User talk:Jeh

From Wikipedia, the free encyclopedia
Jump to: navigation, search
25 days since a lost time accident on this page. (The previous record was 366 days.) Let's try to set a new record!


As I already noted on the talk page, I suspect you are right about the clone vs. IBM. At the time, I never thought about the difference. Are there IBM 2741s with wood grain desks? (See the APL picture.) I don't remember those. Also, do you know where to find either clone or IBM? I know a museum that could be interested in one. I do remember the Itel 7330, clone of the IBM 3330 that would work with S/360. Otherwise, I don't remember much discussion of clones from those years. Or maybe I just wasn't very interested in the distinction. Thanks, Gah4 (talk) 01:55, 19 October 2016 (UTC)

I think we just didn't call such things "clones" in those days. They were certainly around for disk drives. If I remember right the term "plug-compatible" was in vogue. My college was looking at plug-compatible, cheaper replacements for 2311's.
My senior project (EE/CS dual major) was to take a quite different Selectric-based terminal, remove the existing electronics, and build an interface to make it act like a 2741... 'cause at the time I was in love with APL. The real 2741's logic was built with the same sorts of IBM SLT modules that they used in the S/360, plugged into a wire-wrapped backplane, and it used many of those modules. (e.g. there were no UART LSI chips like the WD1402 when IBM designed it so they built the UART, like all the rest of it, out of random logic.)
IBM happily sold me every manual they had on the thing, including a complete set of service info, all the way down to the logic diagrams.
So, the reason I am certain they were all built into that desk is that I've been in the IBM documents catalog. Paper catalog in those days, but they had a KWIC index that made it difficult to miss anything with "2741" in the title. If IBM had re-implemented the logic to not need that box, they would have given the machine a different model number. This was mid-70s so it was about 10 years since the 2741 was introduced... and there were certainly no user's, maintenance, repair, etc., manuals in the catalog for any other version of a "2741". And IBM being IBM, they would have been there. (Heck, you could even order the IBM "Think" sign!)
Sometime after the 2741 was introduced there were indeed a few 2741 clones, and eventually they got the logic down to something that would fit either in an extra-thick baseplate for the Selectric, or a module that hung on the back. There was one manufacturer that sold a "slice" - open your Selectric, take the top cover off, remove the mechanism from the baseplate (this is all a no-tools job, it takes seconds), drop the "slice" on the baseplate, put the Selectric mechanism on top of the slice, and put the cover back on. The slice had the necessary solenoids to work the bails and switches to read the keyboard. The result looks like a Selectric about 1.2 inches higher than usual with the extra height in a metal band between the baseplate and the top cover. (Using a standard office Selectric in this way was alas a great way to wear out the machine's main clutch and other components as they were not rated for continuous duty.) The one I remember using had a built-in paper tape punch and reader, all in a desktop machine, enabling it to function a lot like an ASR-33. Jeh (talk) 08:27, 19 October 2016 (UTC)

I was last year working on an IBM 360/20. As we didn't have a card reader or punch for it, and not the cables or documentation for the printer, I built a virtual card reader and card punch. The first thing to learn was that SLT is signal compatible with TTL. Then what the signals look like, and what they do. But it nicely runs with a virtual card reader and punch. So, I now know all about SLT and IBM logic diagrams.

See: and select minicomputers.

One other thing I remember about 2741s is a piece of plastic that goes over the top, where the mechanism is. (Sold separately, and probably not IBM.) There were stories about people getting their tie (remember IBM and their dress code) wrapped around the rod inside that rotates once per character printed. It seems it doesn't take long to pull it all the way in.

Any thoughts about the wood grain table? Gah4 (talk) 09:31, 19 October 2016 (UTC)

As you have seen I answered that over at talk:IBM 2741.
I never knew that SLT was signal-compatible with TTL. For the 2741 project I didn't look at the logic diagrams much as doing that design was really part of my project; simply copying it in TTL DIPs would have been uncool. I also had the aforementioned UART and cheap PROMs to work with, which IBM had not at the time. In retrospect I suppose I could have used a 6502...
These particular logic diagrams were done on a line printer, with every element a square box with "&", "|", etc., inside it. IBM had a package called "Flowchart" that would take a sort of markup language as input and emit line printer'ed program logic flowcharts. It was clear that IBM had modified it so it could do logic diagrams too. Jeh (talk) 09:52, 19 October 2016 (UTC)

Yes, those line-printer logic diagrams are what we had for the 360/20. Somewhat harder to follow than the usual gate shaped diagrams. Actually, the ones for the 360/20 don't have & and |, but A and O. (You need the TN print train for |, I am not sure about &.) N for inverters.

As far as I understand it, the wiring is computer controlled, such that the same input file does the wiring and diagram. They use both PC board traces and wire-wrap lines.

Talking about UART, when working on the 360/20, and while trying to understand the output data for the card punch, I had a PMOS UART (that needs -12V and +5V power supplies) connected. The final virtual card reader and punch are in FPGA logic, but I did much of the testing and understanding with real TTL gates on breadboards. Gah4 (talk) 10:13, 19 October 2016 (UTC)

I looked at the 2741 manuals decades ago - they may well have used A and O. But I'm pretty sure & was in the commercial character set going back to the 1401, at least.
Was the 360/20 for a museum or something? 'cause otherwise I can't imagine why you didn't virtualize the whole machine! Jeh (talk) 22:05, 19 October 2016 (UTC) yes it is a museum, but unlike some museums, they like the computers to run. And if you feel like programming on a VAX or PDP-10: which you can do without going near the museum. The biggest project is a CDC 6500, similar to the 6600, but not quite as fast. But no 2741s. Gah4 (talk) 23:07, 19 October 2016 (UTC)

I spent a lot of my professional career in kernel mode on VMS and I still have a VAXstation here. But the museum looks terrific. :) Jeh (talk) 01:15, 20 October 2016 (UTC)

I have a MicroVAX 3100/30 that I sometimes run. I have an Integrity (Itanium) machine that runs VMS, though it takes too much power to run for a long time.

But yes, the museum has a VAX 11/780-5 that runs, pretty much, all the time, that people can log into and run programs on. Lots of fun machines to play with. Gah4 (talk) 01:38, 20 October 2016 (UTC)