First Author: Sias Mostert
Sias Mostert is the SUNSAT Microsatellite Development Manager at Stellenbosch University, South Africa. His primary research fields are Real-Time System Engineering and Computer Networks. He graduated with a Bachelor of Engineering degree in 1987 and received a Master of Engineering degree in 1990, both from University of Stellenbosch. In 1992 he completed the International Space University (ISU) summer program in Japan. He lectured in computer systems for five years before joining the SUNSAT team as Development Manager in 1993, and is one of the digital systems engineers on SUNSAT and responsible for the day to day development of the satellite.
Second Author : Arnoldus Schoonwinkel
Arnold Schoonwinkel is Professor at the Department of Electrical Engineering at Stellenbosch University, South Africa. His primary research field is Automatic Control Systems, with Computer Systems as a supplementary field of interest. He received the Honours and Masters Degrees in Electronic Engineering from Stellenbosch University in 1978 and 1981 respectively. He also completed a PhD in Aeronautical and Astronautical Engineering at Stanford University in 1987. In 1993 he received an MBA degree from the Graduate School of Business at Cape Town University. He practised as a control systems design engineer and R&D division manager in private industry for a total of 10 years. He is the originator of the SUNSAT microsatellite development programme at Stellenbosch University. Apart from his academic tasks, he currently manages the marketing, contracting and financial aspects of the programme.
Third Author : Garth William Milne
Garth Milne is Professor in Earth Satellite Engineering at Stellenbosch University, South Africa. His research fields are Systems Identification, Communications and Analogue Electronics. He received a B. Sc. degree in Electrical Engineering from the University of Natal in 1969, and a Masters engineering degree from Stellenbosch University in 1973. A PhD degree in Electrical Engineering at Stanford University was awarded to him in 1987. He has had a career of more than 17 years in aerospace industry, where he practised as an RF and analogue design engineer, systems engineer and R&D manager. He is the author a State Space Systems Identification Tool Box, which enjoys international sales. Apart from his current academic tasks, he is programme manager, and does optics, communications and overall systems engineering for the SUNSAT microsatellite development programme.
Abstract
This paper describes the total system configuration, flight model components and pre-flight tests of the SUNSAT microsatellite which is to be launched in 1997. A very modular satellite mechanical structure has been constructed and environmentally tested. The performance measurements of the communication payloads (VHF, UHF, S and L band communications systems) are reported. The operation of the attitude determination and control system and other bus systems is briefly described, indicating that a very complex microsatellite was developed as the first venture into space by a South African university. A major spin-off from the programme is an established electronics systems laboratory where students can gain multidisciplinary engineering experience on a challenging project.
This paper describes the total system configuration, flight model components and pre-flight tests of the SUNSAT microsatellite which is to be launched in 1997. A very modular satellite mechanical structure has been constructed and environmentally tested. The performance measurements of the communication payloads (VHF, UHF, S and L band communications systems) are reported. The operation of the attitude determination and control system and other bus systems is briefly described, indicating that a very complex microsatellite was developed as the first venture into space by a South African university. A major spin-off from the programme is an established electronics systems laboratory where students can gain multidisciplinary engineering experience on a challenging project.
SUNSAT, an advanced microsatellite developed by graduate students at Stellenbosch University in South Africa, will be launched by NASA from Vandenberg Airforce Base on a Delta II rocket currently manifested for August 1997. The Flight Model hardware is nearly complete and is now being pre-flight tested. This paper commences with an overview of the overall system, which comprises of three main payloads, namely a push broom type imager, packet communications systems and scientific experiments for NASA and the CSIR.
Laboratory measured performance data are provided for the 15 m resolution 3 colour CCD imager of SUNSAT, developed jointly by Stellenbosch University, the Council for Scientific Industrial Research (CSIR) in South Africa and the Korean Advanced Institute of Science and Technology (KAIST). The realisation and laboratory tests of VHF and UHF data collection and dissemination communications modules are described, as well as the operation of an S-band down link transmitter for image data and an L-band up link for rapid file transfers.
The various satellite bus systems is also described. The paper is concluded with a summary of the benefits and future of the microsatellite development laboratory at Stellenbosch University.
The SUNSAT programme set out in 1992 with three major goals, namely
Now, 5 years after commencing the programme, more than 50 graduate engineering students at Master's and PhD. level have contributed to the endeavour through research and development work. Several apprentice technicians have also completed their practical training in the SUNSAT laboratory.
International collaboration has provided a launch opportunity for SUNSAT in 1997. NASA has agreed to launch SUNSAT free of charge as an auxiliary payload on their P-91 Delta mission2. In exchange, the SUNSAT microsatellite has been modified to incorporate a NASA supplied GPS receiver, and a set of laser retro-reflectors. Under this agreement the SUNSAT microsatellite will supply NASA scientists with data from their instruments, which will enable them to study gravitational effects by means of orbit perturbations.
The appeal of space applications to school children is being used to involve them in the SUNSAT programme3. Inside the structure of the microsatellite, room has been reserved for several projects designed and built in school science laboratories. Projects received so far from high school pupil teams include sensors that measure temperatures in space, structural acoustic noises and the effects of radiation on electronic devices.
The SUNSAT architects deliberately decided that students should build as much of South Africa's first satellite as possible. It was also decided in 1992 that SUNSAT would not be a copy of any existing microsatellite. Attempts have been made to achieve improved technical performance in selected microsatellite subsystems, as described in the sections to follow.
Mission specification
SUNSAT 1 is to serve as a technology demonstrator. The programme's originators decided to take a high risk approach by developing a fairly complex microsatellite in their first attempt. The aim is to test as many subsystems as possible in a 60 kg, 45 x 45 x 60 cm satellite, which has an average power availability of approximately 30 W from body mounted solar panels.
Figure 1 shows an engineering drawing of SUNSAT 1 in the operational configuration with its gravity gradient boom deployed.
The SUNSAT microsatellite will carry remote sensing, communications and scientific payloads that will be described later in this paper.
The orbit was chosen to suit the Danish Ørsted satellite, whose primary mission is to map the earth's magnetic field. Due to an inclination of 97° the orbit drifts at approximately one hour in 70 days relative to the earth-sun vector. The altitude varies between 450 km and 860 km. Even though this orbit is not ideal for SUNSAT's remote sensing payload, is was accepted because it triggered close interaction with NASA researchers and permits SUNSAT's design in a sunsynchronous orbit to be evaluated.
The Mission to Planet Earth scientific payloads are the NASA provided GPS receiver and laser retroreflector. A scientific magnetometer is provided by the Hermanus Magnetic Observatory (South Africa). This sensor is co-located with a star camera on the tip of SUNSAT's gravity gradient boom to provide the magnetic field vector. This data will be analysed in conjunction with magnetic data from the Danish Ørsted satellite which is to be launched in a twin orbit with SUNSAT.
The design life time for SUNSAT is 4 to 5 years, with the number of recharge cycles of the NiCd batteries expected to be the limiting factor.
Overall system design
SUNSAT follows many of the concepts of the University of Surrey's UoSAT series4, but with a few distinct differences. The major research contribution is upgrading the imaging performance and incorporating a high accuracy attitude control system.
The camera system on SUNSAT is designed to provide stereo or side view images in three colours. Each colour has 3456 pixels per line with 15 m ground resolution from 800 km orbit height. This resolution is of the same order of magnitude as the SPOT 2 and Landsat satellites. The quality and variety of SUNSAT images will be unique for a university microsatellite.
The communications payloads comprise high speed data links and as well as Amateur Radio transmitters and receivers in the 145 and 435 MHz amateur radio bands. A separate S-band downlink will convey image data at up to 60 Mbit/s to the 4.5 m diameter dish antenna at Stellenbosch and possibly at other locations. An L-band receiver on SUNSAT permits uplinking of data at 2 Mbit/s, and can be coupled to the S Band downlink for data gateway experiments. This means that large data files can be exchanged among remote locations on earth which are not well served by other Amateur communications systems.
The satellite orientation and control system uses reaction wheels and magnetorquers to point the satellite camera's boresight to within 1 km accuracy on the ground. This accurate position control system on SUNSAT for fine pointing of payloads is a novel addition to this class of gravity stabilised microsatellites.
The usual bus systems exist such as the power system, on-board flight computers, telemetry and telecommand systems. More details on these Flight Model subsystems are given in the sections that follow.
Each tray, as well as the top and base plates, have holes on their corners through which a tie rod passes. Nuts on the threaded ends of the tie rods are torqued to press the satellite body together.
Each of the 11 trays houses particular electronic and mechanical components. A typical tray forms one subassembly and consists of 4 sides, 2 support beams and a printed circuit (PC) board. Although a complete tray could be machined from a solid block to avoid problems in joining processes and to follow the minimum component count philosophy, the decision was made in SUNSAT to machine the four sides separately. This reduced manufacturing costs and improved functional flexibility. This flexibility proved to be a significant advantage when the additional GPS (Global Positioning System) electronics had to be accommodated at a late stage in design.
The structural configuration of SUNSAT also provides easy adaptability for future missions.
One of the main research goals of SUNSAT is to maximise the imaging capability of low cost microsatellites. This has to be achieved while constrained by cost and by launcher interfaces, and by a desire to retain a modular tray construction in the satellite. The goals and consequent compromises led to a three-colour pushbroom imager concept using three Texas Instruments TC104 linear CCD sensors. These sensors have 3456 linearly arranged pixels with 10.7 micron spacing, and maintain a good MTF (modulation transfer function or spatial frequency response) over the visible and near infra-red band.
The imager is located in the bottom tray of the satellite, and comprises a single 12 cm diameter optical tube assembly containing a 45 degree mirror, lens system, pentaprism with dichroic colour splitter, three vertically mounted linear CCD's, and their clock drivers and output buffers. The optical tube is mounted diagonally across the bottom of the satellite on bearings, and can be rotated by a stepper motor.
Stereo images are taken with the optical tube horizontal and normal to the velocity vector. The CCD pixels then form a line on the ground that is normal to the velocity vector, and is able to be pitched forward or backwards by up to 22 degrees to obtain various stereo base/height ratios. By orbiting with the tube parallel to the velocity vector, images to the left or right of the ground track can also be taken.
On-board digital image handling
The 64 Mbyte RAM enables a three-colour square image to be stored for later downlinking. A FPGA based data compressor can be activated for real-time image compression. Only 3456 of the 3490 pixels are active in each image line. Eight of the unused bytes of digitised information can be filled with data from the flight computers. The bytes will be filled with frame synch words, time, orbital, and attitude information to ease configuration management of archived image information.
Further details of experimental work are in a paper specifically on the imager performance8.
Overall, the complexity of developing space quality optics was far more difficult than anticipated. The major hurdles that had to be overcome were aligning the 3 CCD's on the exact same optical path and compensating for the differential thermal expansions between the CCD elements and the optical components.
Figure
2 : UHF Communications System Block DiagramSUNSAT's communications provide for amateur radio communications (VHF, UHF, L, S bands), data downlinking (VHF, UHF, S-Band), data collection (VHF, UHF), and command and control. Different frequencies are at times required for the different services, but allocations of nearby frequencies will permit use of common satellite equipment. Final frequency allocations are still in process in South Africa, so are not available at present. The design can handle uplinks from 144-148, 400-403, 435-438, 1265 MHz, and downlinks at 136-138, 143-146, 400-403, 427-430, 435-438, 2250 MHz. Synthesisers, and easily changed crystals make a late change to specific frequencies feasible. Previous papers1 described the basic communications system on SUNSAT. This section will add further information, particularly on the antenna system.
Figure 1 shows the underside of the satellite which contains an S-band helix, L-band monopole, VHF monopole and canted UHF turnstile. The deployed VHF turnstile can be seen on the top of the satellite.
The proliferation of antennas is to provide redundancy. Figure 2 shows a block diagram of the UHF communications system. All RF modules are located in the UHF tray. The VHF system is very similar, so will be described simultaneously.
The receiving system design provides for full redundancy. The phase network of the turnstile antenna splits the transmitter and receiver paths. The received signal passes through the VHF trap into the UHF preamplifier and into receiver RX4. This has two crystal frequencies and a synthesiser feed. The monopole is also trapped and fed into the top preamplifier channel and into receiver RX3. Either antenna, preamplifier, or receiver can fail and communications will still be maintained, albeit with less desirable antenna patterns. By toggling the preamplifier relay, the receivers can also be swapped between preamplifiers for further flexibility. The VHF system has a similar layout, but replaces each UHF preamplifier with a pair of pin-switchable preamplifiers. The extra redundancy is provided since the VHF receivers are the prime uplink receivers for amateur use.
The transmitters have duplicated exciters labelled as TX3 and TX4. The exciters are passively combined with the UHF QPSK exciter output and fed to both power amplifiers, PA3 and PA2. PA2 will normally be the operating transmitter. The UHF relay can switch the turnstile to PA3 if PA2 fails. The additional relay enables the two PA's to transmit via the monopole if the cable to the hybrid fails.
The I and Q channels of the QPSK modulator are fed by a DSP output, enabling complex modulations to be generated via software. SUNSAT will thus be able to test complex modulation schemes in future research programs. The VHF system has no QPSK modulator, but instead has a receiver 455 kHz IF output that feeds the DSP system to enable new DSP demodulators to be evaluated. The other VHF receiver has four FM demodulators to provide many uplink channels to the satellite. The four channels will reduce receiver blocking when amateurs beyond terrestrial link range simultaneously attempt to access the satellite.
All receivers have been optimised for sensitivity with 15 kHz IF bandwidth. A few dB's are lost due to traps, protective diodes, combiners and cables. Spurious responses have also been controlled, so that at all frequencies except the desired frequency and main image frequency, received signals of -50 dBm will not influence reception9. The system has been designed for full duplex reception at VHF and transmission at UHF. The reverse operation far away from the third harmonic may also be possible, but needs further testing.
The power amplifiers have 5W output at VHF and S-band and 10W output at UHF, and can be commanded into a lower power mode to reduce power consumption.
The S-band downlink will permit QPSK transmission of the data described in the imager section. The transmitter can also be fed from a PAL TV camera on the satellite, which will be used for school interest, and possibly for visual orienting of the satellite during commissioning. The L-band uplink is capable of 2 Mbit/s data uploads for upgrading software. In addition, it can be linked to the S-band power amplifier to act as an amateur transponder into the 2400-2450 MHz amateur/ISM band. The received signal levels will be small, and possibly interfered with by the 2400 MHz wireless LAN operations, depending on location9.
SUNSAT will be an earth pointing satellite (body Z-axis towards nadir) to keep the imager in a nominal direction for usage and to provide acceptable antenna gain. The gravity gradient boom and tip mass will stabilise the satellite nominally earth pointing, thus requiring minimum control energy. The satellite will be kept in a slow Z-spin during normal operation (not during imaging), for improved solar thermal distribution. The four solar panels on the X/Y facets will thereby receive an even solar illumination, resulting in an improved life span of the solar cells and better thermal distribution on the satellite. A simple momentum transfer to a Z-axis reaction wheel will despin the satellite before imaging sessions.
Hardware Architecture
The satellite consists of two general purpose on-board computers (OBCs) and a dedicated attitude determination and control processor (ADCS). The two onboard computers are an Intel 188EC and Intel 386EX. The ADCS processor is a T800 transputer. In addition, there are numerous support processors (mostly 80C31 micro controllers) with dedicated functions.
The fault-tolerant design philosophy adopted for the satellite dictates that each communication channel between two hardware subsystems must be backed up by at least one other communication channel.
Software
This section describes some aspects of the engineering process followed in creation of the SUNSAT software and some points of the application level software are discussed.
Design and specification
The space segment software structure has been designed using the mechanism described in a separate paper5. The design method is built around a number of complementary visual languages. The languages include a data flow language, state charts, visual project management status and hierarchical class diagrams (from object oriented analysis and design methods).
Application software
Executing different versions of the same software was disbanded early on in the SUNSAT project due to the amount of man power resources the construction and maintenance it consumes. It was rather decided to invest more time in creating software that is re-usable across all SUNSAT hardware platforms. This led to far better software re-use between different groups of students designing software.
The application software consists of the following modules:
The telecommand system is centralised and distributes the switching signals to each subsystem on the satellite via its own wire interface. The telecommand system is twice three times redundant providing a six times redundancy. Any four of the duplicated access mechanisms can fail without degrading the mission. The two duplicated major channels are combined using an XOR function to enable a stuck line condition to be bypassed. Each of the on-board computers (OBCs) has access to one of the two duplicated systems and can listen to commands received on the other major system.
The telemetry system consists of a centralised data collection and packetising processing board with distributed data sampling and conditioning subsystems. The telemetry channel address is fed to each of the distributed modules with an address bus. Digital and analogue telemetry values are multiplexed onto a digital and analogue bus.
The data collection mechanism of the telemetry system is dual redundant. Selection between the two systems is done via telecommand. Telemetry packets can be transmitted through an on-board modem to the ground without any on-board computer intervention. However, in the normal operating scenario, the on-board computer will collect the information and transmit and store selected packets.
The SUNSAT power system consists of the following elements: a photovoltaic element, a storage element, a control element and a regulating element. The photovoltaic element consists of four solar panels located on the four sides of the satellite (the top and bottom sides excluded). Each of the GaAs solar panels has ten strings containing 19 solar cells per string. The power available from the solar panels is 30W/panel in full sunlight at normal incidence. The solar panels are connected to the power bus and batteries through diodes to prevent the batteries from discharging through the solar panels during eclipse.
Two battery packs, consisting of five NiCd cells each, make up the storage element of the power system. The two packs are connected in series to provide the unregulated bus voltage of 12 to 14V. The batteries will be used to deliver power to the satellite during eclipse. Batteries provide enough power to operate during peak loads when the remote sensing imager and the S-band transmitter are operating.
A shunt topology is used to control battery charging. Voltage regulation is done by several distributed regulators. To avoid a single point of failure, each subsystem has its own power regulator.
From the technical description above it is evident that many engineering and scientific disciplines are involved in building SUNSAT. A large number of graduate students received excellent engineering training by participating in this multi-disciplinary project. They have also gained systems engineering, project management and teamwork experience, which is very relevant to their future work in industry.
As an outgrowth of the SUNSAT project, the Electronics Systems Laboratory (ESL) has been established. Procedures have been developed to produce low cost engineering prototypes by utilising existing expertise and facilities at the university. Thanks to the high standard of science and engineering training in South Africa and the low cost of bursaries (compared to USA, Europe, etc.), local industry has a competitive advantage when tasking graduate students to perform research and development work as part of their studies.
The ESL has now set a goal of doing sufficient industrially relevant work to ensure long term viability. The goal is to retain approximately 50% of the effort on space related projects, mostly in the form of joint development of microsatellite subsystems with other countries. The rest of the ESL's work will focus on emerging opportunities in communications and process control system development. Spin-off projects currently performed from the SUNSAT technology base include digital signal processing of microwave signals, mineral extraction monitoring through froth imaging, and software development for digital TV signal compression.
The ESL enables academics to participate in large industry projects, where a substantial amount of the work is performed by university staff and student teams.
This paper has given a description of the first microsatellite developed in South Africa. It has been an ambitious task for a team with very limited space system design experience and even more limited financial resources. The progress so far is a tribute to a dedicated team of students and co-workers at Stellenbosch University. In particular it is a tribute all the SUNSAT supporters nationally and internationally who were prepared to give us a change to do something really exciting and challenging. We trust that SUNSAT will fulfil many of its operational expectations once launched in 1997.