Mars probe fails due to unit error

On 23 September 1999, the Mars Climate Orbiter approached Mars at an altitude of approximately 57 km. NASA had aimed for between 140 and 150 km, and the probe could not survive below approximately 80 km.

Unit conversion error caused the failure

The cause of the failure relates to a unit conversion that was never made. The ground software from Lockheed Martin, the manufacturer, expressed the impulse of the small thruster firings in pound-seconds. The navigation from the Jet Propulsion Laboratory (JPL), however, read these same values as newton-seconds.

Difference in units led to miscalculations

A pound-force is approximately 4.45 newtons, and the discrepancy accumulated during the nine-month journey. The Lockheed Martin software expressed the impulses in pound-seconds. The JPL navigation read them in newton-seconds, a difference of 4.45 for 1.

Frequent thruster firings worsened the problem

The desaturation of the reaction wheels was 10 to 14 times more frequent than expected. The probe descended to approximately 57 km, 23 km below the survival threshold. Exact figures, read in the wrong unit.

Routine maintenance became dangerous

These small firings are a routine maintenance procedure. The probe's solar panel, asymmetrical, is subjected to the pressure of sunlight, which causes its reaction wheels to spin until saturation. Then, the thrusters must be fired briefly to "desaturate" them.

Misinterpretation led to incorrect trajectory

According to a recent analysis of the report, these desaturations were 10 to 14 times more frequent than what the navigators had planned, and each was recorded in a file produced by the application SM_FORCES. The specification required newton-seconds. The code provided pound-seconds, so that each thrust modeled by the navigators weighed less than a quarter of its actual value.

No one noticed the growing error

Neither calculation was wrong in itself. Lockheed produced exact values in its unit, and the JPL applied exact formulas for its own. Even the on-board version of the calculation, in metric, was correct: only the copy remaining on Earth was off, which left the probe better informed about its route than those who were piloting it.

Nine months of drift with no correction

Nine months of drift that no one read. Launched on 11 December 1998 from Cape Canaveral by a Delta II rocket, the probe was supposed to study the Martian climate. During the cruise, each misinterpreted firing slightly shifted the actual trajectory from the one calculated by the computers.

Unresolved issues led to disaster

Anomalies were detected in the ground software files, then discussed informally, without ever being resolved. No one linked them to the trajectory. On 15 September, eight days before arrival, a final correction was supposed to place the probe at 226 km from the surface for orbital insertion.

Final signal lost too early

On 23 September 1999, a signal lost 49 seconds too early. The main engine ignited at 9:00:46 UTC to slow the probe. The teams expected to lose contact when it passed behind Mars, but the signal disappeared at 9:04:52, 49 seconds earlier than expected. It was never recovered, even after the 21 minutes of announced occultation had passed.

Calculations with correct values show the error

The calculations redone with the correct values place the lowest point of the trajectory at approximately 57 km, 23 km below the survival threshold. The probe disintegrated in the Martian atmosphere.

Tiny error grew over time

On a journey of more than 600 million kilometers, the discrepancy of about 90 km with the low target represents less than a millionth of the journey. Mars Climate Orbiter therefore did not miss its target by chance, but due to a tiny error, repeated at each firing, which had nine months to grow.

Mission cost was significant

Mission cost: approximately 327.6 million dollars. The amounts cited vary depending on the scope, from 125 million for the probe alone to 327.6 million including the launch and operations. The loss also complicated the subsequent NASA Martian program, as the probe was supposed to serve as a communication relay for the Mars Polar Lander lander.

Investigation highlights multiple failures

What the committee criticized. Published on 10 November 1999, the report from the Mars Climate Orbiter Mishap Investigation Board identifies the direct cause: the non-compliance with the metric system in the coding of SM_FORCES. It does not stop there. It points out insufficient testing, inadequate communication between the JPL and Lockheed teams, and incomplete training of operators.

Further report on NASA management

All of these nets should have caught the error. At Lockheed Martin, propulsion engineers had become accustomed to expressing force in pounds, while space mission practice required a conversion to metric, which the JPL teams assumed was done. The committee extended its work with a second report, dedicated to NASA project management, dated 13 March 2000.