9

When Atari was designing the 800 series of computers, they needed a way to connect peripherals. To comply with FCC regulations on RF emission, it needed shielded cables. To achieve low cost of entry for home users, it needed to be daisychained, rather than requiring the purchase of a server. To this end, they designed the SIO.

Another option would have been to use IEEE 488, a.k.a. HPIB, GPIB. IEEE 488 met all of the criteria; it could be daisychained with shielded cables, and the electronics required in each device were cheap. Its use for connecting peripherals to computers had already been proven by the Commodore PET.

Presumably Atari saw some advantage to be gained from designing their own peripheral bus, or they wouldn't have incurred the cost and schedule risk. What advantages did they see SIO having over IEEE 488?

5
  • 4
    Where do you take the "and the electronics required in each device were cheap" claim from? That's absolutely not the case.
    – tofro
    Commented Mar 4 at 10:36
  • 2
    I am sure those 16914buss connectors wasn't the cheapest on the market back in the days... And the fact that Commodere didn't continue to use IEEE 488 when moving from the buisness field to the home computer field might indicate soemthing about price also.
    – UncleBod
    Commented Mar 4 at 14:30
  • It's possible there's some confusion of IEEE-488 with RS-485, which really is an advanced serial bus. Just one from 1983.
    – davolfman
    Commented Mar 4 at 17:30
  • 1
    @tofro I agree. I looked at buying an all-in-one HPIB support chip in about '79 and it would have been £40... that would be ITR of £200 today. Commented Mar 5 at 7:52
  • 1
    The very same reason why Commodore switched from (limited) IEEE-488 with PETs to their IEC for the Vic-20: Cost, cost and no real benefit as Homecomputer rarely need to access expensive measurement devices.
    – Raffzahn
    Commented Mar 5 at 14:18

2 Answers 2

21

IEEE 488 may look simple enough for a home computer ("It's just an 8-bit bus"), but is way more complicated than you might think:

  • Lots of wires - A proper HPIB cable needs 24 wires and extensive shielding, which means unwieldingly thick cables. AN SIO cable needs about half that (and doesn't even use most of these, so depending on your device, you can slim your cable down to four wires).
  • Complex connectors - The stacking connectors are complex to handle, bulky and large, thus very expensive (today, it's easy to spend middle-three digit-$$$ on a proper HPIB cable). The SIO connector is way simpler.
  • The protocol is not as easy as you might think: Specifications say that a unit must react within 200ns (!) to an ATN signal, including switching bus drivers from send to receive. Nothing you could do with a home computer CPU at that time (even low-end modern µCs have problems with that timing). To fit that specification, you had to use (expensive) proprietary IEEE-488 chipsets.
  • Massive Overkill IEEE-488 implements star and bus (nice to have, but not necessary for Atari), multi-master, and independent data transfer between devices. A complex setup could make the bus controller (which would have been the Atari) pretty busy. Atari only needed point-to-point, single-master/multi-slave comms, communication between slaves wouldn't have been necessary, SIO is much simpler than IEEE-488.

HPIB complexity is very often underestimated. No, it's not "just an 8-bit bus". A full implementation would have been total overkill for what Atari needed, and a somewhat down-sized implementation would have been proprietary, but if you go proprietary anyhow, you can just as well go with a massively simpler implementation.

I guess there is a reason why Commodore went to something (significantly) simpler after the PET.

4
  • 3
    IEEE-488 was poorly implemented by many (non-HP) instrument makers initially. In the early 1980s I encountered multiple instruments that worked if and only if they were the sole instrument on the bus. Which, of course, defeats the point of a bus. If kilo dollar instruments had a hard time getting it right, consumer gear had no chance.
    – Jon Custer
    Commented Mar 4 at 14:43
  • 2
    The one thing I'd add to that is that Commodore apparently had a serial variant of IEEE-488 (AKA HPIB/GPIB) which obviously reduced the cost enormously. However the complexity of even modest state and command handling would take memory, which could obviously be used for other things. Commented Mar 5 at 7:56
  • 1
    @MarkMorganLloyd That's the "something simpler" I referred to in my last sentence. It has, however, absolutely nothing to do with the question. And "serial variant of IEEE-488" is a serious misnomer. The solution CBM used is much closer to Atari SIO and has not much to do with IEEE-488.
    – tofro
    Commented Mar 5 at 8:11
  • 1
    @tofro In that case I suggest fixing up en.wikipedia.org/wiki/Commodore_bus to which I found myself referring (because of discussion elsewhere) a few days ago. Commented Mar 5 at 9:46
18

According to Joe Decuir (one of the designers of the SIO bus), he wasn’t aware of another serial bus when he had to design a solution for external peripherals in a radio-silent system. Make of that what you will; but the result has a few advantages over IEEE-488:

  • it’s much simpler to implement, as detailed in tofro’s answer;
  • it provides power (+5V on all systems, along with +12V at first on the 400 and 800);
  • it allows peripherals to provide their own “device drivers” (this could presumably be done over IEEE-488 too but I don’t think it’s part of the protocol).

All said however, Decuir considers that the SIO bus, and the costs associated with it, played a major part in the failure of Atari computers in the market; the competition there wasn’t full-blown IEEE-488, but integrated peripherals, card-style connectors, and other, simpler busses (such as that used on later Commodore computers).

4
  • 1
    I'm really hesitant to call the CBM serial bus "a variant of IEEE-488". (I suspect that was sort of a marketing blurb of CBM to suggest continuity when they changed from IEEE-488 to their "new" bus - although I'm struggling to find a CBM reference of that claim). Technically, there really are more substancial differences than similarities between the interfaces.
    – tofro
    Commented Mar 5 at 12:23
  • 1
    Thanks, I’m not sure what gave me that impression (the Wikipedia page possibly). Commented Mar 5 at 13:16
  • 1
    @StephenKitt it's an image often colported by Commodoristas - supported by the maintaining a similar protocol/API - as well mentioned by Wiki (not sayin Wiki being in any way authoritative)
    – Raffzahn
    Commented Mar 5 at 14:15
  • 1
    I was looking through a number of old C64 brochures and found it advertised as "Intelligent peripheral bus". No mention of IEEE-488
    – tofro
    Commented Mar 5 at 14:20

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .