EF233X DEGREE PROJECT IN SPACE TECHNOLOGY, SECOND CYCLE, 30 CREDITS
STOCKHOLM, SWEDEN 2019
Test bench for
Nanosatellite A tude Determina on and
Control System (ADCS) devices
Design and manufacture of a Merri Cage
Álvaro Cano Torres
A tude Determina on and Control System (ADCS) is o en a complex system on-board any satellite which needs valida on and tes ng to prove its operability and verify its so ware compa bility with hardware and other subsystems. One failure in orbit is extremely expensive in terms of cost and me due to payload prepara on and launch.
The ideal test bench would be the one that perfectly simulates the space environment and all its main factors such as weightlessness, Earth’s Magne c Field (EMF), vacuum, neutral par cles, plasma and radia on, among others. The target in this case was the Earth’s Magne c Field (EMF), solved with a Helmholtz Cage in a Merri Configura on, and weightlessness, not implemented but analysed in detail where different alterna ves are proposed, similar to market solu ons.
As derived from literature and simula ons executed along this M. Sc. Thesis, the Merri Cage seems beneficial against any other configura on in terms of magne c field uniformity and effec ve volume. A er the design and assembly of the test bench, both proper es were verified and successfully achieved, despite the lack of calibra on, not executed because of me limita on, and ny issues encountered along the full evolu on of the project.
Keywords
Helmholtz, Merri , Cage, Test, ADCS, Nanosatellite, Magne c, Orbit, Propaga on, Weightlessness, Air, Bearing.
Abstract
A tude Determina on and Control System (ADCS) är o a e komplicerat system ombord på alla satelliter som behöver validering och testning för a bevisa dess användbarhet och verifiera dess programvarukompa bilitet med hårdvara och andra delsystem. E fel i omloppsbana är extremt dyrt med avseende på kostnader och d på grund av förberedelse och lansering av ny olast Den ideala testbänken skulle vara den som perfekt simulerar rymdmiljön och alla dess huvudfaktorer såsom viktlöshet, Earth’s Magne c Field (EMF), vakuum, neutrala par klar, plasma och strålning, bland andra.
Målet i de a fall var EMF, löst med en Helmholtz-bur i en Merri -konfigura on, och viktlöshet, inte implementerad men analyserad i detalj där olika alterna v föreslås, liknande marknadslösningar.
Som härstammar från li eratur och simuleringar u örda längs denna M. Sc. Avhandling verkar Merri Cage vara gynnsam mot annan konfigura on när det gäller magne ältens enhetlighet och effek v volym. E er konstruk onen och montering av testbänken, var båda egenskaperna verifierade och framgångsrikt uppnådda, trots bristen på kalibrering, inte genomförda på grund av dsbegränsning, och små problem som uppstod under projektets fulla utveckling.
Nyckelord
Helmholtz, Merri , bur, Test, ADCS, Nanosatellit, Magne sk, Bana, Förökning, Tyngdlöshet, Lu lager.
To Steven, Geert Henk and Nickolay,
whose inspiring ideas and advices guided me along this project, and to my parents,
whose values they ins lled in me as a child have served me well throughout my life.
Author
Álvaro Cano Torres alvarocanotorres@gmail.com Aerospace Engineering. Space Explora on KTH Royal Ins tute of Technology
TU Del Technical University
Place for Project
Hyperion Technologies B.V.
Del , Netherlands
Examiner
Tomas Karlsson
Division of Space and Plasma Physics KTH Royal Ins tute of Technology
Supervisor
Mykola Ivchenko
Division of Space and Plasma Physics KTH Royal Ins tute of Technology
External Supervisor
Geert Henk Visser Design Engineer
Hyperion Technologies B.V.
Contents
1 Introduc on 1
1.1 Problem statement . . . . 1
1.2 Methodology . . . . 2
1.2.1 Deliverables . . . 3
1.2.2 Time plan . . . 4
2 Defini on of requirements 6 2.1 System Requirements . . . . 6
2.2 Func onal Requirements . . . . 7
2.2.1 Earth’s Magne c Field problem . . . 7
2.2.2 Fric onless rota on problem . . . 7
3 Market Analysis 9 3.1 Decision criteria . . . . 9
3.2 Earth’s Magne c Field problem . . . 10
3.2.1 Helmholtz Cage . . . 10
3.2.2 Alterna ve approaches . . . 15
3.2.3 Conclusion . . . 15
3.3 Fric onless rota on problem . . . 16
3.3.1 Wire suspension . . . 17
3.3.2 Air bearing . . . 18
3.3.3 Gimbal suspension . . . 23
3.3.4 Rotary table . . . 25
3.3.5 Water bearing . . . 27
3.3.6 Magne c levita on . . . 27
3.3.7 Alterna ve approaches . . . 28
3.3.8 Conclusions . . . 29
v
Test bench for Nanosatellite ADCS devices Álvaro Cano Torres
3.3.9 Research focus . . . 30
4 Fri onless Rota on 31 4.1 Trade-off . . . 31
4.1.1 Hemispherical Air Bearing . . . 31
4.1.2 Rotary table . . . 38
4.1.3 Wire suspension . . . 40
4.1.4 Gimbal suspension . . . 40
4.2 Design . . . 41
4.2.1 Dimensions . . . 41
4.2.2 Structure . . . 42
4.2.3 Equipment . . . 42
5 Helmholtz Cage 44 5.1 Trade-off . . . 44
5.1.1 Two square-coil configura on . . . 45
5.1.2 Four square-coil configura on . . . 45
5.2 Simula on . . . 50
5.2.1 Helmholtz cage (two circular coils) . . . 53
5.2.2 Merri cage (four square coils) . . . 56
5.2.3 Lee Whi ng cage (four circular coils) . . . 61
5.2.4 Tetracoil cage (four circular coils) . . . 65
5.2.5 Two octagonal coil cage . . . 68
5.2.6 Four octagonal coil cage . . . 70
5.2.7 Conclusion . . . 73
5.3 Design . . . 74
5.3.1 Considera ons . . . 74
5.3.2 Dimensions . . . 78
5.3.3 Components . . . 79
5.3.4 Wiring . . . 101
5.3.5 Cage controller . . . 104
5.4 Manufacturing process . . . 108
5.4.1 Bill of Materials . . . 108
5.4.2 Assembly . . . 110
5.4.3 Disassembly . . . 122
6 Verifica on and results 123 6.1 Coil specifica ons . . . 123
6.2 Ambient magne c field and room survey . . . 124
6.3 Generated magne c field . . . 125
6.3.1 Small coils placed along x-axis . . . 126
6.3.2 Uniformity . . . 127
6.3.3 Dynamic Cage Control . . . 130
6.4 Conclusion . . . 130
7 Conclusions and recommenda ons 132 7.1 Upgrade with fric onless rota on system . . . 132
7.2 Future work . . . 132
7.3 Final Words . . . 135
Bibliography . . . 136
Test bench for Nanosatellite ADCS devices Álvaro Cano Torres
List of Figures
1.2.1 V-diagram summarising Project Milestones. . . 3
3.2.1 Helmholtz Cage at Georgia Ins tute of Technology. . . 12
3.2.2 Helmholtz Cage at Massachuse s Ins tute of Technology. . . 12
3.2.3 Helmholtz Cage with Sun simulator at DLR Berlin. . . 13
3.3.1 Wire suspended nanosatellite inside Helmholtz cage concept. . . 17
3.3.2 Planar air bearing. Courtesy: ESA, ESTEC . . . 19
3.3.3 COTS Hemispherical air bearing. Courtesy: PI . . . 20
3.3.4 Concept of dumbbell (le ) and umbrella (right) configura on. . . 21
3.3.5 Experimental spherical hollow air bearing at Cal Tech. . . 22
3.3.6 Spherical air bearing structure held by gimbal suspensions [13]. . . 24
3.3.7 Gimbal suspension (le ) with ac ve compensa on motors (right). . . 26
3.3.8 Air-bearing Direct-drive rotary table. Courtesy: Aerotech. . . 26
3.3.9 Magne c levita on testbed for nanosatellites. . . 28
4.1.1 Absolute rotary encoder. Courtesy: Renishaw. . . 34
4.1.2 Darkfield laser sensor used on computer mice. Courtesy: Logitech. . . . 34
4.1.3 Stereo Cameras system. Courtesy: Naval Postgraduate School. . . 35
4.1.4 Concept of stereo Cameras system, on Air bearing. . . 36
4.1.5 HTC Vive Tracker (le , middle), and Laser lighthouse (right). Courtesy: HTC 36 4.1.6 Two axis inclinometer. Courtesy: Level Developments Limited. . . 37
4.1.7 Painted sphere with infrared sensors. . . 38
4.1.8 Air Bearing Rotary Table RT-150S. Courtesy: LAB Mo on Systems. . . 39
4.1.9 Three-axis Turntable iTURN 3 S1. Courtesy: iMAR Naviga on and Control 39 4.2.1 Preliminary design for hemispherical air bearing ball. . . 42
5.1.1 Cage configura on of two square-coil (le ) and four square-coil (right). . 44
5.1.2 Magne c field lines by FEM Simula on. Courtesy: COMSOL . . . 45
5.1.3 Uniformity of the axial field in the four-coil Helmholtz cage. . . 46
5.1.4 Concept of structure for a Merri Cage. . . 47
5.2.1 Scheme of Cage with Circular coil configura on. . . 51
5.2.2 Scheme of Cage with Square coil configura on. . . 52
5.2.3 Scheme of Cage with Octagonal coil configura on. . . 52
5.2.4 Resultant magne c field ⃗B. Magnitude at the centre is 200 µT. . . 54
5.2.5 Devia on of magne c field ⃗B. Same scale as Figure 5.2.4. . . 54
5.2.6 Devia on of magne c field ⃗B in y-axis direc on. . . 55
5.2.7 Effec ve volume side length def for d = 1500 mm. . . 55
5.2.8 Resultant magne c field ⃗B. Magnitude at the centre is 200 µT. . . 57
5.2.9 Devia on of magne c field ⃗B. Same scale as Figure 5.2.8. . . 57
5.2.10Devia on of magne c field ⃗B in y-axis direc on. . . 58
5.2.11Effec ve volume side length def for d = 1500 mm. . . 58
5.2.12Devia on of ⃗Bfor inaccuracies of 5 mm. Same scale as Figure 5.2.8. . . . 60
5.2.13Devia on of ⃗Bin y-axis direc on for inaccuracies of 5 mm. . . 60
5.2.14Effec ve volume side length deffor d = 1500 mm for inaccuracies of 5 mm. 61 5.2.15Devia on of ⃗Bfor inaccuracies of 10 mm. Same scale as Figure 5.2.8. . . 62
5.2.16Devia on of ⃗Bin y-axis direc on for inaccuracies of 10 mm. . . 62
5.2.17Effec ve volume side length def for d = 1500 mm for inaccuracies of 10 mm. . . 63
5.2.18Resultant magne c field ⃗B. Magnitude at the centre is 200 µT. . . 63
5.2.19Devia on of magne c field ⃗B. Same scale as Figure 5.2.18. . . 64
5.2.20Devia on of magne c field ⃗B in y-axis direc on. . . 64
5.2.21Effec ve volume side length def for d = 1500 mm. . . 65
5.2.22Resultant magne c field ⃗B. Magnitude at the centre is 200 µT. . . 66
5.2.23Devia on of magne c field ⃗B. Same scale as Figure 5.2.22. . . 66
5.2.24Devia on of magne c field ⃗B in y-axis direc on]. . . 67
5.2.25Effec ve volume side length def for d = 1500 mm. . . 67
5.2.26Resultant magne c field ⃗B. Magnitude at the centre is 200 µT. . . 68
5.2.27Devia on of magne c field ⃗B. Same scale as Figure 5.2.26. . . 69
5.2.28Devia on of magne c field ⃗B in y-axis direc on. . . 69
5.2.29Effec ve volume side length def for d = 1500 mm. . . 70
5.2.30Resultant magne c field ⃗B. Magnitude at the centre is 200 µT. . . 71
Test bench for Nanosatellite ADCS devices Álvaro Cano Torres
5.2.31Devia on of magne c field ⃗B. Same scale as Figure 5.2.30. . . 71
5.2.32Devia on of magne c field ⃗B in y-axis direc on. . . 72
5.2.33Effec ve volume side length def for d = 1500 mm. . . 72
5.3.1 Free area between frames in Merri Cage, for d = 1500 mm. . . 76
5.3.2 Movable coils to increase cage accessibility. . . 77
5.3.3 Height of effec ve volume, at eye level. . . 77
5.3.4 Minimum effec ve volume, containing CubeSats up to 12U. . . 78
5.3.5 General required dimensions of the cage (first itera on). . . 79
5.3.6 Defini ve dimensions of the cage. . . 80
5.3.7 T-slo ed Profiles of 40× 40 mm. Courtesy: Alux Profile. . . 81
5.3.8 Anchor pin connec ng T-slot profiles at 90◦. Courtesy: Alux Profile. . . 82
5.3.9 Screw connector connec ng T-slot profiles at 90◦. Courtesy: Alux Profile. 83 5.3.10Angle piece connec ng T-slot profiles at 90◦. Courtesy: Alux Profile. . . . 83
5.3.11Cage using T-slot profiles at corners as s ffeners. . . 84
5.3.12Triangular plates used at the cage corners as s ffeners. . . 85
5.3.13Cage using triangular plates at corners as s ffeners. . . 85
5.3.14Square frames connected with an L-shaped piece, usually brazed. . . 87
5.3.15Detail of frame connec on made with U-profile at 45 degrees. . . 88
5.3.16Integra on of square frames connected with a U-profile at 45 degrees. . . 88
5.3.17Coil connector adding s ffness to the square frame. . . 89
5.3.18Detail of integra on of the coil connectors in the final assembly. . . 90
5.3.19Corner piece linking both U-profiles to form a square frame. . . 91
5.3.20Corner piece used to link two U-profiles to form a square frame. . . 91
5.3.21Integra on of square frames connected with laser-cut corner pieces. . . . 92
5.3.22Smallest ver cal coils hanging from intermediate ver cal coils . . . 93
5.3.233D-printed pieces a aching coils to the main structure. . . 94
5.3.24Detail of integra on of the 3D-printed holding pieces. . . 95
5.3.25Holding pieces made of sheet metal linking coils to main structure. . . 96
5.3.26Detail of integra on of the holding pieces. . . 97
5.3.27Integra on of full cage with selected alterna ves for each component. . . 98
5.3.28Front view of the integra on of full cage. . . 99
5.3.29Top view of the integra on of full cage. . . 100
5.3.30Outer (le ) and inner (right) coil, 15× 15 × 3 mm U-profiles. . . 103
5.3.31Outer (le ), 25× 25 × 3, and inner coil (right), 20 × 20 × 3 mm U-profiles.104
5.3.32Distribu on and numera on of coils in Merri Cage configura on. . . 106
5.4.1 Assembly of main structure made of extruded profiles. . . 110
5.4.2 Main structure made of extruded profiles. . . 111
5.4.3 Inclusion of horizontal triangular gussets. . . 111
5.4.4 Addi on of holding pieces for largest (horizontal) coils. . . 112
5.4.5 Posi oning of largest (horizontal) coils. . . 112
5.4.6 Inser on of medium (ver cal) coils for subsequent posi oning. . . 113
5.4.7 A achment of the outer medium coils with U-shaped pieces. . . 114
5.4.8 A achment of the inner medium coils with holding pieces. . . 114
5.4.9 Posi oning of medium (ver cal) coils. . . 115
5.4.10Detail of free space to insert the remaining coils. . . 116
5.4.11Inser on of small (ver cal) coils for subsequent posi oning. . . 116
5.4.12Small (ver cal) coils placed inside the cage ready to be posi oned. . . 117
5.4.13Addi on of holding pieces for small (ver cal) coils. . . 118
5.4.14Posi oning of small (ver cal) coils. . . 118
5.4.15Inclusion of lower holding pieces for medium (ver cal) coils. . . 118
5.4.16Inclusion of lower holding pieces for small inner coils. . . 119
5.4.17Inclusion of lower holding pieces for ver cal outer coils. . . 119
5.4.18Comple on of posi oning of the twelve different coils. . . 120
5.4.19Inclusion of triangular gussets at lower corners. . . 121
5.4.20Addi on of triangular gussets at upper corners. . . 121
6.3.1 Varia on in the magne c field during opera on of the x-axis coils. . . 127
6.3.2 Selected loca ons to es mate devia on of the generated magne c field. 128 6.3.3 Devia on of the generated magne c field at selected loca ons. . . 129
6.3.4 Devia on of the generated magne c field at selected loca ons. . . 130
6.3.5 Devia on of the generated magne c field at selected loca ons. . . 131
7.1.1 Two implementa ons to increase DOF (fric onless rota ons). . . 133
C.1.1 Devia on of magne c field ⃗B in x-axis direc on. . . 149
C.1.2 Devia on of magne c field ⃗B in z-axis direc on. . . 149
C.1.3 Effec ve volume side length def f for d = 1.5 m. . . 150
C.2.1 Devia on of magne c field ⃗B in x-axis direc on. . . 150
Test bench for Nanosatellite ADCS devices Álvaro Cano Torres
C.2.2 Devia on of magne c field ⃗B in z-axis direc on. . . 151
C.2.3 Effec ve volume side length def f for d = 1.5 m. . . 151
C.2.4 Devia on of magne c field ⃗B in x-axis, inaccuracies of 5 mm. . . 152
C.2.5 Devia on of magne c field ⃗B in z-axis, for inaccuracies of 5 mm. . . 153
C.2.6 Devia on of magne c field ⃗B in x-axis, for inaccuracies of 10 mm. . . 153
C.2.7 Devia on of magne c field ⃗B in z-axis, for inaccuracies of 10 mm. . . 154
C.3.1 Devia on of magne c field ⃗B in x-axis direc on. . . 154
C.3.2 Devia on of magne c field ⃗B in z-axis direc on. . . 155
C.3.3 Effec ve volume side length def f for d = 1.5 m. . . 155
C.4.1 Devia on of magne c field ⃗B in x-axis direc on. . . 156
C.4.2 Devia on of magne c field ⃗B in z-axis direc on. . . 156
C.4.3 Effec ve volume side length def f for d = 1.5 m. . . 157
C.5.1 Devia on of magne c field ⃗B in x-axis direc on. . . 158
C.5.2 Devia on of magne c field ⃗B in z-axis direc on]. . . 158
C.5.3 Effec ve volume side length def f for d = 1.5 m. . . 159
C.6.1 Devia on of magne c field ⃗B in x-axis direc on. . . 159
C.6.2 Devia on of magne c field ⃗B in z-axis direc on. . . 160
C.6.3 Effec ve volume side length def f for d = 1.5 m. . . 160
D.1.1 Determina on of c.o.m. for U-shaped Profiles. . . 163
D.1.2 Determina on of c.o.m. for H-shaped Profiles. . . 165
D.2.1 Diagram Case 1 of different possibili es to hang coils. . . 166
D.2.2 Equilibrium of forces in Bending for Case 1. . . 167
D.2.3 Diagram Case 1 of different possibili es to hang coils. . . 169
D.2.4 Equilibrium of forces in Bending for Case 2. . . 170
D.2.5 Equilibrium of forces in Buckling for Case 1. . . 172
F.4.1 SPI Communica on Protocol block diagram. . . 178
F.4.2 SPI Communica on over LVDS (RS-422), two wires per channel. . . 179
F.4.3 Basic LVDS circuit opera on. . . 179
F.4.4 SPI Communica on over LVDS (RS-485), three wires per channel. . . 180
F.4.5 I2C (or TWI) Communica on Protocol block diagram. . . 180
F.4.6 I2C Communica on Protocol slave ming diagram. . . 181
F.5.1 Measured EMF by PNI RM3100#1 without perturba ons around. . . 182
F.5.2 Measured EMF by PNI RM3100#2 without perturba ons around. . . 182
F.5.3 Measured EMF by PNI RM3100#1 with perturba ons around. . . 183
F.5.4 Measured EMF by PNI RM3100#2 with perturba ons around. . . 183
F.6.1 Setup of experiments with Magnetometers PNI RM3100. . . 185
G.0.1 Simula on of EMF in Hiber-2’s orbit, with STK. . . 188
G.0.2 Simula on of EMF of 500 km al tude, 25◦inclina on orbit, with STK. . . . 188
G.0.3 Simula on of EMF in Hiber-2’s orbit, with Tudat, MATLAB. . . 189
G.0.4 Simula on of EMF of 500 km, 25◦orbit, with Tudat, MATLAB. . . 190
G.0.5 Resultant of EMF for orbits 350− 1000 km, 0 − 90◦. . . 191
H.0.1 Magne c field values, under nominal and (de)tumbling condi ons. . . 193
H.0.2 PSD of the induced electric signal for nominal condi ons simula on. . . . 193
H.0.3 PSD of the electric signal under (de)tumbling condi ons. . . 194
Test bench for Nanosatellite ADCS devices Álvaro Cano Torres
List of Tables
3.2.1 Comparison of exis ng Helmholtz Cages. . . 14
3.2.2 Helmholtz cage cost es mate. . . 14
3.2.3 Decision matrix for EMF problem. . . 16
3.3.1 Wire suspension es mated cost. . . 18
3.3.2 Hemispherical Air bearing cost es mate, in tabletop configura on. . . 23
3.3.3 Quota ons of air bearing rotary tables. . . 27
3.3.4 Decision matrix for the fric onless rota on problem. . . 29
4.1.1 Poten al candidates to solve the telemetry issue. . . 33
4.1.2 Poten al COTS Rotary table mee ng the requirements. . . 38
5.1.1 Alterna ves for number of turns in Merri Cage. . . 47
5.1.2 Electric specifica ons for 25/15 Merri configura on. . . 49
5.1.3 Electric specifica ons for 85/36 Merri configura on. . . 49
5.2.1 Summary of effec ve volume for different cage configura on. . . 51
5.3.1 Available T-slo ed profiles offered by local suppliers. . . 82
5.3.2 Different supplier for the U-shaped profiles. . . 86
5.3.3 Gauge possibili es and available local suppliers. . . 102
5.3.4 Specifica on of coils to be controlled by the cage controller. . . 105
5.3.5 Averaged electric specifica ons for the controller defini on. . . 107
5.4.1 Bill of materials for the final design itera on. . . 109
6.1.1 Resistance of each coil before and a er their connec on to the driver. . . 124
A.0.1 Distribu on of Periods along academic year 2018-2019. . . 143
A.0.2 Weekly hours devoted to M.Sc. Thesis during first period. . . 143
A.0.3 Weekly hours devoted to M.Sc. Thesis during second period. . . 143
A.0.4 Weekly hours devoted to M.Sc. Thesis during third period. . . 143
A.0.5 Weekly hours devoted to M.Sc. Thesis during fourth period. . . 144
A.0.6 Logbook hours devoted to this project along academic year 2018-2019. . 145
B.0.1 Wire configura on of different Helmholtz cages . . . 147
C.7.1 Generic Helmholtz Cages specifica ons . . . 161
E.0.1 Material compa bility for 3D printer Ul maker Original. . . 174
E.0.2 Differences between ABS and PLA. . . 175
E.0.3 Specifica ons of Ul maker Original. . . 175
F.2.1 Specifica ons of sensors RM3100 and LSM303DHDL . . . 177
F.4.1 Slave address for RM3100 depending on pins 3 and 28. . . 181
G.0.1 CubeSat Hiber-2 used as orbit design. . . 187
Test bench for Nanosatellite ADCS devices Álvaro Cano Torres
List of acronyms
ABS Acrylonitrile Butadiene Styrene
ADCS A tude Determina on and Control System AWG American Wire Gauge
BOM Bill of Materials
CCA Copper Clad Aluminium CNC Computer Numerical Control c.o.m Centre of mass
COTS Commercial off-the-shelf DOF Degree(s) of Freedom EMF Earth’s Magne c Field FDM Fused Deposi on Modelling FEM Finite Element Method I2C Inter-Integrated Circuit
IGRF Interna onal Geomagne c Reference Field IMU Iner al Measurement Unit
LEO Low Earth Orbit
LVDS Low-voltage Differen al Signaling MDF Medium Density Fibreboard MISO Master Input Slave Output MOSI Master Output Slave Input PCB Printed Circuit Board PEEK Poly-Ether-Ketone PLA Polylac c Acid
PMW Pulse Width Modula on POM Poly-Oxy-Methylene PSD Power Spectral Density PVC Polyvinyl Chloride SCL Serial Clock Line SCLK Serial Clock SDA Serial Data Line
SEET Space Environment and Effects Tool SLS Selec ve Laser Sintering
SPI Serial Peripheral Interface SS Slave Select
TWI Two-Wire Interface WMM World Magne c Model
Introduc on
Every satellite is equipped with an A tude Determina on and Control System (ADCS), which measures, es mates and controls its a tude thanks to its sensors (Sun sensors, star trackers, magnetometers, etc.) and actuators (reac on wheels, magnetorquers, etc.). Consequently, any ADCS, similarly to any other component and system, requires valida on and tes ng to prove its operability and verify so ware compa bility with hardware and other subsystems before its opera on on-orbit.
The efforts required to set up nanosatellite ADCS ground tes ng are generally comparable to the efforts needed for building a CubeSat. Every element of the system can be individually verified before flight and computer simula ons may be performed, but only the opera on in space environment shows if all subsystems work together correctly.
The present project has been funded by the space company Hyperion Technologies B.V.1 and developed at its facili es in Del (The Netherlands).
1.1 Problem statement
Despite CubeSats are rela vely inexpensive, one failure in orbit is extremely expensive in terms of cost and me due to payload prepara on and launch. The necessity of ground tes ng system for nanosatellite-class ADCS is clear in order to avoid those mission failures. Nevertheless, there is no clear consensus about which ground tests are more
1Hyperion or the Company are some mes used along this report to refer to Hyperion Technologies B.V.
Test bench for Nanosatellite ADCS devices Álvaro Cano Torres
efficient if me, cost, accurate results and similarity with real condi ons are considered into the op miza on problem.
The ideal test bench would be the one that perfectly simulates the space environment and all its factors such as weightlessness, EMF, vacuum, radia on, plasma and neutral par cles, among others. However, the scope of this project focuses on the two first factors: Weightlessness and EMF. More specifically, it is limited to the imita on of the EMF on-orbit since the off-loading system (to mimic weightlessness) was analysed and preliminarily designed but not manufactured or implemented. Vacuum condi on simula ons are already executed by an external supplier company. On the other hand, neutral par cles, plasma and radia on go beyond the scope of this project.
1.2 Methodology
Along this Chapter 1, the general problem is described and the scope of this M. Sc.
Thesis is defined. Star ng with a market research of current and available solu ons, it is determined which of those proposals can be be er adapted to the ADCS-type devices that Hyperion has developed. Different aspects are analysed such as quality of the results or me and cost of manufacturing in an a empt to determine which of those test benches is op mal for the specific device. Versa lity is considered as an asset so that the bench can be modified and adapted to a new and different device. New and disrup ve concepts have been also considered and its feasibility is assessed.
Ini ally, this test bench will be used to test Spacecra sensors (mainly gyroscopes and magnetometers) and actuators (reac on wheels and magnetorquers). The ground test sta on was intended to solve fric onless mo on (rota ons), as one possible solu on for weightlessness, and EMF in space. For the first problem, wire suspension, air bearing or magne c levita on, among others, were taken into considera on. For the la er, different Helmholtz cage configura ons were the candidates to face the genera on of similar space condi ons.
The project was also conceived to ease the adapta on and upgrade in order to test Sun sensors and Star trackers, also manufactured by Hyperion, as part of future work (not included in this document).
A more quan ta ve defini on of the objec ves is listed in Chapter 2. Before any design, a
Test Bench Design Technology
Selection
Manufacturing
Final Test
Subsystem Test
Validation
Detailed Model Initial
Model
Req.
check
Specific Test
failed passed
passed
failed passed
failed
Req.
check Requirements
Req.
check
Figure 1.2.1: V-diagram summarising Project Milestones.
trade-off between market solu ons and different alterna ves, together with the decision criteria to follow, are analysed throughout Chapter 3. Once the feasible solu ons have been considered, a customised test bench is designed in Chapters 4 and 5 for the two main space environment features: weightlessness and EMF. Performance of the implemented test bench (Merri Cage to address the EMF) is described along Chapter 6. Possible upgrades of the test bench as well as main conclusions and lessons learnt are stated in the final Chapter 7. A brief summary of the process followed throughout these ten months, described in the remaining chapters, is expressed in Figure 1.2.1.
In this report, different quan ta ve as well as qualita ve methods, acquired from the M.
Sc. Courses taken at KTH Stockholm and TU Del , have been used in order to reach the final goal of this M. Sc. Thesis, an efficient test bench for nanosatellite ADCS.
1.2.1 Deliverables
The ul mate objec ve is to have a fully opera onal test bench, adaptable to different devices and able to simulate several scenarios present in space and during on-orbit opera ons of the Nanosatellite, only around Earth. Detumbling condi ons are discussed a erwards but it is not one of the design requirements for the test bench.
Test bench for Nanosatellite ADCS devices Álvaro Cano Torres
1.2.2 Time plan
The me span for the whole project was the Academic year 2018 – 2019. At the end of it, this project will be presented at KTH Stockholm to be en tled to obtain the Master’s Degree in Aerospace Engineering. The project dura on, equivalent to a 30 ECTS Master Thesis, must be 30 ECTS× 25 h/ECTS = 750 h, at least.
A calendar with the most suitable working hours for the student is a ached in Appendix A. Apart from the present project, the student, Álvaro Cano Torres, had to take courses with a workload of 30 ECTS at the Technical University TU Del (Netherlands) as an indispensable part of the Erasmus Exchange he was en tled to. Those courses were not evenly divided along the academic year.
The deliverables and main milestones of the project were:
1.- Project Plan.
Submission of Project Plan: End of first week of September 2018.
2.- Defini on of requirements and 3.- Market research, about the different alterna ves of the test bench.
Submission of Defini on of requirements and Market research: End of September 2018.
4.- Selec on among available and affordable technologies for the Test bench.
Decision about which type of Test bench will be developed: End of October 2018.
5.- Design of the Test bench.
Submission of final design itera on with all the required modifica ons to be adapted for Hyperion’s devices, including final choice of COTS subsystems: End of December 2018.
6.- Purchase of necessary components and subsystems to build the Test bench.
Start the construc on and assembly of the Test bench for ADCS devices: Mid of March 2019 (if all components are in-house).
7.- Construc on and manufacturing of the Test bench.
Assembly of the full Test bench for ADCS devices: End of May 2019.
8.- Test phase of Hyperion’s devices.
Successful tests using the final Test bench: End of June 2019.
Chapter 2
Defini on of requirements
The aim of the present chapter is the defini on of requirements which were used as the basis for the design and manufacture of the test bench. ADCS sensi vity and actuators capabili es, both developed by Hyperion, were taken into considera on.
2.1 System Requirements
Req. 1 The test bench must be able to allocate a complete 6U CubeSat, extendable to 12U CubeSat (with a maximum weight of 12 kg [17]). Satellite Solar Panels are assumed stowed.
Req. 2 The test bench must be able to simulate similar space environment condi ons on ground. In this par cular case, features such as fric onless rota on and EMF, similar to the one that the CubeSat would experiment on-orbit, define the baseline for the test bench design.
Req. 3 The test bench should be easily adaptable (such as movable coils and possibility to add addi onal gadgets) to future CubeSats designs (different shapes and/or volumes) and ADCS devices.
Req. 4 The test bench has to operate autonomously for the dura on of, at least, one full revolu on of the design orbit, assumed Low Earth Orbit (LEO).
Req. 5 Desirably, the test bench shall be easily disassembled in case it needs to be stored or moved to a different loca on.
2.2 Func onal Requirements
2.2.1 Earth’s Magne c Field problem
The EMF in space is o en reproduced with a Helmholtz cage. Any requirement referred to this subsystem uses the term “cage”.
Req. 6 The cage shall generate magne c field to dynamically test the magnetometers and magnetorquers used at Hyperion, being detectable by these devices.
Req. 7 The cage must generate a uniform magne c field within the design volume requirement (a.k.a. effec ve volume), i.e. 36× 24 × 12 cm, 6U CubeSat maximum dimensions [17]. Uniformity levels are defined as±1% accuracy in the cage centre and±7% in the effec ve volume [35].
Req. 8 The cage must be able to cancel the ambient magne c field out, in the range from
±25 µT to ±45 µT, along each spa al axis [1].
Req. 9 The cage must be able to simulate a dynamic on-orbit geomagne c field similar to a possible future mission [40]. Values up to±60 µT around each axis are expected.
Req. 10 Desirably, the cage shall be able to simulate EMF varia ons under (de)tumbling condi ons of the Satellite. Rates may go up to±180 deg/s.
2.2.2 Fric onless rota on problem
The op mum solu on for the off-loading system was not selected finally in this document and hence, any requirement referred to this subsystem uses the term “system”. The following requirements are included as a star ng point for future upgrades of the test bench.
Req. 11 The system should be lightweight and equipped with caster wheels (or similar) to be easily moved by only one person.
Req. 12 The system must be made of non-ferromagne c material so it does not interfere with the magne c field produced by the other subsystem, i.e. the cage.
Req. 13 The system shall mimic weightlessness and thus must provide nearly fric onless rota on in, at least, 1 Degree(s) of Freedom (DOF) (desirably 3) in order to test
Test bench for Nanosatellite ADCS devices Álvaro Cano Torres
the actuators (reac on wheels and magnetorquers) of the system. Weightlessness condi on is achieved when disturbance torques around all three axes have an order of magnitude of 10−5Nm (a.k.a. near torque-free) [35]. These DOF are elaborated in the following two requirements.
Req. 14 The system must able to provide unconstrained rota on (±180◦) in one of the rota on axis (usually ver cal axis, Z – Yaw).
Req. 15 Desirably, the system should provide freedom in the other two rota on axes (usually horizontal axes, X – Roll and Y – Pitch). These rota on movements may be constrained but not less than±30◦[35]. It is assumed that Centre of mass (c.o.m.) and centre of rota on of the satellite are properly aligned.
Req. 16 The system must be equipped with telemetry to accurately measure the a tude externally (feedback) to be compared with the ones provided by the internal devices. The precision for which Hyperion Technologies B.V. is aiming has an order of magnitude of 1 arc sec. This high accuracy would be useful for future poin ng payload-related requirements, such as high precision cameras, trackers or laser communica on.
Req. 17 The system must be able to wirelessly communicate between the laboratory computer and the ADCS device to send and receive data in real- me, if required.
Consequently, ba eries needs to be opera onal during tes ng ac vi es since Solar Panels would not be deployed.
Req. 18 Desirably, the system shall tolerate ini al disturbances (for instance, induced by hand) to simulate on-orbit a tude irregulari es or, if possible, detumbling condi ons.
Req. 19 Desirably, the moment of iner a of the system shall be as low as possible to minimally modify the total moment of iner a (off-loading system plus device being tested).
Market Analysis
The test bench shall be able to simulate the magne c field, characteris c of the space environment. Sun and celes al sphere were omi ed and postponed for future work.
Even though the off-loading solu on was also not manufactured at the end, it is analysed and preliminarily designed in the present and following chapters, as star ng point for future development of the test bench (Req. 2).
During this sec on, state-of-the-art technologies, available and suitable for this type of test bench are analysed and assessed in order to determine which could be the op mal alterna ve for this par cular case. Disrup ve technologies not being implemented yet have been also studied. Nevertheless, feasibility was one of the main criteria to judge a technology.
3.1 Decision criteria
The different perspec ves that have been taken into considera on and analysed in detail to make the final decision are presented here:
Feasibility: Technology which has been already implemented, proven and results have been successful (ease of implementa on). This way, the risk was reduced as much as possible.
Accuracy: Technology achieving necessary accuracy. Quality standards have to be similar to the Company’s.
Test bench for Nanosatellite ADCS devices Álvaro Cano Torres
Manufacturability: Technology which could be manufactured and assembled by the Company itself, except for some Commercial off-the-shelf (COTS) subsystems which may be purchased directly in order to save me and resources in the cases where Hyperion had no experience in their design and/or development.
Versa lity: Technology which has ease of modifica on, scalability and flexibility (Req. 3).
Cost: Technology whose price is affordable by the Company, with high price- performance ra o.
In order to objec vely evaluate different op ons, a decision matrix is used to systema cally rate the performance between sets of values and informa on. Decision criteria previously listed are graded from 1 to 5, where 1 means absolutely unimportant and 5 means very important in the final assessment.
3.2 Earth’s Magne c Field problem
Solu ons to the EMF problem already implemented are shown, explained and analysed according to the defined criteria.
3.2.1 Helmholtz Cage
A Helmholtz cage is basically a set of coils aiming to produce a region of nearly uniform magne c field, usually one pair, parallel to each other, along each of the three spa al axes. These coils, named a er the German physicist Hermann von Helmholtz, consist of mul ple windings1usually in a circular or square shape. Biot-Savart Law (described later in Sec on 5.2) states that a wire carrying a current produces a magne c field around it (Req. 6). With the help of a good controller, ambient EMF can be cancelled out (Req. 8) and realis c simula ons of the dynamic magne c field on-orbit can be achieved (Req. 9).
In order to prevent interac on of any cage component with the generated magne c field, the system must be made of plas cs, aluminium, stainless steel (only non-ferromagne c types, such as A2 and A4) or any non-ferromagne c material (Req. 10).
There are some different configura ons concerning number and shape of the coils. Main
1Windings and turns are used interchangeably along this report.
effects are the uniformity of the resultant magne c field and size and/or shape of the effec ve volume2. In order to know more about the real effects of these combina ons, real examples of Helmholtz cages, managed by public organisa ons or private companies, are analysed in this sec on.
It should be noted that uniformity of the magne c field is defined by the difference in percentage between the magne c field magnitude at furthest corner of the effec ve volume, where uniformity is the worst, and the central point of the cage.
Square-shaped coils
The main advantage of this shape is that they are easier to manufacture and hold than circular coils. Effec ve volume is usually larger than circular-shape configura ons [26] but the magne c field does not achieve those levels of uniformity, depending on the setup (2, 3, 4, or 5 coils per axis). This aspect is explained in depth along Simula on Sec on 5.2.
An implementa on of this configura on is shown in Figure 3.2.1, with two square coils per axis.
Merri Four-Coil A par cular case with four square coils per axis, rather than only two, is known as Merri Cage. Some examples of this configura on are the cages at MIT Space Systems Laboratory [31] (see Figure 3.2.2) and at Nurmijärvi Geophysical Observatory [29], among others. According to [23], Merri ’s design op mizes the area of uniform magne c field in one dimension, yielding the largest volume of uniform field. This is beneficial since versa lity (variety of devices to be tested, (Req. 3)) increases. With respect to the common configura on, only two square coils per axis, the generated magne c field is also more uniform. All previous facts, extracted from literature, were proven and contained in Sec on 5.2.
On the other hand, design and manufacture might be more complex. More extensive analysis and be er cost es ma on are carried out along the upcoming Chapter 5.
2Effec ve volume is defined as the design region inside the full size cage where the magne c field is considered uniform (in this case, 1 % in the cage centre and 7 % in the cube of roughly 400 mm side-length).
Test bench for Nanosatellite ADCS devices Álvaro Cano Torres
Figure 3.2.1: Helmholtz Cage at Georgia Ins tute of Technology.
Figure 3.2.2: Helmholtz Cage at Massachuse s Ins tute of Technology.
Figure 3.2.3: Helmholtz Cage with Sun simulator at DLR Berlin.
Circular-shaped coils
Circular coils with similar dimensions produce smaller homogeneous field than square coils, but uniformity levels reached are greater. Manufacturability and mooring techniques (to hold the coils at the right posi on) may become complicated, as observed in Figure 3.2.3, with two circular coils per axis.
Testbeds
Mul ple Helmholtz Cage designs were found at different organisa ons, e.g. TU Del , Air Force Ins tute of Technology, University of Michigan, among others. Main features are listed in Table 3.2.1 to compare real specifica ons from different cage configura ons and be able to be er assess them later on.
The minimum effec ve volume required for this par cular case is 0.064 (0.43)m3(Req.
1, rounded up). This implies that COTS from MEDA together with the Helmholtz cages in CSSWE Colorado or LTU Luleå may not be a good example given that volume requirement and uniformity are not achieved.
Test bench for Nanosatellite ADCS devices Álvaro Cano Torres
Table 3.2.1: Comparison of exis ng Helmholtz Cages.
Ins tu on
or organiza on Design Dimensions
[m]
Capability [µT]
Effec ve
volume [m3] Uniformity TU
Del [30]
2 square coils
per axis 2× 2 × 2 (modifiable) ±750
1 (13) 0.512 (0.83) 0.125 (0.53)
25.04%
10.61%
1.69%
AFIT Ohio [6]
2 square coils
per axis 2.4× 2.4 × 2.4 - 0.512 (0.83) 2.5% (es ma on)
LTU Luleå [4]
2 square coils
per axis 2× 2 × 2 ±120 0.064 (0.43)
0.0008 (0.23) 10%
3.7%
AGH Krakow [12]
2 square coils
per axis 1.5× 1.5 × 1.5 - 0.0008 (0.23) 0.7%
DLR Berlin [35]
2 circular coils
per axis 2.1× 2.1 × 2.1 ±210 0.729 (0.93)
0.216 (0.63) 7%
1%
MIT
Cambridge [32] 4 square coils per axis 1.5× 1.5 × 1.5 - 0.343 (0.73) 0.125 (0.53)
2%
0.1%
CSSWE Colorado [15]
2 square coils
per axis 0.6× 0.6 × 0.6 - 0.027 (0.33) 1%
MEDA HCS-01 (COTS) [18]
2 square coils
per axis 2× 2 × 2 ±200 0.044 (0.353)
0.0008 (0.23) 0.1%
0.03%
Bar ngton HC2 (COTS) [21]
2 square coils
per axis 1.3× 1.2 × 1.3 - 0.014 (0.243)
0.0025 (0.143) 1%
0.1%
Table 3.2.2: Helmholtz cage cost es mate.
Part Quan ty Cost (€)
Structure 1 800
Acrylic walls (if desired) 4 500 12 AWG wire (speaker) 1400 m 500
Power Supply, Controller 1 500 ÷ 2000
Magnetometer 1 100 ÷ 1000
Microcontroller 1 50
Other 1 300
Total 2350 ÷ 5750 €
For further details about wiring configura on of previous test benches, it should be referred to Appendix B.
Budget
Cost has not been stated in Table 3.2.1 due to the lack of informa on for all the compared cages. Nevertheless, a es mate from the collected data has been elaborated in Table 3.2.2.
Commercial cages (COTS) directly purchased to suppliers have much higher sale prices.
For instance, the simplest and cheapest model that the company MEDA offers, HCS-01, has an ini al cost of 78, 750 $, without adding shipping and calibra on labour hours. On
the other hand, Helmholtz Cage from Bar ngton HC2, whose effec ve volume would be ght for a 6U CubeSat (Req. 1), has a price of 46, 670 $. It is clear that performance and uniformity of the magne c field are cer fied but the ques on here is whether it is worth the price. Furthermore, Helmholtz cages achieving current requirements have been designed and manufactured by Universi es and/or public organisa ons themselves while saving money.
COTS subsystems such us power supplies (o en computer-controlled) or high-precision magnetometers are usually the vast majority of the total hardware cost of the Helmholtz cages being studied. To bear with it, devices that the Company already owned were reused if specifica ons met the requirements.
3.2.2 Alterna ve approaches
Even though magne c fields can be generated by different methods such as permanent magnets or thermomagne sm, those alterna ves would not allow to effec vely control the generated magne c field as required to imitate the EMF on-orbit. That is the reason why they were not analysed in this report. An exhaus ve research was carried out in order to find alterna ves to the Helmholtz cage but only less efficient and/or unfeasible solu ons were found.
Cage configura ons with three square coils per axis (a.k.a. Merri 3) are also possible but real implementa ons were not found. However, as derived from [16], its performance is worse than that of previous cases. There is another possible configura on with five square coils per axis (a.k.a. Ruben 5), whose design would become even more complex while performance is not significantly improved [24]. Both setups have not been considered due to the stated reasons.
Permanent magnets have been assessed in the Matrix decision with the only purpose of comparing them with the Helmholtz cage solu on, but it must be understood as an unfeasible method.
3.2.3 Conclusion
As men oned earlier, each decision criterium varies from 1 to 5. If an alterna ve is graded with a value of 5 for the Cost, it means that it is much cheaper than the one rated with
Test bench for Nanosatellite ADCS devices Álvaro Cano Torres
Table 3.2.3: Decision matrix for EMF problem.
Criteria (Ra ng) Feasibility (5) Accuracy (4) Manufacturability (3) Versa lity (2) Cost (3) Total
2 square coils 5 3 5 4 5 75
Helmholtz Cage 4 square coils 5 5 4 5 4 79
2 circular coils 5 4 3 4 3 67
Magnet Permanents 2 1 2 3 2 32
a value of 1. Values assigned to Cost are not directly propor onal to price of purchase and/or manufacture itself, rather the opposite.
The decision matrix, Table 3.2.3, shows that permanent magnets can be totally discarded in the presence of the Helmholtz cage alterna ve, as derived from its poor rate. Secondly, a design made of square coils, instead of circular coils, is probably the most convenient op on since manufacturability is simpler while uniformity (considered in the Accuracy column) does not decrease significantly, only in very confined regions around the central point [16]. Nevertheless, final decision is s ll to be taken since results for two square coils per axis or four coils are sufficiently close.
3.3 Fric onless rota on problem
Apart from drop towers or parabolic flights, that are very expensive and only provide a limited number of seconds (maximum a couple of minutes) of weightlessness, there are some other strategies to simulate fric onless rota ons, such as off-loading systems.
Solu ons allowing up to six DOF (three transla ons and three rota ons) exist but they are not analysed since only free rota ons are sought (Req. 13).
In order to be suitable with the proposed solu ons for the EMF problem, the Helmholtz cage, any component forming the system to solve the fric onless rota on problem must be made of a non-ferrous material so they do not interfere with the magne c field produced by the Helmholtz cage (or any of its configura ons) (Req. 12).
Furthermore, telemetry providing accurate angular posi on in real- me (desirable order up to 1 arc sec level) would be necessary in order to compare outcome from CubeSat’s sensors with an external feedback measurement (Req. 16).
Alterna ves to this problem are shown, explained and analysed here, according to the
Figure 3.3.1: Wire suspended nanosatellite inside Helmholtz cage concept.
defined criteria.
3.3.1 Wire suspension
This would be the simplest solu on where a wire or string is linking the satellite (or a stack of different subsystems) through its centre of gravity to the test bench, usually called piñata. It provides nearly fric onless rota on around one single axis (gravity direc on), under condi on that the angular displacement is small. If not, the wire might introduce a torque if twisted too many mes. Important to remember is that, as it is happening with the magne c levita on explained later, the solu on only allows one DOF (Req. 14).
In addi on, transla ons (string not aligned with the gravity vector) might occur during opera on that need to be prevented for valid tes ng outcomes.
A real implementa on is used by the ExoCube satellite (developed by Cal Poly), based in California, which uses a monofilament string with tensile rated to hold up to 22 kg with a hook on the end to hold the test ar cle [32]. The concept in summarised in Figure 3.3.1.
Test bench for Nanosatellite ADCS devices Álvaro Cano Torres
Table 3.3.1: Wire suspension es mated cost.
Part Quan ty Cost [€]
Harness device (to hold satellite) 1 10
Monofilament string 1m 10
Telemetry System 1 100÷ 500
Total 120÷ 520 €
Budget
Cost es mate for wire suspension alterna ve is analysed and tabulated in Table 3.3.1, following the same methodology as in the preceding Sec on 3.2.
3.3.2 Air bearing
Being a good trade-off between performance and cost, air bearings are a non-contac ng system where a pressurized gas film, typically air, is pumped between the two surfaces in rela ve mo on where it acts as the lubricant. Air bearings have been used for spacecra a tude determina on and control hardware verifica on and so ware development for nearly 45 years, virtually coincident with the beginnings of the space race.
In cases where iner a of the air bearing part a ached to component is significant, so ware and/or hardware modifica ons should be implemented since system response would be slower than during nominal opera on. Complex solu ons imply the use of motors or reac on wheels to compensate the addi onal moment of iner a of that upper part. Further explana on and examples are given along Gimbal suspension, Sec on 3.3.3.
The non-contac ng surface could be planar, allowing near-fric onless transla ons along the two horizontal axes and rota on around the ver cal axis; or spherical, allowing near- fric onless rota on (only) around the three axes. Important to keep in mind is that these transla ons and rota ons are usually restricted to certain range, e.g. end of the horizontal table or hemisphere. A good air bearing surface should achieve roughness level around 1 µm, which makes manufacturing process very costly and elaborate [8].
COTS systems are also analysed along the present sec on.
Some mes, air bearing experiments are run in vacuum, so that the viscosity of the air can be also eliminated. However, in this par cular case, vacuum chamber is not available
Figure 3.3.2: Planar air bearing. Courtesy: ESA, ESTEC
so air resistance is a drawback to bear with, not very significant though.
Planar
The typical example would be an air-hockey table, which provides one rota onal and two transla onal DOF. Planar mo on is of interest for simula ons of rendezvous and docking. In almost every case, the test body carries its own air supply and produces its own cushion of air, allowing it to hover on a polished surface, unlike the air-hockey table where air film is coming out from the planar surface itself.
ESA (with its planar air-bearing microgravity simulator at ESTEC, shown in Figure 3.3.2), NASA, Massachuse s Ins tute of Technology, Stanford University, Tokyo Ins tute of Technology, among many others ins tu ons, have implementa ons of these air bearing benches.
However, since the requirements demand free rota ons and no par cular interest on transla ons is shown, this strategy was thus omi ed.
Hemispherical
It is the most widely used type of air bearing. It is achieved by a aching a hemisphere in a cup (a.k.a. base, spherical shape also) with a constant supply of flow, shown in Figure 3.3.3. Because of the geometrical constrains of such design, rota onal freedom is limited
Test bench for Nanosatellite ADCS devices Álvaro Cano Torres
Figure 3.3.3: COTS Hemispherical air bearing. Courtesy: PI
to roughly±45◦while having full freedom of spin (yaw, ver cal axis). These restric ons have to be considered since they may affect the efficiency of the testbed and the reliability of the ADCS ground tests.
One of the challenges of using an spherical air bearing setup is the extreme precision required when balancing the test ar cles. Given that the hemisphere operates in free float, it is cri cal that the c.o.m. of the device to test is co-located with the centre of rota on. If the c.o.m. is above the centre of rota on, the system would be unstable (roll and pitch angles probably restricted to less than±30◦). If the centre of mass is below, the system would be pendulum stable (higher roll and pitch angles than preceding case). The op mal situa on is when both are coincident since the system would stay at any desired orienta on and respond like it would be in a microgravity environment.
In cases where this condi on is not easily achieved, i.e. c.o.m. of the system (nanosatellite plus upper part of air bearing, hemisphere) and centre of rota on of the air bearing are not coincident, a mass balancing system would be necessary. These systems are based on three independently controlled3masses designed to move parallel to each of the three principal axes of the system.
Instead of using a hemisphere (a.k.a. tabletop), there are two other types of air bearing
3Usually with stepper motors
Figure 3.3.4: Concept of dumbbell (le ) and umbrella (right) configura on.
tables which are umbrella and dumbbell (see Figure 3.3.4). Tabletop and umbrella have pitch and roll constrained to less than±90 degrees while full rota on in yaw axis. With the dumbbell table, free mo on in yaw and roll can be achieved but constrained in pitch.
Since dumbbell has two self-balanced ends, tes ng of separated components is more convenient than tes ng of full satellite.
Dumbbell concept is be er for tes ng of individual components and so is umbrella configura on. This design keeps the centre of mass of the system very near to the centre of rota on by suspending components very carefully below the umbrella, which most of the cases become extremely sensi ve and tricky. This is the reason why these two setups are not taken into account in the upcoming comparisons.
Spherical
One of the most ideal situa on is when there are three DOF with unconstrained rota ons (360◦around the three axis) (Req. 14, 15). Spherical air bearings fit this requirement due to the con nuous surface of the ball.
The air-levitated sphere usually contains only a microcontroller, sensors, ba eries and a flywheel, but not the full-scale satellite because the sphere would have to be much bigger and the holding mechanisms to keep the centre of mass of the satellite coincident with the centre of rota on of the sphere would be very challenging. Moreover, having this bulky and heavy sphere would influence in the moment of iner a of the whole system which is always undesirable (Req. 19).
Test bench for Nanosatellite ADCS devices Álvaro Cano Torres
Figure 3.3.5: Experimental spherical hollow air bearing at Cal Tech.
In the case of a 6U (or even 12U) CubeSat (Req. 1), the sphere would be prohibi vely big and heavy.
These may be the reasons why there is li le informa on about real projects or experiments, except for the one in Cal Tech California [36], shown in Figure 3.3.5.
Just as a curiosity, the spherical air bearings in Figure 3.3.5 were part of a test bench allowing six DOF unconstrained via floa ng spherical air bearing (the one described in this sec on) coupled with a fric onless precision floor and a cylindrical li [36]. The main goal of this test facility was to produce real me simula ons of a full mission meline.
Budget
Cost of hemispherical air bearings are analysed and es mated following same methodology as earlier. As men oned before, planar and spherical air bearings do not seem to fulfil as many requirements as the hemispherical one does. That is the reason why the following cost es ma on will only approach the hemispherical air bearing solu on with the tabletop configura on, expressed in Figure 3.3.3.
Ini ally, 3D prin ng was considered as a possibility to manufacture all the parts, mee ng the required flatness for the cup and hemisphere. If not, addi onal process such
Table 3.3.2: Hemispherical Air bearing cost es mate, in tabletop configura on.
Part Quan ty Cost [€]
3D printed Pedestal 1 10
3D printed Cup 1 20
3D printed Hemisphere 1 30
Extra manufacturing processes 1 200
Telemetry system 1 200÷ 1000
Three Stage Filtra on system 1 100÷ 200
Air pump 1 500÷ 2000
Air piping - 50
Total 1110÷ 3510 €
as polishing or grinding could be used to improve the roughness. Some poten al issues are that the bearing ball (upper part of the system) mass should be well- balanced and centred, and calibra on might be hard to perform once the component is manufactured.
Air Bearing manufacturers, such as PI, Nelson Air or New Way, were requested a quota on for their COTS Air bearings. Since the cheapest models have a price star ng from 3000 €, designing and manufacturing in-house, with local suppliers or partners would be the ini al strategy. Nevertheless, if the flatness requirement was not easy to meet with 3D printers or local suppliers, acquiring a hemispherical air bearing would be the second alterna ve.
3.3.3 Gimbal suspension
A system of mul ple gimbals can be used for the same purpose as an air bearing table but such arrangements introduce the problem of gimbal lock. This alterna ve has been analysed and studied more in depth here.
There is s ll the same drawback as with air bearings, angular posi on must be accurately measured and addi onal system ought to be used.
Most of the informa on in this sec on has been extracted from the research that University of Montpellier was doing for a few years ago [13], which proposes two different alterna ves, explained below.
Test bench for Nanosatellite ADCS devices Álvaro Cano Torres
Figure 3.3.6: Spherical air bearing structure held by gimbal suspensions [13].
No compensa on of disturbance forces
The design described in this sec on has the advantage of hollow (spherical) air bearing but without the problema c massive structure. Same spherical surface is formed with only a few small air bearings, which have one common centre of rota on. Figure 3.3.6 shows this concept. That way, all unnecessary segments of the sphere are eliminated and weight is vastly reduced.
Just as spherical air bearings, this design allows unconstrained three DOF (only rota ons).
The external frame with the air bearing pucks has to spin freely and follow inner part with the CubeSat mounted on it. The posi on of these bearing pucks has to be corrected rapidly based on the current posi on of the satellite (op cal sensors may be the most convenient tracking sensor due to its speed and accurate coordinates). The suspension of the external frame is realized by gimbals.
It is easily seen that the moving parts of the structure are much smaller and lighter than that of the full sphere. Size and mass of this part is cri cal because it leads to unwanted moment of iner a. With this solu on, although it yields addi onal mass, the external part of the structure is independent from the CubeSat mounted on the test bench and therefore makes no influence on the test results.
Nevertheless, some drawbacks are important to consider and listed below:
Versa lity (Req. 3) is compromised with the gimbal suspension solu on since modifica ons of the effec ve volume are extremely complex, implying a totally
new structure for different sizes of satellites.
Precise balancing of the CubeSat inside the holding frame is required so that geometrical centre of the inner sphere (centre of rota on) is perfectly coincident with the c.o.m. of the nanosatellite.
Gimbal structure must accomplish different analyses and tests first, such as s ffness, strength and thermal stability, to be approved and validated in terms of structural demands.
Feasibility might be the most obvious issue since it has never been implemented before.
In addi on, it should be noted that this type of gimbal suspension is s ll making use of air bearings providing contactless rota ons, adding their drawbacks to the list above.
Ac ve compensa on of disturbance forces
Another gimbal suspension solu on, simpler structurally speaking than the one presented before, is analysed in this sec on and shown in Figure 3.3.7. Final test bench has same capabili es, i.e. unconstrained rota on around the three spa al axes.
Nevertheless, it requires a subsystem, composed by a sensor and a motor, to compensate the moments of iner a of the holding rings (significant when comparing with the satellite’s iner a). Previous described drawbacks are s ll present in this alterna ve except for the simpler structure.
It is important to keep in mind that feasibility and possible manufacturing issues of this configura on has not been solved yet either.
3.3.4 Rotary table
A rotary table is a precision work posi oning device used to accurately turn around a fixed axis (ver cal or horizontal). Rotary tables that would be interes ng for this par cular study are those which make use of air bearings to provide free and smooth rota on around their single axis. Precision telemetry system (goniometer, encoder, or similar) is integrated into the table to provide precise angular posi on feedback.
Only one DOF, rota on about gravity axis, may be too limited for certain applica ons.
Test bench for Nanosatellite ADCS devices Álvaro Cano Torres
Figure 3.3.7: Gimbal suspension (le ) with ac ve compensa on motors (right).
Figure 3.3.8: Air-bearing Direct-drive rotary table. Courtesy: Aerotech.
Nevertheless, this alterna ve is analysed since very high accuracy feedback is achieved, up to arc seconds level (Req. 16). The single DOF disadvantage might be solved by combining two (even three) rotary tables (a.k.a. stages) to increase the DOF although it would be a complex implementa on task since mass of these tables is not negligible and would significantly interfere in the final iner a of the system. In addi on, air piping would be hard to achieve when these tables have different stages. Similarly, mass balancing to get that axis of rota on of each table coincident with the c.o.m. of the satellite would be necessary.
This solu on would be directly purchased as a COTS subsystem since technology to
