Paper Punching Machine Looks Like Cute Piece Of Computer History Past

Computing used to run on punch cards. Great stacks of cards would run middling programs, with data output onto more punched cards in turn. [Nii] has built a machine in this vein, capable of punching binary into paper tape. 

The machine is run by a stepper motor, which is charged with feeding the paper tape through the machine in steady steps. A series of vertically-actuated solenoids punch holes in the paper tape as directed. The machine buzzes and clicks away like the best electromechanical computing devices of the mid-century era.

To what end, we couldn’t possibly say. One user noted the machine was punching seemingly random binary into the paper tape, and [Nii] has not provided any explanation as to the machine’s higher purpose. Regardless, whatever it is doing, it looks like it’s doing it well. Feel free to speculate in the comments.

Impressively, the petite device will be demonstrated at MF-TOKYO, the 7th Annual Metal Forming Fair in Tokyo this year. We’re sure the clickity-clack will be muchly appreciated in person.  Video after the break.

22 thoughts on “Paper Punching Machine Looks Like Cute Piece Of Computer History Past

    1. I wonder if this is to make tape for some old machine (cnc, textile loom, etc) to keep it running.

      seems unlikely. in the good ol’ paper tape, the sprocket hole is smaller than the data holes; some readers used that fact to clock in the data.

      1. Reminds me of the paper tape machines in old newspaper offices, circa early-mid 1980s. I remember making “confetti” by running the machine with the gibbberish typing of a 6 year old lol.

  1. It would be rather interesting to see something like a Z80 running code directly off tape (rather than, e.g. buffering the tape’s contents into RAM)… crazy-extended use of the WAIT signal! Of course we’d be talking IPS, or maybe even SPI, instead of MIPS or KIPS… Of course, the addressing system would be a feat of TTL, heh. And coding with minimal jumps would probably be ideal.

    1. CDP1802 had the LOAD mode which allowed direct streaming of bytes from external device (paper tape reader, hobbyist flipping switches like on the original “ELF”) directly into RAM (CPU R0 counted up at each byte receive signal, this was the address to write to). No ROM was needed, this was inspired by the mini computers of the era which other commenters mentioned here.

      An “infinite” paper tape loop (two ends tied together) in fast rotation could be a “ROM” – when the counter driven by sprocket matches the address bus, WAIT is released, otherwise it is asserted. In a way, this is how floppies or magnetic hard drives work – each track is a 1-bit wide bitstream that repeats at each rotation, and the read/write is engaged when the sector under the magnetic head matches the sector number coming from the controller.

  2. FWIW… If this machine operates like some of the old mainframe equipment I’ve read about, it is NOT the solenoids that actually punch the paper. Rather, the purpose of the solenoids is to lock a moving “knuckle” associated with each corresponding punch/die. Locking the knuckle allows force to be transferred to the punch from the real driving force, the stepper.

    The stepper thus performs two functions: It provides motive force for the punches and also the tape feed.

    1. Was going to post the same comment. This unit whilst very nicely made is not like the best punches back in the day. Having the solenoids punch directly is very hard on the units and they would overheat at high speed.

      As pointed out above the original devices had a beefy motor that drove a cam and the solenoids acted much like the base of a mechanical transistor, moving small levers in and out of the cam path for each punch row cycle. This is how they achieved a high punch throughput.

      You can see this in the illustrated Roytron punch manual (mechanical section) at

  3. I began my career as a product engineer at Motorola Semiconductor in Phoenix the early 1970’s. The final test equipment for our TO-92 transistors was controlled by PDP-8, and later PDP-11, computers and the test routines (drive levels, spec limits) for those computers were programmed via paper tape almost exactly as shown in the picture. Each line had places for 8 holes, with a separate indexing hole just like above. Each transistor part number had its own little roll of tape … just a few turns about an inch in diameter and secured by a paper clip. We stored the rolls in an open faced wooden cabinet with a bunch of little cubby holes for the tapes. The tapes were read by an optical reader connected to the computers. We were constantly punching different tapes to change the specs and their priority according to the mix of products in the current backlog. The only difference was that our tape was black.

    It was actually much more efficient than punch cards.

  4. “To what end, we couldn’t possibly say”… really? Didn’t you at least do some research? The PDP-8 I used in college in the early seventies could load and write programs on paper tape, the ASR 33 teletypes we used had a tape punch right beside the keyboard. Someplace around here I have a BASIC and a few games on paper tape, left over from the old days.

    1. I learned programming on an Elliot 803 (1960’s), normal input was paper tape only. I never saw the computer, you slowly typed and re-typed your program on a teletype, left your final version paper tape with the computer guardians and next morning picked up a pile of fan fold print-out if you were lucky or your paper tape back for debugging if you were unlucky.

  5. Not exactly a gnt or fanuc high speed punch but good enough for a short program or message. Paper tape and cards are a lost art but paper is the one medium proven to last thousands of years.

    1. I am in the throws of completely restoring one of these (PC11). The really neat bit that helps this machine go so fast is that in part of the punch cycle the cam moves all of the solenoid plates onto the electromagnets. The selected solenoids are then energised to hold them in that position. This means that the energy to draw the plate onto the coil is bypassed and just holding current is needed. This also means that they are perfectly synchronised at punch time.

      1. I remember another approach (must have been Teletype). The code bars would initially be at all zero state, which retained a group of spring-loaded pawls, one for each code bit. If the code bit was a “1”, the code bar retracted, allowing the pawl to move upwards under spring tension. In the upper position, the pawl enabled the corresponding punch pin. A cam actuated all the enabled punch pins simultaneously, while a second cam reset all the pawls and codebars to their “0” positions.

        Teletype mechanisms were very clever. I repaired them in college as a part time job, back when ASR-33s and mainframes were “a thing”. They took an incredible amount of abuse and were quite easy to disassemble and repair. I don’t think Teletype ever foresaw their use by college students, but they were the ideal terminal: cheap, easy to use, and produced in vast quantities. Video terminals were better, but they had to wait for microprocessors.

Leave a Reply

Please be kind and respectful to help make the comments section excellent. (Comment Policy)

This site uses Akismet to reduce spam. Learn how your comment data is processed.