IBM 1130 498

One technical reason is that I don't think ROM was yet available (other than diode-arrays or bed-of-nails for a punched card). The other is that existing devices were easier to interface without any hardware translation. And it allowed for fun things like block-lettering across the paper tape and punched cards. One of my favorite programs took console keyboard input and punched it in block letters on punched cards for quick signs! The console switches selected center-left-right justify, underline-overline-both-neither, number of spaces between characters. I think that was a gift from Stuvescent High School to the Cooper Union. I don't remember getting to swap programs with Mr. Berlin of Jamaica High School.

I just pulled out my GX26-3566-7 "IBM 1130 reference handbook: programming reference charts" for it lists the data formats for each device and how the 16 bits were mapped (not all bits were always used).

I don't recall what manual listed those but I clearly remember that there were 2 families of translation subroutines.

One was table based, kinda like the 'tr' (translate) instruction of the IBM 360. I don't recall if it was one subroutine that took the table as an argument, or a family of subroutines for all the translation tables (card-code to EBCDIC, EBCDIC to Selectric, etc.)

The other was specific to the 1442 card reader and auto-synchronized to the card reader. The 1442 card reader was really DUMB! but amazing technology for the time: fiber optics illuminated the reader and photo-interruptor for even lighting and eliminating bulb vibration It interrupted once per column (for reading or punching: could not perform both at once because all the cards advanced together). Interrupt level 0 was the highest level interrupt (levels were 0-5) and was ONLY used by the 1442 card reader since it could NOT backspace: miss that interrupt and it's time to put the card back to re-read! Once the motion clutch engaged, the entire card advanced for reading or punching. NO stopping midway! The 1130's core was word oriented: 16 bits (and parity), so only 12 bits were filled: the other 4 were zero filled. So the conversion subroutine ONE FILLED the low 4 bits and spun-waited for them to become zero as the interrupt routine moved in each column. That way, each column was converted as it arrived using the otherwise dead time between arriving columns.

Just as the system 1130 used parts from earlier machines, the 1442 card reader was used in many subsequent machines. I saw one attached to a System 3 instead of the little square cards.

Having 2 stackers (where the cards come out) was really nice for applications such as deck duplication: it alternated the outputs between decks and marked the deck end with a card with "END" across it. A simple application sorted out blank cards before recycling decks.

Sigh: I never knew they'd all go away so fast or I'd have taken more photos and recorded the sounds. Ah, to have had a camcorder back then!