Reading the Boundary of an Apollo Computer Replica

Graphite illustration of an unmarked display-and-keypad console connected to an open processing module with a cyan circuit board.

Filed under: Workshop Notes

Type VERB 35 ENTR into an appropriately configured Apollo display-and-keyboard interface and the documented operation is a lamp test. It is an appealing place to begin: the panel responds, the lamps become the subject of attention, and a historical computer seems suddenly approachable. It is also the moment to ask what, exactly, the demonstration has shown.

A convincing enclosure, a responsive keypad and a running emulator are related achievements. They are not the same achievement. Following a simple command through those layers makes a replica more interesting than treating its appearance as proof that an entire spacecraft system has been recreated.

The panel is an interface to the computer

The Virtual AGC project’s DSKY introduction describes the unit as the display and keyboard through which a user supplies input to the Apollo Guidance Computer and receives visual output. It explains a numbered verb-and-noun interface: a verb requests an operation and a noun identifies the associated data. Its basic command list includes V35E for the lamp test and V16N36E for displaying ground elapsed time.

Read the latter sequence as a compact instruction rather than a mysterious string. The action and the data are encoded separately; the final Enter submits the request. The same visible digits on a panel can therefore play different roles depending on where they appear and what the software is asking the operator to enter.

That distinction gives a workshop its first useful exercise. Ask participants to explain the roles of the input before trying to memorise it. Someone who understands which part specifies the operation can ask a better question about an unexpected response than someone who only remembers a sequence of buttons.

A modern reconstruction preserves some things by changing others

Raspberry Pi’s 22 October 2026 account of Michael Newton’s Recreate Apollo project describes a full-size printed DSKY powered by modern electronics and Virtual AGC software. Newton adapted internal mounting and display support to accommodate the components. The article presents the build as a combination of physical reconstruction, electronics and software work.

Those adaptations are worth discussing rather than hiding. A mounting change may help preserve the external arrangement while departing from the original internal construction. Software emulation addresses a different question again. The object can be valuable as an interface reconstruction without the casing itself demonstrating the behaviour of the original computer.

For a visitor, the external panel is the part available to touch and inspect. For a builder, its response depends on a chain that extends beyond that surface. Keeping both viewpoints visible is a useful way to teach what a replica represents.

Trace a request through three boundaries

Consider a hypothetical workshop in which a key is pressed but the expected response does not appear. No fault in Newton’s build is being reported here. The example is a way to choose what to investigate next.

Questions for an invented input-to-display investigation
Boundary Question What the answer would establish
Physical input to interpreted key Did the interface register the key that was actually pressed? Evidence about this input event, rather than the meaning of the whole command.
Key sequence to software response Did the running configuration receive the intended sequence in the relevant state? A basis for comparing the response with the matching documentation.
Software response to visible output Did the expected output reach the display as intended? Evidence about the presentation path, rather than every calculation inside the system.

If the wrong key is registered, changing the command description will not repair that input problem. If the intended sequence reaches the software but the configuration does not match the guide, replacing a display may address the wrong boundary. The point is not to prescribe a particular diagnostic tool. It is to make the competing explanations separable.

Two similar sequences can ask different questions

The same Virtual AGC guide lists V25N36E as an operation for setting ground elapsed time. Compare it on paper with V16N36E. The noun is the same, but the verb changes: asking to see a value and asking to set it are different interactions.

This pair offers more teaching value than a long list of unexplained codes. Give a participant the two documented descriptions and ask which sequence is appropriate for a read-only question. Then ask what further instructions would be needed before entering data. A command’s short notation can identify an operation without containing the entire procedure for using it correctly.

For an actual replica session, the builder should select documentation that matches the installed software and establish the starting condition. The exercise here is an interpretation of the published guide, not a claim that we have run either command on Newton’s hardware.

What a lamp test can — and cannot — settle

A successful, documented lamp test can provide evidence about a particular interaction and its visible response. It does not by itself establish navigation accuracy, the availability of spacecraft sensor inputs or the behaviour of a simulated mission. Those claims require observations aimed at those functions.

Conversely, a demonstration does not become worthless because its scope is narrow. If a participant can explain why the display is not the whole computer, identify the operation and data in a command, and locate the next boundary to investigate, the replica has supported a substantial lesson.

End the workshop with a precise account: which configuration was used, what starting condition was established, which interaction was attempted and what was actually observed. Keep predictions separate from results. The historical panel then becomes a way to reason about a working system, rather than a prop whose familiar appearance is asked to prove everything at once.

This article reads the linked build account and Virtual AGC introduction. The fault investigation is hypothetical; no hardware test, emulator run or mission simulation was performed for this publication.