• Please review our updated Terms and Rules here

Multibus open hardware?

NobodyIsHere

Veteran Member
Joined
Dec 21, 2006
Messages
2,403
Is there such a thing? Would enough people be interested in it to warrant it? There doesn't seem to be any open hardware Multibus projects AFAIK.

The situation seems ripe for a Multibus open hardware project but I don't want to push something no one else or very few are interested in pursuing.
 
We still run MULTIBUS 1 equipment at work.

In fact, I am re-manufacturing some of it as we speak (a 4-port intelligent synchronous/asynchronous communications card to be exact)...

What are you thinking of?

I would be interested in potentially two things at this point in time. Either:

A MULTIBUS 1 interface to some other (= more modern) bus architecture (e.g. PCI or VMEx64) or

An FPGA-based MULTIBUS 1 card. A 'decent' FPGA - one with some large BRAM, embedded PowerPC processor to run Linux and a fiber or ETHERNET port with a couple of RS232 serial ports.

Is that what you are thinking - or am I reading more into your post than you were thinking of?

Dave
 
Hi

I was thinking more along the lines of a hobbyist open hardware approach somewhat similar to what we did on S-100 or ECB.

Right now (AFAIK) there are no open hardware Multibus designs or the infrastructure to support it so any project involves Multibus basically means starting completely from scratch. For instance, board outlines, part libraries for P1 and P2, example schematics in KiCAD (open source) format, etc. All those things you would need to even start a project. Also there doesn't seem to be any much, if any, hobbyists building their own Multibus boards so there are skills and knowledge would need to be built up as well. Without a Multibus template each project has to carry all the burden and I think that suppresses a lot of people from even starting or considering their own.

I think your ideas are great for Multibus projects but I think they are more graduate level where I was thinking more like freshman introductory level projects to "prime the pump" and get things rolling. [I asked a similar question on the S-100 Google Group so I'll weave in part of my thread from there since it is similar]

What I am suggesting is more of a "crawl, walk, run" approach. IMO, there are really three basic kinds of boards for any parallel bus in rising order of complexity: IO, memory, and processor.

In order to get started with a something well documented and low risk technically but important from an infrastructure perspective. Like the simple IO board from Application Note AP-28A which is really just a pair of 8255 PPIs. Develop that board which would provide the board outline, connectors, and general schematic building blocks for later more sophisticated IO boards like IDE, serial, etc. Or maybe start with an even lower threshold like a prototyping board.

Once IO boards are working and reliable build static RAM and Flash ROM memory boards. Again, keep it simple as possible building on the lessons from the IO boards and build up that open hardware Multibus infrastructure.

Finally build processor boards starting with 8085, move up to 8088, 8086, etc. Once those are working then attempt the non-Intel processors or more aggressive designs like bus bridges, FPGAs, etc. People are doing stuff like that on S-100 now but only after many years of slow progress going up the ladder of complexity.

Obviously this is a big long term project but the model has worked at least twice in the past. Once with the ECB board collection and again with the S-100 boards. It is too big for a single individual and would require significant investment in time, effort, and money.

I would be happy to design a simple open hardware Multibus board and put the design out there but I am doing some "market research" first before I spend time on it. Multibus is different than S-100 and ECB in that those busses are dominated by hobbyists while Multibus systems are still in use in industry and in some cases commercially available.

There are things I'd be happy to do like build a KiCAD version of an IO board but not chase down other people to build it, buy boards, distribute boards, etc. In other words, for this to work there has to be some level sponsorship and group participation or it is just me pushing an idea no one else really wants.

I don't know what the level of interest is out there or what people's capabilities are. Maybe jumping to a PowerPC processor board is feasible as a first step but it seems overly ambitious to me and is way beyond my skill level.

Thanks, Andrew Lynch
 
Last edited:
Is that what you are thinking - or am I reading more into your post than you were thinking of?

Dave

Hi Dave, since you are still running Multibus at work can you answer a few questions for me?

What a people running Multibus doing for mass storage? Are IDE/CF/SATA controllers available for Multibus? If not are people still running their original ST506/412 controllers with legacy MFM type hard disks? Floppy drives? What about networking?

The Multibus controllers I saw in industry were booting from solid state ROM boards to load their controller code. No spinning media due to the environment. The boards probably cost a mid-size fortune back in the day though.

I think simple hobbyist level boards are certainly achievable. For instance, parallel interface, IDE/CF interface, serial interface, prototyping board, simple backplane, simple PS/2 keyboard & mouse, possibly simple FDC, possibly simple 8-bit Ethernet, even possibly simple 8-bit VGA. Simple SRAM and Flash ROM boards too.

Projects can be kept simple to increase their likelihood of success. Runaway complexity kills these projects. Thanks, Andrew Lynch
 
Hi, No. Thanks for the brochure. Interesting stuff.

OK, say you have a legacy Multibus I embedded controller. It's been running for many years and the hard drive is slowly failing. SCSI, IDE, ST506/412 or whatever. Can you get a replacement IDE controller and keep going or do you need a replacement CPU with drive controller? Its not a LMGTFY as I've already tried.
 
You can probably find a Multibus SCSI controller, but IDE is so little silicon to interface, that it's doubtful that you could find it alone on a card. Perhaps as a daughterboard...
 
Andrew,

All of our Multibus nodes are diskless. They boot over a network from a host machine (non MULTIBUS). A small EPROM (located on the CPU card of each Multibus node) contains the Intel iSDM 286 debug monitor and our bespoke boot code via our communications cards.

Incidentally, I have the Intel relocatable object code for the 286 debug monitor. No sources, but all the 'device specific' configuration for the debug monitor is done outside of the relocatable object code. I also have the relevant tools (PLM, assembler, linker etc.) running on DOS. I also have cross development tools that run under RSX-11M; so when I have built my J11 S100 card...

The intelligent communications card I am finishing off contains an 8086 CPU, two dual UARTS, local EPROM and SRAM (only small amounts), two Intel counter timer chips. Mainly used as baud rate generators for the UARTS - but one is used as a real time clock interrupt for the CPU. The card can become a bus master (to transfer data packets to/from MULTIBUS memory) and as a bus slave to receive configuration commands etc. From another bus master (a dedicated control CPU).

I have been involved with MULTIBUS 1 equipment since the early 80's - so more than willing to help out where I can here.

Regards,

Dave
 
Last edited:
Is your communications controller proprietary or open hardware? Is it Multibus I?

The Multibus controllers I've seen have been mostly diskless due to the environment. Occasionally there are disks but usually indoors in a lab setting.

It seems the diskless systems can run practically forever.

I'd like to start a hobbyist Multibus project with simple, single function Multibus boards. Sort of like when John and I did with the S-100 boards. However, I got really frustrated when the boards just kept getting more and more complicated and almost indecipherable. There is goodness in keeping things simple.
 
I would be interested in participating at some level. My electronic design career was primarily focused on "backplane busses"; Qbus, STD, VMEbus, Multibus, CPCI. I wasn't a particular fan of Intel but we once designed a 68000-based Multibus board for a customer. I look forward to following this thread as it develops.
-Ken
 
Thinks like standard bus arbitrators in fpga

They don't do FPGA

Eric Smith just went through the exercise within the past year of designing a Multibus arbiter, we'll see if he posts here.
There apparently are some non-obvious things the Intel part does.
 
They don't do FPGA

Eric Smith just went through the exercise within the past year of designing a Multibus arbiter, we'll see if he posts here.
There apparently are some non-obvious things the Intel part does.

Hi Al
If Eric Smith has tried to design his own Multibus bus arbiter has he or anyone else (CCTALK?) made their own open hardware Multibus boards?

Maybe I am just late to the open hardware Multibus party and didn't know it.
 
Maybe jumping to a PowerPC processor board is feasible

I was in Apple hardware for most of the PowerPC run. The easiest part to try to build a board from will be the pin grid modules out of a G3 desktop, which came
with and without L2 cache chips. Everything else is BGA except the 601, which I would stay away from. We also used the Moto XPC106 north bridge (Grackle)
in that system, but it's a BGA.
 
OK, well I'll just retreat to my den and do whatever.

In the meantime, does anyone know what the 74LSxxx equivalent of an Intel 8098 is? It appears to be a tri-stating inverting bus buffer of some sort. The closest I can fathom is some akin to a 74LS125 (quad tri-stating, non-inverting) or 74LS240 (octal tri-stating inverting bus buffer)

Also can anyone confirm that an Intel 8205 is really masquerading as a 74LS138?
 
Our stuff is not 'open' but I will talk to my bosses and see if I can make the files for the Multibus 1 backplane available in the first instance.

Yes, our stuff is MULTIBUS 1.

As far as I am aware, the Intel 8205 is a 3 to 8 decoder as you surmise.

I thought the Intel 8098 was a micro controller though?

Dave
 
Also can anyone confirm that an Intel 8205 is really masquerading as a 74LS138?

the 8205 and 3205 are Schottky parts (18ns prop delay). The outputs can sink a 10 ma, a little more than typical xx138s

you'll need to compare the prop delays and output current capabilites of the parts in different logic
families.
 
They don't do FPGA

Eric Smith just went through the exercise within the past year of designing a Multibus arbiter, we'll see if he posts here.
There apparently are some non-obvious things the Intel part does.

It would not surprise me if there is some serious FM going on in the 8289 bus arbiter. I read the application note AP-51 and it is completely amazing.
 
I ran across a few printed scans of Intel Multibus schematics.

iSBC 80/10-8102A-4 (drawing # 1000949 Rev F)
Logic Diagram - Backpanel, BLC 606 (drawing # 870305380 Rev C)
Printed Wiring Assy, Communications Expansion Board (drawing # 1001197 Rev F)

Not sure if those have turned up already. But if there is interest, I will scan them in this week.
 
Back
Top