Building a Braille Printer
A project from 1980
the Making Stuff collection
In the late 1970s, I worked as a research assistant in the Medical Computing Laboratory of the University of Illinois School of Basic Medical Sciences in Urbana, Illinois. The lab had significant experience with S-100 computers (we built Altair serial number 34 and used it as a 4-channel protocol converter). Working with Professor Peter Maggs in the Law School, the Lab configured a Sol 20 computer as a speaking computer terminal; Maggs had done some of the text-to-speech work on the Kurzweil Reading Machine. The lab hired a blind computer operator who used that terminal, and he found it frustrating because he had no way to make hard copy.
We looked into the possibility of buying a Braille printer, or embossing machine, and we found two on the market, but they were expensive and did not offer very impressive performance. That led me to suggest that we try to build a Braille printer. Andrew Appel and I did all of the machine-shop work and most of the software for Grade 1 English Braille. It is not clear how much progress we made on Grade 2 Braille, but our printer could emboss about 13 characters per second and we were confident that improved control software could do better.
42 years later, as I was cleaing out my office at work, I found the report that Andrew Appel and I wrote about the printer before I left the University of Illinois in August, 1980, along with photos I took of the machine during construction. At least one drawing has gone missing, along with the page of Braille embossed by our printer that I had kept as a souvenir.
A scanned PDF of our report is included here. I used free OCR software, ocrmypdf to make the PDF somewhat accessible, and on exploring the result, I found that it frequently scrambled the word order. On seeing my results, Andrew Appel tried using Adobe Acrobat Pro to OCR the text. The result was also marginal, Acrobat broke many words into fragments. Since a document about a Braille printer ought to be accessible, I converted the text to HTML and manually repaired it. While I was at it, I extracted the drawings from the PDF and tagged each drawing with the part number from the parts list in the report.
Curiously, the HTML text will still cause problems with screen readers, and the reason is, it contains Braille examples expressed using the Unicode's Braille codes. Sadly, many screen readers do not work well outside of the basic Latin character set.
In doing all of the above, I found many errors. I did not correct errors that were typos or poor descriptions in our original, but I did correct errors that involved incorrect representation of the design. That included cleaning up some of the drawings to make them more legible and making even bigger changes to the schematics.