A Satellite's Sudden Spin: The Single-Event Upset Checklist

Archi4 min read

satellitesspaceradiationtelemetry

Contents
  1. What Causes a Sudden Attitude Disturbance
  2. The Particle That Does the Damage
  3. The Initial Data to Pull Before Changing Anything
  4. How to Read the Patterns
  5. The Checklist for Recovery
  6. What the Checklist Misses
  7. Why the Root Cause Matters

What Causes a Sudden Attitude Disturbance#

A satellite operator sees an alarm at 02:14 UTC. The star tracker reports lost track. Reaction wheel speeds are climbing, and the sun sensor shows an angle that cannot be right. The first assumption is often a debris impact or a structural failure. In many cases, the cause is none of those.

The technical root cause is a single-event upset. A charged particle strikes a storage cell in the attitude control electronics and flips a bit. The corrupted bit changes how software reads a sensor or writes a command. The effect appears as a false attitude estimate or a bad actuator command. The satellite begins to rotate along a vector it was never commanded to follow.

This article is a practical checklist for the moments after such an anomaly. It assumes you have telemetry access and a command link. It does not cover mechanical failures.

The Particle That Does the Damage#

Cosmic rays and solar protons carry enough energy to pass through the satellite's shielding. When one ion crosses a silicon junction, it leaves a trail of charge. If that charge is collected by a memory node, the stored value can change. A single changed bit in a 64-bit attitude word can shift the computed pointing direction by a large angle.

Radiation effects are not equal everywhere. Polar orbits pass through higher radiation flux. Geostationary satellites receive fewer protons but still see cosmic rays. Older components with larger feature sizes actually store more charge and are more resistant. Newer chips with smaller transistors need less charge to flip.

But the operator sees no particle. The operator sees a tumbling platform.

The Initial Data to Pull Before Changing Anything#

Check these items in order.

  1. Star tracker status words and the last valid quaternion.
  2. Sun sensor readings and the predicted sun vector from the onboard ephemeris.
  3. Gyro rates and magnetometer field values.
  4. Reaction wheel tachometers and motor current limits.
  5. Watchdog reset counters and software reboot logs.
  6. Memory error counters in the attitude computer's EDAC logic.
  7. Power bus voltage and solar array string current.

The last item matters because it separates an upset from a latch-up. A latch-up creates a low-resistance path in a chip and draws excessive current. Power bus voltage drops and the resets stop. An upset does not change power draw. If solar array current is stable, the suspect is the software state, not the electronics rail.

How to Read the Patterns#

A single-event upset usually leaves a signature. The gyro rates might be smooth and continuous, while the attitude quaternion changes in a single telemetry frame. That pattern means the sensor hardware is fine. The computer consumed a corrupted data word.

Another signature appears in the star tracker. A tracker that loses track for a short interval, then returns with a rotated attitude, points to a bit flip in the tracker interface. A tracker that stays valid while the attitude solution disagrees with its output points to a bit flip in the main computer memory.

Watchdog resets are common. Many flight computers use a register to store the cause of the last reset. Retrieve that register before clearing it. The register might say "external reset" or "software trap." Neither implies a permanent fault.

The Checklist for Recovery#

Once you have the patterns, do the following. , Switch the attitude controller to Sun acquisition mode. This removes reliance on corrupted star data and gives power-positive orientation., Upload a new star tracker command set and force a re-initialization. Do not trust the onboard cache., Scrub the memory regions used for the attitude filter. If the system has an EDAC scrub command, run it twice., Verify all reaction wheel tachometers against commanded wheel speed. Then disable any wheel that shows a current anomaly., Compare the ground-generated ephemeris with the onboard ephemeris. Replace the onboard block if the comparison is off., Re-enable normal pointing only after two consecutive star tracker solutions agree with the ground calculated attitude., File a report with the telemetry timestamps, the reset cause register, and the memory error counters. This report is the basis for a future mitigation.

What the Checklist Misses#

This checklist covers upsets, not permanent radiation damage. Total ionizing dose accumulates over years. When a satellite reaches its end-of-life, threshold voltages drift and performance fades. That is a different failure mode with a longer timeline.

It also does not cover external torque from solar pressure or gravity gradient. Those torques are predictable. If the angular acceleration follows a known model, the cause is environmental torque, not a bit flip.

The most common misdiagnosis is orbital debris. A debris strike can also cause sudden rotation, but it usually shows a small attitude rate change and a change in solar array current or pressure. With an upset, power is perfectly stable and the rotation is arbitrary.

Why the Root Cause Matters#

Operators who misdiagnose an upset as a debris impact spend days checking hardness telemetry and windows. The actual fix is a memory scrub and a software reload. The same upset can happen again on the next large solar proton event. Knowing the root cause tells you whether to update the radiation monitoring threshold, adjust the safe mode logic, or add a redundant voting scheme in software.

The next anomaly may appear suddenly with no external warning. The telemetry pattern, reset register, and memory counters will identify it before any physical inspection can.

More articles