IBM 029 Keypunch

IBM 029 Keypunch

The keypunch was the main form of data and program entry in an era when keyboards were not directly attached to computers. One would first punch a line of 80 characters text or data on a punch cards, assemble a stack of punch cards into a program, each line being a card, and feed it to the computer for compilation via a card reader. The IBM Model 029 was IBM's latest generation of keypunch in 1964. It is a descendant of the IBM 024 non-printing, then the IBM 026 printing keypunch. It shares much of its basic mechanism with these two earlier machines. No more tubes in the machine though, it is entirely operated by relays.

Most importantly, the IBM 029 introduced the expanded character set beyond the original 42 character "BCD" scheme punched by the 024 and 026. Now called the EBCD (for Extended BCD) the set had 62 characters. It included new codes and new characters that we needed for programming. Most of these new characters used new 3-holes punch codes. It was a superset of both the BCD Commercial and BCD Fortran sets that could be had on the 026 (you had to choose one or the other).

Here is the EBCD set of character and codes punched by the 029. Note the many added 3-punch special characters added at the end of the card.

Which can be compared with the BCD set punched by the IBM 026 "Commercial"

And the BCD set punched by the IBM 026 "Fortran"

There are two main implementations of the 029, one with Reed relays which were found to be unreliable, and was eventually re-implemented with the tried and true telecom inspired wire relays. Mine fortunately had the wire relays.

My unit, rescued by Carl Claunch during a haul in Kansas in 2014 (see, turned out to be a C model with the Interpreter feature. It can take a card that has been already punched but with no print on the top line, such as cards that would be produced by our IBM 1402 high speed punch, read them in and add the missing print at the top.

The machine arrived in pretty good cosmetic condition, but absolutely nothing worked. Just about function in my machine had failed, mostly because of bad contacts, of which there a large amounts spread all around the machine: card sensors, cams, end of travel sensors, card hole detectors, they were all bad. The wires in the printing head were also stuck in gooey ink, and the keyboard developed a contact fault. However, it forced me to learn and understand every part of the machine.

It also came with a pretty complete documentation, which I have scanned and is reproduced below.

Video Series

The 10 episodes of the IBM 029 Keypunch restoration cover everything from getting the non-functional machine to having it fully restored and functional.


The documents below include the scans of the documentation that came with the machine, and is pertinent to the Model C with the Interpreter feature.