CubeBug and BugSat: learn before scaling up

Satellogic’s early missions form a useful example of learning through successive spacecraft. CubeBug-1, known as Capitán Beto, launched in April 2013; CubeBug-2, known as Manolito, followed that November. BugSat-1, called Tita, followed in June 2014. The company lists these as prototypes before the NewSat imaging series. These names identify separate vehicles in a development sequence, rather than three alternative names for one operational satellite.

The company’s retrospective describes engineers developing cameras and spacecraft hardware as they developed the system. A prototype — open definition can reveal an unexpected interaction that a component test missed. Its purpose should still be defined before launch: demonstrating a radio link, collecting an image and delivering a dependable customer service are different achievements. One successful experiment reduces a particular uncertainty; it does not prove every requirement of a later constellation.

Consider a fictional radio test with a requirement to recover at least ninety percent of one hundred transmitted packets during a defined contact. If the receiver decodes ninety-four packets correctly, the measured recovery fraction is ninety-four percent. That passes this stated threshold for that contact. It does not show that the same result will occur at every distance, orientation or ground station, because the test conditions limit the conclusion.

Now imagine that the spacecraft reports a healthy transmitter, but the ground receiver decodes only sixty packets. The team should investigate the whole link: antenna pointing, received signal strength, interference, receiver configuration and packet validation. Replacing a component at random is less informative than comparing observations with predicted behavior. These invented counts are a teaching exercise, not recovered CubeBug telemetry or a statement about an actual failure on any of the prototypes.

Separate a test from a service

Suppose engineers change the antenna mounting after the investigation. The next test needs a recorded hardware version, a repeatable procedure and a comparison with the earlier result. Otherwise an apparent improvement might come from a different ground station or a more favorable pass. Configuration control connects a lesson learned to the actual system that incorporates it; a notebook saying that something worked is not enough to reproduce the result.

Moving from prototypes to repeated production adds another question: can different units meet the same requirements? A successful first article does not remove the need for acceptance tests on later spacecraft. In a classroom review, write one requirement, one measurement and one remaining uncertainty for the fictional radio experiment. Keeping those three statements separate makes the historical progression understandable: early flights can build capability while leaving substantial work before a reliable imaging service exists.