• Please review our updated Terms and Rules here

Does the hobby world need another 8-bit kit?

glitch

Veteran Member
Joined
Feb 1, 2010
Messages
5,051
Location
Central VA
I'm considering building up another project soon -- a bus-oriented 8-bit computer. This is sort of an extension of my 8085 SBC project. Having worked with several 8-bit bus architectures, I've naturally come up with what I think would be the "best" way to do it, from a hardware/hobbyist perspective...a combination of the S-100 bus, OSI-48 bus and Apple II bus. Functions would be on separate plugboards with a passive (well, actively terminated) backplane. I'd imagine there would be an 8085 processor board, RAM boards of some capacity, a UART board with boot ROM for monitor, et c.

While the first processor board would be an 8085-based one, being my favorite 8-bit processor, I'd really like to produce other 8-bit processor boards too. A friend of mine is interested in developing a 6502 board for it, and I'd like to build a NSC800 board as well. The system bus would be designed in a fairly generic manner -- 8-bit bidirectional data, 16-bit address, probably an IO/M control line, reset line, clock line, interrupt line(s)...I'm pretty sure I could fit it into a 50-line bus. To handle the problem of memory-mapped vs. programmed I/O, I have in mind a board that would map part of memory in the top 1K to I/O ports (asserting the IO/M line, et c.). This would allow boards that work with 8085 programmed I/O to be used with a 6502 processor board.

As to the hobbyist angle, like my 8085 SBC, the boards would be laid out in EAGLE CAD Lite, which is free for hobby use. I know I can pack individual functions onto the 3x4" area limit without too much trouble. The bus connector would be an IDC type, allowing use of cheap connectors for board interconnects (either stacking or backplane).

Does the world really need me to bother to do this though? The machine will get built anyway, as I want to do this for the learning experience, but I don't know that it's worth the time/money to have boards commercially etched when I can build them with wire-wrap or point-to-point construction. Hobbyists already have a few 8-bit kits to choose from -- the N8VEM SBC being a great example. I want to follow the designs of earlier systems though, mostly through limiting the functionality of each plugboard to a single task.

So, I guess the question is, does the world need/want another 8-bit kit?
 
Hi! You may want to consider a "standard" bus like ECB. It is not perfect but it would allow you a wider range of options for using legacy boards. The bus control scheme you are describing is well within the ECB standard of 16 bit address, 8 data bits, and all Z80 style control signals (clock, reset, interrupt, etc). The Intel signals are also available on the B-Row. So far, we've placed a Z80, 80C188, and 6809 CPUs on the bus. There is a new host processor in the works too with 6809/6502/6802CPUs with a stand alone ECB

ECB uses the DIN 41612 96 pin connectors which I consider some of the best and most reliable connectors ever made. They are *much* more reliable than IDC or dual row header connectors. They are cheap, plentiful, nearly indestructible, *and* gas-tight which means they are equivalent to a wire-wrap connection -- super reliable! You could then re-use ECB backplanes and boards. They are fairly common in Europe but less so in the US. You could also use N8VEM boards in your design to help prototype and develop new boards.

Regardless of whether you start your own system project I wish you the best of luck! It is a lot of fun and I think you'll enjoy it! You are always welcome to use the N8VEM information for your project. I started the N8VEM home brew computing project with helping builders make their own systems as a primary goal but it is certainly not the only way to do it. I hope we can collaborate on some common projects! Please let me know what I can do to help!

Thanks and have a nice day!

Andrew Lynch
 
Last edited:
Sorry for going a bit off-topic, but out of curiosity, how old is the ECB? I found one reference which said the bus was "defined in 1984 by the German company Kontron", but I believe the bus was around many years prior to that? Perhaps there only were 8-bit variants and the first 16-bit addition was made in 1984?

One member on the (old) VCF IRC today posted scans from Swedish company Scandia Metric who once made a Z80 ECB system called Metric Card. I looked at the pics to determine its age and suggested 1981-82 based on what the broschure says. As this company had a previous history of importing designs, refine and relabel them, I have a feeling one might find some ECB system very similar to theirs, but sold internationally by some other company.
 
I'd thought about using the ECB, but I think for my project it's a little more than I need...it'd also reduce the available board space in trying to keep the cards within the EAGLE CAD 3x4" limitation. We used to use it at work for some of our control boards because of the high-reliability connections it's known for. Using IDC headers means they're easy to source, and using gold-plated interconnects provides reliable connections.

I'll probably take a closer look at the N8VEM information in designing these boards, even if just for personal use. It would be nice to have my design be software-compatible, at least to some degree, with the N8VEM ECB (using the same UART, memory map, et c.). I'll probably take a look at your floppy controller design, as I'd like to add one to mine once I get a basic set of boards up and running. The N8VEM ECB is something that's on my to-get list, and has been for a while, anyway!

There will definitely be community collaboration in what I end up designing...I hope to make something generic enough, and with the right feature set, to enable hobbyists to choose my bus-oriented system for small (modern) embedded projects that would require a more expensive microcontroller or off-the-shelf embedded system. That was the primary reason in designing the 8085 SBC!
 
Hi Anders! Thanks!

Kontron defined the ECB in the early 1980's but it may well have been in use prior to formal definition. Rolf Harmann in Germany has about the best reference site on ECB I've seen. Rolf has helped me and the N8VEM project many times. He is a great guy!

http://www.harrmann-rolf.de/

Also Wolfgang (also from Germany) has a lot of background information on ECB along with a working knowledge of its design and a lot of experience with it. He has a folder on the N8VEM wiki with ECB background information. The meaning of the various ECB signals are a frequent discussion topic on the mailing list.

Do you have a link to the Scandia Metric scans? I would like to see them.

Thanks and have a nice day!

Andrew Lynch
 
Hi! If you want you could use both the ECB and a unique bus and support both types of peripherals. The 8085 to ECB interface is probably going to take about 5 or 6 chips for everything.

Thanks and have a nice day!

Andrew Lynch
 
Going on-topic, it might be nice to see support for the less-known CPUs, but the problem is finding the chips. The National INS8900 periodically shows up on eBay, but not in quantity, for example. SC/MPs are rarer. I have a GI CP1600, but have never run across a Fairchild 9440. The Soviets used a clone of the LSI-11, but those also seem to be hard to come by.

And that's really a shame--the old chps seem to have gone off to be ground up for their gold content.
 
I'd thought about using the ECB, but I think for my project it's a little more than I need...it'd also reduce the available board space in trying to keep the cards within the EAGLE CAD 3x4" limitation.
I'd sure vote for dual-row headers, cards stacked with spacers and connected with ribbon cable; no card cage, no bus board, cheap, flexible, and if you use the 80-wire 40-pin PATA cable idea you'll even have free 'shields'; you could even have simple transition cards to connect to other buses like ECB or even S-100.
 
Here's an idea--why not take one of the Cortex M3 ARM CPUs and equip it with conventional memory and I/O interfaces and then offer emulators for whatever CPU your heart desires? Cheaper than FPGA and "C" is a whole lot easier to get right than a bunch of Verilog or VHDL.
 
I'd sure vote for dual-row headers, cards stacked with spacers and connected with ribbon cable; no card cage, no bus board, cheap, flexible, and if you use the 80-wire 40-pin PATA cable idea you'll even have free 'shields'; you could even have simple transition cards to connect to other buses like ECB or even S-100.

That was my thought...keeping to a 40- or 50-pin header would allow the use of IDE or SCSI cables as the "backplane" -- that's why I used a 40-pin connector on the 8085 SBC. Or you could replace the dual row header with a stacking connector, similar to those found on PC/104 boards (I used machine pin SIP sockets with really long pins on the 8085 SBC for this purpose).

The bus transition card was also an idea I'd had...especially since I want to eventually wire-wrap an 8085 S-100 processor board. You could actually use the 3x4" module as a wire-wrappable module on such a board.

Regardless of the final bus interconnect, it sounds like there might be a fair bit of interest in a system designed to be processor-inspecific. I think it would be really neat to have a bunch of RAM boards, serial boards, disk controllers, et c., already in existence so that if someone wanted to play with a processor, they wouldn't have to design the entire system -- just grab a bit of protoboard, a 25x2 male header, and your support chips and throw together a processor card. By keeping all of the function decode logic on the processor board or maybe a support card, it should be possible to feed each option board generic signals (address, data, data direction, I/O vs memory access, clock, reset).

The idea for a generic bus system with option cards came about as a result of the fair quantity of one-off processors I have. It'd be fun to play with them, but building up a set of support boards for each one is time-consuming, when it's basically the same stuff every time. The OSI-48 bus was a big inspiration as to what would be on the bus, since it's generic enough to support the Z80, 6502 and 6800 (as well as the IMS-6100...extra data lines for the 12-bit words).
 
Using ribbon cable instead of stacking headers does have the advantage that you can conveniently "unfold" the boards for testing and troubleshooting; relative price would depend on whether you've got a box full of suitable ribbon cables & connectors in your junk pile ;-)
 
Using ribbon cable instead of stacking headers does have the advantage that you can conveniently "unfold" the boards for testing and troubleshooting; relative price would depend on whether you've got a box full of suitable ribbon cables & connectors in your junk pile ;-)

Didn't the Jolt use that scheme?
 
Jolt was a minimalist system--you got the cards, some spacers and through-bolts and a ribbon "bus" connector. No enclosures or anything fancy.

You could carry the simplification further. Use "stack though" PC-104 connectors as your bus. For debugging, you could provide a PC-104 ribbon cable that would allow you to break a stack open. It's really a very nice setup--bus length is kept to a minimum and there aren't any unsightly ribbon cables hanging off the thing.

And why use some oddball bus structure? PC-104 is essentially ISA and is well understood. PC104-to-edge-connect ISA cards exist as well.
 
Jolt was a minimalist system--you got the cards, some spacers and through-bolts and a ribbon "bus" connector. No enclosures or anything fancy.

You could carry the simplification further. Use "stack though" PC-104 connectors as your bus. For debugging, you could provide a PC-104 ribbon cable that would allow you to break a stack open. It's really a very nice setup--bus length is kept to a minimum and there aren't any unsightly ribbon cables hanging off the thing.

And why use some oddball bus structure? PC-104 is essentially ISA and is well understood. PC104-to-edge-connect ISA cards exist as well.
Yeah, I think I recall reading about the Jolt in some magazine.

I've always liked the PC-104 stacking concept myself, except for the inconvenience of getting at the cards, but as you say, as long as the headers are a standard configuration then a cable with a couple of connectors is pretty trivial and certainly easier and cheaper than a bus extender card for a board-mounted bus like S-100. With proper design you could have it either way.

And I was also going to mention that ISA bus compatibility might be a good idea, as long as it's also compatible with the various other processor mappings that are anticipated; if not, there's still the option of a ??-bus<>ISA transition card or cable. I see that Andrew's N8VEM group has sort of rejected ISA compatibility and prefers to create their own ECB NICs etc., but I can also see an advantage in being able to use an ISA NIC, I/O or video card, whatever, especially if these boards don't have to fit into a card cage.
 
The OP mentioned an 8-bit kit using Z80, 8085, 6502. PC-104 or ISA doesn't seem like a good fit based on clock speed alone. But the stacking form factor is still neat. Plugging in an ISA network card or graphic card would be nice.

The STD bus (http://en.wikipedia.org/wiki/STD_Bus) might be a better fit electrically. It was of the same era as these 8 bit processors and the 8-bit version of the bus only has 56 pins. Prototyping boards are still available. There seem to even be companies that still sell various STD Bus cards. The form factor with edge connectors, backplane and card cage might leave something to be desired.
 
Back
Top