8
\$\begingroup\$

Attached is a picture of the keyboard controller from a 1985 IBM 6770 typewriter which I'm working on.

None of the chip numbers make any sense. The manufacturer prefix don't match the manufacturer logo (e.g. there's a National Semiconductor chip with a NK prefix, which IUIC should be DM/LM/NN). The serial numbers are all oddly similar --- U1 and U2 have serial numbers which differ by 1, even though they're by different manufacturers! --- and worst of all, trying to look up any of the chip IDs fails. Even U4, which is clearly a microcontroller, carries no useful information.

What's going on here? Have they been anonymised --- if so, why, and is there any way to find out the normal chip codes (I need data sheets)? Is it just a very old alternative numbering system? Or what?

enter image description here

\$\endgroup\$
7
  • 7
    \$\begingroup\$ Interesting. My (wild) guess is that in 1985 IBM still had the clout (and paranoia?) to have the IC manufacturers mark the chips with a 7 digit internal Intel part number. \$\endgroup\$
    – td127
    Feb 5 at 20:24
  • 2
    \$\begingroup\$ Once in a very rare while I have actually found more useful information on the pcb side of an IC than the topside. Have you tried pulling the socketed one yet? \$\endgroup\$
    – Bryan
    Feb 5 at 21:20
  • 1
    \$\begingroup\$ I hadn't even noticed it was socketed! I'll certainly remove it and see if I can dump it. \$\endgroup\$
    – David Given
    Feb 5 at 21:49
  • 3
    \$\begingroup\$ It's an Intel 87C51 and I have a ROM image. That will be incredibly useful --- thanks very much for the prompt! \$\endgroup\$
    – David Given
    Feb 5 at 22:06
  • 1
    \$\begingroup\$ Turns out that I just had a bad ROM dump. Every odd page was just a copy of the preceding even page... It took me an embarrassing amount of time to notice! It turns out to be a vanilla 8051 chip with no interesting features. \$\endgroup\$
    – David Given
    Feb 9 at 0:22

3 Answers 3

Reset to default
6
\$\begingroup\$

Not sure what you mean by serial numbers. Programmable parts often had stickers on them that identified the code / rev / whatever they contained, but not serial numbers. All of the parts on your board have date codes, but I don't see anything that looks like a serial number.

Way back, it was a very common practice for large manufacturers to have ICs "house branded". Back in the day, bulletin boards had text files that cross-referenced house branded part numbers to their industry standard part numbers. Surplus vendors would sell these parts with the equivalent datasheet.

I worked on Data General, DEC, and IBM gear, and this was a constant headache.

\$\endgroup\$
1
  • \$\begingroup\$ Re serial numbers: I was thinking of the first line (e.g. 476 7634 for U1). The second line looks like a part number and the third is a date code and batch. From what you've said it sounds like that first line could actually be the part number. I don't suppose you know where I can find the cross-reference for mid-eighties IBM parts, do you? \$\endgroup\$
    – David Given
    Feb 5 at 20:32
4
\$\begingroup\$

These typewriters are rare, so you're going to have trouble digging up much info. There's an old advert here which says they use the same microprocessor as the IBM PC. Since the large chip is also a DIP40, it's possible that it is an 8086 or 8088, although it could well be an 8051.

I was unable to find any additional info on the chips as marked, but I doubt that the part numbers were intentionally obfuscated. Given the era, and the fact that IBM was involved, I would make a guess that the chips were a custom order, which is why they have unusual markings. It could well be that the main chip is an 8051 or some other DIP40 MCU that came with a custom ROM from the factory.

There's a pretty great guide to vintage Intel chips and yours don't appear anywhere on there. CPU-World also has some great info on old Intel MCUs and CPUs. Unfortunately I was unable to find anything that obviously matched.

If Y1 is a crystal that is connected across pins 18 and 19 of the DIP40, this is a hint that it's probably an 8051. If Y1 is instead a standalone oscillator that provides a 12MHz clock signal to pin 19, then it might be an 8086/8.

Another quick test would be to see if both pins 1 and 2 are tied directly to ground. On the 8086/8 this will be the case, but on the 8051 pin 1 is P1.0 instead.

It's interesting that the 4-digit numbers (8502, 8506, 8510, 8514, 8516) seem to be duplicated across chips from different manufacturers and in different packages. It's possible that these are internal references to IBM project numbers, which would make some sense given the layout.

Whatever it is, it's probably quite rare, and it'd be great to see it documented!

\$\endgroup\$
3
  • \$\begingroup\$ The base unit has the main processor in it, which is also incomprehensibly labelled, but I've dumped the ROM and it's indeed an 8086. Other IBM keyboards of the era used 6800 but I traced the power pins and this isn't one --- it could well be an 8051. It's the small chips I'm particularly interested in, though. \$\endgroup\$
    – David Given
    Feb 5 at 21:46
  • \$\begingroup\$ An 8051 would certainly make sense if they essentially derived this from the PC design. \$\endgroup\$
    – Polynomial
    Feb 5 at 22:07
  • \$\begingroup\$ Someone downvoted; could you explain why? If there's something incorrect in my answer I'd like to fix it. \$\endgroup\$
    – Polynomial
    Feb 6 at 20:31
1
\$\begingroup\$

If the circuit is like the IBM keyboards of the era (maybe ask on retro computing):

  • U7 and U9 are quad op-amps - LM339 or similar
  • U8 is a 8 to 1 multiplexer - 74LS151 or similar
  • U6 appears to be implementing the row scan which is a 1 to 16 or 1 to 20 de-multiplexer - it looks like a custom ROM replacing three off-the-shelf chips
  • RP1 and RP2 should each be 8x 10kohm resistor packs all resistors with one common terminal

Now I am making even wilder guesses but

  • U2 and U3 are quite likely to be hex inverters
  • U1 I have no idea, but would suspect a 74LSxx and almost has to be either a decoder or a wide n-to-1 multiplexer looking at where it is in the circuit - I am tempted to suspect a 74LS154 or 159 if it driving a display on connector J2
\$\endgroup\$
1
  • \$\begingroup\$ There is indeed a display on J2 --- which doesn't work. I suspect a negative voltage power line. This is routed through an orange block on the base station which is identical to L1 here. If this was a later era I'd suspect a potted DC to DC converter, but it's clearly not. An inductor, perhaps? \$\endgroup\$
    – David Given
    Feb 9 at 0:26

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge that you have read and understand our privacy policy and code of conduct.

Not the answer you're looking for? Browse other questions tagged or ask your own question.