Google-Apps
Hauptmenü

Post a Comment On: Ken Shirriff's blog

"A database of SMS cards: The technology inside IBM's 1960s mainframes"

16 Comments -

1 – 16 of 16
Blogger Unknown said...

I thought it was interesting that they did use PCBs even then; somehow I had thought that everything was wire wraps and other harnesses back in the day.

The use of a generic PCB populated different for several card types (and the fact that they didn't use PCBs for "gate" or mainframe backplanes) speaks to what must have been a very high setup cost for a PCB layout.

It looks like these were all 2-layer boards?

March 8, 2015 at 6:18 AM

Blogger Ken Shirriff said...

Hi Dan! The boards are all 1-layer, i.e. one layer of traces on the back, and components and jumpers on the front.

I think the wire-wrapping on the backplanes was used for flexibility. Field engineers would actually move wires around if a user purchased a new feature. Users of the 1401 could buy additional instructions such as comparison, multiplication, or indexing, and the engineer would add more cards and change a few wires.

March 8, 2015 at 11:16 AM

Anonymous Anonymous said...

I used to fix these machines. On a 5320, we would upgrade the machine from 9.2 MB to 13.4 MB by moving a wire from one position to another. We also did engineering changes (ECs)by cutting the pin connection and rewiring. It was a great time to be a field service engineer.

March 8, 2015 at 2:43 PM

Blogger Unknown said...

I can see where you got the pictures :-)

I should have two or three SLT cards somewhere.

March 11, 2015 at 1:06 AM

Anonymous Anonymous said...

Very interesting, thank you!

March 13, 2015 at 10:58 AM

Blogger Eric Wilner said...

Ah, memories of Radio Shack, and the Big Bin o' Boards.
If memory serves, RS sold these things with the connectors broken off.
I may still have a few, stuck in old coffee cans out in the garage, in varying states of component removal.
And, looking at the PCB layout, I'm having flashbacks to Bishop Graphics, decals, and tape....

March 13, 2015 at 7:43 PM

Blogger Ken Shirriff said...

Eric, if you find any SMS cards in your garage that aren't on my list, please send me information and I'll add them.

March 13, 2015 at 8:29 PM

Blogger berto said...

There's actually a lot more SMS gold than that. The socket pins also need (and have) gold where the card fingers contact them.

But it's not plated, rather it's a tiny solid button. They are not visible unless the socket is broken open.
My assessments have varied, but there are about 450 of these dots to 1 gram, though they also contain 10-20% silver.

I have over 7000 SMS sockets that I've been dragging around for decades, but after 35 years, I'm finally getting around to recovering the gold in them.

These came from a 1401 and a 7044 that were installed at Melbourne University until the late 1970's.
I wrecked the SMS cards for the gold many years ago, but only later appreciated their historic interest, though I did keep the core memory gate of the 1401 intact.
My very old SMS card website:-
http://members.optushome.com.au/intaretro/SMSCards.htm
Cheers,
Rob Storey

March 18, 2015 at 6:50 AM

Blogger David Galloway said...

I have no citation, but it feels like this setup was the inspiration for the cinematic core of HAL9000. Dave, disengages small cards when 'killing' HAL

May 12, 2015 at 2:33 PM

Blogger Unknown said...

w@David Galloway, there's also the idea of "isolinear chips" in Star Trek: The Next Generation (e.g., http://en.memory-alpha.wikia.com/wiki/Isolinear_chip) which were often shown in large numbers, arranged in colorfully illuminated banks behind wall panels.

May 27, 2015 at 7:28 AM

Blogger Unknown said...

I have that very same memory card I removed from an IBM 2821 controller. It was one of many in an array. We used to shmoo the memory buffer by adjusting the voltage up and down while running a print diagnostic to get the sweet spot. The 2821 ran 1403 printers and 2540 card reader / punches on IBM 360/ 370 mainframes .

May 21, 2019 at 3:56 PM

Blogger Joe and Julie said...

I was a Field Engineer with IBM. The SMS cards were also used in the 2415 9-track Tape Drives. Got a call from a customer one day with one of the drives getting read errors. Turns out we found the pre-amp for track 1 was clipping the positive side of the signal; visible on an oscilloscope. That particular SMS card was out of stock at the regional parts center. A "Code A" parts order would have take two days to arrive. We had a set of diagrams on microfiche such as you have in your database and analyzed the defective circuit. We removed the defective diode and transplanted with another one from an unused circuit on the same card. Voila! We had it fixed on just under an hour. The customer was extremely joyful to get back into production for his payroll application.

March 24, 2021 at 7:30 PM

Blogger Roland Ronja said...

I recently found a box of 225 cards on a flea market. I have no idea how they came to be there, but I bought the lot. It seems I have a few cards that are missing in the database. Would it be interesting for you to add them to the data base? I do not know their function at all, but that might be someone elses knowledge maybe?

Two of them are marked FHA and FRF, two others are not punched at all, but written on. One says YFH, and the other does not have any text like that, but numbers.

I have about a hundred cards more to go through, so there might be more.

Best regards
Roland

May 24, 2021 at 8:54 PM

Blogger Ken Shirriff said...

Roland, yes, if you have cards that aren't in my database, I'd be interested in adding them. What would work best is if you can send photos of both sides, along with the alphabetical code if any (since it's hard to read from photos). The part number (6 digits, starting with 3) next to the alphabetical code would also be useful. My email is firstname.lastname@gmail.com.
Thanks,
Ken

May 28, 2021 at 10:37 AM

Blogger Unknown said...

A few more details about the SMS boards:

(1) The boards only had traces on one side, the bottom. Many wire jumpers were used on the top side where traces couldn't be run directly.

(2) The molded inductors were used to slightly improve the waveform rise times. You see the transistors were slow, with risetimes around two microseconds, and the backplane wires had high capacitance, so a little inductance helped improve the transition times.

(3) Many cards had a primitive type of PROM: a row of like twelve snippable jumpers. You could change the board logic function by snipping away some jumpers.

(4) The board material was rather cheap paper and phenolic with the traces rather loosely bonded to the board. The boards could not tolerate a hot soldering iron, the traces would just peel off. Of course back then we didn't have temperature controlled soldering irons, just 47-watt irons.

June 13, 2022 at 7:07 AM

Blogger Unknown said...

About servicing these cards-- Around 1974 I was playing with a CDC 160A computer that was dead. Poking with the console switches I found out that one bit of the memory always read back as a "1". The logic was on smaller than SMS cards, like playing-card sized with test points at the card edge. Pretty quickly I figured out one of the core memory sense amps was dead. Since there were 12 of them it was easy to compare voltages and I quickly figured out that one little germanium diode was very leaky. I scavenged a glass diode from a random very different PC board that was lying around and that immediately fixed the problem. Still remember that victory 47 years later! There were lots of less pleasant experiences before and after that that resulted in smoke and sparks.

June 13, 2022 at 7:12 AM

You can use some HTML tags, such as <b>, <i>, <a>

Comment moderation has been enabled. All comments must be approved by the blog author.

You will be asked to sign in after submitting your comment.
Please prove you're not a robot