NASA's Goldstone 70 meter antenna investigation finds software weaknesses, a previously damaged hydraulic limit, and site specific cultural pressure compounded into a single over rotation event during a Juno track — and points to specific, owned fixes.
On the night of Sept. 16, 2025, the 70-meter DSS-14 antenna at NASA's Goldstone Deep Space Communications Complex in the Mojave Desert was tracking NASA's Juno mission at Jupiter when its control system lost an accurate read on where the dish was pointing. The antenna rotated past its limits, fire-suppression water lines ruptured under stress, and water flooded portions of the structure. No one was injured. The dish has not transmitted since.
In April 2026, the Mishap Investigation Board submitted its final report to NASA leadership, and the agency accepted the report as final. The investigation classified DSS-14 as a Type A mishap — NASA's most serious category — based on total damage cost, and assigned an estimated repair range of $4.1 to $4.6 million pending a full systems assessment.
The report's central value for readers outside the agency is not the dollar figure. It is the specific causal chain investigators say produced it, and the named, owned corrective actions that follow. Both deserve to be read carefully rather than skimmed past.
Investigators describe a sequence that begins the day before the over-rotation and compounds at each step.
A Sept. 15 electrical issue left DSS-14's antenna control system misreporting the dish's actual rotation state. When operations resumed the next day during the Juno track, that misreport repeatedly triggered the antenna's built-in limit stops. The hydraulic limit system — designed as the final mechanical safeguard against over-rotation — failed to engage.
Troubleshooting intended to clear the false limit-stop alarms instead bypassed software and hardware safeguards that would normally have stopped the drive. With the limits already passed, an attempted stow command to park the antenna drove the structure further past its safe envelope. The result was the over-rotation, the ruptured water lines, and the flooding described in the agency's release.
The hydraulic limit's failure was not a sudden degradation. According to the final report, the safeguard had been inoperable on Sept. 16 because of damage from an undocumented prior incident, and it had not been adequately tested for an undetermined period before the Juno track. The mechanical backstop that should have caught the misreport chain was already out of service, and the system had no procedure that would have caught that absence before operations began.
The board named three categories of root cause: software weaknesses, human error during the troubleshooting that bypassed safeguards, and the undetected failure in the hydraulic limit system. Each is a specific, technical claim, not a general safety finding.
Layered on top of the technical causes, the report identifies procedural and cultural contributing factors. Procedures were inadequate in places; operations relied on undocumented practices and tacit knowledge passed between personnel; antenna control logic contained gaps that were not surfaced by the procedures that should have exercised them. Workplace pressure, the board found, pushed operators at Goldstone to work expeditiously and beyond their usual roles, expertise, and training.
That cultural finding is site-specific. The board explicitly stated these conditions were not present at the network's other two DSN complexes, in Canberra and Madrid. The fix to one antenna is therefore not a fix to a uniform problem; it is a fix to a localized set of conditions that allowed a particular failure mode.
The recommendations, as summarized by the agency, emphasize training, technical rigor, operational procedures, system design, clear roles and responsibilities, and safety assurance. The agency says it is applying the lessons from the Goldstone investigation across all three DSN sites, not only at Goldstone — though active implementation at Canberra and Madrid has not been independently confirmed beyond NASA's own reporting.
Two named officials spoke on the record in the agency release. Joel Montalbano, acting associate administrator of NASA's Space Operations Mission Directorate, and Kevin Coggins, deputy associate administrator of the SCaN Program, are the public faces of the response.
DSS-14 itself remains offline. It will enter a previously scheduled extended maintenance and upgrade period originally set to begin in August, with the agency projecting completion in October 2028. The 2028 date is NASA's current expectation rather than a hard contractual milestone, and downstream mission planning that depends on DSS-14 capacity should be attributed to the agency's stated timeline rather than to a confirmed return-to-service.
The Deep Space Network's remaining 13 antennas, split between California, Australia, and Spain, are continuing to provide full coverage for more than 40 missions, according to the agency's release. A dedicated scheduling team is maintaining coverage during the outage. The redundancy is real, and it is being actively managed, not relied on passively.
The final report is published in redacted form, with proprietary and privacy material removed. Specific operator names are not in the public release, and details of the control-logic interactions that produced the misreport chain are not disclosed at a level an outside engineer could fully audit. The undocumented prior incident that damaged the hydraulic limit is described in the report's findings but not in operational detail.
For an outside reader evaluating the investigation, that means the chain of causes is NASA-characterized. The board's work is internal to the agency, the technical claims have not been independently re-examined by an outside engineering body in the materials available here, and the corrective actions are NASA's commitments to itself. None of that invalidates the report. It sets the right frame for reading it: the agency has named the failures, but the verification of those named failures, beyond NASA's own concurrence, lies elsewhere.
Three things will tell readers whether the report's recommendations are actually being absorbed.
The first is testing protocol. The hydraulic limit was inoperable because of an undocumented prior incident and an absence of adequate testing. If the corrective action plan produces a test cadence that would have flagged that absence on Sept. 15, the fix addresses the specific failure mode rather than the general one.
The second is procedural documentation. The board found reliance on undocumented practices and tacit knowledge. The corrective action's value is tied to whether tacit knowledge is converted into written procedure, or whether the corrective action substitutes a different form of training that still relies on institutional memory.
The third is the site-specific cultural finding. The board said the workplace conditions at Goldstone were not present at Canberra or Madrid. A network-wide corrective-action plan that does not address the localized pressure at Goldstone is, on the report's own terms, incomplete.
The October 2028 return-to-service window is the visible clock. The work between now and then is the part that will determine whether the next DSS-14-class event reads as a system that learned, or as a system that replaced one set of undocumented practices with another.