kwgst@unix.cis.pitt.edu (Filip Gieszczykiewicz) (01/27/91)
(Please use e-mail to reply.) Greetings. This message is for people who remember S100 buss and can, perhaps, give me it's advantages and disadvantages, for a particular application, as compared with the AT ISA, Multibus and QBus. If not interested, hit "n" now. I am in the process of planning a small IBM AT add-on. It will have it's own case and be connected to the AT-buss though an interface card. The reason why I don't want to put it on one board (or a set) Is that it's components are big real-estate users. Most of the stuff will be "custom", basicly a lot of PLAs and such. Also, a lot of "discrete" LSI and SSI. It also has to be expendable in a big way. My question is really on the buss connctors and layout. I would like to use the S100 for a number of reasons: 1) It's well known and relatively reliable (and cheap) 2) Lot's of people have S100-based enclosures 3) There are a lot of pins available (lot's of possible grounds) 4) wire-wrap boards go for ~$25 each. 5) The idea of having an on-board voltage regulator is a good idea in systems that run fast and don't like noise in DC supply. Perhaps a little info on what the buss has to handle will help in your criticizms :-) I already have a 386 (33MHz) but am not impressed with the way the power of the cpu is bogged down, to a standstill, by the use of the ISA buss (AT buss) I can't really do much about it, but I _can_ do something about the way I add more features to my system. I want to have lots of digital inputs (8255s) and analog channels. I also want a way to design my own projects (such as coprocessors, controllers, etc.) I have been told, by people I consider gurus, that such things do not belong on the system buss. They belong in their own "realm" to which the system downloads problems, goes back to it's main job, and only returns to transfer the processed data back to main memory. Back to the question: I don't want people to misunderstand me. I'm not trying to use an existing buss from the compatibility point of view. All the pins in my S100 project will be my very own. What I want is a reliable and proven array of connectors, pin spacings, noise problems, proto-boards, and other matters which make the final system reliable. So, can anyone suggest a better "card cage"? I was looking at the Multibus I and II, but they are much too expensive. The card cage also is hard to come by. I have also looked at the QBus, but finding proto-card can be an adventure. And, last but least :-), there is the original AT connector and wire-wrap board. The only problem is that it's small, compared to the above board, and has many less pins. Not that I plan to use them all. I want at least 20 or so for grounds and, with the AT connector, there aren't that many left. One more thing, I want the buss to be able to handle fast clocks on parts of it. That is, there may be a set of pins that operate at a high speed, compared to the rest, without interference. I plan to use the "not used" pins, on the opposite side of the card-edge, as grounds. One thing though, why _did_ the S100 buss die? If IBM used the S100, would it be the most popular "standard" now? Or was(is) something just _wrong_ with it? Take care. P.S. Please don't just say "You're an idiot, stick with NuBus" or the like. I would like people to give me "reasonable" comments. Also, please e-mail. I will archive and make it available. -- ------------------------------------------------------------------------------ If reply bounces, try "kwgst@pittvms"... no? try "fmgst@unix.cis.pitt.edu" Did you hear? Some Poles burned the Russian flag near the Soviet embassy, in Poland. :-) :-) "Nothing is impossible if you don't have to do it yourself."
jws@cica4.mlb.semi.harris.com (James W. Swonger) (01/28/91)
An S-100 cage will be very big. Very, very big. That's OK if you really want to use all of that volume for electronics. However, nowadays that takes a lot of parts. When the S-100 was first popular it took a whole board just for the processor & support - another for 16K or RAM, etc. A stuffed card cage would have you running a 64K 8080 system with (gasp!) a printer and (wow!) a TVT and matbe you could have room for one card of experimentation real estate. The S-100 decreased in popularity because of its physical size (both the enclosure and the cards themselves - no manufacturer wants to make anything bigger than they have to) and the obsolescence of the backplane standard. The S-100 signals are just TTL-level with some passive termination at one end. There was bitching about the S-100 crosstalk even in the days of 4MHz Z-80 systems - they were just too darn fast. With edge rate control you can push the performance into the low 10's but the S-100 backplane is pretty unsophisticated and I'd expect you to see some flaky performance if you try to go full speed (like 25MHz). I have a 68000 based S-100 (8MHz) system in the attic. It measures 2' by 2' ( terminal not included ) and weighs in at about 80 lbs. If you want a smaller card cage standard with higher capability I'd say the VME is one to look at. There is a high number of pins, good availability of enclosures and function, processor and memory cards, and the bus seems to be capable of supporting higher-speed systems. The drawback to VME as I see it is that it is a commercial standard, not a hobby standard, and you can therefore expect to pay commercial, not hobby, prices for your hardware. The blank boards and card cages are only moderately expensive, but something like PC capability costs like a PC or maybe 2X. You end up paying for compactness on this end. I also have a dual-processor VME system in the attic. The cage is 6" by 8" by 2'. It is unenclosed though. I'm not sure whether the signal standard is TTL or some other line X/R thing. Since you plan to assign your own signals you may have some trace cutting or component removal to do if you start with a populated backplane. The bus standard may not be what you want in some cases. Then again, the venerable STD bus (44-pin) still has a lot of activity, mainly in industrial control type stuff. If you can get by with the low bus width you at least can get the proto cards down at Radio Shaft. PC-bus cards are available which insert various other processors into a PC system. Prototyping cards are also available. There may be manufacturers of PC bus-extender chassis' out there - I vaguely recall seeing that kind of thing; they usually appear after the product has been out long enough for the hardware freex to get edgy but not long enough for the next generation or clones to arrive. Probably a good place to sort this out would be a PC magazine and/or Byte. any of those are 80+% ads. Some of them will have to appeal to you. The Byte issues for December of each year used to have index sections for the year's articles - you can probably find stuff about the various buses in there, in more detail and religious fervor than you may want.
kenw@skyler.arc.ab.ca (Ken Wallewein) (01/30/91)
In article <1991Jan28.140632.20240@mlb.semi.harris.com> jws@cica4.mlb.semi.harris.com (James W. Swonger) writes: > > An S-100 cage will be very big. Very, very big. That's OK if you really want Sorry, I can't agree. The cards are no larger (smaller than most, in fact) than any other type, and the inter-card distance is no different. I've seen cages raging from 3-slot to 21-slot. The power supplies, on the other hand, were usually big, simple, and heavy :-/. > to use all of that volume for electronics. However, nowadays that takes a lot > of parts. When the S-100 was first popular it took a whole board just for the > processor & support - another for 16K or RAM, etc. A stuffed card cage would > have you running a 64K 8080 system with (gasp!) a printer and (wow!) a TVT > and matbe you could have room for one card of experimentation real estate. Yup. 64K bytes worth of 4k bit chips took a lot of cards. Nowadays you can buy megabytes on a card, just like for any other bus. >.. > The S-100 signals are just TTL-level with some passive termination at one > end. I nearly always had active termination. > There was bitching about the S-100 crosstalk even in the days of 4MHz > Z-80 systems - they were just too darn fast. With edge rate control you > can push the performance into the low 10's but the S-100 backplane is > pretty unsophisticated and I'd expect you to see some flaky performance > if you try to go full speed (like 25MHz). > ... Agreed. >... > Probably a good place to sort this out would be a PC magazine and/or Byte. > any of those are 80+% ads. Some of them will have to appeal to you. The Byte > issues for December of each year used to have index sections for the year's > articles - you can probably find stuff about the various buses in there, in > more detail and religious fervor than you may want. There is a magazine devoted to various bus architectures, including the IEEE/696 (S100). If there is interest, I can post the essentials. -- /kenw Ken Wallewein A L B E R T A kenw@noah.arc.ab.ca R E S E A R C H (403)297-2660 C O U N C I L