Orbiting

Now nearing the end of our first day in lunar orbit. We had a surprise just before noon EDT on Tuesday. As described in the LRO blog, the Orbiter automatically protected itself (entering SunSafe Mode) because a flight software "actionpoint" (AP22) had evaluated a sizeable discrepancy between how the spacecraft was oriented and where it expected it should be (specifically an angle greater than 25 degrees).

Actionpoints are defined in a flight software "task" called "Limit Checker," which monitors specified onboard sensor data against thresholds that we set in order to protect various onboard systems, including the spacecraft itself. When an actionpoint trips, it triggers a predefined sequence of commands intended to respond properly (as best as can be predicted ahead of time) to the detected situation.

What caused the attitude discrepancy? Our Guidance, Navigation & Control (GN&C) colleagues believe that one of the star trackers was busy detecting stars, identifying them, and using that information to return attitude knowledge to the onboard computer when the Moon impinged on its field of view. The tracker is designed not to be confused by this event, but perhaps evaluated sunlit mountain peaks as stars and made a "match" that indicated a star field in another part of the sky, returning a very different spacecraft attitude than the true one. The Limit Checker caught the problem based on the Attitude Control task's overall attitude knowledge (which uses gyros and sun sensor information in addition to star tracker data).

After the circumstances of this incident were established by the LRO team, the Orbiter was commanded back to its usual attitude-control mode ("Observing Mode"). For the rest of the day, star trackers were intentionally commanded out of the control loop whenever the Moon's limb came near the field of view. Instead, the gyros were used at those times.

A few hours ago, "Kalman Filter" control was enabled for the first time. This will prevent the problem from happening again. Onboard ephemeris data will be used in conjunction with the star tracker to indicate when the star tracker will be blocked by the Moon (or Earth) and not be used for attitude knowledge.

The reason KF control wasn't used earlier was that various settings were still requiring fine tuning.

Having said all this, I should add a disclaimer that I am not an attitude-control specialist and my description should be considered a "rough draft" with no guarantees of complete accuracy. I have included my inferences with tentative findings that may not be the complete story. I hope to convey the general aspects of spacecraft attitude control however.