TRACE THE
FAILURE
SPACECRAFT NAVIGATION

NASA’s Mars Climate Orbiter.
Lost over a units mix-up.

A unit mismatch distorted the navigation model. The deeper failure was allowing that mismatch and unresolved warnings to survive to arrival.

Follow the evidence
Illustrative spacecraft with solar panels above Mars
AI-generated spacecraft illustration used for the episode artwork. Not an archival photograph or an exact reconstruction of Mars Climate Orbiter’s trajectory.
THE COMPANION FILMMars Climate Orbiter Failure: NASA's $327M Units Mix-Up Watch episode 4:27

What was the Mars Climate Orbiter unit error?

Mars Climate Orbiter’s ground software produced thruster-impulse data in pound-force seconds, while the navigation software expected newton-seconds. Without the required conversion, the model underestimated the effect of these small thruster firings by a factor of about 4.45, according to NASA’s investigation. Phase I report, p. 16.

The spacecraft was lost during its arrival at Mars on September 23, 1999. The investigation examined why interface specifications, software testing and navigation checks had failed to expose the units error before arrival. Phase I report, pp. 6–7.

Watch the explanation

Watch the episode’s explanation of thruster impulses and navigation with the source correction in mind.

What the “Small Forces” file represented

The spacecraft used reaction wheels to control its orientation. Momentum accumulated in those wheels, and periodic thruster firings removed it. These angular-momentum desaturation events also gave the spacecraft small changes in velocity that the navigators needed to include in their trajectory calculations. Phase I report, pp. 6–7, 17–18.

The file recorded impulse, or force accumulated over time. The same impulse has a different numerical value in pound-force seconds and newton-seconds. Reading the number in the wrong units made the modeled push too small.

The interface documentation already specified metric units. The delivered ground-software output did not follow that specification, and the receiving process failed to catch the mismatch. Phase I report, p. 16.

The spacecraft’s asymmetric solar array contributed to more frequent momentum-desaturation events than the operations navigation team expected. More events made the incorrect small-force modeling more consequential. Phase I report, pp. 6–7, 21.

What happened to the course corrections?

NASA’s report records unresolved differences between navigation solutions during the approach. TCM-4 was executed as planned on September 15, 1999. The contingency maneuver discussed shortly before orbit insertion, but never performed, was TCM-5. Phase I report, pp. 13 and 19.

The report describes incomplete analysis, tests and procedures for that contingency, a team unprepared to execute it, and the priority of the onboard orbit-insertion timeline. It does not support reducing the decision to a cancellation of TCM-4 to save fuel or money.

These findings correct the earlier episode script. After contact was lost, investigators reconstructed the trajectory from the available tracking information and corrected force values. They could not directly observe an exact breakup sequence.

What investigators could establish

The signal disappeared 49 seconds earlier than predicted and was not reacquired. After the loss, corrected navigation estimates placed the first periapsis (the closest approach to Mars) at about 57 kilometres. The Board judged that altitude too low for survival. Investigators derived it after the event; the spacecraft did not transmit a measurement from that altitude. Phase I report, pp. 13–14.

The Board identified weak communication, staffing and training problems, inadequate systems engineering during the transition to operations, and verification that did not adequately cover ground software. It recommended unit-consistency checks and a software-interface audit for the next Mars mission. Phase I report, pp. 7, 16, 21–24.

The units were specified correctly, but the teams had not verified that the complete system used them. Incorrect units crossed the software interface, and the surrounding safeguards failed to catch the resulting navigation error.

The source record

NASA Mars Climate Orbiter Mishap Investigation Board: Phase I Report

November 10, 1999 · Pages 6–7: root and contributing causes; pp. 13–14: arrival, TCM-4 and reconstructed periapsis; pp. 16–19: units and TCM-5; pp. 21–24: systems engineering, communication and verification. Printed and PDF page numbers match.

Research checked September 21, 2026. Illustrative episode images are not primary evidence.