Damn I know this is an old video. I stumble across it just now, an I use to have to repairs these when I went to work with IBM in 1980. I actually had a few accounts still using these machines along with 0557, 0026, 0029, 0059, 0129. Hated when I had to work on them, BUT today I actually miss having to think and analyze why they were failing. These machines made you think and be a mechanic to fix them. Great days! Thanks for the trip down memory lane to an old retired IBM Service Rep.
In the late 70's, I worked on an IBM 1401 system that had one of these. One time I had to do a multi-column sort on about 30,000 punch cards. As each pocket filled, its cards had to be set aside and kept in order for the next pass. Fun times.
WARMANGO, At least You had a computer! All we had was an IBM 402 Accounting machine. YESSS! Multi-pass sorts on large numbers of cards could be stressful. If something went wrong, You had to start all over again!
Pierre Clouthier Yes, I think that's correct. Mentioning EBCDIC really takes me back. I wonder if IBM even uses it anymore. Everywhere else uses ASCII or UTF-8 or such.
You are correct, the 084 was twicw as fast; in fact it had a vacuum assist on the card feed to hold the cards onto the feeder knifes as the ywere called.
There used to be one of these in the 'print' (drawing) library in the pipe space at TI; all the various drawings could be accessed via card files that had one microfiche 35 mm image attached in the IBM card, with the drawing number 'punched in to allow sorting at the end of the day so the drawing cards pulled that day could be first sorted then put back in the card files ...
Yes, the 029 would have helped as would a 557 intrepreter but if your employer wouldn't lease a printing card punch he probably wouldn't have rented a 557 either. I was a Tab Operator at North American Aviation from 1968-1972. I was the only operator in our QA group and my installation consisted of an IBM 084 card sorter (2000 per minute), 514 reproducing punch, a 087 alpha-numeric collator, 557 interpreter, and the bad boy 407 accounting machine. I wired boards and set up and ran all the jobs.
BRACE, Wish that I could have seen the 407. Probably would have helped it You had HAD a 587 (?) collator. I only had experience with a 402 (IBM( accounting machine) for running reports. I certainly could have used a collator at the bank dept where we had receipts based on tab card formats.Not fun putting NEW receipts in among the old stock by HAND!
Brace67 When I was a wee lad, I actually thought that wiring those patch panels was computer programming. Image my surprise when I took my first FORTRAN course as a senior in high school! I did get the opportunity to operate some of the unit record equipment when I worked at Scott Paper in the early 1970s. Lots of fun to watch the card sorter sort my cards!
@abaduck How to sort: 1./ set sortcolumn to Last column. 2./ Run. All cards go from top hopper to the side pockets. Collect pockets in order into one pile, place in top. 3./ Decrement column.. Repeat 1-3 until column number = First. Done. It's easier with numbers than with letters.
You could have explained how it worked and shown a demo that actually sorted on, say, a 3 digit number so we could see the cards falling into the different bins. After the third run, you could have demonstrated that the now sorted cards actually all fell into the same bin showing that they were, indeed, sorted. Of course, I must assume that you know how the machine works and that you actually know what you are doing which was not indicated in the video.
actually, no. they would all end up at the end in the same bin if the sequence numbers all had the same high order (hundreds) digit. but the selection of the card column to sort on was unexplained which is a major flaw in this demonstration in showing how a multicolumn sort would be performed with multiple passes.
Definetly for data, not for programas. Unless you took the care of numbering program cards. But that wasn't a practical idea since programs are subject to a lot of changes and additions. On the other hand, they used to be generally small, while data cards would probably be thouthands.
Normally start with least significant digit.. however.. if you had a very large number of cards i.e. multiple thousand and if you had more than one sorter, you could sort the most significant digit first example first character of last name; then sort all of the A's on one sorter while the other sorter does the B's.. over simplified but it was called block sorting; then sort normally.
Now that is pretty cool though. It's got limited logic in order to be able to to do the sort too. Now of course in the Linux world there's the 'sort' command or 'sort -r' if you want to reverse it.
Not only in the Linux world. "Sort" started also in mainframes. The need for a sorter machine was due to the fact that data input was processed on the go, as disk storage was not available (and some time later, too expensive to be massive). Once magnetic rewrittable storage was available, software sorting eliminated the need of a sorter machine.
Cause the programs would have to process the data in sequence. There was very little disk space if there was any at all, so programs use tu process data input on the go directly. When disk technology became cheaper and more available, the data was then loaded as-is, and latelly pre-sorted by a "sort" program. And then, the Data Bases appeared and the indexes did all the work.
JOHN, You had to sort cards according to what field that You were interested in ( like name or address) The cards were often used in primitive computers or even electro-mechanical accounting machines. If the cards were not in order per certain criteria, it made things difficult for the other machines to handle or compare on a specific field. In the company that I worked for, we often sorted on a cotton bale number ( 7 digit Numeric and then You sorted on 6 digit warehouse location code. Total of 13 passes through the sorter.Hope that I haven't confused You.
@@albertopena5644 it's true that there was very little memory and no disk space, so pre-sorting the data was required. but also, for programs, the cards had sequence numbers in columns 73 to 80 (typically). you'd sort them after you drop the deck.
I must correct you on your premise of three passes per column to sort alpha data on IBM cards. In fact, it took two passes per column. Alphabetic data was represented by two punches in a column. The top portion of the card was called the zone and consisted of the 12, 11, and 0 punching positions. In conjunction with a digit in the column, an alphabetic character was represented. First pass on the column sorted the numeric portion, second pass sorted the zone. I was a Tab Operator 1968-1972.
I actually ran one of these along with an IBM collator when I worked at our local library in the 70's. These machines were really fun to operate. I had a ball working with them.
Unless the long carboard box where you put the sorted cards fall to the ground and you had to start again! Hahaha. I also used both the 083 and the 077 with its wire panel! We used to "program" the panels, with a punched sheet on it so when the "program" was tested ok, we could "store" them for future use by copying the wire links on it with a marker. It seems stone age now! But it was fun.
Sharon, Using the ole unit record machines COULD be fun. Sorters, Collators. Interpreters, card punches. When I started work, our company had a 402 accounting machine ( IBM) I learned to wire the boards for that machine. The most difficult deal that I ever attempted was "cross-footing".You took 2 fields punched into the tab cards and produced a result by adding the 2 fields together. I could have certainly used a collator in one dept that we had at a local bank!
This could sort 1000 cards per minute- that's about sixteen per second. It uses photoelectric sensing to do so. It wasn't the fastest card reader made, but it is the most common one AFAIK. The 084 was twice as fast.
The demonstration was NIL. You have to choose WHICH coluumn to sort , one column per pass. Since you did not choose a columm, the cards simply fell into the "reject" column, with no sorting occurring.
First job I had in data processing was in local city govt. I had to sort trays of city tax record cards. They were held in 5 ft. tall metal cabinets 20 trays to a cabinet. Unfortunately there were 12 cabinets! Took three weeks of sorting to get the cars in sequence so we could do property tax rolls. Fun = card jams, dropped cards, worn and dog-eared cards, paperclips, food crumbs (keypunch operators eating while returning cards to trays) and BOREDOM! Luckily I had no problems with this machine and still love it to this day even after 49 years (34 1/2 yrs in the field).
Quite a trip down memory lane. I used to operate one of these in my first job as a 'computer operator' in Winnipeg back in 1979. Yes - these machines (and the cards) were still very common in the '70's and early '80s. I also operated an IBM7010, IBM370/145 DOS/VSE - quite the hay day!
Thanks for trip down memory lane, and all the responders naming some other peripheral equipment about which Google is strangely silent, as there is very little data out there on the old ways.
I worked on one of these machines in Halifax, NS, 1970-72. You sorted on the least significant digit first, then worked your way up. Heaven help you if you got a card jam...
You could do it backwards starting with the highest digit but you have to return the cards to the hopper starting with the 9s digit. Back in the day, I worked with an IBM1130 that had a card reader/punch. It had only 2 output hoppers. So what I did is write a small FORTRAN program to read the sequence numbers punched in columns 73-80, convert it to binary on the fly and then route the card to either the 0 or 1 bin depending on the value. I sorted units digit first and worked up from there. You just kept running the cards through and when they all fell into the same bin, they were done! Note that had all 8 decimal digits been used, it would have taken 256 passes to get them all sorted! Of course, that would have been 100 million cards all told!
The IBM 083 sorter in this video was a very common piece of punched card equipment in a tab room of old. I operated one and also its faster brother the 084. That baby operated at 2000 cards a minute using a vaccum assisted feed and a light source to read the punches instead of a wire brush. Sorting of card fields was minor to major, one column at a time. One pass for numeric data and two passes for alpha data. Card sorting was monotonous but necessary to get the date files in proper sequence.
All the cards had punches in them. The holes were read by a wire brush that was set on the column to be sorted. What ever row the hole punched in that is the pocket that that card was sent to. I worked for IBM from 1966 until 1996, If IBM made it I fixed it! The sort speed was 1000 cards per min. until it jammed then it was a mess to get the cards out.
And when there was a jam (usually b/c the cards weren't riffled prior to loading the tray) someone (me) had to punch a new card to replace the damaged one. Fun times!?! In a bad jam the machine would have to be fixed - way above my pay grade!
Play nice guys. jx was just making the point that it's the technology that's interesting, what it was used for isn't germane to this discussion. At least that's how I read it; I think you read it wrong Brace.
Ummmm I haven't quite figured that part out yet! I just have a deck of random cards. Hey I'm not really into unit record equipment. Have to RTFM in my copious free time....
I ran one in actual production. Sorters were actually fun to operate and required a degree of manual dexterity. Card would be generated in our shop by RDE (remote data) Shop would have machine that read work center card and employee badge. Cards were then produce by O26 remote key punch. Cards were then sorted by job and time and then sent to tabulator which produced report for cost accountant. Replaced big tubs full of hand written time cards. System was also tied to time card system for pay.
Sorters were generally used for data, but most programming languages of the era reserved columns 73 to 80 of each card for a sequence number. Production program decks were typically numbered in this way so that if a deck was accidentally dropped, it could be sorted back into the proper order using one of these machines.
cool! they should teach more about this in CS classes in college, and take students to actually see these machines. someone also should make videos explaining old school computing.
My dad has one of these babies. I'll post a video of it sorting cards when I'm at his house next. There can't be to many of these around nowadays. Thanks for the demo.
Yes, come to think of it, you are right: two passes would be sufficient for two-punch alpha info. I started programming in the 70's. My employer was so cheap we couldn't afford printing on our 029 keypunches. I learned to sight-read the cards from the holes :o)
join the conversation Yes! The vacuum feed was so good that no card weight was needed even as the stack of unfed cards approached zero! And it was twice as fast as the 083.
That's great. I read a PDF the other day about the history of IBM (from 1920 - today). Machines like this were IBM's backbone for many years. My only (minor) complaint is that you didn't show a close up of the cards... I thought they were all blank until I noticed (at 2:24) that some were punched. &eB
Dam use to repair them in 1976 what i was with the US Army There was one Valve or Tube that when it was running the top of it it glow went around in a circle
