The square deployment frame becomes more useful when it is repeated in depth. A transparent sacrificial parasol can stand in front of a hard-tile photovoltaic field, while a second parasol behind the electronics carries the thermal barrier and radiator. Between them sits one independently operable unit of orbital compute.
The interactive model separates the planes so their jobs remain visible. Drag to orbit, move the separation control, select a component, or switch on the energy paths to follow sunlight into the cell and waste heat back out to space. It shows the selected four-metre envelope, four two-metre structural quadrants, and the provisional 60-tile physical layout around the central service well.
The closed functional cell
From the Sun-facing side to the anti-Sun side, the architecture is:
| Plane or component | Primary job |
|---|---|
| Transparent sacrificial parasol | Break up small hypervelocity particles before they reach the cells |
| Forward standoff | Give an impact plume room to spread and prevent contact with the PV surface |
| Hard-tile photovoltaic field | Convert AM0 sunlight through 60 independently replaceable and isolatable physical tile domains |
| Compute and power core | Consume 1.519 kW nominal including six refined CM250 modules and cell housekeeping |
| Rear standoff and thermal links | Mechanically isolate the planes while conducting compute heat to the radiator |
| Reflective thermal barrier | Keep infrared emission from the hot PV blanket from loading the radiator |
| High-emissivity radiator face | Reject the 2.13 kW inherited allocation through its anti-Sun surface; trajectory closure remains open |
This is not one material doing seven jobs. It is one deployment envelope containing several deliberately separated functions.
A natural power–compute unit
The first complete ledger changes the original rule of thumb. Annual-average sunlight is not the binding case: near the worst eclipse season, each 96.687-minute orbit contains 20.685 minutes of shadow. The sunlit interval must run the live load and restore the battery before the next eclipse.
The current parametric result is:
| Cell balance | Provisional value |
|---|---|
| Selected square side | 4.00 m |
| Structural subdivision | Four 2 × 2 m quadrants |
| Physical tile topology | 500 mm bay pitch; 60 × 499 mm tiles in an 8 × 8-minus-2 × 2 layout |
| Tile population | 12 × 12 cells per tile; 8,640 cells total |
| Complete tile article | 1.444 kg grown; 11.36 mm body; 25.36 mm including rear service hardware |
| Local lattice-bay receiver | 1.5125 kg grown; 220 mm rear service aperture |
| Complete 60-tile population | 86.64 kg |
| Sixty tiles plus sixty local receivers | 177.39 kg before quadrant truss and deployment hardware |
| Total active cell area | 13.133 m² |
| Clear opening around the 700 mm rack | 970 mm |
| Shielded silicon-HJT BOL output | 3.277 kW |
| Five-year EOL output with one physical tile isolated | 2.712 kW |
| Complete nominal cell load | 1.519 kW |
| PV required during sunlight for the maximum-eclipse orbit | 2.003 kW |
| EOL one-tile-out worst-orbit margin | 35.38% |
| Maximum-eclipse recharge time | 30.83 minutes |
| Shared operational/bootstrap storage floor | 1.833 kWh |
The former 3 × 3 m case still produces 1.868 kW BOL behind the bumper, but falls to 1.529 kW at five-year EOL with one equal-power domain out. It needs 2.003 kW during the worst eclipse-season orbit to carry nominal compute and recover storage, so its energy margin is −23.67%. Even switching compute off during eclipse leaves the faulted EOL case below the separate 10% margin requirement.
The revised accounting rule is therefore:
under the current provisional assumptions, one 4 m shielded cell sustains one 1.519 kW nominal compute domain through the reference maximum eclipse
This is still not a flight rating. The 3.550 m calculated physics threshold depends directly on unqualified silicon-cell degradation, coverglass and shield transmission, temperature, pointing, battery efficiency, and the first-order eclipse model. The jump to 4 m is a structural choice: it makes the envelope an integer 2 × 2 grid of 2 m quadrants. The physical reclosure then replaces the parent’s idealized 64 domains with 60 whole tiles and still retains 35.38% margin.
The structural and thermal sides are now the next coupled gates. A first receiver trade shows that the 499 mm tile cannot be pushed through a meaningful fixed lattice on a 500 mm pitch. It therefore mounts on the Sun-facing lattice surface while Hexadrone services the rear grapple through a 220 mm aperture. The handoff now has a provisional mechanism: one 18.0 kg captive polar presenter rises through the 970 mm center well, carries the tile above the array, and holds four structural edge lands until the rear service head confirms passive capture. All 60 bay centers pass the reach screen, but the 50.9 mm worst-corner reach margin and 10 mm parked center-well margin are tight. The 90.75 kg repeated-receiver allocation must also be absorbed into a deployable quadrant truss. Separately, the existing 2.13 kW radiator allocation exceeds the 1.974 kW peak electrical load by only about 154 W; equilibrium temperature, view factors, coatings, and whether the rear surface follows the 4 m power frame remain unresolved.
The rear parasol has two opposite surfaces
Calling the rear plane a heat shield is only half of its job. Its two faces need opposing optical properties:
- The inward face is reflective and low-emissivity. It rejects thermal radiation arriving from the warm backside of the PV plane.
- The outward face is high-emissivity and has a clear anti-Sun view. It radiates heat delivered from the compute core.
The membrane therefore needs an in-plane heat-spreading network. Flexible graphite straps, metallic foils, or segmented heat pipes connect the core to radial spreaders in the deployed surface. Without that thermal connection, the rear parasol is only a shade.
The conservative 250 W/m² figure treats the original rear surface as effectively one-sided. More capable two-sided or higher-temperature radiators could reduce the required area, while a 4 m common frame could increase it. Neither possibility is credited until the trajectory thermal balance closes.
Protection is split by threat
The front parasol is a particle bumper, not universal armor:
| Environment | Assigned response |
|---|---|
| Micrometeoroids and small debris | Transparent sacrificial bumper, standoff distance, cell segmentation, and bypass paths |
| Atomic oxygen | Thin barrier coatings on every exposed susceptible surface |
| Ultraviolet exposure | Cell-local optical barrier and stable membrane chemistry |
| Electrons and protons | Tolerant cells and electronics, local shielding, and architectural redundancy |
| Punctures that reach the array | Cell bypass plus tile isolation so damage is contained to one replaceable tile |
| Rear-side impacts | Local radiator segmentation and redundant thermal paths |
Separating these responses keeps the bumper replaceable without pretending that it replaces cell passivation, radiation design, or fault isolation.
Common mechanism, different skins
All three planes use the same geometric vocabulary: a central hub, four half-diagonals, four corner nodes, and a tensioned square perimeter. The reusable parts are the deployment kinematics and interfaces. The functional membranes differ:
- optically transparent, low-absorption, and tear-arrested in front;
- electrically interconnected and locally bypassed in the middle;
- reflective inward, emissive outward, and thermally conductive behind.
Common corner geometry lets a central spine carry short normal standoffs between the planes. It also gives each surface a clear service boundary. A bumper or radiator may be replaced during a maintenance operation without replacing the photovoltaic field or compute core; replacement is not attempted while the cell is operating.
Deployment sequence
- Release the packaged cell. The central equipment spine establishes the Sun-to-radiator axis.
- Open and latch the photovoltaic plane. It becomes the reference geometry for the other two surfaces.
- Extend the forward standoffs. The transparent bumper moves away from the cells and reaches final tension.
- Extend the rear standoffs. The thermal parasol moves into the anti-Sun direction.
- Tension thermal links and spreaders. Flexible conductors establish a continuous path from compute core to radiator.
- Verify separation and insulation. The system checks latches, electrical isolation, temperatures, and membrane clearance.
- Energize one domain at a time. PV strings, power conversion, compute modules, and communications enter service incrementally.
The model shows the operational stack and an exaggerated separated view. It does not yet represent the launch fold or select a standoff length.
Scaling the data centre
The shielded power–compute cell is the atomic capacity unit, even when many cells share buses, networking, attitude control, and maintenance vehicles:
| Shielded cells | Selected PV envelope area | Approximate continuous nominal power |
|---|---|---|
| 1 | 16.00 m² | 1.519 kW |
| 8 | 128.00 m² | 12.15 kW |
| 64 | 1,024.00 m² | 97.23 kW |
| 66 | 1,056.00 m² | approximately 100 kW |
| 659 | 10,544.00 m² | approximately 1 MW |
Larger pods should be assembled from independently disconnectable cells. A puncture, failed power converter, degraded PV string, or thermal fault can then remove one 1.5 kW domain rather than propagating through the data centre.
Model boundary
This is a provisional architectural closure, not a qualified spacecraft design. The next iteration must determine solar-weighted bumper transmission and absorption, HJT BOL/EOL performance, hypervelocity impact behavior, membrane charging, shield service parking, presenter boom and center-well integration, coupled Hexadrone handoff dynamics, radiator operating temperature, in-plane thermal conductance, flexible-link fatigue, cell string topology, battery cycle life, quadrant-lattice mass, and common deployment-hardware mass.
The important result is narrower and more useful than the original concept: the current energy threshold is 3.550 m, while the selected deployable unit is 4 m because it resolves cleanly into four 2 m structural quadrants. Inside it, an 8 × 8 bay grid with the centered 2 × 2 bays empty produces 60 identical 499 mm tiles, 15 per quadrant. A complete article allocates a 1.444 kg, 11.36 mm CFRP/aluminum-honeycomb tile with rear handling hardware. Its provisional 1.5125 kg receiver places that tile on the Sun-facing lattice surface and keeps the centered rear grapple accessible through a 220 mm aperture. A shared 18.0 kg center-well presenter now closes the passive Sunward/rear handoff geometrically. The next gate is the complete quadrant truss with receiver and presenter loads represented—not another change to tile count.