A.

Around 98percent of SSTL’s turnover results from exports outside the UK. Customers include the United States Air Force, Centre National des Techniques Spatiales of Algeria, Beijing Landview Mapping Information Technology, Information Technologies and Electronics Research Institute of Turkey, Federal Ministry of Science and Technology of Nigeria, as well as the British National Space Centre.

But, as Maarten Meerman says: “Every project is different, depending on whether the need is to travel through deep space or orbit earth or another body. And, if an earth orbit is required, its height makes a difference. Even then, things can be complicated by local conditions. For example, there is an area over the South Atlantic Ocean with higher radiation so, when satellites pass through that area, special care has to be taken to prevent computer problems. Space Shuttles and the International Space Station do not schedule spacewalks when they are due to pass through this region.”

Looking ahead five years, he says: “I think the future looks very exciting. Although we have developed nanosatellites weighing around 6.5kg (Fig.6) and have carried out research into credit-card-sized swarming satellites, I anticipate that most of our business will remain with microsatellites and small geostationary communications satellites around the size of the 660kg Giove-A. There seems to be little commercial pressure for satellites to get smaller, but the processing power and technical capability improves year-on-year.” 

For more information, visit www.sstl.co.uk

"

Recent Issues