Delfi-C3: a Student Nanosatellite as a Test-bed for Thin Film Solar Cells and Wireless Onboard

Communication W.J. Ubbels'!', A.K. Bonnema"', k.D. van Breukelen"', J.H. Doom"', R . van den Eikhofql', E. Van

der Linden"', G.T. Aalbers'"', J . Rotteveel'", R.J. Hmann"', C.J.M. Verhoeven"'

'"Faculty of Aerospace Engineering '*'Faculty of Electrical Engineering, Mathematics and Computer Science

Delft University of Technology The Netherlands

Abxfrac/-- For the past decade, satcllite design has been an important discipline at thc Faculty of Aerospace Engineering at Delft Universiry of Tcchnology. A major change of t i i t predominantly theoretical nature of the satcllitc projects camc into bcing in 2004. with thc completion of an in-house CIass 1UO.OOD Clean Room. Along with equipmcnt in laboraturics located at the Facult) of Electrical Engineering, this facility enables Delft University to not only design, but to actually produce and assemble a small student satellite in housc as well. Learning about this affordable access t o spaet, the space company Dutch Space approached tht Faculty of Aerospace Engineering to discuss the possibility to tcst a new type of thin film solar cells in the space environment. I n addition, the Dutch research inslilulr TNO Science LG Industry showed interest to join in with an autonomous Sun Sensor using a wireless link for data transfer. A third uew technology is an advanced high efficiency transceiver sized for application in pico- snd nanosatcllites. With these three new technologim as primary payloads and the spacc proven CubeSat conccpt as a basic principle of design, the Delfi-C3 nanorntclliie is to be the first satellite that i s designed and realized at Dclft University of Technology. The paper discusses thc Delfi-C3 mission and its design philosophy. Furthermore, B description of its subsystems snd payloads is given. Delfi-C3 is scheduled for a piggyback launch by the end of 2006.

1. INTRODUCTION In recent days much effort is invested in research and

development programs for Micro Systems Technology for space applications. A major objective of these research programs IS to reduce size, mass and powcr consumption of equipment, as these parameters have a direct relation to launch cost and space vehicle development cost. In addition, Micro Systems Technoiogy (MST) and Micro-Electronics (ME) enables new design solutions for space vehicle or instrumentation functions, oRen with bener performance.

Although a standard qualification approach prior to launch is followed, which is generally judged to be a satisfactor). guarantee that the equipment will hnction in B space

0-7803-8977-81OS/S~0.00 G2005 IEEE.

environment, demonstration of the actuat performance during a real space flight is of course more convincing and improves acceptance of the technology for new missions.

For MST and ME, which is characterizcd by relatively srnall size, low mass and low powcr demand, nanosatellites with a mass of I to 3 kg. a volume of I to 3 dm' and a typical power of 1 to 3 Watt can be a suitable qualification plstfom. During recent years a standard has been dcvelopcd for such satellitcs. Thc CubeSat concept has been initiated in 2000 by California Polytechnic State University San Luis Obispo (Cal Poly) and Stanford University's Space Systems Developmcot Lab, leading to a first launch of a number of university satellites in June 2003 [I]. Since then the concept has becn adopted by many universities and by industry. Today, s!andard CubeSat kits and subsystems are available, that can be uscd as a basis for a complete sateilite [2]. Launch services arc brokered by e.g. Cal Poly and University of Toronto Institute for Aerospace Studies for as few as 540,000 for a srrtellitc of I kg and a size of lOxlOx10 cm.

Although originally intended to offer educational institutions an affordable means to access space, the samc sei- up can be uscd to prequalify new MST and ME based systems or other new technology, on a small scale, in space. That approach has been chosen for Delfi-C3, a two-ycar satellite development project at the Faculty Aerospace Engineering (AE) and the Faculty Electrical Engineering, Mathematics and Computer Sciences (EEMCS) of Delft University of Technology in. the Netherlands, in co-operation with Dutch Spacc and TNO Science & Industry.

11. DELFI-C3 The idea for Delfi-C3 originated from the desire of Dutch

Spacc to have an early flight opportunity for innovativu thin film solar cells. Main objective of the space mission i s to measure the characteristic currcnt-voltage (IV) cuwcs of cight of these solar cells as a function of operating tempcrature and angle of incidence of the solar radiation.

167

Although the solar cells themselves are not MST and ME based systems, flying them provides the opportunity to pre- qualify such technology. Candidate MST and ME applications were collected from the MISAT program [3 1, 141. a five-year program starting in 2005 to develop micro- and nanosatellites that dcmonstrate a widc rangc of such applications for use in spacccraft. From this program, two applications were selected io bc flown on the Dclfi-C3 mission: an autonomous Sun senior with its own power supply and wireless data connection (TNO Science and industry), and a high eficiency Advanced Transceiver (TU Delfi EEMCS).

The transceiver is used to receive commands from the Delfi ground station and transmit measurement and housekeeping data back to the ground segment. There will be no on-board data storage, so all measured data will be transmitted directly to the ground segment. The satellite will make use of thc amateur radio frequency bands, so the transceiver will be compatible with the AMSAT standards and operate as a h e a r transponder that can be used by radio amateurs. In return the radio amateurs are asked to collect the Delfi-C3 data and send it to the Deifi ground station in Delft. In this way, data can be collected from the satellite even when it is not within range of the Delfi ground station.

Besides the technical mission objectives mcntioncd before, the Delf1-C3 project has educational objectives as well. Delfi- C3 is a Master's Thesis project, its main educational objective is to provide MSc students as a team the oppomnity to gain hands-on intcrdisciplinary engineering experience with the design and realization of both mission and systems of a satellite by providing a challenging real-world application

Fig. I. T h e DelfK3 satcllite

111. DESIGN PHILOSOPHY

The Delfi-C3 mission, initiated in November 2003, is an ambitious low-cost, short time-todelivery mission. The goal to launch the satellite by the end of 2006 results in an overall development time of two years. Adhering to such a short timc frame without losing track of completeness and precision requires a Systems Engineering (SE) approach. Where applicable, the standard SE approach is reduccd and adapted to the overall design philosophy of being innovative while

Fig. 2 : DslfiC3 System Breakdown.

To avoid delays as a result of tcchnology readiness issues, an important design requirement is to be independent of payioads and other new technologies fl0n1-1. For convenient organization rcasons, payloads are considercd as separate subsystems, as can bc seen in Fig. 2. This way, time tracking and control of the development of these payloads can be done at the highest level. Furthermore, to accelerate the whole design process from concept studies to verification and tcsting. both system and subsystems design are performed concurrently, with overlapping design phases as a result. The team, consisting of students with different bur interdisciplinary backgrounds, is cooperating in a ccntral office facing the in- house Clean Room facilities, improving communications and therewith easing design difficulties like dccision making and irrtcrface control.

11:. MfSSION CONCEPT AND HIGH-LEVEL REQUIREMENTS

Important high-level system requirements dictate that thc thin film solar cells (TFSCs) shall no! bc body-mounted due to thermal constraints. Temperature range requirements arc mainly dictated by the autonomou?: Sun Sensor's temperature range. These and other requirements Icd to thc decision to use

a 3-unit CubeSat measuring 1tiO~lOOx325 mm as the structural basis for Delfi-C3. The TFSCs thermal constraints necessitate a configuration with fou, dcployable panels, each containing a frame with an array o f two TFSCs and five GaAs-cells as the satellite's power : upply. For simplicity, no active attitude control is implcmentc.

Furthermore, the design does n. incorporate a battcry. Since the TFSC and Autonomou Wireless Sun Sensor {AWSS) payloads depend on the prq m c e of solar radiation, there is no need for the satellite to i+c operational in eclipse. This avoids the additional complex; y that is involved with using a battery.

Finally, the satetlitu will have a :;dundant architecturc, to ensure that the mission does nu; depend on the new technologies flown.

Since the Delfi-C3 satellite adherct to the CubeSat standard, this implies the use of the Poly Picn drrllite Orbital Deployer

Thc P-POD is the standard dcp ,qment mechanism for ChbeSats. The P-POD (Fig. 3) was d i .igned by Cal Poly to be

(P-POD) .

168

a reliable and cost effective deployment system. which can be mounted on most current launchers like thc Eurockot Launch Vehicle and the DNEPR.

The P-POD is basically an aluminium box with a spring loaded door. A maximum of three 1-unit CubeSats or one 3- unit CubeSat can bc accummodatcd in the P-POD. The satellites are ejected out of the P-POD by means of a large spring loaded plunger, giving the satcllite an exit velocity of 2 mh [ I ] . The CubeSats are kept in position by means of l'cflon-coated guide rails. Thc use of these guide rails creates a~ envelopc between the P-POD wall and structure that can be used to storc solar panels or other deployablc systems. The walk of the P-POD can not be used as a hold down system for the deployables themselves, they need to have their own hold down system that is activated some time after ejection out of

FiL;. 3. The 3-unit CubeSat structure and the P-POD (Source: Pumpkin lnc , Car Poly)

V. SUBSYSTEM OVERVIEW

A. Srrucrure und mechunisnis The satellite's structure is bascd on the 3-Unit CubeSat

structure (Fig. 3) from Pumpkin Inc., USA. The structure is madc of an aluminum alloy, which has been alodined on all surfaces to make the structure conductive. Furthermore, it has feet on both the top and bottom surfaces providing the mechanical interface with the P-POD. The four solar panels o f the satellite are deployed to an angle of 35 degrees with respect to the body, which is the optimum angle in order to guarantee the required minimum powcr provision in any attitude. Each panel i s 325 mm long, 85 mm wide and is made of 1.5 mm thick glass-fibre reinforced plastic (GFW). The TFSC payload is placed at the end of the panel in a frame. (Fig. 4). While in the P-POD, the panels will be held down by using synthetic wire. The solar panels arc deployed by melting synthctic wircs that arc holding thcm down. A spring mechanism at the hinge ofthe pancl cnsures deployment to the dcsired 35-degree angle. The deployment status of the solar panels \rill be verified by means of micro switches at the back

of the pancls.

Fig 4. GaAr panel and TFSC payload

Eight whip antennas (Fig. 5 ) are used on the satellitc. Four on the top surface and four on the bottom surface in ordcr to create a near omni-directional patiern. The antennas art: mounted a: 45-degree angles with respect to the satellite body. Two of thc antennas on cach surface are used for the downlink and the other two are used for the uplink. The uplink antennas have a length of 18 cm. The downlink antennag have a length of 50 cm, which i s longer than thc structure itself. The antennns will be rolled up to fit within the envelope a: the top and bottom of the structure. stajlng clear of thc AWSS payload. Currently, two design options remain for thc antennas; 0.3 mm piano wire and 6 mm wide tapc spring material. The measuring tape is mounted on hinges to separate the orientating and unfolding motion of the antennas. Thc antennas will also he deployed by using electrical heating wires that melt the synthetic wires holding down the antennas.

The electrical heating wires will be made redundant. Furthermore, in th: case of the solar panels, each pair of opposing panels will be held down with the same pair of wircs to minimize the amount of wires.

Fig. 5 . D e l t X 3 top view

B. EIecriical Power Subsysrem The Electrical Power Subsystem (EPS) has to be able to

provide sufficient power to the subsystems under all possible satellite attitudes. There will be a provision for power buffering to be abIc to avoid continuous initialization of the on-board computer in case of failure of one complete solar panel. This buffering may also compensate for the fluctuation

I69

in available power as a result of the satellite’s tumbling modulation, using the amateur standard AX25 protocol. The motion, The four solar panels accommodate five triple COMMS consists o f two redundant systems, the Advanced junction Gallium-Arsenide (GaAs) solar cells, delivering a Transceiver (ATRX) payload and an AMSAT platform. The minimum power of 3 W. Power from these cells is convertcd Advanced Transceiver payload will be discussed in section VI. by the EPS to provide 10 V DC to the communications The AMSAT platform functions as a backup option for the subsystem, and 5 V DC to the other subsystems. Advanccd Transceiver. f t is built using standard commercial-

off-the-shelf (COTS) components, providing the samc

C. Commnnd and Dora Hurlding ,Svsfenr The Command and Data Handling System (CDHS) is

designed around an ultra-low power Texas Instruments MSP430 rnicrocontroller, which is embedded in the fhgh7 microcontrolIer unit board. The microcontroller and the board are both supplied by Pumpkin Inc. ab pan of the CubeSat kit. The connections from the CDHS LO the other subsystems will be organized according to a star architecture, making the CDHS the central subsystem of the satellite. The main functions of the CHDS are collecting mission and science data for transmission to the ground stations, controlling the deployment of the antennas and solar panets, providing the ability to exccutc commands that have been uploaded from thc ground station and providing some measure of robustness in order to cope with failing subsystems.

The sofrware running on thc microcontroller will bc supported by the SALVO real-time opcrating system (RTOS), which is also provided as a part of the CubeSat kit. The SALVO RTOS was developed by Pumpkin Inc. to run in environments where little memory is available, while i t is still providing support for multi-tasking and events to developers. The software will be flexible but only in a limited fashion, allowing the ground team to change certain parameters but not the actual code itself.

Fur redundancy reasons, the CDHS incorporates a backup system that has been kept as simple as possible. Redundancy for antenna and solar panel deployment control will reside in the EPS, and the On-Board Computer (OBC) will be provided with an extcrnal watchdog. In case of irreversible OBC failurc, the data from the Thin Film Solar Cell experiment will be converted for downlink by an analog system. An analog multiplexer will feed the measurement data (voltages) into a Voltage Controlled Oscillaror which is built using discrete components. It will generate an audio tone with a frequency that is proportional to the measured voltage from the TFSC payload, which can easily be decoded on the ground The AMSAT platform will provide the downlink.

D. Communications Subsywm The communications subsystem (CQMMS) will perform

two functions. Firstly, it will provide the telemetry downlink and telecommand uplink fhctionality. Besides that, it will provide the aforementioned linear transponder service which can be used by amateur radio operators to communicate with each other. The satellite will have an uplink in the UHF amateur radio frcqucncy band (435438 MHz) and a downlink in the VHF amateur radio frequency band (145.8-146 MHz). Telemetry downlink will be done using 1200 bit/s BPSK

telemetry, telecommand and linear transponder functionality as the Advanced Transceiver. Such an AMSAT platform is a proven concept; it has flown on many amateur satelliteb before. Thc linear transponder will relay a 40 KRz wide passband from the UHF band to the VHF band. This provides a very flexible system, which can be used by amateurs using a wide variety of modulation schemcs. Furthermore, the linear transponder can be used for Doppler tracking in order to accurately determine the satellite’s orbital parameters. Last but not least, a linear transpondcr can easily be combined with a telemetry downlink. The eight antennas of the antenna subsystem will be phased to provide circular polarization to counteract Faraday rotation in the Earth’s atmosphere.

i

E. Thermal Subsystem The satellite will be covered by Multi-Layer Insulation

(MLI) which will have a black coating in order to prevent reflection of incident solar radiation against the back side of the Thin Film Solar Cell payload. The passive thermal subsystem has to cope with a relatively large intcrnal dissipation compared to the satellite’s volume. The amount of power that is to bc dissipated is mainly dictated by thc effciency of the communications system power amplifier. The top and bottom panels of the satellite act as thermal radiators.

F. Attitude Determination and Coiitrol Subsystem The satellite will not have active attitude control, becausc

this is not required to fulfill the primary mission objectives. In fact, a slow tumbling motion is required, since it allows all four TFSC panels to be exposed to solar radiation. However, a limit is posed on the satellite’s rotation rates in order to obtain reliable test results for thc TFSC payload. Thcrcforc, rotation rate limiting using magnetic hysteresis materials is applied.

The attitude determination subsystem uses the supplicd current and voltage from each individual GaAs solar panel to determine the sun vector with respect to the satellite’s referccnce frame. Four side mounted Si solar cells are used as an extra reference for the attitude reconstruction algorithm. When thc sun i s within the 90 degrees square field of view o f one of the two Autonomous Wireless Sun Sensor (AWSS) payloads, the reconstructed anitudc data i s augmented with the high accuracy data generatcd by the AWSS. The AWSS payload will be discussed in section VI.

170

VI. PAYLOAD OVERVJEW

A. Thio Film Solar Celts The Thin Film Solar CelI payload is based on the latesi

dcvclopmcnt in the area of photovoltaic cells by the Dutch space company Dutch Space. The cells consist of a C1GS photovoluic layer which is deposited by evaporation on a titanium bdse layer of 25 micrometers. Thc cdls will bc integrated tile-wise, ensuring a minimal loss of active cell a m . Thc intcrconnccts arc covered by thc ncxt cell. with an ovcrlap of 5 mm at lhe longest sidc (Fig 5). The mechanical interconnection has a low rcsistance (- 1 mil); it is placcd between thc gilded contact pads under low compressive pressure.

Fig. 6 TFSC lntegrntron Scheme

The aim of this new type of solar cell is to create a light- weight and low-cost product for future space applications. The target is a 50% cost reduction of solar arrays, while improving the power to mass perhrmancc with 50% compared to conventional solar cells. Cost IS projected to be lower than 35!J Euro pcr Watt at solar array level, and the power is expected to be more than 100 Watt per kilogram. The cell will have no need for a cover glass, but ir will have an emissivity-enhancing and cncapsulating dietecrric coating. The efticiency will be more than 12 % under AM0 light conditions. The performance of the TFSC payload will be tested by determining their characteristic TV-curves and cell temperature per TFSC pancl, which consists of two cells. Thc IV-curves will be measured by means of a programmable current sink; thc temperatures will bc mcasurcd by determining the electrical resistance of a dummy titanium cell mounted close to the actual TFSC’s. This measured data is then transferred to the OBC where i t i s formatted for transmission to the ground segment.

B. Autonomous Wircircless Sun Sensor

Delfi-C3 houses two analogue sun sensors (located at opposite sides of the satellite) that will be fully autonomous and wireless. With the typical internal volume o f a nanosatellite already being quite limited, reducing wiring is an important challcnge. The sensors will have a half-sized GaAs solar cell as their own power supply (Fig. 6), making them independent of the satellite’s electrical power system. This

autonomy is accompanied by a wireless radio frequency link. This link wit1 be either an adapted commercid off-the-shelf transceiver, operating in the 868 MHz band or a custom designed Ultra-Wide Band connection, depending on data rates and technology readincss. In Fig. 7, a block diagram of the AWSS is given.

Sun Sensor Aoenure --n

Fig. 7. Autonomous Wirrless Sun Smsoi

Although wireless data communication on board such small satellitcs might seem a bit supcrfluous in itself, implementing this technology adds to modularity and results in a flexible “plug and play” system. The autonomous sensor unit, in this case measuring approximately 60x40~20 mm, is a predcccssor o f an even smalIer digital version that can be mounted on c.g. a solar panel sacrificing just a single solar cell.

The in orbit expcriment will conccntratc on demonstrating the feasibility of the wireless link (immunity for disturbances; no interference with other equipment) and thc opcrttion of thc sun sensor under variable power supply.

Q u a d m i

I

Fig. 8. Autonomous Wireless Sun Sensor block diagram

C. High EfjicicncjJ Advanced Transceiver Thc other MST/ME experiment on board will be the high

efficiency Advanced Transceiver. Nanosatellites have limited available space for solar cells and batterics, so very limitcd power is available. Often the power system cannot provide the transmission power required by thc antenna, especially because of the limited efficiency of thc powcr amplifier (PA) in thc transceiver. Methods exist to optimize the efficiency of

171

the PA, but they tend to limit the operating bandwidth of the amplifier and the choice of modulation scheme. This makes application of these PA's in nanosatellites problematic. Furthermore, nanosatellites may perform a mission together in close cooperation. This implies that apart from communication to a ground station at various frequencies and data rates and apart from internal wireless communication with the on-board payloads, there wit1 also be inter-sakllitc communication at variable frequencies and data rates. So there will be many frequencies, bandwidths and modulation schemes in use in one nanosatellite formation. Restrictions on mass, volume and power consumption make it impossible 10 fly separate radios for each of the wireless lirks. One transceiver should be able to accommodate for all wireless communication. A crucial part in this is the availability of a wide-band, highly efficient and very linear PA. In this mission a FA will be tested that utilizes over-all double loop ncgative feedback, using integrated transformers in the feedback network and in the frequency compensation network. It has been shown that integrated transformers work very wcll in a wide band when used as feedback eiement IS]. Since negative feedback is a very powerful means to linearize an amplificr, the more non- linearity of the active part -especially the power stage can be rolcrated. When more non-linearity can be tolerated, it is possiblc to increase the efficiency of the power stage. Several cxperiments have shown that both a higher efficiency and a higher linearity can bc obiaincd using this technique. Space- qualification of this technique, where integrated transformers are optimized for wide-band use in feedback amplifiers, i s crucisl. The circuit diagram is shown in Fig.8. Table I shows the improvement that has been achieved in a first experiment comparing amplifiers with identical active parts (and efficiency), one with and one without feedback.

I

Fig.9. A mnrformcr based double loop negative feedback amp!ifirr.

TABLE I COMPARISON BETWEE7.i AUPLIFIERS WITH AND WITHOUT FEEDBACK

1x43 0.9CHz ~.OGHZ FB m a c 43dBc No FB -29dk -33dBc Improvement 17dB I OdB

decoding software will be made available to panicipating radio amateurs so that they can decodc and process thc telemetry data locally. This data u<!l then be sent to thc centralized Delfi command ground station via an Internel connection for furthcr analysis.

The Delfi command ground station will act as the command ground station. This ground station is already operational, and allows for valuable experience in satrllitc communication to be gained.

VIII. PROJECT PARTNERS

A. Dutch Space Dutch Space (formerly Fokker Space) is a space company

based in Leiden, The Netherlands. Dutch Space specializcs in three areas of space technology: solar arrays. stmctures CC: mechanical systems and advanced systems & engineering.

B l"0 Science d? Industry TNO Science 6t Industry is the Dutch govcrnmental

scientific research Institute, specialized in thc development of complete instruments or instrument subsystems for Earth observation and scicnce space projects. Other compctcnccs include precision engineering and mechanisms, calihration key-data and attitude sensors Tor satellites.

REFERENCES )lttD://CUbeSZst&&&&

httn:ilw~a.oumokininc .com W. Jorrgkirid, "The Dutch MST program MicroNed and its C I U S ~ C T MISAT". (submitted for publication 81 lhe ICMENS 2005). F. van Keulcn et.at. "MicroNed project plan", hemof Donimerir. Dclfi. Feb. 2005 Kocn van Haninpveldt, Michiel Kouwenhoven. Chris Verhoeven. "H1' low noise amplifiers with integrated transfur" feedback", ISCASZOO?, USA

VII. GROUND SEGMENT The ground segment consists of two parts, the distributed

ground station network, formed by amateur radio operators and universities around the world, and the Delfi command ground station at Delft University of technology. Telemetry

I72