You are on page 1of 4

SCARLET: DESIGN OF THE FRESNEL CONCENTRATOR ARRAY FOR NEW MILLENNIUM DEEP SPACE 1

David M. Murphy and Michael I. Eskenazi AEC-ABLE Engineering, Inc. 7200 Hollister Avenue, Goleta, CA 93117

ABSTRACT
The primary power for the JPL New Millennium Deep Space 1 spacecraft is a 2.6 kW concentrator solar array. This paper surveys the design and analysis employed to combine line-focus Fresnel lenses and multijunction (GalnPJGaAdGe) solar cells in the second-generation SCARLET (Solar Concentrator Array with Refractive Linear Element Technology) system. The array structure and mechanisms are reviewed. Discussion is focused on the lens and receiver, from the optimizations of optical efficiency and thermal management, to the design issues of environmental extremes, reliability, producibility, and control of pointing error.

BACKGROUND
SCARLET is a concentrator solar array for space applications which uses linear refractive Fresnel lenses to focus sunlight onto spaced rows of solar cells. For a given power level, the SCARLET optical system reduces the required solar cell area by approximately a factor of seven. The decreased cell area significantly reduces solar array system cost and weight especially in high radiation environments where thick cell coverglass is required. The array is scaled up from the prototype SCARLET array that was built for the METEOR satellite in 1995. [ l ] Due to the failure of the Conestoga launch vehicle, this will be the first flight of a modular concentrator array. The BMDO Innovative Science and Technology Directorate is sponsoring development of this secondgeneration SCARLET solar array - which incorporates advanced technologies such as dual-junction solar cells and an improved structural design for use on the first JPL New Millennium Deep Space flight, called DSl, which is to be launched in July of 1998. SCARLET will provide 2.6 kW to DS1 to directly drive an ion propulsion engine for a 2-year mission which rounds Mars and travels out to 1.6 AU to observe the comet West-Kohoutek-Ikemura. [2]

result is a simple cable-synchronized structure which deploys flat. The major advantages are fewer piece parts, simplified pointing control analysis, reduced stowed volume, and simplified yoke structure. Additionally, the lens panels are held securely between power panels in the stowed condition. Basic proven mechanisms such as release assemblies, tiedown cup-cones and cable pullers, and hinges of SCARLET I were utilized again for SCARLET II, but optimized to minimize weight. [3] Once the fundamental structural and component building blocks were in place, the task was to optimize their interaction. The optimization involved a complex matrix of quantifiable variables entailing optical alignment, thermal management, mass mininjzation, reliability enhancement, and parVassembly tolerance control traded against cost, producibility, and less tangible criteria such as technical risk and the role of the perceived risk in future marketability. The resulting design performance is shown below: Wing Power: Weight: Specific Power: Stiffness: Strength: Wing: 1300 W minimum required

61 Ib (including tiedowns)
48 W/kg 0.2 Hz Deployed, 70 Hz Stowed 0.01 5 g Deployed 206 in. x 45 in.

Mechanical Description The DS1 SCARLET solar array consists of two wings of four panels each. The wings are delivered fully integrated with tiedowns, gimbal drive assembly, and spacecraft interface plate (See Fig. 1).

SYSTEM DESIGN
Overview The first generation SCARLET array was a melding of ABLES standard planar array structure, PUMA [3], with concentrator optics. That structural baseline was reassessed for the DS1 SCARLET design to improve the union between the cell substrates and lens panels. The

Figure 1. DSl SCARLET Wing

86 1

0-7803-3767-0/97/$10.00 0 1997 IEEE

26th PVSC; Sept. 30-0ct. 3,1997; Anaheim, CA

Deployment of a wing is initiated when power is applied to the high output paraffin (HOP) linear actuators in each of two tiedown assemblies. A resistive load causes the paraffin to heat and change phase, which forces a pin forward releasing the restraint arm on a torsion tube. Tiedown cables are wrapped and captured in fittings on either end of the tube, so when a torsion spring revolves the tube, both cables are released. When the second release mechanism has actuated, the wing unfolds driven by double-wound torsion springs distributed on each hingeline. The hingelines are synchronized by a system of cables which are wound over static pulley cams. The synchronization transfers the deploy torque to the root where redundant rotary viscous dampers retard the deployment rate.

The cell receiver module consists of five series cells, each with bypass diodes, affixed to a circuit on a high thermal conductivity substrate as depicted in Figure 2. The modules are joined using overlapping redundant tabs with reflowed solder to form 50 cell strings that generate 40 watts at an operating voltage of 90 volts at 1 AU.
Fresnel Lens Glass over Slllcona

Electrical Description
The cells used by DSI SCARLET are about Icm wide (4 cm long) and are spaced in rows 8.6 cm apart. The low cell area per watt needed beneath the concentrator greatly lowers cost and also eases the risk of utilizing emerging, high-performance cell technologies. For this reason, BMDO elected to specify the procurement of an entirely multibandgap-cell-basedsolar array. In 1996 production quantities of GalnPJGaAslGe dual-junction cells were not yet available. Tecstar was the only cell vendor willing to participate, and the DSI team was cautious about the difficulties of bringing new cell technology into production. So to mitigate risk, and to set performance criteria for the flight build, an engineering build quantity of 100 cells was procured. The Ill-V cell design, termed Cascade by Tecstar, had previously been qualified in the standard series of environments for space applications. The only modification required was gridline sizing for the high flux profiles of the concentrator. The engineering evaluation result was very encouraging, with the average efficiency result coming in at 24.25% at 7.5X AMO. The performance criteria for the flight build was set at 23.25%, - partially because of losses anticipated for glazing, but mostly as insurance against the uncertainties of a larger build. During the flight production phase Tecstar experienced a series of setbacks in producing the flight cells. The most persistent problem was shunting (Reviewed in [4]) which reduced the performance of many of the cells to as low as 16% at 1 sun intensity. Fortunately, the high current injection levels created by the lens overrides the fixed magnitude shunts and the performance at concentration is only slightly degraded. After intensive effort by Tecstar, and aided by the synergism of the early dual-junction-cell Mantech program, remarkable improvement in yield and performance were achieved. But schedule delays eventually forced the acceptance of cells with a minimum lot average - under concentration - of 22.6 Yo. The cells were glazed by Tecstar with 0.004-inchthick coverglass with an anti-reflection coating with bluehed filtering (BRR). The reflection of the near infrared lowers the operating temperature of the cell by 11C.

Blocking Dlode

Flex Clrcun Bonded


to Carbon-Carbon Module Base
Bypass Diodes ( 5 )

Figure 2. SCARLET Power Module Cells in the module are interconnected along both their long edges. Given the long aspect ratio (4:1), the most probable crack direction will never leave a section of the cell isolated. Dual ohmics also provide balanced offtrack performance and lower gridline resistance losses. Cell interconnect reliability is also greatly improved over standard CIC construction because 120 interconnects (in parallel per cell) connect the cell to the circuit board carrier. The automated wire bonder, which stitches at a rate of three cells per minute, results in large cost savings by eliminating hand labor. Engineering modules underwent thermal cycling from -160C to +llOC for 100 cycles to assure a margin of compatibility with the single thermal cycle experienced on the DS1 mission at the start of its interplanetary mission. All modules experienced no visible degradation and comparison of pre- and post-lV curves under the X25 solar simulator at LeRC showed no measurable electrical degradation.

Optical Description
The Fresnel lens is comprised of over 1600 precisely formed individual ridges which refract incident light from a 3.22-inch aperture down to a 0.17-inch strip of light focused in the middle of the 0.40-inch-wide cell strip to leave greater than +2 degrees margin for pointing error. The average optical efficiency of the DS1 lenses, which have no anti-reflective coatings, (used with t h e Cascade cell described above) has been measured at 89%. The effective concentration ratio, 7.14 (= 0.89 x 3.22/.40), was selected to provide for reasonable pointing error. The purpose was to create a cost-effective system to manufacture and assemble which is compatible with standard gimbal and spacecraft ACS architectures. The linear Fresnel pattern is molded in a continuous roll process using space-grade silicone. Individual lenses are machined-trimmed and bonded to glass superstrates

862

which have been thermally formed into cylindrical sections. The materials chosen for the lens, the bondline, and the glass are well understood: DC 93-500 silicone and ceria-doped borosilicate glass (Corning 0213). The glass protects the lens from particle radiation and with an AR/ITO coating, planned for future programs, the optical efficiency is enhanced and charge buildup is minimized. Optimization of the lens system began with the variables of aperture width and concentration ratio. Aperture width affects cell temperature and the piece part count per watt. As ihcreasedyldth was pursued to reduce part count costs, the corresponding increase in the cell temperature had to be considered. The remarkable conductivity of the pitch-based graphite fiber used in the panel (which for the layup used is higher than aluminum) allowed the use of the desired size without appreciable impact to the cell temperature. Larger lens elements overly quantize the available power levels, which can constrain the layout options on smaller arrays. Wider lenses are taller and therefore negatively impact the overall stowage height. For higher concentration levels pointing error tolerance is reduced. Furthermore, a higher concentration ratio (at the same aperture width) has a diminishing effect on cell cost savings when countered by the increase in part and assembly costs due to tighter tolerancing requirements. The length of the lens was set primarily to allow lens panel structure weight to be minimized, and to maintain a low-risk aspect ratio for manufacturing of the lens superstrate. The space between lenses must be minimized to maximize packing factor (the ratio of the area of light which passes through the lens to the total panel area). On the long side of the lenses the thin lens panel structure (0.045 in.-thick elements) is vulnerable to large amplitude in-plane (IP) vibration. In this mode the lens will roll side-to-side and flex the 0.003-in.4hick glass. The outof-plane (OOP) distortions minimized by the stiffness of the 0.5-in.-tall frame members and the hold-down points at every other frame member intersection. To demonstrate the survivability of the thin glass lens mounted in this minimal structure frame, five lens-in-frame components were tested - successfully - to predicted local acoustic/random levels (29 Grms OOP, 9 IP). The efficiency of the lens overall is a function of the refractive index matching of the lens, superstrate, and optical coating used, as well as the surface finishes and sharpness of the lens teeth. The manufacturing of the lens produces smooth and sharp prisms with small root radii that cause a maximum of 2.6% scattering loss. The close match between the refractive index of the silicone and the glass (1.523 and 1.409 respectively) causes a slight loss of 0.3%. The losses are summarized in Table 1. Well established optical coatings would reduce the large loss the lens outer surface transmittance, but the thermal forming of the lens superstrates occurs at a temperature which is higher than the survival temperature of typical coatings. Application of anti-reflection (AR) coatings to the curved surface of the glass superstrate was developed at OCLI, but not in time to coat all flight lenses.

Table 1. Optical Losses in Lens Assy (without AR coating)


Component Material Interface Reflection
4.5%

Absorptance Scatterina
0.0%

Space Cover Lens Space


Totals

Vacuum Glass
0.3%

0.5%

Silicone 3.0% Vacuum


7.7% Combined Loss:

3.1yo
0.0%
3.6%

11.O%

(Multiplicative along light path) Two coated lenses will fly on the DS1 array in positions where their contribution to module efficiency can be measured and compared to non-coated lenses. The coating developed combines AR performance with electrical conductivity, using IT0 (Indium Tin Oxide) to dissipate charge to the grounded lens frame structure. Component testing of the coated lenses demonstrated a 2% efficiency gain, from 89.5% to 91.5%. Thermal Design The thermal design challenge is to spread the absorbed but unconverted solar energy (heat) from the cell modules out across the panel to engage the full area and high emissivity of the graphite panel to radiate efficiently. The cell, circuit layers, and panel were analyzed with a detailed finite difference model so that material choices and thicknesses could be optimized to reduce the cell temperature. The calculation results, for 1 AU illumination, are shown in Figure 3.
68 67 66 65 64
h

Distance from Cell Centerline

U, E

63 62 61 60
59

a E

8 H
0 0.1 0.2 0.3
0.4
0.5

&

0.6

0.7

0.8

0.9

Figure 3. Temperature Profile Across Cell (Edge at D = 1) The largest temperature rise is caused by the Kapton. The thermally conductive silicone adhesives and the cell carrier present very little resistance to thermal conduction. The carrier, which is used during cell laydown to stabilize the Kapton circuit and thus protect the cells from damage, is made of a high-conductivity composite to match the thermal expansion coefficient (nearly zero) of the panel.

863

This creates minimal strain along the long bondline, and in addition, the material is very light and stiff. As can be seen in Figure 4, once conducted to the panel the heat spreads rapidly out through the facesheets and core. This is because the facesheets are constructed of a ultra-high-conductivity fiber (which also possesses good compressive strength) which has a conductivity of 384 W/m-K (for unidirectional layup, 60% fiber volume), which is 60% higher than pure aluminum.

istance from Module Centerline


7o 65
60
55

7
A

-+-Cell module stack +Facesheet. front

demonstrated on two tap modules of the DS1 SCARLET array, meets this need. A second common mission requirement not required by the DS1 mission is extended thermal cycling. To validate the flight worthiness of the SCARLET hardware for the additional environmental stress, two modules have been 10C. tested for 1650 thermal cycles from -165C to +I NASA and BMDO also continue to develop technologies to improve the performance of SCARLET. These include development of advanced multiple band gap cells to deliver 30 to 35% efficiency in a concentrator array, monolithic polymer concentrator lens materials that can survive both radiation and UV exposure, demonstration of the radiation tolerance of the cells and array modules through ground and flight testing, and high voltage designs for direct drive electric propulsion purposes.

50

0.1

0.2

0.3

0.4

0.5

0.6

0 . 7 0.8 0.9

Fig. 4. Temperature Profile Across Panel (From cell center to centerline between module rows) Typically, SCARLET cells operate about 20C hotter than a planar GaAs design, mostly due to front surface radiation blockage by the lenses and the temperature gradients associated with heat spreading. The thermal modeling has been well-correlated with a balloon flight module [5] and more recently, with a vacuum thermal balance test at NASA Lewis. For the balloon flight the calculated temperature differential, 15.6C (for the non-flight carrier used which has lower thermal conductivity), was in fair agreement with , and the pre-flight that derived from the measured cell V measured cell temperature coefficients - which ranged from 17.2 to 18.OC for all 65 flight data points.

The DS1 SCARLET array represents successful optimization at the system level of reliable mechanization, optical performance, thermal management, mass reduction, and producibility. The manufacturing, qualification testing, and DSI flight demonstration will position the SCARLET array for low-risk implementation in todays commercial satellite applications.

The work reviewed herein was the product of a small team of engineers with key support from BMDO, JPL, and NASA Lewis.

[ l ] Allen, D., Jones, P., Murphy, D., and Piszczor, M., The SCARLET Light Concentrating Solar Array, Conference Record of the Twenty Fifth IEEE Photovoltaic Specialists Conference, IEEE, 1996. [2] Chmielewski, A., et al., The New Millennium Program Power Technology, Proceedings of the 3 1st lntersociety Energy Conversion Engineering Conference, IEEE, 1996.
[3] Wachholz, J., and Murphy, D., SCARLET I: Mechanization Solutions for Deployable Concentrator Optics Integrated with Rigid Array Technology, Proceedings of the Aerospace Mechanisms Conf., 1996.

During the design and fabrication of the DS1 SCARLET array there has been a concurrent review for potential improvements. A number of advancements are being developed as part of ABLES internal R&D effort. The most significant improvements which have been demonstrated to date are the elimination of the carrier used in cell laydown and of the lens clip stampings used in the lens panel assemblies. These improvements contribute equally to a 10% increase in specific power while reducing fabrication costs. A demonstration model has been built which shows the stowed height can be significantly reduced by nesting the lens panels into the substrate panels. For many applications an electrically conductive coating on the lenses is highly desired to reduce discharge arcing. The ARATO coating mentioned earlier, which will be flight-

[4] Eskenazi, M., Murphy, D., Ralph, G., and Yoo, H.,
Testing of Dual Junction SCARLET Modules and Cells

Plus Lessons Learned, Conference Record of the Twenty Sixth IEEE Photovoltaic Specialists Conf, IEEE 1998.
[5] Eskenazi, M., Murphy, D., Anspaugh, B., Brinker, D., and ONeill, M., Balloon and Lear Jet Testing o f SCARLET Modules and Cells, Proceedings of the 15th Space Photovoltaic Research and Technology Conference (SPRAT XV), 1997.

864

You might also like