Over at the Usagi Electric farm, [David Lovett]’s custom 1-bit, vacuum tube-based computer (UEVTC for short) has been coming along well the past years, matching and exceeding the Motorola MC14500B 1-bit industrial control unit (ICU) that it is heavily inspired by. What is still missing, however, is a faster way to get data into the computer than manually toggling switches. The obvious choice is to make a (punched) paper tape reader, but how does one go about this, and what options exist here? With a few historical examples as reference and the tape reader on the impressive 1950s Bendix G-15 which [David] happens to have lounging around, [David] takes us in a new video through the spiraling complexity of what at first glance seems like a simple engineering challenge.
Punched paper tape saw significant use alongside punched paper cards and magnetic tape, and despite their low bit density, if acid-free paper (or e.g. mylar) is used, rolls of paper tape should remain readable for many decades. So how to read these perforations in the paper? This can be done mechanically, or optically, with in both case the feedrate an important consideration.
Right off the bat the idea of a mechanical reader was tossed out due to tape wear, with [David] digging into his stack of photodetector tubes. After looking at a few rather clunky approaches involving such tubes, the photodiodes in the Bendix G-15’s tape reader were instead used as inspiration for a design. These are 1.8 mm diameter photodiodes, which aren’t super common, but have the nice property that they align exactly with the holes in the paper tape.
This left building a proof-of-concept on a breadboard with some incandescent bulbs and one of the photodiode to demonstrate that a valid logic signal could be produced. This turned out to be the case, clearing the construction of the actual tape reader, which will feature in upcoming videos.
Instead of reling on small photodetector that are not always available, use small leds, that are much more common, and a single photodetector.
You can also use the LEDs as a photodetector.
Both cases can be derived from TRIZ inventive principle #13: “Do the opposite action”
Wouldn’t using a single photodetector require some kind of synchronised movement to sense each position? Unless we are talking something like DTMF for holes and the sensor could detect different light frequencies…
Just light each possible hole position individually, measuring the output of the sensor after each.
Sort of addressed in the video, at around 16:30. Using a single detector would require a shift array, which would be a lot of vacuum tubes.
Not to mention using, being period correct, using incandescent lights sequentially will be horribly slow due to waiting for the filaments glow to diminish between bits
Commented before watching the full video.
Given the historical constraints, other principles may apply, for instance, (26/27) ‘copies’/’short living objects’: have many copies of tapes (to be used with the cheap, mechanic reader mechanism).
Maybe principle (17) ,’another dimension’: use a different tape, wider.
Did you miss the part where [Usagi Electric] is rather obviously trying to do this with as many period-correct parts as possible? Please point out which LEDs were available when vacuum-tube computers were around. Go on, I’ll wait.
Yes, missed that constraint in the article.
He wants to be period correct, so cannot use fancy LEDs. See 12:30 on the video.
Well you have cfl
Do they still make incandescent Christmas lights? Those can light up even on 5-6v, use some metal to focus a line across the strip
For each input you have a photodector
With one bit, a clk pulse, you can use a solar panel or two and op amp or two photodiodes/photodector, they did have solar panels back in vacuum tube days, you do realize tubes were used in computer even in 70s and 80s…they didn’t just disappear when Atari and applied and Altair showed up
Use it to encoded as a differential pair of bits, some anti noise and error correct, and sensitive enough to not even use punched holes, just a high enough dpi laser or inkjet printer, better storage density….
You’ve gotten confused. The problem at hand is to detect the light with period-appropriate hardware, not to generate it.
The suggestion for leds was in using them as photodetectors, having missed the bit about “period-appropriate”. Although, if you can use modern photodetectors, you also can readily find ones which will fit…
I think the point of a vacuum-tube computer is to keep this period-specific. So, if LEDs are even available in the tech base they might be IR only.
That’s how it’s done normally
Same with tv remotes
what is the specific period for an mc14500?
Reader? Easy. Now create the punch to go with it.
Paper tape readers and thermal laminated tape printers are a recurring theme on Hackaday. I am waiting for a maker to exploit a cheap laminated label tape printer to use in conjunction with a DIY photo-reader.
Electrochemical printing using potassium iodide etc has been around since to 1800s according to this page on the fascinating WWII British ‘Telewriter’ portable teleprinter:
https://www.nonstopsystems.com/radio/hellschreiber-modes-other-mfrs-UK.htm
And here’s how to build a Morse Code paper tape printer that uses potassium iodide, from the Australian ‘Hobbies Illustrated’ magazine from 1947:
https://www.surfacezero.com/g503/showfull.php?photo=182309
I experimented with building this exact unit back in the late 70s, the iodide paper strips went brown when I applied voltage to a simple print head as described.
Now I wonder if the dark colour of the decomposed iodide produced by this type of printer would be dark enough to be suitably read by Usagi’s proposed reader?
Selenium rectifier plates ( with the paint removed) are photo sensitive. This type of photo detector was used in old time Weston exposure meters. You can score the selenium layer to make several parallel photo detectors. I did this many, many, (60 years) ago. Worked great in the sunlight. I didn’t try it with any other light source, but his light bulbs should work.
Cool project! In 2022 I was working at a machine shop that was still using early CNC machines that read paper tape. They had the whole setup to program and print and run. Some of those parts we made are in space right now so not bad for old school tech. Don’t get me wrong new machines can hold tighter tolerances but you work with what’s available. I haven’t been back to the shop since 2023 but presumably they’re still turning parts until they can’t source any more tape.