Every subsystem is generated from the same configuration files the checks run against, so a
design change and its documentation can never quietly drift apart.
| Subsystem | Architecture |
|---|
| Structure | An aluminium frame built around a 40 mm central spine in a 2×2×3 cell grid (226.3×226.3×340.5 mm envelope). Every card in the spacecraft shares one 100×100 mm outline with a 20×20 mm spine notch and an 8 mm rail relief -- a format invented for this project because standard 90/100 mm boards don't clear the spine. Six hold-down posts per wing stack, not just the corners, push the stowed first mode to about 137 Hz; an earlier corners-only analysis found only 99 Hz. |
| Power | Two independent MPPT/battery chains (A, B) plus a fully independent survival chain (C) with its own dedicated body-mounted solar string; per-card eFuse and a hardware KILL_N line, with no software anywhere in the protection path. |
| Thermal | Fully passive. Analysis found the spacecraft runs cold rather than hot -- the real fight is a -27 °C eclipse case and a -40 °C safe-mode case, held off by splitting roughly half the surface between the solar cells and dedicated MLI/low-e patches. The telescope column gets its own low-α/low-ε coating and an 8 W thermostatic heater (14-17 °C setpoint); the battery vault is isolated (0.1 W/K) with a 6 W heater, and the survival string has its own 2 W heater so it never drops below -20 °C. |
| Attitude & propulsion | Three reaction wheels (30 mN·m·s each), magnetorquers and a star tracker (5 arcsec, ~17 m ground error) close a 90° slew in about 120 s with 3x+ margin on torque and momentum. Propulsion is electric (iodine gridded-ion/Hall-class) rather than a water resistojet for the specific impulse: 0.5 kg of iodine buys ~570 m/s against ~45 m/s from 0.8 kg of water -- enough for 50-year drag makeup, collision avoidance and a controlled deorbit from one tank. The thruster sits on the centre line, canted a few degrees through the centre of mass; an earlier in-column placement produced enough off-axis torque to saturate a wheel in seven minutes. A drag sail is the passive backup. |
| Compute | A five-tier hierarchy, deliberately over-provisioned at the bottom: an MSP430FR supervisor and dual hot-standby STM32H7 flight controllers are rated for the full 50-year mission, because the long-record space-weather science has to keep running long after the Linux side is expected to have failed. A Raspberry Pi CM5 (always on) and an NVIDIA Jetson Orin Nano (imaging, heavy compute) sit above that with 10-15 year design lives. The mass memory unit holds 8×128 GB with 6+2 erasure coding, BLAKE2 hashing and periodic scrubbing. |
| Data & comms | 27 CCSDS message types, every telecommand Ed25519-signed with a counter and validity window (the link itself is unencrypted, as amateur-band rules require). The control plane is dual CAN-FD (1 Mbps/5 Mbps) for redundancy; a gigabit Ethernet plane carries payload data. Downlink is UHF/LoRa housekeeping (~0.5 MB/day) plus an adaptive S-band science link (250 kbps-2 Mbps) -- from the Pamukkale reference station alone, roughly 28 minutes of contact and ~134 MB/day across 3-4 passes. |
| Payload | 26 instruments split cleanly by lifetime. Long-record space-weather/geophysics science (three-axis magnetometer, total-solar-irradiance radiometer, radiation dosimeter, a 16-electron/16-proton particle spectrometer, an X-ray/UV monitor, a CsI+SiPM gamma-ray-burst detector, a chip-scale atomic clock) is acquired directly by the flight controllers, never the Linux nodes. Imaging -- an 85 mm Cassegrain telescope at roughly 5 m ground sample distance, a ~50-band hyperspectral channel sharing the same optics, a thermal-IR microbolometer and a wide-field star/context camera -- runs on the Pi/Jetson side. Every instrument number above is a design target verified so far only by analysis, not by a built and tested instrument. |
| Ground & operations | Two published ground-segment tiers: a $30-80 UHF/LoRa listener anyone can build to receive the beacon, and a ~$2,200-4,500 S-band science station (1.2 m dish, SPID BIG-RAS rotator) for the full downlink. All 27 protocol messages already have a compiled, verified Kaitai Struct decoder, with SatNOGS DB/Decoder integration planned so any station can decode the telemetry without touching this repo. A second anchor ground station is proposed but not yet arranged, and frequency coordination/licensing hasn't started. |
