ARC24 mission spec

International Space Tournament «Орбита» (spaceeducation.info), 2026 season. Decision record, 12 September 2026. Supersedes MAGSEP-mission-spec.md, which was written against the wrong competition.

One board, one question: what happens to a circuit board when the air goes away. Two coupled experiments sharing one independent variable — ambient pressure, swept continuously by the balloon from 1013 mbar to ~29 mbar at 24 km.

They share a board, sensors and a flight, so fly both. CONVECT is also the insurance: no high voltage, no approval needed, still a complete result.

1. Competition facts (official documents)

Tournament page · Regulations PDF, updated 25 June 2026

• Final: 26 October – 1 November 2026. Forum 31 October.
• Preparation: 10 September – 26 October. Kits ship 10–20 September. Minimum 3 general webinars + 3 individual expert consultations.
• Platform: free 3U CubeSat kit without payload, based on IntroSat.
• Stratospheric launch, up to 24 km — the environment IS available at judging.
• Team: exactly 4 (ages 12–18, optionally one university student ≤21) + mentor who may not do technical work.
• Roles: Engineer (payload design, 3D, integration) · Programmer (electronics, data/comms software, MCU) · Researcher (scientific justification, modelling, interpretation, presentation) · Radio Technician (radio module + ground station software).
• Deliverables at final: integrated CubeSat with payload, ground station software + testing documentation, defence before the Expert Council.
• Scoring: preparatory-stage work + defence. Rubric published only on day 1.
• Missing a mandatory assignment on time = disqualification.

44 days to the final. ~8 days to the PCB fab deadline.

2. Why the flight is a sweep, not an exposure

Two or three hours is far too short to change anything — no useful dose, no material ageing, no biology. But the ascent carries the payload continuously through two decades of pressure and ~70 °C of temperature. It hands you a curve, not a before-and-after. Every strong mission treats pressure or temperature as a swept independent variable.

ISA pressure: 1013 mbar ground · 120 at 15 km · 55 at 20 km · 29 at 24 km.

3. ARC-24 physics

Air's breakdown voltage depends on pressure × gap distance and is not monotonic — Paschen's law, minimum ~327 V near 7.5 Torr·mm for air. At 24 km (~22 Torr) a 0.3 mm gap has pd ≈ 6.6 Torr·mm: ordinary PCB clearances sit at their most vulnerable exactly at this competition's ceiling.

Gap sizing (computed for this report; ISA + Paschen with A=15/cm·Torr, B=365 V/cm·Torr, γ=0.01 — design estimates, not measurements):

At a fixed 500 V, each gap breaks down over a band of altitudes:

That staircase — breakdowns appearing at four predicted altitudes in a predicted order, and a narrow gap recovering as it passes its minimum — is the whole experiment. It cannot be explained by "it was high so it arced."

Why not in a lab: reaching these pressures needs a controlled vacuum chamber. The balloon gives a clean, slow, monotonic sweep for free.

The open question: classical Paschen assumes uniform parallel-plate fields. A PCB is not that. The one directly relevant measurement (arXiv 2010.06570) found 405 V at 5 mbar for a real PCB trace vs 327 V classical, plus unexplained saturation below 10⁻² mbar — but in a vacuum chamber, mostly below stratospheric pressures, on one geometry. Nobody has flown an array of PCB gap geometries, coated and uncoated, through the real stratospheric sweep and compared it to both classical Paschen and IPC-2221.

The standard you get to test: IPC-2221 sets conductor clearance by voltage and treats external conductors differently below/above 3050 m, with a smaller allowance for permanently coated conductors. Your flight goes 8× higher. Coated vs uncoated at 24 km asks whether that coating allowance holds where it has never been characterised.

4. Board design (this is the PCB-learning part)

• Gap array: 5 widths (0.10/0.15/0.20/0.30/0.50 mm) × 2 surface treatments (bare, conformal-coated) = 10 cells.
• Geometry held constant: identical pad shape, copper thickness, corner radius across all cells. Geometry shifts the whole curve, so it must be the controlled variable.
• Current limiting: 10 MΩ in series per cell → 60 µA max at 600 V. This is both the safety measure and the measurement principle.
• Detection: low-side sense resistor per cell into a multiplexed ADC at ≥100 Hz; breakdown = current step from leakage floor to limit. Optional photodiode for independent optical confirmation.
• Voltage measurement: 100 MΩ : 100 kΩ divider into its own ADC channel. Measure the applied voltage; the boost converter will drift with temperature.
• Two modes alternating: fixed 500 V hold (gives the altitude staircase) and 0→600 V ramp over ~20 s (gives onset voltage per cell → the curve itself).
• Pressure sensor rated below 29 mbar, logged on the same timestamp. Measure pressure directly — never infer from model altitude.
• Layout: HV section isolated with guard ring + slot, away from OBC and radio, own return path. Discharges radiate broadband.

5. CONVECT (same board)

Four matched heated elements, each at known constant power, each with its own temperature sensor, differing only in surrounding copper: bare pad · large pour · thermal via array to an inner plane · deliberately thermally-relieved pad. R_th = ΔT / P, plotted against pressure through the ascent.

Tells a PCB designer how much of the cooling their ground testing shows is convection that will not exist in space — per layout technique. This is the rigorous version of "map the heat on the board": a measured collapse of a heat transfer mechanism against a swept variable, not a thermal photograph.

6. Experimental design

• Independent: pressure (swept, measured). Secondary: gap width, surface treatment, copper structure, applied voltage.
• Dependent: onset voltage per cell; breakdown state vs pressure; R_th per structure.
• Controls: 2 mm gap (must never fire); identical twin board tested on the ground before and after; unpowered cells monitored for leakage; isolated pad as low-conduction reference.
• Repetitions: ramp repeats every ~60 s → ~100 onset measurements per cell; descent re-traverses the same pressures (built-in repeat + hysteresis check).
• Covariate: Geiger count rate. Cosmic rays supply seed electrons for breakdown initiation, and the Regener–Pfotzer maximum (~15–26 km) sits inside the flight band. Resolving the coupling in one flight is unlikely; proposing and testing it is what a researcher does.
• Uncertainty: propagate pressure accuracy, fab etch tolerance on gap width (typically ±20% — ask your fab house), divider tolerance and its temp drift, ADC noise. Band every plotted point.
• Falsification: if onsets don't follow a Paschen-shaped curve, or narrow gaps don't extinguish above their minima, classical theory fails for this geometry — plausible given the arXiv result, and interesting, not a failure.

Figures: onset voltage vs pd with classical Paschen overlaid (the defence centrepiece) · altitude staircase · coated vs bare at matched pd · ascent vs descent hysteresis · R_th vs pressure for four structures · onset scatter vs Geiger rate.

7. Safety — read before ordering

Several hundred volts. Made safe by current limiting by design, not care: 10 MΩ per cell caps current at tens of µA; HV fully enclosed, no exposed conductors; bleed resistors discharge on power-down; interlocked off on the ground. Mentor must review. Organizers must approve in writing before you travel. If refused → fly CONVECT + context sensors, still a complete mission.

8. Risks

9. BOM (planning estimates, not quotes)

Many common barometric sensors bottom out at 300 mbar — check the low-pressure limit before ordering.

10. Schedule — 44 days

The one date that matters: ~20 September for the PCB order.

11. Work split (the four official roles)

• Engineer — board outline and 3U integration, gap-array geometry, guard ring and isolation slot, HV enclosure, insulation, post-fab gap measurement.
• Programmer — ramp/hold sequencing, ≥100 Hz multiplexed sampling, event detection distinguishing a true current step from leakage drift, SD logging, reduced live telemetry.
• Researcher — the Paschen prediction computed and plotted before flight, IPC-2221 framing, uncertainty propagation, covariate analysis, defence narrative.
• Radio Technician — ground station and live summary downlink, which is the backup copy of the headline result if the payload is not recovered.

All four roles carry something the defence depends on — which matters when each of you is individually assessed.

12. Sources

• Tournament page · Regulations PDF
• Paschen curve for a PCB trace in low vacuum (arXiv 2010.06570)
• Paschen's law — equation and air coefficients
• IPC-2221B conductor spacing tables and the 3050 m threshold
• Using an IPC-2221 calculator for high-voltage design (Altium)
• NASA-HDBK-4007 Spacecraft High-Voltage Paschen and Corona Design Handbook
• Hands et al. (2016), Space Weather — Pfotzer–Regener maximum in dose equivalent
• Measurement of the Regener–Pfotzer maximum with different detectors
• TASEC-Lab — convective heat transfer on stratospheric balloon missions (Acta Astronautica)
• NASA — Ballooning for Biologists: near-space environment by altitude