Did the spacecraft itself use imperial units?

The critical mismatch occurred in the ground software/data interface. SM_FORCES produced impulse data in pound-force-seconds, while the AMD file interface required newton-seconds and the navigation team consumed the numerical values assuming that requirement had been met.

Why is the factor 4.45?

One pound-force equals approximately 4.44822 newtons. Because the time unit is seconds in both lbf·s and N·s, the impulse conversion carries the same factor. A numerical value of 1.0 lbf·s corresponds to about 4.45 N·s, not 1.0 N·s.

Was 57 km the planned orbit altitude?

No. TCM-4 was designed to produce a first periapsis of 226 km after MOI. During the following week navigation solutions moved downward to 150–170 km and then as low as 110 km about an hour before MOI. The 57 km value was calculated only after the loss using corrected small-forces values.

Could TCM-5 have saved the mission?

The mishap board listed TCM-5 not being performed as a contributing cause. NASA's contingency-maneuver lesson says the maneuver was not in the baseline operations timeline, had not been flight-sequence/safety tested, and was absent from operational readiness testing. It therefore should not be presented as a guaranteed rescue.

What should engineers take from this case today?

Use explicit unit-aware interfaces, automated contract tests, independent calculations, closed-loop anomaly tracking, well-rehearsed contingency procedures, and systems engineering that spans organizational boundaries. The goal is not merely to avoid one conversion mistake; it is to ensure ordinary mistakes are detected before they can propagate.

Mars Climate Orbiter Unit Conversion Failure

01 // Interactive mission failure replay

Select a stage or run the sequence.

Mars Climate Orbiter — navigation reconstructionMission milestones and periapsis values are documented; geometry and trajectory shapes are schematic and not to scale.

The Phase I board retained uncertainty about the final fate: MCO either was destroyed in the atmosphere or exited back into heliocentric space after atmospheric passage. AMD-event markers and all trajectory geometry are illustrative, not event counts or a scaled orbit reconstruction.

Navigation / mission state

Replay encoding

Planned / target geometry

Navigation solution / modeled path

Corrected post-loss trajectory

Small-force / AMD events

Key mishap-board facts

Launch mass629 kg

MOI main-engine burnplanned 16 min

TCM-4 target periapsis226 km

MCO minimum survivable80 km

Corrected periapsis57 km

Unit factor1 lbf·s = 4.45 N·s

Mission event console — latest stage first

02 // Unit-interface failure explorer

See exactly what the 4.45× mismatch means.

Interface behavior

SM_FORCES impulse value — lbf·s

1.0 lbf·s

Illustrative number of AMD events — not a historical count

20 events

Physically equivalent impulse

Value navigation consumes

Modeled fraction of impulse

Scalar aggregate mismatch*

Interface check

Magnitude error

Model boundary. This tool demonstrates unit semantics only. The event-count slider is illustrative: the board reported that AMD events occurred 10–14 times more often than the operations navigation team expected, but it did not publish a single historical count here. The scalar aggregate mismatch is not a spacecraft Δv or a trajectory calculation.

03 // Why small forces became a large navigation error

The problem accumulated through repeated, under-modeled momentum-desaturation events.

01 / ASYMMETRIC SPACECRAFTMCO's single solar array created a solar-pressure torque unlike Mars Global Surveyor's more symmetric configuration.

02 / MORE AMD EVENTSReaction-wheel desaturation events occurred 10–14 times more often than the operations navigation team had expected.^[1]

03 / WRONG UNIT AT INTERFACESM_FORCES output entered the AMD file in lbf·s even though the interface documentation required N·s.

04 / ΔV MISMODELEDNavigation consumed the numbers as metric, under-modeling each small impulse by about a factor of 4.45.

05 / DISCREPANCY NOT CLOSEDDoppler residuals and different orbit-determination solutions indicated a problem, but the mismatch was not resolved before Mars arrival.

04 // Navigation figures

Documented milestones only — no invented continuous trajectory.

First-periapsis knowledge before and after loss

Target / plannedNavigation estimateCorrected post-loss estimate

View milestone data

Milestone	Periapsis	Basis
TCM-4 target	226 km	Computed 8 Sep; TCM-4 executed 15 Sep
Week before MOI	150–170 km	Operations navigation processing
~1 hour before MOI	as low as 110 km	More accurate tracking data
Post-loss corrected estimate	57 km	Corrected small-forces values and all data through loss of signal
Minimum considered survivable	80 km	Mishap-board threshold

The 150–170 km value is a documented range, not a single point. The chart therefore renders it as a range bar rather than inventing an intermediate estimate.

Unit mismatch magnitude

Physical impulse equivalentNumerical value if passed through unchanged

View conversion

Input	Correct metric value	Faulty consumed value	Modeled fraction
1.000 lbf·s	4.448 N·s	1.000 N·s	22.48%

The Phase I report rounded the force conversion to 1 lbf = 4.45 N. The calculator uses the standard conversion 4.44822 for the educational arithmetic.

05 // What actually failed?

This was not an onboard spacecraft software unit mix-up. The mishap board identified a ground-software/interface failure.^[1] Lockheed Martin's ground SM_FORCES application reported the impulse-bit data placed in the AMD file in pound-force-seconds, while the Software Interface Specification required newton-seconds. The downstream JPL navigation processing treated the AMD data as though the specified metric units had been supplied.

The unit error was the root cause, not the whole story

The Phase I board listed eight contributing causes:^[1] undetected mismodeling of spacecraft velocity changes; navigation-team unfamiliarity with the spacecraft; TCM-5 not being performed; inadequate systems engineering during the transition from development to operations; inadequate communications; insufficient operations-navigation staffing; inadequate training; and verification/validation that did not adequately address ground software.

There were warning signals

During spring and summer 1999, working-level concerns existed about differences between navigation solutions.^[1] Doppler residuals associated with the more frequent AMD events were noted but only informally reported. During Mars approach, Doppler-only solutions consistently indicated a closer flight path than combined solutions. The discrepancies were not resolved.

TCM-5 was a missed contingency opportunity

TCM-5 was a contingency option,^[4] but NASA's lessons-learned record states that it was not included in the baseline mission-operations timeline, had not been tested for flight-sequence compatibility or mission safety, and was not included in operational readiness testing. With no compelling evidence judged sufficient to accept that late-maneuver risk, the project did not execute TCM-5. The Phase I board listed “TCM-5 not performed” as a contributing cause.

Final fate remains technically uncertain in the Phase I report

After correcting the small-forces data, the board estimated a 57 km initial periapsis, below the 80 km minimum considered survivable.^[1] The Phase I report states that MCO either was destroyed in the atmosphere or re-entered heliocentric space after atmospheric passage. Later NASA summaries commonly describe the spacecraft as destroyed, but this page preserves the board's more cautious wording.

06 // Key mission and mishap data

Parameter	Value / finding
Mission	Mars Climate Orbiter
Launch	11 Dec 1998
Launch mass	629 kg (NASA press kit: 338 kg spacecraft + 291 kg propellant; some secondary sources quote 638 kg)
Prime contractor	Lockheed Martin Astronautics
NASA lead / navigation	JPL
Science instruments	MARCI + PMIRR
MOI planned burn	16 min
TCM-4 target first periapsis	226 km
TCM-4 executed	15 Sep 1999
~1 h pre-MOI estimate	as low as 110 km
Minimum considered survivable	80 km
MOI engine start	09:00:46 UTC, 23 Sep
Loss of signal	09:04:52 UTC
LOS timing	49 s earlier than predicted
Expected occultation	21 min
Unit mismatch discovered	29 Sep 1999
Corrected periapsis estimate	57 km

Mission/event values in this table are drawn primarily from the NASA Phase I mishap report [1]; mission overview/instrument naming is cross-checked against NASA/JPL [5].

07 // The interface that failed

The specification was metric; implementation and verification did not enforce it.

Required interface behavior

Specified unitN·s

Checkunit-aware

Resultconsistent

Observed failure path

Delivered unitlbf·s

Magnitude error×4.45

Detectionfailed

08 // Eight contributing causes identified by the Phase I board

The unit mismatch was the root cause. Mission loss required multiple engineering and organizational barriers to fail around it.

1 / Velocity mismodeling not detectedSmall spacecraft velocity changes were modeled incorrectly and the discrepancy persisted.

2 / Navigation team unfamiliar with spacecraftThe team underestimated how the spacecraft's asymmetric solar-array configuration drove more frequent AMD events.

3 / TCM-5 not performedThe contingency TCM-5 maneuver was not executed before orbit insertion.

4 / Development-to-operations systems engineeringThe transition did not receive adequate systems-engineering attention.

5 / Inadequate communicationProject elements did not communicate discrepancies and interface assumptions effectively.

6 / Operations navigation staffingThe board identified inadequate staffing as a contributing cause.

7 / TrainingTraining was not sufficient for the operational and software-review demands of the mission.

8 / Ground-software verification & validationV&V did not adequately cover the software path that generated and consumed the small-forces data.

Defense in depth: controls that should stop this class of failure

InterfaceMachine-checkable unit contracts

Carry units as explicit metadata or types. Reject a lbf·s value at an N·s interface rather than accepting an untyped number.

VerificationEnd-to-end interface tests

Test producer → file → consumer chains with known values and independently computed conversions.

NavigationIndependent solution reconciliation

Define thresholds that force closure when Doppler-only, range-only and combined orbit solutions diverge materially.

OperationsClosed-loop anomaly reporting

Informal residual concerns should become tracked anomalies with owners, disposition criteria and documented closure.

ContingencyTCM-5 readiness

Late correction maneuvers must have validated procedures and decision gates before the mission reaches the critical window.

OrganizationSystems engineering across interfaces

Maintain enough staffing, training and mission-systems ownership to bridge contractor, navigation, software and operations teams.

09 // Historical chronology

Date / time	Event
1995	Mars Climate Orbiter and Mars Polar Lander identified for the 1998/1999 launch opportunity.
11 Dec 1998	MCO launches from Cape Canaveral on a Delta II.
Spring–summer 1999	Working-level concerns arise over differences between navigation solutions and Doppler residuals associated with AMD events.
8 Sep 1999	TCM-4 computed to target a 226 km first periapsis after MOI.
15 Sep	TCM-4 executed as planned.
Week before MOI	Navigation processing indicates first periapsis has fallen to roughly 150–170 km.
~1 h before MOI	More accurate tracking solution indicates first periapsis as low as 110 km.
23 Sep, 09:00:46 UTC	Mars Orbit Insertion main-engine burn begins.
09:04:52 UTC	Carrier signal disappears at Mars occultation, 49 seconds earlier than predicted.
After expected 21 min occultation	Signal is not reacquired; search continues through 25 September.
27 Sep	Operations navigation consults spacecraft engineers about ΔV modeling discrepancies.
29 Sep	Unit mismatch discovered: AMD impulse data delivered in lbf·s instead of required N·s.
15 Oct	NASA establishes the Mishap Investigation Board.
10 Nov 1999	Phase I report identifies root cause and eight contributing causes.
13 Mar 2000	Phase II project-management report expands the lessons to systems engineering, communication, mission assurance and risk management.

10 // Sources and technical basis

This page uses the NASA Mars Climate Orbiter Mishap Investigation Board Phase I Report as the primary source for mission chronology, navigation values, root cause and contributing causes. The Phase II report and NASA Lessons Learned records provide project-management and software-process context.

[1] NASA — Mars Climate Orbiter Mishap Investigation Board, Phase I Report (10 Nov 1999) — primary source for the mishap chronology, root cause, navigation milestones and contributing causes.
[2] NASA — Mars Climate Orbiter Mishap Investigation Board, Phase II Report (13 Mar 2000) — systems-engineering, project-management, communication, verification and risk-management lessons.
[3] NASA Lessons Learned 740 — Deficiencies in Mission-Critical Software Development for MCO.
[4] NASA Lessons Learned 916 — Inadequate MCO Contingency Maneuver Planning — source for the TCM-5 readiness/decision details.
[5] NASA/JPL — Mars Climate Orbiter mission overview — mission purpose, launch date, instruments and current NASA/JPL summary.
[6] NASA Science — Mars Climate Orbiter mission archive.
[7] NASA Lessons Learned 886 — Mission Assurance During Mars Climate Orbiter Operations — anomaly reporting, lack of end-to-end AMD-file compatibility testing, and unresolved ΔV discrepancies.

Interpretation boundary

The mission replay and unit explorer are explanatory reconstructions. They do not integrate spacecraft equations of motion, reproduce JPL orbit determination, derive the 57 km periapsis, or reconstruct individual AMD event magnitudes. Only values explicitly identified as historical milestones are presented as mission data.

11 // Engineering lessons learned

Mars Climate Orbiter is a systems-engineering case study disguised as a unit-conversion mistake.

Units are part of the data typeA bare number at an interface is incomplete. Quantity, unit, sign convention, reference frame, coordinate system and epoch may all be part of its engineering meaning.

Interface documents must be executable in practiceWriting “N·s” in a specification is not enough if producer, tests and consumer do not verify the requirement end to end.

Small repeated errors can dominateEach AMD event imparted a small impulse, but systematic under-modeling accumulated over a nine-month cruise.

Disagreement between independent solutions is informationPersistent Doppler/range residuals should trigger reconciliation, not normalization of deviance.

Ground software can be mission criticalThe spacecraft performed nominally. Mission loss came through the ground navigation/software chain, demonstrating that criticality follows function, not physical location.

Contingencies must be executable, not theoreticalA maneuver that exists in the mission plan but lacks completed analysis, test, procedures and decision authority may not be usable when the window arrives.

Transition to operations needs systems ownershipInterfaces that were acceptable during development can become mission risks when teams, tools and responsibilities change at handover.

Simple errors require strong organizational barriersThe board's lesson was not that engineers must never make mistakes; it was that projects need processes capable of detecting ordinary mistakes before they become catastrophic.

12 // Frequently asked questions

Common misconceptions and practical lessons.