Self-Pointing Satellite Testbed
- Role
- Independent project
- Dates
- Sep 2026 – present
- Status
- In progress
- Tools
- MATLAB, Simulink, SolidWorks, ANSYS
- Project
- P01
Abstract
A 0.8 kg, 3U-scale satellite model hangs from a 2 m nylon line and turns itself with a brass reaction wheel, the way a spacecraft points without propellant. The maneuver is a 90° turn in 20 s followed by a ±2° hold for 60 s, judged by a camera reading an ArUco marker on the underside that never feeds the controller.
Hand calculations size the wheel to 4.2× the momentum the turn needs and the frame to a 374 Hz bolted first mode against a 150 Hz requirement, 2.5× the figure asked for. An error budget puts the pointing error of the first design at 2.7°; feedforward for the line's twist and a lower-drift gyro bring it to 1.1°.
Thirteen tests are laid out, each with its target written down before it runs, so the comparison between model and hardware is the result rather than an afterthought.
4.2×wheel momentum margin
374 Hzbolted first mode, 150 Hz required
1.1°pointing error with feedforward
13tests in the plan
P01 · Exploded view
Take it apart
A 0.8 kg satellite model that hangs from a nylon line and turns itself with a brass reaction wheel. Scroll to take it apart.
- The assemblyWhat it is, put together.
- Taken apartKeep scrolling. Every part comes off along its own axis, the way it would on the bench.
- What each part doesClick any label, or any part, to see its name and why it is there.
Files
9 folders · 11 files · 9 open in the browser, nothing downloads. 2 files are not uploaded yet.
Admin
P01 / 00_Admin| Type | File | Rev | Date | Status | Open |
|---|---|---|---|---|---|
| XLSX | Risks scored by likelihood × consequence | C | Nov 2026 | Current | |
| MD | Dated design decisions | B | Nov 2026 | Current |
Requirements
P01 / 01_Requirements| Type | File | Rev | Date | Status | Open |
|---|---|---|---|---|---|
| XLSX | Requirements and verification matrix | B | Nov 2026 | Current |
Analysis
P01 / 02_Analysis| Type | File | Rev | Date | Status | Open |
|---|---|---|---|---|---|
| P01_Gate1_Hand_Calcs.pdfWorked hand calculations | — | — | Planned | ||
| XLSX | Functions × options | A | Nov 2026 | Current |
CAD
P01 / 03_CAD| Type | File | Rev | Date | Status | Open |
|---|---|---|---|---|---|
| DWG | Brass flywheel, Rev A | A | Nov 2026 | Current | |
| MD | Every part in the assembly, its spec and what it is for | A | Nov 2026 | Current |
Simulation
P01 / 04_Simulation| Type | File | Rev | Date | Status | Open |
|---|---|---|---|---|---|
| PNG | 90° slew and 60 s hold from the Gate 1 model | B | Nov 2026 | Current |
Fabrication
P01 / 06_FabricationBuild photos go here once fabrication starts.
Test
P01 / 07_Test| Type | File | Rev | Date | Status | Open |
|---|---|---|---|---|---|
| XLSX | T-01 to T-13, target before the run | A | Nov 2026 | Current |
Report
P01 / 09_Report| Type | File | Rev | Date | Status | Open |
|---|---|---|---|---|---|
| P01_Final_Report.pdfDue Dec 2026 | — | Dec 2026 | Planned |
Procurement
P01 / 11_Procurement| Type | File | Rev | Date | Status | Open |
|---|---|---|---|---|---|
| XLSX | Planned vs ordered vs received | C | Nov 2026 | Current |
A1
| Row | A | B | C | D | E | F | G |
|---|---|---|---|---|---|---|---|
| 1 | ID | Requirement | Method | Calc # | Test ID | Pass criterion | Status |
| 2 | R-01 | Rotation about yaw only; swing and tilt kept small | T, A | #6, #7 | T-09 | Swing amplitude and tilt below the value set by calc #6/#7 | Not started |
| 3 | R-02 | 90° slew in ≤ 20 s, settling within ±2° | T | #1, #11 | T-09 | Camera: settle inside ±2° by t ≤ 20 s, repeated runs | Not started |
| 4 | R-03 | Hold ±2° for 60 s after settling | T | #10, #11 | T-10 | Camera: |error| ≤ 2° for 60 s | Not started |
| 5 | R-04 | Truth sensor independent of the control sensor | I, T | — | T-01 | Camera never in loop; camera error within budget on protractor stage | Not started |
| 6 | R-05 | Wheel capacity ≥ 2× slew momentum at the operating speed limit | A, T | #2, #12 | T-13 | Margin ≥ 2.0 (currently 4.2× at 1,000 rpm) | Not started |
| 7 | R-06 | No wiring across the suspension | I | — | — | Visual inspection | Not started |
| 8 | R-07 | First structural mode ≥ 150 Hz | A, T | #8 | T-07 | FEA (2 codes) and tap test ≥ 150 Hz; FEA within 15% of test | Not started |
| 9 | R-08 | Battery supports a full test sequence on one charge | A, T | #13 | T-12 | Full sequence completed above low-voltage cutoff | Not started |
| 10 | R-09* | Gyro and camera logs share a time base to ≤ 50 ms | D | — | T-01 | Sync event (MCU-triggered LED seen by camera) aligns logs | Not started |
| 11 | R-10* | LiPo handled safely: balance charger, fire-safe bag, low-voltage cutoff in firmware | I | #13 | — | Checklist signed before first powered test | Not started |
| 12 | R-11* | Hardware cost ≤ $270 | A | — | — | BOM actual total ≤ budget | Not started |
| 13 | * Suggested additions — keep, change or delete at Gate 1. |
| Row | A | B | C | D | E | F | G |
|---|---|---|---|---|---|---|---|
| 1 | Item | Description | Part / spec | Qty | Unit low ($) | Unit high ($) | Status |
| 2 | 1 | Gimbal motor + FOC driver board | 2204-class gimbal BLDC; driver per TS-2 | 1 | 25 | 45 | To quote |
| 3 | 2 | Microcontroller + magnetic encoder | ESP32 dev board; encoder per TS-2 | 1 | 15 | 30 | To quote |
| 4 | 3 | IMU breakout | Per TS-3 (published noise density) | 1 | 15 | 25 | To quote |
| 5 | 4 | 3S LiPo (~450 mAh) + balance charger + fire-safe bag | Per calc #13 | 1 | 25 | 45 | To quote |
| 6 | 5 | C360 brass rod, 1-1/2 in | Flywheel + spare for V2 | 1 | 25 | 35 | To quote |
| 7 | 6 | 6061-T6 sheet 1/8 in, 5/16 in square bar, M3 hardware | Plates, rails, fasteners | 1 | 25 | 40 | To quote |
| 8 | 7 | Nylon line, stainless eyelet, marker print | Per TS-1 | 1 | 10 | 20 | To quote |
| 9 | 8 | Camera | Existing phone or webcam | 1 | 0 | 30 | To quote |
| 10 | 9 | Suggested: spare IMU + spare encoder | Covers RK-14 | 1 | To quote | ||
| 11 | 10 | Suggested: Al calibration block for bifilar check (T-02) | Machinable from 6061 bar | 1 | To quote | ||
| 12 | 11 | Suggested: LED + resistor for camera time sync (R-09) | 1 | To quote | |||
| 13 | Subtotal | 140 | 270 | ||||
| 14 | Contingency 15% | 21 | 40.5 | ||||
| 15 | Total with contingency | 161 | 310.5 | ||||
| 16 | Budget (program overview) | 140 | 270 |
| Row | A | B | C | D | E | F | G | H |
|---|---|---|---|---|---|---|---|---|
| 1 | ID | Risk (if… then…) | Category | L | C | Score | Level | Mitigation |
| 2 | RK-01 | If residual gyro bias is a few hundredths of °/s, then integrated drift over 80 s uses most of the ±2° budget | Technical | 4 | 4 | 16 | High | Pick IMU on bias stability (TS-3); warm-up + bias cal before every run; Allan variance (T-05); quantify drift vs camera |
| 3 | RK-02 | If the line's restoring torque isn't compensated, then PD leaves ~0.7° steady-state error and the wheel keeps spinning up during holds | Technical | 5 | 3 | 15 | High | k_s·θ_ref feedforward or integral term; return to 0° between runs; model it in Simulink |
| 4 | RK-03 | If the frame's first torsion mode is lower than the 132 Hz estimate, then R-07 fails | Technical | 3 | 3 | 9 | Medium | Re-derive calc #8 (which inertia twists, bays in series); 2 FEA codes; tap test; shear-panel fallback |
| 5 | RK-04 | If nylon creeps or untwists over hours, then the neutral angle and k_s drift between runs | Technical | 4 | 3 | 12 | Medium | Pre-load the line 24–48 h; record neutral angle each session; trade steel wire (TS-1) |
| 6 | RK-05 | If cogging/stiction near 0 rpm causes limit cycles, then pointing chatters | Technical | 3 | 3 | 9 | Medium | FOC; run the wheel at a bias speed so it never crosses zero; spin-down friction model |
| 7 | RK-06 | If the CAD inertia and the real inertia disagree, then gains and slew times are off | Technical | 4 | 2 | 8 | Medium | Bifilar (T-02/T-03) before any tuning; update model with as-built masses |
| 8 | RK-07 | If swing couples into yaw, then camera and gyro readings are corrupted | Technical | 3 | 3 | 9 | Medium | Damp swing before runs; CG trim; separate 0.35 Hz swing in post-processing |
| 9 | RK-08 | If room air currents (HVAC) exert ~1e-5 N·m, then disturbances are as big as the control torque | Test env. | 4 | 3 | 12 | Medium | Test in a closed room, HVAC off if possible; draft shield; measure drift with motor off |
| 10 | RK-09 | If gyro and camera logs aren't time-synced, then drift and error comparisons are wrong | Test env. | 3 | 3 | 9 | Medium | MCU-triggered LED visible to the camera as a sync mark (R-09) |
| 11 | RK-10 | If wireless telemetry drops packets, then data gaps hide events | Electronics | 3 | 2 | 6 | Low | Log onboard too; sequence numbers + timestamps |
| 12 | RK-11 | If a LiPo is damaged or over-discharged, then fire/injury | Safety | 2 | 5 | 10 | Medium | Balance charger, fire-safe bag, firmware low-voltage cutoff, never charge unattended (R-10) |
| 13 | RK-12 | If shop training or access is delayed, then fabrication slips | Schedule | 3 | 4 | 12 | Medium | Book training now; ask about January hours; fallback: order cut plates from a sheet-metal service |
| 14 | RK-13 | If capstone + spring courses crowd out this project, then gates slip | Schedule | 4 | 4 | 16 | High | Fixed weekly hours; front-load analysis in Nov–Dec; buffer weeks around finals and spring break |
| 15 | RK-14 | If wrong or slow parts arrive, then build stalls | Supply | 3 | 3 | 9 | Medium | Order all electronics at once after G2; buy a spare IMU and encoder |
| 16 | RK-15 | If costs exceed the estimate, then scope gets cut | Cost | 2 | 2 | 4 | Low | Priced BOM with quotes at G2; 15% contingency; ask advisor about department funding |
| 17 | RK-16 | If wheel imbalance excites vibration, then gyro noise rises and structure rings | Technical | 2 | 2 | 4 | Low | Runout tolerance from calc #5; knife-edge balance; spin test to 2,000 rpm |
| 18 | RK-17 | If SolidWorks/ANSYS aren't available off campus, then CAD/FEA slip over break | Tools | 2 | 2 | 4 | Low | Check student licences now; SEAS lab hours as fallback |
| Row | A | B | C | D | E | F | G |
|---|---|---|---|---|---|---|---|
| 1 | Test | Name | Verifies | Needs first | Quantity | Unit | Pass criterion |
| 2 | T-01 | Camera accuracy on a protractor stage | R-04, R-09 | — | Angle error (max) | deg | Within the R-02 budget |
| 3 | T-02 | Bifilar method on Al block | calc #9 | T-01 | Block inertia | kg·m² | Within 3% of analytic |
| 4 | T-03 | Platform inertia (bifilar) | R-02 | T-02 | I_p | kg·m² | Within 5% of SolidWorks, or explained |
| 5 | T-04 | Line stiffness (torsional pendulum) | calc #4 | T-03 | k_s | N·m/rad | Value used in model |
| 6 | T-05 | Gyro noise + bias (Allan variance) | R-03 | — | Bias instability | deg/s | Feeds Pointing_Budget |
| 7 | T-06 | Wheel spin-down friction | calc #3 | — | Coulomb + viscous coeffs | N·m, N·m·s | Value used in model |
| 8 | T-07 | Tap test (phone accelerometer) | R-07 | — | First torsion mode | Hz | ≥ 150 Hz; FEA within 15% |
| 9 | T-08 | Open-loop momentum exchange | Obj. 1 | T-03…T-06 | Peak body rate | deg/s | Within 10% of model |
| 10 | T-09 | Closed-loop 90° slew | R-01, R-02 | T-08 | Settling time to ±2° | s | ≤ 20 s on repeated runs |
| 11 | T-10 | Hold for 60 s | R-03 | T-09 | Max |error| (camera) | deg | ≤ 2° |
| 12 | T-11 | Disturbance rejection (tap/impulse) | Obj. 3 | T-09 | Recovery time | s | Recorded vs model |
| 13 | T-12 | Battery endurance (full sequence) | R-08 | T-09 | Voltage at end | V | Above cutoff |
| 14 | T-13 | Wheel spin + balance test to 2,000 rpm | R-05, calc #5 | — | Vibration / gyro noise vs rpm | — | No resonance; noise acceptable |
| Row | A | B | C | D | E | F | G |
|---|---|---|---|---|---|---|---|
| 1 | Function | Option 1 | Option 2 | Option 3 | Option 4 | Option 5 | Concept A |
| 2 | Store momentum | Solid brass disk | Rim-weighted ring (steel ring on Al hub) | Al disk (bigger Ø) | Spoked wheel | Solid brass disk | |
| 3 | Apply torque to wheel | Gimbal BLDC + FOC driver | Coreless brushed DC + encoder | Small stepper | Drone BLDC + ESC | Gimbal BLDC + FOC | |
| 4 | Measure wheel speed | Magnetic encoder (on motor) | Hall sensors | Back-EMF estimate | Optical tachometer | Magnetic encoder | |
| 5 | Sense body rate (control) | MEMS IMU breakout | Fusion IMU module | Fiber-optic gyro (too costly) | MEMS IMU | ||
| 6 | Suspend with ~1 free axis | Nylon monofilament | Thin steel/music wire | Braided PE (Dyneema) line | Kevlar thread | Air bearing (too costly) | Nylon monofilament |
| 7 | Measure true angle (truth) | Camera + ArUco marker below | Laser on mirror → wall scale | Overhead camera + marker on top | Magnetometer | Camera + ArUco below | |
| 8 | Compute + control | ESP32 | ESP32-S3 | Teensy 4.x | STM32 FOC board | ESP32 | |
| 9 | Telemetry / logging | Wi-Fi UDP | BLE | ESP-NOW to a 2nd ESP32 | Onboard flash/SD log | Wi-Fi + onboard log | |
| 10 | Power | 3S LiPo | 2S LiPo | 2× 18650 Li-ion | 3S LiPo | ||
| 11 | Structure | Rails + plates (Al) | Rails + plates + shear panels | 3D-printed frame | Aluminum extrusion | Rails + plates | |
| 12 | Balance / CG trim | Trim screws | Sliding trim masses | Adjustable eyelet position | Trim screws | ||
| 13 | Concept A = the proposal baseline (pre-filled). Concepts B and C are yours to build, e.g. B = 'minimum-friction' (steel wire, coreless DC, laser truth), C = 'robust' (shear panels, Teensy, onboard log). |
Parts list · P01-100P01 · Rev A
| Item | Part | Spec | What it is for |
|---|---|---|---|
| 01 | Nylon line and eyelet | Nylon monofilament, stainless eyelet · k ≈ 3.1E-6 N·m/rad | Hangs the model so it turns freely about one axis. Its small twist stiffness pulls the model back toward zero, which the controller cancels with feedforward. |
| 02 | Top plate | 6061-T6 aluminum, 1/8 in | Carries the eyelet and ties the four rails together so the frame stays stiff in torsion. |
| 03 | Corner rails | 6061-T6 square bar, 5/16 in | The main structure. Bolted to the plates, they lift the first twisting mode to 374 Hz, far above the control band. |
| 04 | Flywheel | C360 brass · Ø35 × 8 mm · 65 g · I ≈ 1.0E-5 kg·m² | Stores the momentum. Spinning it one way turns the body the other way; it holds 4.2× what the 90° turn needs. |
| 05 | Gimbal motor | 2204-class brushless, FOC drive | Spins the flywheel at low speed under field-oriented control. Cogging and friction are handled by a fast inner speed loop on the encoder, which sees and cancels them hundreds of times a second. |
| — | Wheel deck | 6061-T6 aluminum, 1/8 in | Puts the motor exactly on the spin axis so the wheel only torques about yaw. |
| — | Magnetic encoder | On-axis encoder board | Measures wheel speed for the inner speed loop, which fixes the coarse throttle steps found in the PWM analysis. |
| 06 | Flight computer | ESP32 dev board | Runs the controller and logs data on board and over Wi-Fi. |
| 07 | IMU | MEMS gyro + accelerometer | Senses body rate. Chosen on bias stability, because gyro drift is the biggest threat to the ±2° budget. |
| — | Motor driver | FOC driver board | Turns controller commands into motor current. |
| — | Sync LED | MCU-triggered LED | Flashes at the start of each run so the camera and gyro logs line up within 50 ms. |
| 08 | Battery | 3S LiPo, ~450 mAh | Powers a full test sequence on one charge, with a low-voltage cutoff in firmware. |
| — | Bottom plate and trim screws | 6061-T6 aluminum, 4 trim screws | The trim screws move the center of gravity under the line so the model hangs level. |
| 09 | Camera marker | ArUco target, read from below | The truth sensor. A camera reads its angle independently, so the controller never grades its own work. |
| — | Standoffs | Brass, M3 | Space the two decks and stiffen the middle of the frame. |
| — | Electronics deck | 6061-T6 aluminum, 1/8 in | Carries the computer, driver, IMU and battery close to the spin axis. |
Item numbers match the numbered callouts on the exploded view. A dash means the part is on the model but carries no callout.
Slew and hold response from the Gate 1 model. Test data from T-09 and T-10 is added to this plot as it comes in.
Decision logP01
Design decisions
| Date | Decision | Why |
|---|---|---|
| Nov 2026 | Brass flywheel, gimbal BLDC with FOC | Smooth low-speed torque, enough inertia in a small disk |
| Nov 2026 | Add line-twist feedforward | Error budget showed 2.7° against ±2°; feedforward brings it to 1.1° |
| Nov 2026 | Inner wheel-speed loop | PWM steps too coarse for fine torque |
| Nov 2026 | Camera as truth sensor only | The controller never grades its own work |
P01
Not uploaded yet.